End-of-Life Care Intensity for Physicians, Lawyers, and the General Population

Care at the end of life must balance intensity of treatment with quality of life. Previous research has not determined whether physicians, the group most familiar with end-of-life care, receive higher or lower intensity end-of-life treatments compared with nonphysicians.¹

Non–health maintenance organization Medicare beneficiaries aged 66 years or older who died between 2004 and 2011 in Massachusetts, Michigan, Utah, and Vermont were included due to availability of electronic death records and ability to link to Medicare. The Partners human research committee determined the study was not human research.

Death records were not available for Massachusetts in 2011 and Utah for 2004 through 2005, and approximately 10% of decedents could not be linked to Medicare claims. Death certificates documented age, education, Social Security number, and usual industry and occupation (the type of job engaged in for most of his/her working life). Industry and occupation are typically provided by family members to the funeral director and are highly accurate.²

From Medicare records, we obtained data on 5 validated measures of end-of-life care intensity during the last 6 months of life: surgery,³ hospice care,⁴ intensive care unit (ICU) admission; death in the hospital; and expenditures.⁵ Measures were compared between physicians and the general population (excluding other health care workers and lawyers), physicians vs lawyers, who are presumed to be socioeconomically and educationally similar, and lawyers vs the general population.

For unadjusted analyses, Pearson χ² tests of proportions or robust t tests with unequal variances were used. Adjusted analyses were performed for decedents with at least a college education. For dichotomous outcomes, we used logistic regression to calculate percentages adjusting for correlates of end-of-life care intensity: sex, age-adjusted Charlson Comorbidity Index, race/ethnicity, and hospital referral region. For costs, robust linear regression was used.⁶ Statistical analyses were performed with SAS version 9.3 (SAS Institute Inc). A 2-sided P < .05 was considered significant.

There were 2396 deceased physicians, 2081 lawyers, and 665 579 in the general population (Table 1). Physicians and lawyers were more likely to be male and college educated than the general population. In adjusted analyses, physicians were less likely to die in a hospital compared with the general population (27.9% vs 32.0%, respectively; P < .001), less likely to have surgery (25.1% vs 27.4%; P = .01), and less likely to be admitted to the ICU (25.8% vs 27.6%; P = .04) (Table 2). Physicians were less likely to die in a hospital compared with lawyers (27.9% vs 32.7%, respectively; P < .001) but did not differ significantly on other measures.

For 3 of 5 end-of-life care intensity measures, physicians received significantly less intensive care than the general population. Although not large, the differences suggest less aggressive care for physicians. The possible reasons physicians received less intense end-of-life care than others could be knowledge of its burdens and futility as well as the benefits and the financial resources to pay for other treatment options, such as palliative care or skilled nursing required for death at home.

In contrast, only 1 measure differed between physicians and lawyers. Economic resources and education may therefore contribute to end-of-life-care decisions. However, the lower rates of hospital deaths of physicians compared with lawyers suggest that while resources may promote home deaths, actual experience with hospital deaths may differentially motivate physicians to avoid them.

Study limitations include the use of retrospective data and omission of unmeasured confounders, particularly income, although this is partially addressed by controlling for hospital referral region and making comparisons with lawyers. Second, this study does not address complex decision-making processes nor satisfaction with end-of-life experiences. Third, we could not identify physicians’ specialty or professional experience with end-of-life care. Fourth, death data from some states for selected years were not available. Fifth, data were available on only a limited number of states.

These findings could inform how health professionals communicate with patients about end-of-life care choices. For example, family members of critically ill patients sometimes seek reassurance from physicians that their loved one is receiving the same type of care that a physician would receive.

Corresponding Author: Joel S. Weissman, PhD, Center for Surgery and Public Health, Department of Surgery, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St, Ste 4-020X, Boston, MA 02120 (jweissman@partners.org).

Author Contributions: Dr Weissman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Weissman, Cooper, Hyder, Lipsitz, Zinner, Prigerson.

Acquisition, analysis, or interpretation of data: Weissman, Cooper, Hyder, Lipsitz, Jiang, Prigerson.

Drafting of the manuscript: Weissman, Cooper, Lipsitz.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Lipsitz, Jiang.

Administrative, technical, or material support: Weissman, Cooper.

Study supervision: Weissman, Lipsitz, Zinner.

Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Additional Contributions: We thank Bruce Cohen, PhD (Massachusetts Department of Public Health), for help in acquiring state death data and overall guidance, and Nathanael Hevelone, MPH (Covidien/Medtronic), and Richard Sterling Haring, DO (Brigham and Women’s Hospital), for contributions to the construction of the database. We are indebted to the departments of public health in Massachusetts, Vermont, Michigan, and Utah that provided the electronic death certificate data and to Ellen Goodman (Institute for Healthcare Improvement, the Conversation Project) for advice early in the research process. No compensation was provided for any of the persons involved.