Whither economic evaluation in the case of COVID-19: What can the field of mental health economics contribute within the Australian context?

Abstract

COVID-19 has resulted in broad impacts on the economy and aspects of daily life including our collective mental health and well-being. The Australian health care system already faces limitations in its ability to treat people with mental health diagnoses. Australia has responded to the COVID-19 outbreak by, among other initiatives, providing reimbursement for telehealth services. However, it is unclear if these measures will be enough to manage the psychological distress, depression, anxiety and post-traumatic distress shown to accompany infectious disease outbreaks and economic shocks. Decision making has focused on the physical health ramifications of COVID-19, the avoidance of over-burdening the health care system and saving lives. We propose an alternative framework for decision making that combines life years saved with impacts on quality of life. A framework that simultaneously includes mental health and broader economic impacts into a single decision-making process would facilitate transparent and accountable decision making that can improve the overall welfare of Australian society as we continue to address the considerable challenges that the COVID-19 pandemic is creating.

Introduction

There are many elements that need to be quickly considered during the escalating and continuing spread of COVID-19. The urgency is primarily driven by the paucity of information on how this virus will behave over the longer term – particularly in terms of community infection and re-infection rates. A false dichotomy has emerged in both the policy and public discourse (e.g. numerous mainstream media reports) on the supposedly mutually exclusive choices between population health versus the economy; and even between different dimensions of health, such as mortality (risk of death) versus morbidity (e.g. quality of life impacts). We propose the need for a more comprehensive evaluation framework, which considers the cost and benefits of different courses of action that do not disadvantage mental health or other areas of health.

On what basis are decisions currently being made during the COVID-19 pandemic?

Much of the rationing narrative that we are reading about is a direct consequence of the constrained supply of health care resources (such as ventilators, intensive care units and personal protective equipment) that may become scarce relative to needs if infection rates climb. There is also a threat of an unprecedented rise in mortality – something we are observing internationally. Critical decisions need to be made on how limited health care resources should be allocated, thus prioritizing who will live or die during the coronavirus peak. A recent paper has provided six recommendations for developing guidelines to fairly allocate scarce medical resources in the midst of COVID-19 (Emanuel et al., 2020). In Australia, Sydney Health Ethics released a framework to help clinicians, hospital administrators and policy makers decide how to allocate health care resources during the COVID-19 pandemic (Sydney Health Ethics, 2020). Such frameworks are, however, narrowly focused on decision making within the health sector alone.

We have seen confronting examples internationally, where frontline doctors have been forced to deny lifesaving treatment to those with a lower chance of survival. This utilitarian rationing principle seeks to maximize societal benefits (i.e. life years saved). Priority is given to patients who can benefit most from treatment and who have greater chance of survival. This disadvantages older patients and those with pre-existing health problems. Decisions to prioritize the lives of younger people may also be supported by the ‘fair innings’ argument, which invokes a social norm where every person in society is entitled to some ‘normal’ span of life years (Williams, 1997). Another competing ethical principle being observed under current decision-making processes is the ‘rule of rescue’ (McKie and Richardson, 2003), which prioritizes saving lives that are in imminent danger regardless of the cost. Under this ethical principle, a society is willing to allocate additional resources over-and-above what it might otherwise allocate to health conditions with an equivalent mortality rate (but are not as ‘prominent’) or even conditions that might impact morbidity rather than mortality.

Are there other elements that need to be explicitly considered for decision making?

The initial lockdown and social distancing measures introduced in Australia, along with the even stricter lock down measures introduced more recently in Victoria, have led to escalating rates of unemployment, uncertainty and increasing levels of psychological distress – although these appear to have helped mitigate a catastrophic burden on our health care system. While government policy responses, such as the JobSeeker and JobKeeper payments, will help to alleviate some of the short-term economic impacts of the current lockdown measures, the impact on the economy over the medium to longer term is unknown. It is unlikely that all businesses will be able to ‘hibernate’ only to re-awaken after lockdown restrictions are lifted. Many businesses – especially those reliant on face-to-face client interactions and those exposed to international trade and travel – will have insufficient cash reserves to continue operating when no income is being generated. It is clear that Australia will be facing a severe economic recession (the first in 29 years) over the coming months.

Economic downturns are associated with increased psychological distress and poorer mental health outcomes (Frasquilho et al., 2016). Current evidence suggests that these outcomes are largely mediated through impacts on unemployment, job insecurity and indebtedness. For instance, previous systematic reviews have found that unemployed people are significantly more likely to experience psychological distress, depression, anxiety and poorer subjective well-being when compared to those who are employed (Paul and Moser, 2009). There is also a strong link between unemployment and suicide mortality that persists after controlling for prior mental health status (Milner et al., 2014). Beyond unemployment, increased job insecurity during economic downturns may reduce the mental well-being of employed workers who face poorer job prospects, fear the threat of prolonged unemployment and are less inclined to quit dissatisfying jobs (Frasquilho et al., 2016). People who struggle to meet regular loan repayments are more likely to experience poorer mental health, worse subjective health, increased rates of depression and increased suicidal ideation when compared to those without financial hardship (Turunen and Hiilamo, 2014). The mental health impacts of financial hardship due to indebtedness are particularly relevant to Australians who, on average, have the highest levels of household debt relative to household income both by historical standards and compared to other developed nations.

Psychological distress is not only a function of the impending uncertainty of an economic downturn but also a function of the uncertainty that comes from living through a pandemic. Evidence from recent communicable disease outbreaks (e.g. 2003 SARS, equine influenza) has found associated psychological distress and symptoms of anxiety, post-traumatic distress and depression that were still evident up to 3 years later (Wu et al., 2009). Health care workers at high risk of exposure appear to bear greater long-term psychological impacts (McAlonan et al., 2007). Quarantine also has negative psychological effects including post-traumatic stress symptoms (Brooks et al., 2020). Even so, these studies were limited to communicable disease outbreaks without an associated economic downturn.

While still under researched, increasing levels of loneliness may lead to even greater impacts on both physical and mental health, as well as early mortality (Holt-Lunstad et al., 2015). However, the impacts on psychological distress could be moving in opposing directions. Many are likely experiencing increasing distress because of the threat of a contracting economy and increasing feelings of loneliness. By contrast, others who might be in a fortunate position of not being personally impacted by the contracting economy may be experiencing decreasing levels of psychological distress as Australia proceeds to ‘flatten the curve’. Current research should address how our mental health is responding to such seemingly opposing phenomena (i.e. a worsening economy that occurs while infection rates decrease following the successful containment of COVID-19). Furthermore, the pandemic is likely to deepen social inequality, where those working in low-wage, part-time employment are more vulnerable to the COVID-19 lockdowns compared with white-collar workers who have been able to work from home. Evidence has also shown that ethnic minority communities experience a disproportionate burden of morbidity and mortality from COVID-19 (Rentsch et al., 2020). Moreover, the most vulnerable in our society, such as those with chronic health conditions, are at especially high risk of hospitalization, intensive care unit admission and mortality due to COVID-19. Finally, the impacts on children are unclear, although likely to be substantive, and there are currently a range of projects underway evaluating the impacts of COVID-19 on children.

Current mental health policy responses to COVID-19 – are they good enough?

While the response to the mental health impacts of COVID-19 has led to a welcome and necessary scale up of mental health services, both at the Commonwealth and State levels, it is worth pausing to consider the trade-offs. The scale up of this mental health care is happening within a system that is already underfunded and hence delivering sub-optimal care. The plethora of problems plaguing the Australian mental health system prior to COVID-19 have been aptly identified by both the Productivity Commission and the Victorian Royal Commission’s inquiries into mental health care – as presented in their respective interim reports. It follows that the Australian mental health system will struggle to cope more than prior to the pandemic if increased funding is less than proportionate to increased needs.

Delivering mental health services via e-health will limit unnecessary exposure of patients and health professionals to COVID-19 whereby treatment can be safely delivered through a web-based intervention, mobile app, email therapy, chat therapy or video therapy. While there is an existing evidence base for the effectiveness and cost-effectiveness of e-health, the findings from this evidence base is mixed. E-health has been found to be effective in reducing depressive and anxiety symptoms with a moderate to large effect size (Olthuis et al., 2016) although the evidence in some settings was less impressive (e.g. a general primary care setting). There is also mixed evidence to support the effectiveness of telehealth interventions for the treatment of substance use disorders. This is particularly pertinent in the current climate where anecdotal evidence suggests that alcohol consumption and other substance use may be on the rise. Fortunately, recent evidence has also shown that there are no significant differences in outcomes between e-health versus face-to-face (Langarizadeh et al., 2017) as well as unguided versus therapist-supported interventions (at least for anxiety disorders) (Olthuis et al., 2016), although patient satisfaction is generally reported to be higher with therapist-supported interventions. Finally, both local and international evidence has shown that e-health can be highly cost-effective (Donker et al., 2015; Lee et al., 2017) but there remain concerns with adherence to such interventions (Donker et al., 2015). A key policy report released by the Grattan Institute argues that the rapid roll-out of telehealth should ‘have been introduced long-ago’, within the Australian context (p. 55) and that the challenge now is to ensure that telehealth services are not abused and that no perverse incentives are created (Duckett et al., 2020). Furthermore, this report states that telehealth services should remain beyond COVID-19 as an approach to improve access to care, however such continuation needs to be carefully evaluated.

Towards open and accountable decision making that simultaneously includes mental health and broader economic impacts

While government decision making is currently being driven by expected impacts on mortality (i.e. saving years of life), it is clear that the impacts on quality of life, including mental health, are also considerable and likely to grow further. What we are currently observing is largely an exercise of implicit economic evaluation from a societal perspective where the costs of responding to the COVID-19 crisis include both health care and broader economic costs in other sectors (e.g. tourism, hospitality, retail, tertiary education). The costs are largely being considered separately. The benefits being considered are only the cases of COVID-19 that are potentially averted – with resultant life years saved. The only choices being formally and explicitly considered are different levels of community infection under different scenarios of exposure. Impacts on other sectors of the community are given one-off solutions – such as the JobSeeker/JobKeeper allowances for displaced workers and expanded mental health (largely e-health) services for people experiencing mental health distress. From an economic evaluation viewpoint, both the costs and consequences of alternative courses of action should be considered within one evaluation context, including costs within and outside the health care sector as well as health impacts in terms of mortality and morbidity. However, current decision making is only considering pieces of the puzzle separately and in many respects does not meet criteria for a full and robust economic evaluation.

The existing recommendations for developing guidelines for the allocation of scarce medical resources cited above (Emanuel et al., 2020) are silent on larger considerations around how the economy should be treated. They are limited to addressing questions on how scarce health care resources should be distributed (e.g. who gets personal protective equipment, medical supplies) within a limited health care context. They do not consider how to make larger, and inevitably difficult, decisions regarding how we might trade off choices between quantity of life (including both COVID-19 deaths and mortality in other areas such as an increase in suicide risk or even cardiovascular disease due to people being scared to go to hospital) and quality of life (which should include mental health impacts). What would the trade-offs between all aspects of health look like under different policy responses and associated economic scenarios (e.g. a complete lockdown vs a more gradual response)? Questions around which policies are likely to lessen both the mortality AND the quality of life impacts in an acceptable manner need to be asked and answered. This is especially pertinent to Australia which witnessed a flattening of the infection curve, but is also facing the prospect of continued lockdowns for an unknown length of time under conditions of isolation and economic contraction (particularly in Victoria). Decisions on when and how some of these restrictions might be lifted should ideally consider not only the impacts on likely infection rates but continued impacts on other sources of mortality (e.g. suicide), mental health, well-being and associated quality of life. At this stage, we do not know how and to what extent these trade-offs are being considered in our policy responses.

There has also been some discourse in the health economic literature on the views of members of the general public (whose taxes contribute to funding health care) on distributional considerations when making resource allocation decisions in health care. Despite some heterogeneity, there appears to be a consensus on giving greater priority to the young over the old and the more severely ill over the less severely ill; and giving lower priority to people with self-induced illness or higher socio-economic status (Gu et al., 2015). However, it remains unknown as to what extent such societal preferences hold during a pandemic – particularly with respect to illness severity and even age preference. Such preferences could be included within a more robust decision-making framework. The field of health economics has indeed considered ways in which to incorporate such preferences by either weighting benefits accruing to some people (e.g. age-weighting) or separately describing such impacts. Within the current context, where time is of the essence, separately describing who bears the costs and consequences of alternative courses of action will assist with open decision making.

Furthermore, in almost all contexts of formal health technology assessment for decision making, both impacts on length and quality of life are made. This is true for both physical and mental health conditions. Evaluating the joint impacts of both mortality and quality of life is possible through the use of generic outcome measures such as quality-adjusted life years (QALYs) and disability-adjusted life years (DALYs). QALYs are particularly favoured within health technology assessment contexts as the length of life lived in a particular health state is multiplied by a weight denoting the quality of that health state. These weights are anchored on a scale between 0 and 1, where 0 denotes death and 1 denotes perfect health. So preventing a 50-year-old woman from dying saves 35 life years (based on average life expectancy) but if her average quality of life was 0.6 over her remaining life this would amount to 21 QALYs (undiscounted). It strikes us that under the current decision-making context, quality of life impacts are largely being ignored when evaluating the health impacts of alternative courses of action. Policy responses are limited to providing mental health services to ‘mop up’ such impacts without the scale and size of the impacts being formally counted and considered in a more complete economic evaluation. We have shown that poor mental health inevitably leads to reductions in the quality of life weights used in the calculation of QALYs (Mihalopoulos et al., 2018). Therefore, as a first step, this type of research could be used to model the QALY impacts – beyond just potential infection rates and associated health impacts – of the COVID-19 spread.

These estimates could be built into the economic models evaluating different courses of action in terms of the societal cost impacts (both health sector and broader societal cost – such as reduced income) and the health impacts (both mortality and morbidity) by using outcomes such as QALYs. Jonathon Karnon recently developed a simple hypothetical decision tree analysis showing how such modelling might be undertaken when considering immediate lockdown versus ‘gradual steppers’ (Karnon, 2020). The mental health impacts of different policy decisions, which are having such a dramatic effect on our economy, livelihoods and social consequences (e.g. loneliness, impacts on schooling, etc.), can be considered within the same context as years of life lost due to varying levels of infection (without ignoring other sources of mortality such as suicide). Furthermore, the impacts of future economic downturns can also be formally considered in such analyses. Indeed, the tools of economic evaluation should be used to explicitly model different scenarios and use information as it becomes available. Like all modelling exercises, all relevant data are unlikely to be immediately available and there is inherent uncertainty. Just as the evidence base on the epidemiology of COVID-19 is slowly building, so too is the evidence base on the impacts of the lockdown measures in terms of mental health and quality of life. Unfortunately, the long-term economic consequences – in terms of business impacts, unemployment and indebtedness – of the immediate lockdowns versus a more gradual approach are less known; however, different plausible scenarios can be estimated. Therefore, a dynamic modelling approach that updates model parameters as they become available (e.g. actual unemployment rates, impacts on children etc.) as well as a Bayesian approach – which is based on probabilities rather than absolutes – is required. Such explicit and accountable decision making, which is open to discussion and debate, is certainly preferable and more likely to lead to optimal decisions being made to improve the overall welfare of Australian society.