Evidence on the Effects of Nurses' Strikes

In-hospital mortality increased by 19.4 percent and hospital readmissions increased by 6.5 percent for patients admitted during a strike.

U.S. hospitals were excluded from collective bargaining laws for three decades longer than other sectors because of fears that strikes by nurses might imperil patients' health. Today, while unionization has been declining in general, it is growing rapidly in hospitals, with the number of unionized workers rising from 679,000 in 1990 to nearly one million in 2008. In Do Strikes Kill? Evidence from New York State (NBER Working Paper No. 15855), co-authors Jonathan Gruber and Samuel Kleiner carefully examine the effects of nursing strikes on patient care and outcomes.

The researchers match data on nurses' strikes in New York State from 1984 to 2004 to data on hospital discharges, including information on treatment intensity, patient mortality, and hospital readmission. They conclude that nurses' strikes were costly to hospital patients: in-hospital mortality increased by 19.4 percent and hospital readmissions increased by 6.5 percent for patients admitted during a strike. Among their sample of 38,228 such patients, an estimated 138 more individuals died than would have without a strike, and 344 more patients were readmitted to the hospital than if there had been no strike. "Hospitals functioning during nurses' strikes do so at a lower quality of patient care," they write.

Still, at hospitals experiencing strikes, the measures of treatment intensity -- that is, the length of hospital stay and the number of procedures performed during the patient's stay -- show no significant differences between striking and non-striking periods. Patients appear to receive the same intensity of care during union work stoppages as during normal hospital operations. Thus, the poor outcomes associated with strikes suggest that they might reduce hospital productivity.

These poor health outcomes increased for both emergency and non-emergency hospital patients, even as admissions of both groups decreased by about 28 percent at hospitals with strikes. The poor health outcomes were not apparent either before or after the strike in the striking hospitals, suggesting that they are attributable to the strike itself. And, the poor health outcomes do not appear to do be due to different types of patients being admitted during strike periods, because patients admitted during a strike are very similar to those admitted during other periods.

Hiring replacement workers apparently does not help: hospitals that hired replacement workers performed no better during strikes than those that did not hire substitute employees. In each case, patients with conditions that required intensive nursing were more likely to fare worse in the presence of nurses' strikes.

-- Sarah H. Wright