The effect of loss to follow-up on treatment of blunt traumatic thoracic aortic injury

OBJECTIVE: Loss to follow-up (LTFU) can be a major difficulty for any clinical research study. The objective of this systematic review was to assess the extent of LTFU and its potential effect in studies of adult trauma patients with blunt thoracic aortic injuries (BTAIs). METHODS: Studies comparing management of BTAIs were systematically reviewed. Duplicate independent review was used for study selection, data abstraction, and critical appraisals. RESULTS: Thirty-six studies were included for synthesis, of which 94.1% applied a retrospective cohort design to prospective institutional databases. The mean LTFU at 1 year was 26.5% ± 31.6% for endovascular repair and 20.6% ± 34.2% for open repair groups. Not having a surgical/interventional specialist as a first or senior author was associated with a 39.7% higher LTFU at 1 year (P = .002). Studies with a higher risk of bias, later publication year, or North American origin were associated with a significantly higher risk for LTFU at 1 year (P ≤ .001). Nearly half of included studies assessed in-hospital outcomes exclusively. Only 38.2% explicitly reported LTFU data. Eight studies explicitly described the method of dealing with LTFU: eight used survival analysis and one used a national Social Security Death Index. Sensitivity analyses using plausible worst-case LTFU scenarios resulted in 14% to 17% of studies changing direction of effect. CONCLUSIONS: There is significant LTFU in trauma studies comparing operative methods for BTAIs. LTFU is generally handled and reported suboptimally, and sensitivity analyses suggest that study results are sensitive to differential LTFU. This has implications for the evidence-based choice of the operative method. Some protective factors that may aid in reducing LTFU were identified, one of which was involvement of a surgical or interventional specialist as a key author.