Evaluating the effectiveness of care coordination interventions designed and implemented through a participatory action research process: Lessons learned from a quasi-experimental study in public healthcare networks in Latin America.

Background: Despite increasing recommendations for health professionals to participate in intervention design and implementation to effect changes in clinical practice, little is known about this strategy's effectiveness. This study analyses the effectiveness of interventions designed and implemented through participatory action research (PAR) processes in healthcare networks of Brazil, Chile, Colombia, Mexico and Uruguay to improve clinical coordination across care levels, and offers recommendations for future research.

Methods: The study was quasi-experimental. Two comparable networks, one intervention (IN) and one control (CN), were selected in each country. Baseline (2015) and evaluation (2017) surveys of a sample of primary and secondary care doctors (174 doctors/network/year) were conducted using the COORDENA® questionnaire. Most of the interventions chosen were based on joint meetings, promoting cross-level clinical agreement and communication for patient follow-up. Outcome variables were: a) intermediate: interactional and organizational factors; b) distal: experience of cross-level clinical information coordination, of clinical management coordination and general perception of coordination between levels. Poisson regression models were estimated.

Results: A statistically significant increase in some of the interactional factors (intermediate outcomes) -knowing each other personally and mutual trust- was observed in Brazil and Chile INs; and in some organizational factors -institutional support- in Colombia and Mexico. Compared to CNs in 2017, INs of Brazil, Chile, Colombia and Mexico showed significant differences in some factors. In distal outcomes, care consistency items improved in Brazil, Colombia and Uruguay INs; and patient follow-up improved in Chile and Mexico. General perception of clinical coordination increased in Brazil, Colombia and Mexico INs. Compared to CNs in 2017, only Brazil showed significant differences.

Conclusions: Although more research is needed, results show that PAR-based interventions improved some outcomes regarding clinical coordination at network level, with differences between countries. However, a PAR process is, by definition, slow and gradual, and longer implementation periods are needed to achieve greater penetration and quantifiable changes. The participatory and flexible nature of interventions developed through PAR processes poses methodological challenges (such as defining outcomes or allocating individuals to different groups in advance), and requires a comprehensive mixed-methods approach that simultaneously evaluates effectiveness and the implementation process to better understand its outcomes.