A significant portion of EHRs are missing data, according to a study published in Journal of the American Medical Informatics Association (JAMIA).

Researchers from Harvard Medical School and Harvard Pilgrim Health Care Institute reviewed the EHRs of 5,500 patients diagnosed with either depression or bipolar disorder who were insured by a major insurer and who receive their primary care from the leading area multi-specialty group, Harvard Vanguard Medical Associates.

Comparing the EHRs to the insurance claims data revealed that about one-fourth of the depression and bipolar diagnoses and more than half of behavioral health visits were not recorded in the EHR. 

The study focused on behavioral health care to track the completeness of EHR entries as compared to claims data, recognizing that patients frequently seek behavioral health and specialty medical care across a range of provider sites. Nearly 90 percent of acute psychiatric services at hospital facilities were not captured in the EHR. Data gaps were similarly high for outpatient and inpatient events overall.  The missing data could have "profound implications for clinical care, medical research and public health," according to the researchers.  "[W]e found that the lack of integration, interoperability and exchange in U.S. healthcare resulted in a major EHR missing roughly half of the clinical information," the researchers said. "Poly markers should put more focus on the quality and utility of health information and ways these can be improved, instead of simply tallying up EHR purchases and supposed capabilities."  “Federal efforts have been undermined by the inherently fragmentary nature of the U.S. healthcare system and the proliferation of proprietary EHR systems that communicate poorly with each other," according to the study’s principal investigator and senior author, Stephen B. Soumerai, ScD, of the Harvard Pilgrim Health Care Institute. "Going forward, our national policy for investment in EHR needs to be re-examined to surmount the fragmentation that currently exists in U.S. healthcare and to set robust technical standards for interoperability and data quality.”  “With medical decision-making becoming increasingly computerized, care will suffer if there is overreliance on data that are incomplete,” said lead author Jeanne Madden, PhD, now at Northeastern University’s School of Pharmacy. ”Health research will undercount important diagnosed patient populations and their use of services.”  The study was funded by the National Institute of Mental Health.