Neutral theory: applicability and neutrality of using generic health-related quality of life tools in diseases or conditions where specific tools are available

Abstract

Abstract Background Health-related quality of life (HRQoL) tools are limited by the indicators included in the construct and variation in interpretation by different researchers. Neutral Theory describes the ideal construct that includes all relevant indicators and, therefore, complete accuracy, or neutrality. Neutral Theory can thereby provide the framework to develop or test constructs. To assess the application of Neutral Theory, the neutrality of generic tools (SF-36 and EQ-5D) at measuring HRQoL was compared to disease/condition-specific tools, with the latter considered surrogates for the Neutral construct. Methods Full descriptions of all disease/condition-specific HRQoL tools published on PubMed (to 01-Jul-19) were sourced. For each tool, the number of items with and without a direct match within the SF-36 and EQ-5D was recorded and the sensitivity/specificity calculated. Results The SF-36 and EQ-5D did not achieve a sensitivity/specificity both > 50% against any of the 163 disease/condition-specific tools identified. At 20% prevalence of poor HRQoL, the false positive rate (FPR) was > 75% for all but two tools against the SF-36 and six tools against the EQ-5D. Increasing poor HRQoL to 80%, 47 tools for the SF-36 and 48 tools for the EQ-5D had a FPR < 50%. For rare disease tools (< 1/2000 population; n = 17), sensitivity/specificity ranged from 0 to 40%/5–31% for the SF-36 and 0–22%/29–100% for the EQ-5D. For non-rare (n = 75) and symptom-specific tools (n = 71) sensitivity/specificity was: 0–100%/0–100% (SF-36) and 0–50%/0–100% (EQ-5D); and 0–60%/0–19% (SF-36) and 0–25%/0–100% (EQ-5D), respectively. No concordance was recorded for 18% (2/11) of results from studies of rare disease tools versus the SF-36 (no data vs EQ-5D). For non-rare, disease-specific tools, results were discordant for 30% (25/84) and 35% (23/65) of studies against the SF-36 and EQ-5D, respectively. For symptom-specific tools, corresponding results were 36% (24/66) and 16% (5/31). Conclusions Generic HRQoL tools appear poorly correlated with disease/condition-specific tools, which indicates that adoption of Neutral Theory in the development and assessment of HRQoL tools could improve their relevance, accuracy, and utility in economic evaluations of health interventions.