Characteristics of the study population
Table presents demographic, medical and outcome characteristics for the final sample of 20,662 women and 21,255 newborns. Multiple gestations comprised 2.9% of all deliveries. Of all women, 19% (N = 4,004) were formally referred. Among formally-referred birthing women, the majority (52%) came from the regional birth centre (Mawenzi) 3 km away. Demographic characteristics showed that the proportion of teenage mothers (13-19 years) was higher among formally-referred than self-referred birthing women (15.0% versus 8.5%). Formally-referred women were more frequently rural residents and had lower educational attainments than self-referred women (Table ). A history of FGM, a diagnosis of serious maternal morbidity and unknown HIV status were also more prevalent among formally-referred women. As expected, formally-referred women had a higher rate of adverse outcomes such as low birth weight babies, maternal death, neonatal death, low Apgar score and transfer to the neonatal ward (Table ). For other variables, there were no apparent major differences in characteristics according to referral status.
Characteristics of 20,662 women/21,255 infants at KCMC, Tz, 2000-07
Twenty-seven maternal deaths were recorded in the final sample, equivalent to a maternal mortality ratio (MMR) of 131/100,000 births. This does not reflect the true facility-based MMR, as deaths in early pregnancy, in other departments or postpartum were not routinely recorded in the registry. For all births the perinatal mortality rate was 44/1,000, of which stillborn rate was 38/1000 (44% fresh) and early (facility-based) neonatal death rate was 6/1,000, thus less than 1%. Only deaths occurring in the facility were included in the registry.
Caesarean section rates in referred and self-referred birthing women
In the final sample, 6,765 women were delivered by CS; a facility-based CS rate of 32.7%. Emergency CS constituted 80% of all CS deliveries. Less than 2% of all births were operative vaginal births (ventouse or forceps). CS rates rose from 28.5% in year 2000 to 35.5% in 2004, thereafter falling slightly to 31.7% in 2006 (Figure ). Overall, there was a significant increase in CS rates over the study period (χ2 for trend, p < 0.01). The proportion of formally-referred birthing women remained unchanged over the period (χ2 for trend, p = 0.26).
Contribution of selected groups to overall CS rate, KCMC, 2000-06.
Table presents CS rates in the ten groups, sizes of the groups and their relative contributions to the overall rate for all births and by referral status. The proportion of CS was higher overall in the formally-referred group than in the self-referred group, 55% versus 27% (cOR 3.32, 95% CI 3.09-3.57, p < 0.001). Emergency CS was more frequent in the formally-referred than in the self-referred group, 85% versus 79% (cOR 1.50, 95% CI 1.30-1.72, p < 0.01). Elective CS rates were low in general, except in group 5 (previous scar) with singleton cephalic presentation) where both formally-referred and self-referred birthing women had rates > 20% (data not shown). For both referral categories women with previous scar(s) (group 5) and term singleton cephalic nulliparous women (groups 1 + 2) contributed most towards the total CS rate. These three groups comprised 66% of all CS in the facility (Figure ).
Caesarean section rates 2000-07 according to the Ten-Group Classification and referral status, KCMC, Tz
Main admission diagnoses by referral status
Table presents the main admission diagnoses by referral status for the 6,161 CS deliveries, as recorded in the registry. "Previous scar(s)" and "obstructed labour" were the most frequent registered diagnoses in both groups. In formally-referred birthing women these were followed by "cephalo-pelvic disproportion" and "poor progress," whilst in self-referred birthing women, these were followed by "poor progress" and "foetal distress."
Main admission diagnosis recorded among 6,161 Caesarean births, KCMC, Tz
Associations between referral status and post-Caesarean section maternal and neonatal outcomes
Table presents the effect of referral status on cOR and adjusted odds ratio (aOR) for six maternal and neonatal outcomes after CS. By univariate analyses, neither maternal death (N = 7, cOR 1.35, 95% CI 0.30-6.06, p = 0.71) nor obstetric haemorrhage (cOR 1.05, 95% CI 0.73-1.52, p = 0.78) was associated with referral status. Prolonged postpartum stay was associated with formally-referred status (cOR 1.38, 95% CI 1.05-1.81, p = 0.02), but this effect did not remain significant after adjusting for potential confounders such as parity, type of CS, preterm birth and residence.
Associations between referral status and pregnancy outcomes in Caesarean births, KCMC, Tz
Analysis of neonatal outcomes after CS did not find any association between formal referral and neonatal death (aOR 1.37, 95% CI 0.87-2.16). Both low Apgar score (aOR 1.42, 95% CI 1.09-1.86, p < 0.01) and transfer to the neonatal ward (aOR 1.18, 95% CI 1.04-1.35, p < 0.01) were associated with formal referral after adjusting for parity, low birth weight, type of CS and residence. Maternal age was not associated with the outcomes by univariate analysis and was not included as an adjusting factor.
Outcomes after Caesarean section by referral status in the Ten-Group Classification System
Tables A1 and A2 (Additional file 1
) show the distribution of adverse maternal and perinatal outcomes in CS deliveries according to The Ten-Group Classification System. The absolute number of cases per group is small, and data must be interpreted with caution. Obstetric haemorrhage occurred in one out of ten preterm CS deliveries (group 10), and was also prevalent in other obstetric high-risk groups (groups 6-9). Neonatal death rates were > 2% in group 3 (multiparous, spontaneous labour), groups 6 and 7 (breech), group 8 (multiple gestation) and group 10 (preterm singleton cephalic). Frequency of low Apgar score was > 7% in groups 6 and 7 (breech), group 9 (malpresentation) and group 10 (preterm). Among formally-referred CS, low Apgar score was prevalent in groups 6 and 7 (breech) and group 10 (preterm). Among self-referred CS, low Apgar score was most prevalent in group 10 (preterm). Transfer to the neonatal ward occurred in one out of four formally-referred births, compared with one out of five self-referred births. Both formally-referred and self-referred births in group 5 (previous scar) had low frequencies of obstetric haemorrhage, neonatal death and low Apgar score.
Analysis of results before and after introduction of cost-sharing
Due to the gradual introduction of cost-sharing from 2005 (during the period under study), we compared results from before and after January 1st 2005.
There was no difference in the proportion of formally-referred birthing women between the time periods (21.5% versus 21.1%, p < 0.57). Formally-referred women were significantly younger, with a lower level of education and a higher proportion of rural residents in the second time period compared with the first. Self-referred women on the contrary were significantly older, had better educational attainments and a lower proportion of rural residents in the second time period. In both groups the proportion with four or more antenatal visits decreased significantly between the first time period and the second.
The overall CS rate among formally-referred birthing women fell from 58.8% to 49.2% between the first time period and the second (OR 0.68, CI 0.60-0.78, p < 0.001), whilst the rate among self-referred women remained unchanged. The main reasons for referral were the same for the two time periods. For CS births, only neonatal death showed a crude association with the time period, with an increase of overall neonatal deaths from 0.9% to 1.2% (p < 0.03) between the first time period and the second. Adding time period as an explanatory variable in the regression models did not impact on the adjusted estimates.