Customize your JAMA Network experience by selecting one or more topics from the list below.
Lakin JR, Robinson MG, Bernacki RE, et al. Estimating 1-Year Mortality for High-Risk Primary Care Patients Using the “Surprise” Question. JAMA Intern Med. 2016;176(12):1863–1865. doi:10.1001/jamainternmed.2016.5928
Palliative care improves the value of care for seriously ill patients, but resource constraints necessitate targeting palliative care interventions to patients who need them most.1 The “surprise” question (SQ)—“Would you be surprised if this patient died in the next 12 months?”—has emerged as an attractive, simple solution for identifying patients who might benefit from palliative care.2,3 Despite optimism about the potential of the SQ to identify primary care patients who would benefit from palliative care,4 there is no evidence on its performance in this setting.
We identified patients screened for a high-risk care management program at a large academic primary care practice for whom the primary care physicians answered the SQ between August 30, 2012, and February 27, 2014. We assumed a no answer represented physician prediction of high 1-year mortality risk. Our primary outcome was mortality 1 year after SQ response, determined by linkage to Social Security Administration data. We obtained demographics and comorbidities5 from electronic health records. We assessed SQ performance for estimating 1-year mortality using area under the receiver operating characteristic curve, sensitivity, positive predictive value, and odds ratio of a no response for 1-year mortality using univariate logistic regression. To quantify incremental benefit of the SQ for predicting 1-year mortality over and above routinely collected administrative data, we calculated the integrated discrimination improvement6 of adding the SQ response to multivariate logistic regression of mortality on age, sex, and comorbidity score.5 The institutional review board of Partners HealthCare approved this study with a waiver of informed consent. Analysis is based on patient data from medical records and Social Security Administration data (identified records); tabulations, test characteristics, and regression analyses are deidentified data.
A total of 1737 patients were included in the study. Patients were predominantly female (1041 [60.3%]). Mean age was 65 years, and 750 (43.2%) had 3 or more comorbidities. High-risk patients (SQ answer of no) had 4.36 times higher odds of dying than low-risk patients (SQ answer of yes) (95% CI, 2.63-7.22; P < .001; area under the receiver operating curve, 0.57). Table 1 reports the performance of the SQ as a screening test for 1-year mortality. Sensitivity of the SQ was 20.5% and specificity was 94.4%, giving positive and negative likelihood ratios of 3.66 and 0.84, respectively. Given the 1-year mortality rate of 6.6% in this population, positive and negative predictive values were 20.2% and 94.5%, respectively.
In multivariate analysis, a primary care physician prediction of high risk remained more strongly associated with 1-year mortality (odds ratio, 2.52; 95% CI, 1.46-4.34; P = .001; area under the receiver operating curve, 0.79) than age, sex, or comorbidity score (Table 2). Predictive performance of the logistic regression model, however, was not significantly improved by SQ response (integrated discrimination improvement, 0.88; 95% CI, −0.14 to 1.90).
We found that primary care physician prediction of high mortality risk via the SQ failed to identify most deaths at 1 year, making it a poor screening tool for mortality in a heterogeneous primary care population. Adding the SQ response to a validated 1-year mortality prediction model5 did not improve the discriminative ability of that model. More important, even if mortality were wholly predictable, short life expectancy is only 1 of many triggers for palliative care interventions. Prognosis alone does not explicitly account for symptoms and other burdens of serious illness that also indicate a need for palliative care.
Although these results suggest caution in using the SQ in isolation to identify patients with poor prognosis in the primary care setting, the SQ contributes to estimating mortality: it was strongly and significantly associated with 1-year mortality, and this effect was noted over and above known correlates, such as age and comorbidities. These findings are mostly consistent with prior studies2,3 that examined the SQ in renal disease and cancer. Understanding this signal and incorporating it into more advanced predictive algorithms could be useful topics for future research.
Corresponding Author: Joshua R. Lakin, MD, Department of Psychosocial Oncology and Palliative Care, Dana Farber Cancer Institute, 450 Brookline Ave, D-216A, Boston, MA 02215 (firstname.lastname@example.org).
Published Online: October 3, 2016. doi:10.1001/jamainternmed.2016.5928
Author Contributions: Drs Lakin and Obermeyer had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Lakin, Bernacki, Powers, Block, Cunningham, Obermeyer.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Lakin, Robinson, Powers, Cunningham, Obermeyer.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Robinson, Powers, Obermeyer.
Obtained funding: Obermeyer.
Administrative, technical, or material support: Lakin, Robinson, Bernacki, Cunningham.
Study supervision: Lakin, Bernacki, Obermeyer.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Obermeyer is supported by grant DP5 OD012161 from the National Institutes of Health. Part of the analysis was supported by a grant from the National Institute for Health Care Management (Dr Obermeyer and Mr Powers). The development and implementation of the high-risk primary care population management program at Brigham and Women’s Hospital are funded by the Brigham and Women’s Physicians Organization and Brigham and Women’s Hospital.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.