A well-known problem in research is reporting bias, or paying attention to interesting results while ignoring other findings. The risk of reporting bias is higher when results are difficult to measure — when, for example, research subjects are asked how they feel or researchers are making subjective observations of behavior.
Well aware of the problem, Daniel Tawfik, MD, a Stanford instructor in pediatrics, wanted to look at whether there’s a correlation between burnout in care providers and poor care quality.
Such research may be especially prone to reporting bias: Measuring how burned out someone is feeling typically involves survey questions, which are subjective measures. The people answering the surveys are also self-selecting, which can skew the results.
Measuring quality of care is difficult as well: You can record the length of a patient’s hospital stay, for example, but it’s impossible to pinpoint who’s to blame when a dozen providers are responsible for the patient.
So Tawfik and his research colleagues conducted a meta-analysis: They evaluated all the studies they could find on burnout and care quality, 123 of them. “We wanted to figure out how much reporting bias there was,” Tawfik said.
They pored over the studies, which measured burnout among health care providers — mostly nurses and physicians, but also pharmacists, respiratory therapists and others — and compared them with the quality of care their patients received. They also paid attention to potential bias, considering the objectivity of measurements and other factors that could sway results.
The researchers found some reporting bias, but they also found a connection. As they write in an article in the Annals of Internal Medicine, when providers are feeling burned out, the quality of care patients receive is likely to be reduced.
Tawfik cautions against presuming a cause: “The more intuitive way of thinking is that burnout leads to poor care,” he said. “But it can go in the other direction.”
When a patient isn’t getting the highest quality care, that causes a lot of emotional work, which can result in emotional exhaustion or depersonalization. That can happen with small events over time, or with a single, large significant event.
So, how can we determine what causes what? For researchers who study the effects of a medication, the gold standard is to conduct randomized, controlled trials. But, as Tawfik notes, “You can’t randomly allocate burnout like you can a medication.”
Longitudinal studies, which look at which comes first — burnout or poor care — could help answer the question, Tawfik said, though they are expensive and, obviously, take a long time. He’s hoping to clarify our understanding about burnout by developing more objective ways of identifying and following it, such as measuring cortisol, a stress hormone.
Of course, the answer may be that burnout causes poor quality, and poor quality causes burnout: “There have been some studies that seem to show it going in both directions,” Tawfik said.