Supervisión pública de los profesionales de atención sanitaria privados con fines de lucro en países de ingresos medios y bajos
Resumen
Antecedentes
Los gobiernos utilizan diferentes enfoques para asegurar que los servicios de atención sanitaria privados con fines de lucro cumplan con ciertas normas de calidad. Dicha orientación gubernamental, denominada supervisión pública, incluye políticas gubernamentales, mecanismos regulatorios y estrategias de implementación para asegurar la responsabilidad en la prestación de los servicios. Sin embargo, la efectividad de estas estrategias en países de ingresos medios y bajos (PIMB) no ha sido tema de una revisión sistemática.
Objetivos
Evaluar los efectos de la regulación, la formación o la coordinación del sector público al sector privado de la salud con fines de lucro en países de ingresos medios y bajos.
Métodos de búsqueda
Para encontrar revisiones sistemáticas relacionadas, el 28 de abril de 2015 se realizaron búsquedas en la Base de Datos Cochrane de Revisiones Sistemáticas (Cochrane Database of Systematic Reviews) (CDSR) 2015, número 4; la Base de Datos de Resúmenes de Revisiones de Efectividad (Database of Abstracts of Reviews of Effectiveness) (DARE) 2015, número 1; la Base de Datos de Evaluación de Tecnologías Sanitarias (Health Technology Assessment Database) (HTA) 2015, número 1. Todas ellas son parte de La Biblioteca Cochrane. Para encontrar estudios primarios, se realizaron búsquedas en MEDLINE, Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily y MEDLINE de 1946 hasta el presente, OvidSP (búsqueda del 16 de junio de 2016); Science Citation Index y Social Sciences Citation Index de 1987 hasta el presente, y Emerging Sources Citation Index de 2015 hasta el presente, ISI Web of Science (búsqueda del 3 de mayo de 2016 para los artículos que citaban los estudios incluidos); Registro Cochrane Central de Ensayos Controlados (CENTRAL), 2015, número 3, parte de La Biblioteca Cochrane (incluido el registro especializado del Grupo Cochrane para una Práctica y Organización Sanitaria Efectivas Cochrane Effective Practice and Organisation of Care Group Specialised Register (EPOC)) (búsqueda del 28 de abril de 2015); Embase de 1980 hasta 2015 Semana 17, OvidSP (búsqueda del 28 de abril de 2015); Global Health de 1973 a la semana 16 de 2015, OvidSP (búsqueda 30 abril 2015); WHOLIS, OMS (búsqueda 30 abril 2015); Science Citation Index y Social Sciences Citation Index de 1975 a la actualidad, ISI Web of Science (búsqueda 30 abril de 2015); Health Management, ProQuest (búsqueda 22 noviembre de 2013). Además, en abril 2016, se hicieron búsquedas en las listas de referencias de artículos relevantes, la Plataforma de registros internacionales de ensayos clínicos de la OMS, Clinicaltrials.gov y en varias bases de datos electrónicas de la literatura gris.
Criterios de selección
Ensayos aleatorizados, ensayos no aleatorizados, estudios de series temporales interrumpidas o estudios controlados tipo antes y después (before and after studies).
Obtención y análisis de los datos
Dos autores de la revisión evaluaron de forma independiente la elegibilidad de los estudios y extrajeron los datos, y además compararon los resultados y resolvieron las discrepancias mediante consenso. Los resultados de los estudios se expresaron como razones de riesgos (RR) o diferencias de medias (DM) con intervalos de confianza (IC) del 95%, cuando fue apropiado, y la seguridad en la evidencia se evaluó mediante el sistema Grades of Recommendation, Assessment, Development and Evaluation (GRADE). No se realizaron metanálisis debido a la heterogeneidad de las intervenciones y los diseños de los estudios.
Resultados principales
Se identificaron 20.177 registros, 50 de ellos potencialmente elegibles. Se excluyeron 39 estudios potencialmente elegibles porque no incluían una evaluación rigurosa de la formación, la regulación o la coordinación de los profesionales sanitarios privada con fines de lucro en los PIBM. Están pendientes de evaluación cinco estudios identificados después de la presentación de la revisión. Seis estudios cumplieron nuestros criterios de inclusión. Dos estudios incluidos evaluaron solo la formación; uno evaluó solo la regulación; tres evaluaron una intervención multifacética que incluía formación y regulación; y ninguno evaluó la coordinación. Los seis estudios incluidos estaban dirigidos a los trabajadores farmacéuticos privados con fines de lucro en África y Asia.
Tres estudios encontraron que la formación probablemente aumenta las ventas de solución de rehidratación oral (un ensayo en Kenia, 106 farmacias: RR 3,04; IC del 95%: 1,37 a 6,75 y un ensayo en Indonesia, 87 farmacias: RR 1,41; IC del 95%: 1,03 a 1,93) y el despacho de fármacos antipalúdicos (un ensayo en Kenia, 293 farmacias: RR 8,76; IC del 95%: 0,94 a 81,81); evidencia de certeza moderada.
Un estudio realizado en la República Democrática Popular de Lao muestra que la regulación de la distribución y la venta de los productos farmacéuticos registrados puede mejorar los indicadores compuestos de las farmacias (un ensayo, 115 farmacias: mejoras en cuatro de seis indicadores de las farmacias; evidencia de certeza baja).
El desenlace de tres estudios de intervención multifacéticos fue la calidad de la práctica farmacéutica, incluida la capacidad de hacer preguntas, dar consejos y proporcionar un tratamiento adecuado. Los ensayos aplicaron la regulación, el entrenamiento y la influencia de los compañeros en secuencia; y el diseño del estudio no permite separar los efectos de las diferentes intervenciones. Dos ensayos realizados en 136 farmacias de Vietnam descubrieron que la intervención multifacética puede mejorar la calidad de la práctica farmacéutica, pero el tercer estudio, en el que participaron 146 farmacias de Vietnam y Tailandia, descubrió que la intervención puede tener poco o ningún efecto en la calidad de la práctica farmacéutica (evidencia de certeza baja).
Solo dos estudios (ambos realizados en Vietnam) informaron datos sobre los costos, sin una evaluación rigurosa de las implicaciones económicas de implementar las intervenciones en ámbitos de recursos limitados. Ningún estudio informó datos sobre la equidad, la mortalidad, la morbilidad, los efectos adversos, la satisfacción o las actitudes.
Conclusiones de los autores
La formación probablemente mejora la calidad de la atención (es decir, el cumplimiento de la práctica recomendada), la regulación puede mejorar la calidad de la atención y no hay seguridad con respecto a los efectos de la coordinación sobre la calidad de los servicios de atención sanitaria privados con fines de lucro en los PIMB. La probabilidad de que estudios de investigación adicionales encuentren que el efecto de la formación sea significativamente diferente de los resultados de esta revisión es moderada. Esto quiere decir que es probable que se necesite monitorizar el impacto si se implementa la formación. La baja fiabilidad de la evidencia en el caso de la regulación implica que es alta la probabilidad de que estudios de investigación adicionales encuentren que el efecto de la regulación es significativamente diferente de los resultados de esta revisión. Por lo tanto, se justifica una evaluación del impacto si se realiza la regulación gubernamental de los profesionales de atención sanitaria privados con fines de lucro en PIMB. Las evaluaciones rigurosas de estas intervenciones también deben analizar otros desenlaces como las repercusiones sobre la equidad, las implicaciones en el coste, la mortalidad, la morbilidad y los efectos adversos.
Resumen en términos sencillos
Regulación, formación o coordinación gubernamental de la atención sanitaria privada con fines de lucro en países de ingresos medios y bajos
¿Cuál es el objetivo de esta revisión?
El objetivo de esta revisión Cochrane fue evaluar el efecto de la regulación, la formación o la coordinación gubernamental de la atención sanitaria privada con fines de lucro en países de ingresos medios y bajos.
Se obtuvieron y analizaron todos los estudios relevantes para responder a esta pregunta y se incluyeron seis estudios en la revisión.
¿Por qué los gobiernos regulan, forman o coordinan a los profesionales de atención sanitaria privados?
En muchos países de ingresos medios y bajos, el sector público no es capaz de proporcionar servicios de atención sanitaria de alta calidad a todos los ciudadanos y, por lo tanto, los profesionales de atención sanitaria privados desempeñan una función principal. Sin embargo, hay inquietudes respecto a que la atención sanitaria proporcionada por el sector privado no siempre sea de alta calidad y de que no siempre se sigan las prácticas y las guías recomendadas. Por lo tanto, los gobiernos utilizan diferentes enfoques para asegurar que los servicios de atención sanitaria privados con fines de lucro cumplan con ciertas normas de calidad. Este tipo de orientación gubernamental se denomina "gestión pública" y puede implicar, por ejemplo, la formación y educación de los servicios sanitarios privados con ánimo de lucro; la introducción de reglamentos en los que se establezcan y apliquen normas de calidad; y la coordinación entre los servicios sanitarios privados con ánimo de lucro y los del sector público, por ejemplo, creando sistemas de derivación entre ellos.
¿Qué sucede cuando los gobiernos regulan, forman o coordinan a los profesionales de atención sanitaria privados con fines de lucro?
Formación
En dos estudios en Kenia e Indonesia, el Ministerio de Salud ofreció a los vendedores privados de fármacos sesiones cortas de formación sobre la prescripción y el despacho de fármacos. Estos se compararon con vendedores de fármacos a los que no se les dio formación. Los estudios indicaron que probablemente la formación mejora la calidad de los servicios de atención sanitaria.
Regulación
En un estudio en la República Democrática Popular de Lao, el Ministerio de Salud supervisó los servicios de farmacias privadas en ciertos distritos durante un período de tres meses, aplicó sanciones cuando se rompieron las reglas y ofreció información acerca de las áreas que necesitaban mejorías. Estos distritos se compararon con los distritos sin esta supervisión mejorada. El estudio indicó que esta regulación mejorada puede lograr poco o ningún cambio en la calidad de la atención.
Formación y regulación
En tres estudios en Vietnam y Tailandia, las farmacias privadas de algunos distritos recibieron visitas educativas, así como visitas de inspectores de farmacia para hacer cumplir las regulaciones. Estos distritos se compararon con distritos que no recibieron visitas. Los estudios indicaron que estos tipos de visitas pueden mejorar la calidad de la atención.
Coordinación
La revisión no encontró estudios elegibles que evaluaran los efectos de la coordinación sobre la calidad de la atención.
¿Cuál es el grado de actualización de esta revisión?
Los autores de la revisión buscaron estudios que se habían publicado hasta junio 2016.
Authors' conclusions
Summary of findings
| Training compared to no training for improving quality of care | |||
| Population: Private for‐profit providers of healthcare services | |||
| Outcomes | Impacts | No of Participants | Certainty of the evidence |
| Quality of care | Both studies show that training probably improves the quality of healthcare services | 486 pharmacies | ⊕⊕⊕⊝ |
| * We downgraded the certainty of evidence by 1 point, because of a moderate risk of selection bias in included studies. GRADE Working Group grades of evidence | |||
| Regulation compared to no regulation for improving quality of care | |||
| Population: Private for‐profit providers of healthcare services | |||
| Outcomes | Impacts | No of Participants | Certainty of the evidence |
| Quality of care | Regulation may improve quality of care. The study observed an increase of 34% in the availability of essential materials for dispensing and an increase of 19% in mean orderliness (including the presence of advertisements, and storage of drugs in their original packaging away from sunlight) in the intervention pharmacies compared to the control pharmacies | 115 pharmacies | ⊕⊕⊝⊝ |
| * We downgraded the certainty of the evidence by 2 points, because of a high risk of attrition bias and wide confidence intervals around the effect estimate, ranging from a large benefit to important harm GRADE Working Group grades of evidence | |||
| Training and regulation compared to no intervention for improving quality of care | |||
| Population: Private for‐profit providers of healthcare services | |||
| Outcomes | Impact | No of Participants | Certainty of the evidence |
| Quality of care | Training and regulation may improve quality of care | 379 pharmacies | ⊕⊕⊝⊝ |
| * We downgraded the certainty of evidence by 2 points because of a high risk of attrition bias and heterogeneity of intervention effects. Two studies found that training, regulation, and peer influence may improve quality of care while the third study found little or no effects in the quality of care with the intervention GRADE Working Group grades of evidence | |||
Background
The public health sector in low‐ and middle‐income countries is not always sufficiently well‐equipped and financed to provide high quality health care that is accessible to all citizens (Basu 2012; Lagomarsino 2009). The consequence of this public sector failure has been a proliferation in private providers of healthcare services in most of the countries (Forsberg 2011; Levin 2011; Scott 2011). Governments have a responsibility to ensure the quality of healthcare services delivered by private providers, to expand the coverage of existing private providers, and to rationalise this coverage with that of public sector providers (Waters 2003). Such government guidance is referred to as public stewardship. However, there is a paucity of high quality research evidence on the effects of public stewardship on the quality and accessibility of private for‐profit health care in low‐ and middle‐income countries (Patouillard 2007; Waters 2003); thus the need for this review.
Description of the condition
The private health sector is not homogeneous, but consists of not‐for‐profit and for‐profit as well as formal and informal providers of healthcare services (Basu 2012; Sulzbach 2011). Private not‐for‐profit healthcare providers refer to healthcare organisations that use any surplus revenues to achieve their goals, rather than distributing them as dividends. On the other hand, the private for‐profit sector refers to the part of the economy that is run by individuals and companies for profit and is not state‐controlled. The consequence of the expansion in the private health sector in LMICs is that (poor) communities spend outsized amounts of money for private healthcare services; at times when cheaper public sector alternatives are available (Forsberg 2011; Patouillard 2007). However, the quality of the services provided by the private for‐profit healthcare sector in LMICs is increasingly being questioned (Berendes 2011; Waters 2003).
Description of the intervention
The growing concern regarding the technical failures of health care provided by the private for‐profit sector has led to the development of interventions aimed at addressing these limitations, which simultaneously take advantage of the potential for involving the private for‐profit sector to achieve public health goals. This review assessed the public stewardship of private for‐profit healthcare providers in LMICs. Stewardship can be defined as a function of governments responsible for the welfare of their populations (Veillard 2011). It involves policy guidance to the whole health system, co‐ordination between actors and regulation of different functions, levels and actors in the system, an optimal allocation of resources and accountability towards all stakeholders. Although many actors have an influence on stewardship, there is a central role for the government in ensuring equity, efficiency and sustainability of the health system (Van Olmen 2010). Therefore, stewardship entails oversight and guidance of the whole system; ensuring strategic policy frameworks exist and are combined with effective oversight, coalition‐building, regulation, attention to system‐design and accountability. The stewardship function involves the role of the government in health and its relation to other actors whose activities impact on health. Public stewardship encompasses government policies, regulatory mechanisms, and implementation strategies for ensuring guidance and accountability in which healthcare services are delivered; in order to protect the public interest (WHO 2007). While ultimately it is the responsibility of government, this does not mean all stewardship functions have to be carried out by central ministries of health (WHO 2009).
Various strategies have been proposed for improving the functioning of the private for‐profit health sector in order to increase the quality, availability, and affordability of health care for poor people in LMICs (Lagomarsino 2009; Levin 2011; Patouillard 2007; Waters 2003). These strategies include regulation, contracting‐out, social marketing, franchising, use of vouchers, training, pay for performance, accreditation, and co‐ordination. The strategies use various markers of success which are analysed by their association with differences in performance of intermediate goals or outcomes (Travis 2002). We focused on three types of strategic interventions, namely, regulation, training, and co‐ordination of private for‐profit providers. In the context of this review, regulation refers to the setting and enforcing of standards for the private for‐profit sector; training involves educating and supporting private for‐profit service providers; and co‐ordination entails organising and creating alliances between private for‐profit and public sector healthcare providers. We excluded public stewardship interventions which are already covered by other Cochrane reviews; including social marketing and franchising (Koehlmoos 2009), contracting (Lagarde 2009), and pay for performance (Witter 2012).
How the intervention might work
Regulatory interventions take the form of rules, enforcement systems and sanction mechanisms, and can be applied at the levels of the healthcare provider, organisation, or facility. At the provider level, regulation may include requirements for pre‐service training, continuing education, licensing, and certification. At the organisational or facility level, regulation may aim to control the location of facilities, their registration, prices, and minimum complement of staff or facilities. For example, pharmaceutical market regulation aims to limit the availability of harmful drugs and unregistered products, minimise drug misuse, control the sale of specific drugs through prescriptions, and control drug manufacture and importation. Training interventions may involve formal educational sessions (educational meetings and workshops), vendor‐to‐vendor education, distribution of guidelines, printed educational materials, educational outreach i.e. a personal visit by a trained government official to private for‐profit healthcare providers in their own settings, or audit and feedback i.e. a summary of the performance of private for‐profit providers over a specified period of time given in a verbal or written format; alone or in combination (Forsetlund 2009; Jamtvedt 2006; O'Brien 2007; Patouillard 2007). A wide variety of private for‐profit healthcare sector components could be targeted for training, including physicians, pharmacists, midwives, nurses, and traditional healers. Finally, government co‐ordination of private for‐profit health care ensures harmonised minimum standards for health service delivery across geographic areas and social groups (Waters 2003). For instance, the creation of referral systems between the private for‐profit and public sector and ensuring that health professionals in different health sectors understand their roles in disease management. The ultimate aim of government regulation, training, and co‐ordination of the private for‐profit health sector is to promote equity, better health outcomes, and financial protection (Lagomarsino 2009).
Why it is important to do this review
A systematic review published in 2007 found “evidence that effective public‐private partnerships can increase access, improve equity, and raise quality of health services" (Patouillard 2007). However, using the GRADE approach (Balshem 2011; Guyatt 2008), this evidence on the effectiveness of interventions for working with the private for‐profit sector to improve the utilisation and quality of health services for the poor in low‐ and middle‐income countries was found to be of low certainty (Wiysonge 2008). The implication of the low certainty of the evidence is that further research on this topic is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. It is possible that additional primary studies may have been conducted on this topic. Therefore, we reviewed the currently available evidence on public sector efforts to work with private for‐profit health service providers to improve the quality of existing healthcare services as well as expand and rationalise their coverage (Waters 2003).
Objectives
To assess the effects of public sector regulation, training, or co‐ordination of the private for‐profit health sector in low‐ and middle‐income countries.
Methods
Criteria for considering studies for this review
Types of studies
The studies eligible for inclusion in the review were:
-
randomised trials, including individually‐randomised and cluster‐randomised trials;
-
non‐randomised controlled trials i.e. experimental studies in which people are allocated to different interventions using methods that are not random;
-
interrupted time series studies with at least three measurements before and after introducing the intervention; and
-
controlled before‐after studies with at least two intervention groups and at least two comparable control groups, with simultaneous data collection (EPOC 2013).
Types of participants
Studies taking place in low‐ and middle‐income countries as defined by the World Bank. All types of health services provided by for‐profit providers were eligible for inclusion in this review.
Types of interventions
Regulation, training, and co‐ordination of any intensity or duration, implemented by the public sector. The control group received no intervention or an alternative intervention.
Types of outcome measures
The outcomes of interest were as follows.
Primary outcomes
-
Quality of care (defined as adherence to recommended practice or guidelines).
Secondary outcomes
-
Equity
-
Mortality or morbidity
-
Adverse effects (e.g. undesirable impacts on existing public or private services, inappropriate use of services, and distortion in the provision of services)
-
Satisfaction
-
Attitudes
-
Costs of implementing the interventions
Search methods for identification of studies
We developed a sensitive and previously validated search strategy for randomised trials, non‐randomised trials, controlled before‐after studies, and interrupted time series studies combined with relevant medical subject headings (MeSH) and free‐text terms relating to health regulation, training and co‐ordination literature for low‐ and middle‐income countries. We placed no language or date restrictions on the search strategy. We translated the MEDLINE (Ovid) search strategy into the other databases using the appropriate controlled vocabulary.
Electronic searches
We searched the following databases for systematic reviews:
-
Cochrane Database of Systematic Reviews (CDSR) 2015, Issue 4, part of The Cochrane Library. www.cochranelibrary.com (searched 28 April 2015)
-
Database of Abstracts of Reviews of Effectiveness (DARE) 2015, Issue 1, part of The Cochrane Library. www.cochranelibrary.com (searched 28 April 2015)
-
Health Technology Assessment Database (HTA) 2015, Issue 1, part of The Cochrane Library.www.cochranelibrary.com (searched 28 April 2015).
We searched the following databases, with no language or date restrictions, for primary studies:
-
MEDLINE, Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily and MEDLINE 1946 to Present, OvidSP (searched 16 June 2016)
-
Science Citation Index and Social Sciences Citation Index 1987 to present, and Emerging Sources Citation Index 2015 to present, ISI Web of Science (searched 3 May 2016 for papers citing included studies)
-
Cochrane Central Register of Controlled Trials (CENTRAL), 2015, Issue 3, part of the Cochrane Library. www.cochranelibrary.com (including the Cochrane Effective Practice and Organisation of Care (EPOC) Group Specialised Register) (searched 28 April 2015)
-
Embase 1980 to 2015 Week 17, OvidSP (searched 28 April 2015)
-
Global Health 1973 to 2015 Week 16, OvidSP (searched 30 April 2015)
-
WHOLIS, WHO (searched 30 April 2015)
-
Science Citation Index and Social Sciences Citation Index 1975 to present, ISI Web of Science (searched 30 April 2015)
-
Health Management, ProQuest (searched 22 November 2013).
Searching other resources
In April 2016 we searched the following databases and websites for eligible studies:
-
OpenGrey (opengrey.eu)
-
Grey Literature Report (greylit.org)
-
World Bank e‐Library (elibrary.worldbank.org)
-
US National Institutes of Health (NIH) (nih.gov)
-
United Nations Children's Fund (UNICEF) (unicef.org)
-
Alliance for Health Policy and Systems Research (who.int/alliance‐hpsr/en)
-
United States Agency for International Development (USAID) (usaid.gov)
-
Gavi, The Vaccine Alliance (gavi.org)
-
Private Healthcare in Developing Countries (ps4h.org)
-
Population Services International (PSI) (psi.org)
-
Shops (shopsproject.org)
-
United Kingdom Department for International Development (gov.uk/government/organisations/department‐for‐international‐development)
-
Center For Health Market Innovations (healthmarketinnovations.org
-
World Bank (worldbank.org)
Trial Registries
-
International Clinical Trials Registry Platform (ICTRP), Word Health Organization (WHO) (who.int/ictrp/en) (searched April 2016)
-
ClinicalTrials.gov (clinicaltrials.gov) (searched April 2016)
We checked the reference lists of identified reviews (Basu 2012; Berendes 2011; Forsberg 2011; Forsetlund 2009; Levin 2011; Patouillard 2007; Sulzbach 2011; Waters 2003) as well as reference lists of full‐text articles reviewed for inclusion in this review.
Data collection and analysis
Selection of studies
Two authors (Leila Abdullahi and Valantine Ndze, Leila Abdullahi and Charles Wiysonge, or Valantine Ndze and Charles Wiysonge) screened the titles and abstracts of outputs from the searches using a pre‐designed screening guide to identify potentially eligible studies. We retrieved the full text of all publications deemed potentially eligible by at least one of the two authors. The two authors then independently examined each of these for eligibility. Each author compiled a list of studies which he or she believed met the inclusion criteria. Both authors compared the lists and resolved discrepancies by discussion and consensus.
Data extraction and management
The two authors independently extracted descriptive and outcome data from each included study using a pre‐designed data collection form. Both authors compared extracted data, resolving any discrepancies by discussion and consensus, failing which a third author would have arbitrated. One of two authors (Leila Abdullahi and Charles Wiysonge) entered the data into Review Manager (RevMan) 5.3 (RevMan 2014) and the other author performed double checks to ensure that there were no errors in the data entered.
Assessment of risk of bias in included studies
We assessed the risk of bias based on six standard domains:
-
Sequence generation
-
Concealment of allocation
-
Blinded or objective assessment of primary outcome(s)
-
Incomplete outcome data
-
Selective outcome reporting; and
-
Other sources of bias (Higgins 2011a).
We also used three additional criteria specified by the Cochrane Effective Practice and Organisation of Care Group (EPOC) (EPOC 2015):
-
Similar baseline characteristics
-
Similar baseline outcome measures
-
Reliable primary outcome measures; and
-
Adequate protection against contamination.
For each included study, two authors independently reported their assessment of the risk of bias for each domain (i.e. low, high, or unclear) together with a descriptive summary of the information that influenced their judgment. The two authors compared the results of their independent assessments of the risk of bias and resolved any discrepancies by discussion and consensus. Had the two authors failed to reach an agreement, a third author would have arbitrated.
Measures of treatment effect
We grouped measures of treatment effect based on outcome variables and study designs. We recorded and used estimates of effect from the primary analysis reported by the investigators.
We anticipated that there would be important baseline differences between intervention and control groups and planned to base our primary analyses for trials and controlled before‐after studies on estimates of effect that were adjusted for baseline differences. For dichotomous outcomes we planned to calculate the adjusted risk difference as the difference in adherence after the intervention minus the difference before the intervention. A positive risk difference would indicate that compliance with the recommended practice improved more in the intervention group than in the control group (e.g. an adjusted risk difference of 0.11 would indicate an absolute improvement in compliance with targeted behaviours of 11%). For continuous outcomes we planned to calculate the adjusted change relative to the control group as the post‐intervention difference in means minus the baseline difference in means divided by the baseline control group mean. As with the adjusted risk difference, a positive change would indicate that compliance improved more in the intervention group than in the control group. This is a relative effect rather than an absolute effect; the effect size reflects the baseline performance as well as the change in performance and it is not bound between ‐100% and +100%.
We planned to analyse interrupted time series studies using either a regression analysis with time trends before and after the intervention, which adjusts for auto‐correlation and any periodic changes; or any other technique that adjusts for auto‐correlation and periodic changes. We would present results for the outcomes as changes along two dimensions: change in level and change in slope. Change in level is the immediate effect of the policy and change in slope is the change in the trend from pre‐ to post‐intervention. It reflects the long‐term effect of the intervention.
For all measures we planned to calculate 95% confidence intervals (CI).
Unit of analysis issues
We planned that if investigators reported cluster‐randomised trial data as if the randomisation was performed on the individuals rather than the clusters, we would request the intra‐cluster correlation coefficient from the study authors; failing which we would obtain external estimates of the intra‐cluster correlation coefficient from similar studies or available resources. Once established, we would use the intra‐cluster correlation coefficient to re‐analyse the trial data to obtain approximate correct analyses; as described in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). We planned to combine the effect estimates and their corrected standard errors from cluster‐randomised trials with those from parallel group designs using the generic inverse variance method (Deeks 2011). If insufficient information was available to control for clustering in this way, we would enter data into RevMan using individuals as the unit of analysis. We would then perform sensitivity analyses to assess the potential bias that may have occurred as a result of the inadequately controlled cluster‐randomised trials. We planned that we would also perform sensitivity analyses if the intra‐cluster correlation coefficients were obtained from external sources to assess the potential biasing effects of inadequately controlled cluster‐randomised trials.
Three included studies were cluster‐randomised trials based on matched pairs of clusters (Chalker 2002; Chuc 2002; Stenson 2001). We did not re‐analyse these data as matching cannot be taken into account in re‐analyses in such studies unless the raw data are available. The studies, however, conducted appropriate analyses of the data, and we have provided the results as reported in the studies. We have re‐analysed the data for the fourth cluster‐randomised trial (Abuya 2009). We did not conduct a meta‐analysis.
Dealing with missing data
We planned that where necessary, we would contact the corresponding authors of included studies to supply any unreported data. If the corresponding author did not respond within one week of our request, we planned to contact other authors (copying the corresponding author). If a study reported outcomes only for participants completing the trial or only for participants who followed the protocol, we planned to contact the authors and ask them to provide additional information to permit us to conduct meta‐analyses by intention‐to‐treat. We would describe missing data and dropouts for each included study in the risk of bias table, and discuss the extent to which the missing data could alter our results. We planned to conduct sensitivity analyses to assess the effect of missing data on our primary meta‐analyses. If we had at least 10 studies in a meta‐analysis, we would have explored the impact of including trials with high levels of missing data in the overall assessment of intervention effects by using sensitivity analyses.
For the current version of the review, we did not contact the primary study authors for missing data. We identified levels of attrition for included trials and performed analyses for reported outcomes. All participants were analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. We assumed that missing participants did have the outcome of interest, and did not conduct sensitivity analyses imputing values for the outcome status of missing participants.
Assessment of heterogeneity
Given the variation found across studies in relation to the interventions, study design and outcome measures, we have not conducted a meta‐analysis of study results. A statistical assessment of heterogeneity of results was therefore not done.
If we found homogeneous studies of similar interventions that reported similar outcomes, we would have conducted a meta‐analysis, examined statistical heterogeneity between study results using the Chi2 test of homogeneity (with significance defined at the alpha‐level of 10%) (Deeks 2011), and quantified any statistical heterogeneity between study results using the I2 statistic (Higgins 2003).
Assessment of reporting biases
We employed strategies to search for and include relevant unpublished studies in order to reduce possible publication bias. These strategies included searching the grey literature and prospective trial registration databases to overcome time‐lag bias.
Data synthesis
Due to important heterogeneity between studies, a pooled statistical analysis of the results was not possible. Therefore we did a qualitative analysis based on intervention and outcome measures. If we had identified two or more clinically homogenous studies with similar interventions and comparison groups that reported similar outcome measures, we would have used meta‐analysis to estimate the overall effect across those studies. We would have calculated all overall effects, if applicable, using inverse variance methods.
We used the GRADE approach to summarise the certainty of the evidence for each outcome (Guyatt 2008).The GRADE approach results in an assessment of the certainty of a body of evidence as high, moderate, low, or very low. High certainty evidence implies that "further research is very unlikely to change our confidence in the estimate of effect". Moderate certainty evidence means that "further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate". Evidence is considered of low certainty if "further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate", and very low quality if "we have very little confidence in the effect estimate" (Balshem 2011).
Subgroup analysis and investigation of heterogeneity
We stratified analysis by type of intervention. We did not find any studies that were similar enough to combine in a meta‐analysis and, therefore, we did not conduct any subgroup analyses.
Sensitivity analysis
If we had found 10 or more studies that were similar enough that it would be sensible to combine them in a meta‐analysis, we would have conducted sensitivity analyses to investigate the robustness of the results to risk of bias (i.e. omitting any studies with high risk of bias) and method of meta‐analysis (i.e. random‐effects versus fixed‐effect).
Results
Description of studies
Results of the search
Our search yielded 20,177 titles and abstracts. After removing 1,419 duplicates , we screened 18,758 records; 18,708 of which were not relevant. We reviewed the 50 potential eligible articles for inclusion. Six of these studies met our inclusion criteria (Abuya 2009; Chalker 2002; Chalker 2005; Chuc 2002; Ross‐Degnan 1996; Stenson 2001), and we excluded 39 for reasons given in the Characteristics of excluded studies. Five studies were identified after the review was submitted and are awaiting assessment (see Characteristics of studies awaiting classification). We present the search and selection of studies for this review in Figure 1.
Study flow diagram
Included studies
We included six randomised trials on regulation and training of private for‐profit healthcare providers in low‐ and middle‐income countries (Abuya 2009; Chalker 2002; Chalker 2005; Chuc 2002; Ross‐Degnan 1996; Stenson 2001). One study (Ross‐Degnan 1996) had two components; one being a randomised trial, and the other a non‐randomised trial. Five studies assessed training (Abuya 2009; Chalker 2002; Chalker 2005; Chuc 2002; Ross‐Degnan 1996), four studies assessed regulation (Chalker 2002; Chalker 2005; Chuc 2002; Stenson 2001), and no study assessed co‐ordination.
Description of interventions
Two studies (Abuya 2009; Ross‐Degnan 1996) assessed only training interventions (N = 486 pharmacies), one study (Stenson 2001) assessed regulation only (N = 115 pharmacies), and three studies (Chalker 2002; Chalker 2005; Chuc 2002) had a multifaceted intervention which combined training, regulation, and peer influence (N = 379 pharmacies). All six studies targeted private pharmacy workers or drug retailers.
Training
The intervention in the two 'training‐only' studies consisted of short‐duration training sessions of one or two days in Kenya (Abuya 2009; Ross‐Degnan 1996) and Indonesia (Ross‐Degnan 1996). In both studies drug sellers were trained on prescription and dispensing of drugs, and surrogate patients (i.e. simulated clients) were used to assess the effects of the intervention on the quality of care provided by the trained retailers. The training was provided by the Ministry of Health in each country.
Regulation
One study (Stenson 2001) assessed regulation only (N = 115 pharmacies). The regulatory intervention involved three‐month intensive supervision of pharmacy services in the Lao People's Democratic Republic, applying sanctions when rules were violated and providing up‐to‐date regulatory documents and information about particular areas needing improvements (Stenson 2001). The study compared districts with active regulation to districts with only "regular supervision". The 'regular supervision' intervention package was implemented in the way and speed that would have taken place in the absence of the study. The aim was to let the control districts follow their natural course. The intervention was provided by the Ministry of Health with assistance from the United Nations Children's Fund (Stenson 2001).
Multifaceted intervention
Three studies (Chalker 2002; Chalker 2005; Chuc 2002) had a multifaceted intervention which combined training, regulation, and peer influence (N = 379 pharmacies). Each intervention lasted three months, with a gap of four months before the next intervention. The quality of practice after the intervention was assessed through simulated clients. Two studies (Chalker 2002; Chuc 2002) were performed in Hanoi (Vietnam) with the intervention delivered by the Hanoi Health Bureau and the Hanoi Pharmacy Association. One study (Chalker 2005) was performed in both Hanoi (Vietnam) and Bangkok (Thailand); with the intervention delivered by the Hanoi Health Bureau and the Hanoi Pharmacy Association in Vietnam and the Bangkok Health System Research Institute and the Community Pharmacy Association in Thailand. The studies compared pharmacies with the multifaceted intervention to pharmacies without any intervention. All pharmacists who received the multifaceted intervention received all three interventions as a set. Enforcement regulation was performed by pharmacy inspectors while the educational intervention consisted of educational visits by senior researchers. Peer influence involved a number of group leaders and representatives of the pharmacy associations, who attended seminars where the research group informed them about the peer influence strategy and reviewed management.
Co‐ordination
We did not identify any study that assessed the effects of government co‐ordination of private for‐profit healthcare providers, such as the creation of referral systems between the private for‐profit and public sectors.
Description of outcomes
All six included studies reported on change in quality of care. The latter was measured using different dimensions in the different studies. One study (Stenson 2001) that assessed only regulation measured quality of care through change in the quality of private pharmacy practices (using "pharmacy indicators"). The two studies (Abuya 2009; Ross‐Degnan 1996) that assessed only training measured quality of care through correct management of childhood malaria or diarrhoea respectively. The three studies with multifaceted interventions (Chalker 2002; Chalker 2005; Chuc 2002) measured quality of care through change in the correct management of tracer conditions, antibiotic dispensing practices, and correct symptomatic treatment of sexually transmitted infections respectively. Two studies (Chuc 2002; Chalker 2002) reported the cost of implementing the interventions.
No studies reported data on equity, mortality, morbidity, adverse effects, satisfaction, or attitudes.
Excluded studies
We excluded 34 studies (Akoria 2008; Akol 2014; Ali 2011; Ali 2012; Andrianasolo 2012; Bhat 1996; Bojalil 1999; Chakraborty 2000; Contiades 2007; Dholakia 2013; Farsi 1999; Fernandes 2009; Goodman 2007; Grundy 2010; Guiscafre 2001; Harrison 2000; Hongoro 2000; Kangwana 2011; Khan 2006; Kumaranayake 2000; Maiga 2010; Marsh 2004; Minh 2013; Murugesan 2009; Nsimba 2007; Obua 2004; Okonofua 2003; Osterholt 2009; Rutta 2011; Stenson 2001b; Syhakhang 2001; Tavrow 2003; Tumwikirize 2004; Willey 2014) for reasons given in the table of Characteristics of excluded studies. The most common reason for exclusion was an ineligible study design.
Risk of bias in included studies
We have summarised our judgements about the risk of bias in each included study in Figure 2 and Figure 3.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
The methods for generation of the randomisation sequence and allocation concealment were unclear in all six studies (Abuya 2009; Chalker 2002; Chalker 2005; Chuc 2002; Ross‐Degnan 1996; Stenson 2001).
Blinding
Outcome assessors were blinded in two studies (Abuya 2009; Ross‐Degnan 1996), but there was no description of blinding in the rest (Chalker 2002; Chalker 2005; Chuc 2002; Stenson 2001).
Incomplete outcome data
Loss to follow up was moderate to high in the six studies.
Selective reporting
Selective reporting was categorised as unclear since we had no access to the study protocols.
Other potential sources of bias
All studies reported similar baseline characteristics among the intervention and control groups. Two studies (Chalker 2002; Chalker 2005) reported small differences in the outcome measures at baseline while no description was provided on baseline outcome measures in the rest of the studies. In one cluster‐randomised controlled trial there was some degree of contamination in a cluster that was meant to be a control cluster (Abuya 2009), but none of the other studies reported contamination of control clusters with the interventions assessed. We did not have any evidence that other biases were introduced into the remaining studies, over and above the ones reported above.
Effects of interventions
See: Summary of findings for the main comparison Government training of private for‐profit healthcare providers; Summary of findings 2 Government regulation of private for‐profit healthcare providers; Summary of findings 3 Government training and regulation of private for‐profit healthcare providers
Primary outcome
All six included studies reported effects on quality of care; although quality of care was measured using different indicators. Three studies focused on improving treatment of childhood illnesses such as acute respiratory infection, malaria, or diarrhoea (Abuya 2009; Chalker 2002; Ross‐Degnan 1996); one assessed the quality of treatment of sexually transmitted infections (Chuc 2002); and two assessed antibiotic dispensing practices (Chalker 2005; Chuc 2002).
Training
Each of the two studies that assessed training alone (Abuya 2009; Ross‐Degnan 1996) observed improvements in quality of care. The study conducted in Kenya and Indonesia (Ross‐Degnan 1996) showed an overall improvement in the management of diarrhoea among counter attendants in the intervention pharmacies compared to the controls. The sale of oral rehydration solution in the intervention pharmacies increased by 204% in Kenya (1 trial, 106 pharmacies; RR 3.04, 95%CI 1.37 to 6.75: Analysis 1.1) and 41% in Indonesia (1 trial, 87 pharmacies; RR 1.41, 95%CI 1.03 to 1.93: Analysis 1.1); compared to control pharmacies. In Kenya (Abuya 2009), correct prescription and dispensing of anti‐malarial drugs improved substantially (1 trial, 293 pharmacies; RR 8.76, 95% CI 0.94 to 81.81: Analysis 1.2). Using the GRADE approach (Balshem 2011) we judged the certainty of evidence on the effects of training on quality of care as moderate (summary of findings Table for the main comparison). Although the findings were consistent across the studies, we downgraded the evidence because of a moderate risk of bias in the included studies (Figure 3).
Regulation
In the 'regulation only' study, conducted in the Lao People's Democratic Republic, the distribution and selling of registered pharmaceutical products was regulated in order to protect consumers against unfair practices (Stenson 2001). The study observed an increase of 34% in the availability of essential materials for dispensing and an increase of 19% in mean orderliness (including the presence of advertisements, and storage of drugs in their original packaging away from sunlight) in the intervention pharmacies compared to the control pharmacies. "Information given to customers increased from 35% to 51% and the mixing of different drugs in the same package went down from 17% to 9%. The pharmacies in the active intervention districts showed greater improvements for four of the six pharmacy indicators" (Stenson 2001) . Using the GRADE approach (Balshem 2011), we judged the certainty of the evidence on the effects of the regulatory interventions on quality of care as low (summary of findings Table 2). Our main concern with the evidence was the imprecision of the effect estimate.
Multifaceted intervention
In the multi‐faceted intervention studies (Chalker 2002; Chalker 2005; Chuc 2002), the interventions (regulation, training, and peer influence) were applied in a sequence, and the study design does not permit separation of the effects of the different interventions. The three studies provided inconsistent results regarding the effect of the multiple interventions on quality of pharmacy practice; including the ability to ask questions, give advice, and provide appropriate treatment for four tracer conditions (acute respiratory conditions, malaria, diarrhoea, and sexually transmitted infections). In one study conducted in Vietnam (Chalker 2002), knowledge and reported practice among drug sellers improved for three of the four tracer conditions in intervention pharmacies compared to control pharmacies. The second study conducted in Vietnam, (Chuc 2002), found that the intervention pharmacies improved substantially in all tracer conditions compared to the control pharmacies. Chalker 2005 was conducted in both Vietnam and Thailand and had mixed results. Improvements were observed in Vietnam in the dispensers’ behaviour for all tracer conditions, but in Thailand improvements occurred in only one of the tracer conditions. We judged the certainty of the evidence on the effects of the multifaceted intervention as low (summary of findings Table 3), because of concerns regarding inconsistency of findings and high risk of bias in the included studies.
Secondary outcomes
Two multifaceted intervention studies reported the cost of the interventions (Chalker 2002; Chuc 2002). Chalker 2002 reported the cost incurred for the three interventions in 30 pharmacies to be USD 5700. The Chuc 2002 study reported that the costs of treating four tracer conditions increased for both intervention and control pharmacies.
No study reported data on equity, mortality, morbidity, adverse effects, satisfaction, or attitudes.
Discussion
Summary of main results
Our comprehensive search of the literature identified 20,177 records, from which six randomised controlled trials fulfilled our inclusion criteria. In two studies in Kenya and Indonesia, the Ministry of Health offered private drug sellers short training sessions on prescribing and dispensing drugs. These sellers were compared to drug sellers who were not offered training. The studies suggest that training probably improves the quality of healthcare services. In one study in the Lao People’s Democratic Republic, the Ministry of Health supervised private pharmacy services in certain districts over a three‐month period, applied sanctions when rules were broken, and offered information about areas needing improvement. These districts were compared to districts without this enhanced supervision. The study suggests that this enhanced regulation may improve quality of care. In three studies in Vietnam and Thailand, private pharmacies in some districts received educational visits as well as visits from pharmacy inspectors to enforce regulations. These districts were compared to districts that did not receive any visits. The studies suggest that these types of visits may provide mixed results. The review did not find any eligible study that assessed the effects of co‐ordination on quality of care.
Overall completeness and applicability of evidence
Despite the large number of records obtained in our literature search, only six studies with moderate to high risk of bias met our inclusion criteria. All studies were conducted in Africa and Asia; and the results may be applicable to low‐ and middle‐income countries in other continents.
Health worker availability in the public sector is a key barrier to strengthening health systems in LMICs. Effective government interventions to expand the coverage of private for‐profit healthcare providers and rationalise access to their services with that of public sector providers could strengthen health systems in LMICs. However, countries considering the implementation of public stewardship interventions need to assess the human and financial resource capacity of the public sector to properly supervise private providers.
All the studies covered only services of private for‐profit pharmacists. The findings may not be directly transferable to other cadres of private healthcare providers. Therefore, there is a need for studies on other private sector components such as private hospitals, physicians, midwives, nurses, and traditional healers.
There were no studies that reported data on equity, mortality, morbidity, adverse effects, satisfaction, or attitudes. If training or regulatory interventions are directed at providers that serve disadvantaged populations they could help to decrease inequity. However, intervention effects could vary across settings, for example between rural and urban areas, because of the distribution of private providers in these different areas. Expanding the coverage of private for‐profit providers could reduce inequity if, for example, access to the private sector is available where access to the public sector is limited. However, if private for‐profit providers are unavailable in underserved areas, expanding access to private providers may increase inequity between urban and rural areas.
There was no rigorous evaluation of the cost implications of implementing the interventions, thus this review does not provide evidence on investment in private for‐profit providers on quality of care in low‐ and middle‐ income countries. The structure and specific tasks associated with a particular public stewardship function will determine the costs. Given these uncertainties, implementation of public sector regulation, training or co‐ordination of private for‐profit healthcare providers should be accompanied by a robust framework for monitoring the costs and impacts of the interventions.
Much of the currently available literature on training and regulation of private for‐profit providers in low‐ and middle‐income countries is descriptive rather than evaluative, detailing experiences that may have great potential without rigorously testing their effectiveness (Berendes 2011; Patouillard 2007; Waters 2003; Wiysonge 2008). Well‐designed studies evaluating public stewardship functions are therefore needed before these are implemented on a large scale in low‐income countries.
Certainty of the evidence
Using the GRADE approach, we judged the certainty of evidence on the effects of training interventions on quality of care as moderate; which implies that “further research is likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate”. We rated the certainty of evidence on regulation and the multifaceted intervention as low, which means that “further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate” (Balshem 2011). Our main concerns with the evidence were limitations of the included studies, wide confidence intervals around the effect estimates, and heterogeneity of intervention effects.
Potential biases in the review process
We minimised potential biases in the review process by adhering to Cochrane guidelines (Higgins 2011). We conducted comprehensive searches without limiting the searches to a specific language. Two authors independently assessed study eligibility, extracted data, and assessed the risk of bias in each included study.
Agreements and disagreements with other studies or reviews
Despite the widespread availability of private healthcare services in low‐ and middle‐income countries, there is a shortage of systematic reviews that have assessed interventions showing how governments have worked with the private for‐profit providers to achieve public health goals (Brugha 1998; Levin 2011; Patouillard 2007; Peters 2004; Waters 2003). To the best of our knowledge, our review is the most comprehensive and up‐to‐date assessment of the evidence on the effects of training, regulation, and co‐ordination of private‐for‐profit health care in low‐ and middle‐income countries.
In 2003 Hugh Waters and colleagues published a review that assessed the evidence available concerning public sector efforts to work with private health service providers and other components of the private sector, in order to both improve the quality of their services and to rationalise and expand their coverage (Waters 2003). The review focused on interventions aimed at regulating, contracting, financing, social marketing, training, co‐ordinating, and informing private providers in low‐ and middle‐income countries. The review authors searched Pubmed and Popline databases, and reference lists of included articles, for both published and unpublished literature from 1980 onwards. They included 42 studies including six ‘controlled’ trials comparing results in two or more groups; 10 studies with a pre‐post evaluative component, but no comparison group; four cross‐sectional studies; and 22 descriptive case studies. Waters 2003 found that although governments are gaining experience in using the tools of contracting, regulating, financial incentives, training, co‐ordinating, and informing to influence the private sector, the evidence on the effectiveness of these interventions remains weak.
In 2007 Edith Patouillard and co‐workers published a related review, which assessed the effectiveness of interventions on working with the private for‐profit sector to improve utilisation of quality health services by the poor in low‐ and middle‐income countries (Patouillard 2007). Interventions of interest to the review authors included social marketing, use of vouchers, pre‐packaging of drugs, franchising, training, regulation, accreditation, and contracting‐out. They conducted a comprehensive search of peer‐reviewed and grey literature for eligible studies; focusing on studies which evaluated the impact of interventions on utilisation or quality of services, or both, and which provided information on the socioeconomic status of the beneficiary populations. The review authors identified 52 eligible studies; five provided data on the average socioeconomic status of recipient communities and five provided data on the distribution of benefits across socioeconomic groups. Patouillard 2007 concluded that it is not possible to prove from the available literature that private sector interventions benefit the poor and improve equity. However, they argue that the fact that many such interventions have operated successfully in relatively poor settings indicates that the interventions do benefit the poor. The authors went on to recommend that better evidence of the equity impact of interventions working with the private sector is needed for more robust conclusions to be drawn.
There are marked differences between these two previous reviews (Patouillard 2007; Waters 2003) and the current systematic review. Although the authors of the two previous reviews conducted literature searches that were relatively comprehensive, they do not explicitly say whether they undertook duplicate study selection and data extraction and do not report reliable criteria for assessing the risk of bias in included studies. They do not provide appropriate description of the characteristics of included studies and do not seem to synthesise data from included studies using reliable methods. In addition, much of the data reported in the reviews is descriptive rather than evaluative, detailing experiences that may have great potential without rigorously testing their effectiveness. Most of the studies included in the reviews were not set up as research projects (Patouillard 2007; Waters 2003). Furthermore, the two reviews included studies in which the stewardship functions were carried out by the public sector as well as those in which the functions were carried out by non‐governmental organisations (NGO). We acknowledge that while ultimately public stewardship is the responsibility of government, this does not mean all stewardship functions have to be carried out by the public sector. While we recognise the importance of private sector interventions implemented by NGOs, their characteristics and incentives are likely to differ from those implemented by government. We included one study in our review (Abuya 2009), which was published after publication of the two previous reviews (Patouillard 2007; Waters 2003).
To the best of our knowledge, our review is the most comprehensive and up‐to‐date assessment of the evidence on the effects of training, regulation, and co‐ordination of private for‐profit healthcare providers in low‐ and middle‐income countries. Some recent Cochrane reviews have assessed public stewardship interventions not covered in our review; such as social marketing and franchising (Koehlmoos 2009), contracting (Lagarde 2009), and pay for performance (Witter 2012). Koehlmoos 2009 did not find any eligible studies that assessed the effects of social franchising on access to, and the quality of, health services in low‐ and middle‐income countries. We are therefore uncertain about the effects of social franchising as a public stewardship function. Lagarde 2009 examined the effects of contracting out and included three studies, all conducted in low‐ and middle‐income countries. The review found that contracting out services to non‐state not‐for‐profit providers may increase access to and utilisation of health services, improve patient outcomes, and reduce household health expenditures. None of the three included studies presented evidence on whether contracting out was more effective than making a similar investment in the public sector. Witter 2012 found nine studies that assessed the effects of pay‐for‐performance schemes on the provision of healthcare and health outcomes in low‐ and middle‐income countries. The review found that it is uncertain whether pay‐for‐performance improves provider performance, the utilisation of services, patient outcomes or resource use in low‐ and middle‐income countries. Unintended effects of pay‐for‐performance schemes might include adverse selection (for example, excluding high‐risk people from care in order to obtain better performance), deception (i.e. inaccurate or false reporting), and distortion (i.e. ignoring important tasks that are not rewarded with incentives).
Study flow diagram
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Training, Outcome 1 Quality of care.
Comparison 1 Training, Outcome 2 Quality of care.
| Training compared to no training for improving quality of care | |||
| Population: Private for‐profit providers of healthcare services | |||
| Outcomes | Impacts | No of Participants | Certainty of the evidence |
| Quality of care | Both studies show that training probably improves the quality of healthcare services | 486 pharmacies | ⊕⊕⊕⊝ |
| * We downgraded the certainty of evidence by 1 point, because of a moderate risk of selection bias in included studies. GRADE Working Group grades of evidence | |||
| Regulation compared to no regulation for improving quality of care | |||
| Population: Private for‐profit providers of healthcare services | |||
| Outcomes | Impacts | No of Participants | Certainty of the evidence |
| Quality of care | Regulation may improve quality of care. The study observed an increase of 34% in the availability of essential materials for dispensing and an increase of 19% in mean orderliness (including the presence of advertisements, and storage of drugs in their original packaging away from sunlight) in the intervention pharmacies compared to the control pharmacies | 115 pharmacies | ⊕⊕⊝⊝ |
| * We downgraded the certainty of the evidence by 2 points, because of a high risk of attrition bias and wide confidence intervals around the effect estimate, ranging from a large benefit to important harm GRADE Working Group grades of evidence | |||
| Training and regulation compared to no intervention for improving quality of care | |||
| Population: Private for‐profit providers of healthcare services | |||
| Outcomes | Impact | No of Participants | Certainty of the evidence |
| Quality of care | Training and regulation may improve quality of care | 379 pharmacies | ⊕⊕⊝⊝ |
| * We downgraded the certainty of evidence by 2 points because of a high risk of attrition bias and heterogeneity of intervention effects. Two studies found that training, regulation, and peer influence may improve quality of care while the third study found little or no effects in the quality of care with the intervention GRADE Working Group grades of evidence | |||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Quality of care Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 1.1 RCT studies | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.2 Non RCT studies | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 Quality of care Show forest plot | 1 | Odds Ratio (Fixed, 95% CI) | Totals not selected | |
