1. AIDS treatment and the health
workforce crisis in Africa
Task shifting and quality of
care in Mozambique
Kenneth Sherr, MPH, PhDc
Technical Advisor, Health Alliance International
ksherr@u.washington.edu
April 29, 2009
2. Presentation overview
Introduction to „know-do‟ gap and
implementation science
Example of research to impact workforce
policy and planning
– Task shifting and quality of HIV care
Eye towards the future: How can research
strengthen Primary Health Care?
3. „Know-do‟ gap (1)
Advancements in medical science have far
outstripped their application
>10 million annual deaths from diseases
with proven, low cost prevention or
treatment strategies
1 million malaria deaths
6 million preventable child deaths
½ million maternal deaths
3 million HIV-related deaths
4. „Know-do‟ gap (2)
Consider HIV
– Unprecedented financial and political
commitment
> $8 billion spent on HIV programs per year
– 10-fold increase in number on ART from
2001-2007 (to 3 million)
Still only 30% of need
Mortality 28% higher at 6 & 12 months in low-
income vs. high-income countries
– Promising tools (male circumcision, microbicides),
can they be implemented effectively?
5. Translating science into improved
health: where are bottlenecks?
Improved
Delivery
Discovery Development Health
Outcomes
How to move beyond the “Black Box” of delivery?
6. Implementation science (1)
Research that addresses the know-do gap
Defining element is agenda, not
methodology
– Basic Science: What is the pathophysiology?
– Clinical Science: What is the appropriate diagnosis
and intervention?
– Evaluation Science: Does the intervention and
delivery model work in a specific setting?
– Implementation Science: How to best deliver and
scale-up interventions? How to strengthen health
systems?
7. Implementation science (2):
Framework
Economics
Health
Systems Anthropology
Research
Health
Sociology Medicine
Systems
Delivery Operations
Management
Research
Science
Systems Quality
Design Improvement
Adapted from: Kim, J, “Bridging the Implementation Gap in Global Health”, 2009
9. Workforce and the know-do gap
Chronic shortage of trained health workers
– Deficit of 2.4 million physicians, nurses and
midwives
– Workforce expansion of nearly 2.5 times
required to meet MDG goals
– ART expansion highlights weaknesses
Area of research to understand local
dynamics and evaluate solutions
10. Workforce in selected countries
Country Doctors Nurses
(per 100,000) (per 100,000)
Malawi 2 59
Mozambique 3 21
Uganda 8 61
Kenya 14 114
WHO Standard 20 100
South Africa 77 408
Brazil 115 384
USA 256 937
Cuba 591 744
Source: World Health Report, 2006
11. Workforce solutions
Long term: Treat, Train, Retain (TTR)
– Treat: HIV prevention, care and ART for health
workers
– Train: Pre-service and in-service courses
– Retain: Monetary incentives and improved working
conditions
Interim: Task shifting
12. Task shifting: Background
Long history in Africa
– At least 25 countries in SSA have a cadre of
non-physician clinicians (NPC)
– Expanded broadly between 1975-85 with
PHC
Recent concerns about task shifting for
clinical HIV care
14. Task shifting: Advantages (2)
Number of health facilities with ART in Mozambique:
2004-2007
250
Majority of 211
NPC trainings 193
completed
200
155
150
100
47
38
50
29
24
13 13
0
Jan 2004 Jun 2004 Jan 2005 Jun 2005 Jan 2006 Jun 2006 Jan 2007 Jun 2007 Dez 2007
MOH, 2007
15. Task Shifting: Uncertainties
Quality
Cost-effectiveness
Overload an already overburdened staff
“Research on the cost-effectiveness and care
outcomes of task shifting is needed to allow
decision makers to support such deployments.”
– Source: Samb, Celletti, Holloway, Van Damme, De Cock, Dybul. Rapid Expansion of the Health Workforce
in Response to the HIV Epidemic. NEJM 2007: 357; 24.
16. Study:
“Task shifting to mid-level clinical health providers:
an evaluation of quality of ART provided by non-
physician clinicians and physicians in
Mozambique”
Gimbel-Sherr K1,2, Augusto O4, Micek M1,2, Gimbel-Sherr
S1,2, Tomo MI3, Pfeiffer J1,2, Gloyd S1,2
1 University of Washington, Seattle
2 Health Alliance International
3 Ministry of Health, Mozambique
4 Eduardo Mondlane University, Mozambique
Supported by the Doris Duke Charitable Foundation’s Operations Research for
AIDS Care and Treatment in Africa (ORACTA) Initiative
17. Study Aims
1. Evaluate the quality of HIV care provided
by non-physician clinicians (NPCs)
compared with MDs
2. Identify provider-level factors that are
associated with quality of HIV care
18. Study methods (1)
Retrospective cohort study of
patients initiating ART during the
first 3.5 years of the national
ART program (7/04 – 11/07)
Study sites: 2 specialized
(vertical) HIV clinics in Central
Mozambique managed by MOH
– Vertical approach designed to
address high patient volume
and ensure supervision
– Standardized approach to HIV
care
– HIV prevalence > 25%
19. Study methods (2)
Data Sources:
– Routine clinic database
Includes clinical, laboratory, pharmacy, and social
worker visit data
Evaluated consistency against paper charts;
K>0.80 for key variables
– Interviews with clinic providers to gather
information on provider
characteristics, experience and knowledge of
MOH protocols
– Direct observation to determine provider time
in clinic over 4-week period
20. Study methods (3)
„Primary provider‟ defined as first clinical
provider at the clinic
Exclusion criteria – related to primary
provider:
– Children (<15 years)
– Women initiating ART during pregnancy
– Patients in MTCT-Plus
– Patients starting ART before July 2004
21. Study methods (4)
Outcomes:
– Process indicators reflecting country
protocols:
CD4 testing at 90-210 days post ART initiation
CD4 testing at 330-390 days post ART initiation
Frequency of clinical visit (at least 3 of 4 quarters
post ART initiation)
– Also assessed:
Adherence during first 6-months post ART initiation
(≥90% as optimal, based on pharmacy records)
Lost to follow-up & mortality (combined)
22. Study methods (5)
Data analysis:
– Multivariate generalized linear models
extended to the binomial family for
dichotomous outcomes
– Cox Proportional Hazards models for time to
event data
– All models account for provider-level
correlation and adjust for clinic
– Forward stepwise approach to identify patient-
level covariates for inclusion in final models
23. Study results (1)
Table 1: HIV clinic characteristics
Beira Chimoio
Mean monthly new ART initiation (>15 years age) 94 80
Mean patients enrolled in study per month 81 66
Mean number of clinical consults per month 1,110 551
Mean number of clinical consults per month with MD 505 (46%) 234 (43%)
Mean number of clinical consults per month with NPC 606 (54%) 317 (57%)
Observed staffing patterns
Observed MD FTE 1.3 0.5
Observed NPC FTE 1.4 2.5
Total 2.7 3.0
24. Results (2):
Table 2. Characteristics of study providers
MD NPC
N (%) N (%) p
Training detail
NPC NA 15 (100)
General MD 20 (56) NA
Specialized MD 16 (44) NA
Provider sex
Male 27 (75) 12 (80)
Female 9 (25) 3 (20) 0.70
HIV knowledge score 16.4 (82) 16.3 (82) 0.83
N (SD) N (SD) p
Provider age 40.2 (5.6) 38.8 (13.8) 0.71
Days of HIV-related training 38.8 (39.3) 21.5 (11.2) <0.01
Years of experience 12.7 (6.1) 11.6 (11.4) 0.72
Mean number of HIV consults per provider 745.6 (776.1) 3,233.9 (3,325.8) <0.01
25. Results (3)
Table 3. Patient characteristics by provider type
MD NPC
N (%) N (%) p
Study participants 1,799 (30.5) 4,093 (69.5)
Study Clinic
Chimoio 981 (54.5) 1,671 (40.8)
Beira 818 (45.5) 2,422 (59.2) <0.01
Sex
Male 808 (44.9) 1,800 (44.0)
Female 991 (55.1) 2,293 (56.0) 0.51
Distance of Residence from Clinic
<5 km 1,175 (65.3) 2,495 (61.0) -ref-
5-10 km 391 (21.7) 1,175 (28.7) <0.01
>10 km 233 (13.0) 423 (10.3) 0.089
Mean (SD) Mean (SD)
CD4 at enrollment 156.5 (115.6) 151.9 (113.9) 0.16
Age 36.1 (9.9) 35.9 (9.8) 0.44
Years of education 6.7 (3.5) 6.3 (3.4) <0.01
26. Results (4)
Table 4: Primary outcomes by provider type
MD NPC
N (%) N (%) ARR* (95%CI)
CD4 90-210 days post ART-initiation 496 (37.5) 1,210 (44.0) 1.13 (1.01, 1.27)
CD 330-390 days post ART-initiation 198 (18.7) 438 (21.2) 1.12 (0.95, 1.33)
Clinician visit 3 of 4 quarters post ART 926 (87.6) 1,836 (88.7) 1.02 (0.99, 1.05)
Optimal 6-month adherence 986 (74.5) 2,123 (77.3) 1.06 (1.01, 1.10)
(≥90% ARV pickup)
Death/loss to follow-up 504 (28.0) 1,005 (24.6) 0.89 (0.78, 1.02)
*Adjusted for clinic, years of patient education, provider-level correlation
27. Results (5)
Table 5: Study outcomes and provider characteristics
Optimal 6-month
CD4 90-210 days CD4 330-390 days Frequency of clinician Death/loss to
adherence¥
post ART initiation* post ART initiation* visits** follow-up
RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI)
Provider sex (ref=male) 0.62 (0.47, 0.81) 0.67 (0.44, 1.02) 1.03 (0.96, 1.11) 1.12 (1.06, 1.18) 1.11 (0.87, 1.42)
HIV knowledge score 1.00 (0.99, 1.01) 0.99 (0.97, 1.01) 1.00 (0.99, 1.00) 1.00 (0.99, 1.00) 0.98 (0.97, 0.99)
Days of HIV training 0.99 (0.99, 1.00) 1.00 (0.99, 1.01) 1.00 (0.99, 1.00) 1.00 (0.99, 1.00) 0.99 (0.99, 1.00)
Years of Service 1.00 (0.99, 1.01) 1.01 (1.00, 1.02) 1.00 (0.99, 1.00) 1.00 (0.99, 1.00) 1.00 (0.99, 1.01)
Cadre
NPC (ref) -ref- -ref- -ref- -ref- -ref-
General MD 0.93 (0.63, 1.37) 0.65 (0.38, 1.12) 1.06 (0.95, 1.18) 0.90 (0.81, 1.00) 1.68 (1.09, 2.59)
Specialized MD 0.84 (0.75, 0.95) 0.70 (0.61, 0.80) 1.01 (0.97, 1.05) 0.96 (0.92, 0.99) 1.31 (1.09, 1.58)
Total no. HIV consults 1.00 (0.99, 1.00) 1.00 (0.99, 1.00) 1.00 (0.99, 1.00) 0.99 (0.99, 1.00) 1.00 (0.99, 1.00)
*Adjusted for clinic, years of patient education, baseline CD4, provider-level correlation
**Adjusted for clinic, years of patient education, baseline CD4, provider-level correlation
¥
Adjusted for clinic, years of patient education, baseline CD4, SES
28. Discussion (1)
NPCs are important drivers for ART expansion in the
study clinics
Measures of service quality for NPCs were equivalent to
or better to MDs
Inconsistent associations between provider-level
characteristics and service quality
– Provider cadre, sex, and HIV knowledge score associated with
quality of care measures
– No association for days of in-service HIV training, years of
service and experience with HIV patients
29. Discussion (2)
Study limitations
– Switching providers may lead to misclassification of
provider type (23% of patients with multiple providers)
– Unable to account for all patient and clinic-level
characteristics
– Additional indicators of quality not measured
– Generalizability
30. Discussion (3)
Nevertheless…First study to compare quality of
HIV care between NPCs and MDs in
Mozambique
Implications for implementation
Task shifting can expand access with existing
resources
Augurs for training more NPC cadres
Gaps in outcomes identify areas for improvement
System-level interventions
Better training
Development of on-the-job support mechanisms
31. Acknowledgements
Patients at the Beira Central
Hospital and Chimoio
Provincial Hospital
Study providers, MOH
managers and co-
investigators
Doris Duke Charitable
Foundation Operations
Research for AIDS Care
and Treatment in Africa
(ORACTA)
32. Future directions
Ongoing research to strengthen integrated
Primary Health Care
MOH/HAI/UW Operations Research Center in
Sofala, Mozambique
7-year project funded by the Doris Duke
Charitable Foundation‟s African Health Initiative
– Collaboration between:
Ministry of Health
University of Washington DGH, Industrial Engineering, School of
Business
Eduardo Mondlane University School of Medicine
Health Alliance International
33. Duke project: Background
MOH decentralization to district level
management faces multiple hurdles
– Fragmentation
– Underdeveloped management capacity
– Weak data systems
– Lack of resources
34. Duke project: Objectives
Improve health outcomes through stronger
and integrated Primary Health Care sub-
systems
Objectives:
– Improve management capacity
– Strengthen data system quality and use
– Budget support for bottlenecks
– Focused research & program evaluation
35. Duke project: Relationship with
health system
Programmatic Operational
Directorate of Health Promo. & Dis. Control
Directorate of Medical Care
Clinical Care
National Planning & Admin & Finance Human Resources
Comm
Disease
Reprod.
Laboratory Pharmacy
& Mgmnt (HIV) Coop.
Mobilization
Control
Health
(Malaria, TB)
Prov. Medical Officer
Province Planning & Admin & Finance Human Resources
Dept. of Community Health Planning &
Clinical Nursing Coop.
Statistics
Care
District
Management Team
Director
Medical Pharmacist Statistician Administrator
Chief
Facilities
Ambulatory Inpatient Surgery Antenatal WCC Health Ed/
Care Care Care Outreach
Program Outputs
Health Outcomes
36. Conclusion
Approach to overcoming know-do gap
– Focus on data quality and use
– Multiple research approaches
– Institutional collaborations
– Engagement with health service management
Stay tuned!