Table 3

Cox proportional hazards regression for the effect of patient predictors on the probability of discontinuing antidepressant treatment

Unifactorial model, unadjusted for practiceMultifactorial model, adjusted for practice
PredictorHRa95% CIP-valueHR95% CIP-value
Sex
 Female
 Male1.071.03 to 1.1<0.0011.061.02 to 1.090.001
Age, years
 <35b
 36–650.760.73 to 0.78<0.0010.770.74 to 0.79<0.001
 >650.730.69 to 0.77<0.0010.710.67 to 0.74<0.001
Deprivation quintile
 1 Low
 21.091.03 to 1.160.0061.071.00 to 1.140.043
 31.181.12 to 1.25<0.0011.121.05 to 1.19<0.001
 41.191.12 to 1.25<0.0011.171.09 to 1.24<0.001
 5 High1.271.2 to 1.34<0.0011.181.10 to 1.26<0.001
Comorbidityc
 No
 Yes0.990.94 to 1.040.66d
Diagnosis coded
 No
 Yes0.880.86 to 0.9<0.0010.860.85 to 0.88<0.001
Previous treatment
 No
 Yes0.94e0.91 to 0.97<0.0010.920.89 to 0.95<0.001
  • a Hazard ratio (HR) = note that HR greater than one indicates a greater risk of discontinuing treatment at any given time point.

  • b When age <35 years compared to all other patients, hazard ratio = 1.33 (95% CI = 1.28 to 1.37).

  • c Comorbidity includes diabetes, coronary heart disease, and cancer; analysis was restricted to patients aged over 45 years.

  • d Comorbidity was omitted from the multifactorial model as unifactorial analysis was restricted to patients aged over 45 years.

  • e When restricted to patients aged over 35 years, hazard ratio = 0.98 (95% CI = 0.94 to 1.02), P-value = 0.4.