Abstract

Introduction

Decision makers across the world use evidence of treatment benefit from different sources to decide the coverage of new treatments. There is a pipeline of new treatments for SLE in clinical trials and recently completed Phase III trials. Therefore, there is a need for data to inform longer term effectiveness and cost-effectiveness analyses in SLE for decision-making.

Objective

To develop a natural history model to simulate long-term outcomes in Systemic Lupus Erythematosus (SLE).

Methods

Longitudinal data on 1354 patients from the Hopkins Lupus Cohort were included in the analysis. Disease activity, treatments, Systemic Lupus International Collaborative Clinics/American College of Rheumatology Damage Index (SLICC/ACR DI) events, and laboratory tests are reported at every clinic visit. A linear random effects model was used to estimate mean change in SLE Disease Activity Index (SLEDAI) score over a year and to estimate average prednisone dose as a function of SLEDAI score. Parametric survival models were used to estimate organ damage and mortality. A simulation model combined the separate statistical models to assess the validity of the natural history predictions.

Results

The long-term disease activity model finds that previous SLEDAI , previous renal involvement, age, and males predict a reduction in future SLEDAI score. Ethnicity, anaemia, haematological involvement, increased DNA binding and low complement in the previous year predict an increase in disease activity. The annual average prednisone dose increases for every unit increase in annual average SLEDAI . Adjusted Mean SLEDAI was associated with an increased risk of mortality, cardiovascular, renal and peripheral vascular damage. Organ involvement predicted mortality, cardiovascular, renal, neuropsychiatric, pulmonary, gastrointestinal, ocular and skin damage. Corticosteroids increase the risk of gonadal failure, diabetes, cardiovascular, musculoskeletal, neuropsychiatric, and gastrointestinal damage. The simulation reproduces some of the patient outcomes reported in the Hopkins Lupus cohort with good accuracy.

Conclusion

The analysis generates a natural history model that can be used to extrapolate the long-term effectiveness and cost-effectiveness of treatments for SLE . Alternative modelling methods should be explored to improve the description of long term outcomes for future treatment assessments