| Abstract |
Cesarean sections (CS) have become increasingly common in both developed and developing countries, raising legitimate concerns regarding their appropriateness. Since improvement of obstetric care at the hospital level needs quantitative evidence, using routinely collected health data we contrasted the performance of the 11 maternity centres (coded with an alphabetic letter A to L) of an Italian region, Friuli Venezia Giulia (FVG), during 2005-15, after removing the effect of several factors associated with different delivery modes (DM): spontaneous vaginal delivery (SVD), instrumental vaginal delivery (IVD), overall CS (OCS) and urgent/emergency CS (UCS). A multivariable logistic regression model was fitted for each individual DM, using a dichotomous outcome (1 = each DM; 0 = rest of hospital births) and comparing the stratum specific estimates of every term with their respective reference categories. Results were expressed as odds ratios (OR) with 95% confidence intervals (95%CI). The Benjamini-Hochberg (BH) false discovery rates (FDR) approach was applied to control alpha error due to the large number of statistical tests performed. In the entire FVG region during 2005-2015, SVD were 75,497 (69.1% out of all births), IVD were 7,281 (6.7%), OCS were 26,467 (24.2%) and UCS were 14,106 (12.9% of all births and 53.3% out of all CS). SVD were more likely (in descending order of statistical significance) with: higher number of previous livebirths; clerk/employed occupational status of the mother; gestational age <29 weeks; placentas weighing <500 g; stillbirth; premature rupture of membranes (PROM). IVD were predominantly more likely (in descending order of statistical significance) with: obstructed labour, non-reassuring fetal status, history of CS, labour analgesia, maternal age ≥35 and gestation >40 weeks. The principal factors associated with OCS were (in descending order of statistical significance): CS history, breech presentation, non-reassuring fetal status, obstructed labour, multiple birth, placental weight ≥ 600 g, eclampsia/pre-eclampsia, maternal age ≥ 35 and oligohydramnios. The most important risk factors for UCS were (in descending order of statistical significance): placenta previa/abruptio placenta/ antepartum hemorrage; non-reassuring fetal status, obstructed labour; breech presentation; PROM, eclampsia/pre-eclampsia; gestation 33-36 weeks; gestation 41+ weeks; oligohydramnios; birthweight <2,500 g, maternal age ≥ 35 and cord prolapse. After removing the effects of all other factors, we found great variability of DM rates across hospitals. Adjusting for all risk factors, all hospitals had a OCS risk higher than the referent (hospital G). Out of these 10 hospitals with increased adjusted risk of OCS, 9 (A, B, C, D, E, F, I, J, K) performed less SVD and 5 (A, C, D, I, J) less IVD. In the above 5 centres CS was therefore probably overused. The present study shows that routinely collected administrative data provide useful information for health planning and monitoring. Although the overall CS rate in FVG during 2005-15 was 24.2%, well below the corresponding average Italian national figure (38.1%), the variability of DM rates across FVG maternity centres could be targeted by policy interventions aimed at reducing the recourse to unnecessary CS. In some clinical conditions such as obstructed labor, non-reassuring fetal status, breech presentation, history of CS, higher maternal age and multiple birth, consideration may be given to more conservative DM. The overuse of CS in nulliparas and repeat CS (RCS) should be carefully monitored and subject to audit.
|
