Using the Analytic Hierarchy Process to Derive Health State Utilities from Ordinal Preference Data.

BACKGROUND: The EuroQol five-dimensional questionnaire is a standardized instrument used in the economic evaluation of health care to measure health state preferences across disease groups. A time trade-off (TTO) approach is commonly used to elicit preferences from the public. However, there are issues regarding how best to measure worse-than-dead states; at present, extreme valuations are rounded up to more acceptable values. TTO elicitation is also cognitively demanding for respondents and is therefore expensive to investigate. OBJECTIVES: To describe how the analytic hierarchy process approach could be used to generate utilities from the ordinal relationships between the health states instead of the ordinal relationships between health states, allowing potentially useful preference data to be incorporated rather than excluded as they are at present. It was applied to the Measurement and Valuation of Health study data set, measuring health state preferences for the United Kingdom. METHODS: The analytic hierarchy process approach was explained. Five approaches to structure pairwise comparisons of health state preference were described (two concave, two convex, and one linear). RESULTS: All approaches described predicted the rankings of health states well. However, utilities derived followed an unconventional, bunched shape compared with the original Measurement and Valuation of Health TTO study. An approach was identified by optimizing the parameters, minimizing the sum of squared errors between the ordinal "health state ranking" approach and the original TTO-derived utilities. CONCLUSIONS: This approach outlined offers the potential to convert ordinal preference data into cardinal utilities. It is simpler than TTO studies to carry out and removes the need to directly alter results of the preference ranking exercise.

Explaining spatial accessibility to high-quality nursing home care in the US using machine learning.

In this study we measure and map the system-wide spatial accessibility to good quality nursing home care for all counties in the contiguous United States, and use an 'imputed post-lasso' machine learning technique to systematically examine this accessibility measure's associations with a broad range of county-level socio-demographic variables. Both steps were carried out using publicly available datasets. Analyses found clear evidence of spatial patterning in accessibility, particularly by population density, state and the populations of specific racial minorities. This has implications for outcomes that extend beyond the care homes and we highlight a number of policy measures that may help to address these shortcomings. The 'out-of-sample' predictive performance of the machine learning approach highlights the method's usefulness in identifying systematic differences in accessibility to services.

Developing composite indices of geographical access and need for nursing home care in Ireland using multiple criteria decision analysis.

Background: Spatial accessibility has consistently been shown to influence utilisation of care and health outcomes, compared against local population needs. We sought to identify how appropriately nursing homes (NHs) are distributed in Ireland, as its NH market lacks central planning. Methods: We used multiple criteria decision analysis (MCDA) approaches to develop composite indices of both access (incorporating measures of availability, choice, quality and affordability) and local NH need for over 65s (relating to the proportion living alone, with cognitive disabilities or with low self-rated health, estimated scores for activities of daily living and instrumental activities of daily living, the average number of disabilities per person and the average age of this group). Data for need were derived from census data. Results were mapped to better understand underlying geographical patterns. Results: By comparing local accessibility and need, underserved areas could be identified, which were clustered particularly in the country's northwest. Suburbs, particularly around Dublin, were by this measure relatively overserved. Conclusions: We have developed multi-dimensional indices of both accessibility to, and need for, nursing home care. This was carried out by combining granular, open data sources and elicited expert/stakeholder opinion from practitioners. Mapping these data helped to highlight clear evidence of inequitable variation in nursing home distribution.

Data linkage in medical science using the resource description framework: the AVERT model.

There is an ongoing challenge as to how best manage and understand 'big data' in precision medicine settings. This paper describes the potential for a Linked Data approach, using a Resource Description Framework (RDF) model, to combine multiple datasets with temporal and spatial elements of varying dimensionality. This "AVERT model" provides a framework for converting multiple standalone files of various formats, from both clinical and environmental settings, into a single data source. This data source can thereafter be queried effectively, shared with outside parties, more easily understood by multiple stakeholders using standardized vocabularies, incorporating provenance metadata and supporting temporo-spatial reasoning. The approach has further advantages in terms of data sharing, security and subsequent analysis. We use a case study relating to anti-Glomerular Basement Membrane (GBM) disease, a rare autoimmune condition, to illustrate a technical proof of concept for the AVERT model.