Determinants of Length of Stay After Vaginal Deliveries in the Friuli Venezia Giulia Region (North-Eastern Italy), 2005-2015.

Although length of stay (LoS) after childbirth has been diminishing in several high-income countries in recent decades, the evidence on the impact of early discharge (ED) on healthy mothers and term newborns after vaginal deliveries (VD) is still inconclusive and little is known on the characteristics of those discharged early. We conducted a population-based study in Friuli Venezia Giulia (FVG) during 2005-2015, to investigate the mean LoS and the percentage of LoS longer than our proposed ED benchmarks following VD: 2 days after spontaneous vaginal deliveries (SVD) and 3 days post instrumental vaginal deliveries (IVD). We employed a multivariable logistic as well as a linear regression model, adjusting for a considerable number of factors pertaining to health-care setting and timeframe, maternal health factors, newborn clinical factors, obstetric history factors, socio-demographic background and present obstetric conditions. Results were expressed as odds ratios (OR) and regression coefficients (RC) with 95% confidence interval (95%CI). The adjusted mean LoS was calculated by level of pregnancy risk (high vs. low). Due to a very high number of multiple tests performed we employed the procedure proposed by Benjamini-Hochberg (BH) as a further selection criterion to calculate the BH p-value for the respective estimates. During 2005-2015, the average LoS in FVG was 2.9 and 3.3 days after SVD and IVD respectively, and the pooled regional proportion of LoS > ED was 64.4% for SVD and 32.0% for IVD. The variation of LoS across calendar years was marginal for both vaginal delivery modes (VDM). The adjusted mean LoS was higher in IVD than SVD, and although a decline of LoS > ED and mean LoS over time was observed for both VDM, there was little variation of the adjusted mean LoS by nationality of the woman and by level of pregnancy risk (high vs. low). By contrast, the adjusted figures for hospitals with shortest (centres A and G) and longest (centre B) mean LoS  were 2.3 and 3.4 days respectively, among "low risk" pregnancies. The corresponding figures for "high risk" pregnancies were 2.5 days for centre A/G and 3.6 days for centre B. Therefore, the shift from "low" to "high" risk pregnancies in all three latter centres (A, B and G) increased the mean adjusted LoS just by 0.2 days. By contrast, the discrepancy between maternity centres with highest and lowest adjusted mean LoS post SVD (hospital B vs. A/G) was 1.1 days both among "low risk" (1.1 = 3.4-2.3 days) and "high risk" (1.1 = 3.6-2.5) pregnanices. Similar patterns were obseved also for IVD. Our adjusted regression models confirmed that maternity centres were the main explanatory factor for LoS after childbirth in both VDM. Therefore, health and clinical factors were less influential than practice patterns in determining LoS after VD. Hospitalization and discharge policies following childbirth in FVG should follow standardized guidelines, to be enforced at hospital level. Any prolonged LoS post VD (LoS > ED) should be reviewed and audited if need be. Primary care services within the catchment areas of the maternity centres of FVG should be improved to implement the follow up of puerperae undergoing ED after VD.