The study by Pereira-Lima et al was a well-intentioned effort to shed more light on the potential relationship between physician depression and medical errors. Unfortunately, the authors (and perhaps copy editors), by inaccurately conflating key terms such as “medical errors” and “perceived errors” and “depressive symptoms” and “depression” throughout the article, may have inadvertently muddied the field and compounded the confusion regarding this critical issue.
As a result of imprecise use of terms (especially in the title) by these researchers, both lay and popular medical news writers attempting to publicize scientific findings have already widely disseminated the attention-grabbing (but misleading) headline, “Physician depression leads to medical errors.”
The overwhelming majority of the studies analyzed in the study (91%) measured “perceived” or self-reported, as opposed to objective measures of error. Those physicians reporting errors also reported experiencing depressive symptoms. Physicians experiencing features of depression are more likely to recall negative events and to view themselves in a negative light and are therefore more likely to retrospectively assess a clinical event with a suboptimal outcome as an error, and to blame themselves for that error. Such self-reported errors thus may reflect self-judgment and willingness to report, rather than actual differences in objective error rates. This important distinction was noted in the limitations section of the study but was unfortunately completely lost in the titling and throughout much of the article. Only 101 of the 21,517 participants had any kind of external identification of error. In the other 21,416 participants, the existence of error was assumed from retrospective self-report.
This unfortunate conflation of perceived and actual errors was then “justified” in the limitations section by reference to a single study purporting to show that “self-reported errors have been found to be highly correlated with recorded events.” That study, a resident peer survey of purported errors, involved residents reporting on all errors observed on a service, and not personal errors; it was no justification at all for the contention.
The studies included in the meta-analysis used several population-based screens for depressive symptoms, more than half of which are highly sensitive but not specific for a diagnosis of depression. Most of these screens were not subsequently followed by any definitive testing or clinical interviews to establish a diagnosis of depression.
Yet despite this deficiency, throughout the paper, the terms “depression” and “depressive symptoms” are regularly conflated. This confuses the issue of whether the physicians being surveyed (most of them trainees) met clinical criteria for depression, or whether they were instead experiencing depressive symptoms related to burnout, exhaustion, poor sleep or high stress common in residents. Indeed, in the one study where burnout and depression were formally measured, the number of residents with diagnosed depression was rather small relative to those with defined burnout.
Further, while depression screens used in the majority of the studies assess recent symptoms (typically several weeks), elicitation of self-error reports typically covered much longer periods (3 to 12 months). Therefore, the two phenomena being collated (depressive symptoms and self-reported perceived errors) may not even have occurred coincidentally.
Although it is undeniable, as opined by the authors, that “a reliable estimate of the degree to which physicians with a positive screening for depression are at higher risk for medical errors would be useful,” this analysis did not achieve that goal. Instead, it made an estimate as to how often physicians with self-reported depressive symptoms self-report perceived errors or even risk of errors. Such estimates may or may not relate to the actual incidence of errors occurring in depressed physicians.
Last but certainly not least, it was disappointing to see yet again the solidly discredited Institute of Medicine and Makary papers regarding the purported rampancy of generic adverse events in hospitalized patients recycled in support of studies whose focus was to be physician medical errors. Such indiscriminate propagation of erroneous memes as happened with these misapplied statistics in the popular press will now be applied to a new misperception of the roots of the problems of adverse events and medical error.
Indeed, in just 1 week following the release of this study, there have already been at least a dozen iterations of the false meme, “Depressed doctors make more mistakes.”
Although I agree with Dr. Privitera and no doubt the study authors that depressive symptoms in doctors must be addressed in every conceivable supportive and nonpunitive way, in my opinion, in the current regulatory environment, providing fuel for the propagation of salacious memes inciting public fears about the dangers of depressed doctors seems unlikely to help to achieve this end.
Emmons RS, et al. J Am Physicians and Surgeons. 2018; https://www.jpands.org/vol24no2/emmons.pdf.
Fahrenkopf AM, et al. BMJ. 2008;doi:10.1136/bmj.39469.763218.BE.
Institute of Medicine Committee on Quality of Health Care in America; Kohn LT, et al. ”To Err Is Human: Building a Safer Health System” Vol 6. Washington, DC: National Academies Press; 2000.
Mazer BL, et al. J Gen Intern Med. 2019;doi:10.1007/s11606-019-05156-7.
Makary MA, et al. BMJ. 2016;doi:10.1136/bmj.i2139.
Shojania KG, et al. BMJ Qual Saf. 2016;doi:10.1136/bmjqs-2016-006144.
Disclosure: Andrew reports no relevant financial disclosures.
Louise B. Andrew, MD, JD
Founder and principal consultant