Of the 1,637 hospitals in our sample, 166 were designated as high quality, 1318 as intermediate quality, and 153 as low quality (). There were substantial differences in the characteristics of these hospitals: high quality hospitals were more often large compared with low quality hospitals (26% versus 8%, p<0.001), and more often non-profit in ownership (84% versus 49%, p<0.001). High quality hospitals were significantly more likely to be teaching hospitals than low quality hospitals (44% versus 23%, p<0.001), belong to a hospital system (71% versus 55%, p <0.005), located in urban areas (86% versus 59%, p<0.001), and have a dedicated coronary intensive care units (62% versus 28%, p<0.001). Finally, the percentage of patients with Medicaid was substantially lower in the high quality than the low quality hospital cohort (9 % versus 15%, p<0.001).
Baseline Characteristics, by Quality Cohort, among responders to the AHA IT Survey
We found substantial differences in the adoption of EHR functions among the three groups of hospitals (). High quality hospitals more often had electronic nursing notes (81% versus 73% and 68%, p = 0.04) and medication lists (89% versus 79% and 73%, p < 0.01) than intermediate and low quality hospitals, respectively. All “decision support” tools had significantly higher adoption levels in the high quality cohort. The differences between the high and low quality cohorts in adoption of all of these functions ranged from 17% to 20%, and all were significant ().
Proportion of hospitals with selected electronic functionalities implemented in at least one unit in the high, intermediate, and low quality grades
After multivariable adjustment, we found that adoption of 22 of the 24 functions was still higher in high quality hospitals, although most of the differences were no longer statistically significant (). Functions for which the differences across the three quality cohorts were statistically significant included problem lists, medication lists, diagnostic test images, and many of the clinical decision support tools.
Appendix Table 1
Multivariable-adjusted Proportions and Differences of Proportions of Hospitals with Selected Electronic Functionalities Implemented in At Least One Unit in the Highest, Intermediate, and Lowest Quality Groups
In sensitivity analyses, when we examined groupings based on alternative cutpoints, we found that most of the results were qualitatively similar. However, expanding the high and low quality groups to the 30% cutoff decreased the differences between groups, some of which became nonsignificant (see Appendix).
We performed separate factor analyses in each of the three cohorts of hospitals (high quality, intermediate quality, and low quality) and found relatively similar results across all three groups. Within each cohort, there were two factors with relatively high Eigen values (greater than 3). For example, among the high quality cohort, the hospitals differed most in terms of whether they had adopted CPOE and decision support. The second factor clustered together adoption of patient demographics with viewing lab and radiology reports. The patterns were very similar in the intermediate and low quality cohorts (see –).
Among those hospitals which had yet to implement specific EHR functions, we found high rates of hospitals reporting that they had no concrete plans to implement many key functionalities (). For clinical documentation, results viewing, and computerized order entry functionalities, low quality hospitals were generally more likely to report no concrete plans to adopt the functions, although none of the differences were statistically significant. This may have been due, in part, to the fact that the underlying rates of adoption of specific functions were high and the number of non-adopters was relatively low.
Proportion of non-adopting hospitals with no resources or no plans to implement selected electronic functionalities in the high, intermediate, and low quality groups
The patterns for decision support functions were, however, different. We found that nearly two-thirds of all non-adopters in the low quality cohort reported no concrete plans to implement these functions, rates that were significantly higher than those reported by high quality hospitals. For example, low quality hospitals without clinical guidelines were more likely to report having no concrete plans to implement them than intermediate or high quality hospitals (67% versus 55% and 47% p = 0.02). After multivariable adjustment, the lowest quality hospitals were still significantly more likely to report no concrete plans to implement two of the key decision support tools ().
Appendix Table 2
Multivariable-adjusted Proportions and Differences in Proportions of Non-Adopting Hospitals with No Resources or No Plans to Implement Selected Electronic Functionalities in the Highest, Intermediate, and Lowest Quality Groups
Finally, when we examined hospitals’ ability to meet the Meaningful Use criteria, we found a very small percentage of hospitals across all quality categories have adopted the entire set of functions, with modest differences between them: 2.1% of high quality hospitals could meet all 9 of core measures compared to 1.1% of low quality hospitals, a difference that was not statistically significant. In sensitivity analyses, we found that the results were qualitatively similar for the alternative cutpoints (see Appendix).
When we examined individual Meaningful Use criteria, the majority were present significantly more frequently in the high quality group. Among these functions were the ability to report HQA measures to CMS (41% versus 30% and 34%, p=0.02), implement drug-drug and drug-allergy checks (25% versus 17% and 13%, p = 0.02), data exchange capabilities with other facilities (60% versus 54% and 42%, p < 0.01), and the implementation of at least one clinical decision support tool (84% versus 72% and 63%, p < 0.001) ().
Proportion of Hospitals Meeting Selected “Meaningful Use Criteria” in the High, Intermediate, and Low Quality Grades