Improving Patient Access to Malaria and other - …siteresources.worldbank.org/INTTRANSPORT/Resources/336291... · Improving Patient Access to Malaria and other Essential Medicines

Improving Patient Access

to Malaria and other

Essential Medicines in

Zambia Results of a Pilot Project

Prepared for Transport Week Learning Event

Jed Friedman, DECPIwith Monique Vledder, Mirja Sjoblom,

Prashant Yadav, and others

Why was the pilot designed? Drugs are often unavailable in health facilities

• A 2006 survey shows that the key drug (ACTs) for malaria treatment were not available at time of visit in:

– 44% of urban facilities

– 29% of rural facilities

– 15% of hospitals

• Stock-outs on district level are less common indicating that distribution from districts to health facilities is the main bottleneck in the system

Stockout bottlenecks often lie between the district store and the facility

0% 10% 20% 30% 40% 50% 60%

ACT Pediatric (malaria treatment)

ACT Adult (malaria treatment)

Amoxicillin (Antibiotics)

CTX (Antibiotics)

DepoProvera (Contraceptive)

SP (malaria prevention)

DHO

Health Facility

Why is it difficult to deliver drugs?

Improving access to essential medicines in Zambia

Objective

Identify a cost-effective way to improve the availability of drugs through strengthening of the supply chain from center to

districts and health facilities

Approach

The pilot compares the effectiveness of two different supply chain interventions to select one (or a combination/variation) that

can be rolled-out nationally

Two interventions tested

System A:

• Health centers (HCs) place orders to District Health Officer (DHO) who sends aggregated monthly orders to central stores (MSL)

• DHOs store commodities and supply HCs monthly

• Districts are responsible for assembling orders for the HCs and coordinating delivery between the district and HCs

• District logistic oversight conducted by new cadre, Commodity Planner (CP)

System B:

• HCs place orders directly to MSL

• Orders are packed at MSL in sealed packages for each individual facility

• Districts only responsible for coordinating delivery or pick up of orders between the district and HCs, facilitated by CP

Pilot evaluation design

• Districts stratified and randomly selected from 52 peri-urban and rural districts in Zambia

• Total of 24 districts randomly assigned: 8 districts for system A, 8 districts for system B and 8 comparison districts

• Pilot implementation for a one-year period

• Baseline data collected from 250 facilities in Dec-Jan 2008/09 and follow-up data during the same period in 2009/10

• Inventory and stock-out rates of tracer drugs measured at both baseline and endline

• Supplementary information – stocking history, storage conditions – also collected at endline

District Selection

Reduced stockoutsin A system

Comparison of baseline and endline values in A districts

*the reduction in stockout rate is statistically significant with respect to any observed change in control districts

0% 10% 20% 30% 40% 50% 60% 70% 80%

ACT Pediatric (malaria treatment)

ACT Adult* (malaria treatment)

Amoxicillin (Antibiotics)

CTX* (Antibiotics)

DepoProvera* (Contraceptive)

SP (malaria prevention)

A baseline

A endline

Dramatically reduced stockouts in B system

Comparison of baseline and endline values in B districts

*the reduction in stockout rate is statistically significant with respect to any observed change in control districts

0% 10% 20% 30% 40% 50% 60% 70% 80%

ACT Pediatric* (malaria treatment)

ACT Adult* (malaria treatment)

Amoxicillin* (Antibiotics)

CTX* (Antibiotics)

DepoProvera* (Contraceptive)

SP* (malaria prevention)

B baseline

B endline

More people get their lifesaving drugs in B districts

Number of days of stockouts for the last quarter of 2009

0

5

10

15

20

25

30

35

40

ACT Pediatric (malaria

treatment)

ACT Adult (malaria

treatment)

Amoxicillin (Antibiotics)

CTX (Antibiotics) DepoProvera (Contraceptive)

SP (malaria prevention)

comparison districts

A districts

B districts

Summary of results

• System B performs significantly better than system A and comparison districts in terms of availability

• There is a significant and large decrease in number of days of stockouts in B districts compared to control

• Hence unmet demand is significantly lower in B districts compared to A and control

• Reporting rates from district health offices to MSL have increased during the pilot period for both A and B districts to nearly 100%

Expected impact on malaria mortality

If Model B were to be scaled up nationwide, projections indicate:

–16,600 U5 deaths due to malaria could be averted by 2015, as well as 2,200 adult deaths

–Child and adult mortality due to malaria could be reduced by 21% and 25% respectively

–These gains focus only on increased availability of malaria drugs however widespread gains likely from increased availability of all essential drugs

Program costs• The monthly recurrent costs for Model A is US$2832 per

district and for Model B it is US$3325/district.

– Pilot was implemented in remote districts with higher transportation costs.

– Some net savings in B districts are not included

• National scale up of B would increase the supply chain operational cost from 4.1 percent to 8.5 percent of the total pharmaceutical budget

• Total current procurement budget for drugs (partners and MOH): approximately US$100 million/year

Model B is 4 times as cost-effective as Model A

• Cost per day of essential medicine stock-out averted:

– Model A reduces stock-out day of one tracer drug at a cost of $14.5 in additional operating costs

– Model B achieves the same stock-out reduction at a cost of $4.2

• Focusing on possible malaria mortality averted: cost of $22 per YLL averted for a national scale-up of Version B over a 5 year period

• Compares favorably with many other public health interventions

What are the lessons for IE in infrastructure sectors?

• This intervention necessarily implemented at a highly aggregate level – the district

• Common situation for infrastructure projects

• Even though the unit of observation is the HC (stock-outs), potential for much observational dependency within district

• The pilot only had funds to test 2 interventions in 8 districts, risk of underpowered study if within-district outcomes are highly correlated

Lessons (cont)• From base-line data, power analysis suggested study can

identify standardized effect of .49

• Definition: standardized effect = (change in outcome of interest)/(standard deviation of outcome)

• Implication: study could not identify moderate gains that may have real health implications (i.e. reduction in stock-out rate to 20% from 40%)

• In this study model B is a smashing success: a standardized effect of .57 [Got Lucky!]

• however gains from A are often not statistically different from B or control

Lessons (cont)• Need to ensure an adequately powered study

• First best solution is sufficient study scope and funding to observe adequate number of treatment units in an RCT

• When not possible, what can be done to help improve power?

• Pair-matched randomization (Imai, King, and Nall, 2009)

• Quasi-experimental methods (i.e. propensity score matched difference-in-difference)

• However these methods usually come at a cost – stronger assumptions needed to identify causal impact

• No easy answer, but need to be aware of challenges at the start of project design stage

Additional slides

Another way to measure performance: Stockouts in Model B vs. Comparison Districts

0% 10% 20% 30% 40% 50% 60% 70% 80%

ACT Pediatric (malaria treatment)

ACT Adult (malaria treatment)

Amoxicillin (Antibiotics)

CTX (Antibiotics)

DepoProvera (Contraceptive)

SP (malaria prevention)

B districts

comparison districts

System ACommodity Planners

Pull system, monthly delivery

CP receives stock from MSL and manages district stock in district store room and process and packs orders from health facilities

Monthly Twice Monthly

Health facilities receive facility packages from CP

Health facilities place orders to CP

Medical Stores Limited

CP places orders to MSL

One pack per districts (for all health facilities) is compiled

Districts

Health Facilities with adequate storage

space

Health Facilities with limited storage

space

System BCommodity Planners + Sealed Packages

Pull system, monthly delivery

CP receives facility packages from MSL; No stock kept at District Store

Monthly Twice Monthly

Health facilities receive facility packages from CP

Health facilities place orders directly to MSL Districts

Medical Stores Limited

One customized pack for each health facility is compiled

Health Facilities with limited storage space

Health Facilities with adequate storage

space