Variations in cesarian delivery for fetal distress

Journal of Family Practice, Nov, 1996 by William J. Hueston, Richard R. McClaflin

After rising markedly in recent decades, cesarean section rates in the United States appear to have reached a plateau.[1-3] Concerns about the increasing frequency of cesarean delivery have prompted several examinations of the diagnoses given as reasons for this procedure.[4] While cesarean delivery for fetal dystocia is responsible for the some of the increase in cesarean section rates,[4,5] distress has also increased substantially in recent years.[6]

Estimates for the frequency of performing cesarean sections for fetal distress range from 2% to 7%[4,6,7] of all deliveries. Fetal distress, like dystocia, has remained a poorly defined term. Much reliance has been placed on the interpretation of fetal monitor tracings, which has been shown to have great interrater variability.[8] Clear guidelines indicating when operative interventions are appropriate for "nonreassuring" tracings do not exist, leaving the provider with great latitude to determine both what constitutes fetal distress and when distress is serious enough to indicate a cesarean delivery. This lack of objectivity may result in the same variations in practice patterns seen for dystocia, where provider, payer, and institutional biases have been implicated in causing variations in cesarean delivery rates.[9-13]

This study was undertaken to determine to what degree cesarean section for fetal distress varies among institutions and providers, and whether patient and obstetric variables influence the diagnosis of fetal distress and operative delivery. In addition, we sought to determine whether cesarean delivery for fetal distress follows temporal patterns previously reported for dystocia,[14] which suggested that the interpretation of fetal heart rate tracings and decisions based on those interpretations differ depending on the time of day. This could indicate that the diagnosis of fetal distress may be influenced by nonclinical factors such as physician or patient fatigue or the environment in which the physician practices.

METHODS

Data were derived from a retrospective study of deliveries occurring at five institutions in the calendar years 1991 and 1992. The participating institutions ranged in size from 110 to 560 beds and were located in five noncontiguous states. The institutions did not share medical staffs. Four of these institutions were community hospitals, and one facility was the primary affiliate of a state-supported medical school. Resident trainees in obstetrics and gynecology and family practice, or both, were present in all five facilities. Immediate access to anesthesia and pediatric resuscitation was available at two of the five facilities.

To generate the sample population, a clustered random sample of patients who gave birth during the 2 years under consideration was generated as follows. All deliveries occurring each month during these 2 years were considered for analysis. When an institution reported more than 80 deliveries per month, a random sample of 80 patients was selected for analysis based on a random number sequence that included as possibilities the total number of patients who gave birth during that month. Using this technique, a total of 8647 patients were identified and their hospital charts were reviewed.

Because induction of labor generally may be associated with higher cesarean section rates in certain groups of patients[15] and usually follows a temporal pattern of its own, all patients whose labors were induced were excluded from analysis (n=1515). In addition, all women who were admitted for planned cesarean delivery were excluded (n=692). This produced a total sample of 6440 women available for analysis. Because of the theoretical possibility that physicians' knowledge of preexisting maternal or fetal risk factors would alter the interpretation of fetal heart rate patterns and bias toward intervention, women were stratified into low- (n=5621) and high-risk (n=819) categories based on the presence or absence of specific risk factors (Table 1). These risk factors constitute the factors that would identify a woman as high-risk according to the Hollister Risk Form (Hollister, Inc, Libertyville, Ill), which was used by many of the physicians in the study.

TABLE 1
Factors Classifying Patient as High-Risk

  * Hypertension or preeclampsia
  * Preexisting or gestational diabetes
  * Chronic cardiac or renal disease
  * Current drug or alcohol abuse
  * Thrombophlebitis during pregnancy
  * Intrauterine fetal growth retardation
  * Placental abruption or previa
  * Premature labor or rupture of membranes
  * Multiple gestation
  * Active herpes at time of delivery
  * Oligo- or polyhydramnios
  * Abnormal presentation

The findings of this study should be interpreted in light of several limitations. First, the diagnosis of fetal distress could not be independently validated and was based on the primary reason given in the medical record by the attending physician at the time of the cesarean procedure. As noted earlier, however, there is no clear consensus on the fetal heart rate criteria that define fetal distress or necessitate intervention. Since this study was prompted by a concern about interventions based on the physician's interpretation of the fetal heart rate pattern, we believe that the physician's diagnosis of fetal distress is a better indicator of what the physician believed was actually happening than fetal monitor tracings. This belief reflects the opinion that medical decision-making regarding intervention for fetal distress is a complex process involving the patients progress in labor and the underlying risk for poor outcome as well as fetal monitor data.