1e. Individual (Clinical) vs. Population Interventions

Clinical Interventions

Javier Larrea/age fotostock/Getty Images

Clinical interventions are familiar to us all. They receive the vast majority of health related public resources. Restorative clinical interventions include surgery, medications to overcome infections, and physical therapy. Clinical preventive practices include medications to control blood pressure or cholesterol and counselling for diet, physical inactivity, and smoking. Clinical interventions always have two things in common:

1. they are aimed at or performed on individuals, and
2. they are directed at those who have been deemed to be sick or at significant risk of illness. 

In summary, the clinical approach tries to use dichotomous categories as the basis for deciding whether people should receive an intervention. 

Limitations of Clinical Interventions

People Who Will Benefit from Treatment May Not Be Targeted to Receive It

The continuous nature of most risk factors and conditions makes the clinical approach difficult. Indeed, the use of cut-points, whether selected on an individual basis, or through the use of algorithms and multiple risk factor analyses, makes it difficult to properly classify people. In epidemiological terms, these classification systems have relatively low sensitivity and specificity. As a result, large portions of people who might benefit from treatment do not get it, and large numbers of people who won’t benefit are treated. 

Resources Are Not Distributed for Maximum Effect

Another problem with a clinical approach to treatment is that the focus is on selecting an intervention that increases the odds of individual treatment success. There is little concern about the investment required to obtain this success.

For example, clinical guidelines for smoking cessation routinely recommend brief counselling plus pharmacotherapy such as nicotine replacement products. The recommendation is justified on the grounds that helping people to quit smoking is cost effective relative to treatment for tobacco-related disease such lung cancer or COPD. If, for example, brief counselling is effective 13% of the time, and pharmacotherapy plus counselling is effective 18% of the time, then it’s better to recommend the latter. However, while pharmacotherapy and counselling is 40% more effective than counselling alone, it may be twice as expensive. Therefore, the investment required to produce each successful quit will actually be higher for drugs plus counselling.

The larger point is that clinical approaches are not very successful in distributing resources across populations to produce a maximum population effect.

“Upstream” Social Determinants of Health Are Not Addressed

Another limitation of traditional clinical treatment approaches are that they are largely limited to addressing intrapersonal factors. Clinical interventions simply aren’t designed to address more “upstream” social determinants of health such as poverty, income inequality, or the inequitable distribution of health services, healthy food choices, and so forth.

As we have learned, attempts to change behaviour are not just a function of the individual, but also of their environment. A person is unlikely to reduce their intake of fat if they do not have access to healthy foods. A person is less likely to change if they are reminded about the satisfying flavor of unhealthy foods on a minute-by-minute basis through the media. 

Population Interventions

funky-date/E+/Getty Images

An alternative to individually focused clinical interventions is to adopt the population health approach described earlier.

Population intervention is distinctive from clinical intervention by virtue of the focus on a broader set of determinants of health, by concern with the health of the overall population rather than the health status of individuals, and by being aimed at populations rather than groups¹.

Geoffrey Rose

Reproduced with permission of the Library and Archives Service, London School of Hygiene and Tropical Medicine. Retrieved from  http://www.epi.umn.edu/cvdepi/photograph/rose-geoffrey-md-3/

Much of what we now know about the utility of a population health approach to disease prevention is based on groundbreaking work of the British epidemiologist Geoffrey Rose².

The population strategy of prevention starts with the recognition that the occurence of common diseases and exposures reflects the behaviour and circumstances of society as a whole. 

Prevention Paradox

The central finding of Rose’s work, which spanned over three decades, is that for many chronic diseases such as coronary heart disease or diabetes, a large number of people at small risk give rise to more cases of the disease than a small number of people who are at high risk. It therefore follows that you can have a larger impact in reducing preventable morbidity and mortality through interventions targeting entire populations — what we know as the population health approach. The tendency for many more people at low-to-moderate risk to acquire preventable diseases than the small number of people at higher risk is known as the prevention paradox.

Much of Rose’s work focused on the distribution of population health outcomes. These can be plotted on a curve just like the grades for all the students in this course.

Watch the following video that illustrates the results from one of Rose’s studies. The image shows a shift in the broader distribution of population-level factors contributing to increased body mass index (BMI) and cardiovascular disease. 

In summary, in Westernized populations, the entire bell curve of cardiovascular disease risk-factor levels shifted due to lifestyle factors, so even “normal levels” within the population confer cardiovascular disease risk. Thus, the many people at small-to-medium risk give rise to more cases of disease than the small number who are at high risk — this is the prevention paradox. 

The prevention paradox is apparent in more than just risk factors for cardiovascular disease.

Watch the following video that illustrates a similar pattern for depression reported in a 1985 study by Brenner³.

Although Rose emphasized the importance of population-level interventions, he did not argue in favour of neglecting high risk groups. Instead, he recommended a two-pronged approach to disease prevention.

Watch the following video that illustrates the shift in distribution between the population approach and the high-risk approach to disease prevention.

Magnitude of Population-Level vs High-Risk Interventions: An Illustration

Simple mathematics illustrates the power of Rose’s theory.

Say you had to choose between investing in a media-based smoking cessation campaign for an entire region of 500 000 people or subsidized nicotine replacement therapy for 5000 smokers living in that region. 

• If the region-wide campaign is successful in getting just 1% of the population to stop smoking, then you’ve potentially prevented 5000 deaths (1% of 500 000) from tobacco-related illness.
• If the nicotine replacement therapy has a 30% success rate (much higher than the campaign), you’ve still only prevented 1500 deaths (30% of 5000) from tobacco-related illness. 

This may seem like common sense today, but back in the early 1990s, it was a very innovative and radical concept.

Sadly, Geoffrey Rose did not live to enjoy his growing prominence: he passed away of liver cancer just one year after the publication of his landmark work The Strategy of Preventive Medicine.

I’m going to end this module with a question I’d like you to think about. It’s addressed in some of your required readings and will show up in a forthcoming discussion forum.

One of the key concerns arising from Rose’s theory is that population-wide strategies may inadvertently heighten social inequities in health by disproportionately benefiting lower-risk individuals, many of whom may have higher levels of income or education than higher-risk individuals. As a result, population health strategies may increase the dispersion or skew of the population distribution for a particular illness. Do you think this is the case, and, if so, what can be done to deal with this potential limitation?