Can the adoption and implementation of electronic health records (EHRs) be tied to hospital performance and lowered mortality rates? While we might be a bit of time away from being able to make that precise claim, new research does suggest a measurable beneficial relationship.
The findings were revealed by HIMSS Analytics, the research arm of the Healthcare Information and Management Systems Society (HIMSS), and Healthgrades, an online resource for comprehensive information about physicians and hospitals. The value of EHRs has long been discussed, but until now evaluations have lacked comprehensive clinical data, according to HIMSS officials.
Using HIMSS Analytics’ Electronic Medical Record Adoption Model (EMRAM) and mortality rate measures collected by Healthgrades across 19 unique procedure and condition based clinical cohorts, the analysis found that hospitals with advanced EHR capabilities (as reflected in high EMRAM scores) demonstrated significantly improved actual mortality rates, most notably for heart attack, respiratory failure, and small intestine surgery.
Most cohorts experienced improvement in predicted mortality rates when compared to hospitals with lower EMRAM scores. The predicted mortality rate is an indicator of the level of documentation and capture of patient risk factors that are correlated to increased risk of mortality. In total, 4,583 facility records were selected from the HIMSS Analytics database, representing the total number of facilities with complete data from 2010 through 2012.
Collecting the Data
One such facility that participated in the study was the Charlottesville-based University of Virginia Health System, which includes a 604-bed hospital, Level I trauma center, cancer and heart centers, and primary and specialty clinics throughout central Virginia. According to UVA Health System’s CIO, Richard Skinner, who is also a board member for HIMSS Analytics, while the EMRAM model has enabled healthcare systems to see where they rank as far as EHR maturity, any kind of data that describes the impact of implementing an EHR on clinical performance has been missing until now. “The reason for this study was to describe potential benefits from the EHR, and preliminary results say there are benefits,” Skinner says.
For years, HIMSS Analytics has collected a very detailed data set from each hospital in the U.S. with the exception of some very small ones; the model has very specific criteria for which capabilities a facility needs to have for each stage (0-7) on the scale. “Every year, [HIMSS] will call someone from each hospital and ask them to renew that data set. They ask questions such as, ‘Do you have an electronic medical administration record and do you do CPOE (computerized physician order entry)’, for example. With all of that data in hand, HIMSS can then say Hospital A is at Stage 4,” says Skinner. Then, Healthgrades takes Centers for Medicare & Medicaid Services (CMS) data and looks at people who have died in a specific facility, and CMS’ grading of if those people in the aggregate were expected to have died given their diagnoses and so forth, Skinner says.
According to Skinner, to date, the study has shown that those facilities that are higher on the EMRAM (in Stages 6-7) have a better ratio of actual mortality to expected mortality than do hospitals that are lower down on the scale. But Skinner does say that a deeper dive of the data is coming, and that the analysis is very preliminary. “We don’t know why that is yet, but to date that’s what the data has showed us. And you might ask about other factors—‘Are the ones higher on the EMRAM better funded, bigger, and in urban areas?’ There are a host of factors that can come into play. But again, the preliminary data shows a correlation between mortality rate and implementation of EHRs,” says Skinner.
As of today, the study hasn’t gotten down to institutional level to see what happened at a given organization, Skinner says. “And it might not, because the power of the study is the size of the sample’ and it’s the size that enables being able to discover the correlation,” he says.”If you did it at one hospital, there would be so many other variables that statistically, you couldn’t make that association.”
At UVA Health System, Skinner says he has looked at the organization’s clinical performance indicators over time and whether they are improving or not. “For some of those indicators, it’s clear there is at least an association with having better data and having that data in front of clinical decision makers. For others, it’s hard to tell, he says. “Things like urinary tract infection (UTI rates) are getting markedly better, but is that all because of EHRs? No, but you can credit the EHR with at least being able to expose the data and communicate it effectively.”
Skinner says that the reason why such evaluations have lacked comprehensive clinical data is two-fold. First of all, the EHR is a relatively new phenomenon, as most organizations have only implemented a comprehensive EHR in the last few years, and getting it to operate effectively takes some time, he says. The second factor is that the contributing factors to an improvement in clinical performance are, even in the simplest cases, “numerous and interrelated.” So analytically, Skinner says, “It’s difficult to figure out what the most causative variable happen s to be in improvement in expected mortality, or whatever it is that you’re trying to measure. I think the message to the industry is that for hospitals with EHRs, there exists great potential to get further benefits from these tools as we mature in figuring out how to use them,” he says.
Skinner adds that he feels confident in saying there is a “statistical” correlation between advanced EHR capabilities and improved mortality rates. “But again, what part of that correlation is causative awaits further analysis of the data and is not in the preliminary report. All we can say at this point is that there is a correlation. Now, intuitively, it stands to reason that further analysis will filter out those other variables to get to the real contribution to having an EHR.”
What’s more, Skinner does say that the results so far are exactly in line with what he expected. “Of course I am a biased CIO who has a stake in this business,” he says. “But organizations that have spent billions in the aggregate to implement EHRs obviously have the same expectations. While meaningful dollars play a role in that, the entire industry has the expectation that having better information better organized in front of clinical decision makers will lead better results. This study indicates that we’re starting to see that.”
Skinner says he feels that it’s important for the industry to demonstrate this not only because of the magnitude of the investment that’s already been made, but also because there is a huge amount of work left to make truly optimal use of these tools to improve performance. “That’s the hill we are climbing as an industry,” he says.
Clearly a proponent of EHRs, Skinner says that those who criticize the technology for not providing clinicians enough value might not be accurately measuring what the “value” really is. “It may be that a specific clinician hasn’t found much value to him or her, but that doesn’t mean his or her use of the EHR hasn’t proved value to the patient, to the institution as a whole, or to other parts of the institution,” he says.
Skinner notes that the case is easier to make at the organizational level than it is than it is at the individual physician level. “Providers do have a point in that they are being asked to do more and put their hands on a tool they never had to worry about. So there’s no question they have acquired added burdens. But the real question is, ‘Has the institution and its patients gotten sufficient benefit to justify that extra burden?’”