Mortality rates creep up at critical access hospitals

Thirty-day mortality rates for acute MI, heart failure and pneumonia at critical access hospitals rose between 2002 and 2010 while rates at non-critical access hospitals dropped, according to an analysis published April 3 in the Journal of the American Medical Association. Any interventions designed to improve outcomes at critical access hospitals should be tested in a cluster randomized controlled trial, the author of an accompanying editorial advised.

Congress initiated the Critical Access Hospital (CAH) program in 1997 to stem the tide in closures of rural hospitals. The program exempted eligible hospitals from quality improvement programs and prospective payments. Over time eligibility requirements have eased and Medicare has added to CAH payments, which has prompted more hospitals to seek CAH status.

Karen E. Joynt, MD, MPH, of the cardiovascular division at Brigham and Women’s Hospital in Boston, and colleagues looked at mortality trends over time between CAH and non-CAH facilities. Their study focused on Medicare beneficiaries and used numerous data resources to identify nonfederal hospitals and hospital characteristics as well as patient discharge diagnoses for acute MI, heart failure and pneumonia and mortality within 30 days of admission.

In all, they found 3,968 hospitals in the U.S. that provided acute care to Medicare patients in 2002, of which 22 percent were CAHs. The percentage of CAH hospitals increased to 28 percent in 2010. The proportion of CAHs in urban and suburban areas grew between 2002 and 2010 while the proportion in isolated rural areas and small towns decreased.

Non-CAHs’ characteristics remained unchanged for the most part during that period. Changes in patient characteristics between 2002 and 2010 were similar for both CAH and non-CAHs, and the composite baseline mortality was similar for both.

Joynt et al found that mortality increased at CAHs at a rate of 0.1 percent a year and decreased 0.2 percent per year at non-CAHs. By 2010, the mortality rate at CAHs was 13.3 percent compared with 11.4 percent at non-CAHs. During the 2002 to 2010 period, CAH status was associated with 10.4 excess deaths per 1,000 admissions.

Looking at individual conditions, they found that acute MI mortality rates increased at CAHs, from 15.4 percent in 2002 to 19.3 percent in 2010, but dropped at non-CAHs, from 17.2 percent to 14.9 percent. To account for the usual lack of PCI at CAHs, they added an analysis that allowed PCI to occur at any time and any hospital within 30 days of the index admission. Still, they discovered a mortality difference of 4 percent, or 21.2 excess deaths per 1,000 admissions.

The difference for congestive heart failure was 2.3 percent (16 excess deaths per 1,000 admissions) and for pneumonia it was 1.7 percent (8.3 excess deaths per 1,000 admissions).

The overall findings held up in matched analyses, sensitivity analyses and analyses that accounted for changes in CAH eligibility. Based on the average annual change in risk-adjusted mortality, 48 percent of CAHs compared with 68 percent of non-CAHs improved in the study period.  

Joynt et al offered several possible reasons for the increase in mortality rates at CAHs that may be unintended consequences of the CAH model, such as exemption from performance reporting and the consequent lack of feedback and disincentives to look for efficiencies due to cost-based reimbursement. They also speculated that CAHs may not have the high-tech resources of non-CAHs.

“Constraints on care in isolated rural areas can be substantial, and our findings suggest that the supports contained in the CAH program have not been adequate to help these hospitals overcome the challenges imposed by caring for this vulnerable patient population in remote settings,” they wrote. They listed that interventions such as partnering with larger hospitals, teleconsultations and other programs might be helpful.

“Given the substantial challenges that CAHs face, new policy initiatives may be needed to help these hospitals provide care for U.S. residents living in rural areas,” they concluded.

In an accompanying editorial, John P. A. Ioannidis, MD, DSc, of the Stanford University School of Medicine in Stanford, Calif., described the study as “the best study to date on the important issue of outcomes at CAHs” but cautioned that it nonetheless had limitations. Those included the use of a large-scale data set, administrative data and potential misclassifications that may lead to over- and underreporting.

“[T]he most difficult aspect of analyses involving administrative data sets is to make inferences about which modifiable features, if any, might underlie the observed differences in mortality,” Ioannidis wrote. “Here the potential factors are typically numerous, overlapping, and difficult to disentangle. “Before initiating interventions that may or may not improve outcomes (or possibly worsen them), he advocated a cluster randomized controlled trial to confirm the findings of Joynt et al.