231 - HIGH HETEROGENEITY IN PRENATAL DETECTION RATES OF SEVERE CONGENTAL HEART DISEASE BETWEEN PHYSICIANS, HOSPITALS AND REGIONS

Background: We recently showed that only 68% of severe congenital heart defects (CHD) were identified before birth. We hypothesized that this mean detection rate of 68% hides significant heterogeneity in the ability to detect severe CHD by obstetrical morphology scans (U/S) between hospitals, regions, and physicians. Our objective was to measure inter- and intra-hospital variability in the sensitivity of prenatal detection of severe CHDs, and to identify at which level this variability was greatest.

METHODS AND RESULTS: We performed a retrospective observational cohort study of all pregnancy-infant dyads with a severe CHD in Quebec between 2007 and 2015. Prenatal clinical data was extracted from the Quebec congenital heart disease registry. We matched this information with healthcare administrative data to identify regions, hospitals and physicians performing U/S. Generalized estimating equations were used to estimate 95% confidence intervals (95%CI). Estimation of variability in detection rates at each level (region, hospital and physician) was done using multilevel generalized linear mixed models, and random effects models were used to estimate intra-class correlations (ICCs).
There were 1436 severe CHD in our cohort, with 162 (11.2%) identified by a fetal echocardiogram done prior to any U/S. Of the 1274 remaining CHD, 699/1274 CHDs were suspected at the U/S [detection rate of 54.8% (95%CI: 52.1-57.6%)]. After excluding physicians and hospitals with lower volume ( < 5 severe CHD), very significant heterogeneity was found between hospitals (Figure 1) and physicians (Figure 2). The detection rate of U/S in the lowest quartile for physicians, hospitals and regions was 27.4%, 29.0% and 39.8%, respectively. The detection rate in the highest quartile for physicians, hospitals and regions was 87.3%, 71.3% and 63.7%, respectively. The mean difference of detection rate between the lowest and highest quartiles was 63.5 percentage points (95%CI: 58.3-68.7) for physicians, and 46.1 percentage points (95%CI: 37.0-55.2) for hospitals. The greatest variance was found at the level of the physicians (intra-hospital), with ICC for physicians, hospitals and regions of 5.5%, 0.9 % and 1.8%, respectively.


Conclusion: There was considerable heterogeneity between regions, hospitals, and physicians, with detection rates of severe CHDs among physicians in the higher quartile >3 times higher than their counterparts in the lowest quartile. The driver of the heterogeneity seemed to be at the physician level, with higher intrahospital variability. Reducing variability could greatly increase the overall detection rate, and any measures of improvement should be targeted at the physician level.