Table 1 Variables, measures and methods of analysis
Variable/Outcomes | Hypothesis | Outcome measures | Methods of analysis |
|---|---|---|---|
Baseline data: | There is no difference between the two groups | Gender, age, Katz ADL, CCI, nursing home/hospitalisation, smoking status | Absolute numbers, percentages for categorical variables and the minimum, maximum, mean, SD and quartiles for quantitative variables |
Primary: | |||
Function | There is a clinically important difference between the two groups, with improvement comparing baseline data with data from 6 weeks, 12 weeks, and 1 year. Surgical treatment is hypothesised to be superior | NDI (0–100%) [continuous] | Student t test, chi-square, Mann-Whitney, Fisher’s exact test. Time-dependent differences between AUC |
Secondary: | |||
Health-related quality of life | There is a clinically important difference between the two groups | EQ-5D [continuous] | Student t test, chi-square, Mann-Whitney, Fisher’s exact test Time-dependent differences between AUC |
Pain | There is a clinically important difference between the two groups | VAS (0–100) [continuous] | Students t test, chi-square, Mann-Whitney, Fisher exact test Time-dependent differences between AUC |
Non-union | The non-union rate is lower in the surgical group | Bone bridge in CT [binary], mobility on extension-flexion radiographs [binary], | Chi-square test |
Mortality | The survival is greater in the surgical group | all-cause mortality [binary], time to death [continuous, censored] | Kaplan-Meier analysis, Cox regression models, additional subdistribution hazards approach |
Osteoporosis | There is no difference between the two groups | The bone density is > 2.5 standard deviations below normal DXA T-score [binary] | Chi-square |
Subgroup analysis: | |||
septuagenarians vs. octogenarians vs. nonagenarians | Treatment effect is diminished in nonagenarians | ||
Goel-Harms technique vs. Magerl technique | There is no difference between the two groups | ||
Male vs. female | There is no difference between the two groups | ||
Sensitivity analysis: | |||
Per-protocol analysis | All outcomes | Students t test, chi-square, Mann-Whitney, Fisher exact test | |
Adjusting for baseline covariates | All outcomes | Uni-and multivariate adjusted logistic regression and Cox proportional hazard models | |
Adjusting for mortality | All outcomes | subdistribution hazards approach | |