Controlling tuberculosis: Evidence from the first community-wide health experiment

Karen Clay, Peter Juul Egedesø, Casper Worm Hansen, Peter Jensen, Avery Calkins 13 August 2019

a

A

According to the World Health Organization (2018), tuberculosis (TB) is one of the top 10 causes of death worldwide and the leading cause from a single infectious agent. In 2017, there were 1.3 million deaths caused by TB and 10 million new TB incidences. The disease has been treatable by antibiotics since the 1940s, but drug-resistant strains now exist. It is therefore hardly surprising that eliminating TB is part of the UN’s third sustainable development goal. 

While TB is mainly a challenge for developing countries today, historically, the US and Europe also experienced many TB deaths. Yet, striking reductions were observed both in the US and Europe before any effective medical treatments against TB were developed. For example, the TB mortality rate fell from above 200 per 100,000 in the beginning of the 20th century to about 60 per 100,000 in the mid-1930s in the US (Cutler and Meara 2004). 

What caused this historical decline in TB and other infectious diseases prior to the advent of effective medical treatments? McKeown (1976) argued that the main driver was nutrition, with public health interventions playing a limited role. Others, including Preston (1975) and Cutler and Miller (2005), have viewed the role of public health interventions more favourably. Moreover, recent research by Anderson et al. (2018) is partly in line with McKeown and finds that the first campaign against TB had limited success prior to 1918 in the US. By contrast, work by Hollingsworth (2014) and Egedesø et al. (2017) show that some interventions targeted at TB were successful in the pre-antibiotic era.

Our recent research (Clay et al. 2019) contributes to this debate by providing a rigorous quantitative analysis of the effects of the first public health demonstration – the Framingham Community Health and Tuberculosis Demonstration – on TB mortality, total mortality, and infant mortality. The demonstration was carried out from 1917 and 1923 and was regarded as a success,  both at the time and today (e.g. Kannel and Levy 2005). The official evaluation found that the TB mortality rate in Framingham fell 69%, compared with that of seven pre-selected control municipalities in which the fall was 32%. However, using a broader set of comparison municipalities and modern econometric methods, our research suggests that the Framingham Demonstration had little effect on TB mortality.

The Framingham Demonstration was made possible by a donation of $200,000 from the Metropolitan Life Insurance Company to the National Association for the Study and Prevention of Tuberculosis in 1916. Later that year, Framingham, Massachusetts, was chosen as a typical American community. The Framingham Demonstration focused on efforts to control TB through a consultation service, but also included infant health work. The consultation service helped local physicians to diagnose TB and served as a link between physicians and patients. 

The Framingham Demonstration was carried out at a time when there was no medical cure available. Instead, the best treatment was believed to be fresh air, sunlight, good food, and rest. These could be provided at home or at sanatoriums, which were also used to isolate patients. Moreover, ‘good mouth hygiene’ and respecting anti-spitting ordinances were emphasised to prevent the disease from spreading. 

Evaluating the Framingham Demonstration

To evaluate the effects of the Framingham Demonstration, we use newly digitised vital statistics for municipalities in Massachusetts. The official evaluation used Chicopee, Clinton, Fitchburg, Gardner, Marlborough, Milford, and North Adams as control municipalities to measure whether the demonstration reduced TB mortality in Framingham during the demonstration period (1917-1923). 

We extend the number of potential control municipalities within Massachusetts to study whether the demonstration reduced TB mortality and consider all municipalities with populations with at least 5,000 inhabitants and at most 50,000 inhabitants. In addition, our newly collected data allow us to study whether the demonstration had any long-lasting effects by extending the study period into the 1930s.   

Figure 1 TB mortality per 1,000 in Framingham, the official control municipalities, and synthetic control results

The top panel of Figure 1 compares the TB mortality rate in Framingham to that of the average of the official control municipalities. Although TB mortality fell in Framingham during the demonstration period, it also fell in the control municipalities. The mortality rates are similar at the end of the 1920s. We also notice that the rate for Framingham exhibits fluctuations, which suggests that TB mortality is subject to so-called transitory shocks.

To evaluate the effect of the demonstration more rigorously, we apply econometric methods that take into account that the experiment only had one treated unit and transitory shocks in the mortality rates. Specifically, we apply the synthetic control method, pioneered by Abadie and Gardeazabal (2003), a recent extension of this method by Doudchenko and Imbens (2016), and standard differences-in-differences methods. 

The use of synthetic control methods allows us to relax the assumption that TB mortality in Framingham and the average of other Massachusetts municipalities were on parallel trends needed for identification in differences-in-differences estimation. A synthetic control is constructed by putting appropriate weights on the other Massachusetts municipalities to closely match the pre-intervention evolution of the outcome studied. The extension by Doudchenko and Imbens (2016) allows for better matching of the fluctuations in the treated unit’s mortality rates. 

We apply these three econometric methods to measure the effects of the demonstration on the TB mortality rate, the infant mortality rate, the total mortality rate, and the TB case rate. In the bottom panel of Figure 1, we show the evolution of the TB mortality rate in Framingham compared to the synthetic control obtained when using the extension of Doudchenko and Imbens (2016) and note that the graphs follow each other closely in the period prior to the first demonstration year (i.e. the years before 1917). Yet, we find that none of the differences in TB mortality rates after the demonstration was initiated (i.e. the years after 1917) are statistically significant, implying that we cannot reject that the demonstration had no effects on TB mortality. 

The same insights are made for the other mortality outcomes and when using other econometric methods. However, we do find some evidence that the demonstration increased the TB case rate, indicating that the demonstration was successful in finding new cases of TB, but this did not materialise into fewer people dying of TB. 

Conclusion

The implication of the findings described above is that the Framingham Demonstration was not as successful as believed by the National Association and often repeated in the historical narrative. Moreover, our findings tend to support McKeown’s (1976) contention that public health policy was not a decisive factor in the reduction of TB mortality. Yet, it is important to stress that our study does not invalidate the conclusions in Hollingsworth (2014) and Egedesø et al. (2017) as these authors investigate other interventions. Their findings suggest that there may be valuable policy lessons for current developing countries combating drug resistant TB today, but on our evidence, the same cannot be said about the Framingham Demonstration. 

References

Abadie, A, and J Gardeazabal (2003), “The economic costs of conflict: A case study of the Basque Country”, American Economic Review 93(1): 113–132.

Anderson, D M, K K Charles, C L H Olivares and D I Rees (2018), “Was the first public health campaign successful? The tuberculosis movement and its effect on mortality”, American Economic Journal: Applied Economics 11(2): 143–75.

Clay, K, P J Egedesø, C W Hansen, P S Jensen, and A Calkins (2019), “Controlling tuberculosis? Evidence from first community-wide health experiments”, NBER working paper 25884.

Cutler, D, and E Meara (2004), “Changes in the age distribution of mortality over the twentieth century”, in D A Wise (ed.), Perspectives on the economics of aging, Chicago: University of Chicago Press, pp. 333–365.

Cutler, D M, and G Miller (2005), “The role of public health improvements in health advances: The twentieth-century United States”, Demography 42(1): 1–22.

Doudchenko, N, and G Imbens (2016), “Balancing, regression, difference-in-differences and synthetic control methods: A synthesis”, NBER working paper 22791.

Egedesø, P J, C W Hansen and P S Jensen (2017), “Preventing the White Death: Tuberculosis dispensaries”, Discussion paper 17-19, Department of Economics, University of Copenhagen.

Hollingsworth, A (2014), “Controlling TB in a world without antibiotics: Isolation and education in North Carolina, 1932-1940”, working paper.

Kannel, W B, and D Levy (2005), “Commentary: Medical aspects of the Framingham Community health and tuberculosis demonstration”, International Journal of Epidemiology 34: 1187–1188.

McKeown, T (1976), The modern rise of population, Edward Arnold Ltd, Great Britain.

Preston, S H (1975), “The changing relation between mortality and level of economic development”, Population Studies 29: 231–248.

World Health Organization (2018), Global tuberculosis report 2018.

a

A

Topics:  Economic history Health economics

Tags:  public health, public health interventions, TB, infectious disease, US, historical review, health

Professor of Economics and Public Policy, H. John Heinz III School of Public Policy and Management, Carnegie Mellon University

Consultant, Incentive

Associate Professor of Economics, University of Copenhagen

Professor,Department of Business and Economics, University of Southern Denmark

Doctoral candidate, Department of Economics, University of Michigan

Events

CEPR Policy Research