Table 2

Results of multilevel logistic regression using binary (yes/no) variable of increase in MVPA as a dependent variable 

Dependant variable
Increase in MVPA (yes/no)
Yes (n=62)
No (n=54) (reference)
Independent variables (fixed effects)Univariate modelMultivariate model
EstimateErrorP valueEstimateErrorP value2.5% CI97.5% CI
Time in MVPA (100 min)−−−0.529−0.077
Age (10 years)−0.470.270.082
BMI status: normalReference
BMI status: overweight−0.340.490.494−0.870.560.125−2.0300.203
BMI status: obese−0.950.550.083−1.720.660.009−3.090−0.481
Cancer treatment: radiotherapyReference
Cancer treatment: surgery−0.740.380.052−0.770.420.067−1.618−0.045
ADT: noReference
ADT: yes0.380.380.307
Marital status: partnerReference
Marital status: single−1.190.620.056
Employment status: unemployed/retiredReference
Employment status: working−0.470.440.286
IMD decile: 1–3 (most deprived)Reference
IMD decile: 4–6−0.250.650.698
IMD decile: 7–80.020.600.973
IMD decile: 9–10 (least deprived)−0.200.600.738
CCI score: mildReference
CCI score: moderate−0.020.410.966
  • Pharmacy (clustering effect) was included as a random effect. Factors that can potentially contribute to the change in MVPA over time were explored using univariate regression. The following independent variables were included: baseline time in MVPA (continuous), age (continuous), BMI status (categorical; normal, overweight, obese), cancer treatment (categorical; radiotherapy, surgery), ADT (categorical; yes, no), marital status (categorical; single, partner), employment status (categorical, unemployed/retired, working), IMD decile (categorical; 1–3, 4–6, 7–8, 9–10) and CCI score (categorical; mild, moderate, severe). A backward elimination procedure and statistical significance of p<0.1 was used to derive a final multivariate model. P<0.05 was considered statistically significant. ADT, androgen deprivation therapy; BMI, body mass index; CCI, Charlson Comorbidity Index; IMD, index of multiple deprivation; MVPA, moderate to vigorous physical activity.