Study strengths and limitations
Our study has several strengths. We used a large, state-wide cohort derived from 5 years of routinely collected data in Australia, to explore the association between VBAC and significant perineal injury. Whilst there is a growing body of published evidence recognising VBAC as a risk factor for 3rd and 4th degree tear, ours is one of the largest, most recently reported studies to explore this association and provides a more rigorous, updated approach to adjusted analyses than the existing published studies.
The main limitation of our study is the use of retrospective data, which can produce biased estimates of risk due to missing data and limited covariate detail. However, we thoroughly explored the pattern of missing data, specifically for body mass index and epidural for which >10% of data were missing. It was important that data describing these two variables were appropriately accounted for, since high body mass index is thought to be protective against significant perineal injury (35, 36) and epidural increases the likelihood of instrumental birth which is a risk factor for 3rd and 4th degree tear (10, 37, 38).
In contrast to other studies (16), our study accounted for missing data and our model was carefully planned and scrutinised. We attempted to overcome the limitations of routine data and produce the least biased results by imputing missing data, analysed for bias with appropriate diagnostic techniques, and performing sensitivity analyses with non-missing data. The main limitation for our multiple imputation was the absence of auxiliary variables. However, this is a recognised limitation of using routinely collected population data, and not a limitation of our analysis per se. All adjusted models, including complete case models, produced estimates within close proximity of each other, diminishing the likelihood that our final estimate of risk is misleading.