A network study exploring factors that promote or erode interaction among diverse community health workers in rural Ethiopia.

Task shifting in response to the health workforce shortage has resulted in community-based health workers taking on increasing responsibility. Community health workers are expected to work collaboratively, though they are often a heterogeneous group with a wide range of training and experience. Interpersonal relationships are at the very core of effective teamwork, yet relational variables have seldom been the focus of health systems research in low resource, rural settings. This article helps fill this knowledge gap by exploring the dyadic level, or relational, characteristics of community maternal and newborn health workers and the individual and collective influence of these characteristics on interaction patterns. Network data were collected from community health workers (N = 194) in seven rural kebeles of Amhara region, Ethiopia from November 2011 to January 2012. Multiple Regression Quadratic Assignment Procedure was used to fit regression models for frequency of work interactions, a proxy for teamwork. Strong and consistent evidence was found in support of Trust and Past training together as important relational factors for work interactions; less consistent evidence was found across sites in support of Homophily, Distance and Shared motivations. Our findings also point to a typology of network structure across sites, where one set of networks was characterized by denser and stronger health worker ties relative to their counterparts. Our results suggest that the development of interventions that promote trust and incorporate cross-cadre training is an important step in encouraging collective action. Moreover, assessing the structure of health worker networks may be an effective means of evaluating health systems strengthening efforts in rural, low-resource settings.