The rapid increase in cesarean deliveries in the last 2 decades and the demonstration that cesarean delivery rates may not always be explained by clinical indications have led to efforts to reduce the number of cesarean deliveries.1,2 Although an optimal rate has yet to be established, and the possible benefits of elective primary cesarean delivery remain controversial,3,4 it has been assumed that hospitals with lower standardized rates provide a higher quality of obstetric care.5,6 Although low standardized rates of cesarean delivery may serve as 1 measure of the quality of perinatal care, the Healthy People 2010 objectives advise that “in addition to monitoring the rates of cesarean births, the outcomes of these deliveries (for both the mother and the infant) should be watched closely to assure that changes in the mode of delivery do not put women or their infants at risk.”1 To examine whether hospitals with low cesarean delivery rates were putting newborns at risk, this study compares the mortality and morbidity experienced by infants born to low-risk mothers across 3 groups of hospitals, those that have average, lower-than-average, and higher-than-average cesarean delivery rates.
MATERIALS AND METHODS
We used a database developed by the California Office of State Health Planning and Development that combines birth certificate and death certificate information, infant and maternal primary hospitalization discharge data, prenatal hospitalization data, and transfer/rehospitalization data during the first year of life for the period 1998–2000. Individual identifiers are not included in this dataset. Birth certificate information and administrative International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes7,8 obtained from this database were used to identify maternal risk factors, neonatal diagnoses, and therapeutic interventions. The study was approved by the Stanford University Institutional Review Board.
An institution's cesarean delivery rate depends on the risk profile of its pregnant women and its practitioners’ perception of the benefits of cesarean delivery in each clinical situation. There are 2 general approaches for controlling for differences in risk across hospitals, regression analysis9,10 and risk stratification (that is the selection of a population thought to be homogenous for risk).1,11 We used the latter strategy because it allowed us to identify and study a defined subset of low-risk women who account for a large proportion of cesarean deliveries and whose risk for cesarean delivery should be quite similar. The low-risk cohort consisted of women of all parities who did not have a prior cesarean delivery, were in active labor at term (more than 37 completed weeks of gestation) with a singleton, and had no evidence of diabetes, oligohydramnios, chorioamnionitis, placental abruption, or abnormal presentation. Mothers whose infants were born with congenital physical or metabolic anomalies were not included. A further advantage of risk stratification was the potential to directly compare this cohort's outcomes across hospitals whose cesarean delivery rate for these women was average, low, or high. However, using the Pearson χ2 statistic we found significant differences in the distributions of maternal race/ethnicity, age, and parity across the 3 hospital categories. To control for these differences in case mix, logistic regression adjusted for race/ethnicity, maternal age, and parity was used to compare outcomes in low-cesarean delivery and high-cesarean delivery hospitals with those in average-cesarean delivery hospitals. In addition to presenting odds ratios and 95% confidence limits, we also account for multiple comparisons by using P values adjusted by the stepwise Bonferroni method (Westfall P, Tobias RD, Rom D, Wolfinger RD, Hochberg Y. Multiple comparisons and multiple tests using SAS. Cary, NC: SAS Institute; 1999.). All analyses were computed by using SAS 9 (SAS, Cary, NC).
Each of 282 hospitals with at least 200 yearly births was classified as having an average, low, or high cesarean delivery rate by testing (using the hospital's 95% confidence interval [CI]) whether the observed cesarean delivery rate for the hospital's low-risk cohort was equal to the cohort's state average. Sixty-two hospitals (22%) had cesarean delivery rates that were significantly lower than average, 160 hospitals (56.7%) had average rates, and 60 (21.3%) had significantly higher-than-average rates (P < .05).
This low-risk cohort's neonatal outcomes were compared across low-, average-, and high-cesarean delivery-rate hospitals. Low birth weight (less than 2,500 g) and very low birth weight (less than 1,500 g) classifications were obtained from birth certificates. Neonatal mortality (deaths per 1,000 cohort births in the first 28 days or during the initial hospitalization regardless of the number of days) was obtained from the death certificate. Long length of stay obtained from the neonatal discharge record was defined as an length of stay exceeding 3 days for a vaginal birth and 5 days for a cesarean delivery. Morbidity classified as attributable to 1) labor and delivery, 2) asphyxia, 3) gastrointestinal, and 4) other systems, and selected therapeutic interventions were ascertained from ICD-9-CM diagnosis and procedure codes. Because of this population's low incidence of morbidity and procedures, rates are reported per 1,000 live births.
During 1998–2000, 1,540,771 singleton, live births were delivered in 282 California hospitals with at least 200 deliveries per year. The primary cesarean delivery rate for these infants was 23.3%. Our low-risk study population consisted of 748,604 (48.6%) of these singletons. The cesarean delivery rate in this low-risk population was 5.8% and averaged 3.5% in the 62 low-cesarean delivery-rate hospitals, 5.7% in the 160 average-cesarean delivery-rate hospitals, and 8.3% in the 60 high-cesarean delivery-rate hospitals. Table 1 compares maternal characteristics for all singleton live births, the low-risk singleton birth cohort, and the low-risk births across the 3 hospital groupings. Compared with all singletons, the low-risk singleton cohort had slightly more Hispanic mothers (50.4% versus 48.7%), a higher percentage of mothers aged 20–30 years (53% versus 49.6%), and nulliparous mothers (41.3% versus 38%). The distributions of maternal race/ethnicity, age, payer source, and parity differed between all singletons and the low-risk singleton cohort (P < .001) and across hospital classifications (P < .001). Average-cesarean delivery-rate hospitals had more non-Hispanic white deliveries, fewer Medi-Cal deliveries, and tended to have intermediate levels of teenaged, older, and nulliparous mothers. Low-cesarean delivery-rate hospitals had the highest percentage of managed care births.
Mortality, morbidity, and therapeutic interventions were very infrequent in our low-risk population (Table 2). The neonatal mortality rate was only 0.16 deaths per 1,000. Meconium aspiration syndrome (22.1 per 1,000) was the most frequent diagnosis, and continuous mechanical ventilation (13.7 per 1,000) the most frequent intervention.
Table 3 shows the incidence of selected neonatal outcomes in infants born in average-cesarean delivery-rate hospitals and the adjusted odds ratios (OR) and 95% CI for these outcomes in infants born at low-cesarean delivery-rate hospitals. There was no difference in the mortality of infants born in low-cesarean delivery-rate and average-cesarean delivery-rate hospitals (OR 0.66; 95% CI 0.42–1.05; P = .99). For some clinical conditions, observed morbidity was increased in infants born at low-cesarean delivery-rate hospitals. These infants experienced a 65% increase in adverse effects of maternal anesthesia (OR 1.65; 95% CI 1.30–2.11; P < .001), a 38% increase in birth asphyxia (OR 1.38; 95% CI 1.14– 1.67; P < .02), and a 19% increase in meconium aspiration syndrome (OR 1.19; 95% CI 1.15–1.24; P < .001). Hypoxic ischemic encephalopathy and intracranial bleeds were not increased (P > .99); however, tomographic studies of the head (presumably prompted by concern for central nervous system dysfunction) showed a trend toward increase (OR 1.21; 95% CI 1.07–1.38; P = .06). Note that although the 95% CI does not include 1, the statistical significance of the relationship was P = .06 after adjusting for multiple comparisons.
There was no difference (P > .99) in the odds of necrotizing enterocolitis/gastrointestinal hemorrhage (OR 1.05; 95% CI 0.75–1.47) or renal failure (OR 1.36; 95% CI 0.73–2.51). However, renal function may have been compromised because electrolyte and metabolic abnormalities were increased by 10% (OR 1.10; 95% CI 1.05–1.16; P = .01). Feeding difficulties were increased 35%, (OR 1.35; 95% CI 1.28–1.42; P < .004) and may represent sequelae of difficult labor and delivery. Therapeutic interventions were greatly increased in infants born at low-cesarean delivery-rate hospitals. The increased use of injected or infused therapeutic substances (OR 1.72; 95% CI 1.59–1.87; P < .001), pressors (OR 1.74; 95% CI 1.33–2.27; P < .001), and mechanical ventilation (OR 1.55; 95% CI 1.48–1.62; P < .001), reflect the increased morbidity experienced by neonates born in low-cesarean delivery-rate hospitals. The percentage of infants with a prolonged hospital stay (longer than 3 or 5 days for vaginal or cesarean delivery, respectively) was increased 12% (OR 1.12; 95% CI 1.09–1.16; P < .001) in low-cesarean delivery-rate hospitals (Table 3).
To more closely examine the increased morbidity, outcomes were compared by route of delivery in low- and average-cesarean delivery-rate hospitals (Table 3). We hypothesized that increased morbidity in vaginally delivered infants at low-cesarean delivery-rate hospitals might suggest that certain infants delivered vaginally could potentially have benefited from cesarean delivery. Increased morbidity in infants born by cesarean delivery might suggest that there may be deficiencies in either the timing or conduct of the procedure. When compared with infants delivered vaginally at average-cesarean delivery-rate hospitals, infants delivered vaginally at low-cesarean delivery-rate hospitals had significantly higher levels of morbidity (including effects of maternal anesthesia, asphyxia, meconium aspiration, feeding difficulties, and electrolyte problems, and significantly higher use of therapeutic interventions (including infusion of therapeutic agents, use of pressors, and mechanical ventilation). The odds of prolonged length of stay was increased 12% (OR 1.12; 95% CI 1.08–1.16; P < .001)
Table 3 also summarizes differences in morbidity for infants born by cesarean delivery in the 2 settings. With the exception of increased feeding problems (OR 1.41; 95% CI 1.16–1.71; P < .01), the prevalence of morbidity was similar for cesarean delivery births in low- and average-cesarean delivery-rate hospitals. However, infants born by cesarean delivery in low-cesarean delivery-rate hospitals required 39% more mechanical ventilation (OR 1.39; 95% CI 1.19–1.62; P < .001), and length of stay longer than 5 days was not increased (OR 1.19; 95% CI 1.03–1.37; P = .46).
When compared with average-cesarean delivery-rate hospitals, high-cesarean delivery-rate hospitals had similar rates of neonatal mortality (OR 0.75; 95% CI 0.48–1.18; P > .99). Differences in morbidity and interventions were mixed. Although fetal hemorrhage (OR 1.30; 95% CI 1.16–1.45; P < .001), non–central nervous system birth trauma (OR 1.15; 95% CI 1.10–1.19; P < .001), birth asphyxia (OR 1.42; 95% CI 1.18–1.71; P < .001), and electrolyte disorders (OR 1.23; 95% CI 1.17–1.30; P < .001) were increased, there was a 10% decrease in meconium aspiration syndrome (OR 0.90; 95% CI 0.87–0.94; P < .001) (Table 4). Although hospitals with high cesarean delivery rates used more ventilation (OR 1.13; 95% CI 1.08–1.19; P < .001), they used less infused/injected medication (OR 0.70; 95% CI 0.63–0.78; P < .001), and there was no difference in the proportion of infants with a prolonged length of stay (OR 0.97; 95% CI 0.94–1.01; P > .99).
We also investigated differences in outcome by mode of delivery. We hypothesized that if the additional cesarean deliveries in high-cesarean delivery-rate hospitals were performed on infants who were truly at risk of adverse outcomes that could be decreased by cesarean delivery, then delivering these higher-risk infants by cesarean would result in decreased morbidity in the remaining pool of infants delivered vaginally. Paradoxically, compared with infants delivered vaginally in average-cesarean delivery-rate hospitals, the incidence of fetal hemorrhage, trauma, birth asphyxia, electrolyte abnormalities, and the use of mechanical ventilation is actually increased in infants delivered vaginally in high-cesarean delivery-rate hospitals (P < .001, Table 4). This suggests that in high-cesarean delivery-rate hospitals the selection of “additional” infants for cesarean delivery may have excluded a number of infants who would have benefited from this mode of delivery.
Compared with average-cesarean delivery-rate hospitals, the outcomes of infants delivered by cesarean delivery in high-cesarean delivery-rate hospitals are similar with the exception of increased feeding problems (OR 1.41; 95% CI 1.16–1.71; P < .01) and mechanical ventilation (OR 1.39; 95% CI 1.19–1.62; P < .001).
For more than 2 decades, the cesarean delivery rate and more recently the case mix–adjusted cesarean delivery rate have been commonly used as a reflection of the quality of perinatal care. Our findings demonstrate that hospitals with both higher- and lower-than-average cesarean delivery rates for low-risk mothers experience higher levels of neonatal morbidity and therapeutic interventions. These findings serve to emphasize that procedure rates such as cesarean delivery birth rates are only meaningful when evaluated within the context of outcome. They fully support and stress the importance of the Healthy People 2010 recommendation that the outcomes of deliveries “should be watched closely to assure that changes in the mode of delivery do not put women or their infants at risk.”1
In comparing cesarean delivery rates across hospitals it is essential to adjust for differences in case mix.11 To control for differences in case mix, we selected a study population consisting of singleton infants born to a clinically homogeneous, low-risk, laboring group of mothers without prenatal complications. This population, with its low primary cesarean delivery rate of 5.8%, is of practical significance because it accounts for almost half (48.6%) of California's singleton live births and represents the population with the greatest interprovider variation in the use of cesarean births. Furthermore, because this population represents laboring women with no maternal, fetal, or placental complications, our results should generally reflect labor and delivery, or obstetric management, as opposed to the management of clinical conditions. However, even in this low-risk cohort we identified differences in the distribution of maternal race/ethnicity, age, and parity across the low-, average-, and high-cesarean delivery-rate hospitals. To control for these differences we built logistic models that included these covariates. Although our cohort has a low incidence of neonatal morbidity, the statistically significant adjusted ORs experienced in low-cesarean delivery-rate hospitals ranged from 1.10 to 1.74 (Table 3). This translates to a 2–6 per 1,000 increase in the rates of neonatal morbidity and interventions. Associated with a reduction in the cesarean delivery rate of 2 per 100, these increases suggest an effect rate of between 1–3 and 1–10 adverse outcomes per cesarean delivery not performed. The increased morbidity in vaginally delivered infants at low-cesarean delivery-rate hospitals suggests that in these hospitals additional infants may have benefited from cesarean delivery. Our findings also support and extend the contention that “it is not apparent that higher cesarean delivery rates in these lower risk patients results in improved outcomes.”11 As demonstrated in Table 4, the increase in morbidity in vaginally delivered infants in high-cesarean delivery-rate hospitals provides evidence that the selection process in this group of hospitals, though leading to more cesarean delivery, has neglected to include many infants who might have benefited from the procedure.
Our findings confirm the recently reported findings by Bailit et al,12 which are based on risk-adjusted primary cesarean delivery rates for Washington State in 1995–1996. Both studies demonstrate a statistically significant increase in asphyxia in lower-than-expected- and in the higher-than-expected-cesarean delivery-rate hospitals. However, further studies will be required to determine if the increased morbidity is directly related to the lower and higher rates of cesarean delivery or to other associated obstetric practices. Further studies are also needed to determine the extent to which the California and Washington observations can be generalized to other states. For example, although their analyses were based on physicians rather than hospitals, the pattern of morbidity seen in California's low-cesarean delivery-rate hospitals did not include the increase in intracranial hemorrhage reported by Li et al13 in their evaluation of deliveries by low- versus medium- and high-cesarean delivery-rate physicians in New Jersey in 1996 and 1997. It is of concern that 3 studies, using very different approaches, have demonstrated a relationship between low cesarean delivery and increased morbidity. Given these findings, it is important that hospitals with low cesarean delivery rates begin to carefully review their practices and outcomes.
A useful feature of our analytic approach is that its assessment of a low-risk cohort identifies a foundation for obstetric quality improvement that moves beyond case mix–adjusted cesarean delivery rates. For example, our findings suggest that a practical initial strategy to evaluate cesarean delivery use at the local level would be to review the labor and delivery of every low-risk term infant without congenital anomalies who requires the use of mechanical ventilation or develops significant morbidity.
One question that arises from these data is whether a hospital's cesarean delivery practice pattern for the low-risk patients chosen for this study is similar to its practice pattern for the Healthy People 2010 cesarean delivery evaluation population (nulliparous, full term, singleton, vertex). Across our 282 hospitals, these 2 cesarean delivery rates were highly correlated (r = 0.79).
There are several potential limitations to our study. The first is the use of ICD-9-CM administrative data based on hospital discharge billing reports. Although a recent study comparing medical record and ICD-9-CM administrative data in a California hospital for 12 indications for cesarean delivery reported accuracy ranging between 83.9% and 100%,14 there are no published reports on the accuracy of administrative neonatal ICD-9-CM diagnostic or therapeutic intervention codes in California or other states. Although there has been widespread use of neonatal administrative ICD-9-CM data as a general indicator of quality,15 the possibility of underreporting and miscoding represents a potential limitation. We have tried to reduce this limitation by examining multiple neonatal outcomes that could be expected to be related to the mode, timing, and conduct of delivery. The pattern of increased morbidity across these outcomes was very consistent. We also included therapeutic interventions such as mechanical ventilation, transfusion/volume expansion for shock, and infusion of therapeutic agents as indicators of morbidity. Because of their importance to reimbursement and accessibility to audit, interventions would be expected to be reliably coded. The findings for both diagnoses and interventions were quite consistent. We also evaluated the stability of our findings by repeating the analysis presented on low-risk births born in 1995–1997. The 2 sets of analyses were highly consistent (full details of the 1995–1997 cohort available on request).
A second limitation of the study is that we considered all low-cesarean delivery-rate hospitals a single stratum. We suspect that within the low-cesarean delivery-rate group there may be hospitals that have optimized cesarean delivery rates and low morbidity and hospitals whose low rates are at the expense of infants who would have benefited from cesarean delivery. An important next step is to develop a morbidity indicator that incorporates several diagnoses and interventions that will allow us to assess hospitals individually.
A further limitation of our study is that it is impossible to precisely assess what factors contribute to poor outcomes in hospitals with low and high cesarean delivery rates. Although one is tempted to look toward the quality of obstetric care, the quality of pediatric care, as well as structural issues (for example nurse staffing and the availability of anesthesiology), also deserve careful scrutiny.
We conclude that using low case mix–adjusted cesarean delivery rates as a proxy for quality obstetric care without taking infant outcomes into account may not provide a valid reflection of the quality of obstetric care because low cesarean delivery rates may be associated with poorer neonatal outcomes. Furthermore, the quality of care should be assessed in high-cesarean delivery-rate and in low-cesarean delivery-rate outlier hospitals because neonatal morbidity may be increased in both of these settings. The development of a methodology to assess the quality of a hospital's intrapartum care based on practice patterns and their resultant neonatal and maternal morbidity should be a high priority for perinatal medicine.
1. U.S. Department of Health and Human Services. Healthy People 2010. Washington, DC: U.S. Government Printing Office; 2000.
2. Centers for Disease Control and Prevention. Rates of cesarean delivery: United States, 1993. MMWR Morb Mortal Wkly Rep 1995;44:303–7.
3. Minkoff H, Chervenak FA. Elective primary cesarean delivery. N Engl J Med 2003;348:946–50.
4. Sachs BP, Kobelin C, Castro MA, Frigoletto F. The risks of lowering the cesarean-delivery rate. N Engl J Med 1999;340:54–7.
5. de Regt RH, Minkoff HL, Feldman J, Schwarz RH. Relation of private or clinic care to the cesarean birth rate. N Engl J Med 1986;315:619–24.
6. Lagrew DC Jr, Adashek JA. Lowering the cesarean section rate in a private hospital: comparison of individual physicians’ rates, risk factors, and outcomes. Am J Obstet Gynecol 1998;178:1207–14.
7. Gregory KD, Korst LM, Gornbein JA, Platt LD. Using administrative data to identify indications for elective primary cesarean delivery. Health Serv Res 2002;37:1387–401.
8. Henry OA, Gregory KD, Hobel CJ, Platt LD. Using ICD-9 codes to identify indications for primary and repeat cesarean sections: agreement with clinical records. Am J Public Health 1995;85:1143–6.
9. Bailit JL, Dooley SL, Peaceman AN. Risk adjustment for interhospital comparison of primary cesarean rates. Obstet Gynecol 1999;93:1025–30.
10. Keeler EB, Kahn KL, Draper D, Sherwood MJ, Rubenstein LV, Reinisch EJ, et al. Changes in sickness at admission following the introduction of the prospective payment system. JAMA 1990;264:1962–8.
11. American College of Obstetricians and Gynecologists. Evaluation of cesarean delivery. Washington, DC: ACOG; 2000.
12. Bailit JL, Garrett JM, Miller WC, McMahon MJ, Cefalo RC. Hospital primary cesarean delivery rates and the risk of poor neonatal outcomes. Am J Obstet Gynecol 2002;187:721–7.
13. Li T, Rhoads GG, Smulian J, Demissie K, Wartenberg D, Kruse L. Physician cesarean delivery rates and risk-adjusted perinatal outcomes. Obstet Gynecol 2003;101:1204–12.
14. Korst L, Gregory KD, Gornbein JA. Elective primary cesarean delivery: accuracy of administrative data. Paediatr Perinat Epidemiol 2004;18:112–9.
© 2004 The American College of Obstetricians and Gynecologists
15. Schwartz RM, Gagnon DE, Muri JH, Zhao QR, Kellogg R. Administrative data for quality improvement. Pediatrics 1999;103(1 suppl E):291–301.