Journal of Health Care and Prevention
Open Access

Public health; Physical activity; Modeling techniques

Introduction

Given the limited resources at their disposal, decision-makers must commission treatments based not just on their effectiveness but also on their cost-effectiveness. Economic evaluation is frequently used to assist reimbursement decisions for funding interventions if there are several options. Numerous studies that examined the cost-effectiveness data for encouraging physical activity (PA) in the general population discovered that the treatments were, for the most part, cost-effective. Methodological reviews have, however, drawn attention to a number of difficulties with the economic assessment of public health [1-10] initiatives, such as PA. These difficulties have previously been divided into four major categories: attribution of effects, measuring and valuing outcomes, intersectoral costs and consequences, and incorporating equity concerns. They cover every aspect of the evaluation, from research design of the intervention to statistical and economic analyses. A recent study looked at two categories of PA treatments (targeted PA and sedentary behaviour) to see how the four methodological issues mentioned above have been handled in applied studies. This research, which confirmed earlier findings, discovered a general lack of reporting quality and obvious variations in the methodologies used across economic evaluations. It then offered a number of recommendations for the creation, analysis, and evaluation of economic evaluations. The methodological reviews that have already been published did not include a critique of the suitability of the modelling techniques employed to calculate the effects of changes in PA on population health. More precisely, they made no comments about whether the [2,6] structural aspects of the modelling technique were suitable or whether they were consistent with the basic characteristics of the behavior-population health dynamic they were intended to depict. This is crucial because faulty model assumptions and inadequate structure can undermine the reliability of cost-effectiveness findings. More precisely, in terms of quantifying efficacy, the [9, 6] literature has explored a number of difficulties in modelling healthy behaviours for public health economic evaluations, including PA. These difficulties stem from the fundamental complexity of the relationship between behaviour and population health, which necessitates the use of modelling. These difficulties include: 1) the relationship between multiple chronic diseases; 2) the dynamic nature of behaviour; and 3) heterogeneous responses to the intervention. A recent scientific report , which forms the basis of the current UK PA guidelines , describes the relationship between PA and the downstream risk of disease, by assessing the relevant available evidence from systematic reviews and meta-analysis against five criteria (i.e. applicability, generalisability, risk of bias or study limitations, quantity and consistency and magnitude and precision of effect) (i.e. applicability, generalisability, risk of bias or study limitations, quantity and consistency and magnitude and precision of effect). This study confirmed that there is compelling evidence linking PA to conditions related to metabolism (such as type II diabetes), the cardiovascular system (such as coronary heart disease and stroke), genetic mutation (such as colorectal and breast cancer), the mind (such as depression), and the elderly (such as falls). Furthermore, the likelihood of an [11] illness may not be independent of one another. For instance, type II diabetes risk has been linked to Breast and colorectal cancer risks are mostly a result of common risk factors, with PA playing a significant role. From a modelling perspective, this intricacy is a technological challenge since, depending on their PA level and other personal factors, healthy individuals are always exposed to a variety of competing and complementary illness risks. In addition, while some disease risks will be immediately impacted by changes in PA behaviours (such as psychological advantages), others will take years to manifest (such as the incidence of colon cancer). While most people's PA habits remain consistent throughout adulthood, a drop in PA is frequently linked to getting older. Evidence has been demonstrated that seasonal factors or specific life phases might cause natural variations in PA. Furthermore, it is reasonable to anticipate that different people will respond to exposure at the same level in different ways in terms of changes in behaviour, as well as that these changes will persist over time at various rates. Although the size of the effect will vary depending on the type of intervention, it is generally more likely that rebound trajectories will occur than it is to assume that changes brought about by the intervention in the short term will persist over time. Last but not least, it is critical to consider the heterogeneity in the natural history of the PA behaviour-health process and the effects of interventions on this process for two key reasons: to lower the risk of producing bias in the cost-effectiveness results and to align the modelling approach with the goals of the decision-makers the model is intended to inform. Because society values eliminating unfair inequities combined with increasing health, public health decision-makers prioritise minimising existing health disparities in the population, such as those related to socioeconomic variables. Numerous studies have been conducted with the goal of creating taxonomies of the mathematical and epidemiological paradigms that health economic modellers might use. This has also been taken into consideration in order to enlighten modelbased economic evaluations in the field of public health. Based on their capacity to explicitly represent time-dependent effects and interactions between people and their environment, these frameworks have been broadly categorised into cohort and individual level methods. To put it briefly, cohort-level frameworks are typically simpler than individuallevel ones. With decision trees and comparative risks assessments (CRAs), which have the highest modelling capacity, neither time nor interactions can be explicitly taken into account. Instead of representing time in the process implicitly as a series of state [12-15] transitions, Markov chains, which may be applied to both individuals and cohorts, can explicitly express time in the process. More complicated are discrete time events and agent-based models, which have only seen limited use in public health despite their formal capacity to depict changes in states over time and interactions between people (the latter), using either discrete or continuous time frameworks. The implementation of these methods in practise (i.e., the structural assumptions used) can have an impact on the validity of cost-effectiveness conclusions in addition to the applicability of the modelling framework. Similar to how the National Institute of Health and Care Excellence (NICE) assesses models submitted by manufacturers, previews have been made in other public health evaluation settings to question the veracity of the models' fundamental structural assumptions. No rigorous study has, to date, specifically looked into these problems in the PA literature. This gap is to be filled by the current paper.

Methods and Discussion

The search approach, eligibility requirements, study screening, and selection procedures are all given in detail in Appendix I. From the beginning of the database through April 2019, model-based economic evaluations of PA interventions were found in the published literature. Only complete economic evaluations were included due to the review's focus (i.e. cost-utility, cost-benefit, cost-consequences and cost-benefit analyses).

Main findings

This methodological study, which complements earlier reviews , offers an overview and assessment of the modelling techniques used in model-based [12-15] economic evaluations for evaluating implications of changes in PA on public health. The key structural presumptions that underlie the models have been clarified by this assessment, which can help comprehend the cost-effectiveness results and point up potential areas for model development.

Results

There were 25 different models found. The most popular models were cohort models. Across studies analysing comparable populations, there was significant variation in the modelling of downstream diseases. Most of the time, structural assumptions about the dynamics of changing physical activity were erroneous. Only a few research addressed heterogeneity, and writers at best acknowledged the issue of health equity.

Conclusions

The majority of this material is characterised by modelling techniques that do not fully meet the challenges of illustrating the relationship between physical activity behaviour and population health. These sources of uncertainty might be diminished with the creation of a reference model and agreement on how to model the effects of physical exercise on public health.

Acknowledgement

As a member of the National Institute for Health Research Collaboration for Leadership in Applied Health Research and Care Yorkshire and Humber, Paolo Candio received support through the White Rose PhD Studentship Network programme (NIHR CLAHRC YH).

Conflict of Interest

The authors declare that they have no competing interests.

1. Sinclair JR (2019) Importance of a One Health approach in advancing global health security and the Sustainable Development Goals. Revue scientific et technique 38: 145-154.

Indexed at, Google Scholar, Crossref

2. Aslam B, Khurshid M, Arshad MI, Muzammil S, Rasool M, et al. (2021) Antibiotic resistance: one health one world outlook. Frontiers in Cellular and Infection Microbiology 1153.

Indexed at, Google Scholar, Crossref

3. Doherty R, Madigan S, Warrington G, Ellis J (2019) Sleep and nutrition interactions: implications for athletes. Nutrients 11:822.

Indexed at, Google Scholar, Crossref

4. Jagannath A, Taylor L, Wakaf Z, Vasudevan SR, Foster RG, et al. (2017) The genetics of circadian rhythms, sleep and health. Hum Mol Genet 26: 128-138.

Indexed at, Google Scholar, Crossref

5. Somberg J (2009) Health Care Reform. Am J Ther 16: 281-282.

Indexed at, Google Scholar, Crossref

6. Wahner-Roedler DL, Knuth P, Juchems RH (1997) The German health-care system. Mayo Clin Proc72: 1061-1068.

Indexed at, Google Scholar, Crossref

7. Nally MC (2009) Healing health care.J Clin Invest119: 1-10.

Indexed at, Google Scholar, Crossref

8. Weinstein JN (2016) An “industrial revolution” in health care: the data tell us the time has come.Spine41: 1-2.

Indexed at, Google Scholar, Crossref

9. Marshall EC (1989) Assurance of quality vision care in alternative health care delivery systems. J Am Optom Assoc 60: 827-831.

Indexed at, Google Scholar

10. Cutler (2021) Building health care better means reining in costs. In JAMA Health Forum 2: 210117-210117.

Indexed at, Google Scholar, Crossref

11. Lindeque BG (2009) American Health Care System Disaster. Orthopedics 32: 551.

Indexed at, Google Scholar, Crossref

12. Ampomah IG, Malau-Aduli BS, Malau-Aduli AE, Emeto T I (2020) Effectiveness of integrated health systems in Africa: a systematic review. Medicina 56: 271.

Indexed at, Google Scholar, Crossref

13. Irwin MR, Opp MR (2017) Sleep health: reciprocal regulation of sleep and innate immunity. Neuropsychopharmacology 42:129-155.

Indexed at, Google Scholar, Crossref

14. Gamache R, Kharrazi H, Weiner JP (2018) Public and population health informatics: the bridging of big data to benefit communities. Yearbook of medical informatics 27: 199-206.

Indexed at, Google Scholar, Crossref

15. Jandoo T (2020) WHO guidance for digital health: what it means for researchers.

Google Scholar, Crossref