CLICK TO ADD TITLE - Purdue Universityweb.ics.purdue.edu/~aiyer/PR1-3.pdf · There are few peer-reviewed papers… Daff BM, Seck C, Belkhayat H, Sutton P. Glob Health Sci Pract. 2014;2(2):245-252

CLICK TO ADD TITLE

[DATE][SPEAKERS NAMES]

Improving pharmaceutical supply chains in Africa through rigorous data-driven approach

Eric Mallard – World BankCopenhagen, Denmark

How can we address supply chain chronic issues in Africa?

How can we help governments make evidence-based decisions when reforming their supply systems?

What are recent successful supply chain interventions?

Where can we find evidence of what works and does not work?

2

There are few peer-reviewed papers…

Daff BM, Seck C, Belkhayat H, Sutton P. Glob Health Sci Pract. 2014;2(2):245-252 3

Percentage of Facilities Experiencing a Stockout in 2 Comparison Districts, Senegal, January–July 2012

Open database start emerging….

http://nphcda.thenewtechs.com/ 4

Average quality scores of health facilities in Adamawa state, Nigeria, 2012-2014

… and we know a few well-designed pilot projects

05

10152025303540

comparison districts A districts B districts

World Bank. 2012. World Bank policy note : enhancing public supply chain management in Zambia. 5

Average number of days of stock outs at health facilities, Zambia, Q4 2009

There are more “promising” than really “proven” practices in this field

6

Poor data management hinders evidence-based approaches

Limited evidence

Data scarcity Data fragmentation

Data ownership

Poor data use

Think beyond stock-out Integration Open data Data

analytics

7

Data collection and analysis is a prerequisite to fixing supply chains

Chronic supply chain dysfunctions are a complexproblem

Solving complex problems requires to testinterventions, measure the effects, and then learn andadapt

We need to collect more data and to make a better useof them

More rigorous evaluation of supply chain interventionswould greatly benefit the global community

There may be an opportunity to start with aretrospective meta-analysis focusing on pre-existingdata

8

Objectives of this Economic and Sector Work

Inform government and development communitypolicy and strategy on successful supply chaininterventions and reforms

Consolidate supply chain data across countriesand systems

Generate evidence on the performance ofpharmaceutical supply systems in Africa

Propose evidence-based solutions and newmodels

Stimulate public debate and disseminate bestpractices

9

Overall project timelines

2013 2014 2015Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

Partners engagement

Concept note and fundraising

Performance framework and design choices

Data mining and collection

In-depth analysis

Dissemination and policy dialogue

Study extensionProof of approach

Interim output

Final output 10

A data-driven approach

With the support from 11

Our achievements so far

Performance framework developed and vetted by the advisory committee 6 performance areas covering both outcomes and systems characteristics 12 KPIs identified and prioritized, alternative “back-up” indicators available

A total of 19 hypotheses on design interventions that influence performance of supply chains in Africa were identified Across the value chain: forecasting, procurement, storage & distribution, delivery Also including cross-cutting themes – particularly for financing

Full literature review Case studies and data sources mapping Performance framework and KPI comparison

A rigorous methodological approach has been developed Data collection, cleaning and codification process pattern recognition approach to cope with confounders and limited number of

cases

With the support from 12

Key performance areas through which to test supply chain design hypotheses were identified with an expert committee

Availability to patient

Quality

Affordability

Responsiveness

A

B

C

D

Cost-efficiencyE

AdaptabilityF

Out

com

esSy

stem

s C

hara

cter

istic

s

• Stock-out (or inverse availability) rate - SDP or peripheral facility level

• Order fill rate - CMS or central stocking level

• Supply chain procurement mark-up - % price mark-up at importation level - tariffs, etc.

• Supply system total mark-up to retail (from CIF to patient level) - e.g. unregulated in-country prices and variation at end-user pricing

• Existence of a pharmaceutical quality assurance system - (e.g. Laborex/IHS as ISO-certified wholesalers, GFATM VPP, WHO MQAS)

• Quality maintenance throughout the supply chain

• Ratio of supply management / distribution cost to commodity value• % of stock wasted incl. expired (at various levels)

• Average lead time from order to fill at SDP or dispensing facility level (days) - stock records

• % of (final dispensing point / SDP) facilities with buffer stock

• Ability to withstand shock such as market recall, demand spikes e.g. national days or % annual need buffer stock available

• Long-term sustainability - for example, ratio of commodities budget that country is able to self-finance (public or private funds)

6 indicators prioritized – to be

reviewed post-data mining to

affirm both availability of

such data and/or practicality of measurement

6 indicators suggested–

however, some are exploratory given limited

data collection. To be finalized

post-data mining

With the support from13

Maturity of forecasting techniques

Frequency of supply planning

…

Central vs decentralized procurement

Ordering frequency

Framework contracting

…

Outsourcing

Multiplicity & competition

Cross-docking

Integration

…

Licensing multiple small operators

Cash & carry vs credit systems

Community involvement

….

A number of research hypotheses have been defined throughout the supply chain

Government’s role (operator vs stewardship)

Performance-based financing

…

14With the support from

A theoretical performance framework is highly complex but can be simplified into quantitative ‘pattern recognition’

Inputs that directly impact upon supply management and performance but largely separate from policy

Inputs that reflect policy and/or technical design choices (whether deliberate or not) that directly drive supply management+

Availability of accurate and consistently

comparable data

Performance•Beneficiary level outcomes

•System level outputsor characteristics

Contextual factors Hypotheses Supply systems

Independent variable to control for (confounder)

Independent variableto test for

Dependent variableUnit of analysis

n independent supply chains/systems that deliver across …

•Commodities (What?)

•Geographies (Where?)

•Institutions (Who? and to whom?)

Outcomes{O(n)}• Availability

• Cost management

• Quality

Proxyoutputs{P(n)}• Responsiveness

• Cost efficiency

• Adaptability

Set of key design choice hypotheses{H},e.g.,

•H1: Outsourced private procurement vs. public tender vs. mixed system

•H2: Cross-docking or level-jumping

•H3, H4, H5,etc.

Set of environmental contextualfactors{CF},e.g.,

•CF1 = port access or none (landlocked)

•CF2 = per capita income

•CF3, CF4,etc.

1. We can exclude or control for confounders via data cleaning, sub-sampling, etc.

2. We initially run tests (e.g. R2) for pairs of (1) performance indicator matched to (1) hypothesis: Px(n) ∩ Hy(n)“Performance indicator x correlates with design choice y.”

Correlation vs. regression

15

With the support from

We need your support!

Everyone can contribute:•Countries•Donors and their implementing partners•Pharmaceutical wholesalers•Pharmaceutical companies•…

by:•Identifying case studies•Providing data•Championing the project in the respective countries

16

17