We studied busy day effects on obstetric care using daily delivery volume by hospital size category stratified into quiet, optimal, and busy days as the primary exposure. The outcome measures were unplanned CS, instrumental delivery, induction of labour, and EA during labour in different sized delivery hospitals and on the whole obstetric ecosystem level. The only significant difference in pooled analysis comprising the entire delivery hospital network was the less than 10% decrease of instrumental deliveries on quiet compared to optimal days.
Dissecting the data by hospital size brought up at least four interesting findings. First, the unplanned CS rate decrease was 15 to 30% from quiet to busy days in the smallest hospitals (C2, C1), whereas there was no change in the other hospital categories in the same comparison. Second, in C3 hospitals, induction showed a similar decreasing pattern as CSs from quiet to busy days. Still interestingly, in large non-university hospitals (C4), the induction rate showed the opposite and increased significantly up to 50% from quiet to busy days. Third, epidural analgesia was up to 60% higher on busy than quiet days in large non-university hospitals (C4) but not in other hospital categories. Fourth, adjustment for the case mix changed the results only a little, suggesting that the patient profiles in each hospital category on quiet and busy days were very similar.
In small delivery hospitals (C1, C2), unplanned CS rates were higher during quiet days and lowered during busy days. This cannot be explained by medical reasons or case-mix differences since these were considered in the multiple regression and in the study setting, where each hospital category served as its control. Therefore, there is no reason to think that the treatment policy is deliberately changed by daily variation. Unplanned CS is the most intensive intervention in resource use since it takes more time than instrumental delivery, and EA and a surgical team are needed [21]. Therefore, we can speculate that the capacity to perform unplanned CS does not fully cover the demand on busy days in small delivery hospitals. This is in line with previous research results showing that busy times like night and weekend deliveries are associated with unfavourable alterations in some obstetric interventions and perinatal outcomes [22,23,24].
In general, large non-university hospitals (C4) performed more unplanned CSs, instrumental deliveries, inductions, and EAs on busy compared to quiet days. This result was somewhat unexpected and not noticed in other delivery hospital categories. However, it shows that large central hospitals have significant capacity to perform and increase interventions even on busy days. The explanation why this happens is unclear. There may be an attempt to speed up the labour process with the increasing patient flow. Whatever the explanation, it is evident that inductions were organized in a very poorly coordinated way since more inductions were performed on busy than optimal or quiet days.
After pooling all hospital categories, the sum of available data showed that a 10% decrease in instrumental deliveries in quiet compared to optimal days was the only statistically significant finding over the full spectrum of outcomes at the ecosystem level. Still, the observed heterogeneity was high in all pooled analyses suggesting marked differences between hospital categories.
Differences between delivery hospital categories were significant in each outcome, which supports heterogeneity in delivery hospital comparison. In tertiary university hospitals (C5), the changes in the use of all interventions were less than 10% during busy and quiet days compared to optimal days and primarily did not reach nominal statistical significance, suggesting that the university hospitals were resistant to changes in patient flow.
The strength of this study is extensive, population-based national data from reliable MBR of high quality and comprehensive over a lengthy period [25, 26]. Such big data sets are needed to discover small but clinically significant differences caused by daily delivery volume variation, which would easily go unnoticed if only data from one delivery hospital were analysed for self-validation purposes. Furthermore, analyses were performed with a multivariate regression model, and the case-mix adjustment was performed for every outcome.
The quiet, optimal, and busy days were determined in different sized delivery hospitals. These daily delivery volume variation calculations were based on the daily delivery frequency in different sized hospital categories and estimated to the nearest 10% to represent quiet and busy delivery volume days. However, due to the complexity, duration, and nature of the delivery process, the calculations may not be exact, and the daily delivery volume does not describe the actual workload in the hospital during varying daily periods. The other limitation of this study is the hospital categorisation by annual delivery volume. The register data related lacks specific characteristics between different delivery hospitals, such as acuity, patient case-mix, availability of obstetric anesthesia, and detailed information on nursing staff in each unit.