The new frontier in health services research: a behavioural paradigm guided by genetics

Health care insurers are increasingly relying on patient activation (defined as the skills and confidence that equip patients to become actively engaged in their own health care) to help them and their patients improve outcomes and minimize costs. Low activation is associated with unhealthy behaviour, poorer outcomes ¹ and higher costs. ² As a result, there is increasing interest in why individuals display different levels of activation particularly when experiencing multiple, long-term conditions.

It has recently been proposed that certain personality attributes may explain a person’s willingness, persistence or ability to seek and maintain continuity of care for one or more long-term conditions (i.e. their level of activation). ³ For example, impulsivity is a facet of personality and a significant predictor of disease onset and severity, ⁴ mostly as a result of its behavioural correlates, including alcohol abuse, smoking and over-eating. ⁵ Interestingly, individuals with impulsive predispositions are generally less receptive to advice to adopt behaviours targeted at disease prevention ⁶ and may even be prone to loss aversive behaviors. ⁷ Loss aversion describes an inclination to take greater risks to avoid incurring a loss than to acquire a gain (originally described as an endowment effect). ⁸ Thus, when years of impulsive, unhealthy habits culminate in disease states, the desire to delay the loss associated with death may lead unhealthy people to seek, often too late, a range of different forms of health care from different providers. Extensive documentation of diminished continuity of care in late life generally supports this hypothesis, in the United States at least. ⁹ However, this short-sighted response associated with impulsivity may produce short-term benefits to individuals’ longevity, ¹⁰ although at a greater societal cost.

Despite the known robust effects of patient activation, little is known about what distally influences activation or how interventions can be designed to raise it. To further our understanding of these unknowns, it is necessary to create a new paradigm for health services research that not only explains phenomena such as impulsive (and quite costly) behaviour but also harnesses newly available genomics data, which are becoming more widely accessible to clinical and social scientists.

A genetic analysis that references, for example, a pathway of catecholamine synthesis (i.e., dopamine and serotonin) that strongly correlates with behaviour and cognition, and, therefore, patient activation is needed for three reasons. First, such a pathway is frequently implicated in psychiatric conditions, ¹¹ which are associated with multimorbidity ¹² and consequently health care use. ¹³ Second, describing a genetic basis of behaviour advances previous models of health care use ¹⁴ as well as offering a patient-centred focus that positions genetics as more than merely determining treatment selection. Finally, variation in this pathway could conceivably account for some of the population heterogeneity that generates documented differences in outcomes between populations insured within multi-tiered systems. ¹⁵ Differences in outcomes within mixed systems (i.e., publicly and privately insured populations) could arise because commercial insurers, unlike their public counterparts, continually seek to protect their business interests by avoiding individuals most likely to need health care and/or use it ineffectively because of risky behaviours or noncompliance.

Such effects are not likely to be confined to multi-payer, mixed systems. Even in single-payer systems (i.e. European models), similar patient selection effects may appear as a consequence of distinctions in provider organization (i.e. extent of gatekeeping) affecting patient preferences and behaviours in ways that might not be so easily discernable. ¹⁶ For example, an analysis by this author of data from The Commonwealth Fund’s international profiles of health care systems from 2012 to 2015 ^{1 7} using generalized linear mixed models demonstrates that mandating registration with a general practitioner and gatekeeping each had concomitant protective effects on smoking by reducing its prevalence, but negative effects for obesity (data available upon request), suggesting a substitution effect at play. This finding is supported in a recent large UK sample ¹⁸ and highlights past findings regarding primary care physicians’ inability to measurably improve patients’ unhealthy lifestyle choices. ¹⁹ Perhaps, this is because we do not understand preference formation or modification based simply on interactions with health care providers.

There exist genetic variants that routinely predict these types of behaviours. Family studies have demonstrated the heritability of these risky behaviours: for example, the heritability of the maximum number of drinks consumed in a 24-h period ²⁰ was approximately 0.5; the heritability of nicotine dependence ²¹ was estimated to be about 0.75; the heritability of obesity ²² was greater than 0.8. These high heritability coefficients describing the percent of variation accounted for by genetics provide strong impetus to identify genetic variants associated with these impulsive behaviours, which may also correlate highly with health care use.

Genetic variation is operationalized as allele frequencies of various single nucleotide polymorphisms (SNPs) in genes of interest. The human genome has roughly 10 million SNPs, some of which may alter gene functioning. The allele frequency describes the relative frequency of a SNP at a particular base pair in a population, expressed as a fraction or percentage. Allele frequencies assessed at >5% indicate common population variants, which may be useful in describing varying levels of patient activation across tiered payer systems or within provider organizations of single-payer systems.

As just one example of many possible genetic variants, DOPA decarboxylase is a gene that encodes the enzyme aromatic L-amino acid decarboxylase, which catalyzes several decarboxylation reactions, including the production of dopamine and serotonin. The gene and the SNPs within it have been associated with nicotine ²³ and alcohol ²⁴ dependence as well as increased body mass index and a lower preference for perceived healthy food-related decision making. ²⁵ These behaviours are considered consequences of a hypo-dopaminergic state. Recently, psychologists have termed this a “reward deficiency syndrome” to explain the onset and persistence of such addictive traits. ²⁶ So far, however, pharmacologically or nutritionally modulating this insufficient dopaminergic transmission has not been shown to alter this loss aversion, ²⁷ the attribute of close connection to potentially ineffective health care utilization (in the absence of any gatekeeping role) described above.

These behaviours have well-established associations to the impulsivity and risk-seeking proclivities that commercial insurance providers seek to exclude from their covered population or that gatekeepers use to determine referral likelihood. Furthermore, dopamine and serotonin regulate feelings of overconfidence and reduced cognitive capacity, ²⁸ which behavioural economists have implicated in the formation of heuristics by individuals when making complicated decisions related to health insurance plans and related coverage.^29,30

In conclusion, health services researchers need to understand better two issues: (1) the profound dynamic that relates a reward deficiency phenotype with unhealthy behaviours that collectively promulgate declining population health; and (2) the correlation of that phenotype with an activation status governed by loss aversion leading to more ineffective health care utilization to remediate the consequences of prior unhealthy behaviours. Health economists like to assess loss aversion using experiments, but health care utilization is not and can never be an experiment open for analysis. Therefore, the best avenue for examining such a new paradigm is the reliance on genetic data to begin to discern how genes associated with unhealthy behaviours and loss aversion affect health care utilization. This is the hypothesis that could redefine health services research. Its inclusion will allow us to relate findings on the associations of specific SNPs or allele frequencies with excessive or ineffective resource use to the omnipresent research on the genetics of behaviours inducing chronic disease. Doing so will not only provide an objective rationale for the provision of scarce public health dollars to improve population health but also inform the provision and structure of coverage and care among insurers and providers (within either various domains of a multi-tiered system or specific segments of a single-payer system) that will nudge individuals away from addictive behaviours and towards more effective and efficient health care utilization.