Is the use of retrospectively collected radiographic scores an accurate method to determine time to union? A retrospective clinical cohort

OBJECTIVES: To evaluate the accuracy of time to union (TTU) in a cohort of patients after intramedullary nailing (IMN) of tibial shaft fractures where radiographic union was measured retrospectively with uncontrolled intervals between radiographs. METHODS: Design: Retrospective cohort Setting: A single Level I academic Trauma Center Patient Selection Criteria: Included were adult patients (18 years and older) treated with IMN for an acute tibial shaft fracture (OTA/AO 42A, 42B, 42C) between January 2020 to January 2025 with at least 6 week and 12-week post-operative radiographic evaluations. Patients who required a reoperation to promote union were excluded. Outcome Measures and Comparisons: The modified radiographic union scores for tibial fractures (mRUST) were determined at each post-operative radiographic follow-up. The observed TTU was defined as the days from IMN to radiographic union (defined as mRUST ≥ 12). Cohort median time intervals were determined for: the time from IMN to the radiograph immediately prior to union; the time between the radiograph prior to union and union, and the TTU. The inter-quartile range to median ratio (IQR:median) was calculated for these intervals to normalize the time interval ranges and was used to define accuracy. Accuracy was defined as follows: low accuracy = IQR:median ≥0.5; moderate accuracy = IQR:median 0.2-0.5; and high accuracy = IQR:median ≤0.2. A predicted TTU for the cohort was calculated by considering all mRUST scores from the entire cohort over time to estimate the time at which 50% of the population would be expected to achieve radiographic union. This was a way to account for the changing nature of mRUST scores with time in between clinical visits. To determine if the observed TTU over- or underestimated the predicted TTU, the predicted TTU was compared to the observed TTU by taking the difference between the two. RESULTS: 89 patients with a mean age of 43.1 years (SD = 13.7 years) and 65.9% male were included. The median observed TTU was 96 days (mean 208.6, SD 363.8 days) with an IQR:median of 0.9, indicating low accuracy. The median time from surgery to the radiograph immediately prior to union was 47 days (mean 58.7, SD 37.3 days) with an IQR:median of 0.4 (moderate accuracy). The median time between the radiograph immediately prior to union and union was 49 days (mean 150, SD 358 days) with an IQR:median of 1.1 (low accuracy). The predicted TTU was 61.8 days. The observed TTU over-estimated TTU by 34.2 days. CONCLUSIONS: In patients treated with tibial IMN, retrospectively determining radiographic time to union (TTU) with uncontrolled interval radiographic evaluations was an inaccurate method to determine TTU. The observed TTU over-estimated a predicted TTU by 34.2 days. These findings highlight the inaccuracy of retrospectively determined TTU in orthopaedic clinical cohorts. LEVEL OF EVIDENCE: Level III