Advancing the application of systems thinking in health: why cure crowds out prevention

Abstract

Introduction: This paper presents a system dynamics computer simulation model to illustrate unintended consequences of apparently rational allocations to curative and preventive services.

Methods: A modeled population is subject to only two diseases. Disease A is a curable disease that can be shortened by curative care. Disease B is an instantly fatal but preventable disease. Curative care workers are financed by public spending and private fees to cure disease A. Non-personal, preventive services are delivered by public health workers supported solely by public spending to prevent disease B. Each type of worker tries to tilt the balance of government spending towards their interests. Their influence on the government is proportional to their accumulated revenue.

Results: The model demonstrates effects on lost disability-adjusted life years and costs over the course of several epidemics of each disease. Policy interventions are tested including: i) an outside donor rationally donates extra money to each type of disease precisely in proportion to the size of epidemics of each disease; ii) lobbying is eliminated; iii) fees for personal health services are eliminated; iv) the government continually rebalances the funding for prevention by ring-fencing it to protect it from lobbying.The model exhibits a "spend more get less" equilibrium in which higher revenue by the curative sector is used to influence government allocations away from prevention towards cure. Spending more on curing disease A leads paradoxically to a higher overall disease burden of unprevented cases of disease B. This paradoxical behavior of the model can be stopped by eliminating lobbying, eliminating fees for curative services, and ring-fencing public health funding.

Conclusions: We have created an artificial system as a laboratory to gain insights about the trade-offs between curative and preventive health allocations, and the effect of indicative policy interventions. The underlying dynamics of this artificial system resemble features of modern health systems where a self-perpetuating industry has grown up around disease-specific curative programs like HIV/AIDS or malaria. The model shows how the growth of curative care services can crowd both fiscal and policy space for the practice of population level prevention work, requiring dramatic interventions to overcome these trends.

Figure 1

Subsystem 1: the population model. A susceptible population is at risk for either dying from disease B or transitioning temporarily into disease A. Abbreviations: A Acumu DB, Accumulated DALYS from A; B Acumu DB, Accumulated DALYs from B; Total Accumu DB, Total accumulated DALYs.

Figure 2

Subsystem 2: doctors (left) and PCWs (right) accumulate “power” in the form of earnings. A public fund (center bottom) is allocated to public payments to doctors (arrow pointing left) and public payments to PCWs (arrow pointing right). There are also two non-governmental organizations (NGOs) named NGO-D, which donates to doctors, and NGO-P, which donates to PCWs. Consultation fees (bottom left) in proportion to the number of patients with disease A also supplement doctor’s earnings.

Figure 3

Subsystem 3: lobbying process – doctor power can be depleted by spending either on providing patient services or on lobbying for more resources. The proportion of power spent on lobbying is a constant fraction in each run of the model. The more doctors spend on patient services the shorter the duration of disease A for patients who contract that disease.

Figure 4

Subsystem 3: lobbying process – PCW power is spent on lobbying and on shortening the incidence of death from disease B. The proportion of power spent on lobbying is a constant fraction in each run of the model.

Figure 5

System dynamic model showing the factors affecting resource allocation. This figure displays how all of the previously introduced sub-systems relate.

Figure 6

Baseline scenario: results of NGO donation policies at various fixed levels of NGO donation per epidemic case of disease A from 0 to 30. Points are labeled with D_A:D_B, which represent, respectively, the $ per additional DALY of disease A and of disease B. At all additional payments to the prevention workers, paying more reduces the burden of disease.

Figure 7

Baseline scenario: evidence of population harm from expanding payments to doctors. The parts of the cost effectiveness curves sloping up and to the right indicate options that cost progressively more and increase the burden of disease. These undesirable options occur beyond a threshold of $20 per DALY of disease A. Points are labeled with D_A:D_B, which represent, respectively, the $ per additional DALY of disease A and of disease B. D_B is held constant within each iso-policy curve.