Towards a Theoretical Understanding of Diabetes Management
Jon Ettinger
Faculty Advisor: Professor Walter Nicholson
Submitted to the Department of Economics at AmherstCollege in partial fulfillment of the requirements for the degree of Bachelor of Arts with Distinction
April 27, 2007
Acknowledgements
First off, I would like to thank my advisor, Professor Walter Nicholson for his support and guidance throughout the thesis process. In addition to his help on this project, Professor Nicholson’s Law and Economics class really changed the way I think. Thanks to Professor Rivkin for his instruction and help both this semester and the last, and to Jeanne Reinle and her uplifting smile for making the third floor of Converse a great place to work. Thank you also to all the great friends here at Amherst that have been there along the way. Thanks to Lucy Sheehan for her timely edits and encouragement, and of course, huge thanks to all the 2007 Converse Hall-Stars for creating such a great community amongst the Econ majors this spring. Finally, I owe a huge debt of gratitude to my parents. Their comments on this paper were invaluable, and their support of my ongoing skirmish with diabetes has been unflappable. Although this may be tautological[I], without my mother and father, none of this would have been possible.
Table of Contents
I. Introduction……………………………………………………………… 3
II. Literature Review………………………………………………………. 9
IIa. Compliance and Empowerment Models IIb. Cost-effectiveness of Diabetes Management
IIc. Behavioral Economics and Modeling Diabetes Management
III. Theoretical Model of Diabetes Management…………….…………… 24
IIIa. The Components of DMRL
IIIb.Discussion of Price Levels
IIIc.The HbA1c Production Function
IIId.Lifetime Utility as a Function of HbA1c Level
IV. Externalities and Informational Problems…………………………… 40
V. Conclusion……………………………………………………………….. 44
Appendix. Depression and Diabetes………………………….……………. 46
References……………………………………………………….…………... 48
I. Introduction
In 2002, the American Diabetes Association (ADA) estimated the economic cost of diabetes to be $132 billion, adding the disclaimer that this figure likely underestimates the true burden of the disease. Direct medical expenditures constituted $91.8 billion of the total cost, up from an estimated $44 billion in 1997. Although the 2002 ADA study represents the most recent comprehensive examination of the economic costs of diabetes in the US, one can safely assume that this figure has increased significantly. Since 2002, the diagnosed population of diabetics has increased by 20.7%, from 12.1 to 14.6 million. An estimated 6.2 million diabetics remain undiagnosed.
Given these staggering statistics on the prevalence, growth trends, and economic burden of diabetes, it is essential that policy makers, healthcare professionals, and diabetic patients themselves find ways to reduce the impact of diabetes. Many studies have examined the cost-effectiveness[1] of various treatments and regimens in an effort to reduce the total cost of diabetes. While there is great variation in the efficacy from one program to another, it seems clear that there are diabetes intervention measures that are very cost-effective, yet continue to be under-utilized by the diabetic population that they are intended to help. To the consternation of healthcare professionals, many diabetics fail to comply with treatment regimens that are intended to minimize the progression of the disease and the occurrence of long-term complications associated with poor diabetes management[2]. Healthcare economists contend that the implementation of better diabetes management would lower the economic burden of diabetes.
The notion that improved diabetes management, and the associated increase in diabetes management expenditure, might lower the total cost of diabetes requires an understanding of how the economic costs of diabetes are incurred. Diabetes is a general term used to describe two distinct but similar diseases: type 1 diabetes, also known as juvenile diabetes, and type 2 diabetes, or adult-onset diabetes[3]. Type 1 diabetes is an auto-immune disease in which the immune system attacks and destroys the cells in the pancreas that produce insulin. Type 2 diabetes, which accounts for 90-95% of all diabetes cases, is a metabolic disorder characterized by insulin resistance, or the body’s inability to use insulin efficiently. Both diseases eliminate the body’s ability to effectively regulate glycemic levels in the blood, resulting in hyperglycemia or an excessively high blood-sugar level. Chronic hyperglycemia causes gradual cellular damage throughout the body, while acute hyperglycemia leads to diabetic ketoacidosis, a potentially life threatening condition that causes further damage to bodily organs.
The Diabetic Control and Complications Trial (DCCT), conducted from 1983 to 1993, gave the medical community conclusive evidence of what they had long believed: it is hyperglycemia caused by diabetes, rather than diabetes in and of itself, that leads to the development of costly long-term complications and the associated increase in morbidity and mortality amongst diabetics. Fortunately, through oral medications and intensive blood-glucose management (IBGM)[4], diabetics can manage their glycemic levels and avoid hyperglycemia. Early treatment of developing complications can reduce their impact, and close attention to diet and exercise helps blood-sugar levels stay within a normal range. Regular appointments with healthcare professionals may help diabetics formulate a plan for controlling their diabetes. In essence, through vigilant diabetes management, diabetics can regulate their blood-sugar levels and mitigate the development of costly long-term complications.
Thus, the severity and cost of diabetes related complications is a function of the amount of resources devoted to diabetes management, and the effective use of those resources. Formulating diabetes treatment strategies begins with this fundamental relationship. Cost-effectiveness analyses of various diabetes management techniques exploit this principle, comparing the cost to implement a specific intervention with the costs saved in diabetes complications. Cost-effectiveness modeling has become quiteaccurate and is a crucial tool for informing medical professionals and diabetics about how best to allocate their diabetes management resources.
The most effective measure of diabetes control is the HbA1c level. HbA1c refers to glycosylated hemoglobin, a measurement for the level of hemoglobin exposure to high plasma levels of glucose. HbA1c levels approximate the average glycemic level over a roughly three month period – higher average blood-glucose raises HbA1c – and can therefore be used to measure the quality of overall diabetes care. Normal HbA1c levels in non-diabetics range from 4.0%-5.9%, whereas the HbA1c average for diabetics is roughly 8.0% with outliers well over 10%. Because HbA1c levels reflect the degree of chronic hyperglycemia in an individual, they are a very strong predictor of the extent of long-term diabetes complications.
Knowing how to best distribute diabetes management resources is only half of the picture in preventing costly diabetes complications. The other half is selecting the level of resources to devote to diabetes management. While efficient strategies of resource allocation arelargely determined by the medical community, the level of diabetes management maintained by an individual diabetic is a personal consumption choice. Despite the significant implications associated with choosing a level of diabetes management consumption, no theoretical work has been done to approach this issue. Conventional wisdom merely asserts that diabetics do not devote enough resources to diabetes management, causing the total cost of diabetes to be higher than it could otherwise be. However, from the perspective of rational choice theory, diabetics should devote a personal utility-maximizing level of resources to their disease management, contradicting the notion of underconsumption given no externalities and complete information.
Measuring the amount of resources a diabetic commits to diabetes management is made difficult by the fact that there are several types of resources that diabetics devote to their disease management. I identify three broad categories of resources in order to simplify this problem; blood-glucose management (BGM) resources, professional healthcare (PHC) resources, and changes in lifestyle that diabetics adopt to control their diabetes, which I call habits. Each of these categories can be further decomposed into their various components. From this point on, I refer to the total quantity of resources as the diabetes management resource level, or DMRL. The concept of resources as it is used here can be equated with “costs” and includes both financial and non-financial costs to the individual of diabetes management. Changing one’s habits to manage glycemic levels, or dealing with the pain of an insulin injection are both examples of costs to the individual diabetic. The value of these suffered costs is understood to be a component of the DMRL an individual chooses. These non-financial costs are implicitly converted to dollar amounts using willingness to pay (WTP) schemes so that the cost of DMRL is given in consistent units.
There are several reasons to believe that rational diabetic agents choose utility-maximizing DMRLs that do not simultaneously maximize social welfare. Diabetes management creates certain externalities; the costs of diabetes are not endured exclusively by diabetics. Costs are also borne by employers, family members and friends, and health-insurance plans (which are generally structured in such a way that creates moral hazard problems). Self-control problems brought on by inconsistent time preferences as well as systematic sources of imperfect information may also contribute to inefficiently low DMRLs.
The main question I investigate in this paper is how diabetics choose the level of resources they devote to diabetes management from a rational choice, utility-maximizing perspective. Secondarily, I look at possible explanations for why this level of resources may not be optimal for social welfare maximization. The basis for my analysis is a theoretical model I create to understand the consumption decision diabetics face when they choose a level of diabetes management.
Section 2 provides a literature review and background to cost-effectiveness research on diabetes management as well as behavioral economics concerns in creating a diabetes management model. Section 3 presents a theoretical model of diabetes management from an individual utility maximizing perspective. In Section 4, I discuss the potential externalities of diabetes management, the systematic gaps in information that hinder rational choice, and the policy implications they create. I conclude the paper with a summary of my diabetes management theory and its potential importance in understanding the true total economic burden of diabetes.
II. Literature Review
Most of the literature on diabetes management focuses on a compliance-based model of diabetes care and on the effectiveness of specific medical practices in treating diabetes. Little work has been done that accurately portrays the agency of informed diabetics in determining their own utility maximizing strategies for treating diabetes. This section will characterize the literature on patient agency and existing models on the cost-effectiveness of diabetes management. It includes a review of the literature on expected utilityhealthcare models, as well as the issues that are relevant to diabetes in particular.
IIa. Compliance and Empowerment Models
Many medical studies investigate the issue of patient compliance with assigned diabetic management regimens. They examine the risk factors associated with non-compliance and give suggestions to medical professionals about how to encourage compliance in their patients.
Recent empirical evidence suggests that the compliance model of diabetes care leads to poor diabetes management. The patient compliance model minimizes the individual agency that diabetics have in their own self-management of diabetes. It places the responsibility of creating a diabetes management regimen primarily on the physician. However, maintaining steady, normal glycemic levels is a more complicated and dynamic process than the compliance model assumes[5]. Not allowing diabetics the ability to allocate their diabetes management resources or adjust their DMRL to suit their own personal preferences causes a perceived lack of control that further undermines compliance (Dunn, 2005).
Problems with the compliance model have led to a new model for diabetes management, referred to as the “Empowerment Model” (Meetoo and Gopaul, 2005). This new philosophy of diabetes treatment involves collaboration between physician and patient. In this model, the healthcare professional provides the patient with information about the consequences of poor diabetes management and how he or she can most effectively manage the disease. A central concern of the Empowerment Model is diabetes education. One of the main objectives of professional healthcare is to train the patient to utilize effective diabetes management techniques and to teach the patient about the long-term cost of complications resulting from poor glucose control. Proper instruction of these two concepts lays the foundation from which diabetic patients can make the rational decisions regarding their diabetes management that are essential to patient empowerment. Equipped with a better understanding of diabetes and how best to manage blood-sugar levels, the patient can then customize a management regimen that maximizes his or her lifetime utility.
The new, patient-based approach to diabetes management necessitates a better understanding of how informed patients choose their DMRL. Existing compliance literature focuses on the healthcare professional, but the agency in diabetes management has shifted to the consumer. Perhaps owing to the relatively recent shift in treatment philosophy, no clear framework has been made to analyze the patient’s decision from a lifetime utility maximizing perspective.
IIb. Cost-effectiveness of Diabetes Management
The ability to measure the cost-effectiveness of various diabetes management practices is severely limited by the difficulty of isolating the effects of a specific treatment and the long time-frame over which diabetes-related costs are incurred[6]. Nevertheless, many studies have conducted cost-effective analyses (CEA) of various diabetes interventions, the most relevant of which being intensive blood-glucose management (IBGM). In a 2000 article,Klonoff and Schwartz conducted a meta-analysis of CEA studies spanning 17 different interventions for diabetes. Because of the difficulty in evaluating the cost-effectiveness of these interventions, most of their results were inconclusive. However, they found that improved glycemic controlwas a clearly cost-effective intervention. The cost-effectiveness of glycemic control through IBGM has been corroborated by many other studies (See Rubin et al., 1998; Steffens, 2000; Sidorov et al., 2002). These studies measured improvements in glycemic control owing to IBGM by looking at HbA1ctest resultsand determining the average cost of diabetes complications associated with variousHbA1clevels. Blood-glucose control is by far the most important aspect of diabetes management, and it is also the aspect over which a diabetic has the most control. For this reason, I use the terms diabetes management and DMRLs to refer to an individual’s attempts to regulate blood-sugar and avoid hypo and hyperglycemia.
A 2006 paper by Beaulieu et al. gives the most recent and comprehensive CEA of diabetes disease management. They examine the incentives for health plans to offer comprehensive diabetes management programs according to new chronic care guidelines that follow through with the Empowerment Model of disease management[7]. As health plans bear most of the financial cost of diabetes care spending for individuals they insure, net savings caused by greater diabetes disease management will be passed on to the healthcareplan. Adverse selection problems and patient turnover[8]may potentially negate the cost savings to the health plan, and indeed Beaulieu estimates that HealthPartners Minnesota roughly breaks even on their diabetes disease managementprogram over a ten-year period[9]. Meanwhile, they calculate the total societal benefit of the ten-year program to be $64,000 per diabetic patient.
Beaulieu arrives at this figure by estimating the value of the three primary benefits from improved diabetes management: improved quality of life, long term cost savings from avoided complications, and workplace productivity gains. To valuate improved quality of life, Beaulieu assumed the HealthPartners program brought HbA1c down from 10% to 7.2%, and furthermore assumed that this yielded an increase of 0.87 quality adjusted life years (QALYs). Beaulieu valued a QALY at $100,000, and after accounting for discount rates, calculated the net present value of the improved quality of life to be $59,000. Healthcare savings from avoided complications were estimated to be another $5,000, and workplace productivity gains were left out of the estimation due to lack of data. One can easily argue with several of Beaulieu’s assumptions, especially the relatively high valuation of a QALY. Still, his conclusion that disease management can lead to significant societal benefits is quite robust.
The large benefits of the diabetes management program studied by Beaulieu evidence the poor disease management Habits of many diabetics. Even if one adjusts Beaulieu’s assumptions and accounts for additional costs (such as time costs) that the study leaves out, it is hard to avoid the conclusion that some underconsumption of diabetes management is occurring. Assuming that patients rationally choose the level of resources they commit to diabetes, underconsumption may be indicative of large market failures for diabetes care and in medical care market at large. Patients have imperfect information about the benefits of diabetes management and how to efficiently allocate their DMRL. As Beaulieu points out, health plans have insufficient incentives to provide programs that resolve the patient informational problems. Because the market for healthcare does not function like competitive markets with certainty (Arrow, 1963), additional frictions may impede the ability of patients to consume their desired quantity of professional care or switch to a healthcare provider that has sufficient diabetes information to help patients allocate their DMRLs efficiently[10].
Although the model I develop is based on the agency of a patient to optimize his DMRL under the assumption that healthcare professionals are a known quantity as they fit into a patient’s overall allocation of diabetes management, there are problems with this assertion as it plays out in actuality. Healthcare professionals are paid by schemes that assume the non-contractibility of health outcomes (Dranove and White, 1987). Physician effort is thus called into question, as physicians may have incentives to take on more patients and give each a lesser degree of care. In the case of diabetes, where effective management requires frequent collaboration between patient and doctor, this can be doubly problematic. As a simplifying assumption to assess how patients choose their DMRLs, I ignore the incentive problems inherent to healthcare professionals delivering medical care. Instead, I treat professional care in the same way that I treat commodity goods that are part of a diabetes management budget; I assume that there is no extra uncertainty in the quality of professional care than there is in the integrity of insulin.