AIM: This study aimed to identify and characterise cross-sectoral co-financing models, their operational modalities, effectiveness, and institutional enablers and barriers.
METHODS: We conducted a systematic review of peer-reviewed and grey literature, following PRISMA guidelines. Studies were included if data was provided on interventions funded across two or more sectors, or multiple budgets. Extracted data were categorised and qualitatively coded.
RESULTS: Of 2751 publications screened, 81 cases of co-financing were identified. Most were from high-income countries (93%), but six innovative models were found in Uganda, Brazil, El Salvador, Mozambique, Zambia, and Kenya that also included non-public and international payers. The highest number of cases involved the health (93%), social care (64%) and education (22%) sectors. Co-financing models were most often implemented with the intention of integrating services across sectors for defined target populations, although models were also found aimed at health promotion activities outside the health sector and cross-sectoral financial rewards. Interventions were either implemented and governed by a single sector or delivered in an integrated manner with cross-sectoral accountability. Resource constraints and political relevance emerged as key enablers of co-financing, while lack of clarity around the roles of different sectoral players and the objectives of the pooling were found to be barriers to success. Although rigorous impact or economic evaluations were scarce, positive process measures were frequently reported with some evidence suggesting co-financing contributed to improved outcomes.
CONCLUSION: Co-financing remains in an exploratory phase, with diverse models having been implemented across sectors and settings. By incentivising intersectoral action on structural inequities and barriers to health interventions, such a novel financing mechanism could contribute to more effective engagement of non-health sectors; to efficiency gains in the financing of universal health coverage; and to simultaneously achieving health and other well-being related sustainable development goals.
METHODS: The review protocol was registered on PROSPERO (CRD42022270039), and reporting followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. Relevant studies were identified through searches in six generic and specialized bibliographic databases, i.e. PubMed, Embase, NHS Economic Evaluation Database, PsycINFO, Health Economic Evaluations Database, tufts CEA registry and EconLit, from their inception to 23 October 2022. The cost-effectiveness of adherence interventions is represented by the incremental cost-effectiveness ratio (ICER). The quality of studies was assessed using the quality of the health economics studies (QHES) instrument. Data were narratively synthesized in the form of tables and texts. Due to the heterogeneity of the data, a permutation matrix was used for quantitative data synthesis rather than a meta-analysis.
RESULTS: Fifteen studies, mostly conducted in North America (8/15 studies), were included in the review. The time horizon ranged from a year to a lifetime. Ten out of 15 studies used a micro-simulation, 4/15 studies employed Markov and 1/15 employed a dynamic model. The most commonly used interventions reported include technology based (5/15), nurse involved (2/15), directly observed therapy (2/15), case manager involved (1/15) and others that involved multi-component interventions (5/15). In 1/15 studies, interventions gained higher quality-adjusted life years (QALYs) with cost savings. The interventions in 14/15 studies were more effective but at a higher cost, and the overall ICER was well below the acceptable threshold mentioned in each study, indicating the interventions could potentially be implemented after careful interpretation. The studies were graded as high quality (13/15) or fair quality (2/15), with some methodological inconsistencies reported.
CONCLUSION: Counselling and smartphone-based interventions are cost-effective, and they have the potential to reduce the chronic adherence problem significantly. The quality of decision models can be improved by addressing inconsistencies in model selection, data inputs incorporated into models and uncertainty assessment methods.