Analyzing implementation dynamics using theory-driven evaluation principles: lessons learnt from a South African centralized chronic dispensing model.

BACKGROUND: Centralized dispensing of essential medicines is one of South Africa's strategies to address the shortage of pharmacists, reduce patients' waiting times and reduce over-crowding at public sector healthcare facilities. This article reports findings of an evaluation of the Chronic Dispensing Unit (CDU) in one province. The objectives of this process evaluation were to: (1) compare what was planned versus the actual implementation and (2) establish the causal elements and contextual factors influencing implementation.

METHODS: This qualitative study employed key informant interviews with the intervention's implementers (clinicians, managers and the service provider) [N = 40], and a review of policy and program documents. Data were thematically analyzed by identifying the main influences shaping the implementation process. Theory-driven evaluation principles were applied as a theoretical framework to explain implementation dynamics.

RESULTS: The overall participants' response about the CDU was positive and the majority of informants concurred that the establishment of the CDU to dispense large volumes of medicines is a beneficial strategy to address healthcare barriers because mechanical functions are automated and distribution of medicines much quicker. However, implementation was influenced by the context and discrepancies between planned activities and actual implementation were noted. Procurement inefficiencies at central level caused medicine stock-outs and affected CDU activities. At the frontline, actors were aware of the CDU's implementation guidelines regarding patient selection, prescription validity and management of non-collected medicines but these were adapted to accommodate practical realities and to meet performance targets attached to the intervention. Implementation success was a result of a combination of 'hardware' (e.g. training, policies, implementation support and appropriate infrastructure) and 'software' (e.g. ownership, cooperation between healthcare practitioners and trust) factors.

CONCLUSION: This study shows that health system interventions have unpredictable paths of implementation. Discrepancies between planned and actual implementation reinforce findings in existing literature suggesting that while tools and defined operating procedures are necessary for any intervention, their successful application depends crucially on the context and environment in which implementation occurs. We anticipate that this evaluation will stimulate wider thinking about the implementation of similar models in low- and middle-income countries.