Measuring quality of life in opioid-induced constipation: mapping EQ-5D-3 L and PAC-QOL

Anthony James Hatswell, Stefan Vegter

In health economic evaluations, quality of life should be measured with preference-based utilities, such as the EuroQol 5 Dimension 3-level (EQ-5D-3 L). Non-preference-based instruments (often disease-specific questionnaires) are commonly mapped to utilities. We investigated if the relationship observed between the Patient Assessment of Constipation Quality of Life (PAC-QOL) and the EQ-5D-3 L in patients with chronic idiopathic constipation (CIC) also applies in opioid-induced constipation (OIC). EQ-5D-3 L patient-level data from a clinical study of lubiprostone in OIC (n = 439) were scored using the UK tariff. A published mapping between the PAC-QOL and the EQ-5D-3 L was tested using these data. New mapping formulas were analysed, including PAC-QOL total and subscale scores. The root mean square error (RMSE), the adjusted R(2) and predicted/observed plots were used to test the fit. The utility measured with the EQ-5D-3 L was 0.450 ± 0.329, with a distinctly bimodal distribution. This significantly improved if patients responded to treatment (defined as an increase of three spontaneous bowel movements per week, with no rescue medication taken). The published mapping in CIC performed poorly in this OIC population, and the PAC-QOL could not be reliably mapped on to the EQ-5D-3 L even when re-estimating coefficients. This was shown in our two mappings (using PAC-QOL total score, and subscale scores) by a high RMSE (0.317 and 0.314) and a low R(2) (0.068 and 0.080), with high utilities underestimated and low utilities overestimated. Patients with OIC have a low quality of life which does improve with the resolution of symptoms. However the PAC-QOL cannot be used to estimate the EQ-5D-3 L utility - potentially as the PAC-QOL does not capture the all relevant aspects of the patients quality of life (for example the cause of the opioid use).