Older black patients are much more likely than whites to be readmitted to the hospital after a hip or knee replacement. The reasons are complex, but more attention from nurses may help address the problem, a new study from the University of Pennsylvania School of Nursing's Center for Health Outcomes and Policy Research suggests.

Readmission rates matter both because no one wants to go back into the hospital, and because the rates are a Medicare quality measure. Hospitals are penalized financially when too many of their patients return after they are sent home.

The new study may be an argument for "precision nursing" or tailoring care to specific patients' needs, said study author Karen Lasater, a postdoctoral fellow at the research center.

Readmission rates for black patients were 40 percent higher than for whites after controlling for factors that the Centers for Medicare and Medicaid Services uses for risk adjustment, including patients' other illnesses, Lasater said. They were still 30 percent higher when her team controlled for the socio-economic status of patients and the financial health of hospitals where they got their care.

On average, 7.5 percent of black joint-replacement patients were readmitted, compared with 5.6 percent of white peers. The study found that each additional patient assigned to a nurse increased the odds of readmission by 8 percent for white patients and by 15 percent for black patients.

The study found a correlation between staffing and readmissions, but it did not prove that staffing caused the differences.

Black patients were more likely to receive care in large, urban, nonprofit hospitals that are not academic medical centers. Notably, the hospitals were more likely to have nurses with relatively low caseloads - indicating that mere numbers may not be the issue. In the study, which was based on Medicare records and surveys of nurses, hospital nurse staffing ranged from 2.9 to 9.8 patients per nurse. The average was 5 patients per nurse.

"Even in the context of good staffing levels," Lasater said, "we still find there is a disparity in outcomes."

She did not have information about staffing ratios on the particular units where joint-replacement patients stayed. Nor, she said, is there scientific information about what the ideal nurse-to-patient ratio is.

Why black readmission rates are higher is not entirely clear. These were elective surgeries, so all the patients were relatively healthy. The black patients, though, averaged two years younger and had more health problems. They were also more likely to be poor.

Lasater said the black patients may have had more health-care access problems or less support at home.

Lower nurse-to-patient ratios give nurses more time to talk with individual patients and monitor their response to surgery, she said. That might help nurses spot "subtle warning signs" of infections or bleeding problems, two of the most common reasons for readmission, before they get worse. It might also help nurses spend more time teaching patients about what to do once they get home, and to get patients up and moving, reducing the odds of blood clots while improving strength. Plus, it might increase the chances that patients would reveal the challenges they're facing at home.

Shouldn't doctors and social workers already know how much help patients are going to get at home before they take someone to the operating room? Lasater said patients may be asked about support before surgery, but they may not always be honest about how many other things their potential caregivers are juggling.

"I often find that patients can say one thing to physicians and surgeons and another thing to their nurse," Lasater said. "Patients feel a sense of trust and openness with nurses."

The implication of the study is that hospitals worried about readmission should consider giving black patients more nursing attention. But would the financial benefit be worth the investment in hiring?

"That's the million-dollar question," Lasater said. "That we haven't gotten to yet."

215-854-4944 @StaceyABurling