Full Mortality risk in patients with adrenal insufficiency using Prednisolone or Hydrocortisone: A Retrospective Cohort study.

Context Prednisolone has been recommended rather than hydrocortisone for glucocorticoid replacement in adrenal insufficiency due its longer duration of action and lower cost. Objective To determine mortality rates with prednisolone versus hydrocortisone. Design Observational study. Setting A UK primary care database (Clinical Practice Research Datalink). Participants Patients with primary and secondary adrenal insufficiency, treated with either prednisolone or hydrocortisone, and controls individually matched for age, sex, period and place of follow-up. Interventions Nil. Outcomes Mortality relative to individually matched controls. Results As expected, mortality in adrenal insufficiency irrespective of cause was increased, based on 5478 patients (4228 on hydrocortisone; 1250 on prednisolone) and 54314 controls (41934 and 12380, respectively). Overall, the adjusted hazard ratio (HR) was similar with the two treatments (prednisolone, 1.76 [95% CI, 1.54-2.01] vs. hydrocortisone 1.69 [1.57-1.82]; p=0.65). This was also the case for secondary adrenal insufficiency. In primary disease (1405 on hydrocortisone vs. 137 on prednisolone:13965 and 1347 controls, respectively), prednisolone-users were older, more likely to have another autoimmune disease and malignancy, and less likely to have mineralocorticoid replacement. Nevertheless, after adjustment, the HR for prednisolone-treated patients remained higher than for those taking hydrocortisone (2.92 [2.19-3.91] vs. 1.90 [1.66-2.16]; p=0.0020). Conclusions In primary but not in secondary adrenal insufficiency mortality was higher with prednisolone. The study was large, but the number of prednisolone-treated patients was small, and they had greater risk factors. Nonetheless the increased mortality associated with prednisolone persisted despite statistical adjustment. Further evidence is needed regarding the long-term safety of prednisolone as routine replacement.