Long Term Effects of Payment for Performance: evidence from Pwani, Tanzania Gemini Mtei 24 th November RBF a health systems perspective. White Sands Hotel, Dar es Salaam.
Rationale Global implementation of payment for performance (P4P) - currently operating in over 30 African countries P4P in Africa introduced to accelerate progress to MDGs4 and 5. Limited robust evidence of impact of P4P in low and middle income countries Focus on incentivised services Knowledge gaps Pathways: motivation, quality of care, user costs Unintended consequences Long term effects
P4P in Tanzania Aim: A pilot introduced in 2011 focusing on MCH service coverage to inform a national programme Location: Pwani region of Tanzania Implementers: MOHSW and CHAI Funder: Government of Norway
Scheme Design Facility level targets: ANC: IPT2; % HIV+ women on ART Institutional delivery rate % of newborns with OPV0 in first 2 weeks % infants with Penta 3; measles vaccine % of PNC visit w/n 7 days CYP HMIS reports correctly filled and submitted on time + use of partograms District regional level targets: % of maternal/perinatal deaths audited on time % of facilities with stock outs
Bonus Payouts made every 6 months (in USD) Total % staff % facility improvement Hospitals 6,790 60 (RCH) 30 (non-rch) 10 Upgraded health centres 4,380 75 25 Health centres 3,220 75 25 Dispensaries 820 75 25 District managers 2,800 100 0 Regional managers 3,000 100 0
P4P in Tanzania: a package of interventions Bonus Payments Financial autonomy Performance monitoring / supervision Improved HMIS
Introduction of P4P Training of key stakeholders and provision of guidelines, opening bank accounts Bonus payment to health workers Bonus payment to District and Regional managers Positive Effects Increased motivation of health workers and managers Negative Effects Damage intrinsic motivators Re-allocate resources Improve quality of care / increased patient satisfaction Reduced user charges Coercive strategies to increase utilisation Mis-reporting performance Increased utilisation of targeted health services Crowding out of non-targeted health services Reduced quality of care
Focus of presentation Effect on incentivised service use at the facility level Change pathways: Quality of care: adherence to clinical guidelines for ANC, provider kindness during delivery; drug stock out in past 90 days (IPT, oxytocin) Probability of paying out of pocket at public facilities Unintended consequences Non-targeted service use (outpatient care for under 5s; over 5s) Long term effects
Study Design Design: Controlled before and after study design 7 intervention districts 4 neighbouring control districts Comparable poverty, literacy, rate of institutional deliveries, IMR, pop. per health facility, no. of children < 1 yr Timing: -Baseline in January-February 2012 -Midline in March-April 2013 (13 months) short term effects -Endline in January-February 2015 (36 months) long term effects
7 P4P districts 4 districts with no P4P Only include facilities eligible for first cycle payment 150 health facilities, 75 in each arm incl. 6 hospitals 16 health centres 53 dispensaries 1 facility survey at each facility Non-targeted service users: Outpatient for malaria, diarrhoea, ARI 10 exit interviews with patients at each facility 20 interviews with women who delivered in past 12 months, from the catchment area of each facility Targeted service users: ANC, under 1 vaccinations/pn C
Impact on Targeted Services Variables Baseline Short term impact P4P Control Diff % effect of P4P Long term impact % effect of P4P ANC 2+ doses of anti-malarial (%) 49.5 56.7-7.2*** 10.3*** 5.3 Treated for HIV/AIDS (%) 7.8 6.8 1.0-0.3 4.0 Intra-partum care Delivery in a facility (%) 84.7 86.8-2.1 8.2*** 4.8** Postpartum care PNC < 7 days after birth (%) 21.6 16.9 4.7** 0.6 1.8 Family planning current (%) 37.1 39.8-2.7-0.9-1.3 OPV0 (%) 77.4 78.5-1.1 5.6* 4.1 DPT 3 (%) 76.4 79.9-3.5 2.4 4.7 Measles (%) 51.4 53.3-1.9 9.6-6.1
Pathways Variables Baseline Short term impact ANC P4P Control Diff % effect of P4P Long term impact % effect of P4P Any ANC (%) 97.1 99.9-2.8 3.3*** -0.2 Quality of ANC (index) 0.78 0.75 0.03** 0.00-0.01 Stock out of IPT (%) 27.1 17.8 9.3-10.0* -12.4 Pay for ANC (%) 8.1 7.5 0.6-2.7 2.8 Intrapartum care Staff kindness at delivery (index) Stock out of oxytocin (%) 7.2 7.6-0.4*** 0.49* 0.41** 42.9 18.1 25*** -40.0*** -30.6*** Pay for delivery (%) 16.5 11.9 4.6** -5.0** -10.3***
Unintended Consequences Variables Baseline Short term impact Outpatient < 5 years visits Outpatient < 5 years in dispensaries visits Outpatient > 5 years visits Outpatient > 5 years in dispensaries -visits P4P Contr ol Diff 223 9 193 7 30 2 164 8 172 6-7 8 359 5 287 3 72 2 P4P effect Long term impact P4P effect -41.1-1.3-57.5** -9.0-15.8 74.9 276 8 235 4 41 4-90.8*** -50.5
Evolving implementation status Funding context Norwegian funding came to an end in Dec 2013 Delays securing agreement for World Bank take over of scheme funding Implementation status Bonus payments for previous 12 months had not been received. Payment for the June-Dec 2013 cycle received in June 2014. Withdrawal of CHAI support on implementation since Dec 2013 Verification scaled down and no feedback sessions in past 12 months Attitudes towards the scheme Perceptions and acceptability of the scheme remains high Incentive effect maintained but resource effect
Summary Effect on deliveries and IPT in the short term. Delivery effect sustained in long term (size reduced) Delivery pathways: provider kindness, payment for delivery and availability of drugs - sustained over time Reduction in the probability of paying for delivery ANC pathways: Ability to procure IPT reduced over time; no effect on probability of paying Negative effects on non-targeted services at dispensaries in short term, disappears in long term
Conclusions Reduction in effect over time why? Does not appear to be due to reduced incentive effect...yet Delays in payments and less intensive support in the last 12 months of implementation affect ability to procure IPT reduced effect on non-targeted services Timing of evaluation important for complex interventions that are dynamic over time Importance of documenting implementation alongside outcome evaluation
Acknowledgements Josephine Borghi - LSHTM Peter Binyaruka - LSHTM Powell-Jackson T - LSHTM Patouillard E - LSHTM Torsvik G CMI Mayumana I IHI Masuma Mamdani - IHI Lange S - CMI Maestad O - CMI