Clinical Effect of Burnout Difficult to Measure: Meta-analysis

Tara Haelle

October 07, 2019

Though the effect size may be smaller than what previous studies have found, a recent meta-analysis of studies on burnout leaves little doubt about the association of burnout with poorer quality care. The systematic review, published online October 7 in the Annals of Internal Medicine, assessed potential bias in more than a hundred studies employing a wide range of burnout measures and quality-of-care outcomes.

The study's primary purpose was "to inform the burnout research community and provide recommendations in how to move the research forward in this field," but it has clinical implications, too, lead author Daniel S. Tawfik, MD, MS, an instructor of pediatric critical care medicine at the Stanford University School of Medicine in Palo Alto, California, told Medscape Medical News.

He found it "particularly striking" that "even though there are myriad different ways of measuring quality care outcomes, the majority" show some relationship between burnout and quality of care.

"That relationship seems to span a wide variety of quality-of-care domains," Tawfik said. "Even if there is some excess in the literature and some exaggerated effects, even based on the most rigorous studies, there does appear to be at least a moderately strong effect."

The researchers searched MEDLINE, PsycINFO, Health and Psychosocial Instruments (EBSCO), Mental Measurements Yearbook (EBSCO), Embase (Elsevier), and Web of Science for all peer-reviewed studies published in any language through May 28, 2019, that assessed healthcare provider burnout and its association with quality of care. Researchers identified 123 studies, published from 1994 to 2019, involving 241,553 healthcare providers. Half the studies involved physicians, 59% involved nurses, and 13% involved other providers. (Some included multiple provider types.)

The studies covered a wide range of care types, care settings, provider types, and patient ages. Burnout measures were classified as overall burnout, emotional exhaustion, depersonalization, or low personal accomplishment.

For each study, researchers measured two potential biases in the studies: excess significance using the Ioannidis test and small study effects using the Egger test.

These assessments revealed excess significance across the whole sample: although 62% were predicted to have statistically significant results, 73% of studies actually did (P = .011). "This indicator of potential bias was most prominent for the least-rigorous quality measures of best practices and quality and safety," the authors note. The Egger test revealed no effect overestimation from small studies, though.

The studies included subjective and objective quality-of-care metrics, including medical malpractice allegations and reports from patients, providers, or health systems. Quality of care fell into five categories: best practices (n = 14 studies), communication (n = 5), medical errors (n = 32), patient outcomes (n = 17), and quality and safety (n = 74).

The authors acknowledge subjective metrics to be more susceptible to bias, especially recall bias.

Objective metrics, likewise, have their own limitations, noted Albert W. Wu, MD, director of the Johns Hopkins Center for Health Services and Outcomes Research and faculty at the Johns Hopkins Armstrong Institute for Patient Safety and Quality in Baltimore, Maryland.

"It is inherently difficult to measure quality of care, and it's particularly difficult to measure safety since unsafe care is hard to capture," Wu, who was not involved in the study, told Medscape Medical News. Though patient outcomes may be easier to measure objectively, many of them are uncommon or "not necessarily attributed to safety problems or quality."

Tawfik agreed, adding that it is difficult to detect when, for example, a provider does the right thing for the wrong reason, or catches themselves and rights an error immediately, or a medical error occurs that doesn't result in harm.

The findings reflect the studies' heterogeneity. Among 114 burnout-quality combinations, 58 found an association with poor-quality care, 50 found no significant effect, and six found an association with high-quality care.

Most of the 50 showing no effect, Tawfik said, were "pointing in the same direction" as the 58. "I wouldn't take them as proof that there's no effect, but they failed to show an effect," he said, adding that quite a few were underpowered to demonstrate a significant effect.

Despite the finding that effect sizes may be smaller than previously reported, "any effect on patient safety is unacceptable and merits some sort of remedy," regardless of the effect's magnitude, Tawfik explained. Still, important practical reasons exist for knowing the strength of effects.

"If researchers are designing studies based [on] a faulty estimate of what's out there, they may design underpowered studies," Tawfik said. Intervention studies could also suffer if designed based on inappropriate expectations of effect sizes.

Further, hospitals have limited resources when looking for ways to improve quality of care. "They would love to address every issue, but they need to address what's most pressing an issue. This can help them figure out where the relationship of burnout and quality of care falls in relation to other aspects of healthcare delivery that affect quality of care."

But he doesn't expect hospitals to reduce resources for physician burnout based on these findings.

"First, even if it's not as bad as people thought it was, it's still really bad. Even the lower estimates of what that relationship is still show there's a strong relationship between burnout and quality of care," Tawfik said. Second, he said, many other reasons exist to address burnout, including associations with substance abuse, suicidality, relationship problems, morale, and financial implications relating to reduced patient satisfaction.

Yet the authors did not include patient satisfaction as an "important outcome," stating that "it is not consistently indicative of care quality or improved medical outcomes" and may arise from "factors outside the provider's immediate control, such as facility amenities and access to care."

That omission is unfortunate, Wu suggested.

"I would be really interested in knowing what the impact of burnout was on patient satisfaction," Wu, also a practicing internist and a professor in the Johns Hopkins Bloomberg School of Public Health, explained. "They argued that patient satisfaction is not indicative of healthcare quality. I would argue that's not true, that it's indicative of quality from the patient's perspective."

He said patient satisfaction measures are reliable and typically show less error than other outcomes included in the study.

"Many of the effects of burnout are not just on performance and functioning but on attitude and compassion, not so much on what you do to care for people, but how much you care about what you're doing," Wu said. "If you're having compassion fatigue and have become cynical, and you don't have enough energy to do all the things you need to do, some of the things more likely to go by wayside are caring and comforting" — exactly the kinds of effects that would show up in patient satisfaction surveys.

Whether satisfaction is included or not, the authors said their work revealed the need to come up with at least some consistent metrics.

"If you're trying to answer the same question as another researcher and you're all using different ways of measuring that outcome, it creates difficulty in trying to synthesizing all that evidence in the future," Tawfik said. "The more we can distill it down to a key set of outcomes that we can all agree on, the more we can design studies to intervene with burnout and to improve quality of care."

Since the studies were observational, they cannot show causation or directionality. Researchers and previous studies have suggested a bidirectional influence, and these authors suggested a multifactorial relationship.

"Providers who have burnout may have less time or commitment to optimize the care of their patients, may take more unnecessary risks, or may be unable to pay attention to necessary details or recognize the consequences of their actions," the authors note. "Conversely, exposure to adverse patient events or recognition of poor-quality care may result in emotional or other psychological distress among providers."

The latter, often called secondary trauma, typically occurs due to "sentinel events or important safety incidents" but can also result from "repeated minor incidents," the authors write.

Delivering high-quality healthcare is not easy, Wu said.

"It is stressful and technically and emotionally difficult, and it's in everyone's interest to have the workforce be as healthy and happy as they can possibly be," he said. "We could use as many good, happy, healthy physicians as we can get."

The research was funded by the Stanford Maternal and Child Health Research Institute. Study author Jochen Profit, MD, MPH, has received honoraria for speaking about burnout at scientific meetings. Tawfik, the remaining study authors, and Wu have disclosed no relevant financial relationships.

Ann Intern Med. Published online October 7, 2019. Abstract

Follow Medscape on Facebook, Twitter, Instagram, and YouTube.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.