How Responsive is Mortality to Locally Administered Healthcare Expenditure? Estimates for England for 2014/15

An Approach for Combining Clinical Judgment with Machine Learning to Inform Medical Decision Making: Analysis of Nonemergency Surgery Strategies for Acute Appendicitis in Patients with Multiple Long-Term Conditions

BACKGROUND

Machine learning (ML) methods can identify complex patterns of treatment effect heterogeneity. However, before ML can help to personalize decision making, transparent approaches must be developed that draw on clinical judgment. We develop an approach that combines clinical judgment with ML to generate appropriate comparative effectiveness evidence for informing decision making.

METHODS

We motivate this approach in evaluating the effectiveness of nonemergency surgery (NES) strategies, such as antibiotic therapy, for people with acute appendicitis who have multiple long-term conditions (MLTCs) compared with emergency surgery (ES). Our 4-stage approach 1) draws on clinical judgment about which patient characteristics and morbidities modify the relative effectiveness of NES; 2) selects additional covariates from a high-dimensional covariate space (P > 500) by applying an ML approach, least absolute shrinkage and selection operator (LASSO), to large-scale administrative data (N = 24,312); 3) generates estimates of comparative effectiveness for relevant subgroups; and 4) presents evidence in a suitable form for decision making.

RESULTS

This approach provides useful evidence for clinically relevant subgroups. We found that overall NES strategies led to increases in the mean number of days alive and out-of-hospital compared with ES, but estimates differed across subgroups, ranging from 21.2 (95% confidence interval: 1.8 to 40.5) for patients with chronic heart failure and chronic kidney disease to -10.4 (-29.8 to 9.1) for patients with cancer and hypertension. Our interactive tool for visualizing ML output allows for findings to be customized according to the specific needs of the clinical decision maker.

CONCLUSIONS

This principled approach of combining clinical judgment with an ML approach can improve trust, relevance, and usefulness of the evidence generated for clinical decision making.

HIGHLIGHTS

Machine learning (ML) methods have many potential applications in medical decision making, but the lack of model interpretability and usability constitutes an important barrier for the wider adoption of ML evidence in practice.We develop a 4-stage approach for integrating clinical judgment into the way an ML approach is used to estimate and report comparative effectiveness.We illustrate the approach in undertaking an evaluation of nonemergency surgery (NES) strategies for acute appendicitis in patients with multiple long-term conditions and find that NES strategies lead to better outcomes compared with emergency surgery and that the effects differ across subgroups.We develop an interactive tool for visualizing the results of this study that allows findings to be customized according to the user's preferences.

Systematic review of the impact of health care expenditure on health outcome measures: implications for cost-effectiveness thresholds

OBJECTIVE

Empirical estimates of the impact of healthcare expenditure on health outcome measures may inform the cost-effectiveness threshold (CET) for guiding funding decisions. This study aims to systematically review studies that estimated this, summarize and compare the estimates by country income level.

METHODS

We searched PubMed, Scopus, York Research database, and [anonymized] for Reviews and Dissemination database from inception to 1 August 2023. For inclusion, a study had to be an original article, estimating the impact of healthcare expenditure on health outcome measures at a country level, and presented estimates, in terms of cost per quality-adjusted life year (QALY) or disability-adjusted life year (DALY).

RESULTS

We included 18 studies with 385 estimates. The median (range) estimates were PPP$ 11,224 (PPP$ 223 - PPP$ 288,816) per QALY gained and PPP$ 5,963 (PPP$ 71 - PPP$ 165,629) per DALY averted. As ratios of Gross Domestic Product per capita (GDPPC), these estimates were 0.376 (0.041-182.840) and 0.318 (0.004-37.315) times of GDPPC, respectively.

CONCLUSIONS

The commonly used CET of GDPPC seems to be too high for all countries, but especially low-to-middle-income countries where the potential health losses from misallocation of the same money are greater.

REGISTRATION

The review protocol was published and registered in PROSPERO (CRD42020147276).

Achieving cardiac rehabilitation uptake targets: What is the value case for commissioners? A UK case-study

Cardiac Rehabilitation (CR) has become an established intervention to support patient recovery after a cardiac event, with evidence supporting its effectiveness and cost-effectiveness in improving patient health and reducing future burden on healthcare systems. However, this evidence has focussed on the national value case for CR rather than at the point at which it is commissioned. This analysis uses the UK as a case-study to explore variation in current CR engagement and disassemble the value case from a commissioner perspective. Using data collected by the National Audit of CR (NACR), and an existing model of cost-effectiveness, we present details on the current level of CR uptake by commissioning region (Specialist Clinical Networks) in light of the current UK target of achieving 85% uptake. We then interrogate the value case for achieving the target at a commissioner level, highlighting the expected profile of health benefits and healthcare system costs over the long-term. Importantly we consider where this may differ from the national value case. Each commissioning region has a unique level of CR uptake and sociodemographic profile. Concurrently, the value case for commissioning CR relies on the upfront cost of the service being offset by long-term healthcare savings, and health improvements. The shift in the UK and internationally to more localised commissioning necessitates evidence of cost-effectiveness that better reflects the realities of those decision makers. This paper provides vital additional data to facilitate such commissioners to understand the value case in increasing CR uptake in line with national policy.

The impact of different types of NHS expenditure on health: Marginal cost per QALY estimates for England for 2016/17

English data from 2003 to 2012 suggests that it costs the NHS £10,000 to generate an additional quality-adjusted life year (QALY).  This estimate relates to all NHS expenditure and no attempt was made to explore possible heterogeneity within this total.  Different types of expenditure - such as secondary care, primary care and specialized commissioning - may have different productivities and estimates of these may help policymakers decide where additional investment is most beneficial.  We use the two-stage least squares estimator and data for 2016 to explore the mortality response to three types of healthcare expenditure.  Three specifications are estimated for each type of expenditure: backward selection and regularized regression are used to identify parsimonious specifications, and a full specification with all covariates is also estimated.  The regression results are combined with information about survival and morbidity disease burden to calculate the marginal cost per QALY for each type of expenditure: the most conservative results suggest that this is about £8,000 for locally (CCG) commissioned services, while estimates for specialized commissioning and primary care are more uncertain.  When this heterogeneity is taken into account, the estimated marginal cost per QALY for all NHS expenditure increases slightly, from about £6,000 to £7,000.  Our results suggest that additional investment is likely to be most productive in primary care and in locally commissioned services.

The cost-per-QALY threshold in England: Identifying structural uncertainty in the estimates

Introduction

There are increasing numbers of estimates of opportunity cost to inform the setting of thresholds as ceiling cost-per-quality-adjusted life year (QALY) ratios. To understand their ability to inform policy making, we need to understand the degree of uncertainty surrounding these estimates. In particular, do estimates provide sufficient certainty that the current policy "rules" or "benchmarks" need revision? Does the degree of uncertainty around those estimates mean that further evidence generation is required?

Methods

We analyse uncertainty and methods from three papers that focus on the use of data from the NHS in England to estimate opportunity cost. All estimate the impact of expenditure on mortality in cross-sectional regression analyses and then translate the mortality elasticities into cost-per-QALY thresholds using the same assumptions. All three discuss structural uncertainty around the regression analysis, and report parameter uncertainty derived from their estimated standard errors. However, only the initial, seminal, paper explores the structural uncertainty involved in moving from the regression analysis to a threshold. We discuss the elements of structural uncertainty arising from the assumptions that underpin the translation of elasticities to thresholds and seek to quantify the importance of some of the effects.

Results

We find several sets of plausible structural assumptions that would place the threshold estimates from these studies within the current National Institute for Health and Care Excellence (NICE) range of £20,000 to £30,000 per QALY. Heterogeneity, an additional source of uncertainty from variability, is also discussed and reported.

Discussion

Lastly, we discuss how decision uncertainty around the threshold could be reduced, setting out what sort of additional research is required, notably in improving estimates of disease burden and of the impact of health expenditure on quality of life. Given the likely value to policy makers of this research it should be a priority for health system research funding.