There are some strengths and possible bias in this kind of register-based studies. The most important strength was that the register-based data offered a comprehensive picture of OASR risks among the total vaginally delivered population during the study period, and thus there were no selection bias. Data concerning caesarean sections were excluded from the final analysis since women delivered by Cesarean do not have OASRs. In addition, multiple deliveries having a very different risk profile from singletons were also excluded. These exclusions may have had an impact on the results, since there may have been treatment differences in these high risk groups at night and in holiday periods. Therefore the present results can only be generalized for singleton, vaginal deliveries. Possible bias in our study could also result from the fact that this kind of register-based study might include errors and missing values because the data have been collected for administrative purposes. However, the present data were a subset of the national, population-based Medical Birth Register, which has excellent coverage and is of high quality [20, 21]. This is especially relevant to incidents that result in surgical repair with specific codes of diagnosis, requiring extra days of hospital care. However, some demographic characteristics and outcomes such as spinal analgesia (marked with * in the Tables) were not collected for the whole study period, and therefore we could only incorporate these in the multivariate analyses for a limited time span, i.e. 2004 onwards. The information on anal ruptures was not available in the MBR before year 2004, but the data were taken from Hospital Discharge Register. Also this register in mandatory for all hospitals and its completeness and quality has been shown to be high . In 2006-2007, for example, it covered 95% of OASRs registered in the MBR.
OASRs are consequences of a complex interplay of maternal (age), fetal (birth weight), human (clinical skills), institutional (policy relating to interventions), and administrative (staffing, workload) factors. This probably applies even to those OASRs that were deemed to be caused by a single, simple factor such as fetal macrosomia. This study concentrated on the human and administrative factors rather than the maternal, fetal and institutional ones, by evaluating OASR rates at night and on weekends and holidays. The synergistic effect caused by maternal, fetal and institutional factors was taken into account in the multivariate analysis.
Based only on the risk of exposure to OASR, the safest time to give birth between 1997 and 2007 in Finland was in July or at night. Herein, we report significantly (p ≤ 0.001) lower OASR rates during night time. After adjustment for patient-mix or use of interventions, the risk of OASR was shown to be 11% lower during night time and 15% lower in July. We expected the opposite result because during holidays the competence of the professionals on duty might be lower due to temporary staff, and during emergency duties (weekends and nights) fewer professionals especially obstetricians are responsible for the service. In addition, particularly at night, fatigue might have an influence on maternal outcomes. Accordingly, it may be speculated that the differences can partly be explained by treatment policy, since during emergency duties and in July obstetricians were more willing to allow women to delivery naturally without intervention. For example, use of vacuum assistance was 0.8-1.3% lower during the night time, and in July the rate of induction was among the lowest in comparison with other months. Consequently, our results somewhat confirmed those of one previous study suggesting that higher OASR rates were associated with higher use of obstetric interventions . Lower OASR rate in July could not be explained by higher cesarean section rate, which was among the lowest in comparison with other months, as shown in Additional file 1. Further, the OASR rate in April (0.49%) was almost as low as in July, and we suggest that it may be to some extent due to Eastern, which increased the number of emergency duties. However, our primary aim was to analyze whether deliveries during July, which is the most popular holiday month in Finland, had higher OASR risks, and therefore we did not perform further analysis concerning lower OASR rate in April. If women had received the same level of care irrespective of the month and if the risk of OASRs had been equal, the total number of OASRs in the eleven-year study period would have decreased by 374 (-13%). Correspondingly, if the same risks had been present during the night and the day, the total number of OASRs would have been 329 lower (-12%).
We believe that the general applicability of the current results is likely to be high for similar health care systems with free access to antenatal and obstetric services. The results are probably less applicable to countries in which the majority of antenatal care and deliveries are given in private healthcare facilities. There may be significant differences in teaching status, staff skills and experience and the use of interventions between these two systems. In Finland, the service is provided equally by trainees and senior doctors also in holiday periods with trainees being on call at hospital and senior doctors at home. Thus, it was not necessary to control for the level of supervision or training. However, national data on staffing patterns of obstetricians and midwifes do not exist, and therefore our conclusions about the OASR rates with respect to seasonal patterns and different times of the day are only speculative reflections about the competence of the staff and the role of human errors. Further, since the adverse event itself is not simply the direct harm associated with OASRs but also the consequences such as undesirable outcomes following repair, all aspects should be considered before drawing any firm conclusions about the safest time to give birth, since the inexperience of staff is likely to be expressed in all aspects of their work.