Hospital Mortality Rates Fall After EHR Adoption, Eventually

July 10, 2018

Adopting electronic health records (EHRs) is initially linked to a minor increase in 30-day mortality rates, particularly if more functions are present at adoption, but rates then decline over time and continue to drop with each new EHR functionality adopted, according to a study published online July 9 in Health Affairs.

The mortality rate improvements were most pronounced in small hospitals and nonteaching hospitals, whose decreases in mortality rates drove the overall trend, the study shows.

"Our findings suggest that EHR adoption affects performance improvement through multiple pathways," write Sunny C. Lin, MS, a doctoral candidate from the Department of Health Management and Policy at the University of Michigan in Ann Arbor, and colleagues.

"[T]he relationship between EHR adoption and performance is not as simple and straightforward as, 'Does it work?' The relationship varies not only by hospital characteristics but also with time," they write. "Much of the performance improvement we observed in our study occurred either with the maturation effect of baseline functions or with the addition of functions each year."

The researchers analyzed three sets of data to examine how adoption of EHR might be associated with 30-day mortality rates. These sources included national Medicare hospital claims for 2008 to 2013, the American Hospital Association Annual Survey Database information for 2014, and the Information Technology Supplement for that survey for 2008 to 2013.

These data provided hospital EHR and individual EHR function adoption dates, hospital characteristics, and the hospital's mortality performance for 5992 US hospitals. The authors limited their analysis to nonfederal acute care hospitals not missing any years' mortality data and not missing more than 3 years of EHR adoption data, resulting in data for 3249 hospitals.

The researchers temporally assessed mortality rates as hospitals adopted each of 10 basic EHR functions. Mortality rates only included 15 common conditions and were combined into an overall rate risk-adjusted for sex, age, and comorbidities. Hospital characteristics included location (urban/suburban vs rural), size, teaching status, and safety net status.

At baseline, hospitals initially adopted EHR systems with an average 5.7 basic EHR functions and had an average 30-day mortality rate of 13.46 per 100 admissions. Rates were initially higher at baseline by 0.11 percentage points per EHR function added.

Subsequently, mortality rates then dropped an average 0.09 percentage points per year per function. Each added function after that was associated with a drop in mortality rates of 0.21 percentage points.

It was primarily small and nonteaching hospitals that "experienced worse performance with a greater number of baseline functions but improved performance from the maturation effect and new function adoption," the authors report.

EHR adoption was also linked to lower mortality rates overall over time: Hospitals that adopted EHRs had 0.67 fewer deaths per 100 admissions per year compared with nonadopting hospitals. Further analysis of hospital characteristics suggested that "mortality reduction from adoption was significantly greater for nonteaching hospitals," the authors report.

The greater effect seen in small and nonteaching hospitals lends credence to the hypothesized "ceiling effect," in which hospitals already making quality improvements may have less room to improve before somewhat plateauing.

"For small and nonteaching hospitals, EHR adoption may have represented a large, highly visible quality improvement initiative that also prompted broader conversations about quality," the authors write. "Large and teaching hospitals have the resources to continuously engage in quality improvement, both before and after EHR adoption, so the EHR itself might not have resulted in similarly substantial gains in performance."

Or, the authors propose, the findings could explain why small and nonteaching hospitals struggled with worse mortality rates when an EHR was initially adopted with multiple functions but then improved. "[A]nother explanation may be that these hospitals had limited resources and experience to support a high-quality initial EHR implementation," the authors write. "Therefore, they may have done a worse job with implementation of baseline functions but learned how to improve new implementation efforts over time."

The study's limitations include not considering the order in which different types of EHR functions were added and the fact that baseline data about a hospital's use of EHR in 2008 could not take into consideration how long the hospital had been using the system.

The former limitation means some observed effects may have resulted from adopting more advanced EHR functions rather than simply more functions in general. The second "could explain why we found that large teaching hospitals did not benefit from EHR adoption," the authors note. "Such hospitals, which had higher levels of baseline EHR adoption, could have achieved most EHR-related gains before our study period, though we do not think that this is likely."

Further, the nonresponding hospitals excluded from the analysis were more likely to be "small, nonteaching, safety-net institutions" and "may be less likely to adopt EHRs, use them effectively, or both."

The research was funded by the John A. Hartford Foundation. The authors have disclosed no relevant financial relationships.

Health Aff. Published online July 9, 2018. Abstract

For more news, join us on Facebook and Twitter


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.