Acute Respiratory Distress Syndrome in Pregnant Women
For straightforward questions, such as rate of cesarean delivery by month, the database works.2 For complex issues, accuracy is questionable. Examples: comparison of Nationwide Inpatient Sample with chart review showed accurate indication of spinal fusion indication in only 48% of patients; comparison of Nationwide Inpatient Sample with Nationwide Surgical Quality Improvement Program showed a fivefold difference in conversion from laparoscopic to open colectomy.3,4
The authors confirm risk factors for mortality—prolonged intubation, renal failure, and amniotic fluid embolism. A stunning conclusion was a remarkably low mortality rate, less than 5% in 2012. Could transferred patients be counted twice, once at the referring hospital and again at the receiving? The authors report a combined vaginal and cesarean delivery rate of 56.6%. Were more than 40% of women really discharged undelivered?
In rare obstetric complications, huge databases have huge potential, but data validation is key. For example, we recently found less than 50% accuracy of International Classification of Diseases, 9th Revision, diagnosis of vasa previa. Until we can trust large databases, we may have to continue to learn through tedious chart review.