Sharing the burden of healthcare: providing care to our sickest patients

One of the major challenges to affordable, universal health insurance is the high cost of providing care to the sickest patients. According to Roy Vaughn, senior vice president at BlueCross BlueShield of Tennessee, “just 5 percent of the company’s marketplace customers had accounted for nearly 75 percent of its claims costs.” What is the cost of healthcare for the typical person in the United States?Distribution of per capita US health expenditures 2012

Data from 2012, the last year for which a full analysis is available, presents a complex and confusing picture. The graph above shows per capita expenditures by percentile starting with the highest per capita expenditure. 10% face expenditures of at least $10,250. The median per capita expenditure was $854. The mean average per capita expenditure was $4309 – five times the median – and “the top 1 percent ranked by their healthcare expenses accounted for 22.7 percent of total healthcare expenditures with an annual mean expenditure of $97,956″. In brief, there is no typical person: since the bottom 50% accounted for 2.7% of total expenditures, the average per capita expenditure of the top 1% was 420 times that of the bottom 50%. There really is no typical person in terms of healthcare expenditures.

Pareto/ power law distribution of healthcare costs

This extreme distribution of healthcare costs (approximately an “80/20”, Pareto/ power law distribution) poses a major challenge to providing universal healthcare through traditional insurance models based upon risk pooling. Prior to the Affordable Care Act (ACA), the US health insurance industry addressed these challenges with risk selection – adjusting premiums or denying insurance to patients with high predicted risks, such as those with pre-existing conditions, and imposing caps on annual and/or lifetime benefits, much like the way the auto insurance industry sets premiums and limits benefits to address extreme differences in projected driver risks. Come back tomorrow for another blog post with more technical details about the Pareto distribution and healthcare costs.

Risk selection is illegal but prevalent

The ACA makes both caps on benefits and risk selection based upon pre-existing conditions illegal. In particular, US insurance carriers are required to provide coverage to all, at rates independent of pre-existing conditions, a requirement which President-Elect Donald Trump would like to keep.

However, the extreme distribution of healthcare costs means that “Targeting the highest spenders represents the greatest opportunity to have a significant impact on overall spending”; an opportunity for insurance carriers as well as for public policy. Moreover, there are good predictors for high spending: age and end of life, chronic conditions, and high spending in a previous year. For example 44.8% of the top decile in 2008 healthcare expenditures “retained this top decile ranking with respect to their 2009 healthcare expenditures”; a fact cited in an extensive Forbes report. Swiss and Dutch experience found risk selection prevalent and persistent. However, with every adult paying the same premium – within a given fund for the same type of contract – but expected healthcare expenditure (HCE) varying widely, strong incentives for risk selection are created in the absence of an adequate risk adjustment scheme. Although risk selection is illegal, it is prevalent. Swiss conglomerates of insurance carriers have been reported to achieve risk selection by assigning applicants to “specific carriers based on their risk profiles.”

Removing the economic incentives for risk selection

There is one clear way to avoid built-in economic incentives for risk selection (incentives which seem to drive insurance company behavior); that is, a single payer system, universally or as excess coverage for significant, predictable expenses. The United States now has several parallel single payer systems, namely Medicare for the elderly, Medicaid for the very poor and CHIP for children; thus, in effect, a public/private partnership in healthcare. These pre-existing single-payer systems might serve as models for a more inclusive US single payer system. Alternatively, the United States might act as an insurer of last resort, providing umbrella insurance covering individual expenses above some relatively high limit, or for costly but treatable conditions using the End Stage Renal Disease (ESRD) Program, passed in 1972 as a model. This approach would also remove extreme costs from the health insurance risk pool, as both Medicare and the ESRD Program do now, by providing near-universal coverage for our sickest patients outside the private insurance system (elderly US citizens and those with severe chronic kidney disease, respectively).

Tomorrow I will return to the Pareto-like distribution of healthcare expenditures and its consequences for any competitive insurance program. But for now, a few conclusions. Medicare and the ESRD program provide models for a smooth transition from health insurance pre-ACA with its caps and limitations to a more universal system. Medicare can be expanded to a broader public alternative. Universal coverage for additional treatable but high-risk conditions can be modeled on the ESRD program. These steps should provide the basis for further evolution of the present public/private partnership into a more universal, more cost-effective system.

In my opinion, the extreme distribution of healthcare expenditures and the ability to perform risk selection, even though illegal, present a strong, essentially irrefutable argument for a single payer system; either overall, or for chronic conditions and expenditures predictable through risk selection. In the US, Medicare and the ESRD program provide illustrative, successful and useful models.


Chris Sampson’s journal round-up for 3rd October 2016

Every Monday our authors provide a round-up of some of the most recently published peer reviewed articles from the field. We don’t cover everything, or even what’s most important – just a few papers that have interested the author. Visit our Resources page for links to more journals or follow the HealthEconBot. If you’d like to write one of our weekly journal round-ups, get in touch.

Using discrete choice experiments with duration to model EQ-5D-5L health state preferences: testing experimental design strategies. Medical Decision Making [PubMedPublished 28th September 2016

DCEs are a bit in vogue for the purpose of health state valuation, so it was natural that EuroQol turned to it for valuation of the EQ-5D-5L. But previous valuation studies have highlighted challenges  associated with this approach, some of which this paper now investigates. Central to the use of DCE in this way is the inclusion of a duration attribute to facilitate anchoring from 1 to dead. This study looks at the effect of increasing the options when it comes to duration, as previous studies were limited in this regard. In this study, possible durations were 6 months or 1, 2, 4, 7 or 10 years. 802 online survey respondents we presented with 10 DCE choice sets, and the resulting model had generally logically ordered coefficients. So the approach looks feasible, but it isn’t clear whether or not there are any real advantages to including more durations. Another issue is that the efficiency of the DCE design might be improved by introducing prior information from previous studies to inform the selection of health profiles – that is, by introducing non-zero prior values. With 800 respondents, this design resulted in more disordering with – for example – a positive coefficient on level 2 for the pain/discomfort dimension. This was not the expected result. However, the design included a far greater proportion of more difficult choices, which the authors suggest may have resulted in inconsistencies. An alternative way of increasing efficiency might be to use a 2-stage approach, whereby health profiles are selected and then durations are selected based on information from previous studies. Using the same number of pairs but a sample half the size (400), the 2-stage design seemed to work a treat. It’s a promising design that will no doubt see further research in this context.

Is the distribution of care quality provided under pay-for-performance equitable? Evidence from the Advancing Quality programme in England. International Journal for Equity in Health [PubMedPublished 23rd September 2016

Suppose a regional health care quality improvement initiative worked, but only for the well-off. Would we still support it? Maybe not, so it’s important to uncover for whom the policy is working. QOF is the most-studied pay-for-performance programme in England and it does not seem to have reduced health inequalities in the context of primary care. There is less evidence regarding P4P in hospital care, which is where this study comes in by looking at the Advancing Quality initiative across five different health conditions. Using individual-level data for 73,002 people, the authors model the probability of receiving a quality indicator according to income deprivation in their local area. There were 23 indicators altogether, across which the results were not consistent. Poorer patients were more likely to receive pre-surgical interventions for hip and knee replacements and for coronary artery bypass grafting (CABG). And poorer people were less likely to receive advice at discharge. On the other hand, for hip and knee replacement and CABG, richer people were more likely to receive diagnostic tests. The main finding is that there is no obvious systematic pro-poor or pro-rich bias in the effects of this pay-for-performance initiative in secondary care. This may not be a big surprise due to the limited amount of self-selection and self-direction for patients in secondary care, compared with primary care.

The impact of social security income on cognitive function at older ages. American Journal of Health Economics [RePEc] Published 19th September 2016

Income correlates with health, as we know. But it’s useful to be more specific – as this article is – in order to inform policy. So does more social security income improve cognitive function at older ages? The short answer is yes. And that wasn’t a foregone conclusion as there is some evidence that higher income leads to earlier retirement, which in turn can be detrimental to cognitive function. In this study the authors use changes in the Social Security Act in the US in the 1970s. Between 1972 and 1977, Congress messed up a bit and temporarily introduced a policy that made payments increase at a rate faster than inflation, which was therefore enjoyed by people born between 1910 and 1916, with a 5 year gradual transition until 1922. Unsurprisingly, this study follows many others that have made the most of this policy quirk. Data are taken from a longitudinal survey of older people, which includes a set of scores relating to cognition, with a sample of 4139 people. Using an OLS model, the authors estimate the association between Social Security income and cognition. Cognition is measured using a previously developed composite score with 3 levels: ‘normal’, ‘cognitively impaired’ and ‘demented’. To handle the endogeneity of income, an instrumental variable is constructed on the basis of year of birth to tie-in with the peak in benefit from the policy (n=673). In today’s money the beneficiary cohort received around $2000 extra. It’s also good to see the analysis extended to a quantile regression to see whereabouts in the cognition score distribution effects accrue. The additional income resulted in improvements in working memory, knowledge, languages and orientation and overall cognition. The effects are strong and clinically meaningful. A $1000 (in 1993 prices) increase in annual income lead to a 1.9 percentage point reduction in the likelihood of being classified as cognitively impaired. The effect is strongest for those with higher levels of cognition. The key take-home message here is that even in older populations, policy changes can be beneficial to health. It’s never too late.


Does political reform really reduce child mortality?

Measuring causal effects is a tricky business. But, it’s necessary if we want to appropriately design effective policies and interventions. Many things are not amenable to manipulation in an experiment and so we rely upon a toolbox of statistical tools to try to identify the effect of interest. These methods are often ingenious, finding sophisticated ways of exploiting different types of variation, but they are essentially uninterpretable without an underlying causal theory. To illustrate this, let’s consider a paper that was featured a few weeks ago in the journal round-up: Effect of democratic reforms on child mortality: a synthetic control analysis.

A large number of countries have undergone democratic reform over the last 20 years. This article aimed to estimate how that reform has impacted upon child mortality. To do this a synthetic control method was used.

Synthetic control

The synthetic control method was formalised by Alberto Abadie, Alexis Diamond, and Jens Hainmueller in an article in the Journal of the American Statistical Association. It’s particularly useful in the situation where there is one area or cluster or country that has undergone a change, and multiple potential countries or clusters to act as controls that did not undergo the change. The eponymous synthetic control is a weighted average of the potential control sites where the weights have been chosen to best replicate the pre-intervention trend in the intervention site. The example given by Abadie and colleagues is estimation of the impact of tobacco control reform in California on tobacco consumption. The other US states are all potential controls. Bayesian synthetic control methods have also been established (by a team at Google), which we will make use of later.

The synthetic control method therefore seems appropriate to analyse the impact of democratic reform in a given country. Measurement of democratic reform was based on a change in the Polity2 index that rates the degree of autocracy/democracy in countries; a switch in the index from negative to positive (the index runs from -10 to 10) was considered ‘democratic reform’. Of the 60 countries identified as having reformed, 33 met the inclusion criteria, and 24 counterfactuals were able to be constructed. The primary outcome is the relative reduction in child mortality after ten years; the results from the 24 countries are shown in the histogram below (Figure 1). It would seem that, on average, democratic reform seems to reduce child mortality.


Figure 1. Histogram of results from the 24 included countries from Pieters et al.


Perhaps one of the factors that have limited research in the area of political economy and health is the complexity of the relationships between the various macro, micro, economic, and political effects. For example, on the basis of the evidence presented above, we would still not be able to say whether, for a given country, introducing democratic reform would have any impact on child mortality. Let’s consider a couple of examples to explore why.


Figure 2. Results from synthetic control analyses of the impact of democratic change on child mortality. Data from UN Inter-agency Group for Child Mortality Estimation.


Mozambique was engaged in a civil war between 1977 and 1992 as the communist Frelimo battled the anti-communist Renamo for control of the country. At the cessation of hostilities in 1993, wide sweeping reforms were enacted by Joaquim Chissano, and an election was held. We can consider 1993 as the year of democratic reform and conduct our own synthetic control analysis (using the aforementioned Bayesian approach). The results are shown in the figure above (Figure 2). Clearly, there is a significant reduction, but is this due to democratic reform or simply the end of civil war? A counterfactual approach is used as the theory of causation behind much statistical inference. Had an autocratic regime followed the civil war would there have still been declines in child mortality? I would conjecture that there would have been. Democratic reform in this case is either not a cause or a redundant cause.


Kenneth Kuanda was removed as the president of Zambia in 1991. Following this, reforms for multiparty democracy were enacted. Figure 2 above reports the estimated impact upon child mortality. A decline is clearly evident, however, this decline does not start until 2003, when the copper price tripled (Figure 3 below; copper constitutes 88% of Zambia’s exports) and GDP per capita almost doubled. Again, whether democratic reform can be inferred as a cause in this instance is questionable, especially when reduced to questions of counterfactuals.


Figure 3. Copper price per metric ton (US$). Data from IMF Cross Country Macroeconomic Statistics Database.

South Africa

The apartheid regime was ousted in 1994 following election of the ANC in South Africa. This regime change opened up the political institutions to the majority of South Africans who had previously been excluded. However, as Figure 2 above shows, this reform appeared to have little impact on child mortality. Indeed, South African healthcare still faces significant challenges, and large structural inequalities in access to quality healthcare persist even today. Taken together, this illustrates that democratic reform is not a sufficient condition for improvement of population health.


Finally, consider Zimbabwe, which became more autocratic in 1986 following a deal between Zanu and Zapu. If democratic reform improves child mortality then it seems reasonable that autocratic changes would worsen child mortality. Figure 2 above reports the synthetic control results for Zimbabwe. No change is evident.

What can we conclude?

Democratic reform is neither a sufficient nor necessary condition for improvements in child mortality. We cannot understand the evidence without an underlying theory. The study discussed here is a good data analysis; decent analyses in this area should be encouraged. But, the theory should come before the data.