March 2013, The Lancet; The Global Burden of Disease Study (GBD) 2010 made a major contribution to population health by igniting a debate about how best to reduce human suffering and premature death. The timing was impeccable. As governments, health agencies, and civil society go into conclave to elect a successor framework to the Millennium Development Goals (MDGs), the GBD put health high on the policy agenda. We live in a world of horrendous health inequalities despite a vast array of effective interventions. In 2010, healthy life expectancy in men was 30 years in Haiti but 70 years in Japan.1 The corresponding figures for women were 37 years and 72 years, respectively. Never before has our obligation to reduce suffering been accompanied by such potential to do so.
Beyond disease burden: towards solution-oriented population health
The challenge for policy makers is to maximise population health with the resources at their disposal, taking into account equity and social values. From this perspective, taxonomies of solutions are more useful than are taxonomies of disease burden. The important problems are the ones that we can do something about, those for which we have effective interventions. Because of budget limits, a decision to invest in a particular set of interventions means that we are implicitly deciding not to invest in others. By prioritising cost-effective interventions we make the most of available resources. However, budgets are not the only constraint. Because ecosystem disruption is a major threat to survival, ecological limits must also be considered.2, 3 Maximisation of health gain requires a focus on cost and carbon effectiveness.4
The information we need for allocation decisions is less about disease burden and more about the costs, benefits, and environmental effect of potential solutions. Priority setting should be informed by the marginal improvement in health and the marginal resource intensity of interventions.5 Methods to allocate resources within and between health programmes to optimise population health within resource constraints are available.6, 7 They allow explicit inclusion of equity concerns, although to be vague about equity is easier than to be explicit.
Although the GBD authors acknowledge that priority setting should be guided by the costs and benefits of intervention, this point is largely lost given that their focus is on disease burden. “The GBD challenges us to be rigorous and clear in our arguments about the criteria that should guide programming and investment decisions” writes the President of the World Bank.8 “Accurate assessment of the global, regional, and country health situation and trends is critical for evidence-based decision making for public health” writes the WHO Director-General.9 These views can be challenged. Accurate data for disease burden are not essential to set priorities and it is not necessary to accurately count how many people have died prematurely to prevent premature death. A preoccupation with disease burden could move resources from highly cost-effective solutions to less cost-effective ones, reducing health gain. The best way to improve population health is to think less about the problem, disease burden, and more about the solution—cost-effective and carbon-effective interventions.
A solution-oriented approach should also be taken for research investment. In deciding what research to fund, funding bodies should consider the costs of the research in relation to the expected value of the information that it will provide.10 Although the expected benefits of information increase with the size of the population whose choice of intervention can be informed by additional evidence, traditional measures such as incidence and prevalence (in the case of chronic diseases) are needed rather than summary estimates of disease burden. Often there is only a tenuous link between research questions and the decision problems faced by policy makers aiming to maximise health. Starting with a comprehensive decision model and examining the extent to which reduction of uncertainty in model parameters would affect decision making would be a more appropriate framework for research funding than would a simple relation with disease burden.
Investment in health without monitoring the return on the investment and without holding the recipients of health funding accountable would be foolish. Although repeated assessment of burden would allow comparisons to be made between populations and over time, because mortality and morbidity are multicausal, any changes in burden are difficult to attribute to actions taken by the health sector. A given health system might achieve the best possible population health given its budget, but disease burden could increase because of changes in other causes of disease (eg, changing food supply or climatic conditions). Similarly, a system might provide substandard care while disease burden falls. Even in high-income countries the correlation between quality of care and mortality is low.11
A solution-oriented approach to monitoring and accountability has many advantages. For example, early administration of tranexamic acid to bleeding trauma patients reduces the risk of bleeding to death by about a third.12, 13 The treatment is highly cost effective in all countries irrespective of income level.14 Because the causal link between tranexamic acid and mortality is established, we can monitor whether trauma patients receive the drug. Such monitoring is easier than monitoring trauma case-fatality, for which case-mix differences necessitate scrupulous risk adjustment. Even if trauma mortality is low and patients are not being treated, we can infer that it would lower still if they were.
Health services have more control over process than outcome and so are better able to do something about process. Process measures stimulate action from all health services with potential for improvement, not only those with poor outcomes, avoiding stigma.15 For example, some UK hospitals fail to give tranexamic acid, whereas some hospitals in India do. On this dimension, some Indian hospitals provide better care than do UK hospitals. Of note, there is no direct link between the potential to benefit from tranexamic acid and any particular disease burden category: some traffic injury victims can benefit, as can some victims of violence and some patients with falls. Similar arguments apply to policy interventions. We know that smoking is common and causes poor health and early death, so we implement non-smoking policies and smoking cessation interventions. Although monitoring the ongoing smoking burden is an (indirect) method by which to estimate the effectiveness of policy and other interventions, to monitor the degree of implementation of processes that have been shown to improve cessation rates is more useful.
Although disease seems an obvious enemy and disease burden a worthy adversary, the notion of disease is also a slippery one. According to Rose “there is no disease that you either have or don’t have—except perhaps sudden death and rabies. All other diseases you either have a little or a lot of.”16 In non-communicable diseases, medical care decision making is moving away from diagnosis towards prognosis and treatment. Because most of us have some cardiovascular disease, the role of the health worker is to predict the individual’s risk of future adverse health events and to offer treatment to those for whom the treatment benefits exceed the harms, taking into account costs and patient preferences.17
The traditional notion of disease has been useful because it allowed doctors to dichotomise the population into those at high and low risk of adverse health outcomes. Having diagnosed disease, they can focus on the high-risk group, the group that they call patients, and forget about the others. Making diagnoses enables them to allocate resources more easily. Diagnosis of disease typically involves placing people into binary categories on the basis of a somewhat arbitrary threshold on a continuous sale.18 For example, people with a fasting blood glucose greater than 6·9 mmol/L are called diabetics and when they die it might be said that diabetes is the cause of death. However, this arbitrary dichotomy does not reflect biology and ignores other predictors of poor health outcomes such as smoking, blood pressure, and cholesterol levels.18 Indeed, many people without diabetes with only moderately raised blood glucose concentrations but high levels of other vascular risk factors will be at greater risk than will many patients diagnosed with diabetes.
Disease would be a useful construct if risk prediction was univariate but it seldom is, with most ill-health related to complex long-term disorders. The increasing use of prognostic models that take several factors into account limits the use of disease labels. Disease burden cannot be used to monitor trends and set priorities (even if this approach were appropriate) when diseases change position in disease burden rankings because of changing labelling conventions. Some current diseases (hypertension, type 2 diabetes, obesity) might largely disappear as they come to be accepted as continuous risk (causal) factors for vascular and metabolic disorders. It would be a pity if public health professionals adopted a disease focus just as medical care moves away from it.
The idea of disease as a link in a causal multifactorial chain does not only apply to non-communicable disease. For example, a man dies after a road traffic injury. Although the injury itself is clearly part of the chain of causation, it might also include road design factors, poor safety enforcement, excessive speed, driver fatigue, depression, or drunkenness, and the absence of effective trauma care. Moreover, upstream of these proximal causes are ecological, social, and economic factors such as fossil fuel energy policies, land use planning, and oil prices. To say that death was caused by road traffic injury is to select one link from a long chain.
Much premature death and suffering can be prevented by tackling its causes. Removal of upstream (distal) causes is often more cost effective than is removal of proximal medical causes, because upstream causes bring about a plethora of downstream sufferings. Several years ago one of us was invited to lead a series in The Lancet on road traffic injury. The offer was declined on the basis that a narrow focus on road traffic injury neglected other important health effects of transportation fossil fuel energy use. Instead, the London School of Hygiene and Tropical Medicine assembled a team of researchers to estimate the health effects of climate change mitigation strategies.19 The team showed that limiting car use and increasing walking and cycling would reduce heart disease, stroke, breast cancer, dementia, and depression, and reduce the carbon emissions that threaten the integrity of the ecosystems on which life depends.20 These benefits would bring large cost savings to health services.21 Restriction of livestock production to reduce methane emissions would reduce the amount of cardiotoxic saturated fat flowing into the food system.22 Tackling of upstream causes is also likely to be more sustainable than is tackling of proximal causes, and because we have no choice other than to address the threats to the viability of our ecosystems, alignment of health and sustainability objectives makes tackling of upstream causes more cost effective since we only have to consider the additional cost of the health interventions. Nevertheless, some medicines are highly cost effective, although the chemical industry that many depend on is highly carbon intensive.23 A solution-based approach to better health would prioritise the most cost-effective and least carbon-intensive interventions, whether upstream or downstream.
A solution orientation means reduction of premature death and suffering through a concerted focus on removal or modification of causes, in the most efficient and sustainable way. It means monitoring the upstream ecological, economic, and social determinants of health, setting targets, and holding governments to account for reductions in hazards. It also means monitoring and management of the implementation of cost-effective and carbon-effective downstream interventions and holding health services to account for their implementation. Action will be required in all countries. The focus of population health decision making would be on finding the most appropriate package of policies and health services in view of the constraints of environment, cost, and equity. Research should be oriented towards provision of accurate data for the cost and carbon effectiveness of sustainable interventions and on optimum allocation to maximise population health. The best way to tackle problems is through a resolute focus on solutions.