The Canadian Institute of Health Information (CIHI) Discharge Abstract Database (DAD) is the main source of routinely collected data for gynaecologic surgery in Canada and is increasingly used for research. These data are prone to error as they were originally collected for administrative purposes, and they therefore should be validated for clinical research. The objective of this study was to validate hysterectomy codes from the DAD at a single institution. This was a retrospective observational study using an existing hospital database. We obtained a consecutive sample of all gynaecologic procedures performed at The Ottawa Hospital from April 2016 to March 2017 using the DAD. Patient data, including diagnosis, procedure type, concomitant procedure, and surgical approach, were reabstracted from records. These data were compared with the DAD Canadian Classification of Health Interventions (CCI) codes using sensitivity, specificity, positive predictive value (PPV), and κ coefficient with associated 95% confidence intervals (CIs). Of 1068 gynaecologic procedures, 639 hysterectomies were performed: 39.2% vaginally, 35.4% laparoscopically, and 25.4% abdominally. Median patient age was 46 years (IQR 41-54 y). The κ, sensitivity, specificity, and PPV for all hysterectomies were 0.92 (95% CI 0.90-0.95), 95.1% (95% CI 93.2-96.7), 97.9% (95% CI 96.6-99.3), and 98.5% (95% CI 97.6-99.5), respectively. The κ coefficients for vaginal, laparoscopic, and abdominal hysterectomy were 0.91 (95% CI 0.88-0.94), 0.92 (95% CI 0.89-0.95), and 0.92 (95% CI 0.89-0.95), respectively. Agreement for sub-total hysterectomy and bilateral salpingectomy with oophorectomy was excellent, with κ exceeding 0.80. The level of agreement for salpingectomy alone was poor, though specificity and PPV were high. Our study suggests that hysterectomy-associated CCI codes in CIHI's DAD have a high level of validity for clinical research purposes.
Read full abstract