Search

Decision-analytic modelling is used to synthesise clinical and economic data to evaluate health outcomes and costs of novel healthcare interventions. Diabetes is associated with disabling and costly complications which develop over time, and such models are crucial to capture the impact of treatment over a person's lifetime. Most diabetes simulation models to date are based on risk equations estimated using data from the UK Prospective Diabetes Study (UKPDS), a study conducted more than two decades ago, and may not reflect contemporary event rates given changes in treatment and management of diabetes and its complications. A Study of Cardiovascular Events iN Diabetes (ASCEND) recruited 15,480 participants with diabetes without previous cardiovascular disease, and is one of the largest diabetes trials conducted to date. Using data from the ASCEND study, I found the UKPDS risk equations to overpredict cardiovascular events and mortality, further adding to the evidence base that the UKPDS risk equations are unsuitable for use in contemporary cohorts. I developed a new framework for modelling long-term health outcomes and costs which can be used to assess cost-effectiveness of treatments for people with diabetes without previous cardiovascular disease. New risk equations were estimated using the ASCEND data, and these were integrated into a simulation model which predicts the likelihood of occurrence of cardiovascular, bleeding, cancer, lower-limb amputation, end-stage renal disease and death events over the lifetime of a patient given their baseline characteristics. The simulation model was externally validated in a cohort of 18,250 participants with diabetes without cardiovascular disease identified from the UK Biobank, a longitudinal prospective observational cohort. The simulation model performed well generally. Where discrepancies between model predictions and observed event rates were found, the reasons behind the discrepancies were identified. A novel issue-driven approach was developed to update the risk equations. New estimates of hospital costs and health-related quality of life (QoL) associated with the adverse events in the simulation model were also derived from the ASCEND study and incorporated into the simulation model to quantify the impact of treatments on lifetime cost and QoL. To demonstrate an application of the framework, I assessed the cost-effectiveness of aspirin for the primary prevention of cardiovascular disease in people with diabetes. In this population, it is still uncertain if the cardiovascular benefits of aspirin use outweigh the bleeding risks. Using the latest evidence available, I demonstrate that aspirin is likely to be cost-effective in the cohort of UK Biobank participants with diabetes without previous cardiovascular disease. However, significant uncertainties in the results warrant further research.

The rapid economic, demographic, and epidemiologic transitions that occurred in China in recent decades have increased the burden of non-communicable diseases, including cardiovascular diseases (CVD), as well as health expenditures and health inequalities. In 2009, China launched major healthcare reforms to provide equal and affordable basic healthcare for all by 2020. This thesis aims to study healthcare use for CVD across socioeconomic groups in China by examining: (i) trends in hospital care utilisation and outcomes for stroke and ischaemic heart disease (IHD) by socioeconomic groups over an 8-year period; (ii) gender differences in the management of acute IHD; and (iii) impact of health insurance (HI) cost sharing and socioeconomic characteristics on choice of hospital tiers for stroke and IHD. The analyses involved half a million adults in the China Kadoorie Biobank, recruited in 2004-2008 and followed-up until 2017 through linked HI records, and disease and death registries. After adjustment for demographic, socioeconomic, lifestyle and morbidity factors, hospitalisation rates for stroke and IHD each increased by 4-5% per year in 2009-2016. Higher socioeconomic groups had higher hospitalisation rates, but the annual proportional increases were greater among those with lower levels of education or income, urban or rural resident HI schemes, and in rural areas. Lower socioeconomic groups had higher stroke and IHD case fatality rates, but achieved greater proportional reductions in these during 2009-2016 than higher socioeconomic groups. The average length of hospital stay decreased by 2% annually and to a greater extent in higher than lower socioeconomic groups. There were marked gender differences in the use of invasive coronary tests and procedures for acute IHD in 2004-2016. While the use of cardiac enzymes did not differ between women and men with acute myocardial infarction, women had 20% lower rates of coronary angiography and 15% lower rates of revascularisation. Compared to men, women with angina or other IHD had 3%, 13-34% and 39-44% lower rates of having cardiac enzyme tests, coronary angiography and revascularisation, respectively. However, socioeconomic and health system factors did not explain the gender differences in the use of invasive tests and procedures for IHD. In China, hospitals are classified into three tiers, with higher tier hospitals being larger, better equipped, and typically located in urban areas. While over 60% of urban residents with first stroke or IHD admissions in 2009-2017 chose tier 3 hospitals, only around 30% of rural residents chose tier 3 hospitals. In urban areas, higher reimbursement rates were associated with higher odds of choosing all hospital tiers, and higher tier 3 deductibles with lower odds of choosing tier 3 hospitals. In rural areas, higher tier 3 reimbursement rates were associated with higher odds of choosing tier 3 hospitals, and higher tier 1 deductibles with lower odds of choosing tier 1 hospitals. However, higher tier 2 and 3 deductibles in rural areas were associated with higher odds of choosing these hospitals. Individuals with higher socioeconomic status and, separately, more severe disease were more likely to choose tier 3 than tier 1 or 2 hospitals. These findings should inform further strategies to reduce socioeconomic and gender disparities and facilitate choice of appropriate hospital tiers in the use of hospital care for different types of CVD across China.

Previous studies have shown that use of effective and widely recommended treatments, specifically statin and antiplatelet therapy, for the secondary prevention of atherosclerotic cardiovascular disease (ASCVD) is suboptimal. However, knowledge regarding the underlying reasons for suboptimal medication use at different treatment stages, including the role of particular individual characteristics, and the variations in medication use across different population groups and ASCVD categories, is limited. I undertook the first systematic literature review of studies that estimated the proportions and determinants of statin and antiplatelet therapy for the secondary prevention of ASCVD among myocardial infarction and ischaemic stroke patients at every stage of the treatment pathway, namely medication prescription on discharge, initiation of treatment after discharge, adherence to and persistence with treatment. The review presented evidence of gradually increasing, yet suboptimal medication use at almost every stage of the pathway for both statin and antiplatelet therapy. Multiple patient and provider- related risk factors of suboptimal medication use were identified. Women were less likely to receive statin and antiplatelet therapy (except for aspirin) on discharge, and less likely to initiate statin. A U-shaped association was observed for risk of statin discontinuation with age, where elderly individuals aged 69 years or older and those younger than 64 years were at greater risk. Prior CVD and presence of specific physical comorbidities were associated with suboptimal statin use at various treatment stages. Most of these studies had a limited follow-up time and examined a maximum of two treatment stages, thereby limiting the ability to generalise results and insights into the full extent of suboptimal use across the entire treatment pathway and over time. Using linked NHS Scotland administrative hospitalisation and prescription information data on all individuals in Scotland ever hospitalised for an ASCVD event since 2009 with a follow-up time of up to eight years, I estimated the extent of use and patient characteristics associated with suboptimal statin and antiplatelet therapy use. About 82% and 84% of individuals hospitalised for ASCVD event subsequently initiated statin and antiplatelet therapy, respectively, with rates varying significantly by ASCVD type. Although more than 90% of individuals who initiated statin and/or antiplatelet therapy were adherent while on medication, one quarter of individuals eventually discontinued treatment. Of those who discontinued, 50% of individuals ceased statin and/or antiplatelet therapy within 1.5 years, and 80% within 3.5 years, despite clinical guidelines recommending lifelong treatment. Among individuals who discontinued treatment, 38% and 37% of individuals re-initiated some type of statin and antiplatelet therapy at some point in time, and on average within 1.1 years since discontinuation. Patient phenotypes that were consistently associated with higher risks of non-initiation and discontinuation of treatment included women, individuals aged below 50 years or above 70 years (compared to those aged 60 to 69 years), individuals with multiple physical morbidities, individuals with a previous record of receiving specialist mental health care, patients residing in the most deprived areas (compared to those living in the least deprived areas), and patients hospitalised in the earlier time period in study (i.e. before 2015). In addition to patient-related factors, this Thesis showed that the intensity of statins prescribed was significantly associated with market supply forces, such as the market entry of lower cost generic atorvastatin in 2012, and further accelerated via health care system demand-side measures, such as clinical guidelines and educational initiatives advocating for the use of higher intensity statin treatment. The overall prescribing budget and the notion of relative cost-effectiveness of different statin types are likely key influencing factors in prescribing decisions. These findings will be useful to healthcare policy makers and providers in making investment and prioritisation decisions, and underline calls for interventions designed to improve drug initiation and persistence to improve health and curtail rising costs due to preventable vascular events.

Introduction: Stroke, a major cause of disability and death worldwide, can be caused by carotid artery stenosis. Detection and treatment of carotid stenosis has substantial health and economic impact worldwide. It is therefore important to evaluate the long-term effects and cost-effectiveness of treatment methods for carotid stenosis, namely medical therapy (MT), carotid endarterectomy (CEA) in addition to MT, and carotid artery stenting (CAS) in addition to MT. Methods: The literature was systematically reviewed for economic evaluations comparing treatment methods for carotid stenosis, to summarise current evidence and identify limitations of available economic evaluations. Randomised Controlled Trials (RCTs) and large observational cohort studies were systematically reviewed to summarise information on the clinical effectiveness of treatment methods. Information from this evidence was then combined with information from the literature on quality of life and healthcare costs relevant to carotid stenosis patients, to inform a new decision analytical model, evaluating the cost-effectiveness of all available treatment methods for carotid stenosis patients in the UK. Results: There was a clear need in the literature for an economic evaluation comparing all available treatment methods for carotid stenosis in a single framework. Other limitations were identified relating to model structure, input parameters used, and reporting of results. Results synthesised from RCTs show that CEA has lower procedural stroke/death risks and lower overall stroke risks when compared to CAS, with differences mostly attributable to the larger number of minor strokes following CAS. When compared to MT alone, CEA reduced the risk of the combined endpoint of any stroke or procedural death. Observational studies showed that procedural stroke/death risks of CEA have decreased overtime, whilst procedural stroke/death risks of CAS patients appear to be stable. Results from the cost-effectiveness framework comparing all treatment methods show that, at willingness to pay for a quality adjusted life year (QALY) of £20,000, CEA in addition to MT is the most cost-effective strategy in patients up to the age of 75 years. MT alone was the most cost-effective option for patients 75 years or older. Treatment decisions based on the cost-effectiveness analysis remained the same when considered by gender or symptomatic status of patients. The value of information analysis indicated that further information about the relative post-procedural stroke risks of CEA plus MT vs MT alone is most valuable to reduce uncertainty in treatment decisions. Conclusion: The cost-effectiveness framework and evidence developed in this thesis addresses directly the limitations in the literature. It used information about procedural CEA risks in recent observational cohort studies and relative treatment effects from meta-analysis of RCTs, showing that optimal treatment decisions depend on the age of patients, with CEA plus MT being cost-effective in patients up to the age of 75 but MT alone likely to be more cost-effective in older people. These findings should be highly relevant in the development of future treatment guidelines.

Rates of overweight and obesity worldwide have increased substantially in recent decades. In England, over 60% of adults are now overweight or obese. Because excess weight is associated with higher risks of chronic conditions, including type-2 diabetes, cardiovascular disease, osteoarthritis, and certain cancers, as well as with premature death, it constitutes a major health burden. It is also associated with higher total healthcare spending, but how the associations differ in different healthcare settings is not well understood. I undertook a systematic literature review of studies that used individual participant data to estimate annual healthcare costs in relation to body mass index (BMI; in kg/m2). Compared to adults at healthy weight (i.e. BMI 18.5 to <25 kg/m2), total annual healthcare costs were found to be 12% and 36% higher for overweight (BMI 25 to <30 kg/m2) and obese (BMI >=20 kg/m2) adults, respectively. The associations were strongest for costs of medications (18% for overweight and 68% for obese adults), followed by inpatient care costs (12% and 34%), and ambulatory care costs (4% and 26%). Most of these studies used data from the United States, and were based on small-to-moderate numbers of participants, limiting their ability to reliably estimate healthcare costs in relation to grades of obesity or for different health conditions. There was no reliable evidence from individual participant data pertaining directly to the United Kingdom. Using data on over one million middle-aged and older women in the prospective Million Women Study, linked to routine administrative data on deaths, inpatient and day-case admissions, and primary care services, I estimated annual rates and costs of hospital admissions, primary care consultations, prescription items issued, and monitoring and diagnostic tests in relation to body mass index. Every 2 kg/m2 higher BMI beyond 20 kg/m2 was associated with 7.4% higher annual hospital admission costs, 5.2% higher primary care consultation costs, and 9.9% higher prescription costs, but no clear association with test costs was identified. Projecting these results to the total population of women aged 55 to 79 years in England in 2013, 15% (£662 million) of hospital costs, 11% (£229 million) of primary care consultation costs, and 22% (£384 million) of prescription costs were attributable to overweight and obesity. Type-2 diabetes, and cardiovascular and musculoskeletal diseases were the major contributing conditions to the total excess weight attributable costs. These findings will be useful to healthcare policy makers, commissioners, and providers in making investment and prioritisation decisions, and underline calls for greater investments in cost-effective interventions to reduce excess weight and prevent weight gain.