Wojahn RD, Foeger NC, Gelberman RH, Calfee RP. Long-term outcomes following a single corticosteroid injection for trigger finger. J Bone Joint Surg Am. 2014 Nov 19;96(22):1849-54. doi: 10.2106/JBJS.N.00004. PubMed PMID: 25410501; PubMed Central PMCID: PMC4238395.
Background: Corticosteroid injection (CSI) mainstay of nonoperative treatment of trigger finger. 61%-84% success in short term followup. One prior study demonstrated linear failure in the first year but not long term results have been demonstrated.
Methods: Retrospective review of patients treated 2000 – 2007 with minimum f/u of 5 years. Pts identified via CPT and ICD-9 codes and confirmed via manual chart review. All CSI performed by hand fellowship trained attendings at A1 pulley 1mL 40mG/mL depo-medrol w/ 1 mL lidocaine.[AMAZONPRODUCTS asin=”1455740241″]
Data collected: Age, sex, date of injection, digit injected, presenting symptoms according to Green classification, presence of multiple trigger fingers, presence of DM. If multiple injections occurred on multiple at subsequent visits or in same visit, only first injection was considered (attempt to limit bias). Records reviewed for treatment failure (subsequent injection or surgical release). Pts lost to f/u were contacted by telephone.
Stats: Chi squared for categorical variables. T-test for continuous variables.
Results: 1049 injections in 867 patients total. 366 met exclusion criteria, had complete data with >5 year f/u.
Injections most common in thumb > long > ring > index > small fingers. Grade I 19%, grade II 59%, Grade III 22%
54.6% had treatment failure with subsequent repeat CSI (64% of failures) or surgical release (33% of failures). 45.4 % did not require further treatment. 1.9% experienced no immediate relief. Mean time to treatment failure was 45 months.
– CSI more successful in females: 49.2% vs 38.1% (p < 0.05)
– CSI more successful with single affected digit: 50.7% vs. 38.5% (p < 0.05)
– Age, symptom type, DM were not statistically significant.
– Kaplan-Meier curve demonstrates 54.5% of failures occur within one year, 83.6% occur within 2 years.
– Log-rank analysis showed non-significant difference in time to treatment failure between genders.
– Log-rank analysis showed significant association in time to failure with multiple affected digits.
– Multivariable cox regression model demonstrated significant interaction between gender and multiple digits.
Conclusions: Highest likelihood of success for single CSI injection is for females with a single affected digit. Patients who experience relief beyond 2 years are likely to maintain long-term success.
Discussion: Weaknesses: retrospective. Potential bias in patients who developed recurrent symptoms but didn’t seek further treatment. No control for adjunct treatment. Chart review biases toward treatment failure because patients without symptoms are less likely to follow up. Did not consider 2 or 3 injection regimens. Treatment sought at outside facilities were not accounted for.
Strengths: large sample size, thoughtful attempt at controlling confounding factors.
Chi squared test: – Compares two samples of categorical variables (A vs. B), tests difference in proportions
T test: – Compares two samples of continuous variables (numbers), tests difference in means
Kaplan-Meier curve: – Evaluates survival of a population after an intervention. Can compare multiple samples.
– Data becomes “right censored” as subjects are lost to follow-up. Noted by vertical ticks on curve.
Log-rank analysis: – Compares “hazard functions” to evaluate subject survival of two Kaplan-Meier curves, accounts for right censoring.
Multivariate cox – Evaluates the effect of multiple variables on outcome. Generates an equation explaining each variable’s
regression model: affect on outcome. (y=m*x+b etc) Can determine which variables are independently significant.