Limitations
The present study has several limitations. First, as a single site
study, the conclusion may not be applicable in general because of
difference in practice patterns. Second, patients were not randomly
assigned to either group. Therefore, selection bias may affect our
findings. To mitigate the effect of this selection bias on outcomes,
propensity score matching was used to identify well matched LVD and NLVF
patients for comparison. Third, the size of our cohorts are somewhat
small for comparison, but the statistical power of the analysis of
operative outcomes, which was performed to determine whether there was
enough power in this model. Last, incomplete follow up may bias our
results when patient who drop out are different from those who complete
follow-up. We have minimized this bias by incorporating innovative
prevention and retention strategies (such as proactive social media
interaction) into our design and implementation. We performed a
structured survey by telephone interview in patients who were lost to
follow-up to address this issue.