Response to “Impact of HIT on burnout remains unknown – for now”

Dear Editors,

We appreciate both the expertise and the passion that Zsenits and colleagues bring to the conversation about electronic health records (EHRs) and clinician burnout. We hope that by responding to some of the points in their letter that we can provide clarification and perhaps some reassurance.

First, this was not a case-control study, nor did we test for correlation. The Department of Health samples all of the physicians in the state for public reporting purposes. Our study population was not generated by identifying cases and controls; rather, we categorized physicians’ responses into relevant groups for the planned analysis. We used multivariable regression to identify which covariates were independently associated with burnout symptoms among respondents. The term “predictor” can be used in cross-sectional studies in the biomedical sciences without implying causation; in this context, it serves as a synonym for an independent variable.^1,2 Fortunately, we were able to include many clinician and practice characteristics in the model; as with any observational study, we were not able to include unmeasured factors, nor would we attempt to determine causation. Dr Guyatt's evidence-based papers³ are superb, but the comparison studies the authors speak of are not comparable to the study design in our paper. They would apply more to longitudinal cohorts or randomized trials.

Second, we understand Zsenits and colleagues’ frustration with how our article might be perceived as “fuel[ing] a self-fulfilling negative attitude toward EHRs” among physicians. We are neither champions nor disparagers of EHRs. Our work seeks to illuminate the experience of clinicians actively using the technology to care for patients. We hope that, by exploring factors associated with stress and burnout, we can provide information that will facilitate change to the benefit of physicians and their patients. We appreciate the authors’ suggestions for “trainers, personalization, and trusted and agile improvement processes.” These would surely help address some of the challenges identified by our respondents.

We respectfully disagree with the critique by Zsentis and colleagues, and we have confidence in the methodology and the findings of our paper. The next iteration of the survey will more systematically capture strategies that clinicians identify as reducing technology-related stress. We look forward to reporting those findings in the future.