Using direct clinical observation to assess the quality of cesarean delivery in Afghanistan: an exploratory study

BACKGROUND As part of a National Emergency Obstetric and Newborn Care (EmONC) Needs Assessment, a special study was undertaken in July 2010 to examine the quality of cesarean deliveries in Afghanistan and examine the utility of direct clinical observation as an assessment method in low-resource settings. METHODS This cross-sectional assessment of the quality of cesareans at 14 facilities in Afghanistan included a survey of surgeons regarding their routine cesarean practices, direct observation of 29 cesarean deliveries and comparison of observations with facility records for 34 additional cesareans conducted during the 3 days prior to the observation period at each facility. For both observed cases and record reviews, we assessed time intervals between specified points of care-arrival to the ward, first evaluation, detection of a complication, decision for cesarean, incision, and birth. RESULTS All time intervals with the exception of "decision to skin incision" were longer in the record reviews than in observed cases. Prior cesarean was the most common primary indication for all cases. All mothers in both groups observed survived through one hour postpartum. Among newborns there were two stillbirths (7%) in observed births and seven (21%) record reviews. Although our sample is too small to show statistical significance, the difference is noteworthy. In six of the reviewed cesareans resulting in stillbirth, a fetal heart rate was recorded in the operating theater, although four were recorded as macerated. For the two fresh stillbirths, the cesarean surgeries were recorded as scheduled and not urgent. CONCLUSIONS Direct observation of cesarean deliveries enabled us to assess a number of preoperative, postoperative, and intraoperative procedures that are often not described in medical records in low resource settings. Comparison of observations with findings from provider interviews and facility records allowed us to infer whether observed practices were typical of providers and facilities and detect potential Hawthorne effects.