Limitations
Our study has several limitations. Firstly, the ANZSCTS Database collects different data points to major international registries such as the STS and the European database utilised for creating the EuroSCORE II. In contrast to the STS database, procedure type is categorised differently, and ethnicity data, previous mediastinal radiation, cancer and chronic liver disease are not captured [15]. Unlike EuroSCORE II, ANZSCTS does not capture pulmonary hypertension as a risk factor [16]. Consequently, we could not correctly calculate the EuroSCORE II nor STS score for our patients, limiting direct comparability of risk profiles to international data.
Secondly, by focussing on the most recent decade of procedures to fit contemporaneous standards of best practice, we have limited our follow-up period to 30 days. This limits the capacity to assess any longer-term outcome difference between patients who underwent repair versus replacement. Nonetheless, this study is founded on the a priori assumption that repair is preferable to replacement with respect to long-term mortality. Additionally, a 30-day follow-up does not allow for assessment of reoperation rate or repair durability, relevant to both the quality of repair and the durability of replacement prostheses. As directed by the European guidelines for management of chronic primary MR, “achieving a durable valve repair is essential [2]l;” a competent valve intraoperatively or at 30 days does not necessarily equate with durability.
Finally, the presence of only 83 30-day mortalities limited the capacity to assess contributing variables, and in particular to assess for the effect of procedural volume. In comparison, Vassileva et al’s similar study using Medicare Provider Analysis and Review data from the US analysed a cohort of 125,079 patients aged 65 and over; a hospital mortality rate of 10.7% equated to more than 12,000 patients [8]. This much larger cohort and much higher mortality rate facilitated assessment of impact of procedural volume on mortality, which we were unable to achieve to statistical significance.