doi:10.1111/j.1365-3156.2008.02012.x
Systematic Review
Summary
objectives To (a) summarise opinion about what supervision of primary health care is by those
advocating it; (b) compare these features with reports describing supervision in practice; and (c) to
appraise the evidence of the effects of sector performance.
methods Systematic review. Reports were classified into three groups and summarised using appropriate methods: policy and opinion papers (narrative summary), descriptive studies (systematically
summarised) and experimental or quasi-experimental studies (design and outcomes systematically
summarised). Data presented as narrative summaries and tables.
results 74 reports were included. In eight policy and opinion papers, supervision was conceptualised as
the link between the district and the peripheral health staff; it is important in performance and staff
motivation; it often includes problem solving, reviewing records, and observing clinical practice; and is
usually undertaken by visiting the supervisees place of work. In 54 descriptive studies, the setting was the
primary health care (PHC) or specific services and programmes. Supervisor-supervisee dyads were
generally district personnel supervising health facilities or lay health workers. Supervision mostly meant
visiting supervisees, but also included meetings in the centre; it appeared to focus on administration and
checking, sometimes with checklists. Problem solving, feedback and clinical supervision, training and
consultation with the community were less commonly described in the descriptive studies. Supervision
appears expensive from studies that have reported costs. In 12 quasi-experimental trials, supervision
interventions generally showed small positive effects in some of the outcomes assessed. However, trial
quality was mixed, and outcomes varied greatly between studies.
conclusions Supervision is widely recommended, but is a complex intervention and implemented in
different ways. There is some evidence of benefit on health care performance, but the studies are
generally limited in the rigor and follow up is limited. Further research delineating what supervision
consists of and evaluating it in the context of unbiased comparisons would guide the implementation of
effective supervision as part of the management of PHC.
keywords supervision, developing countries, health services administration, health workers, training,
management, performance, quality of care
Introduction
Why should district health team members be motivated to commit themselves to the tedious task of
administrative supervision and consider this as a
normal part of their duties? (Meuwissen 2002).
Primary health care (PHC) emphasises universal access to
health services (WHO-UNICEF 1978a). The movement
was accompanied by an expansion of basic health services
in developing countries provided by paramedical and
community health workers (Cueto 2004). Inevitably some
health workers worked in isolation, in remote rural areas
Performance
Feedback
Problem
identification &
solving
PHC worker
Compare service
delivered against
norms
Activities e.g.
immunisation
Examine information
against expected
outputs
Outputs e.g.
vaccine
coverage
Health outcomes
e.g. mortality
Figure 1 Management framework for supervision.
370
371
372
Health
area
PHC
NA
Flahault et al.
(1980)
Gibson &
McClelland
(1990)
Larsen (1987)
PHC
Main focus
Fendall (1980)
Author year
NA
Peripheral units
Experienced doctor
(1) Establishment of
contacts; (2) review of
targets; (3) observation of
the health workers; (4)
identification of gaps; (5)
consultation with the
community; (6) reporting
to the health team. Follow-up and solutions of
problems
Regular
Six components:
and
communication, teaching,
frequent seeing patients, tour of the
health centre, assessing
performance, and follow
up. Assess performance,
follow-up training; communication and observation of health facilities.
Feedback provided.
Checklists
Visits
Control of staff, problem
solving, follow-up of
training, review of
information, management
of resources and supplies,
and community
involvement. Human
relationships. Incentives.
Community
Visits
Health Workers
Regular
Focus on evaluation,
schedule correction of errors,
reinforce understanding,
review on information
and collect options from
the community.
Recording grid
Visits and Problem solving, review of
meetings information. Also reports,
statistics, protocols,
circulars, courses.
Visits
Health centres.
District level
Anyone is
entitled
Village health
agent;
health teams
NA
Supervisee
Supervisor
Clinical supervision
and observation of
practices
Visits
(1) Level A
(community
level)
(2) Level B
NA
Nineteen studies mentioned the tools used for supervision, and this included recording grids, checklists, written
instructions, standardised forms, revision of achievements
and rapid assessment tools.
In 24 studies, the authors provided an opinion about the
effects of supervision on quality of health care or utilisation
of services: fifteen were positive about the effects and nine
reported poor implementation.
Eight studies reported on the costs of supervision, which
were estimated in a variety of ways. In half of the studies,
authors considered costs involved in supervision as
expensive (Grosskurth et al. 2000 notes that supervision
was the single most expensive component of recurrent
costs, for example).
A number of studies described barriers in implementing
effective supervision (Box 2).
Objective 3. Effectiveness of supervision
NA-not available.
Waterson (1982)
PHC
(1) Level B
(local council
sub areas)
(2) Level C (district)
Doctor in PHC
Author year
Table 1 (Continued)
Health
area
Main focus
Supervisor
Supervisee
Visits
Components
and tools
Percentage
54
100
27
15
8
3
1
50
28
15
6
2
21
14
5
5
3
6
39
26
9
9
6
11
19
18
17
35
33
31
38
9
2
7
70
17
4
13
(9)
19
16
12
11
10
8
7
7
6
5
35
30
22
20
19
15
13
13
11
9
15
9
30
8
28
17
56
15
374
Health service
and resources
Supervisors
Supervisee and
Community
The box below summarises the features of supervision from field reports and surveys.
Box 3 Summary of the features of supervision from field reports and surveys
Mostly consists of central supervisors visiting peripheral health
care staff, but can include visits to the centre;
Focuses on administration and checking, sometimes with
checklists; less common is problem solving, feedback and
clinical supervision.
Sometimes includes training and consultation with the
community.
Is expensive, and has a number of barriers to effective
implementation.
376
MCH
The Gambia,
19821996
Vietnam,
19941996
Hill et al.
(2000)
Chalker
(2001)
Prescribing
Malaria
Delacollette
Zaire,
et al. (1996) 19851987
Author year
ITS
DSS
CBA
Study
design
Intervention and
control groups
I: complex PHC
intervention,
including
supervision of
village health
workers and
traditional birth
attendants by
Community
Health nurses
C: no
intervention
I: Money for
217 Commune
drugs, donation
health workers
in 12 districts of for equipment;
monthly visits
Hai Phong
province
by district
supervisors to
community
health stations.
C: rolling
controls
Data from
Demographic
Surveillance
system in 40
villages
Sample size
Output
Output
Impact
Output
Impact
Impact
Outcome
category
Table 3 Comparative studies, ordered by strength of the outcome category and by author name
95% CI:
3.0 to 80.1; 1.4 to 2.7
2.4 to 15.3; 1.1 to 3.1
I: 4.9; C: 2.0
I: 6.0; C: 1.9
Ranges
only significant in
in several
1989-92
periods: 0.74 to for infants: 0.67 (95%
1.14 (infant)
CI:
and 0.74 to
0.50 to 0.90).
1.07
(childhood);
Not reported
95% CI:
1.6 to 2.1
1.7 to 2.2
not significant
Statistical
significance
1.9
1.9
-
Results
Main outcomes
Nepal
Uganda
Kafle et al.
(1997)
Kafuko
(1997)
Reproductive
Health
Miller et al.
(1996)
Kenya, 1989
and 1995
PHC
Loevinsohn
Philippines,
et al. (1995) 19911992
Prescribing
Prescribing
Author year
Table 3 (Continued)
CBA
CBA
Cluster
RCT
CBA
Study
design
Intervention and
control groups
99 (year 1989)
and 147 (year
1995) Service
Delivery Points
Intervention: 21
HUs; control:
42 HUs. 30
prescriptions in
each HU and 9
observations of
encounters
I: National
reproductive
health
programme. C:
same in 1989
I: supervision by
District Public
health officer +
audit feed-back
at the end of
each
supervision,
with guidelines
and tools
C: no
intervention
15-36 patients
I1: treatment
encounters in
guidelines +
127 HC
training + 12
monthly
supervision
visits of HUs by
members of the
District Health
team;
I2: guidelines +
training;
I3: only
guidelines
C: no
intervention
Intervention: 56 I: integrated
HUs; control 68 supervisory
HUs
checklist
C: supervision
without
checklist
Sample size
Output
Output
Output
Score (maximum
60)
summarising 20
indicators at 0
and 6 months
Proportion of
cases treated
according to
guidelines; same
for malaria
cases
SDP receiving
supervisory
visits
Improvement in Significant in I1
all groups but C (P = 0.007) and I2
(P = 0.047)
I: from 26 to 37 P = 0.003
points;
C: from 27 to 32
points
Significant
improvements
(p not reported)
Statistical
significance
Output
Not reported
Results
Decrease
injections,
increase
consultation
time and
prescription
according to
guidelines
Main outcomes
Output
Outcome
category
377
378
Zimbabwe,
1996
South Africa
Trap et al.
(2001)
Uys et al.
(2005)
Zeitz et al.
(1993)
PHC
Prescribing
Author year
Table 3 (Continued)
Intervention and Outcome
control groups
category
I: training of
supervisors on
diarrhoea
management.
C: preintervention.
Output
Output
Output
I1: supervision
based on matrix
model; I2:
supervision
based on Centre
for Health and Output
Social Studies
C: usual training
on supervision Output
Output
24 HUs for each I: Quarterly
supervisory
arm:
visits on drug
intervention A
management
(stock
supervisors (I1)
management),
and standard
intervention B
Output
(guidelines) and treatment
guidelines (I2)
control.
C: no supervision
Sample size
ITS
29 Health
(March and workers in
May 1992) March and
31 in May
CBA
CBA
Study
design
Statistical
significance
I1: 7 to 26
I2: 0 to 27
C: )16 to 4
Statistical differences
found in 17 of the 27
items.
No difference in I1
No difference in I1 and
I2
No difference in I1 and
C (no data on B)
4 indicators significant
1 indicators significant
1 indicator significant
Results
I1: from 33 to
32; I2: 30
(only after);
C: from
20 to 36
Score of job
I1: from 122 to
satisfaction
75; I2: 75.5
(only after)
Score of quality I1: from 11 to
12; I2: from 10
of care and
nursing records to 11
standards
Score of patient I1: from 11 to
satisfaction
15; I2: from 14
to 14.
Differences in % )1.7 to 62.1
of health
workers
performing
clinical
tasks, two
months apart
Median change
in score of nine
stock
management
indicators
(scores)
Median change
in scores of four
adherence to
standard
treatment
guidelines
indicators
Score of
perceived
supervision
Main outcomes
Country, year
Nepal, 1992
Author year
Curtale et al.
(1995)
Table 3 (Continued)
MCH
Health area
Controlled
Study
design
1080 mothers in
each group;
several
Community
Health
Volunteers
(CHV)
Sample size
% Of mothers
with
appropriate
health seeking
behaviour
% Full
immunisation,
anti-helminths
treatment
coverage,
correct
treatment for
ARI, growth
monitoring
uptake
% Of mothers
using ORT
Knowledge
Knowledge
Output
Output
% of CHV
answering
knowledge
items on night
blindness and
ARI
% Of mothers
utilising services
Knowledge
I: complex
nutrition
education
intervention
where health
post staff
supervises
CHV
C: usual care
Main outcomes
Outcome
category
Intervention and
control groups
I: 7894%;
C: 4376%
I: 22% to
84%;
C: <1% to
66%
I: 21% to
92%;
C: 4% to
24%,
I: 35%;
C: 4%
I: 33% to
99%;
C: 3% to
64%,
Results
P < 0.001
All items
P < 0.001
P < 0.001
All items
P < 0.001
All items
P < 0.001
Statistical
significance
379
380
Output
C: control; CBA: controlled before and after study; CI; confidence interval; DSS: demographic surveillance system (repeated measures); HU: health unit; I: intervention; MCH:
maternal and child health; ORT: oral rehydration therapy; RCT: randomised control trial.
I: training about
40 HUs in each
client-provider
group and 869
and 859 users in information
exchange and
intervention
training of
and control
supervisors
groups
visiting
respectively
providers
C: usual care
CBA
Reproductive
Health
(family
planning)
Costello et al. Philippines,
(2001)
1997
Table 3 (Continued)
Study
design
Sample size
Intervention and
control groups
Outcome
category
Main outcomes
Results
Statistical
significance
381
382
Corresponding Author Xavier Bosch-Capblanch, Swiss Centre for Tropical Health Swiss Tropical Institute, Socinstrasse 57,
CH-4002 Basel, Switzerland. Tel.: +41 (0) 612848319; Fax: +41 (0) 612848103; E-mail: x.bosch@unibas.ch.
383