Detecting data variation in disparate perinatal clinical systems using a triangulated approach of data concept analysis, clinician perception study, and patient record review

Delivery of high quality health care requires access to complete and accurate patient information. Variation in data context and content across disparate clinical systems adversely affects the integration of information needed for effective patient care and outcomes research. This study detects the extent and nature of data variation across three disparate clinical systems used along different points of the perinatal care continuum at Intermountain Health Care (IHC). Three analytical methods were used to examine data variation: data structure analysis; clinician perception of missing data elements; and patient record review of key data values. Knowledge acquisition techniques and consensus among clinical domain experts were used to select sample data elements for the data structure analysis. Findings revealed only 17% of the sample data elements had ompatible structure and meaning across the prenatal, labor and delivery (L&D), and newborn intensive care (NICU) clinical data systems. Impact on clinician efficiency from missing and contradicting information in nonintegrated perinatal systems was captured and analyzed using a Critical Incident Technique-based clinician survey. In a 1-month period, 75% of responding clinicians reported missing data and 34% reported contradicting data. The time taken to resolve problems from 1 month's missing data was estimated to be 231 hours for 23 clinicians. Data values from patient records for eight laboratory results were compared across the three perinatal systems. The best match across any two systems was 88% (blood type) and the worst was 0% (antibody screen, chlamydia). The highest incidence of contradicting data was 2.5% for blood type. Comparing agreement of the three methods, triangulation,"" gave additional insight into IHC's data variation problem. The data model study and the patient record review study showed missing data element problems beyond what clinicians perceived. In all, the consistency of data capture in the three perinatal systems at IHC is worse than expected. The data necessary to computationally execute the logic of the perinatal care process models is intermittent and unreliable. Rework of the perinatal applications based on a uniform data model and standard terminologies will provide an infrastructure to achieve IHC's vision of interdisciplinary care.""