19
Influencing Behavior of Providers and Pa4ents: The MicroFinancial Environment Jessica L. Cohen Harvard School of Public Health & Brookings Ins4tu4on October 20, 2011

cohen_oct202011

Embed Size (px)

DESCRIPTION

http://cddep.org/sites/cddep.org/files/cohen_oct202011.pdf

Citation preview

Influencing  Behavior  of  Providers  and  Pa4ents:  The  Micro-­‐Financial  

Environment  

Jessica  L.  Cohen  Harvard  School  of  Public  Health  &  Brookings  Ins4tu4on  

October  20,  2011  

The  Pa4ent:    Incen4ves  to  Use  RDTs  

Ø  An  RDT  is  a  “consump4on  good”:    consume  informa4on  about  true  cause  of  illness  

 

Ø  “Value  of  informa4on”  =  willingness  to  pay    for  info,  prior  to  decision    

Ø  Value  =  Benefit  –  Cost      

Ø  Direct  and  indirect  costs  of  purchasing  informa4on.      

Ø  No  direct  benefits  to  RDTs  per  se—benefit  is  the  aversion  of  costs  –  One  benefit  is  poten4ally  avoiding    cost  of  unnecessary  an4malarial.  –  Another  is  avoiding  an  illness  that  progresses  in  severity    

Ø  Pa4ents  who  want  informa4on  on  true  cause  of  illness  have  op/ons,  of  which  RDTs  are  only  one,  each  with  an  associated  cost  and  benefit  –  Presump4ve  treatment  of  malaria  is  one  way  of  purchasing  info  –  “Wait  and  see”  is  a  form  of  informa4on  acquisi4on  too  

Condi4ons  for  Pa4ents  to  Want  RDTs  I.  There  is  Some  Value  to  the  Informa4on  

Ø  RDTs  only  valuable  in  condi4ons  of  uncertainty    

Ø  Value  of  informa4on  is  low  when:    1)  People  are  sure  they  know  when  symptoms  equal  malaria  or  symptoms  do  not  equal  malaria.    2)  People  don’t  believe  test  is  accurate.  Or  at  least  more  accurate  than  subjec4ve  assessment  of  probe(malaria|symptoms).  

 

Ø  The  RDT  most  valuable  (i.e.  WTP  highest)  when  least  certain  in  own  assessment  and  most  certain  about  accuracy  of  test.    

Cost:      

Ø  Price  of  the  RDT    Ø  Opportunity  cost  of  4me  Ø  Non-­‐economic  (e.g.  psychological,  sociological)—fear,  reluctance  to  try  new  

technology,  etc.      

Benefit:    

Benefit  is  in  aversion  of  costs  incurred  if  don’t  take  RDT.    Directly  linked  to  what  alterna/ves  people  have.    

E.g.,  if  relevant  alterna4ve  is  presump4ve  treatment,  poten4al  benefits  are:    

Ø  Avoiding  the  cost  of  the  an4malarial  Ø  Avoiding  mis-­‐diagnosis  (poten4ally  lead  to  worse  illness)  Ø  Non-­‐economic  (e.g.  psychological)—uncertainty  avoidance    

Condi4ons  for  Pa4ents  to  Want  RDTs  II.  Benefit  of  the  Informa4on  Exceeds  its  Cost  

Costs  are  lowered  when:    

Ø  RDTs  subsidized  and/or  measures  taken  to  reduce  mark-­‐ups  

Ø  RDTs  are  easily  accessible  &  available  (OTC,  provided  by  CHWs,  etc.)  

Ø  They  become  familiar,  there  is  learning-­‐by-­‐doing  or  learning  from  peers    

Benefits  are  increased  when:    

Ø  An4malarials  are  expensive  (or  have  side  effects)  

Ø  Public  facili4es  are  far,  inconvenient,  not  well-­‐stocked,  etc.    Ø  Probability  it  is  not  malaria  is  high  &  consequences  of  delay  in  trea4ng  

not-­‐malarial  illness  is  high  (e.g.  for  pneumonia  in  children)  

Ø  Alterna4ve  types  of  learning  unreliable—e.g.  microscopy  high  error  rate  

Condi4ons  for  Pa4ents  to  Want  RDTs  II.  Benefit  of  the  Informa4on  Exceeds  its  Cost  

Op4on  1:  Do  Nothing/Wait  Costs:  Zero  today.  Possibly  high  future  costs  if  illness  gets  worse.    

Benefits:  Avoid  the  cost  of  medicine  and  opportunity  cost  of  seeking  care.    

Op4on  2:  Treat  Presump4vely  with  OTC  An4-­‐malarial  Costs:  Direct  cost  of  meds.  Possibly  high  future  costs  if  not  malaria  &  gets  worse.      Benefits:  Avoid  the  cost  of  seeking  care  at  facility  &  possibility  of  missed  malaria.    

Op4on  3:  Buy  an  RDT  at  Drug  Shop  Costs:  Direct  cost  of  RDT  &  of  meds  if  test  posi4ve.  Possibly  high  future  cost  if  RDT  is  

nega4ve  and  true  illness  gets  worse.    Benefits:  Avoid  the  cost  of  seeking  care  at  facility  &  possibility  of  missed  malaria.    

Op4on  4:  Go  to  Health  Facility  Costs:  Direct  costs  of  health  facility  visit  (depends  on  context;  usually  an4malarials  

free).  Direct  &  indirect  costs  of  travel  &  wai4ng,  stockouts.    Benefits:  Possibly  most  accurate  diagnosis  (including  assessment  &  treatment  of  

alterna4ve  illness)—avoid  missed  illness.  Avoid  costs  of  an4malarial.    

Condi4ons  for  Pa4ents  to  Want  RDTs  III.  Net  Benefit  of  RDTs  Higher  than  Alterna4ves  

Cohen,  Dupas  &  Schaner  (2011);  Western  Kenya    

•  Pre-­‐AMFm,  gave  households  cards  allowing  purchase  of  subsidized  ACTs  &  RDTs  in  local  drug  shops  

•  Random  varia4on  in  price  of  ACTs  &  RDTs;  RDTs  varied  from  $0  -­‐  $0.20  •  Measured  WTP  &  impact  on  treatment  seeking    

Some  basic  results:    

Ø  Doubled  rate  at  which  illnesses  diagnosed  (from  22%  to  44%)    Ø  Did  not  crowd  out  diagnosis  (microscopy)  at  health  center  Ø  Good  distribu4onal  effects  :  illiterate:  14%-­‐>  31%;  literate:  27%  -­‐>  52%  Ø  RDT  seeking  increasing  with  “predicted  posi4vity”  (Figure  )  

Ø  80%  of  the  4me,  when  people  bought  ACTs,  bought  RDT  also.  Willing  to  pay  (at  these  low  prices).  

Ø  People  no  more  likely  to  use  RDT  when  ACT  price  was  higher.    Ø  99%  of  RDT-­‐posi4ve  bought  ACT;  60%  of  RDT-­‐nega4ve  bought  ACT  

Pa4ents:  Evidence  

Cohen  (2011);  Central  Uganda      

•  Pre-­‐AMFm.  Subsidized  ACTs  offered  for  sale  with  varying    packaging/  messaging  to  encourage  adherence  

•  RDTs  randomly  offered  to  some  pa4ents  when  arrived  at  shops  to  buy  ACT  •  Measuring  impact  of  packaging  and  RDT  on  adherence  to  ACTs  •  Measure  beliefs  about  whether  the  illness  was  malaria  prior  to  doing  RDT    

Some  basic  results:  Ø  There  is  significant  uncertainty  about  the  likelihood  of  malaria  (Fig1)  Ø  People  are  decent  guessers  (but  not  perfect)  about  chances  (Fig2)  Ø  Over-­‐treatment  much  more  likely  for  adults  (Fig  3)  Ø  RDT-­‐posi4ve  25%  more  likely  to  adhere,  but  not  necessarily  the  effect  of  RDTs    

Pa4ents:  Evidence  

Cohen,  Dupas,  Schaner  (on-­‐going);  Western  Kenya  •           Ask  about  chances  the  illness  is  malaria  before  and  ater  RDT.  Ø       People  revise  upward  when  test  posi4ve  (but  not  to  1),  revise  downward  when  test  nega4ve  (but  not  to  0);  suggests  some  belief  in  test  but  not  perfect  

Ø  Commonly  heard  that  shops  wouldn’t  want  to  sell  RDTs.  Why?  –  Well-­‐known  that  people  “over-­‐treat”  malaria  –  Shops  could  lose  money  on  these  unnecessary  an4malarial  purchases    

Ø  True,  but  they  can  make  addi4onal  profits  from:  –  Selling  the  test  –  Selling  alterna4ve  treatment  when  test  is  nega4ve  –  Selling  the  an4malarial  at  a  higher  price  (WTP  could  increase  w/  certainty)    

Ø  As  profits  from  these  products  increase,  drug  shops  become  more  willing  to  sell  RDTs  

 

Ø  Profit  will  depend  on  what  they  can  charge,  which  in  turn  depends  on  the  things  highlighted  on  the  part  of  consumers  –  E.g.  price  of  RDT  can  be  higher  as  belief  in  test  increases    

Ø  Will  depend  on  market  condi4ons  (compe44on)  as  well  as  things  like  value  of  reputa4on  (avrac4ng  repeat  customers).  

Providers:    Incen4ves  to  Sell  RDTs;  Components  of  RDT  Prices  

Externali4es  from  over-­‐treatment  

or   Consumer  probability  belief  >  than  shop  owner’s  

or   Subsidies  for  an4malarials  

Shop  under-­‐provides  RDTs  

No  externali4es  to  over-­‐treatment  

&  Shops  &  customers  have  accurate  beliefs  about  malaria  probability  

Sell  RDTs  under  all  condi4ons  we  would  want  them  to  (max  welfare)  

Providers:  Evidence  I.  Cohen  &  Dickens  (2011):  Economic  Model  of  Shop  Behavior  

Mechanisms  to  increase  RDT  provision:  1.  BCC  targe4ng  beliefs  about  malaria  prevalence  &  accuracy  of  test  2.  Increase  compe44on  between  shops  3.  Introduce  RDT  subsidy  Op4mal  RDT  Subsidy  =    (Rate  of  overtreatment)*[(Subsidy  for  an4malarial)  +  (Social  Cost  of  Over-­‐treatment)]  

Some  results  so  far:    

Ø  RDTs  are  converging  in  price  to  around  $1  (100%  mark-­‐up)  Ø  Rural  shops  more  likely  to  sell/promote  the  RDTs  than  “urban”  ones  (stated  

reason:  proximity  of  urban  shops  to  health  centers)  Ø  RDTs  more  oten  purchased  for  children  than  adults    Ø  Shops  are  crea4ve  in  promo4ng  the  RDTs  (for  example…)  Ø  RDTs  maintaining  quality  (based  on  lot  tes4ng  ater  4me  in  field)  Ø  Shops  easily  trained  for  the  most  part  &  following  protocol  Ø  Some  crea4ve  shops  are  offering  a  “bundle  price”  for  the  RDT  and  treatment  

Providers:  Evidence  

Cohen,  Fink,  Dickens  (Uganda;  Ongoing)      

•   Sell  subsidized  RDTs  ($0.50)  through  wholesaler  to  shops  in  8  districts.  •   Implement  BCC  campaign  (with  Uganda  Health  Marke4ng  Group)  to  test  key  messages  on  RDT  use  and  adherence  to  test  results  

 

•  Not  really  enough  to  know  willingness  to  pay  for  RDTs.  Why?    Ø  This  is  likely  to  change  over  4me  as  people  revise  expecta4ons  about  the  

prob(malaria|symptoms)  and  the  belief  in  the  test.    Ø  Very  high  rates  of  over-­‐treatment  &  ignoring  test  results  could  be  “short-­‐

run”  •  What  are  the  key  behavioral  messages  to  encourage  diagnosis  and  

adherence  to  test  results?  This  could  speed  up  acceptance  of  RDTs.  •  Cri4cal  need  for  bever  evidence  on  how  treatment  seeking  varies  with  

underlying  endemicity    –  For  example,  oten  assume  RDTs  more  cost-­‐effec4ve  in  low  endemicity  

se|ngs,  but  this  is  only  if  people  are  really  bad  at  guessing.    •  How  do  RDTs  affect  adherence  to  ACTs?    

–  Important  for  cost-­‐effec4veness  calcula4ons  which  are  majorly  determined  by  probability  of  resistance  

•  What  determines  shops’  RDT  pricing  decisions?  Are  there  any  policy  levers  we  can  pull  to  influence  them?  

•  Are  there  technological  improvements  in  RDTs  that  bever  fit  consumer/provider  beliefs/incen4ves  ?  

Summing  Up:  Some  Key  Missing  Pieces  

.4.6

.81

lpol

y sm

ooth

: J5:

wen

t to

chem

ist

0 .2 .4 .6 .8 1

No Subsidy ACT Subsidy

0.2

.4.6

Sha

re

0 .2 .4 .6 .8 1Predicted Positivity

A. All

0.2

.4.6

Sha

re

0 .2 .4 .6 .8 1Predicted Positivity

B. Illiterate Head

0.2

.4.6

Sha

re

0 .2 .4 .6 .8 1Predicted Positivity

C. Literate Head

Illness was treated with ACT

Back  

Cohen,  Dupas,  Schaner  2011.  (Kenya)  

.2.4

.6.8

1

lpol

y sm

ooth

: J5:

wen

t to

chem

ist

0 .2 .4 .6 .8 1

ACT Subsidy Only ACT + RDT Subsidy

0.2

.4.6

Sha

re

0 .2 .4 .6 .8 1Predicted Positivity

A. All

0.2

.4.6

Sha

re

0 .2 .4 .6 .8 1Predicted Positivity

B. Illiterate Head

0.2

.4.6

Sha

re

0 .2 .4 .6 .8 1Predicted Positivity

C. Literate Head

Took Malaria Test

Back  

Cohen,  Dupas,  Schaner  2011.  (Kenya)  

Cohen  (2011);  Central  Uganda  

05

1015

20P

erce

nt

0 2 4 6 8 10Chances (1 - 10) that you have malaria

Histogram of Beliefs About Likelihood Patient Has MalariaHistogram  of  Beliefs  that  Pa4ent  Has  Malaria    

Back  

010

2030

0 5 10 0 5 10

Negative Positive

Per

cent

Chances (1 - 10) that you have malariaGraphs by RDT Positive

Histogram  of  Beliefs  that  Pa4ent  Has  Malaria  by  RDT  Result  

Back   Cohen  (2011);  Central  Uganda  

0.2

.4.6

.81

Sha

re M

alar

ia P

ositi

ve

0 20 40 60 80Age

kernel = epanechnikov, degree = 0, bandwidth = 4.08

Actual  Posi4vity  Among  Subsidized  ACT  Buyers  Local  Linear  Regressions  of  Posi4vity  on  Age  

Cohen,  Dupas  and  Schaner  (2011);  Western  Kenya  

Cohen  (2011);    Central  Uganda  

Back  

Back  

Back