This article is more than 5 years old. Information may no longer be current.

Empiric assessment of mortality risk 'a poor screening tool' in primary care

A question based on a clinicians’ judgment designed to identify patients for palliative care failed to predict mortality in a primary care population, according to findings published in JAMA Internal Medicine.

Joshua R. Lakin, MD, from Dana Farber Cancer Institute, and colleagues, noted, however, that the question aided in estimating mortality.

"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," Lakin and colleagues wrote. “The ‘surprise’ question — ‘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."

The researchers attempted to quantify evidence for the use of the surprise question in primary care. They identified 1,737 patients who had been previously screened for inclusion in a high-risk care management program. Primary care physicians answered the surprise question about these patients between August 2012 and February 2014. The researchers used Social Security Administration data to determine mortality and electronic health records to collect comorbidity and demographic data.

Results showed that in cases where the clinician answered ‘No,’ patients were 4.36 times more likely to die than in cases where the clinician answered ‘Yes’ (95% CI, 2.63-7.22; P < .001).

After 1 year, Lakin and colleagues reported the following:

• 23 patients were correctly predicted to die
• 91 patients were incorrectly predicted to die
• 89 patients were incorrectly predicted to live; and
• 1,534 patients were correctly predicted to live.

This resulted in a sensitivity of 20.5% and specificity of 94.4%.

In addition, they found that physician prediction of risk was "more strongly associated" with mortality at 1 year (OR = 2.52; 95% CI, 1.46-4.34; P = .001) than scores for comorbidities, age or sex.

The researchers acknowledged that the surprise question was "a poor screening tool for mortality in a heterogeneous primary care population," as physician risk prediction was unsuccessful in identifying the majority of deaths.

"Although these results suggest caution in using the [surprise question] in isolation to identify patients with poor prognosis in the primary care setting, the [surprise question] 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," Lakin and colleagues concluded. "These findings are mostly consistent with prior studies that examined the [surprise question] in renal disease and cancer. Understanding this signal and incorporating it into more advanced predictive algorithms could be useful topics for future research." – by Chelsea Frajerman Pardes

Disclosure: The authors report no relevant financial disclosures.