The impact of user fees on health service utilization in low- and middle-income countries: how strong is the evidence?
Mylene Lagarde, Natasha Palmer
Volume 86, Number 11, November 2008, 839-848
Table 3. Main characteristics of studies on the introduction of user fees and its effects on health service utilization in low- and middle-income countries, according to literature review
| Study | Study setting | Study design | Intervention | Quality assessment | Overall risk of bias |
|---|---|---|---|---|---|
| Ridde (2003) |
Burkina Faso – 9 intervention and 5 control health centres | ITSa | Introduction of user fees in PHC facilities compared with some control facilities. National policy change | Presence of confounding factors; differences in control and treatment groups; time of intervention varied slightly across facilities; use of routine data, potentially unreliable; data reanalysed to account for their longitudinal nature | High |
| Mbugua et al. (1995) |
Kenya – 1 hospital and 2 health centres and 3 free dispensaries (control) | ITSa | Introduction of user fees in hospitals and health centres. National policy change | Presence of confounding factors; few observation points; control sites not equivalent; use of routine data, potentially unreliable; data reanalysed to account for their longitudinal nature | High |
| Collins et al. (1996) |
Kenya – 4 district hospitals and 3 provincial hospitals | ITSa | Introduction of user fees in hospitals and health centres. National policy change | Presence of confounding factors (economic hardship); few observation points; use of routine data, potentially unreliable; data reanalysed to account for their longitudinal nature | High |
| Moses et al. (1992) |
Kenya – Nairobi’s special treatment clinic for STIs | ITSa | Introduction of user fees in the national referral structure for STIs. National policy change | Presence of confounding factors; few observations after the intervention; specific to one referral centre for STIs; use of routine data, potentially unreliable; data reanalysed to account for their longitudinal nature | High |
| Benjamin et al. (2001) |
Papua New Guinea – 1 general hospital and urban clinics (controls) | ITSa | Introduction of user fees for antenatal care in a hospital. National policy change | Presence of confounding factors; potential secular changes; use of routine data, potentially unreliable; data reanalysed to account for their longitudinal nature | High |
| Kremer & Miguel (2007) |
Kenya – 75 schools (25 randomly selected to introduce cost recovery) | C-RCT | Introduction of user fees for preventive deworming drugs. Experimental study | Slight difference in time of pre-intervention exposure to free drugs between some control and intervention sites | Low |
| Diop et al. (1995) |
Niger – primary care facilities in 3 districts (2 intervention sites, 1 control) | CBA | Introduction of user fees + quality improvements in PHC facilities. Pilot study | Differences in control and intervention sites (potentially affecting health-seeking behaviours); pre-existence of informal fees in the control sites; statistical analysis not always appropriate | High |
| Litvack & Bodart (1993) |
Cameroon – 5 health centres (2 control, 3 intervention) | CBA | Introduction of user fees + quality improvements in PHC facilities. Pilot study | Selection of control and treatment facilities unclear; no details provided on characteristics of treatment and control sites; statistical analysis not always appropriate (failure to test for statistical significance of comparisons; inappropriate econometric analysis of variations across socioeconomic groups) | High |
CBA, controlled “before and after”; C-RCT, cluster randomized controlled trial; ITS, interrupted time series; PHC, primary health care; STIs, sexually transmitted infections.a Longitudinal data were reanalysed by the authors of the review, so that the results do not necessarily reflect the conclusions and views of the authors of the original paper.