Doctors' Depression, Medical Errors Go Hand in Hand

Med Page Today
30 Nov, 2019 ,

Physicians with signs of depression made more medical errors, and those who reported more medical errors also had more depressive symptoms later on, according to a meta-analysis.

Source
Full content

Across 11 surveys of mostly early-career physicians, depressive symptoms were associated with nearly twice the rate of self-reported medical errors, like prescribing the wrong medication (RR 1.95, 95% CI 1.63-2.33).

The association appeared bidirectional, reported Karina Pereira-Lima, PhD, of the University of Michigan Medical School in Ann Arbor, and colleagues in JAMA Network Open.

Reported medical errors were associated with subsequent depressive symptoms in four studies (RR 1.67, 95% CI 1.48-1.87). Depressive symptoms were associated with subsequent medical errors in six studies (RR 1.62, 95% CI 1.43-1.84).

"These findings underscore the need for us to do systematic efforts to prevent and reduce depressive symptoms among physicians," Pereira-Lima told MedPage Today. "This is not only important to physician health, but also impacts the quality of care they are presenting to patients."

Early-career physicians have been shown to have higher rates of depression than the general population, and this appears to particularly afflict medical students and residents, with an estimated prevalence of close to 30% among the latter.

Medical errors committed by clinicians may contribute to feelings of guilt or depression, which can be compounded by the threat of legal action, according to Thomas Rodziewicz, MD, of Michigan State University in East Lansing, and John Hipskind, MD, of Kaweah Delta Medical Center in Visalia, California, writing in a recent eBook on medical error prevention.

"Governmental, legal, and medical institutions must work collaboratively to remove the culture of blame while retaining accountability," they wrote. "When this challenge is met, health care institutions will not be constrained from measuring targets for process improvement, including all errors, even with adverse outcomes."

The majority of the studies included in this meta-analysis took place in the U.S. (82%), with one conducted in Japan and another in South Korea.

Notably, 73% involved only physicians in training. Two were conducted among internal medicine residents, one surveyed pediatric residents, and another involved anesthesiology residents. The rest recruited doctors from all specialties and at any career level.

Medical errors were evaluated through self-reported measures in all but one study, and about three-quarters of the studies only asked about medical errors that had occurred within the past 3 months (73%). One asked if clinicians made a harmful error with negative consequences for the patient; another asked if they committed any type of medical error. In one study, researchers reviewed charts to determine the rate of medical errors.

The studies involved in the primary analysis were highly heterogeneous (χ2=49.91), but a sensitivity analysis showed no individual study impacted the estimation of errors by more than 0.12 points.

On the other hand, the studies used to determine the direction of the associations were not very heterogeneous (χ2 = 5.77), they noted.

Moreover, the authors stratified the data by a variety of characteristics to determine which factors may be playing a role in the association detected, and found that the association held through each analysis.

For example, longitudinal studies were associated with lower risk ratios than cross-sectional studies (RR 1.62 vs 2.51), but the association was observed in both study designs.

Surgical specialties also showed higher risk ratios than nonsurgical specialties, as did non-U.S. countries versus the U.S.

The primary limitation of this study was its reliance on self-reported measures, the authors noted. The studies involved also asked general questions about "major, harmful, or any" medical errors, which may have limited the reliability of participants' recall. The sample sizes in some of the subgroup analyses were also relatively small, they added.