Direct healthcare costs associated with sickle cell disease complications: a retrospective cohort study using routinely collected healthcare data in England.

objectivesDue to the multisystemic nature of sickle cell disease (SCD), complications can occur together and thus discerning costs associated with individual complications requires a methodology that can estimate the costs of a given complication while accounting for the presence of other complications. In this study, we aimed to estimate period-based incremental costs associated with specific chronic complications in patients with SCD in England while accounting for multimorbidity. DESIGN/

settingAll-cause primary and secondary care healthcare resource utilisation (HCRU) was obtained for a retrospective cohort of patients with SCD using Clinical Practice Research Datalink (CPRD) Aurum linked to Hospital Episode Statistics (HES) datasets. Annualised HCRU and costs were calculated, dividing patient-level events by patient-level time (in years) to obtain per person per year estimates. A series of generalised linear models were used with adjustment for demographic factors and proportion of follow-up time with each complication to estimate the costs associated with 10 chronic SCD-related complications of interest. For these costs, annual equivalent costs can be obtained by dividing by the median follow-up time of 4.74 years.

participantsPatients with a diagnosis of SCD, with or without complications, in CPRD or HES with at least 12 months follow-up. OUTCOME MEASURES: Period-based all-cause direct healthcare costs.

resultsOf the 1271 patients with SCD included in the study, 49.9% (n=634) had at least one complication and of these 41.3% (n=262) had two or more complications either at baseline or during follow-up. Patients with complications had higher all-cause healthcare costs compared with patients without complications (mean (SD) annualised cost £16 058 (£21 488) vs £4399 (£6635)). Patients with complications had four times the number of annualised inpatient admissions (6.1 vs 1.5 admissions) and more than double the number of annualised bed days in hospital (8.3 vs 3.8 days) over a median 4.74 years of follow-up. Of the complications evaluated, end-stage renal disease had the highest estimated incremental cost of £252 083 (95% CI £214 478 to £283 745) over 4.74 years; this is in addition to the £18 547 period-based cost among patients with SCD without complications. Osteonecrosis was the most common complication with an estimated incremental cost of £27 399 (95% CI £6417 to £43 319) over the same period.

conclusionEstimating the cost of complications, while accounting for multimorbidity, is essential to determine the true direct cost of SCD. The modelling method presented in our study provides period-based estimates of cost and hospital admissions for individual complications in patients with SCD, accounting for multimorbidity. This approach can be used and extended to other diseases with multisystemic complications to estimate the direct HCRU and costs of individual complications.