Introduction UK guidelines recommend pirfenidone and nintedanib as antifibrotic therapy in idiopathic pulmonary fibrosis (IPF); both are effective in slowing disease progression. Randomised controlled trials have shown discontinuation rates of 14%–33% due to adverse events (including side-effects, disease progression and death); clinical audit in our regional centre for IPF showed a rate of 18.4%. We aimed to establish the main clinical, diagnostic and social factors, which predict early discontinuation of antifibrotic therapy using a mixed methodology approach.

Methods We collected data on 170 patients with MDT-diagnosed IPF (2012–16) prescribed either pirfenidone (n=139) or nintedanib (n=31). Retrospective data was collected from electronic records and telephone interviews. Data included demographics, social factors (eg. access to social support, distance from hospital, socio-economic status and level of education), diagnostic tests (HRCT result, pulmonary function tests, broncheoalveolar lavage and biopsy results), which antifibrotic was prescribed, treatment duration and side-effects. We undertook focus groups and patient interviews, which were analysed thematically.

Results Commonest side-effects were nausea and vomiting (50%) and appetite loss (40%; pirfenidone) and diarrhoea (79%) and weight loss (25%; nintedanib). Factors associated with early discontinuation at 30 days on univariate analysis included age ≥60 (p=0.03), female gender (p=≤0.001), DL_CO≤40% at diagnosis (p=0.001), gastrointestinal (p=0.007) and skin side-effects (p=0.004), with similar results at 90 days. In multivariate analysis, DL_CO≤40% at diagnosis (OR=2.44) showed increased risk of early discontinuation at 30 days, with male gender (OR=0.39) and skin side-effects (OR=0.09) both associated with reduced risk. At 90 days, systemic side effects (OR=2.50) increased the risk of early discontinuation, whereas male gender (OR=0.39) and gastrointestinal side effects (OR=0.42) reduced the risk. Thematic analysis of interview scripts suggested side-effects were tolerable with support, and social factors (eg. distance to hospital, barriers to travel) were not considered barriers by patients.

Conclusion Age, female gender and DL_CO≤40% at diagnosis were strong predictors of early discontinuation, when compared with social factors and side-effects. Given low DL_CO has previously been shown to predict increased mortality, it follows that progressive disease would increase the risk of early discontinuation. Side effects were usually manageable, in keeping with previously reported trial outcomes.