Willem A Odendaal, Kim Ward, Jesse Uneke, Henry Uro‐Chukwu, Dereck Chitama, Yusentha Balakrishna, Tamara Kredo

Contracting out of governmental health services is a financing strategy that governs the way in which public sector funds are used to have services delivered by non-governmental health service providers (NGPs). It represents a contract between the government and an NGP, detailing the mechanisms and conditions by which the latter should provide health care on behalf of the government. Contracting out is intended to improve the delivery and use of healthcare services. This Review updates a Cochrane Review first published in 2009. To assess effects of contracting out governmental clinical health services to non-governmental service provider/s, on (i) utilisation of clinical health services; (ii) improvement in population health outcomes; (iii) improvement in equity of utilisation of these services; (iv) costs and cost-effectiveness of delivering the services; and (v) improvement in health systems performance. We searched CENTRAL, MEDLINE, Embase, NHS Economic Evaluation Database, EconLit, ProQuest, and Global Health on 07 April 2017, along with two trials registers - ClinicalTrials.gov and the International Clinical Trials Registry Platform - on 17 November 2017. Individually randomised and cluster-randomised trials, controlled before-after studies, interrupted time series, and repeated measures studies, comparing government-delivered clinical health services versus those contracted out to NGPs, or comparing different models of non-governmental-delivered clinical health services. Two authors independently screened all records, extracted data from the included studies and assessed the risk of bias. We calculated the net effect for all outcomes. A positive value favours the intervention whilst a negative value favours the control. Effect estimates are presented with 95% confidence intervals. We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach to assess the certainty of the evidence and we prepared a Summary of Findings table. We included two studies, a cluster-randomised trial conducted in Cambodia, and a controlled before-after study conducted in Guatemala. Both studies reported that contracting out over 12 months probably makes little or no difference in (i) immunisation uptake of children 12 to 24 months old (moderate-certainty evidence), (ii) the number of women who had more than two antenatal care visits (moderate-certainty evidence), and (iii) female use of contraceptives (moderate-certainty evidence).The Cambodia trial reported that contracting out may make little or no difference in the mortality over 12 months of children younger than one year of age (net effect = -4.3%, intervention effect P = 0.36, clustered standard error (SE) = 3.0%; low-certainty evidence), nor to the incidence of childhood diarrhoea (net effect = -16.2%, intervention effect P = 0.07, clustered SE = 19.0%; low-certainty evidence). The Cambodia study found that contracting out probably reduces individual out-of-pocket spending over 12 months on curative care (net effect = $ -19.25 (2003 USD), intervention effect P = 0.01, clustered SE = $ 5.12; moderate-certainty evidence). The included studies did not report equity in the use of clinical health services and in adverse effects. This update confirms the findings of the original review. Contracting out probably reduces individual out-of-pocket spending on curative care (moderate-certainty evidence), but probably makes little or no difference in other health utilisation or service delivery outcomes (moderate- to low-certainty evidence). Therefore, contracting out programmes may be no better or worse than government-provided services, although additional rigorously designed studies may change this result. The literature provides many examples of contracting out programmes, which implies that this is a feasible response when governments fail to provide good clinical health care. Future contracting out programmes should be framed within a rigorous study design to allow valid and reliable measures of their effects. Such studies should include qualitative research that assesses the views of programme implementers and beneficiaries, and records implementation mechanisms. This approach may reveal enablers for, and barriers to, successful implementation of such programmes.