Missing Data in Retrospective Upper Extremity Surgery Research – A Single US-Institution Analysis

Background: Missing data are a recognised challenge in clinical research, yet their extent and handling are rarely reported transparently. We aimed to quantify the frequency of missing data in a retrospective cohort of upper extremity surgical patients and identify factors associated with incomplete records.

Methods: Datasets from four surgical cohorts from our integrated health network (1. arthroscopic rotator cuff repair, 2. open reduction and internal fixation (ORIF) of acute scaphoid fractures, 3. ORIF of distal radius fractures and 4. open carpal tunnel release) were pooled into a dataset of 2,180 patients. The extent of missing data was calculated for each of the following variables: sex, age, body mass index (BMI), race, diabetes mellitus (DM), current smoking status and laterality of the primary procedure. Differences in missingness rates were assessed using the chi-square test. Bivariate analyses were performed to assess whether each variable was a predictor of missing data.

Results: Data missingness was highest at 8.9% (±0.61%) for race, followed by 3.4% (±0.38%) for BMI, 1.4% (±0.28%) for laterality, 0.5% (±0.15%) for smoking status, 0.2% (±0.1%) for DM and no missingness for sex or age data. No patient-level factors were significantly associated with missingness.

Conclusions: Race was the most frequently missing variable, reflecting a broader tendency to underreport social and demographic factors in orthopaedic research. Transparent reporting of missing data, alongside systematic efforts to collect it, remains essential for producing reliable and generalisable surgical evidence.

Level of Evidence: Level V (Therapeutic)