Abstract
The Universal Health Coverage (UHC) is a movement that promote health for all and is included among the Sustainable Development Goals. The World Health Organization (WHO), the World Bank and other institutions have proposed ‘effective coverage’ as an indicator for tracking progress toward UHC. Effective coverage is an indicator that combine two parameters: the coverage of a healthcare intervention and the quality of it. According to Shengelia et al.’s framework for effective coverage, quality must be measured through the health gains associated with the utilization of a healthcare service. However, none of these organizations above mentioned have properly measured effective coverage due to the lack of information about the quality of healthcare services.In a systematic scoping review aimed to identify the use of effective coverage in the scientific literature, I found 128 studies preforming 246 assessments of healthcare interventions in 138 countries – 81% low and middle-income countries. Only one assessment included health gains into the parameter of quality. The aim of this thesis is to develop a practical procedure to estimate the effective coverage considering health gains.
Calculating health gains is related with valuating health states. There are two main approaches for valuing health states: those that look for social preferences or utilities (e.g. standard gamble, time-trade off, person -trade-off), and those that look for a direct measure of the health status or disability (e.g. paired comparison). The first approach produces ‘health-utility weights’, and the second approach produces ‘disability weights’. Traditionally, disability weights are used for calculating disability-adjusted life years (DALY) a measure of burden of disease, while health-utility weights are used for calculating quality-adjusted life years (QALY) a measure used in cost-effectiveness analysis.
Using data from the Chilean National Health Survey 2009-2010 (Ch-NHS 2009-2010) and the Health States Description questionnaire included in that survey, I calculated a latent variable of disability. I argue that through a regression model applied to a latent variable of disability or health-states utilities, it is possible to estimate disability weights (or health-utility weights) for different health states associated with a disease, adjusting by comorbidities and other confounders.
The attributable fraction encompasses a family of epidemiological estimators that combine relative and absolute effect sizes. Attributable fractions have been used mainly for exploring the effect of risk factors on diseases. Using the attributable fraction metrics applied to a continuous outcome such us a latent variable of disability or health-state utility, I present a new way of calculating the burden of disability (or the loss of health-state utilities) associated with diseases. This methodological proposal would be more straightforward to be carried out than the standard methodological alternative (i.e. years lived with disability, a component of DALYs). Two approaches to calculate the burden of disability attributable to diseases are presented: the population average-level and the individual-level.
I also argue that the procedure to calculate the burden of disability (or loss of health-state utilities), described above, can be used to estimate effective coverage. I define effective coverage as the fraction of avoidable disability (or loss of health-state utilities) attributable to a disease, avoided by using a healthcare intervention. I also propose a definition for other related indicators: health benefit, quality, relative effective coverage (r-EC) and absolute effective coverage (a-EC). While effective coverage results from the combination of the coverage and quality, the r-EC results from the combination of the coverage and the health benefit (i.e., effectiveness). a-EC is defined as the fraction of the disability attributable to a disease in the entire population that is avoided by the healthcare intervention. This indicator is suitable to be combined with costs associated with healthcare services. The procedure to estimate these indicators is tested initially using data from the Ch-NHS 2009-2010 applied to the case of treatment for depressive disorder.
A more comprehensive appraisal of the performance of the procedure to calculate effective coverage and other indicators is also carried out using cross-sectional data from WHO study on global ageing and adult health (SAGE), Wave 1, undertaken between 2007-2010 in China, Ghana, India, Mexico, the Russian Federation, and South Africa. Three healthcare interventions were explored: treatment for depressive disorder, treatment for hypertension and treatment for osteoarthritis.
The methodological proposal for calculating effective coverage achieves estimating health gains into a parameter of quality using cross-sectional data. Among the strengths of the proposal developed in this thesis I highlight: (1) the concept of effective coverage is expanded through new indicators; (2) the procedure is straightforward to be implemented; (3) it depends on only one source of information, which ensures consistency between parameters; and (4) it can be used indistinctly with different outcomes: disability or health-state utilities.
However, its main limitation is that the effect size attributable to the healthcare intervention is weak because the procedure proposed in this thesis is based on cross-sectional data. To improve the methodological proposal of this thesis, I highlight the following challenges for future research: (1) exploring other procedures to obtain a better proxy of the effect size of healthcare interventions using cross-sectional data (e.g. propensity score matching, instrumental variables); (2) including fatal consequences; (3) including an equity perspective in the outcome; and (4) exploring combining a-EC with the costs of healthcare interventions.
Regarding tracking progress toward UHC, I argue that a-EC would be a more adequate indicator than effective coverage. a-EC includes in a single metric the effectiveness of healthcare services, the coverage of it, the disability associated with a disease, and the prevalence of such disease. Moreover, a-EC can be added across different healthcare- services in a simpler way than with other indicators. Finally, it can be combined with the cost of the healthcare interventions, which is appropriate to inform decision makers.
Date of Award | 1 Nov 2022 |
---|---|
Original language | English |
Awarding Institution |
|
Supervisor | Ricardo Araya Baltra (Supervisor) & Ioannis Bakolis (Supervisor) |