Related Experiment Videos
Validity of diagnostic codes for early pregnancy loss in Emergency Department records
Lisa J Zhang1, Jennifer A Hutcheon1, Regina Renner1
1Department of Obstetrics and Gynecology, University of British Columbia, Vancouver, Canada.
Background:
Early pregnancy loss (a loss before 20 weeks) is an important reproductive health outcome, but the classification accuracy of diagnostic codes for early pregnancy loss in administrative data is poorly understood. We determined the validity of early pregnancy loss diagnostic codes from emergency department visits compared with medical chart review.
Methods:
We conducted a validation study of females aged 15-45 years presenting to two emergency departments in Canada (one community and one tertiary hospital) in April 2018 - December 2021. We used random sampling to identify visits with and without early pregnancy loss codes and reviewed medical records to determine the presence or absence of a confirmed early pregnancy loss based on the physician's final clinical impression. The sensitivity, specificity, positive predictive value and negative predictive value of early pregnancy loss codes were calculated with 95% confidence intervals and weighted using probability of sampling weights.
Results:
Our sample of 376 visits included 241 with a diagnostic code for early pregnancy loss code in the administrative data and 135 without. Compared with medical chart review, early pregnancy loss codes had good sensitivity (87.4 [95% CI: 68.6 to 95.7]) and specificity (99.3 [95% CI: 99 to 99.5]), but only modest positive predictive value 66.2% [95% CI: 59.5 to 72.3]. Nearly half (49%) of false positive cases were pregnancies of unknown location.
Conclusion:
Early pregnancy losses are accurately identified in emergency department visit diagnostic codes, but the codes include a sizeable fraction of false positive cases, most notably from pregnancies of unknown location.
