This is a follow up of my blog post couple years ago on this topic.

Positive deviance is a social change strategy that has been used successfully to solve seemingly intractable and complex social and public health problems. It is based on the observation that there are certain individuals or groups (positive deviants) within a community, whose uncommonly positive behaviors and strategies enable them to find better solutions to problems than their peers, while having similar access to resources. Applying these principles and lessons from public health in health care infection prevention is both interesting as well as tricky.

We tested the effectiveness of this approach in improving patient safety culture and preventing healthcare-associated infections (HAI) in hospital wards. It was the most complicated research study I ever undertook. Our paper reporting this study is available online. Our team was very diverse. Dr. Lucia Dura has expertise in communications and positive deviance. Maria Eva Fernandez was our research coordinator who also conducted a large proportion of discovery and action dialogues. Dr. Kristina Simacek is a sociologist who oversaw the social network analyses and interpretation. Dr. Tom Fomby is an economics professor with expertise in time series modeling. Dr. Abu Minhajuddin performed the remaining statistical analyses. I originally met Dr. Brad Doebbeling on a collaborative engaged in gathering qualitative challenges in implementation. We wrote four grant applications, out of which one was funded. As fate would have it, the start of the grant coincided with a failed regulatory survey by the Centers for Medicare and Medicaid Services at our study institution.

We conducted an observational study in six medical wards at Parkland Memorial Hospital, an 800-bed public academic safety-net hospital in the United States. In three of these wards, we conducted positive deviance inquiry among the healthcare personnel for nine months. Three wards of a similar size serving medical inpatients served as control and did not receive the intervention. We measured the monthly rates of healthcare-associated infections during the six-month period before start of intervention for baseline. We prospectively measured the monthly rates of HAI in all wards during the nine months of positive deviance intervention and for nine months after the intervention ended. We surveyed patient safety culture and social networks among healthcare personnel at 6, 15, and 24 months.

We were somewhat surprised by the results. The study was, without a doubt, conducted at a time when there a great deal of organizational stress. What I learned during the positive deviance inquiry or ‘discovery and action dialogues’ was eye-opening. Even though the wards receiving intervention and not receiving intervention were randomized, they did not have similar patient safety culture per the survey. We were surprised by the significant differences at baseline. The wards in which the personnel received positive deviance maintained their measured level of positive safety culture, whereas the control wards experienced a significant decline. Social network maps suggested that nurses, charge nurses, medical assistants, ward managers, and ward clerks served as nodes of information exchange. It was something that made sense in hindsight. The infection rates declined in both groups of wards, most likely because of secular trends from aggressive hand hygiene interventions we implemented in response to the failed regulatory surveys.

We discussed the lessons learned in our paper. The staff members and unit managers we interviewed remember us even today. They modeled complex workflows to each other and came up with several great ideas. They seemed to lean on external help to sort the ideas and channel them forward. One staff nurse who had been around for two years paced back and forth past me a few times before gathering the courage to ask me where she can find the policies. That was a particularly ouch moment for me because we had pushed several educational initiatives throughout the system in the preceding months. One of the patient care assistants in the intervention wards who came up with a detailed list of skills that all personnel must possess in order to prevent infections effectively, made his peers jealous. It was an interesting problem because positive deviance approach seemed to depend on whether the positive outliers are accepted by their peers or not. The group seemed to want to embrace the new behavior demonstrated by the positive outlier only if they accepted the person and didn’t become jealous. This probably means that groups that recognize their positive outlier peers and embrace better behaviors demonstrated by them have a better chance of improving their outcomes. So, what came first, the positive culture or the positive deviant? They likely have an amplifying effect on each other, but it’s great food for thought.

My own takeaway from the thousand odd hours of thinking and working hard on this project is that positive deviance is yet another tool in a change maker’s armamentarium. One of the big lessons I learned in thirteen years as a healthcare epidemiologist charged with improving rates of healthcare-associated infections is that any intervention intended to improve delivery of healthcare requires adaptation to local social and cultural conditions. I reviewed these social adaptation approaches in this article available at It is a summary of my recent grand rounds talk at my institution. The science on best ways to adapt a particular intervention to the local organizational context is still evolving. Before planning to ‘socialize’ an intervention, it is particularly important for change leaders to arrive at a correct diagnosis of their organizational context. The ‘problem-solving journey’ continues.