Overuse and Underuse of Acute Healthcare: Evidence from Mali

videoType

dynamic-media

videoDmUrl

keyFrameImage

timestamp

00:01 So,

00:01 hello everybody.

00:02 My name is Dion Filmer.

00:03 I'm um

00:05 the director of the World Bank's Development Research Group.

00:08 Um,

00:08 welcome to this policy research talk.

00:10 Uh,

00:10 these talks,

00:11 as some of you may know,

00:12 give us an opportunity to present work coming out of the department,

00:15 the research department.

00:17 With the goal of sharing the findings with each other,

00:19 with colleagues inside the bank,

00:21 as well as uh

00:22 others outside of the World Bank.

00:24 Um,

00:25 just welcome

00:26 our online audiences,

00:27 both on Webex as well as on YouTube.

00:30 Um,

00:32 today my colleague Anya Salman.

00:34 Who is a research economist who actually joined us relatively recently,

00:37 but,

00:38 uh,

00:39 she's a research economist in the human development team of the department.

00:42 Uh,

00:43 she's going to be presenting insights from recent

00:45 research on healthcare demand and utilization in Mali.

00:50 Her talk will focus on the allocation of acute

00:53 curative healthcare

00:54 and,

00:55 uh,

00:56 to what extent demand side policies can affect this allocation.

01:00 Um,

01:01 I'm very grateful

01:02 to have,

01:02 uh,

01:03 Gaston Sorgo here today to discussant.

01:05 Uh,

01:06 Gaston is a practice manager in the Health,

01:08 Nutrition and Population Global Practice of the Human Development Practice Group

01:12 here at the bank.

01:13 He joined the World Bank in 1999 and has previously served as a country manager

01:18 for Mauritania and lead public health specialist and

01:21 human development sector leader based in Cameroon.

01:23 Gaston is a physician with a Master's of

01:26 Public Health from the University of Amsterdam,

01:28 and a doctorate in Public Health from Harvard University.

01:32 So the run of show today is gonna be,

01:34 um,

01:35 Anya will talk for about 45 minutes,

01:37 after which we'll hear from Gaston for 10 to 15 minutes,

01:41 after which we'll,

01:42 uh,

01:42 have questions and answers from the audience.

01:45 Um,

01:45 if you have a question,

01:47 please use the raised hand function in,

01:49 in WebEx or just signal to me in the chat that you have a question and I will call on you.

01:54 If you're following on YouTube,

01:56 uh,

01:56 please put your

01:57 uh question in the chat over there and that will get relayed to me.

02:01 So with that,

02:02 uh,

02:02 over to you,

02:03 Anya.

02:06 Thank you so much,

02:06 Dionne.

02:07 Give me one second,

02:08 I'll share my screen.

02:17 Is this,

02:18 uh,

02:18 is,

02:18 can you all see my screen?

02:22 Great.

02:24 Um,

02:25 so,

02:26 yeah,

02:26 thank you,

02:26 Dion.

02:27 Thank you very much for hosting,

02:28 uh,

02:28 this talk.

02:29 Thank you,

02:29 Gaston,

02:30 for being here as a discussant and thank you also to the audience for,

02:33 for joining us.

02:35 Um,

02:36 as Dion said,

02:36 this talk is,

02:37 uh,

02:38 about overuse and underuse of acute healthcare

02:41 with evidence from,

02:42 with empirical evidence from Mali.

02:45 And before I,

02:46 uh,

02:46 go into this evidence,

02:47 I would like to motivate a little bit why we should

02:50 care about overuse and underuse and specifically in acute care.

02:55 As I'm sure many in the audience know,

02:57 um,

02:57 global health has seen great advances in preventive care in the last two decades.

03:01 Um,

03:03 And there is

03:04 something of a policy consensus today

03:06 that in order to make further significant improvements,

03:09 we really need to focus attention on

03:12 providing better care for acute illness in addition

03:15 to preventive care,

03:16 acute illness and curative care.

03:19 Now,

03:20 unlike in preventive care,

03:21 however,

03:21 in curative care,

03:23 um,

03:23 whether and when any given individual needs treatment varies,

03:26 and it varies based on the person's health status.

03:30 Uh,

03:31 this health status is often unobserved to the outside observer,

03:34 um,

03:34 but it might even be unknown to some degree to the patient themselves.

03:38 Many healthcare decisions are made in the home

03:40 and,

03:40 uh,

03:41 uh,

03:41 and oftentimes,

03:42 um,

03:43 when you

03:43 present with some diffuse symptoms,

03:45 you might not know what your,

03:46 what your illness exactly is.

03:48 In this context,

03:49 misallocation,

03:50 that is the underuse and overuse

03:52 of care is a core challenge for efficiently and with high quality provides

03:57 acute care.

03:59 And this misallocation question is also a measurement challenge because,

04:03 um,

04:03 again,

04:03 as outside observers,

04:04 as researchers,

04:05 but also as policymakers,

04:07 we have to define who needs what type of care

04:10 and then measure if they actually receive this care.

04:13 And of course,

04:13 this allocation question

04:15 is closely linked to the Good Health and well-being SDG

04:18 which mentions among other things,

04:19 pay for performance,

04:20 sustainable financing,

04:22 quality of care,

04:23 and so on.

04:25 So why,

04:26 um,

04:27 uh,

04:28 look at both underuse and overuse,

04:30 um,

04:30 in low and middle-income country contexts?

04:33 For surely,

04:34 I don't need,

04:34 uh,

04:34 don't need,

04:35 don't need to motivate,

04:36 um,

04:36 uh,

04:37 the question of underused.

04:38 Underuse remains,

04:39 uh,

04:40 even today a challenging problem.

04:41 Despite substantial progress made,

04:44 just to give one number,

04:45 in 2019,

04:47 5.2 million

04:49 children under five years have been estimated to have died,

04:52 mostly from preventable and treatable causes such as diarrhea,

04:56 malaria,

04:56 and pneumonia.

04:58 Part of the policy response uh to this underused problem

05:03 has been a

05:04 fundamental shift from a user fee model

05:07 to providing free care for children and mothers in particular.

05:12 Uh,

05:12 this is a pivot away from the Bamako consensus in 1987.

05:16 And the same thing has actually happened in Mali,

05:19 where in 2019,

05:21 the Ministry of Health

05:22 announced the intention to provide free children's and mother's healthcare

05:26 um

05:27 by 2022.

05:30 Having said all of this,

05:31 however,

05:32 um,

05:32 there's at the same time,

05:34 a lot of evidence that there's also an overuse problem.

05:36 So medication misuse has been documented since at least,

05:39 uh,

05:40 1990.

05:41 Um,

05:42 there is a WHO medicines use summary report in 2006

05:46 that shows some of this evidence,

05:48 um,

05:48 and what the report

05:50 says,

05:50 among other things,

05:51 that is that there's substantive overuse of injections and antibiotics,

05:55 a lot of instances of polypharmacy and so on.

05:58 Polypharmacy is,

05:59 uh,

05:59 when

06:00 the same set of symptoms is treated with treatment for different kinds

06:03 of illnesses such as malaria treatment and antibiotics at the same time.

06:08 More recently,

06:08 there has been also a lot of experimental and

06:10 other study evidence of these misuse and overuse problems.

06:14 Many of those studies came actually out of the World Bank Research Department,

06:18 and they show that um malaria drugs and antibiotics are overused,

06:22 and more generally that there's a lot of unnecessary and even harmful treatments

06:26 and insufficient diagnostics.

06:27 a variety of conditions.

06:30 Oftentimes this evidence shows that this type of overuse

06:33 is the result of subsidies on the patient's side

06:36 as well as financial incentives on the doctor's side.

06:40 And this raises the question,

06:42 if we try to combat underuse by introducing subsidies for patients,

06:48 could this reinforce financial incentives for providers and lead

06:51 to greater rates of overtreatment at the same time?

06:58 So,

06:59 when we think about the allocation of care and the correct use of care,

07:03 um,

07:03 this is certainly a challenging problem and the reason is

07:06 that there are multiple and complex decisions involved

07:09 from visit,

07:10 from the decision to visit a

07:12 provider through the diagnostic and prescription process

07:15 to actually purchasing and then adhering to treatment.

07:18 Um

07:20 And as a result,

07:21 there are also many potential sources of misallocation.

07:25 Um,

07:26 on the household side,

07:27 um,

07:27 certainly,

07:28 we are acutely aware that there are financial constraints,

07:31 meaning that families

07:32 may not be able to raise funds to cover emergency,

07:35 emergencies and acute care.

07:37 There might also be informational constraints,

07:39 for example,

07:40 because households are unable to assess

07:42 symptoms or correctly administer treatments.

07:44 And uh a general recognized problem in healthcare

07:48 is that there's often misaligned incentives when allocating care

07:52 between the government,

07:53 the funder,

07:54 often of care,

07:55 but providers,

07:56 but also caregivers and patients.

07:59 And these incentives are often due to uh

08:01 externalities and uninternalized costs and benefits of care.

08:07 One thing that reinforces all these issues

08:09 is that there is a fundamental asymmetric

08:11 information problem which is termed in health

08:13 economics the problem of the informed expert.

08:16 Meaning that there is one

08:18 person in this process,

08:20 uh,

08:20 or,

08:20 or a very small set of persons,

08:22 the healthcare provider,

08:24 who can actually diagnose and assess

08:25 the illness and make treatment recommendations.

08:28 And everyone else in the process has to rely on this provider's,

08:32 um,

08:32 uh,

08:33 advice.

08:34 But this can lead to issues if um

08:37 If there is,

08:38 if there is misaligned incentives.

08:41 Now,

08:42 what is,

08:42 what comes on top of this since there are so many decisions involved,

08:45 these different issues may interact or mutually reinforce each other and

08:50 um one example could be that if if if patients due to

08:54 um the provider having different

08:56 interests from them,

08:58 lack trust,

08:59 this could lead to a lack of effort on the parts of the providers

09:03 um to diagnose the illness and give uh and give specific advice.

09:09 So,

09:09 in this talk,

09:10 I'm certainly not gonna be able to talk about all of the issues that come in here,

09:13 but I would like you,

09:15 uh,

09:15 I would like to present you with 3 studies plus 1 unpublished study from Mali

09:20 where we,

09:20 um,

09:21 collected experimental evidence on underuse and overuse of curative care.

09:25 And then I would also like to take the

09:27 opportunity to point out some new measurement approaches,

09:30 uh,

09:30 that we've taken.

09:33 So there's two sets of studies.

09:34 Um,

09:35 the first two papers with my co-authors,

09:38 uh,

09:38 Christine Blander,

09:39 Sam Brown,

09:39 and Mark Dean

09:41 covers the decision about care at home,

09:44 uh,

09:44 specifically children's care at home.

09:46 And,

09:47 uh,

09:47 these,

09:48 these,

09:48 these studies are part of an RCT

09:51 where we,

09:51 uh,

09:52 study two common policy interventions to improve access to care,

09:55 specifically subsidies

09:57 and health information provided by community health workers,

10:00 CHW for short.

10:02 What's novel about this study is that we conduct a very detailed home survey

10:07 where we have 9 weeks of daily health diary data on symptoms,

10:11 consultations,

10:12 and so on.

10:14 And there's two papers that I would like to talk about.

10:17 The first is the actual randomized control trial,

10:19 RCT.

10:20 Here,

10:21 we examine the decision to seek care,

10:23 so the extensive margin of care,

10:24 should we get care or not

10:26 for children under 5.

10:27 And in order to understand overuse and underuse,

10:30 we construct illness spells,

10:32 spells

10:32 and assess the needs for provider evaluation using the community

10:37 version of the Integrated Management of Childhood illness guidelines,

10:40 IMCI.

10:41 Um,

10:42 which are guidelines for community health workers with rule charts

10:45 that indicate when a child should see a doctor.

10:49 I'm gonna talk much more about these studies.

10:50 This is more a summary of what I'm gonna show you.

10:53 The second study,

10:54 uh,

10:55 primarily examines,

10:56 uh,

10:57 gender differences,

10:58 but,

10:58 um,

10:59 also gives us some insight on overuse and underuse

11:02 because it looks at antibiotics adherence.

11:05 And what we do is construct

11:07 spells of antibiotics treatments and then compare these

11:10 with WHO guidelines for minimal treatment length.

11:14 The second set of studies I want to talk

11:16 about is about behavior during the provider visit.

11:19 So what happens when you actually go at the clinic.

11:22 Um,

11:22 and what we did here is to conduct a randomized controlled trial

11:27 that isolates the contributors to misallocation

11:29 of here specifically subsidized malaria medication.

11:35 How we measure,

11:35 how we approach the measurement,

11:37 uh,

11:37 question here is by conducting a clinic survey,

11:40 we conduct entry and exit interviews,

11:43 but then we also do a home follow-up where we carry

11:45 out a malaria test so we are able to actually see

11:48 whether patients who have malaria did get treatment and

11:51 those who didn't have malaria did not get treatment.

11:55 The first study looks at the first

11:57 of the two cross-randomized interventions in this RCT

12:01 where we handed out vouchers for malaria treatments,

12:04 uh,

12:04 artemisinin combination therapy,

12:06 ACTs,

12:07 either directly to patients or to the provider,

12:10 and we use the difference between those two arms to ask,

12:14 when doctors have discretion to decide about giving a

12:17 voucher or not giving a subsidy or not,

12:19 do they use that discretion to steer patients towards more or less use?

12:24 And again,

12:24 I'm going to talk more about these studies and the results.

12:26 The second intervention tried to approach this issue from a different angle,

12:30 and we asked how information

12:33 can encourage adherence to test results

12:35 and,

12:36 and improve the allocation of malaria treatment.

12:38 And we did this by providing information on malaria tests,

12:42 uh,

12:42 rapid detection test,

12:43 RDT efficacy,

12:44 either to doctors through a day-long training or to patients through a short

12:49 video.

12:51 Now,

12:52 um,

12:53 in,

12:53 uh,

12:54 to give you a preview of some of the findings

12:56 and to summarize a little bit how these two,

12:59 these studies all speak to overuse and underuse,

13:01 let me summarize for you.

13:03 Um,

13:04 we find that in care seeking underuse

13:07 as um measured by the IMCI is,

13:10 is definitely alleviated but by no means eliminated by subsidization.

13:15 And,

13:16 um,

13:16 despite this increase in use that is mostly um uh an alleviation of underuse,

13:22 there's also a significant share of overuse

13:25 in those newly created provider visits from the subsidy,

13:28 around 30%.

13:29 And on top of this,

13:30 information does not

13:32 improve the allocation of care for the information provided by the health workers.

13:37 Moreover,

13:37 when we look at the clinic,

13:39 when costs are low and the subsidies available,

13:42 providers seem to fail in their gatekeeping function,

13:44 keeping prescription medications out of the hands of patients who don't need them.

13:49 And one significant factor in this seems to be patient pressure who,

13:52 patients who are asking for these medications.

13:56 We find among other things that informed doctors who know informed,

13:59 in quotation marks,

14:00 doctors who received the training on RD on RDT tests

14:04 are better gatekeepers,

14:05 but this actually leads to a reduction in patient satisfaction.

14:09 And then finally,

14:10 on the treatment side,

14:11 on the adherence to medication side,

14:14 we find that even after the medication was purchased here,

14:16 we're looking at antibiotics,

14:18 but the cost should be low,

14:19 and yet,

14:20 there is a lot of underuse and it is even higher for girls than for boys.

14:26 So this raises a number of questions,

14:28 among other things,

14:29 how do we direct

14:30 subsidized care to those who need it the most and make

14:33 sure that resources that are used serve their intended purpose.

14:37 But in addition,

14:38 what these pieces of evidence highlight is that

14:41 there are multiple margins of overuse and underuse,

14:44 and I will argue that this raises a number of important open questions.

14:48 For example,

14:49 do these margins interact and possibly reinforce each other,

14:52 and then what are the actually the net effects on health outcomes of these,

14:56 of these misallocation issues?

14:59 So what I'm going to do in the rest of the talk is I'm going to do,

15:02 to introduce the setting and show you some summary statistics of our samples.

15:06 And then I'm gonna talk about each of the studies in turn,

15:09 the RCT on subsidies and healthcare,

15:12 um,

15:12 workers,

15:13 health workers,

15:13 community health workers,

15:14 sorry,

15:15 um,

15:15 the study on overuse of malaria drugs and then the adherence study for antibiotics,

15:20 and then I'm going to conclude.

15:23 So to start with the setting,

15:25 um,

15:26 Public health in Mali is provided uh through

15:29 decentralized community-based care funded by user fees.

15:32 This is in accord with the Pharmaco consensus.

15:35 And a lot of this care is provided by so-called SSCO's Community Health clinics.

15:41 Um,

15:41 and these clinics are commonly used,

15:43 so 47% of mothers,

15:45 uh,

15:45 reported in the DHS

15:47 in Bamako that they took a child with a fever or cough to a CSO.

15:52 And there's around 70 of such,

15:54 of such clinics and associated reference hospitals all over Bamako.

15:58 Our,

15:58 uh,

15:59 malaria overtreatment study um

16:02 uh was conducted at 60 of these clinics,

16:04 and we have a census

16:06 that gives us some insight into what these clinics look like.

16:10 So,

16:10 they have about 30,

16:12 they have about 30 patients a day on average.

16:16 Um,

16:16 most of these clinics actually have a

16:18 basic laboratory where they can conduct microscopy.

16:22 And many of,

16:23 uh,

16:24 all of these uh SSOs,

16:25 at least in Bamako in the capital,

16:27 have a dispensary that has essential medicines available for sale.

16:32 Around 11 staff have um the,

16:35 the ability to write prescriptions for things such as malaria

16:39 treatments.

16:40 Um,

16:40 and it should be said these are,

16:41 these are doctors and,

16:42 uh,

16:43 and,

16:43 and nurses,

16:44 but also medical interns who staff these clinics after hours.

16:48 And you can see that there is a resource constraint.

16:51 Uh,

16:52 many of these clinics report temporary stock-outs,

16:55 but only in 1.7% of cases,

16:57 they are really stocked out of all the malaria drugs on a given day.

17:02 So our samples come from the two RCTs and the first is the

17:06 Action for Health RCT that studies the

17:09 subsidy and community health worker intervention.

17:12 Here,

17:12 we conducted a home survey with 1,768 children in 642 compounds.

17:17 A compound is a living,

17:19 a sort of a living unit,

17:21 uh,

17:21 where several households share some common infrastructure.

17:24 And,

17:25 um,

17:25 in the sample included,

17:26 are all families eligible for the program.

17:29 This means they have children under 5,

17:31 they passed a proxy means test,

17:33 and,

17:33 um,

17:34 they lived in the expansion area of Action for Health,

17:36 which was a program by the,

17:38 um,

17:39 by the health

17:40 organization Mali Health.

17:41 Um,

17:42 the second sample is the clinic data with 2,055 patients

17:46 with acute systems collected at the 60 SECOs in Bamako.

17:51 And as you can see,

17:52 these samples are actually quite fairly similar.

17:55 So our action for health data comes from the household heads,

17:59 but you can,

18:00 and the,

18:00 and the,

18:01 and the clinic data comes from patients.

18:03 There are some differences in literacy rates because

18:05 uh women generally have lower literacy rates,

18:08 but you can see overall literacy is low.

18:10 Only 13% of our action for health households have salaried employment.

18:15 Um,

18:15 the households are relatively large.

18:18 But a little bit smaller in the Action for Health program,

18:21 probably because these are uh relatively young families,

18:24 but they are similar in average monthly income per person

18:28 between $22,000 and $26,000 safer.

18:31 So in US dollars,

18:33 around $40,000 to $45 per person.

18:37 And as you can see,

18:38 at least in Action for Health,

18:39 this is an urban area,

18:40 so this means these households live relatively close to the clinic,

18:43 only around 1/3 of a mile or 0.5 kilometer away on average.

18:50 So,

18:50 let's start with the first study,

18:52 care seeking for children.

18:53 Um,

18:55 To motivate the original study,

18:58 what we were interested in is to understand

19:01 the effects of subsidies and how they affect the allocation of care.

19:05 Subsidies

19:06 have the potential to reduce underuse due to uh

19:09 access barriers such as credit constraints and incorrect beliefs,

19:13 but because they monotonically increase use,

19:16 they may also lead to overuse,

19:17 and in that case,

19:19 information may help improve the allocation.

19:21 Information is here provided by community

19:23 health workers who visit families biweekly

19:26 and

19:27 discuss health um and healthy and symptoms

19:30 and health needs with parents.

19:32 In the experiment,

19:33 we randomize compounds into these two treatment arms,

19:38 um,

19:39 cross-section,

19:39 uh,

19:40 we do a cross-randomization,

19:41 so some households receive both and some only eat one of each of the interventions.

19:46 The first one is the subsidy intervention where,

19:49 um,

19:50 Action for Health pays

19:51 for clinic user fees

19:53 for the five most common conditions for children under 5,

19:57 and then the information intervention of the community health worker visits.

20:01 Um,

20:03 One puzzle in the literature on these types of interventions is that policies that

20:08 reduce the cost of acute care often

20:10 actually have relatively little health outcome effects.

20:12 And we don't always know why.

20:14 And so what we did in this paper is

20:17 that we measure health utilization directly

20:20 using the diary data I mentioned before against the IMCI benchmark.

20:25 So what we have to assume in order to define overuse and

20:28 underuse is that compliance with the IMCI actually is important for health.

20:33 So in other words,

20:34 when the IMCI says by the rule chart that a child should seek care,

20:38 if this child does not receive care,

20:40 we think of this as underuse of,

20:42 of,

20:43 of the doctor.

20:45 And in analyzing this data,

20:47 we will treat the decision to seek care as basically a stopping problem.

20:51 The parents make a decision every day watching the illness of their child,

20:55 um,

20:56 whether they should go today or wait for another day to seek care.

20:59 And the IMCI works kind of in a similar way because it gives,

21:02 um,

21:03 it gives guidelines for how long a kid should a kid should have,

21:07 should wait before care should be sought,

21:09 for example,

21:09 for things like diarrhea.

21:11 And so what we do is we use the MCI to classify a child during an an illness spell

21:16 as either early for care,

21:17 so you don't

21:18 yet need care

21:20 or care required uh and care should be sought

21:23 today or should have been sought in the past.

21:27 So now for the data collection,

21:29 um,

21:30 we had surveyors visit the households 9 months after the action for Health Start.

21:34 Um,

21:36 And they collected,

21:37 they,

21:37 they constructed detailed health diaries with

21:39 uh the mothers.

21:41 And just to give you an idea,

21:42 out of about 60 days of illness,

21:44 we uh 60 days of,

21:46 of 60 diary days that we um collected for each child,

21:51 around 16 days,

21:53 the average child had some symptoms that have not yet received treatment.

21:58 And the five symptoms listed here are the most common symptoms,

22:01 and you can see that colds and coughing are very,

22:04 very common among those illness days.

22:06 And those are not as serious and in many cases do not require care.

22:10 However,

22:11 in 32% of cases,

22:13 there's also a report of,

22:15 uh,

22:15 unusually hot or cold skin,

22:16 so in other words,

22:17 potentially a fever.

22:19 In 9% of cases,

22:20 there are more than 3 loose stools,

22:22 or in other words,

22:22 diarrhea,

22:23 and in 5% of cases,

22:25 uh,

22:25 the child is reported to vomit everything,

22:28 which is actually one of the 4 danger signs

22:30 we collect and would require immediate medical attention.

22:35 On top of these symptoms records,

22:37 we have records on doctor consultations and medication purchases,

22:42 and this is the first um,

22:45 Impression of how the subsidy has affected care seeking.

22:49 So you see here that only 89

22:52 times

22:53 our children see a private provider.

22:56 But 646 times they actually go and see and see a a provider at the SESCO.

23:03 And what's striking is that the vast majority of these visits,

23:06 508 take place in the groups that received the healthcare subsidy.

23:10 And you can see part of the reason is,

23:12 of course,

23:12 the cost difference between the two visits.

23:15 Um,

23:16 the costs of the visit are much lower than there is a subsidy,

23:19 and I should say the costs are always a little lower than

23:21 the value of care received or the total price of care.

23:23 And part of the reason is that many households seek outside help to finance,

23:27 uh,

23:27 their care needs.

23:29 Um,

23:29 but you also can see,

23:30 I'm not gonna talk much about this,

23:32 but you can see that a vast majority of healthcare is actually obtained in informal,

23:36 from informal providers,

23:37 informal sources of medication

23:40 without any medic any consultation of a medical professional.

23:46 So this slide shows you the main results of this um study.

23:51 Um,

23:51 what you see here is the 4 treatment groups.

23:54 Um,

23:54 on the left side,

23:55 you see the groups without subsidies and on the right with the subsidy.

23:59 And then on the top,

24:00 you see the groups,

24:01 um,

24:01 without the health worker visits and in the bottom with the health worker visits.

24:05 We have a total of nearly 4000 illness spells for these children.

24:10 And these spells last on average 6,

24:12 a little over 6 days and have 3 care required days.

24:15 There's huge amounts of variation,

24:17 so some,

24:18 some,

24:18 some reason for the high,

24:19 uh,

24:20 high number of average,

24:21 the high average length

24:22 is actually that there's some real outliers

24:24 where kids are sick almost the entire period

24:27 that we interview.

24:29 Um,

24:29 but what you see are basically three main points.

24:33 The first is,

24:34 you see this in the control group even,

24:36 um,

24:37 So,

24:37 sorry,

24:37 I should say what's on the graph.

24:39 So on the X-axis,

24:40 you see the date of the illness spell.

24:44 And then the two lines represent the probability of seeking care on

24:48 either a day that is early for care according to the IMCI

24:52 or where care is required according to the IMCI and

24:55 on the Y axis is the probability of care seeking.

24:58 And what you see is that um the probability of care

25:00 seeking is certainly a lot higher than care is actually required.

25:05 What you also see is going from the left to the right,

25:08 um,

25:09 as you,

25:09 as these households receive the subsidy,

25:11 care-seeking increases a lot,

25:14 um,

25:15 and,

25:15 and this increase is concentrated on care-required days,

25:18 but by no means is care-seeking at 100% on care-required days.

25:23 And then finally,

25:24 what you see is that when you go from top to bottom,

25:26 that the community health workers actually had

25:29 surprisingly little effect on care-seeking behavior.

25:32 And in particular,

25:33 what they do not do is

25:35 direct children better towards uh care when care is required

25:38 versus not receiving care when care is not required.

25:41 So they do not seem to improve the allocation very much.

25:45 So to summarize the findings,

25:47 um,

25:48 we estimate these formally with a hazard model,

25:51 and we find there is an over 5 times higher probability of care

25:54 on care required days than on early days.

25:57 So what this seems to indicate is that parents can

25:59 clearly distinguish differential need for care by child's health status.

26:03 They just underuse care overall.

26:06 Um,

26:06 the subsidy is successful in increasing care seeking.

26:10 Um,

26:10 the probability increases by 250%.

26:14 And as a result of the differential care

26:16 seeking and care required versus early days,

26:19 about 70% of the new visits are actually IMCI recommended care,

26:24 which is good news.

26:25 However,

26:26 it is still the case that 71% of care required spells

26:30 still don't receive

26:31 any care,

26:32 even in the subsidy condition and care is almost entirely free.

26:38 So what we see is that underuse is alleviated,

26:40 alleviated but not eliminated by these subsidies.

26:43 And moreover,

26:44 because underuse and overuse increase at the same rate,

26:48 despite the very low probability of overuse on any given early day,

26:53 there is still a significant share of utilization,

26:55 uh,

26:56 share of doctor visits that actually um constitute overuse,

27:00 um,

27:00 according to the IMCI.

27:02 And,

27:03 um,

27:03 in addition,

27:04 the information provided by the community health

27:07 workers does not improve this allocation.

27:10 So what,

27:11 what this,

27:11 what that this raises,

27:13 uh,

27:13 raises some questions,

27:15 in particular,

27:15 how could we direct care better,

27:17 um,

27:18 the subsidized care better towards these care-required illness episodes versus,

27:22 uh,

27:22 care not required episodes.

27:26 Now for the 2nd study on patient demand for treatment,

27:30 this is now a study

27:31 at the clinic,

27:33 and we are focusing on prescription and purchasing behavior.

27:37 The motivation was the policy challenge of overtreatment

27:41 and wrong treatment obtained at clinic visits,

27:43 even in very low resourced settings.

27:46 And our data gives a good impression of this.

27:49 So from our home visit data where we conducted malaria tests,

27:52 we know that 42% of malaria negative patients received

27:56 treatment.

27:57 And on top,

27:58 20% of malaria negative and 53% of malaria positive patients

28:03 received treatment for severe malaria.

28:05 So to give you an impression,

28:06 severe malaria is a,

28:07 is a crisis state of malaria,

28:09 so to speak.

28:10 It usually comes with convulsions,

28:12 um,

28:13 organ failure,

28:14 sometimes coma,

28:15 and,

28:16 um,

28:16 and it is an imm,

28:17 it requires immediate medical,

28:19 uh,

28:20 very,

28:20 very significant medical intervention.

28:23 Um,

28:23 it's treated typically with injections and then

28:26 an immediate referral to the hospital.

28:28 It is typically estimated that out of all malaria cases,

28:31 around 10% escalate into severe malaria,

28:35 and because we interviewed these patients as patients at home,

28:38 they were clearly not referred to the hospital.

28:40 So there's good reason to believe that a lot of

28:42 this severe treatment is actually overtreatment of some form.

28:46 Now,

28:46 the question is,

28:48 is this overtreatment driven by providers?

28:51 Um,

28:52 the reason why this question is prevalent is that

28:54 a lot of health economics focuses on the financial

28:57 incentives for providers to overprescribe and increase the budget

29:00 of the clinic or increase their own income.

29:03 Um,

29:04 there is some evidence in the data for this possibly being the case.

29:08 First of all,

29:08 patients in our data do not always,

29:11 only in about 2/3 of the cases actually buy

29:14 a prescription for an anti-malarial that they receive.

29:17 And moreover,

29:18 severe treatment is over 2 times more expensive than simple treatment,

29:22 even though,

29:22 and it is prescribed very often,

29:24 and a lot of this,

29:26 a lot of the fees are actually fees for labor,

29:28 so they would go directly into the clinic salary budget.

29:33 On the flip side,

29:34 when you actually ask providers why over-prescription happens,

29:37 they will tell you something different.

29:39 They will say the patients come to the clinic knowing

29:41 what they want and they ask for this treatment.

29:44 And again,

29:44 we have evidence for this in our data.

29:47 Before the consultation,

29:48 55% of patients we interviewed believe specifically that

29:52 they have malaria and need treatment for it.

29:55 And 57% of our health workers report that they were,

29:59 that they feel pressured to prescribe medications and oftentimes they mention.

30:03 Anti-malarials specifically.

30:05 What's remarkable about this is because these

30:08 doctors,

30:08 these providers have prescription power,

30:11 this requires to some degree reluctant compliance by them giving

30:15 in to the patient's demands against their own better judgment.

30:21 So in order to test which of these two explanations is prevalent,

30:24 we conducted a,

30:26 an,

30:26 a randomized trial where we rotate each clinic through 3 pricing conditions

30:31 and the price variation is for

30:34 the simple treatment for the malaria treatment that

30:37 is standard and recommended for simple cases,

30:39 the artemisinin-based combination therapies,

30:42 um,

30:43 these are simple tablets taken over 3 days.

30:46 On control days,

30:47 there are no

30:48 malaria discounts available at all,

30:51 but then there are two additional interventions.

30:54 The patient voucher intervention

30:56 gives out these free vouchers

30:59 for simple malaria treatment,

31:00 and these vouchers are directly given to patients.

31:04 Um,

31:04 the vouchers are not valid for severe

31:07 treatment or any treatment that involves injections,

31:10 and they are received importantly before the patient even sees the doctor,

31:13 so they come into the consultation

31:15 knowing this treatment is free for them.

31:18 On doctor voucher days,

31:20 by contrast,

31:21 the doctors obtain this exact same voucher.

31:24 It has the same terms and conditions as in the PV arm.

31:28 However,

31:29 they have the discretion to give the voucher to patients as they see fit.

31:33 The patients do not know in advance

31:35 that the ACT is free before they uh consult with the doctor.

31:42 Now,

31:43 how does this help us distinguish uh who drives overtreatment?

31:46 Well,

31:47 there are different predictions for treatment effects under the,

31:50 under the two treatment arms.

31:53 First of all,

31:54 imagine we give these vouchers for patients.

31:56 Um,

31:57 this reduces the cost of simple malaria treatment and

32:00 keeps the cost of all other treatment the same.

32:02 So what this should do is,

32:04 it should increase

32:05 the demand for the simple ACT tablets

32:08 at the expense of not receiving any malaria treatment

32:12 or receiving severe treatment,

32:14 receiving this treatment that involves an injection.

32:18 Now from the doctor's side,

32:20 when we give the vouchers to doctors,

32:21 if they show the voucher to the patient,

32:23 that should have the exact same effect on patient demand.

32:26 However,

32:27 they can,

32:28 they can decide not to show the voucher to the,

32:30 to the patient and use their information advantage

32:33 to steer the patient's demands.

32:36 And one option they have is they could

32:38 keep overall demand lower than in the PV condition

32:42 by

32:43 not showing the voucher to patients who would otherwise buy no treatment,

32:47 but showing the voucher,

32:48 giving the voucher to those who would request severe

32:51 treatment and thereby reduce demand for severe treatment only.

32:55 If we see this,

32:56 this would be evidence of patient-driven overtreatment,

32:58 patients actually pushing doctors and doctors trying to keep demand low.

33:03 On the flip side,

33:03 what they could also do is they can use

33:05 their information and advantage to increase overall demand,

33:08 so they could show the voucher,

33:10 give the voucher to patients who would otherwise not buy treatment,

33:13 and,

33:13 and,

33:14 uh,

33:14 increase purchasing rates and prescription rates,

33:17 but at the same time,

33:18 not use the voucher for those patients who would otherwise get severe

33:22 treatment and therefore keep demand for the severe treatment option high.

33:26 If this is what we see,

33:27 then we would conclude that this is,

33:28 that the overtreatment we see is doctor-driven and there's a

33:32 significant push to sell more on the doctor's side.

33:36 So what do we find?

33:37 Well,

33:37 we find the overwhelming evidence that

33:39 there's actually significant patient-driven demand,

33:42 demand at the margins where we,

33:44 where we,

33:45 where we see changes.

33:46 Um,

33:47 we find that in the,

33:49 in the patient voucher arm,

33:51 significantly more vouchers were used than in the doctor voucher arm,

33:55 despite prices being the same.

33:57 We also found that

33:58 that doctors make significantly more anti-malarial prescriptions

34:02 in the patient voucher arm than in the doctor voucher arm,

34:05 and you can see

34:06 that there is not a large difference

34:08 to the control where the price discount was not received at all.

34:12 And then finally,

34:12 we find a significant increase corresponding to this prescription increase

34:17 in anti-malarial purchases.

34:19 So many more patients

34:21 actually receive an anti-malarial in PV

34:24 over DV.

34:25 We also see,

34:26 and this is what I mentioned before,

34:27 we see that you can see the large difference in purchasing rates

34:31 in the,

34:32 in the control group,

34:33 so many patients actually do not purchase treatment,

34:35 and there is a significant increase when we

34:37 receive when when vouchers are actually available.

34:40 Many patients are now able to afford the treatment.

34:45 So,

34:46 what,

34:47 what,

34:47 uh,

34:47 what else do we find?

34:49 Um,

34:49 first of all,

34:50 we,

34:51 in contrast to these results on,

34:52 on the patient demand side,

34:54 we find no evidence that there's doctor-driven demand.

34:57 In particular,

34:58 we find that both in DV and in PV the treatment reduces

35:02 significantly how much severe treatment is used.

35:05 So doctors substitute patients from the more

35:08 expensive to the less expensive treatment option.

35:11 And what we also find in general using our home testing

35:14 data is that there's a really poor match between treatment and illness

35:18 in both directions.

35:19 So some patients who have malaria in the home test did not receive an anti-malarial,

35:24 and

35:24 many patients,

35:25 I mentioned this already,

35:26 who do have,

35:27 did not have malaria did receive treatment.

35:30 But what we find in particular is that this illness treatment match is worse

35:34 in the patient voucher than in the doctor voucher condition,

35:37 even though there is actually some improvement in the underuse

35:41 component of this mismatch

35:43 from greater purchases that result from the fact that

35:46 in the voucher conditions,

35:47 treatment is now free.

35:51 What we also find is,

35:52 uh,

35:53 this is the 2nd paper,

35:54 the unpublished paper that um that I mentioned

35:56 earlier in the intervention where we provided training.

36:00 When doctors receive additional training on the efficacy

36:03 of tests and using those tests for,

36:06 for making treatment decisions,

36:08 they actually make better prescriptions.

36:10 The treatment illness match improves,

36:12 but we find that this lowers patient satisfaction,

36:14 which is another piece of evidence that these patients are pushing

36:18 for more medications.

36:21 So to summarize,

36:22 we find that at least some of the overuse observed

36:26 at the clinic is surprisingly driven by patient demand.

36:30 Um,

36:31 we find that information in the form of the training

36:33 we provided had an effect on the doctor's side,

36:36 but the video we showed at least on the patient's side,

36:38 did not improve allocation.

36:42 And what those two things point to

36:45 is that there is a twin problem,

36:47 a surprising twin problem where we observe underuse at the extensive margin,

36:51 meaning whether care is received at all,

36:53 whether a patient sees a doctor.

36:56 But then there's overuse at the intensive margin,

36:58 namely,

36:58 how much care is actually received.

37:01 And

37:02 one way of,

37:03 of thinking about this is that there's at least a possibility that these two

37:07 problems reinforce each other.

37:08 I mentioned this before.

37:10 So,

37:11 suppose only a small subset of patients go to the clinic.

37:14 These might be exactly the patients with the highest and

37:17 the strongest demands for some kind of um powerful treatment.

37:21 So in order to meet these patients' preferences,

37:23 the doctors tend to overprescribe,

37:25 and that makes care on average more expensive.

37:28 But this could have this paradoxical effect that even fewer people go to the clinic.

37:32 And on top of this,

37:33 those who do go to the clinic,

37:35 uh,

37:36 have less trust

37:37 in the doctor's recommendations.

37:39 And that in turn,

37:40 and this is the subject of the who signed it al.

37:42 paper that just recently came out,

37:45 this in turn might mean that doctors have less

37:47 of an incentive to actually put effort into diagnostics

37:50 and giving good treatment recommendations,

37:52 but rather they just follow what the patients want because

37:54 they anticipate that patients will not follow their advice closely,

37:58 um,

37:59 anyways.

38:00 So,

38:01 this is at this point,

38:02 of course,

38:02 we need more evidence that these,

38:04 these chains of reasoning hold,

38:06 but there is um at least some reason to be concerned about them.

38:11 If true,

38:11 then this is a really challenging policy issue

38:14 because of the aforementioned informed expert problem.

38:17 The provider is really the one who diagnoses

38:19 the patients and knows what treatment is needed,

38:22 and we rely on the provider to both

38:25 uh give

38:26 medications when needed,

38:27 but also gatekeep medications that are not needed.

38:30 And monitoring this behavior would really require observation by yet

38:33 another qualified provider and that's just very often not feasible.

38:37 And so one interesting

38:40 potential that arises from what we found on our information

38:43 treatments is that paradoxically actually empowering providers might here have

38:47 played a role in improving

38:49 the allocation of care.

38:53 So the last study is on treatment and adherence,

38:56 and it originally focused on gender

38:58 differences in antibiotics treatment adherence,

39:01 but it also has some

39:03 things to say on overuse and underuse.

39:07 So what we do in this paper is we use the same

39:09 health diary data from before and we construct antibiotics treatment spells.

39:14 And benchmark exit from the spell or stopping the treatment

39:17 against WHO recommendations.

39:19 And we can use our data with all our detailed

39:22 symptom records and so on to estimate on the,

39:24 on the one hand,

39:25 the probability of actually starting a treatment spell

39:28 and then conditional on having started

39:30 completing that treatment spell,

39:32 and we can look at the role of gender and

39:34 other variables um in this decision to end the treatment.

39:39 The table below just gives you an

39:41 overview over our antibiotics spell observations.

39:45 Um,

39:45 we have a total of 712 instances of antibiotics used in our data,

39:51 and as you can see,

39:52 most of the standard recommendations involve essentially 5

39:55 days of antibiotics treatment with some exceptions for,

39:58 for other,

39:59 for specific conditions.

40:01 So this graph gives you kind of the the headline results.

40:05 Um,

40:05 it shows on the x axis again are spell days,

40:09 and it shows you how many children are still taking the antibiotic on each day.

40:14 And of course,

40:15 if we,

40:16 if this is an antibiotic where the 5-day treatment recommendation holds,

40:19 we should see 100% use up to day 5.

40:22 And what we see is that unfortunately,

40:24 that is not at all what we find.

40:26 Um,

40:27 the girls in this data set have only about 38% completion rates of antibiotics,

40:33 and boys have higher rates,

40:35 but by no means,

40:36 um,

40:36 full full adherence,

40:38 but it's a significant difference of 9 to 12% um between the two genders.

40:44 And what we find is surprisingly,

40:47 that antibiotic

40:49 that this completion advantage for boys accrues almost

40:52 entirely to the sons of educated household heads.

40:56 What you see in this graph here is the gender effect,

40:59 um,

41:00 the difference in antibiotics adherence

41:02 for boys and girls,

41:04 um.

41:06 Split up by education level of the household head.

41:10 The reference category is no formal schooling.

41:12 And what you see is that in particular,

41:14 in highly educated households,

41:16 this gender difference exists,

41:17 whereas in,

41:18 in,

41:18 in households where the household head is not as highly educated,

41:22 um,

41:22 there is not much of a gender difference.

41:25 One interpretation of this is that um is as follows.

41:29 Education is essentially needed in order to make good adherence to decisions.

41:33 This could be because it comes with higher literacy rates and so on.

41:37 Household heads are typically much more educated,

41:39 and they are also typically male.

41:42 And more generally,

41:42 there's a huge education gap between women and men in,

41:46 in Mali in our data.

41:48 And what might be happening here is that these male household heads who,

41:51 who are able to read and follow the treatment instructions,

41:55 just pay very different attention to the care of

41:57 girls in the household over the care of boys.

42:02 And um so,

42:04 so to summarize with an eye to overuse and underuse,

42:08 despite a really low cost of adherence,

42:11 the antibiotic is already purchased,

42:13 um,

42:14 it's in the hands of the household.

42:16 There is still significant underuse in the sense

42:19 that many,

42:20 um,

42:21 that many households do not adhere to the full treatment length.

42:24 It also seems that the presence of a,

42:26 of a,

42:27 of a literate,

42:27 educated parents matters significantly.

42:30 And this suggests that there are subtle biases in the,

42:33 in the use of treatment

42:35 that may affect other aspects of treatment success that we may not have a lot

42:38 of information on because they happen in the home and we don't very often see,

42:43 uh,

42:43 see these behaviors.

42:45 Now,

42:46 what's really interesting and puzzling and a challenging problem is that um

42:51 we know that there are really high levels of overprescription of antibiotics.

42:54 We saw

42:55 the example of malaria,

42:56 but similar things are true for antibiotics.

42:58 And um and I think there is some hint of this in this data.

43:02 I mentioned there are 712 antibiotic spells.

43:06 And remember,

43:06 we only had about 730 formal instances of care in the first place.

43:12 So,

43:13 a lot of times,

43:14 children are prescribed an antibiotic.

43:16 So,

43:17 um,

43:19 Low adherence to the antibiotic

43:21 treatment recommendation is of course underuse in the sense

43:24 that parents don't follow the treatment instruction,

43:27 but many of the spells we see may actually constitute overuse in

43:31 the sense that the child's illness is not caused by bacteria.

43:34 So the policy problem here is simultaneously

43:37 make sure that

43:38 many of these children don't use an antibiotic at all because they don't need it,

43:42 but then for the other children that they actually adhere

43:45 to the full treatment recommended length.

43:50 So,

43:50 to conclude,

43:52 um,

43:54 We have seen that,

43:55 uh.

43:57 Uh,

43:57 in this,

43:57 in this decision chain,

43:59 there are multiple margins of misallocation possible and,

44:03 and documented.

44:05 And what we've also seen is that many of the policy instruments we have

44:09 are in some ways,

44:10 uh,

44:10 blunt instruments in the sense that,

44:12 uh,

44:13 they might address some of these margins of overuse and underuse but not others.

44:17 Just to give the example of subsidies,

44:19 um,

44:19 subsidies may help address underuse by

44:22 bringing more children to the doctor in our case,

44:25 underuse at the extensive margin.

44:27 But they might at the same time exacerbate overuse at the intensive margin,

44:30 meaning that when,

44:31 once the patient does show up at the doctor,

44:34 they actually receive

44:35 too much care.

44:38 Um,

44:39 what this suggests is that there is a need for additional research.

44:42 Um,

44:43 we would like to know in much

44:44 more detail how different incentives and policy interventions

44:48 change utilization at these different decision margins,

44:51 and then it is important to understand actually,

44:54 and this is something that,

44:55 that is in some ways missing here.

44:58 How do these changes,

44:59 both individually and in the aggregate,

45:02 actually map into what we care most about,

45:04 namely the welfare of the patients,

45:06 their health,

45:07 but also their income and uh happiness.

45:10 And what I would like to advocate for in this talk is that we

45:14 need better and more detailed measurements of

45:17 the allocation and quality of curative care.

45:21 Researchers and policymakers already have uh ways to measure these.

45:25 Um,

45:25 for example,

45:26 in economics research,

45:27 people have used audits,

45:28 that is direct observation,

45:30 um,

45:30 as well as standardized patients.

45:32 So these are trained actors who present with

45:34 specific conditions and then record what the doctor,

45:37 uh,

45:38 does in terms of diagnostics and prescriptions.

45:41 And actually,

45:41 the widespread use of these methods was pioneered by World Bank researchers.

45:45 Uh,

45:46 but in this talk,

45:46 I also showed you two potential new approaches.

45:49 One is to use health diary data of illness

45:51 and medication spells to understand behavior at home,

45:54 and the other is to conduct,

45:56 uh,

45:56 not only facility entry and exit interviews,

45:59 but combine these with home testing.

46:02 For malaria,

46:03 we can actually conduct home tests

46:05 that can help us assess the match of illness and treatments,

46:09 um,

46:09 that,

46:10 that actually occurred in this specific doctor consultation.

46:15 This is all I have.

46:16 Thank you very much.

46:18 Um,

46:18 I'm gonna

46:19 hand back over to the.

46:21 Thanks,

46:22 Anya,

46:22 for that um

46:24 really detailed and insightful set of,

46:26 set of results,

46:27 um,

46:28 which I,

46:28 you know,

46:28 I,

46:29 I hope moving over to Gaston now and maybe the discussion,

46:32 there's some real policy implications here that are kind of hard to grapple with.

46:36 So,

46:36 um,

46:37 Gaston,

46:37 uh,

46:38 over to you for,

46:39 for 1015 minutes and then we'll open up the questions.

46:42 Just a reminder,

46:43 if you want to ask a question,

46:44 either just raise your hand in the chat,

46:46 chat or signal that you have a question,

46:48 and I'll come to you later.

46:49 Uh,

46:49 Gaston,

46:50 over to you.

46:52 Thank you very much,

46:53 uh,

46:53 Dion.

46:53 Uh,

46:54 NJ,

46:54 uh,

46:55 you are the author of this study,

46:56 and I would like to thank you for inviting me,

46:59 Dion,

47:00 and the entire deck team

47:02 for asking me to join this,

47:03 this event.

47:05 I am a practice manager who

47:06 deals with policy dialogue and operational interventions

47:11 with our client countries and,

47:12 uh,

47:13 I'm working in close to 7,

47:15 17 countries in West Africa.

47:18 So,

47:18 I,

47:19 I looked at your,

47:20 your study with a very practical angle in answering the question,

47:24 how does this,

47:26 these studies

47:27 speak,

47:27 speak to me?

47:29 And Andre,

47:30 really,

47:31 congratulations for having nicely

47:33 sensitized 4 studies into

47:36 a concise and excellent presentation

47:39 that it's very difficult for anybody

47:42 to grasp the full content and to comment very meaningfully unless

47:47 you reduce yourself to a specific aspect that you can handle,

47:50 and that's exactly what I'm going to do.

47:53 You managed to,

47:54 to really address a complex problem.

47:57 Uh,

47:57 because measuring healthcare,

47:59 overuse and underuse depends on

48:02 the definition of adequate healthcare,

48:04 uh,

48:05 which integrates individual preference as well as needs.

48:08 So,

48:09 uh,

48:10 some scholars,

48:11 for example,

48:11 Vikas Susani defines operation care as.

48:15 Uh,

48:15 care that is needed,

48:16 wanted,

48:17 clinically effective,

48:18 affordable,

48:20 socially equitable,

48:21 and responsible,

48:22 responsible in its use of resources.

48:24 Of course,

48:24 uh,

48:25 this is kind of a challenging definition,

48:28 obviously,

48:28 but

48:29 it doesn't,

48:30 it doesn't prevent us from,

48:32 from really looking at the issue of healthcare overuse and underuse because

48:36 overuse and underuse has several implications.

48:38 Have you mentioned some of them and

48:40 I would like to go

48:42 at a little bit higher level because

48:45 you,

48:46 you,

48:46 you are really focusing on

48:49 demonstrating

48:49 what is happening

48:51 and what I would like to speak at is at what are the implications in terms of the,

48:56 the,

48:56 the,

48:56 the something,

48:57 someone like me

48:59 who is really dealing with operations and

49:01 how do I

49:03 look at those

49:04 findings

49:05 that you just

49:06 brought into the picture.

49:08 So first,

49:09 for,

49:09 for patient well-being and safety,

49:11 health systems should,

49:12 of course,

49:12 bring solutions,

49:13 but often we fail to address patients

49:15 suffering and even sometimes we create more problems,

49:18 like you just mentioned it,

49:20 from the data set.

49:21 This is really critical.

49:22 Second,

49:23 for equity reasons as well,

49:24 because every time one patient unnecessarily use,

49:27 uses a resource,

49:28 another patient is deprived from the same resources.

49:31 So,

49:32 very critical and very important.

49:33 And third,

49:35 Of course,

49:35 from the economy side,

49:37 because overuse and overuse imply

49:39 misuse of equipment

49:41 and human and financial resources,

49:43 particularly overuse deflects resources from public health

49:47 and other social spending,

49:48 and this

49:49 is aggravated by the lack of adherence to treatment.

49:51 So

49:53 in total,

49:53 you know,

49:54 we can see that these past years,

49:58 overuse and underuse have attracted the attention

50:00 of the century committee.

50:02 And other health public health professionals.

50:05 For example,

50:06 the,

50:07 The Lancet published the Right Care series in 1917.

50:12 Which highlight that overuse occurs across a range

50:15 of medical specialties.

50:18 It concerns procedures such as cardiovascular procedures,

50:21 hysterectomy,

50:22 etc.

50:23 and most studies on overuse have been conducted in high income countries.

50:29 And although it's increasingly recognized in low income countries,

50:32 it's not yet well documented in those settings.

50:35 So

50:36 your work

50:37 fills an important gap in the literature,

50:41 and it deals with key parts of the subject of inadequate care.

50:45 You talked about

50:47 financial

50:48 barriers to care,

50:50 drivers of overuse,

50:52 adherence to treatment,

50:52 etc.

50:54 So

50:55 you have marked my mind with key numbers that you brought into light,

51:00 and let me just refer to,

51:02 to,

51:02 to,

51:02 to 3 of them.

51:04 You concluded in your research that

51:06 with subsidies,

51:08 underuse and overuse increase at the same time,

51:11 at the same rate.

51:12 That was a surprise to me.

51:15 Other aspect you mentioned that 42%

51:17 of malaria negative patients receive treatment.

51:20 What a scandal,

51:21 should I say,

51:22 as a clinician.

51:23 And you also highlight the fact that 20% of malaria negative patients

51:28 and 53% of malaria positive case received treatment for severe malaria.

51:33 That's really a real problem.

51:35 And

51:36 more importantly,

51:38 You concluded that 57% of healthcare

51:41 workers

51:42 report pressure to prescribe unnecessary treatment

51:45 and

51:46 that some overuse,

51:48 mostly dry,

51:49 is driven by patient demand.

51:52 So,

51:54 I already said that your study

51:56 fills a gap,

51:57 but

51:58 Additionally,

51:59 it's timely and useful for evidence-based decisions,

52:03 as many countries now aim to achieve investor health coverage,

52:07 and we at the World Bank

52:09 are supporting the global

52:11 and individual country move towards investor health coverage.

52:14 Now let me go a little bit

52:16 deeper into,

52:17 into,

52:17 into the research.

52:19 Some part of your results

52:21 are

52:22 encouraging

52:23 since they show that once financial barriers are removed,

52:27 demand for care

52:29 when it's medically indicated increases significantly,

52:32 which is great.

52:34 But other parts of your results beg a fundamental

52:37 question again using my perspective as practice manager.

52:41 And the question is,

52:43 how can countries achieve the ambitious

52:45 agenda of universal health coverage while ensuring

52:48 care resources are adequately used

52:51 when

52:52 we also observed that 50% of health workers

52:55 in your study were pressured to prescribe unnecessary treatment.

53:00 We all agree that this is concerning,

53:04 and by that specific

53:06 finding.

53:07 It's great.

53:08 It raises our attention because

53:11 seriously,

53:12 That's a huge question for all of us,

53:15 and it's it's concerning because when you look

53:17 at the framework of universal health coverage,

53:19 of course,

53:20 it addresses three dimensions,

53:22 which are population coverage,

53:24 yes,

53:24 who is covered,

53:26 service coverage,

53:27 which services are covered,

53:28 and financial coverage,

53:29 what proportion of the cost is covered.

53:31 But

53:31 obviously

53:33 one dimension that is

53:35 maybe missing certainly is the issue of the quality,

53:37 and this is exactly

53:39 To what your,

53:39 your findings is pointing to.

53:43 Because,

53:43 uh,

53:43 and I believe seriously that

53:45 in our countries,

53:47 we'll certainly need to develop a strong quality control measures.

53:53 In particularly,

53:54 some specific measures will be needed looking at what you have found in your,

53:59 in your research.

54:00 Uh,

54:01 first,

54:03 we need certainly a strong communication

54:05 for behavior change and in short,

54:07 let me say education because

54:09 while the principle of universal health coverage

54:12 include the need to ensure that all citizens

54:15 have access to data and information on ESA.

54:19 As part of societal dialogue,

54:21 and dialogue and participation processes,

54:23 yet

54:24 your self study highlighted that a specific role of education in care decisions.

54:28 I think

54:29 we would need to integrate some sort of communication strategies into our approach

54:35 to make

54:36 the population realize and understand that

54:38 universal health care is not about getting care simply about

54:42 Getting care that is

54:43 needed,

54:44 that is useful,

54:46 and that is timely.

54:47 So this is

54:48 something we need to think deeply about because so far

54:52 when we help countries we do not necessarily

54:56 make this linkage with the issue of the that your side has highlighted.

55:00 Second thing I think we need also to consider is the issue of

55:04 the shared decision making as it goes end in end with the first one that I mentioned,

55:08 the first

55:09 measure that I mentioned.

55:10 Shared decision making is a process whereby clinicians collaboratively help

55:14 patients to reach evidence informed and value concrete medical decisions,

55:19 and several studies have shown

55:21 that

55:22 some patients know when care is not needed

55:25 and that there's a lack of shared decision making in this process.

55:28 And this is really critical and very important.

55:31 And again,

55:32 When we are moving at this pace whereby we

55:35 can ensure services to the maximum of population,

55:38 we need to make sure that decisions are made in a sound manner.

55:42 So this will call for us really to be more regarded and more

55:46 attention to some of the aspects because

55:49 many studies are found,

55:51 and this is corroborate what you just expressed in your study,

55:54 that for them,

55:55 in the US 17%,

55:57 meaning that 1 in 6 patients.

55:59 Reported that a healthcare provider recommended a medication

56:02 test or procedure that they felt they didn't need it,

56:06 and 50% of the adults

56:08 who had received a recommendation for a test they didn't think they need it

56:12 will end up getting it anyway.

56:14 You know,

56:15 so this is really even in developed countries this is happening.

56:19 So

56:19 what about our countries

56:22 where the situation is even worse?

56:23 So it really says that we need to be

56:27 more focused and pay more attention to those aspects that you have mentioned

56:31 in your study.

56:32 There are more studies on this,

56:33 for example,

56:34 in South Africa,

56:35 just to be close to Africa.

56:37 Well,

56:37 patients do not always request antibiotics and we're willing to discuss,

56:40 they're willing to discuss with the,

56:42 the doctor

56:43 how to manage the condition without them.

56:45 Doctors in private clinics

56:47 often go straight and prescribe

56:50 to medical,

56:51 to,

56:51 to,

56:51 to the patient,

56:51 and they will often go and get the medicines.

56:55 So,

56:55 those are really critical questions that you have highlighted through your,

56:58 your study.

57:00 So you also mentioned the use of the informed expert.

57:02 I wouldn't elaborate too much on it,

57:03 but it's a real question that we have to integrate as well

57:06 in all whatever we are going to do

57:09 when we move toward universal health coverage.

57:10 And then we need really to be,

57:12 to get some of the initiatives like such as choosing Wise,

57:16 the initiative that

57:17 promotes patient physician conversation about unnecessary medical tests

57:21 and procedures.

57:23 So the 3rd measure is the the the resource use monitoring system.

57:27 And I think that based on what you have shown and demonstrated,

57:31 definitely we need to be more careful and more looking at

57:35 what we are doing when we just subsidize subsidize services.

57:40 It's not enough to just pour resources

57:43 and even drugs into the healthcare facilities.

57:45 It will not.

57:47 resolve the problem

57:49 unless we have means to measure and to assess

57:52 how those is used and to give guidance,

57:55 guidance to the practitioners

57:56 and also to the clients

57:58 to get the best from whatever we are putting in the services.

58:01 So this is something very important and in many countries there are

58:04 already different tools that have been in utilization.

58:07 We may either adopt

58:08 or create new ones,

58:10 but definitely

58:11 we have something to do around this issue as well.

58:14 So let me

58:15 finish by a practical way forward suggestion.

58:18 So I know it's certainly done,

58:20 but I would like

58:21 to see DC teams reach out more directly to practitioners active on the ground.

58:27 And what I mean is,

58:28 for example,

58:29 I would like to invite you

58:31 to join my

58:32 next monthly unit meeting

58:35 with the staff

58:36 so that you can present the findings of your research to the staff,

58:40 gather the comments from the TTAs.

58:43 Who are working with the government to to increase access to services

58:47 and discuss the implications of the findings of your research

58:51 on the police and operational ground for World Bank Tail working in Africa.

58:55 So this is very critical because

58:57 this webinar is certainly interesting,

58:59 but we can go deeper

59:00 and more into detail

59:02 with the units and help them really

59:04 grasp this kind of information.

59:06 It's so critical

59:07 and it's It's not obvious that people will be exposed to this information.

59:10 It's so vital,

59:12 and I'm very happy that I was invited to comment.

59:14 I wouldn't even look at it,

59:16 even if I see it passing around.

59:18 But now that I have the chance to look at it,

59:21 to listen and to see,

59:23 I realize how important and critical it is

59:26 for our entire policy dialogue with the countries,

59:29 and really I'm amazed

59:30 to see this

59:32 this correct.

59:33 And its complete implication between research and operation on the ground.

59:38 So again,

59:39 having said that,

59:40 I would like to stop there

59:41 and welcome the Q&A question.

59:43 And again,

59:43 congratulations to you and your team

59:45 for the work that makes us think about how to effectively use results

59:50 from field studies

59:51 to meaningfully increase our chances to achieve coverage in our client countries.

59:55 Thank you very much.

59:56 Back to you,

59:57 day one.

59:59 Thanks,

1:00:00 Gaston,

1:00:00 and,

1:00:00 and,

1:00:01 and

1:00:01 based on previous conversations with Anya,

1:00:03 I know she'll jump at the opportunity to,

1:00:05 to join you at your upcoming uh staff meeting and,

1:00:08 and I,

1:00:09 I know she'll be happy to,

1:00:10 to have the kinds of discussions that you anticipate.

1:00:12 Um,

1:00:13 and,

1:00:13 and thanks for saying that because,

1:00:14 uh,

1:00:14 you know,

1:00:15 this really is the one of the purposes of these talks is

1:00:17 to engage people who might not have paid attention to some of these

1:00:21 results,

1:00:21 um,

1:00:22 and it gives us an opportunity to,

1:00:24 to highlight them and,

1:00:25 and,

1:00:25 and,

1:00:25 and really engage with,

1:00:26 with the likes of you and,

1:00:27 and,

1:00:28 and others.

1:00:29 Um,

1:00:30 Anya,

1:00:31 before turning it over,

1:00:31 do,

1:00:32 do you have any quick reactions to any of the points that Gaston made or,

1:00:35 or should we go straight to Q and A?

1:00:38 Uh,

1:00:38 yeah,

1:00:38 let me just,

1:00:39 uh,

1:00:39 say a couple of things.

1:00:40 So thanks,

1:00:41 Gaston,

1:00:41 for your comments and I really appreciate,

1:00:43 um,

1:00:44 uh,

1:00:44 how passionate you are about this and,

1:00:46 and agreed with Leon.

1:00:47 I would love to,

1:00:48 to,

1:00:49 to join you,

1:00:50 um,

1:00:50 in,

1:00:50 in discussing this more and with a smaller team.

1:00:54 But um

1:00:56 I think you make two really great points that I want to just quickly comment on.

1:00:59 So,

1:01:00 first of all,

1:01:01 you said that,

1:01:01 um,

1:01:02 and this is something that I think is,

1:01:04 is true in,

1:01:05 in,

1:01:05 in our setting as well,

1:01:06 but you said that there's evidence at least from the US and Europe

1:01:10 that patients often suspect that their doctors overprescribe.

1:01:14 And I think what this speaks to is exactly this issue of trust

1:01:17 that I mentioned before and,

1:01:18 and you.

1:01:19 You have said that you think that it's really important

1:01:22 to uh come to a model of,

1:01:24 of joint communication and establishing trust between patient

1:01:27 and provider to improve the allocation of care.

1:01:30 And then you've mentioned a second,

1:01:32 um,

1:01:32 a second need and that is monitoring resource use.

1:01:36 And I'd like to connect those two a little bit because I think what we find in our data,

1:01:41 at least,

1:01:41 and I should say,

1:01:42 I mean,

1:01:42 these are small studies and they are from one country,

1:01:44 from one,

1:01:45 city,

1:01:45 so I think there's the caveat of to what degree we can generalize.

1:01:50 But what we find in general is that these

1:01:51 providers are clearly interested in their patients' welfare.

1:01:55 They are not as guided by financial

1:01:56 incentives as maybe some health economic studies,

1:01:59 uh,

1:01:59 suggest,

1:02:00 but they do give in patient pressure.

1:02:03 And I think one question is how can we use this to improve care?

1:02:08 And I think one thing that has struck me

1:02:10 is that um in order to establish better communication,

1:02:14 greater trust,

1:02:15 one other scarce resource is rarely mentioned,

1:02:18 and this is time.

1:02:19 Um,

1:02:20 I think these,

1:02:21 these issues do require time.

1:02:23 We see that patients are,

1:02:25 are constrained in that,

1:02:26 in that way,

1:02:27 because we see that they don't even go to the clinic when the care itself is free.

1:02:31 This speaks to some time constraints.

1:02:33 But we also know that when a doctor really tries to communicate

1:02:37 their,

1:02:38 their insights,

1:02:38 their advice to a patient,

1:02:40 this actually takes a lot of time to teach the patient

1:02:42 what they need to do when they go home and so on.

1:02:45 So I agree with your,

1:02:47 with your,

1:02:48 with your thought that we should have better

1:02:51 data collection systems to monitor resource use.

1:02:54 But I think um

1:02:56 this also provides an opportunity perhaps to introduce systems

1:02:59 that cannot just

1:03:01 Monitor,

1:03:01 but can also facilitate communications and improve processes and save

1:03:06 both patients and providers valuable time that they can then use

1:03:10 to communicate better about the things that really matter,

1:03:13 such as how to use antibiotics when you actually come home with your kid.

1:03:16 And I think there is a huge opportunity here.

1:03:19 So I remember from my,

1:03:20 my research collection in Mali,

1:03:22 from my data collection in Mali,

1:03:23 um,

1:03:24 in the first few years,

1:03:25 we actually visited a lot of clinics and we saw

1:03:27 how doctors keep track of their visits.

1:03:29 They have these beautiful ledgers,

1:03:31 handwritten ledgers filled with fountain pen.

1:03:34 And all this data in them is basically not

1:03:37 used very well for any kind of decision making,

1:03:39 and it seems really important

1:03:41 when we introduce systems to monitor better,

1:03:44 to also use that information for the

1:03:46 doctors and for the patients' benefits to actually

1:03:49 improve the care that they receive in each visit.

1:03:53 Um,

1:03:56 Really,

1:03:56 yeah,

1:03:56 I'm gonna stop here.

1:03:57 This is,

1:03:57 this is,

1:03:58 these were my main comments that I was gonna make.

1:04:00 Um.

1:04:04 Thanks,

1:04:05 Anya.

1:04:05 Gaston,

1:04:06 I'm gonna give you a chance to come in a bit later again,

1:04:08 sort of maybe after some discussion,

1:04:09 unless you really want to come in immediately right now.

1:04:13 Um,

1:04:14 or,

1:04:14 or should we just take a few questions and then

1:04:16 come back?

1:04:16 No,

1:04:16 let us move for the,

1:04:17 for the questions,

1:04:18 and I'm happy to hear that.

1:04:19 And

1:04:20 just to mention that I have a chance to live in Mali for Mali for 2 years

1:04:24 at the World Bank office,

1:04:25 and I was there,

1:04:25 the,

1:04:27 like,

1:04:27 how do I call it,

1:04:28 the detail for the operation there.

1:04:29 So

1:04:29 I have a little sense of what's going there,

1:04:31 but it's cuttered some time ago.

1:04:33 Back to you.

1:04:33 Thanks.

1:04:35 OK,

1:04:35 great.

1:04:35 So I'm gonna,

1:04:36 so there's a couple of questions.

1:04:37 I'll,

1:04:37 I'll just mention the names and maybe we'll take,

1:04:39 uh,

1:04:40 we'll take 3 questions,

1:04:41 um,

1:04:41 with the,

1:04:42 the 3 people have signaled,

1:04:43 um,

1:04:43 I'll go in turn and then maybe Anya you can respond,

1:04:46 uh,

1:04:46 uh,

1:04:46 in one go.

1:04:47 I'm gonna take the chair's privilege and

1:04:49 actually interject 11 little question beforehand,

1:04:52 which is,

1:04:53 you,

1:04:54 on one of your slides,

1:04:54 you,

1:04:55 you,

1:04:55 you,

1:04:55 you pointed to the high use of informal care.

1:04:59 And you said it to me,

1:05:00 we're not gonna talk about that today,

1:05:01 but I did,

1:05:02 it was very,

1:05:03 I,

1:05:03 I,

1:05:03 I'm very curious about what's the content of that.

1:05:05 Is that pharmacies,

1:05:06 is that other care,

1:05:07 and how does that interact with this kind of cares seeking behavior that we did see?

1:05:11 And I realized that's maybe a discussion for another day,

1:05:13 but,

1:05:14 but just a little bit of insight on that,

1:05:15 I think would be useful.

1:05:17 Um,

1:05:18 so in terms of the questions,

1:05:20 Ellen,

1:05:20 are you still there?

1:05:21 You had two interventions,

1:05:22 so maybe if you could sort of.

1:05:24 Give each of them,

1:05:25 each of them briefly,

1:05:27 uh,

1:05:27 do you want to turn on your mic,

1:05:28 Alan,

1:05:29 and maybe your camera and ask your questions?

1:05:32 Uh,

1:05:34 sure,

1:05:34 Dion,

1:05:34 can you hear me?

1:05:35 Yes,

1:05:36 we can hear you well.

1:05:36 OK.

1:05:37 Uh,

1:05:37 OK,

1:05:37 yeah,

1:05:38 I feel I'm just gonna repeat now what I already thought,

1:05:40 but

1:05:41 thanks for the interesting presentation.

1:05:42 I was just also

1:05:44 like in line with,

1:05:44 I think,

1:05:45 what Gaston was saying,

1:05:46 trying to translate your findings to

1:05:48 the operational work,

1:05:49 um,

1:05:49 that we are sort of engaged in on a,

1:05:51 on a,

1:05:52 on a daily basis.

1:05:53 Um,

1:05:54 I work in the global financing facilities,

1:05:56 for not a specific country,

1:05:57 but involved in many operations,

1:05:58 especially in West Africa.

1:06:00 And I don't know whether I understood sort of your conclusion on the first part well,

1:06:04 but to me it sounded like indeed

1:06:06 demand size subsidies make people go more to facilities,

1:06:09 but that's not a guarantee that actually care gets better allocated,

1:06:13 which resonated,

1:06:14 but at the same time I was thinking,

1:06:16 um,

1:06:16 like putting myself in the shoes of a TTLO,

1:06:18 so what,

1:06:19 like what,

1:06:20 what is it?

1:06:20 Tell me,

1:06:20 if we seem to be struggling with only supply sides,

1:06:23 um,

1:06:24 strengthening through results-based financing.

1:06:26 Demand side subsidies struggle from this shortcuting like what,

1:06:29 what would be the,

1:06:30 the sort of guidance or what would the questions be to ask when we're developing

1:06:34 a project under a limited timeline and limited resources instead of how,

1:06:38 how do we,

1:06:38 how do you see we could mitigate this?

1:06:41 Um,

1:06:42 what to watch out for or build into these projects to

1:06:45 at least try to,

1:06:46 um,

1:06:47 alleviate some of these concerns.

1:06:49 And then I guess a second um related comment to the second part

1:06:53 of the presentation where I fully appreciate the point you're trying to make that

1:06:57 even in low-income settings it's not only about

1:07:00 low utilization rates and underuse but also inefficiencies

1:07:03 through overuse,

1:07:04 I think.

1:07:05 Probably

1:07:06 most of my colleagues would agree with that,

1:07:08 but then again,

1:07:09 sort of operationally,

1:07:10 what do we,

1:07:11 what do we do with this if you work in a country

1:07:13 like Mali where overall utilization is still so low for very basic,

1:07:17 uh,

1:07:18 services

1:07:19 and we have limited resources and efforts,

1:07:21 where are we gonna put most of our

1:07:24 efforts on?

1:07:24 I,

1:07:24 I,

1:07:25 I unders I sort of,

1:07:26 I was thinking,

1:07:27 uh,

1:07:27 given everything you described on working through.

1:07:30 Community health workers and that it's really quite a difficult agenda to address,

1:07:33 for example,

1:07:34 overuse or misuse of antibiotics.

1:07:36 It would almost require a project in itself,

1:07:38 right,

1:07:38 because it's

1:07:39 financial,

1:07:40 behavioral,

1:07:40 it's a complex thing to change.

1:07:41 I even

1:07:42 see that in high income countries it's really a problem.

1:07:44 So

1:07:45 again,

1:07:45 sort of them thinking,

1:07:46 do we,

1:07:47 do we,

1:07:47 should we,

1:07:48 do we want to invest in this in countries like Mali?

1:07:50 Is this where we get the biggest bang for the buck,

1:07:52 given the sort of limited resources?

1:07:55 Um,

1:07:55 or should we sort of

1:07:56 work on this agenda depending on where countries are

1:08:00 and the extent of the,

1:08:01 of the problem.

1:08:02 Just again thinking that the low health outcomes

1:08:04 you see in a country like Mali are very

1:08:06 much driven by the very low use of care and perhaps not so much by the,

1:08:10 by the misuse of care,

1:08:11 um,

1:08:11 but you might disagree.

1:08:13 I was just wondering,

1:08:14 uh,

1:08:14 what your

1:08:15 thoughts were on,

1:08:15 on these sort of more operational questions.

1:08:17 So thank you.

1:08:18 Over.

1:08:20 Thanks,

1:08:20 Ellen.

1:08:21 Um,

1:08:21 Khuram,

1:08:22 are you still there?

1:08:23 You had a very,

1:08:24 one very specific question.

1:08:25 Do you want to ask that?

1:08:32 Um,

1:08:33 maybe Kuram,

1:08:34 uh,

1:08:34 no,

1:08:34 Kuram,

1:08:34 I see he's still on,

1:08:36 um.

1:08:38 Um,

1:08:39 sorry,

1:08:39 uh.

1:08:42 My another question about uh the overuse of antibiotics and uh that is

1:08:46 uh in a country where there is a high level of pollution.

1:08:50 Don't you think that the doctors uh

1:08:53 Prescribe for overuse of antibiotics to

1:08:56 to protect the patients and uh

1:08:59 uh and uh to

1:09:01 uh lessen the effect of illness.

1:09:06 So might they be doing something

1:09:08 good anyway

1:09:09 by overprescribing

1:09:10 because of other conditions?

1:09:11 OK.

1:09:12 And then Adna,

1:09:13 are you there?

1:09:14 Would you like to ask a question?

1:09:21 Uh,

1:09:21 I think I don't know may have left.

1:09:25 Um,

1:09:28 OK,

1:09:29 so,

1:09:29 I know,

1:09:30 do you want to take those,

1:09:31 those two?

1:09:36 Sure,

1:09:36 thank you.

1:09:36 Um,

1:09:37 thanks so much for those questions.

1:09:38 Um,

1:09:40 So,

1:09:40 Dion,

1:09:41 to start with your question on,

1:09:43 on informal care.

1:09:44 So you're right,

1:09:45 we are seeing a ton of these informal care

1:09:48 users.

1:09:49 So we collected,

1:09:50 um,

1:09:51 we collected absolutely everything.

1:09:52 These are peddlers who come to the home with,

1:09:54 um,

1:09:55 with little

1:09:56 packets of,

1:09:57 of treatment options.

1:09:59 These are market stall visits and so on.

1:10:01 Um.

1:10:02 What you could also see is that a lot of these,

1:10:05 these,

1:10:06 these informal options had incredibly small costs,

1:10:08 so 200 safer,

1:10:09 that's

1:10:11 Divide by 500,

1:10:12 that's a really small amount in US dollars.

1:10:15 Um,

1:10:15 so,

1:10:16 what most of these visits actually involve or what,

1:10:18 what these,

1:10:19 what these purchases involve are

1:10:21 really um remedies for symptoms such as Tylenol or ibuprofen.

1:10:26 They are not often actual treatment,

1:10:27 for example,

1:10:28 like anti-malarials and antibiotics.

1:10:30 Moreover,

1:10:31 they are often,

1:10:32 um,

1:10:33 if they are labeled an antibiotic,

1:10:35 we are not sure what the quality of that care actually is.

1:10:37 So we would certainly

1:10:39 And when we encourage care use by subsidizing,

1:10:42 we really want families first to see a qualified provider

1:10:45 to diagnose before we want them to buy anything.

1:10:48 But you're certainly right to ask about the

1:10:49 interactions of the existence of these informal options

1:10:52 and,

1:10:53 and the use of formal care.

1:10:54 And the,

1:10:55 the short answer,

1:10:56 at least for our data is that we are just not powered to say much about that,

1:10:59 but I think it's a really important question for future research to think about.

1:11:04 Now,

1:11:05 uh,

1:11:05 Ellen,

1:11:06 thank you.

1:11:06 You asked a lot of really good and interesting questions,

1:11:09 important questions.

1:11:10 Um.

1:11:12 Your first question was,

1:11:14 well,

1:11:14 now that we know that patients,

1:11:16 um,

1:11:17 go more to the,

1:11:18 go to the doctor,

1:11:19 go to the provider more when we give subsidies,

1:11:22 and we've also learned that some of these visits might actually be overused,

1:11:25 well,

1:11:25 what should we do?

1:11:27 Um,

1:11:27 I think,

1:11:28 well,

1:11:28 the first answer is that I,

1:11:31 that we had hoped when we studied the Action for Health program

1:11:34 that one intervention that could actually

1:11:36 improve this allocation would be precisely

1:11:38 the visits from these community health workers.

1:11:41 One of their,

1:11:41 this is not the only thing they do for sure,

1:11:43 they also do some prevention and so on,

1:11:45 but

1:11:45 Uh,

1:11:46 one of the things that they do is they actually go through the symptom rule charts

1:11:50 and,

1:11:50 um,

1:11:51 and try to assess if a child currently needs care.

1:11:54 And on top of this,

1:11:55 they are also supposed to teach parents how to take better care of their children,

1:11:58 for example,

1:11:59 how to respond to diarrhea and when the child should see a doctor versus not.

1:12:04 Um,

1:12:05 unfortunately,

1:12:05 this is not what we find.

1:12:06 We did not find that in this particular program,

1:12:09 the community health workers improved the allocation.

1:12:12 This is certainly not to say that this must be always the case.

1:12:15 There could be

1:12:16 measures to be taken to improve

1:12:18 how community health workers steer the use of formal care,

1:12:22 but I think it is fair to say that this is an open

1:12:24 question how we could improve the use of care at the home.

1:12:28 Um,

1:12:29 I also think it is really,

1:12:31 you make one important point

1:12:33 that it could be that this utilization problem,

1:12:36 this misallocation

1:12:38 is the reason why we don't see as many health effects.

1:12:41 But I think it's also fair to say that we don't know that either yet,

1:12:44 for sure.

1:12:44 So the IMCI has been tested many times in public health context,

1:12:48 and it has has been found to be effective.

1:12:50 But the IMCI set of interventions is actually a whole set of things that happen.

1:12:55 Uh,

1:12:55 not only these community health worker visits,

1:12:57 but other things as well.

1:12:58 And so,

1:12:59 pinpointing how the exact time when you go see a doctor affects

1:13:04 later health outcomes is going to be another open question to,

1:13:07 to answer with more research.

1:13:10 Um,

1:13:10 so this is not as conclusive an answer,

1:13:12 I'm afraid,

1:13:13 as,

1:13:14 as,

1:13:14 um,

1:13:14 as

1:13:15 you might have hoped for.

1:13:17 But I think it's fair to say that there's many things we just don't know yet.

1:13:21 Um,

1:13:22 you asked the second question,

1:13:23 which I think is a really good question.

1:13:25 And that is one that I can probably and should probably not answer.

1:13:29 And that is the question,

1:13:30 how should we prioritize between the two things?

1:13:32 You're making the very good point that because underuse is such a big problem,

1:13:36 maybe

1:13:37 this should be taking priority over

1:13:40 any problems of Overuse.

1:13:42 And I think there's many,

1:13:43 many areas and settings where this is definitely true.

1:13:45 So

1:13:46 Mali itself is just an example.

1:13:48 So the study that we conducted is in peri-urban and urban areas

1:13:52 where clinics are actually relatively frequent.

1:13:55 Most clinics have uh a provider on staff,

1:13:58 a qualified medical trained doctor,

1:14:01 um.

1:14:02 They are relatively,

1:14:03 they have good opening times and the providers are typically actually present.

1:14:07 I think there is a huge,

1:14:09 a huge,

1:14:10 a hugely different problem of access in rural areas,

1:14:13 for example,

1:14:14 in Malian rural areas are certainly a case in point where the clinics are really far

1:14:19 and even when you go to the clinic,

1:14:20 it might be closed or the provider might not be on site and so on.

1:14:24 So

1:14:24 these underused problems certainly

1:14:27 I'm almost surely.

1:14:28 Thinking that we should take priority.

1:14:31 But once we go to urban areas and settings where overuse is more of a potential issue,

1:14:37 I think we should also think about overuse.

1:14:39 And the reason is simply that resources are so incredibly scarce.

1:14:43 Um,

1:14:43 you saw the data from,

1:14:45 uh,

1:14:45 some of these clinics in,

1:14:47 in Bamako where the dispensary,

1:14:49 the,

1:14:49 the,

1:14:50 the national.

1:14:50 Dispensary is actually closed and they can,

1:14:52 they have

1:14:53 relatively good access.

1:14:55 And yet many of these clinics run out of treatment regularly and have stock outs,

1:15:00 which means that because some people receive treatment that they didn't need,

1:15:04 there will be days where just no pre-treatment is available,

1:15:06 and some folks come in

1:15:08 and do not get the malaria treatment that they

1:15:10 would have needed in order to cure the infection.

1:15:13 So,

1:15:15 I am

1:15:16 probably not in a good place to give definitive recommendations

1:15:19 of how limited resources and policy making should be allocated.

1:15:23 But I do think that there are potentially mutually reinforcement issues here,

1:15:28 and I mentioned one pathway in which this could happen

1:15:31 if the quality of care at the clinic is,

1:15:33 is low and overuse

1:15:35 or overprescription by providers is really high.

1:15:37 This might actually discourage people from coming to the clinic,

1:15:40 and it might also lead to greater distrust between providers and patients

1:15:45 and overall substantial misuse of resources.

1:15:50 So,

1:15:50 I think of these as joint problems in many ways,

1:15:53 and I would,

1:15:54 I would,

1:15:54 well,

1:15:56 I would advocate for thinking about overuse as well,

1:15:58 but I totally take your point

1:16:00 that we need to potentially make

1:16:02 set priorities given how scarce all our resources are.

1:16:07 For the last question,

1:16:09 Koram,

1:16:09 um,

1:16:11 You asked about potential overuse as a preventive measure,

1:16:14 and I think you make a really

1:16:16 good point.

1:16:18 In fact,

1:16:18 uh,

1:16:19 for antibiotics,

1:16:19 I'm,

1:16:20 I'm less familiar with,

1:16:21 with

1:16:22 exact medical treatment recommendations,

1:16:24 but for malaria,

1:16:25 it's certainly the case that treatment recommendations

1:16:27 have changed over the last few years.

1:16:29 So,

1:16:30 um,

1:16:31 so,

1:16:31 originally in malaria endemic countries,

1:16:34 the recommendation was actually,

1:16:35 especially when testing is not available,

1:16:38 to simply treat any fever as if it was a malaria case.

1:16:41 And the reason is simply that the downside risk of not

1:16:44 treating a malaria case is so much higher than the,

1:16:46 the,

1:16:47 the slight risk of overtreating.

1:16:49 However,

1:16:49 in recent,

1:16:50 in recent years,

1:16:51 we have become much more attuned to the problem of creating resistances.

1:16:56 We are already on the 2nd generation of anti-malarials

1:16:59 and 1st generation anti-malarials have often become ineffective,

1:17:02 so

1:17:03 parasite resistance is a real concern,

1:17:06 and,

1:17:06 uh,

1:17:07 malaria treatment recommendations have been updated

1:17:09 to really only provide treatment when.

1:17:12 There has been a test.

1:17:13 For antibiotics,

1:17:14 this is a much harder problem because it is harder to test for bacterial infections.

1:17:19 But the,

1:17:19 the,

1:17:20 the rule of thumb is that uh giving an antibiotic does not have preventive effects.

1:17:26 Um,

1:17:26 you really need to have a bacterial infection for

1:17:29 an antibiotic to do something for your health.

1:17:31 And in fact,

1:17:32 we also know that antibiotics generally actually harm.

1:17:35 The,

1:17:35 the natural bacterial flora of the of the body.

1:17:38 So it is not obvious that giving antibiotics preemptively is,

1:17:42 uh,

1:17:43 is particularly helpful for overall health outcomes.

1:17:46 Having said that,

1:17:46 I,

1:17:47 I,

1:17:47 I think in,

1:17:47 in other contexts besides antibiotics,

1:17:49 your point is certainly well taken that when you have little access to diagnostics,

1:17:53 overtreatment might be actually the right course of action,

1:17:56 and we've seen that in the past.

1:17:58 At least in our setting,

1:17:59 this is not the case because rapid detection tests are widely available.

1:18:03 They should be free for patients,

1:18:05 and they are actually Mali's and the WHO's

1:18:08 official malaria policy to use before treating.

1:18:14 This is,

1:18:14 uh,

1:18:15 I think I answered all the questions,

1:18:17 but do let me know if you,

1:18:18 if you have follow-up questions.

1:18:20 You did,

1:18:20 and in,

1:18:20 in fact,

1:18:21 and in passing,

1:18:22 you also

1:18:23 answered Ana's question,

1:18:25 which was about community health workers and how,

1:18:27 you know,

1:18:28 how they could be leveraged or some alternatives to them

1:18:30 could be leveraged to right some of these wrongs,

1:18:32 but you,

1:18:32 you answered that in,

1:18:34 in,

1:18:34 in part in response to Alan.

1:18:36 Um.

1:18:38 So I don't see anybody else um raising their hand or um asking a question,

1:18:43 so I just want to give you a chance,

1:18:45 um.

1:18:47 If not,

1:18:47 Gaston,

1:18:48 I might come back to you just for a closing comment if you had one.

1:18:52 Um,

1:18:53 before we sign off,

1:18:54 um,

1:18:55 so there's nobody asking to ask a question.

1:18:58 So Gaston,

1:18:59 would you like to,

1:19:00 and then Anya,

1:19:00 I'll give you the very,

1:19:01 very final word,

1:19:02 but Gaston,

1:19:02 did you have any final comments and reflections?

1:19:05 No,

1:19:05 OK,

1:19:06 thank you very much for,

1:19:07 for giving me the opportunity to,

1:19:08 to join this,

1:19:09 uh,

1:19:09 this meeting.

1:19:09 As I said,

1:19:10 uh,

1:19:11 I'm amazed to see how relevant,

1:19:13 uh,

1:19:13 those,

1:19:13 those discussions are really for,

1:19:15 for us on the ground.

1:19:17 And to go a little bit further,

1:19:18 I would even start asking myself,

1:19:21 how can we really get a better linkage between deck and,

1:19:25 and operations because

1:19:27 Uh,

1:19:27 yes,

1:19:28 it's good to,

1:19:29 to,

1:19:29 to,

1:19:29 to,

1:19:30 to be connected at the end of the,

1:19:32 the research to get the results,

1:19:33 our findings

1:19:34 and to discuss them,

1:19:35 but

1:19:36 to what extent can we

1:19:38 maybe have more,

1:19:40 I wouldn't say involvement because it may be too much,

1:19:42 but

1:19:43 at least,

1:19:43 certainly there was some kind of communication or things

1:19:47 with the teams in,

1:19:48 in,

1:19:48 in the country,

1:19:49 for example,

1:19:49 in Mali.

1:19:50 But I would love to see the money team join this kind of discussion.

1:19:53 So,

1:19:54 uh,

1:19:55 uh,

1:19:55 it doesn't look like there was many of them around.

1:19:57 Well,

1:19:58 it's very critical and very important for,

1:20:00 for,

1:20:00 for,

1:20:00 for them.

1:20:00 So,

1:20:01 how can we try to,

1:20:02 to improve this kind of,

1:20:03 uh,

1:20:04 communication so that we have

1:20:05 the team,

1:20:06 particularly those who are sitting in those countries where the studies

1:20:09 are,

1:20:09 are done so that they can be part of,

1:20:11 of the process.

1:20:12 So,

1:20:12 just a comment that,

1:20:13 uh,

1:20:14 I would like to make.

1:20:15 Again,

1:20:16 thank you very much for,

1:20:17 uh,

1:20:17 for this opportunity.

1:20:19 I,

1:20:19 I do,

1:20:20 I do appreciate your,

1:20:21 your conclusion when you,

1:20:22 you mentioned the need for,

1:20:24 for more research and the issue of measurements.

1:20:27 And

1:20:29 again,

1:20:29 we can,

1:20:30 even if it's,

1:20:31 if it's a,

1:20:32 a,

1:20:32 a small scale studies,

1:20:34 we can already talk about

1:20:37 policy implication because honestly,

1:20:39 I do believe

1:20:40 that the numbers,

1:20:41 uh,

1:20:42 speaks quite loudly for us and for teams

1:20:45 on the ground.

1:20:46 Because

1:20:47 those kinds of references are very useful for our papers,

1:20:50 for our documents,

1:20:51 for our policy discussions,

1:20:52 etc.

1:20:53 So yes,

1:20:54 this is certainly part of the question that we can start

1:20:57 addressing even with a small study.

1:21:00 So what can be the policy implication.

1:21:03 With this kind of information that we have gathered,

1:21:06 so this is something I think we may also add it

1:21:08 and try to discuss it as well.

1:21:10 So that's what I would like to add again.

1:21:12 Thank you for inviting

1:21:13 me.

1:21:13 It's a real pleasure

1:21:15 and I really appreciate it,

1:21:16 and we'll get together

1:21:17 for the presentation to my unit.

1:21:19 Thank you very much.

1:21:21 Back to you.

1:21:22 Thanks,

1:21:22 Gaston.

1:21:23 Anya,

1:21:23 any last final words?

1:21:26 Um,

1:21:27 yeah,

1:21:27 just to say thank you so much,

1:21:29 Gaston.

1:21:29 I really appreciate it.

1:21:30 Um,

1:21:31 and I agree with you there is,

1:21:32 um,

1:21:33 there's both

1:21:34 a need to communicate these results further and also to,

1:21:38 um,

1:21:38 to,

1:21:39 to think about

1:21:40 working together to,

1:21:41 to design additional studies and,

1:21:43 and additional research that really serves operational purpose.

1:21:46 Uh,

1:21:47 I do wanna say that I conducted all of these

1:21:49 studies before I actually came to the World Bank.

1:21:51 Um,

1:21:51 so all the data collection was done before I was,

1:21:54 uh,

1:21:54 uh,

1:21:54 a member of DCRG.

1:21:56 I was a professor at Brown University before

1:21:59 and that's where all this research programs started,

1:22:01 but I'm really excited to

1:22:03 get more involved in,

1:22:04 in,

1:22:05 in the work that's actually done here,

1:22:07 um,

1:22:08 and,

1:22:08 and have the chance to talk more about these,

1:22:10 these open policy questions and,

1:22:12 um,

1:22:13 and hopefully,

1:22:13 hopefully find solutions as well and not just point out

1:22:16 the many,

1:22:17 many issues we've seen.

1:22:19 Thank you.

1:22:22 Thanks,

1:22:22 Anya.

1:22:23 No,

1:22:23 that's exactly,

1:22:23 I think that's,

1:22:24 that's exactly what this kind of session is supposed to try to

1:22:27 do is to create these linkages that then we can follow up,

1:22:29 um,

1:22:30 follow up with you can follow up with others too.

1:22:33 So with that,

1:22:33 um,

1:22:34 I'm gonna close the session.

1:22:35 Thank you,

1:22:35 Anya,

1:22:35 for a fantastic presentation,

1:22:37 really insightful data and analysis,

1:22:40 and,

1:22:40 and,

1:22:40 and a lot of questions and some,

1:22:43 some

1:22:44 solutions as you,

1:22:45 as you,

1:22:45 as you drew out in the,

1:22:46 in the Q&A,

1:22:47 uh,

1:22:48 but also a lot of food for thought.

1:22:49 And then,

1:22:50 um,

1:22:51 thank you,

1:22:51 Gaston as well for,

1:22:52 for,

1:22:53 for really interesting,

1:22:54 um,

1:22:55 reflections and reactions and great to you,

1:22:58 you know,

1:22:59 Take it,

1:22:59 we'll take you up on your invitation to come and join and,

1:23:02 and work proactively with,

1:23:03 with your teams.

1:23:04 And then thank you all of you for joining and for the really interesting questions.

1:23:08 Uh,

1:23:08 Ryan has just put in the,

1:23:10 in the chat,

1:23:10 um,

1:23:12 a link with all the,

1:23:13 um,

1:23:13 the materials,

1:23:14 the presentation,

1:23:16 the video will be posted there afterwards.

1:23:18 Uh,

1:23:18 you can share with others who weren't able to join today.

1:23:22 So once again,

1:23:22 thank you,

1:23:23 everybody,

1:23:23 and,

1:23:24 uh,

1:23:24 congratulations and have a good rest of your day.

1:23:28 Bye.

1:23:28 Thank you very much.

1:23:29 Bye-bye.

1:23:30 Congratulations again.

1:23:31 Thanks.

1:23:32 Thank you very much.

1:23:33 It was a great pleasure pleasure.

1:23:35 Thank you.

1:23:35 Bye-bye.

1:23:36 Bye-bye.

showAllTimestamps

no

transcript

So, hello everybody. My name is Dion Filmer. I'm um the director of the World Bank's Development Research Group. Um, welcome to this policy research talk. Uh, these talks, as some of you may know, give us an opportunity to present work coming out of the department, the research department. With the goal of sharing the findings with each other, with colleagues inside the bank, as well as uh others outside of the World Bank. Um, just welcome our online audiences, both on Webex as well as on YouTube. Um, today my colleague Anya Salman. Who is a research economist who actually joined us relatively recently, but, uh, she's a research economist in the human development team of the department. Uh, she's going to be presenting insights from recent research on healthcare demand and utilization in Mali. Her talk will focus on the allocation of acute curative healthcare and, uh, to what extent demand side policies can affect this allocation. Um, I'm very grateful to have, uh, Gaston Sorgo here today to discussant. Uh, Gaston is a practice manager in the Health, Nutrition and Population Global Practice of the Human Development Practice Group here at the bank. He joined the World Bank in 1999 and has previously served as a country manager for Mauritania and lead public health specialist and human development sector leader based in Cameroon. Gaston is a physician with a Master's of Public Health from the University of Amsterdam, and a doctorate in Public Health from Harvard University. So the run of show today is gonna be, um, Anya will talk for about 45 minutes, after which we'll hear from Gaston for 10 to 15 minutes, after which we'll, uh, have questions and answers from the audience. Um, if you have a question, please use the raised hand function in, in WebEx or just signal to me in the chat that you have a question and I will call on you. If you're following on YouTube, uh, please put your uh question in the chat over there and that will get relayed to me. So with that, uh, over to you, Anya. Thank you so much, Dionne. Give me one second, I'll share my screen. Is this, uh, is, can you all see my screen? Great. Um, so, yeah, thank you, Dion. Thank you very much for hosting, uh, this talk. Thank you, Gaston, for being here as a discussant and thank you also to the audience for, for joining us. Um, as Dion said, this talk is, uh, about overuse and underuse of acute healthcare with evidence from, with empirical evidence from Mali. And before I, uh, go into this evidence, I would like to motivate a little bit why we should care about overuse and underuse and specifically in acute care. As I'm sure many in the audience know, um, global health has seen great advances in preventive care in the last two decades. Um, And there is something of a policy consensus today that in order to make further significant improvements, we really need to focus attention on providing better care for acute illness in addition to preventive care, acute illness and curative care. Now, unlike in preventive care, however, in curative care, um, whether and when any given individual needs treatment varies, and it varies based on the person's health status. Uh, this health status is often unobserved to the outside observer, um, but it might even be unknown to some degree to the patient themselves. Many healthcare decisions are made in the home and, uh, uh, and oftentimes, um, when you present with some diffuse symptoms, you might not know what your, what your illness exactly is. In this context, misallocation, that is the underuse and overuse of care is a core challenge for efficiently and with high quality provides acute care. And this misallocation question is also a measurement challenge because, um, again, as outside observers, as researchers, but also as policymakers, we have to define who needs what type of care and then measure if they actually receive this care. And of course, this allocation question is closely linked to the Good Health and well-being SDG which mentions among other things, pay for performance, sustainable financing, quality of care, and so on. So why, um, uh, look at both underuse and overuse, um, in low and middle-income country contexts? For surely, I don't need, uh, don't need, don't need to motivate, um, uh, the question of underused. Underuse remains, uh, even today a challenging problem. Despite substantial progress made, just to give one number, in 2019, 5.2 million children under five years have been estimated to have died, mostly from preventable and treatable causes such as diarrhea, malaria, and pneumonia. Part of the policy response uh to this underused problem has been a fundamental shift from a user fee model to providing free care for children and mothers in particular. Uh, this is a pivot away from the Bamako consensus in 1987. And the same thing has actually happened in Mali, where in 2019, the Ministry of Health announced the intention to provide free children's and mother's healthcare um by 2022. Having said all of this, however, um, there's at the same time, a lot of evidence that there's also an overuse problem. So medication misuse has been documented since at least, uh, 1990. Um, there is a WHO medicines use summary report in 2006 that shows some of this evidence, um, and what the report says, among other things, that is that there's substantive overuse of injections and antibiotics, a lot of instances of polypharmacy and so on. Polypharmacy is, uh, when the same set of symptoms is treated with treatment for different kinds of illnesses such as malaria treatment and antibiotics at the same time. More recently, there has been also a lot of experimental and other study evidence of these misuse and overuse problems. Many of those studies came actually out of the World Bank Research Department, and they show that um malaria drugs and antibiotics are overused, and more generally that there's a lot of unnecessary and even harmful treatments and insufficient diagnostics. a variety of conditions. Oftentimes this evidence shows that this type of overuse is the result of subsidies on the patient's side as well as financial incentives on the doctor's side. And this raises the question, if we try to combat underuse by introducing subsidies for patients, could this reinforce financial incentives for providers and lead to greater rates of overtreatment at the same time? So, when we think about the allocation of care and the correct use of care, um, this is certainly a challenging problem and the reason is that there are multiple and complex decisions involved from visit, from the decision to visit a provider through the diagnostic and prescription process to actually purchasing and then adhering to treatment. Um And as a result, there are also many potential sources of misallocation. Um, on the household side, um, certainly, we are acutely aware that there are financial constraints, meaning that families may not be able to raise funds to cover emergency, emergencies and acute care. There might also be informational constraints, for example, because households are unable to assess symptoms or correctly administer treatments. And uh a general recognized problem in healthcare is that there's often misaligned incentives when allocating care between the government, the funder, often of care, but providers, but also caregivers and patients. And these incentives are often due to uh externalities and uninternalized costs and benefits of care. One thing that reinforces all these issues is that there is a fundamental asymmetric information problem which is termed in health economics the problem of the informed expert. Meaning that there is one person in this process, uh, or, or a very small set of persons, the healthcare provider, who can actually diagnose and assess the illness and make treatment recommendations. And everyone else in the process has to rely on this provider's, um, uh, advice. But this can lead to issues if um If there is, if there is misaligned incentives. Now, what is, what comes on top of this since there are so many decisions involved, these different issues may interact or mutually reinforce each other and um one example could be that if if if patients due to um the provider having different interests from them, lack trust, this could lead to a lack of effort on the parts of the providers um to diagnose the illness and give uh and give specific advice. So, in this talk, I'm certainly not gonna be able to talk about all of the issues that come in here, but I would like you, uh, I would like to present you with 3 studies plus 1 unpublished study from Mali where we, um, collected experimental evidence on underuse and overuse of curative care. And then I would also like to take the opportunity to point out some new measurement approaches, uh, that we've taken. So there's two sets of studies. Um, the first two papers with my co-authors, uh, Christine Blander, Sam Brown, and Mark Dean covers the decision about care at home, uh, specifically children's care at home. And, uh, these, these, these studies are part of an RCT where we, uh, study two common policy interventions to improve access to care, specifically subsidies and health information provided by community health workers, CHW for short. What's novel about this study is that we conduct a very detailed home survey where we have 9 weeks of daily health diary data on symptoms, consultations, and so on. And there's two papers that I would like to talk about. The first is the actual randomized control trial, RCT. Here, we examine the decision to seek care, so the extensive margin of care, should we get care or not for children under 5. And in order to understand overuse and underuse, we construct illness spells, spells and assess the needs for provider evaluation using the community version of the Integrated Management of Childhood illness guidelines, IMCI. Um, which are guidelines for community health workers with rule charts that indicate when a child should see a doctor. I'm gonna talk much more about these studies. This is more a summary of what I'm gonna show you. The second study, uh, primarily examines, uh, gender differences, but, um, also gives us some insight on overuse and underuse because it looks at antibiotics adherence. And what we do is construct spells of antibiotics treatments and then compare these with WHO guidelines for minimal treatment length. The second set of studies I want to talk about is about behavior during the provider visit. So what happens when you actually go at the clinic. Um, and what we did here is to conduct a randomized controlled trial that isolates the contributors to misallocation of here specifically subsidized malaria medication. How we measure, how we approach the measurement, uh, question here is by conducting a clinic survey, we conduct entry and exit interviews, but then we also do a home follow-up where we carry out a malaria test so we are able to actually see whether patients who have malaria did get treatment and those who didn't have malaria did not get treatment. The first study looks at the first of the two cross-randomized interventions in this RCT where we handed out vouchers for malaria treatments, uh, artemisinin combination therapy, ACTs, either directly to patients or to the provider, and we use the difference between those two arms to ask, when doctors have discretion to decide about giving a voucher or not giving a subsidy or not, do they use that discretion to steer patients towards more or less use? And again, I'm going to talk more about these studies and the results. The second intervention tried to approach this issue from a different angle, and we asked how information can encourage adherence to test results and, and improve the allocation of malaria treatment. And we did this by providing information on malaria tests, uh, rapid detection test, RDT efficacy, either to doctors through a day-long training or to patients through a short video. Now, um, in, uh, to give you a preview of some of the findings and to summarize a little bit how these two, these studies all speak to overuse and underuse, let me summarize for you. Um, we find that in care seeking underuse as um measured by the IMCI is, is definitely alleviated but by no means eliminated by subsidization. And, um, despite this increase in use that is mostly um uh an alleviation of underuse, there's also a significant share of overuse in those newly created provider visits from the subsidy, around 30%. And on top of this, information does not improve the allocation of care for the information provided by the health workers. Moreover, when we look at the clinic, when costs are low and the subsidies available, providers seem to fail in their gatekeeping function, keeping prescription medications out of the hands of patients who don't need them. And one significant factor in this seems to be patient pressure who, patients who are asking for these medications. We find among other things that informed doctors who know informed, in quotation marks, doctors who received the training on RD on RDT tests are better gatekeepers, but this actually leads to a reduction in patient satisfaction. And then finally, on the treatment side, on the adherence to medication side, we find that even after the medication was purchased here, we're looking at antibiotics, but the cost should be low, and yet, there is a lot of underuse and it is even higher for girls than for boys. So this raises a number of questions, among other things, how do we direct subsidized care to those who need it the most and make sure that resources that are used serve their intended purpose. But in addition, what these pieces of evidence highlight is that there are multiple margins of overuse and underuse, and I will argue that this raises a number of important open questions. For example, do these margins interact and possibly reinforce each other, and then what are the actually the net effects on health outcomes of these, of these misallocation issues? So what I'm going to do in the rest of the talk is I'm going to do, to introduce the setting and show you some summary statistics of our samples. And then I'm gonna talk about each of the studies in turn, the RCT on subsidies and healthcare, um, workers, health workers, community health workers, sorry, um, the study on overuse of malaria drugs and then the adherence study for antibiotics, and then I'm going to conclude. So to start with the setting, um, Public health in Mali is provided uh through decentralized community-based care funded by user fees. This is in accord with the Pharmaco consensus. And a lot of this care is provided by so-called SSCO's Community Health clinics. Um, and these clinics are commonly used, so 47% of mothers, uh, reported in the DHS in Bamako that they took a child with a fever or cough to a CSO. And there's around 70 of such, of such clinics and associated reference hospitals all over Bamako. Our, uh, malaria overtreatment study um uh was conducted at 60 of these clinics, and we have a census that gives us some insight into what these clinics look like. So, they have about 30, they have about 30 patients a day on average. Um, most of these clinics actually have a basic laboratory where they can conduct microscopy. And many of, uh, all of these uh SSOs, at least in Bamako in the capital, have a dispensary that has essential medicines available for sale. Around 11 staff have um the, the ability to write prescriptions for things such as malaria treatments. Um, and it should be said these are, these are doctors and, uh, and, and nurses, but also medical interns who staff these clinics after hours. And you can see that there is a resource constraint. Uh, many of these clinics report temporary stock-outs, but only in 1.7% of cases, they are really stocked out of all the malaria drugs on a given day. So our samples come from the two RCTs and the first is the Action for Health RCT that studies the subsidy and community health worker intervention. Here, we conducted a home survey with 1,768 children in 642 compounds. A compound is a living, a sort of a living unit, uh, where several households share some common infrastructure. And, um, in the sample included, are all families eligible for the program. This means they have children under 5, they passed a proxy means test, and, um, they lived in the expansion area of Action for Health, which was a program by the, um, by the health organization Mali Health. Um, the second sample is the clinic data with 2,055 patients with acute systems collected at the 60 SECOs in Bamako. And as you can see, these samples are actually quite fairly similar. So our action for health data comes from the household heads, but you can, and the, and the, and the clinic data comes from patients. There are some differences in literacy rates because uh women generally have lower literacy rates, but you can see overall literacy is low. Only 13% of our action for health households have salaried employment. Um, the households are relatively large. But a little bit smaller in the Action for Health program, probably because these are uh relatively young families, but they are similar in average monthly income per person between $22,000 and $26,000 safer. So in US dollars, around $40,000 to $45 per person. And as you can see, at least in Action for Health, this is an urban area, so this means these households live relatively close to the clinic, only around 1/3 of a mile or 0.5 kilometer away on average. So, let's start with the first study, care seeking for children. Um, To motivate the original study, what we were interested in is to understand the effects of subsidies and how they affect the allocation of care. Subsidies have the potential to reduce underuse due to uh access barriers such as credit constraints and incorrect beliefs, but because they monotonically increase use, they may also lead to overuse, and in that case, information may help improve the allocation. Information is here provided by community health workers who visit families biweekly and discuss health um and healthy and symptoms and health needs with parents. In the experiment, we randomize compounds into these two treatment arms, um, cross-section, uh, we do a cross-randomization, so some households receive both and some only eat one of each of the interventions. The first one is the subsidy intervention where, um, Action for Health pays for clinic user fees for the five most common conditions for children under 5, and then the information intervention of the community health worker visits. Um, One puzzle in the literature on these types of interventions is that policies that reduce the cost of acute care often actually have relatively little health outcome effects. And we don't always know why. And so what we did in this paper is that we measure health utilization directly using the diary data I mentioned before against the IMCI benchmark. So what we have to assume in order to define overuse and underuse is that compliance with the IMCI actually is important for health. So in other words, when the IMCI says by the rule chart that a child should seek care, if this child does not receive care, we think of this as underuse of, of, of the doctor. And in analyzing this data, we will treat the decision to seek care as basically a stopping problem. The parents make a decision every day watching the illness of their child, um, whether they should go today or wait for another day to seek care. And the IMCI works kind of in a similar way because it gives, um, it gives guidelines for how long a kid should a kid should have, should wait before care should be sought, for example, for things like diarrhea. And so what we do is we use the MCI to classify a child during an an illness spell as either early for care, so you don't yet need care or care required uh and care should be sought today or should have been sought in the past. So now for the data collection, um, we had surveyors visit the households 9 months after the action for Health Start. Um, And they collected, they, they constructed detailed health diaries with uh the mothers. And just to give you an idea, out of about 60 days of illness, we uh 60 days of, of 60 diary days that we um collected for each child, around 16 days, the average child had some symptoms that have not yet received treatment. And the five symptoms listed here are the most common symptoms, and you can see that colds and coughing are very, very common among those illness days. And those are not as serious and in many cases do not require care. However, in 32% of cases, there's also a report of, uh, unusually hot or cold skin, so in other words, potentially a fever. In 9% of cases, there are more than 3 loose stools, or in other words, diarrhea, and in 5% of cases, uh, the child is reported to vomit everything, which is actually one of the 4 danger signs we collect and would require immediate medical attention. On top of these symptoms records, we have records on doctor consultations and medication purchases, and this is the first um, Impression of how the subsidy has affected care seeking. So you see here that only 89 times our children see a private provider. But 646 times they actually go and see and see a a provider at the SESCO. And what's striking is that the vast majority of these visits, 508 take place in the groups that received the healthcare subsidy. And you can see part of the reason is, of course, the cost difference between the two visits. Um, the costs of the visit are much lower than there is a subsidy, and I should say the costs are always a little lower than the value of care received or the total price of care. And part of the reason is that many households seek outside help to finance, uh, their care needs. Um, but you also can see, I'm not gonna talk much about this, but you can see that a vast majority of healthcare is actually obtained in informal, from informal providers, informal sources of medication without any medic any consultation of a medical professional. So this slide shows you the main results of this um study. Um, what you see here is the 4 treatment groups. Um, on the left side, you see the groups without subsidies and on the right with the subsidy. And then on the top, you see the groups, um, without the health worker visits and in the bottom with the health worker visits. We have a total of nearly 4000 illness spells for these children. And these spells last on average 6, a little over 6 days and have 3 care required days. There's huge amounts of variation, so some, some, some reason for the high, uh, high number of average, the high average length is actually that there's some real outliers where kids are sick almost the entire period that we interview. Um, but what you see are basically three main points. The first is, you see this in the control group even, um, So, sorry, I should say what's on the graph. So on the X-axis, you see the date of the illness spell. And then the two lines represent the probability of seeking care on either a day that is early for care according to the IMCI or where care is required according to the IMCI and on the Y axis is the probability of care seeking. And what you see is that um the probability of care seeking is certainly a lot higher than care is actually required. What you also see is going from the left to the right, um, as you, as these households receive the subsidy, care-seeking increases a lot, um, and, and this increase is concentrated on care-required days, but by no means is care-seeking at 100% on care-required days. And then finally, what you see is that when you go from top to bottom, that the community health workers actually had surprisingly little effect on care-seeking behavior. And in particular, what they do not do is direct children better towards uh care when care is required versus not receiving care when care is not required. So they do not seem to improve the allocation very much. So to summarize the findings, um, we estimate these formally with a hazard model, and we find there is an over 5 times higher probability of care on care required days than on early days. So what this seems to indicate is that parents can clearly distinguish differential need for care by child's health status. They just underuse care overall. Um, the subsidy is successful in increasing care seeking. Um, the probability increases by 250%. And as a result of the differential care seeking and care required versus early days, about 70% of the new visits are actually IMCI recommended care, which is good news. However, it is still the case that 71% of care required spells still don't receive any care, even in the subsidy condition and care is almost entirely free. So what we see is that underuse is alleviated, alleviated but not eliminated by these subsidies. And moreover, because underuse and overuse increase at the same rate, despite the very low probability of overuse on any given early day, there is still a significant share of utilization, uh, share of doctor visits that actually um constitute overuse, um, according to the IMCI. And, um, in addition, the information provided by the community health workers does not improve this allocation. So what, what this, what that this raises, uh, raises some questions, in particular, how could we direct care better, um, the subsidized care better towards these care-required illness episodes versus, uh, care not required episodes. Now for the 2nd study on patient demand for treatment, this is now a study at the clinic, and we are focusing on prescription and purchasing behavior. The motivation was the policy challenge of overtreatment and wrong treatment obtained at clinic visits, even in very low resourced settings. And our data gives a good impression of this. So from our home visit data where we conducted malaria tests, we know that 42% of malaria negative patients received treatment. And on top, 20% of malaria negative and 53% of malaria positive patients received treatment for severe malaria. So to give you an impression, severe malaria is a, is a crisis state of malaria, so to speak. It usually comes with convulsions, um, organ failure, sometimes coma, and, um, and it is an imm, it requires immediate medical, uh, very, very significant medical intervention. Um, it's treated typically with injections and then an immediate referral to the hospital. It is typically estimated that out of all malaria cases, around 10% escalate into severe malaria, and because we interviewed these patients as patients at home, they were clearly not referred to the hospital. So there's good reason to believe that a lot of this severe treatment is actually overtreatment of some form. Now, the question is, is this overtreatment driven by providers? Um, the reason why this question is prevalent is that a lot of health economics focuses on the financial incentives for providers to overprescribe and increase the budget of the clinic or increase their own income. Um, there is some evidence in the data for this possibly being the case. First of all, patients in our data do not always, only in about 2/3 of the cases actually buy a prescription for an anti-malarial that they receive. And moreover, severe treatment is over 2 times more expensive than simple treatment, even though, and it is prescribed very often, and a lot of this, a lot of the fees are actually fees for labor, so they would go directly into the clinic salary budget. On the flip side, when you actually ask providers why over-prescription happens, they will tell you something different. They will say the patients come to the clinic knowing what they want and they ask for this treatment. And again, we have evidence for this in our data. Before the consultation, 55% of patients we interviewed believe specifically that they have malaria and need treatment for it. And 57% of our health workers report that they were, that they feel pressured to prescribe medications and oftentimes they mention. Anti-malarials specifically. What's remarkable about this is because these doctors, these providers have prescription power, this requires to some degree reluctant compliance by them giving in to the patient's demands against their own better judgment. So in order to test which of these two explanations is prevalent, we conducted a, an, a randomized trial where we rotate each clinic through 3 pricing conditions and the price variation is for the simple treatment for the malaria treatment that is standard and recommended for simple cases, the artemisinin-based combination therapies, um, these are simple tablets taken over 3 days. On control days, there are no malaria discounts available at all, but then there are two additional interventions. The patient voucher intervention gives out these free vouchers for simple malaria treatment, and these vouchers are directly given to patients. Um, the vouchers are not valid for severe treatment or any treatment that involves injections, and they are received importantly before the patient even sees the doctor, so they come into the consultation knowing this treatment is free for them. On doctor voucher days, by contrast, the doctors obtain this exact same voucher. It has the same terms and conditions as in the PV arm. However, they have the discretion to give the voucher to patients as they see fit. The patients do not know in advance that the ACT is free before they uh consult with the doctor. Now, how does this help us distinguish uh who drives overtreatment? Well, there are different predictions for treatment effects under the, under the two treatment arms. First of all, imagine we give these vouchers for patients. Um, this reduces the cost of simple malaria treatment and keeps the cost of all other treatment the same. So what this should do is, it should increase the demand for the simple ACT tablets at the expense of not receiving any malaria treatment or receiving severe treatment, receiving this treatment that involves an injection. Now from the doctor's side, when we give the vouchers to doctors, if they show the voucher to the patient, that should have the exact same effect on patient demand. However, they can, they can decide not to show the voucher to the, to the patient and use their information advantage to steer the patient's demands. And one option they have is they could keep overall demand lower than in the PV condition by not showing the voucher to patients who would otherwise buy no treatment, but showing the voucher, giving the voucher to those who would request severe treatment and thereby reduce demand for severe treatment only. If we see this, this would be evidence of patient-driven overtreatment, patients actually pushing doctors and doctors trying to keep demand low. On the flip side, what they could also do is they can use their information and advantage to increase overall demand, so they could show the voucher, give the voucher to patients who would otherwise not buy treatment, and, and, uh, increase purchasing rates and prescription rates, but at the same time, not use the voucher for those patients who would otherwise get severe treatment and therefore keep demand for the severe treatment option high. If this is what we see, then we would conclude that this is, that the overtreatment we see is doctor-driven and there's a significant push to sell more on the doctor's side. So what do we find? Well, we find the overwhelming evidence that there's actually significant patient-driven demand, demand at the margins where we, where we, where we see changes. Um, we find that in the, in the patient voucher arm, significantly more vouchers were used than in the doctor voucher arm, despite prices being the same. We also found that that doctors make significantly more anti-malarial prescriptions in the patient voucher arm than in the doctor voucher arm, and you can see that there is not a large difference to the control where the price discount was not received at all. And then finally, we find a significant increase corresponding to this prescription increase in anti-malarial purchases. So many more patients actually receive an anti-malarial in PV over DV. We also see, and this is what I mentioned before, we see that you can see the large difference in purchasing rates in the, in the control group, so many patients actually do not purchase treatment, and there is a significant increase when we receive when when vouchers are actually available. Many patients are now able to afford the treatment. So, what, what, uh, what else do we find? Um, first of all, we, in contrast to these results on, on the patient demand side, we find no evidence that there's doctor-driven demand. In particular, we find that both in DV and in PV the treatment reduces significantly how much severe treatment is used. So doctors substitute patients from the more expensive to the less expensive treatment option. And what we also find in general using our home testing data is that there's a really poor match between treatment and illness in both directions. So some patients who have malaria in the home test did not receive an anti-malarial, and many patients, I mentioned this already, who do have, did not have malaria did receive treatment. But what we find in particular is that this illness treatment match is worse in the patient voucher than in the doctor voucher condition, even though there is actually some improvement in the underuse component of this mismatch from greater purchases that result from the fact that in the voucher conditions, treatment is now free. What we also find is, uh, this is the 2nd paper, the unpublished paper that um that I mentioned earlier in the intervention where we provided training. When doctors receive additional training on the efficacy of tests and using those tests for, for making treatment decisions, they actually make better prescriptions. The treatment illness match improves, but we find that this lowers patient satisfaction, which is another piece of evidence that these patients are pushing for more medications. So to summarize, we find that at least some of the overuse observed at the clinic is surprisingly driven by patient demand. Um, we find that information in the form of the training we provided had an effect on the doctor's side, but the video we showed at least on the patient's side, did not improve allocation. And what those two things point to is that there is a twin problem, a surprising twin problem where we observe underuse at the extensive margin, meaning whether care is received at all, whether a patient sees a doctor. But then there's overuse at the intensive margin, namely, how much care is actually received. And one way of, of thinking about this is that there's at least a possibility that these two problems reinforce each other. I mentioned this before. So, suppose only a small subset of patients go to the clinic. These might be exactly the patients with the highest and the strongest demands for some kind of um powerful treatment. So in order to meet these patients' preferences, the doctors tend to overprescribe, and that makes care on average more expensive. But this could have this paradoxical effect that even fewer people go to the clinic. And on top of this, those who do go to the clinic, uh, have less trust in the doctor's recommendations. And that in turn, and this is the subject of the who signed it al. paper that just recently came out, this in turn might mean that doctors have less of an incentive to actually put effort into diagnostics and giving good treatment recommendations, but rather they just follow what the patients want because they anticipate that patients will not follow their advice closely, um, anyways. So, this is at this point, of course, we need more evidence that these, these chains of reasoning hold, but there is um at least some reason to be concerned about them. If true, then this is a really challenging policy issue because of the aforementioned informed expert problem. The provider is really the one who diagnoses the patients and knows what treatment is needed, and we rely on the provider to both uh give medications when needed, but also gatekeep medications that are not needed. And monitoring this behavior would really require observation by yet another qualified provider and that's just very often not feasible. And so one interesting potential that arises from what we found on our information treatments is that paradoxically actually empowering providers might here have played a role in improving the allocation of care. So the last study is on treatment and adherence, and it originally focused on gender differences in antibiotics treatment adherence, but it also has some things to say on overuse and underuse. So what we do in this paper is we use the same health diary data from before and we construct antibiotics treatment spells. And benchmark exit from the spell or stopping the treatment against WHO recommendations. And we can use our data with all our detailed symptom records and so on to estimate on the, on the one hand, the probability of actually starting a treatment spell and then conditional on having started completing that treatment spell, and we can look at the role of gender and other variables um in this decision to end the treatment. The table below just gives you an overview over our antibiotics spell observations. Um, we have a total of 712 instances of antibiotics used in our data, and as you can see, most of the standard recommendations involve essentially 5 days of antibiotics treatment with some exceptions for, for other, for specific conditions. So this graph gives you kind of the the headline results. Um, it shows on the x axis again are spell days, and it shows you how many children are still taking the antibiotic on each day. And of course, if we, if this is an antibiotic where the 5-day treatment recommendation holds, we should see 100% use up to day 5. And what we see is that unfortunately, that is not at all what we find. Um, the girls in this data set have only about 38% completion rates of antibiotics, and boys have higher rates, but by no means, um, full full adherence, but it's a significant difference of 9 to 12% um between the two genders. And what we find is surprisingly, that antibiotic that this completion advantage for boys accrues almost entirely to the sons of educated household heads. What you see in this graph here is the gender effect, um, the difference in antibiotics adherence for boys and girls, um. Split up by education level of the household head. The reference category is no formal schooling. And what you see is that in particular, in highly educated households, this gender difference exists, whereas in, in, in households where the household head is not as highly educated, um, there is not much of a gender difference. One interpretation of this is that um is as follows. Education is essentially needed in order to make good adherence to decisions. This could be because it comes with higher literacy rates and so on. Household heads are typically much more educated, and they are also typically male. And more generally, there's a huge education gap between women and men in, in Mali in our data. And what might be happening here is that these male household heads who, who are able to read and follow the treatment instructions, just pay very different attention to the care of girls in the household over the care of boys. And um so, so to summarize with an eye to overuse and underuse, despite a really low cost of adherence, the antibiotic is already purchased, um, it's in the hands of the household. There is still significant underuse in the sense that many, um, that many households do not adhere to the full treatment length. It also seems that the presence of a, of a, of a literate, educated parents matters significantly. And this suggests that there are subtle biases in the, in the use of treatment that may affect other aspects of treatment success that we may not have a lot of information on because they happen in the home and we don't very often see, uh, see these behaviors. Now, what's really interesting and puzzling and a challenging problem is that um we know that there are really high levels of overprescription of antibiotics. We saw the example of malaria, but similar things are true for antibiotics. And um and I think there is some hint of this in this data. I mentioned there are 712 antibiotic spells. And remember, we only had about 730 formal instances of care in the first place. So, a lot of times, children are prescribed an antibiotic. So, um, Low adherence to the antibiotic treatment recommendation is of course underuse in the sense that parents don't follow the treatment instruction, but many of the spells we see may actually constitute overuse in the sense that the child's illness is not caused by bacteria. So the policy problem here is simultaneously make sure that many of these children don't use an antibiotic at all because they don't need it, but then for the other children that they actually adhere to the full treatment recommended length. So, to conclude, um, We have seen that, uh. Uh, in this, in this decision chain, there are multiple margins of misallocation possible and, and documented. And what we've also seen is that many of the policy instruments we have are in some ways, uh, blunt instruments in the sense that, uh, they might address some of these margins of overuse and underuse but not others. Just to give the example of subsidies, um, subsidies may help address underuse by bringing more children to the doctor in our case, underuse at the extensive margin. But they might at the same time exacerbate overuse at the intensive margin, meaning that when, once the patient does show up at the doctor, they actually receive too much care. Um, what this suggests is that there is a need for additional research. Um, we would like to know in much more detail how different incentives and policy interventions change utilization at these different decision margins, and then it is important to understand actually, and this is something that, that is in some ways missing here. How do these changes, both individually and in the aggregate, actually map into what we care most about, namely the welfare of the patients, their health, but also their income and uh happiness. And what I would like to advocate for in this talk is that we need better and more detailed measurements of the allocation and quality of curative care. Researchers and policymakers already have uh ways to measure these. Um, for example, in economics research, people have used audits, that is direct observation, um, as well as standardized patients. So these are trained actors who present with specific conditions and then record what the doctor, uh, does in terms of diagnostics and prescriptions. And actually, the widespread use of these methods was pioneered by World Bank researchers. Uh, but in this talk, I also showed you two potential new approaches. One is to use health diary data of illness and medication spells to understand behavior at home, and the other is to conduct, uh, not only facility entry and exit interviews, but combine these with home testing. For malaria, we can actually conduct home tests that can help us assess the match of illness and treatments, um, that, that actually occurred in this specific doctor consultation. This is all I have. Thank you very much. Um, I'm gonna hand back over to the. Thanks, Anya, for that um really detailed and insightful set of, set of results, um, which I, you know, I, I hope moving over to Gaston now and maybe the discussion, there's some real policy implications here that are kind of hard to grapple with. So, um, Gaston, uh, over to you for, for 1015 minutes and then we'll open up the questions. Just a reminder, if you want to ask a question, either just raise your hand in the chat, chat or signal that you have a question, and I'll come to you later. Uh, Gaston, over to you. Thank you very much, uh, Dion. Uh, NJ, uh, you are the author of this study, and I would like to thank you for inviting me, Dion, and the entire deck team for asking me to join this, this event. I am a practice manager who deals with policy dialogue and operational interventions with our client countries and, uh, I'm working in close to 7, 17 countries in West Africa. So, I, I looked at your, your study with a very practical angle in answering the question, how does this, these studies speak, speak to me? And Andre, really, congratulations for having nicely sensitized 4 studies into a concise and excellent presentation that it's very difficult for anybody to grasp the full content and to comment very meaningfully unless you reduce yourself to a specific aspect that you can handle, and that's exactly what I'm going to do. You managed to, to really address a complex problem. Uh, because measuring healthcare, overuse and underuse depends on the definition of adequate healthcare, uh, which integrates individual preference as well as needs. So, uh, some scholars, for example, Vikas Susani defines operation care as. Uh, care that is needed, wanted, clinically effective, affordable, socially equitable, and responsible, responsible in its use of resources. Of course, uh, this is kind of a challenging definition, obviously, but it doesn't, it doesn't prevent us from, from really looking at the issue of healthcare overuse and underuse because overuse and underuse has several implications. Have you mentioned some of them and I would like to go at a little bit higher level because you, you, you are really focusing on demonstrating what is happening and what I would like to speak at is at what are the implications in terms of the, the, the, the something, someone like me who is really dealing with operations and how do I look at those findings that you just brought into the picture. So first, for, for patient well-being and safety, health systems should, of course, bring solutions, but often we fail to address patients suffering and even sometimes we create more problems, like you just mentioned it, from the data set. This is really critical. Second, for equity reasons as well, because every time one patient unnecessarily use, uses a resource, another patient is deprived from the same resources. So, very critical and very important. And third, Of course, from the economy side, because overuse and overuse imply misuse of equipment and human and financial resources, particularly overuse deflects resources from public health and other social spending, and this is aggravated by the lack of adherence to treatment. So in total, you know, we can see that these past years, overuse and underuse have attracted the attention of the century committee. And other health public health professionals. For example, the, The Lancet published the Right Care series in 1917. Which highlight that overuse occurs across a range of medical specialties. It concerns procedures such as cardiovascular procedures, hysterectomy, etc. and most studies on overuse have been conducted in high income countries. And although it's increasingly recognized in low income countries, it's not yet well documented in those settings. So your work fills an important gap in the literature, and it deals with key parts of the subject of inadequate care. You talked about financial barriers to care, drivers of overuse, adherence to treatment, etc. So you have marked my mind with key numbers that you brought into light, and let me just refer to, to, to, to 3 of them. You concluded in your research that with subsidies, underuse and overuse increase at the same time, at the same rate. That was a surprise to me. Other aspect you mentioned that 42% of malaria negative patients receive treatment. What a scandal, should I say, as a clinician. And you also highlight the fact that 20% of malaria negative patients and 53% of malaria positive case received treatment for severe malaria. That's really a real problem. And more importantly, You concluded that 57% of healthcare workers report pressure to prescribe unnecessary treatment and that some overuse, mostly dry, is driven by patient demand. So, I already said that your study fills a gap, but Additionally, it's timely and useful for evidence-based decisions, as many countries now aim to achieve investor health coverage, and we at the World Bank are supporting the global and individual country move towards investor health coverage. Now let me go a little bit deeper into, into, into the research. Some part of your results are encouraging since they show that once financial barriers are removed, demand for care when it's medically indicated increases significantly, which is great. But other parts of your results beg a fundamental question again using my perspective as practice manager. And the question is, how can countries achieve the ambitious agenda of universal health coverage while ensuring care resources are adequately used when we also observed that 50% of health workers in your study were pressured to prescribe unnecessary treatment. We all agree that this is concerning, and by that specific finding. It's great. It raises our attention because seriously, That's a huge question for all of us, and it's it's concerning because when you look at the framework of universal health coverage, of course, it addresses three dimensions, which are population coverage, yes, who is covered, service coverage, which services are covered, and financial coverage, what proportion of the cost is covered. But obviously one dimension that is maybe missing certainly is the issue of the quality, and this is exactly To what your, your findings is pointing to. Because, uh, and I believe seriously that in our countries, we'll certainly need to develop a strong quality control measures. In particularly, some specific measures will be needed looking at what you have found in your, in your research. Uh, first, we need certainly a strong communication for behavior change and in short, let me say education because while the principle of universal health coverage include the need to ensure that all citizens have access to data and information on ESA. As part of societal dialogue, and dialogue and participation processes, yet your self study highlighted that a specific role of education in care decisions. I think we would need to integrate some sort of communication strategies into our approach to make the population realize and understand that universal health care is not about getting care simply about Getting care that is needed, that is useful, and that is timely. So this is something we need to think deeply about because so far when we help countries we do not necessarily make this linkage with the issue of the that your side has highlighted. Second thing I think we need also to consider is the issue of the shared decision making as it goes end in end with the first one that I mentioned, the first measure that I mentioned. Shared decision making is a process whereby clinicians collaboratively help patients to reach evidence informed and value concrete medical decisions, and several studies have shown that some patients know when care is not needed and that there's a lack of shared decision making in this process. And this is really critical and very important. And again, When we are moving at this pace whereby we can ensure services to the maximum of population, we need to make sure that decisions are made in a sound manner. So this will call for us really to be more regarded and more attention to some of the aspects because many studies are found, and this is corroborate what you just expressed in your study, that for them, in the US 17%, meaning that 1 in 6 patients. Reported that a healthcare provider recommended a medication test or procedure that they felt they didn't need it, and 50% of the adults who had received a recommendation for a test they didn't think they need it will end up getting it anyway. You know, so this is really even in developed countries this is happening. So what about our countries where the situation is even worse? So it really says that we need to be more focused and pay more attention to those aspects that you have mentioned in your study. There are more studies on this, for example, in South Africa, just to be close to Africa. Well, patients do not always request antibiotics and we're willing to discuss, they're willing to discuss with the, the doctor how to manage the condition without them. Doctors in private clinics often go straight and prescribe to medical, to, to, to the patient, and they will often go and get the medicines. So, those are really critical questions that you have highlighted through your, your study. So you also mentioned the use of the informed expert. I wouldn't elaborate too much on it, but it's a real question that we have to integrate as well in all whatever we are going to do when we move toward universal health coverage. And then we need really to be, to get some of the initiatives like such as choosing Wise, the initiative that promotes patient physician conversation about unnecessary medical tests and procedures. So the 3rd measure is the the the resource use monitoring system. And I think that based on what you have shown and demonstrated, definitely we need to be more careful and more looking at what we are doing when we just subsidize subsidize services. It's not enough to just pour resources and even drugs into the healthcare facilities. It will not. resolve the problem unless we have means to measure and to assess how those is used and to give guidance, guidance to the practitioners and also to the clients to get the best from whatever we are putting in the services. So this is something very important and in many countries there are already different tools that have been in utilization. We may either adopt or create new ones, but definitely we have something to do around this issue as well. So let me finish by a practical way forward suggestion. So I know it's certainly done, but I would like to see DC teams reach out more directly to practitioners active on the ground. And what I mean is, for example, I would like to invite you to join my next monthly unit meeting with the staff so that you can present the findings of your research to the staff, gather the comments from the TTAs. Who are working with the government to to increase access to services and discuss the implications of the findings of your research on the police and operational ground for World Bank Tail working in Africa. So this is very critical because this webinar is certainly interesting, but we can go deeper and more into detail with the units and help them really grasp this kind of information. It's so critical and it's It's not obvious that people will be exposed to this information. It's so vital, and I'm very happy that I was invited to comment. I wouldn't even look at it, even if I see it passing around. But now that I have the chance to look at it, to listen and to see, I realize how important and critical it is for our entire policy dialogue with the countries, and really I'm amazed to see this this correct. And its complete implication between research and operation on the ground. So again, having said that, I would like to stop there and welcome the Q&A question. And again, congratulations to you and your team for the work that makes us think about how to effectively use results from field studies to meaningfully increase our chances to achieve coverage in our client countries. Thank you very much. Back to you, day one. Thanks, Gaston, and, and, and based on previous conversations with Anya, I know she'll jump at the opportunity to, to join you at your upcoming uh staff meeting and, and I, I know she'll be happy to, to have the kinds of discussions that you anticipate. Um, and, and thanks for saying that because, uh, you know, this really is the one of the purposes of these talks is to engage people who might not have paid attention to some of these results, um, and it gives us an opportunity to, to highlight them and, and, and, and really engage with, with the likes of you and, and, and others. Um, Anya, before turning it over, do, do you have any quick reactions to any of the points that Gaston made or, or should we go straight to Q and A? Uh, yeah, let me just, uh, say a couple of things. So thanks, Gaston, for your comments and I really appreciate, um, uh, how passionate you are about this and, and agreed with Leon. I would love to, to, to join you, um, in, in discussing this more and with a smaller team. But um I think you make two really great points that I want to just quickly comment on. So, first of all, you said that, um, and this is something that I think is, is true in, in, in our setting as well, but you said that there's evidence at least from the US and Europe that patients often suspect that their doctors overprescribe. And I think what this speaks to is exactly this issue of trust that I mentioned before and, and you. You have said that you think that it's really important to uh come to a model of, of joint communication and establishing trust between patient and provider to improve the allocation of care. And then you've mentioned a second, um, a second need and that is monitoring resource use. And I'd like to connect those two a little bit because I think what we find in our data, at least, and I should say, I mean, these are small studies and they are from one country, from one, city, so I think there's the caveat of to what degree we can generalize. But what we find in general is that these providers are clearly interested in their patients' welfare. They are not as guided by financial incentives as maybe some health economic studies, uh, suggest, but they do give in patient pressure. And I think one question is how can we use this to improve care? And I think one thing that has struck me is that um in order to establish better communication, greater trust, one other scarce resource is rarely mentioned, and this is time. Um, I think these, these issues do require time. We see that patients are, are constrained in that, in that way, because we see that they don't even go to the clinic when the care itself is free. This speaks to some time constraints. But we also know that when a doctor really tries to communicate their, their insights, their advice to a patient, this actually takes a lot of time to teach the patient what they need to do when they go home and so on. So I agree with your, with your, with your thought that we should have better data collection systems to monitor resource use. But I think um this also provides an opportunity perhaps to introduce systems that cannot just Monitor, but can also facilitate communications and improve processes and save both patients and providers valuable time that they can then use to communicate better about the things that really matter, such as how to use antibiotics when you actually come home with your kid. And I think there is a huge opportunity here. So I remember from my, my research collection in Mali, from my data collection in Mali, um, in the first few years, we actually visited a lot of clinics and we saw how doctors keep track of their visits. They have these beautiful ledgers, handwritten ledgers filled with fountain pen. And all this data in them is basically not used very well for any kind of decision making, and it seems really important when we introduce systems to monitor better, to also use that information for the doctors and for the patients' benefits to actually improve the care that they receive in each visit. Um, Really, yeah, I'm gonna stop here. This is, this is, these were my main comments that I was gonna make. Um. Thanks, Anya. Gaston, I'm gonna give you a chance to come in a bit later again, sort of maybe after some discussion, unless you really want to come in immediately right now. Um, or, or should we just take a few questions and then come back? No, let us move for the, for the questions, and I'm happy to hear that. And just to mention that I have a chance to live in Mali for Mali for 2 years at the World Bank office, and I was there, the, like, how do I call it, the detail for the operation there. So I have a little sense of what's going there, but it's cuttered some time ago. Back to you. Thanks. OK, great. So I'm gonna, so there's a couple of questions. I'll, I'll just mention the names and maybe we'll take, uh, we'll take 3 questions, um, with the, the 3 people have signaled, um, I'll go in turn and then maybe Anya you can respond, uh, uh, in one go. I'm gonna take the chair's privilege and actually interject 11 little question beforehand, which is, you, on one of your slides, you, you, you, you pointed to the high use of informal care. And you said it to me, we're not gonna talk about that today, but I did, it was very, I, I, I'm very curious about what's the content of that. Is that pharmacies, is that other care, and how does that interact with this kind of cares seeking behavior that we did see? And I realized that's maybe a discussion for another day, but, but just a little bit of insight on that, I think would be useful. Um, so in terms of the questions, Ellen, are you still there? You had two interventions, so maybe if you could sort of. Give each of them, each of them briefly, uh, do you want to turn on your mic, Alan, and maybe your camera and ask your questions? Uh, sure, Dion, can you hear me? Yes, we can hear you well. OK. Uh, OK, yeah, I feel I'm just gonna repeat now what I already thought, but thanks for the interesting presentation. I was just also like in line with, I think, what Gaston was saying, trying to translate your findings to the operational work, um, that we are sort of engaged in on a, on a, on a daily basis. Um, I work in the global financing facilities, for not a specific country, but involved in many operations, especially in West Africa. And I don't know whether I understood sort of your conclusion on the first part well, but to me it sounded like indeed demand size subsidies make people go more to facilities, but that's not a guarantee that actually care gets better allocated, which resonated, but at the same time I was thinking, um, like putting myself in the shoes of a TTLO, so what, like what, what is it? Tell me, if we seem to be struggling with only supply sides, um, strengthening through results-based financing. Demand side subsidies struggle from this shortcuting like what, what would be the, the sort of guidance or what would the questions be to ask when we're developing a project under a limited timeline and limited resources instead of how, how do we, how do you see we could mitigate this? Um, what to watch out for or build into these projects to at least try to, um, alleviate some of these concerns. And then I guess a second um related comment to the second part of the presentation where I fully appreciate the point you're trying to make that even in low-income settings it's not only about low utilization rates and underuse but also inefficiencies through overuse, I think. Probably most of my colleagues would agree with that, but then again, sort of operationally, what do we, what do we do with this if you work in a country like Mali where overall utilization is still so low for very basic, uh, services and we have limited resources and efforts, where are we gonna put most of our efforts on? I, I, I unders I sort of, I was thinking, uh, given everything you described on working through. Community health workers and that it's really quite a difficult agenda to address, for example, overuse or misuse of antibiotics. It would almost require a project in itself, right, because it's financial, behavioral, it's a complex thing to change. I even see that in high income countries it's really a problem. So again, sort of them thinking, do we, do we, should we, do we want to invest in this in countries like Mali? Is this where we get the biggest bang for the buck, given the sort of limited resources? Um, or should we sort of work on this agenda depending on where countries are and the extent of the, of the problem. Just again thinking that the low health outcomes you see in a country like Mali are very much driven by the very low use of care and perhaps not so much by the, by the misuse of care, um, but you might disagree. I was just wondering, uh, what your thoughts were on, on these sort of more operational questions. So thank you. Over. Thanks, Ellen. Um, Khuram, are you still there? You had a very, one very specific question. Do you want to ask that? Um, maybe Kuram, uh, no, Kuram, I see he's still on, um. Um, sorry, uh. My another question about uh the overuse of antibiotics and uh that is uh in a country where there is a high level of pollution. Don't you think that the doctors uh Prescribe for overuse of antibiotics to to protect the patients and uh uh and uh to uh lessen the effect of illness. So might they be doing something good anyway by overprescribing because of other conditions? OK. And then Adna, are you there? Would you like to ask a question? Uh, I think I don't know may have left. Um, OK, so, I know, do you want to take those, those two? Sure, thank you. Um, thanks so much for those questions. Um, So, Dion, to start with your question on, on informal care. So you're right, we are seeing a ton of these informal care users. So we collected, um, we collected absolutely everything. These are peddlers who come to the home with, um, with little packets of, of treatment options. These are market stall visits and so on. Um. What you could also see is that a lot of these, these, these informal options had incredibly small costs, so 200 safer, that's Divide by 500, that's a really small amount in US dollars. Um, so, what most of these visits actually involve or what, what these, what these purchases involve are really um remedies for symptoms such as Tylenol or ibuprofen. They are not often actual treatment, for example, like anti-malarials and antibiotics. Moreover, they are often, um, if they are labeled an antibiotic, we are not sure what the quality of that care actually is. So we would certainly And when we encourage care use by subsidizing, we really want families first to see a qualified provider to diagnose before we want them to buy anything. But you're certainly right to ask about the interactions of the existence of these informal options and, and the use of formal care. And the, the short answer, at least for our data is that we are just not powered to say much about that, but I think it's a really important question for future research to think about. Now, uh, Ellen, thank you. You asked a lot of really good and interesting questions, important questions. Um. Your first question was, well, now that we know that patients, um, go more to the, go to the doctor, go to the provider more when we give subsidies, and we've also learned that some of these visits might actually be overused, well, what should we do? Um, I think, well, the first answer is that I, that we had hoped when we studied the Action for Health program that one intervention that could actually improve this allocation would be precisely the visits from these community health workers. One of their, this is not the only thing they do for sure, they also do some prevention and so on, but Uh, one of the things that they do is they actually go through the symptom rule charts and, um, and try to assess if a child currently needs care. And on top of this, they are also supposed to teach parents how to take better care of their children, for example, how to respond to diarrhea and when the child should see a doctor versus not. Um, unfortunately, this is not what we find. We did not find that in this particular program, the community health workers improved the allocation. This is certainly not to say that this must be always the case. There could be measures to be taken to improve how community health workers steer the use of formal care, but I think it is fair to say that this is an open question how we could improve the use of care at the home. Um, I also think it is really, you make one important point that it could be that this utilization problem, this misallocation is the reason why we don't see as many health effects. But I think it's also fair to say that we don't know that either yet, for sure. So the IMCI has been tested many times in public health context, and it has has been found to be effective. But the IMCI set of interventions is actually a whole set of things that happen. Uh, not only these community health worker visits, but other things as well. And so, pinpointing how the exact time when you go see a doctor affects later health outcomes is going to be another open question to, to answer with more research. Um, so this is not as conclusive an answer, I'm afraid, as, as, um, as you might have hoped for. But I think it's fair to say that there's many things we just don't know yet. Um, you asked the second question, which I think is a really good question. And that is one that I can probably and should probably not answer. And that is the question, how should we prioritize between the two things? You're making the very good point that because underuse is such a big problem, maybe this should be taking priority over any problems of Overuse. And I think there's many, many areas and settings where this is definitely true. So Mali itself is just an example. So the study that we conducted is in peri-urban and urban areas where clinics are actually relatively frequent. Most clinics have uh a provider on staff, a qualified medical trained doctor, um. They are relatively, they have good opening times and the providers are typically actually present. I think there is a huge, a huge, a hugely different problem of access in rural areas, for example, in Malian rural areas are certainly a case in point where the clinics are really far and even when you go to the clinic, it might be closed or the provider might not be on site and so on. So these underused problems certainly I'm almost surely. Thinking that we should take priority. But once we go to urban areas and settings where overuse is more of a potential issue, I think we should also think about overuse. And the reason is simply that resources are so incredibly scarce. Um, you saw the data from, uh, some of these clinics in, in Bamako where the dispensary, the, the, the national. Dispensary is actually closed and they can, they have relatively good access. And yet many of these clinics run out of treatment regularly and have stock outs, which means that because some people receive treatment that they didn't need, there will be days where just no pre-treatment is available, and some folks come in and do not get the malaria treatment that they would have needed in order to cure the infection. So, I am probably not in a good place to give definitive recommendations of how limited resources and policy making should be allocated. But I do think that there are potentially mutually reinforcement issues here, and I mentioned one pathway in which this could happen if the quality of care at the clinic is, is low and overuse or overprescription by providers is really high. This might actually discourage people from coming to the clinic, and it might also lead to greater distrust between providers and patients and overall substantial misuse of resources. So, I think of these as joint problems in many ways, and I would, I would, well, I would advocate for thinking about overuse as well, but I totally take your point that we need to potentially make set priorities given how scarce all our resources are. For the last question, Koram, um, You asked about potential overuse as a preventive measure, and I think you make a really good point. In fact, uh, for antibiotics, I'm, I'm less familiar with, with exact medical treatment recommendations, but for malaria, it's certainly the case that treatment recommendations have changed over the last few years. So, um, so, originally in malaria endemic countries, the recommendation was actually, especially when testing is not available, to simply treat any fever as if it was a malaria case. And the reason is simply that the downside risk of not treating a malaria case is so much higher than the, the, the slight risk of overtreating. However, in recent, in recent years, we have become much more attuned to the problem of creating resistances. We are already on the 2nd generation of anti-malarials and 1st generation anti-malarials have often become ineffective, so parasite resistance is a real concern, and, uh, malaria treatment recommendations have been updated to really only provide treatment when. There has been a test. For antibiotics, this is a much harder problem because it is harder to test for bacterial infections. But the, the, the rule of thumb is that uh giving an antibiotic does not have preventive effects. Um, you really need to have a bacterial infection for an antibiotic to do something for your health. And in fact, we also know that antibiotics generally actually harm. The, the natural bacterial flora of the of the body. So it is not obvious that giving antibiotics preemptively is, uh, is particularly helpful for overall health outcomes. Having said that, I, I, I think in, in other contexts besides antibiotics, your point is certainly well taken that when you have little access to diagnostics, overtreatment might be actually the right course of action, and we've seen that in the past. At least in our setting, this is not the case because rapid detection tests are widely available. They should be free for patients, and they are actually Mali's and the WHO's official malaria policy to use before treating. This is, uh, I think I answered all the questions, but do let me know if you, if you have follow-up questions. You did, and in, in fact, and in passing, you also answered Ana's question, which was about community health workers and how, you know, how they could be leveraged or some alternatives to them could be leveraged to right some of these wrongs, but you, you answered that in, in, in part in response to Alan. Um. So I don't see anybody else um raising their hand or um asking a question, so I just want to give you a chance, um. If not, Gaston, I might come back to you just for a closing comment if you had one. Um, before we sign off, um, so there's nobody asking to ask a question. So Gaston, would you like to, and then Anya, I'll give you the very, very final word, but Gaston, did you have any final comments and reflections? No, OK, thank you very much for, for giving me the opportunity to, to join this, uh, this meeting. As I said, uh, I'm amazed to see how relevant, uh, those, those discussions are really for, for us on the ground. And to go a little bit further, I would even start asking myself, how can we really get a better linkage between deck and, and operations because Uh, yes, it's good to, to, to, to, to be connected at the end of the, the research to get the results, our findings and to discuss them, but to what extent can we maybe have more, I wouldn't say involvement because it may be too much, but at least, certainly there was some kind of communication or things with the teams in, in, in the country, for example, in Mali. But I would love to see the money team join this kind of discussion. So, uh, uh, it doesn't look like there was many of them around. Well, it's very critical and very important for, for, for, for them. So, how can we try to, to improve this kind of, uh, communication so that we have the team, particularly those who are sitting in those countries where the studies are, are done so that they can be part of, of the process. So, just a comment that, uh, I would like to make. Again, thank you very much for, uh, for this opportunity. I, I do, I do appreciate your, your conclusion when you, you mentioned the need for, for more research and the issue of measurements. And again, we can, even if it's, if it's a, a, a small scale studies, we can already talk about policy implication because honestly, I do believe that the numbers, uh, speaks quite loudly for us and for teams on the ground. Because those kinds of references are very useful for our papers, for our documents, for our policy discussions, etc. So yes, this is certainly part of the question that we can start addressing even with a small study. So what can be the policy implication. With this kind of information that we have gathered, so this is something I think we may also add it and try to discuss it as well. So that's what I would like to add again. Thank you for inviting me. It's a real pleasure and I really appreciate it, and we'll get together for the presentation to my unit. Thank you very much. Back to you. Thanks, Gaston. Anya, any last final words? Um, yeah, just to say thank you so much, Gaston. I really appreciate it. Um, and I agree with you there is, um, there's both a need to communicate these results further and also to, um, to, to think about working together to, to design additional studies and, and additional research that really serves operational purpose. Uh, I do wanna say that I conducted all of these studies before I actually came to the World Bank. Um, so all the data collection was done before I was, uh, uh, a member of DCRG. I was a professor at Brown University before and that's where all this research programs started, but I'm really excited to get more involved in, in, in the work that's actually done here, um, and, and have the chance to talk more about these, these open policy questions and, um, and hopefully, hopefully find solutions as well and not just point out the many, many issues we've seen. Thank you. Thanks, Anya. No, that's exactly, I think that's, that's exactly what this kind of session is supposed to try to do is to create these linkages that then we can follow up, um, follow up with you can follow up with others too. So with that, um, I'm gonna close the session. Thank you, Anya, for a fantastic presentation, really insightful data and analysis, and, and, and a lot of questions and some, some solutions as you, as you, as you drew out in the, in the Q&A, uh, but also a lot of food for thought. And then, um, thank you, Gaston as well for, for, for really interesting, um, reflections and reactions and great to you, you know, Take it, we'll take you up on your invitation to come and join and, and work proactively with, with your teams. And then thank you all of you for joining and for the really interesting questions. Uh, Ryan has just put in the, in the chat, um, a link with all the, um, the materials, the presentation, the video will be posted there afterwards. Uh, you can share with others who weren't able to join today. So once again, thank you, everybody, and, uh, congratulations and have a good rest of your day. Bye. Thank you very much. Bye-bye. Congratulations again. Thanks. Thank you very much. It was a great pleasure pleasure. Thank you. Bye-bye. Bye-bye.

showAllTranscripts

no

duration

PT1H23M46S

scene7File

worldbank/MarginsOveruseUnderuseHealthcare

scene7Domain

scene7FileAvs

worldbank/MarginsOveruseUnderuseHealthcare-AVS

title

MarginsOveruseUnderuseHealthcare

description

MarginsOveruseUnderuseHealthcare

showTimestampAndTranscript

yes