Minimal Pain Decrease Between 2 and 4 Weeks after Nonoperative Management of a Displaced Midshaft Clavicle Fracture Is Associated with a High Risk of Symptomatic Nonunion

BACKGROUND: The main long-term benefit of operative treatment of displaced midshaft clavicular fractures is the reduction in nonunion risk, and as this risk is generally low, the ideal approach would be to operate only patients at high risk of nonunion. However, most current surgical decision models use baseline variables to estimate the nonunion risk, and the value of these models remains unclear. Pain in the early weeks after fracture could be potentially be an indirect measurement of fracture healing, and so it is a potential proxy variable that could lead to simpler prediction models.

QUESTIONS/PURPOSES: (1) Is pain a possible proxy variable for the development of symptomatic nonunion after nonoperative treatment of midshaft clavicular fractures? (2) How reliable is the model we created that uses pain as a proxy variable for symptomatic nonunion of nonoperatively treated clavicle fractures?

METHODS: In this secondary retrospective analysis of an earlier randomized trial, we studied prospectively collected data from 64 nonoperatively treated patients aged 18 years to 60 years. In the original randomized trial, we compared operative and nonoperative treatment of displaced midshaft clavicular fractures. In all, 150 patients were included in the study, of whom 71 received nonoperative treatment. Patients were predominantly males (75%, 48 of 64) with a mean age of 38 ± SD 12 years; most fractures were comminuted and shortened more than 1 cm. All 71 patients who were nonoperatively treated were potentially eligible for this secondary analysis; of those, 11% (8 of 71) were lost to follow-up, leaving 63 patients from the nonoperative treatment arm and one patient from the operative treatment arm (who declined surgical treatment after randomization but was followed in this group according to the intention-to-treat principle) for analysis here. Nonunion was defined as lack of callus formation, persistent fracture lines and/or sclerotic edges of the bones at the fracture site on plain radiographs at 6 months follow-up. Nonunions were regarded as symptomatic if pain, tenderness, and local crepitation were present at the fracture site. Seventeen percent (11 of 64) of patients had symptomatic nonunions. After investigating differences in early pain scores between the union and nonunion groups, we defined the VASratio as the VAS pain score at 4 weeks divided by the VAS pain score at 2 weeks. Week 2 VAS pain score was chosen as baseline after visual inspection of a linear mixed model that showed increased divergence in pain scores between union and nonunion group at 2 weeks after fracture. Week 4 was chosen as the cutoff because we wanted a reasonable time frame for the detection of pain reduction and did not want to delay surgical treatment more than necessary. Odds ratios for various risk factors were calculated using logistic regression analyses. We used a receiver operating characteristic curve analysis to identify cutoff values for the VASratio.

RESULTS: An increase in absolute pain score at 4 weeks after fracture (odds ratio 1.8 per 1 point increase [95% confidence interval 1.1 to 3.4]) was associated with an increased risk of nonunion 6 months after fracture. Likewise, we found that an increasing VASratio (OR 1.02 per 0.01 point increase [95% CI 1.002 to 1.06]) was also associated with nonunion. Receiver operating curve analysis found that the best cutoff value of VASratio was about 0.6. Patients with a VASratio above 0.6 had a relative risk of developing nonunion of 18 (95% CI 2 to 130) compared with patients with a VASratio below 0.6. Sparse-data bias could be present, as is evident from this wide confidence interval, though even at the low end of the confidence interval, the relative risk was 2, which may still improve surgical decision-making.

CONCLUSION: A pain score that exhibits no or minimal change from 2 to 4 weeks after nonoperative treatment of a displaced midshaft fracture of the clavicle is associated with a high risk that symptomatic nonunion will develop. Patients with no or minimal change in pain in the early weeks may be candidates for surgery to reduce the risk of symptomatic nonunion. As this was a retrospective study, with a risk of sparse-data bias, the predictive value of the VASratio needs to be further investigated in large prospective studies before clinical use.

LEVEL OF EVIDENCE: Level III, diagnostic study.