- Email: [email protected]
The role of individual time preferences in health behaviors among hypertensive adults: a pilot study
Journal of the American Society of Hypertension 3(1) (2009) 35– 41
Research Article
The role of individual time preferences in health behaviors among hypertensive adults: a pilot study R. Neal Axon, MD,* W. David Bradford, PhD, and Brent M. Egan, MD The Medical University of South Carolina, Charleston, South Carolina, USA Manuscript received April 24, 2008 and accepted August 5, 2008
Abstract An economic framework incorporating patients’ time-value preferences may help explain individual variation in preventive health behaviors. We conducted a pilot study to examine the relationship between health discount rates and preventive health practices. A group of 422 hypertensive individuals were assessed by written survey regarding their actual or likely preventive health behaviors, and they were posed a series of time preference questions. Regression methods that account for the interval nature of the time preference responses were used to estimate individual respondents’ discount rates. Dichotomous regression analyses (using probit models) adjusted for gender, age, race, income, and health status revealed mean health discount rates of 0.438 or (43.8%) per year (standard deviation [SD], 0.07). Analyses adjusted for age, gender, race, income level, insurance status, and health status indicated that a 1% increase in discount rate increased the likelihood respondents would not check their BP by 3.5% (P ⫽ .003), not alter diet and exercise habits by 0.6% (P ⫽ .004), and not follow doctors’ treatment plans by 1.6% (P ⫽ .05). Compared to the four lowest quintiles, patients in the highest quintile of discount rates (annualized discount rates between 50% and 57.2%) tended to have lower likelihood of ever checking blood pressure (BP) at home (42.5% vs. 47.6%; P ⫽ .36), of not using their physician’s office for sick care (16.5% vs. 27.6%; P ⫽ .01), and of not altering their diet and exercise habits in response to a diagnosis of hypertension (6.8% vs. 12.4%; P ⫽ .07). These preliminary data indicate that the degree to which individuals discount the future has a significant impact on their health behaviors. J Am Soc Hypertens 2009;3(1): 35– 41. © 2009 American Society of Hypertension. All rights reserved. Keywords: Health discounting; hypertension; rates; health status.
Introduction An estimated 65 million Americans have hypertension, representing almost 29% of the adult population, and recent data reveal that the prevalence of hypertension has not changed from 1999 to 2006.1– 4 Despite improvement in recognition and treatment rates for hypertension in recent years, control rates in the population still lag behind goals set by experts.5,6
This study was supported by the Department of Health and Human Services entitled Community-Focused Initiative to Reduce Burden of Stroke (SBEMP040001-02-0, Drs. Egan and Bradford) and the Agency for Healthcare Research and Quality (AHRQ) “Understanding and Eliminating Health Disparities in Blacks” (P01HS1087-01, Dr. Egan). Conflicts of interest: none. *Corresponding author: R. Neal Axon, MD, Assistant Professor of Medicine and Pediatrics, The Medical University of South Carolina, 135 Rutledge Avenue, MSC 591, Charleston, South Carolina 29425. Tel: 843-792-2900; fax: 843-792-6355. E-mail: [email protected]
Barriers to effective hypertensive care include patient, provider, and system-specific factors.7 Provider barriers, including time constraints, practice patterns, fear of adverse drug effects, and complexity of managing comorbid conditions as well as therapeutic inertia have received significant attention in recent years.7–10 System-specific barriers to care including geographic location, transportation, physician supply, insurance coverage, and costs of care have also been well described.11–15 Patient-specific barriers to care have been conceived as including predisposing, enabling, and reinforcing inputs.15 Such predisposing factors include knowledge of hypertension, attitudes regarding care, and self-reported health status.16 One attitudinal factor that has received relatively little attention to date is the degree to which individual patients value the future relative to the present. Economic theory states that individual time preferences are a fundamental personal characteristic and that people discount future value to varying degrees. In general, we expect that people with higher rates of discounting future value will shift consumption of economic goods to the
1933-1711/09/$ – see front matter © 2009 American Society of Hypertension. All rights reserved. doi:10.1016/j.jash.2008.08.005
36
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
present relative to people who place greater value/emphasis on the future. Grossman17 proposed a model in which health is a component of human capital, which depreciates over time and in which investments can be made. Subsequent economic research has investigated many dynamic aspects of health production and health care demand.18 –22 Recent research has refined the methods for measuring individual discount rates.23,24 To date, no comprehensive model has sufficiently explained the high degree of variability observed in individual preventive health behaviors and medical adherence suggesting that as-yet unmeasured factors contribute to this phenomenon. Recognizing that preventive health behaviors in hypertensive patients are often suboptimal, an economic framework that features time-value preferences can be useful for understanding attitudinal predisposing factors. We present a pilot study to examine the relationship between the rate at which hypertensive patients discount future value and their actual or likely health behaviors. We also make suggestions regarding how this new method of assessment might be incorporated into existing models of health promotion planning.
Methods Setting/Recruitment This study was conducted as a part of the South Carolina Excellence Centers to Eliminate Ethnic/Racial Disparities (EXCEED) Study sponsored by the Agency for Healthcare Research and Quality (AHRQ) in conjunction with the Medical University of South Carolina. This study was approved by the Institutional Review Board at the Medical University of South Carolina. Telephone invitations for group surveys were extended to hypertensive adult clinic patients aged 20 and above seen for primary care in community-based clinics affiliated with the Hypertension Initiative of South Carolina.25 In an effort to limit potential bias, group surveys were conducted at locations outside the clinic setting. All group sessions were completed between June 13, 2003 and October 7, 2004.
Table 1 Health behavior questions and responses Do you test your own blood pressure using a blood pressure cuff at home? 1 ⫽ No, never 0 ⫽ Sometimes 0 ⫽ Yes, regularly Where do you usually go for health care when you are sick? 0 ⫽ Mainly to a doctor’s office or clinic 1 ⫽ Mainly to an emergency room 1 ⫽ Sometimes to a doctor’s office and sometimes to an emergency room 1 ⫽ Other, specify: Everyday I follow the treatment program my doctor set up for me. 1 ⫽ Strongly disagree 1 ⫽ Disagree 0 ⫽ Agree 0 ⫽ Strongly agree What would you rather do? 1 ⫽ Eat, drink, and live my life the way I want and have poorer health in 5 years 0 ⫽ Give up eating, drinking, and living the way I want and have better health in 5 years Imagine you smoke a pack of cigarettes daily, what would you be likely to do? 0 ⫽ Participate in 12 weekly quit smoking classes, which cost $120 and meet near your home in the evening 0 ⫽ Participate in the quit smoking classes and take pills, which cost an extra $120 but which will ensure you really stay off cigarettes (class and pills total cost $240) 0 ⫽ Quit smoking on my own 1 ⫽ Cut back smoking on my own 1 ⫽ I probably would not quit or cut back until I feel a negative effect from smoking
related to patients time-value preferences. These were of the general form: “If you win the lottery today, which prize would you prefer? $1,000 cash paid today OR $1,500 paid in cash 5 years from today?”
Data Collection/Group Surveys Patients presenting for group sessions were given written survey forms to complete. All survey items were pretested, and the tool included selected and previously validated items from the Perceived Efficacy in Patient-Physician Interactions (PEPPI) questionnaire.26 Study personnel were available to provide assistance to individuals in interpreting and answering questions upon request. In addition to basic demographic information, survey topics included access to and preferences for primary care, quality of physician-patient interactions, knowledge of hypertension and its complications, and patient self-efficacy. Additional questions
Four follow-up survey items refined the patients’ timevalue preferences by presenting higher values for the hypothetical future payment to more accurately define individual discount rates. Finally, patients were asked a series of hypothetical questions regarding their actual or likely health behaviors. Topics included likelihood of checking BP at home, likelihood of altering diet and exercise patterns, likelihood of following a doctor’s treatment plans, likelihood of smoking cessation, and likelihood of seeking sick-day care outside of the primary care home. Relevant questions and the coding scheme for answers are shown in Table 1.
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
Statistical Analysis After survey completion, all demographic variables were entered into a secure database prior to analysis. Summary statistics were generated for demographic variables including age, gender, race, income level, and self-reported health status. From a theoretical perspective, we assumed that each respondent had an underlying discount rate that is not directly observable. However, the range in which the underlying discount rate must fall can be inferred. For example, if a person said they preferred $1,500 in 5 years over $1,000 today, then we would know that the rate at which this person discounts the future must be below the 8.5% annual interest rate implied by the $500 increase in value over 5 years; we assume that the discount rate is above 0%. Thus, while the persons rate of discounting is unobservable, it falls within the [0.0, 0.085] interval. Further, we assume that this discount rate is a function of the person’s characteristics, which can be used as explanatory variables in a regression model. Once we have estimated the regression model, we can use the parameters to predict the actual discount rate for each person. We used a generalized linear modeling technique called “interval regression,” or grouped regression, to estimate these parameters. (Note that this is a maximum likelihood model that uses upper and lower bounds on the implied discount rate to estimate the model; as such there is no “dependent variable” in the traditional sense of the word.) Respondents were then grouped into quintiles based on their estimated discount rates comparing the group with the highest rates of discounting future value to those in the lower four quintiles. The proportion of patients willing to forego individual health behaviors was then compared between groups using 2 tests. Additional dichotomous regression analyses using probit models adjusted for age, gender, race, income, health status, and insurance status were performed to model the incremental influence of a change in discount rate on the likelihood of forgoing specific health behaviors. All statistical calculations were performed using Stata version 10 (StataCorp, College Station, TX).
Results A total of 1,465 individuals were contacted via telephone to participate in group surveys; 744 (50.8%) agreed to participate with 422 attending (28.6%) and completing the full survey. Basic demographics for study participants are listed in Table 2. Overall, respondents were ⬃75% AfricanAmerican, 75% female, and 55% were younger than 60 years old. Approximately half had household incomes of less than $15,000 per year, and 70% had some form of public health insurance (ie, Medicare or Medicaid). Seventy-eight percent said their doctor answered their questions in a way they understood. Reported self-efficacy was high, but
37
Table 2 Demographic characteristics of respondents Variable
Total
%
Sample Race African-American White Hispanic Asian Native American Gender Male Female Age 20–30 31–40 41–50 51–60 61–70 71–80 81–90 School 1–8 9–12 13–16 17–20 Income ⬍$15,000 $15,000 – $30,000 $30,000 – $50,000 $50,000 – $75,000 ⬎$75,000 Health Insurance Medicare Medicaid Private US Government State Government None
n ⫽ 439 Total 303 112 11 2 2 Total 104 335 Total 13 38 72 119 123 60 9 Total 62 237 109 19 Total 217 117 51 26 11 Total 177 87 132 21 58 73
% of 430 70.47 26.05 2.56 0.47 0.47 % of 439 23.7 76.3 % of 434 3.1 8.83 16.59 27.67 28.6 13.95 2.07 % of 427 14.52 55.5 25.53 4.45 % of 422 51.42 27.73 12.08 6.16 2.61 % of 439 40.32 19.82 30.11 4.78 13.21 16.63
nonadherence to self-management and pill taking for their hypertension was admitted by one-third of patients. Over 70% reported taking some home remedy for hypertension. A total of 98% were willing to take more medications and/or make lifestyle changes if the link to better outcomes was clearly stated. The results of the interval regression model to estimate discount rates are presented in Table 3. The overall discount rate in our cohort was 0.438 or (43.8%) per year (standard deviation [SD], 0.07). Compared to the four lowest quintiles, patients in the highest quintile of discount rates tended to have lower likelihood of ever checking BP at home (42.5% vs. 47.6%; P ⫽ .36), of not using their physician’s office for sick care (16.5% vs. 27.6%; P ⫽ .01), and of not altering their diet and exercise habits in response to a diagnosis of hypertension (6.8% vs. 12.4%; P ⫽ .07). The respondents in the highest quintile discount rate group had discount rates between 50% and 57.2% per year. The proportion of patients likely to continue smoking was larger in
38
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
Table 3 Regression model estimating health discount rates Variable
Coefficient
Race (black ⫽ 1) Race (white ⫽ 1) Gender (male ⫽ 1) Age (40–69) Age (70⫹) Education (at least high school ⫽ 1) Income (⬍$15,000/y ⫽ 1) Health status (very good ⫽ 1) Year of data collection (y ⫽ 1) Constant Number of observations ⫽ 422
0.0776042 0.0626919 ⫺0.0669549 0.0176825 0.0569559 ⫺0.006622 0.0764141 ⫺0.1067717 0.0435879 0.3169918
P value .182 .301 .057 .695 .222 .856 .03 .009 .209 0
Wald chi-square ⫽ 22.26; Prob ⬎ chi-square ⫽ .0081
the higher discounting group though not statistically significant (2.5% vs. 3.2%; P ⫽ .72). There were no differences in the proportion of patients unlikely to follow physician’s care plans between groups. Subsequent dichotomous regression analyses used probit models adjusted for gender, age, race, income, health status, and insurance status. These marginal effects models suggested the incremental contribution of one additional percentage point in discount rate to the likelihood respondents would not engage in specific health behaviors. Presented in Table 4, the model predicted that increasing discount rate by one percentage point decreased the likelihood respondents would check their BP by 3.5% (P ⫽ .003). Presented in Table 5, the model predicted that increasing discount rate percentage point decreased the likelihood respondents would alter diet and exercise habits by 0.6% (P ⫽ .004). Presented in Table 6, the model predicted that increasing discount rate by one percentage point increased the likelihood respondents would not follow doctors’ treatment plans by 1.6% (P ⫽ .05). Table 4 Marginal effects model for likelihood of not checking BP at home Variable
dF/dx
P value
Discount Race (black) Race (white) Gender (male) Age (40–69) Age (70⫹) Education (at least high school) Income (⬍$15,000/y) Insurance (Medicare) Insurance (Medicaid) Insurance (private) Insurance (no insurance) Health status (very good) Number of observations ⫽ 422
0.03536 ⫺0.14619 ⫺0.16436 0.28357 ⫺0.07564 ⫺0.32705 0.04534 ⫺0.16936 0.022771 0.04733 ⫺0.0251 ⫺0.07439 0.29286
.003 .313 .191 .003 .350 .001 .430 .108 .746 .514 .696 .357 .028
BP, blood pressure.
Table 5 Marginal effects model for likelihood of not altering diet and exercise habits Variable
dF/dx
P value
Discount Race (black) Race (white) Gender (male) Age (40–69) Age (70⫹) Education (at least high school) Income (⬍$15,000/y) Insurance (Medicare) Insurance (Medicaid) Insurance (private) Insurance (no insurance) Health status (very good) Number of observations ⫽ 422
0.00651 0.38182 0.97125 0.14999 ⫺0.02319 ⫺0.04325 0.01392 ⫺0.00876 0.00262 0.01649 ⫺0.01418 ⫺0.02392 0.09467
.107 .069 .000 .085 .282 .134 .360 .758 .876 .437 .378 .088 .408
Discussion These preliminary data indicate that the degree to which individuals discount the future has a significant impact on their self-reported health behaviors. This insight is particularly important to make in a hypertensive population where control rates remain less than ideal. Significant research has been appropriately directed at addressing provider and health system-specific barriers to improvement in chronic disease care over recent years in part on the assumptions that patient-specific barriers are relatively more difficult to predict and relatively more intransigent. However, an economic framework for understanding individual time preferences offers new potential for understanding the variations observed in individual health behavior and compliance with hypertension care. It seems intuitive that discount rates should affect behavior. Support for this notion comes from Brown et al,27 who
Table 6 Marginal effects model for likelihood of not following physician treatment plan Variable
dF/dx
P value
Discount Race (black) Race (white) Gender (male) Age (40–69) Age (70⫹) Education (at least high school) Income (⬍$15,000/y) Insurance (Medicare) Insurance (Medicaid) Insurance (private) Insurance (no insurance) Health status (very good) Number of observations ⫽ 422
0.06709 ⫺0.33987 ⫺0.45551 0.38974 ⫺0.21869 ⫺0.78358 ⫺0.09759 ⫺0.64560 0.33589 0.11182 ⫺0.04824 0.45070 0.53239
.052 .401 .231 .199 .360 .021 .571 .047 .123 .600 .798 .054 .210
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
found that when compared to more “future-oriented” persons, those with a more “present-oriented” outlook perceived themselves to be less susceptible to the consequences of hypertension, believed more in home remedies, and were more skeptical of the benefits of prescribed medication. Importantly, the present study is the one of the first to apply this type of quantitative economic modeling in the analysis of health behaviors and lifestyles. Additional support for this concept comes from earlier work by our group with a large longitudinal database, the most recent wave of the Health and Retirement Survey (2004).28 In addition to core components of that survey including questions about healthrelated behaviors, basic socio-demographics, insurance characteristics, comorbidity, and patients’ recent (past two years) health screening activities, a selected group of 1,039 respondents (out of 20,129) aged 24 to 65 years old were surveyed via telephone with questions on time preferences. In this cohort, the interval regression model yielded a mean imputed annual discount rate of 0.26 (26%) per year (SD, 0.05), which is lower than our current findings. Compared to respondents with moderate-to-low imputed discount rates, respondents in the highest imputed discount rate percentile group had a statistically lower rate of mammogram use (0.669 vs. 0.802; P ⫽ .0003), Papanicolaou (PAP) smear use (0.689 vs. 0.764; P ⫽ .03), prostate exam use (0.520 vs. 0.717; P ⫽ .002), dental visits (0.488 vs. 0.722; P ⬍ .001), cholesterol testing (0.716 vs. 0.797; P ⫽ .014), flu shot usage (0.438 vs. 0.511; P ⫽ .06), rates of vigorous exercise (0.219 vs. 0.378; P ⬍ .001), and higher rates of smoking (0.294 vs. 0.221; P ⫽ .03). Taken together, these results argue for more in-depth research in this area. Overall, the rates of discount we measured were quite high, at nearly 44% per year. It is important to note that clinics for the EXCEED Study were selected from clinics in the Hypertension Initiative that were known to provide services for disproportionately more low income and minority patients. The intentional selection bias may have affected the findings. Notwithstanding, such rates of future depreciation far exceed rates that would be acceptable as interest rates for mortgage loans or as depreciation rates on vehicles. They are not excessive when compared to the annualized interest rates on many types of consumer credit such as some credit cards or “payday lending” outlets. Payday loans, which are banned in several U.S. states, are a form of predatory lending also called “cash advances” where the annualized interest rates can reach 400%.29 However, such high discount rates for health may begin to rationally explain the apparently short-sighted decisions made by hypertensive patients regarding their health behaviors. Furthermore, the marginal effect of a 10% increase in discount rate on the probability of foregoing health behaviors in our models appears to be at least moderate in size. For example, a 35% increase in the likelihood of foregoing home BP checks observed in our model was similar in magnitude to the 32% decrease in probability observed with
39
moving from the lower age range (40 to 69 years) to the higher age range (70 years and up). Similarly, the 16% increase in the likelihood of not following a doctor’s treatment plans observed in our model was similar in magnitude to the 15% decrease in probability observed with moving into the lower income group from the other income groups. This research has significant implications for clinical care and for policy making. First, if the estimated discount rates are accurate, there is little hope that some individuals will voluntarily undertake preventive activities that entail significant near-term costs in terms of time, money, or effort if the perceived benefits accrue many years in the future. In that case, even very large future benefits to the individual would be discounted heavily in present value terms, and clinical attempts to alter patients’ behavior by stressing the health benefits accruing 10, 15, or 20 years in the future would likely be unsuccessful. For policy makers, the observation that high rates of discount are associated with negative effects on health behavior implies that increased advocacy and education for prevention may not be universally effective tools in certain patient groups. It will be important to incorporate this new knowledge into existing theoretical frameworks for health promotion such as the Predisposing, Reinforcing, and Enabling Constructs in Educational Diagnosis and Evaluation–Policy, Regulatory, and Organizational Constructs in Educational and Environmental Development (PRECEDE-PROCEED) model of Green and Kreuter.30 This dominant model of health promotion during the past three decades outlines processes for the program planning and evaluation. Individual time preferences might logically be incorporated into this framework as one of the predisposing factors surveyed as educational and ecological inputs during phase 4 of the PRECEDE process. Such factors include a “person or population’s knowledge, attitudes, beliefs, values, and perceptions that facilitate or hinder motivation for change.” This concept is illustrated graphically in the Figure. Health discount rates are necessarily individual characteristics, and prior research suggests that these are relatively stable within individuals over time.23 That is, individual discount rates may be somewhat inflexible and thus, not as amenable to traditional educational efforts. However, knowledge of mean discount rates for specific populations may help inform the types of interventions best suited to populations with different levels of health discounting. For instance, programs that promote more aggressive provider-driven disease management for hypertension and other chronic diseases may be emphasized over education and self-management strategies in groups with high levels of discounting. An alternative or complementary strategy would be to identify and promote potentially favorable near-term benefits, eg, improved exercise tolerance or to link long-term benefits to a clear outcome, eg, participating in a child or grandchild’s graduation or marriage ceremony.
40
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
4.
5.
6. Figure. Conceptual framework for incorporating health discount rates into the PRECEDE-PROCEED model of health promotion planning. Adapted from Green LW, Kreuter M. Health Program Planning: An Educational and Ecologic Approach. 4th ed. New York, New York: McGraw Hill, 2005.30
7.
8. Our study has several limitations. First, the survey was limited to 422 individuals. While this limited number of patients was adequate for the number of covariates we evaluated, more complex modeling was not possible with this data set. Secondly, some elements of our survey tool were not validated prospectively. However, the questionnaires were pretested, and selected query items were drawn from a previously validated survey tool. Additionally, our measure of health behaviors was indirect, relying on patient self-report. It is possible that patients’ actual behaviors could differ from their survey responses, but this measure represents a necessary first step along this line of investigation. Future inquiry in this area will directly test this relationship by ensuring a larger sample size and controlling for multiple potential confounders. Such study will allow for further refinement of the valuation methodology in order to make it more clinically applicable. One can imagine that a parsimonious and prospectively validated survey tool for clinical and research use could be useful in promoting preventive health behaviors.
9.
10.
11.
12.
13.
14.
15.
References 1. Kannel WB, Wilson PW. Cardiovascular risk factors and hypertension. In: Izzo JL BH, editor. Hypertension Primer. 3rd ed. Philadelphia, Pennsylvania: Lippincott, Williams, and Williams, 2003:235–39. 2. Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of hypertension in the United States, 1988 –2000. JAMA 2003;290:199 –206. 3. Fields LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, Sorlie P. The burden of adult hypertension in the United
16.
17. 18.
States 1999 to 2000: a rising tide. Hypertension 2004;44:398 – 404. Ostchega Y, Yoon SS, Hughes J, Louis T. Hypertension awareness, treatment, and control-continued disparities in adults: United States, 2005–2006. NCHS Data Brief 2008;3:1– 8. Chobanian AV, Bakris GL, Black HR, Cushman WC, Izzo JL. Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure. National High Blood Pressure Education Program Coordinating Committee. Hypertension 2003;42:1206 –52. Healthy People 2010. U.S. Department of Health and Human Services, 2007. Available at: http://www. healthypeople.gov/. Accessed: April 30, 2007. Oliveria SA, Lapuerta P, McCarthy BD, L’Italien GJ, Berlowitz DR, Asch SM. Physician-related barriers to the effective management of uncontrolled hypertension. Arch Intern Med 2002;162:413–20. Phillips LS, Branch WT, Cook CB, Friedman RH, Glickman M, Kader B, et al. Clinical inertia. Ann Intern Med 2001;135:825–34. Berlowitz DR, Ash AS, Hickey EC, Doyle JP, El-Kebbi IM, Gallina DL, et al. Inadequate management of blood pressure in a hypertensive population. N Engl J Med 1998;339:1957– 63. Okonofua EC, Simpson KN, Jesri A, Rehman SU, Durkalski VL, Egan BM. Therapeutic inertia is an impediment to achieving the Healthy People 2010 blood pressure control goals. Hypertension 2006;47:345–51. Brand FN, Smith RT, Brand PA. Effect of economic barriers to medical care on patients’ noncompliance. Public Health Rep 1977;92:72– 8. Lurie N, Ward NB, Shapiro MF, Brook RH. Termination from Medi-Cal— does it affect health? N Engl J Med 1984;311:480 – 84. Shulman NB, Martinez B, Brogan D, Carr AA, Miles CG. Financial cost as an obstacle to hypertension therapy. Am J Public Health 1986;76:1105– 8. Kotchen JM, Shakoor-Abdullah B, Walker WE, Chelius TH, Hoffmann RG, Kotchen TA. Hypertension control and access to medical care in the inner city. Am J Public Health 1998;88:1696 –99. Hill MN, Bone LR, Kim MT, Miller DJ, Dennison CR, Levine DM. Barriers to hypertension care and control in young urban black men. Am J Hypertens 1999;12: 951–58. Lawrence W, Green MWK. Health Promotion Planning: An Educational and Environmental Approach. 2nd ed. Mountain View, California: Mayfield Publishing Co, 1991. Grossman M. On the concept of health capital and the demand for health. J Polit Econ 1972;80:223–55. Wagstaff A. The demand for health: some new empirical evidence. J Health Econ 1986;5:195–233.
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
19. Van Doorslaer E. Health, Knowledge, and the Demand for Medical Care: An Econometric Analysis. Maastricht, The Netherlands: van Gorcum and Comp, 1987. 20. Wagstaff A. The demand for health: an empirical reformulation of the Grossman model. Health Econ 1993; 2:189 –98. 21. Grossman M, Kaestner R. Effects of education on health. In: Behrman J, Stacey N, The Social Benefits of Education. Ann Arbor, Michigan: University of Michigan Press, 1997:69 –123. 22. Zweifel P, Breyer F. Health Economics. New York, New York: Oxford University Press, 1997. 23. Frederick S, Loewenstein G, O’Donoghue T. Time discounting and time preference: a critical review. J Econ Lit 2002;XL:351– 401. 24. Bradford W, Kleit A, Krausel-Wood M, Re R. Doublebounded dichotomous choice method to assess willingness to pay for telemedicine. J Telemed Telecare 2004; 10:325–30. 25. Lackland DT, Lin Y, Tilley BC, Egan BM. An assessment of racial differences in clinical practices for hy-
26.
27.
28.
29.
30.
41
pertension at primary care sites for medically underserved patients. J Clin Hypertens (Greenwich) 2004;6:26 –31; quiz 2–3. Maly RC, Frank JC, Marshall GN, DiMatteo MR, Reuben DB. Perceived efficacy in patient-physician interactions (PEPPI): validation of an instrument in older persons. J Am Geriatr Soc 1998;46:889 –94. Brown C, Segal R. Ethnic differences in temporal orientation and its implications for hypertension management. J Health Soc Behav 1996;37:350 – 61. Bradford W. The role of individual time preferences on patients health maintenance habits. In: Center for Health Economics and Policy (CHEPS) Working Paper. Charleston, South Carolina: The Medical University of South Carolina, 2007:26. Center for Responsible Lending: A Resource for Predatory Lending Opponents. 2008. Available at: http://www. responsiblelending.org/. Accessed: June 19, 2008. Green LW, Kreuter M. Health Promotion Planning: An Educational and Ecologic Approach. 4th ed. New York, New York: McGraw Hill, 2005.
Research Article
The role of individual time preferences in health behaviors among hypertensive adults: a pilot study R. Neal Axon, MD,* W. David Bradford, PhD, and Brent M. Egan, MD The Medical University of South Carolina, Charleston, South Carolina, USA Manuscript received April 24, 2008 and accepted August 5, 2008
Abstract An economic framework incorporating patients’ time-value preferences may help explain individual variation in preventive health behaviors. We conducted a pilot study to examine the relationship between health discount rates and preventive health practices. A group of 422 hypertensive individuals were assessed by written survey regarding their actual or likely preventive health behaviors, and they were posed a series of time preference questions. Regression methods that account for the interval nature of the time preference responses were used to estimate individual respondents’ discount rates. Dichotomous regression analyses (using probit models) adjusted for gender, age, race, income, and health status revealed mean health discount rates of 0.438 or (43.8%) per year (standard deviation [SD], 0.07). Analyses adjusted for age, gender, race, income level, insurance status, and health status indicated that a 1% increase in discount rate increased the likelihood respondents would not check their BP by 3.5% (P ⫽ .003), not alter diet and exercise habits by 0.6% (P ⫽ .004), and not follow doctors’ treatment plans by 1.6% (P ⫽ .05). Compared to the four lowest quintiles, patients in the highest quintile of discount rates (annualized discount rates between 50% and 57.2%) tended to have lower likelihood of ever checking blood pressure (BP) at home (42.5% vs. 47.6%; P ⫽ .36), of not using their physician’s office for sick care (16.5% vs. 27.6%; P ⫽ .01), and of not altering their diet and exercise habits in response to a diagnosis of hypertension (6.8% vs. 12.4%; P ⫽ .07). These preliminary data indicate that the degree to which individuals discount the future has a significant impact on their health behaviors. J Am Soc Hypertens 2009;3(1): 35– 41. © 2009 American Society of Hypertension. All rights reserved. Keywords: Health discounting; hypertension; rates; health status.
Introduction An estimated 65 million Americans have hypertension, representing almost 29% of the adult population, and recent data reveal that the prevalence of hypertension has not changed from 1999 to 2006.1– 4 Despite improvement in recognition and treatment rates for hypertension in recent years, control rates in the population still lag behind goals set by experts.5,6
This study was supported by the Department of Health and Human Services entitled Community-Focused Initiative to Reduce Burden of Stroke (SBEMP040001-02-0, Drs. Egan and Bradford) and the Agency for Healthcare Research and Quality (AHRQ) “Understanding and Eliminating Health Disparities in Blacks” (P01HS1087-01, Dr. Egan). Conflicts of interest: none. *Corresponding author: R. Neal Axon, MD, Assistant Professor of Medicine and Pediatrics, The Medical University of South Carolina, 135 Rutledge Avenue, MSC 591, Charleston, South Carolina 29425. Tel: 843-792-2900; fax: 843-792-6355. E-mail: [email protected]
Barriers to effective hypertensive care include patient, provider, and system-specific factors.7 Provider barriers, including time constraints, practice patterns, fear of adverse drug effects, and complexity of managing comorbid conditions as well as therapeutic inertia have received significant attention in recent years.7–10 System-specific barriers to care including geographic location, transportation, physician supply, insurance coverage, and costs of care have also been well described.11–15 Patient-specific barriers to care have been conceived as including predisposing, enabling, and reinforcing inputs.15 Such predisposing factors include knowledge of hypertension, attitudes regarding care, and self-reported health status.16 One attitudinal factor that has received relatively little attention to date is the degree to which individual patients value the future relative to the present. Economic theory states that individual time preferences are a fundamental personal characteristic and that people discount future value to varying degrees. In general, we expect that people with higher rates of discounting future value will shift consumption of economic goods to the
1933-1711/09/$ – see front matter © 2009 American Society of Hypertension. All rights reserved. doi:10.1016/j.jash.2008.08.005
36
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
present relative to people who place greater value/emphasis on the future. Grossman17 proposed a model in which health is a component of human capital, which depreciates over time and in which investments can be made. Subsequent economic research has investigated many dynamic aspects of health production and health care demand.18 –22 Recent research has refined the methods for measuring individual discount rates.23,24 To date, no comprehensive model has sufficiently explained the high degree of variability observed in individual preventive health behaviors and medical adherence suggesting that as-yet unmeasured factors contribute to this phenomenon. Recognizing that preventive health behaviors in hypertensive patients are often suboptimal, an economic framework that features time-value preferences can be useful for understanding attitudinal predisposing factors. We present a pilot study to examine the relationship between the rate at which hypertensive patients discount future value and their actual or likely health behaviors. We also make suggestions regarding how this new method of assessment might be incorporated into existing models of health promotion planning.
Methods Setting/Recruitment This study was conducted as a part of the South Carolina Excellence Centers to Eliminate Ethnic/Racial Disparities (EXCEED) Study sponsored by the Agency for Healthcare Research and Quality (AHRQ) in conjunction with the Medical University of South Carolina. This study was approved by the Institutional Review Board at the Medical University of South Carolina. Telephone invitations for group surveys were extended to hypertensive adult clinic patients aged 20 and above seen for primary care in community-based clinics affiliated with the Hypertension Initiative of South Carolina.25 In an effort to limit potential bias, group surveys were conducted at locations outside the clinic setting. All group sessions were completed between June 13, 2003 and October 7, 2004.
Table 1 Health behavior questions and responses Do you test your own blood pressure using a blood pressure cuff at home? 1 ⫽ No, never 0 ⫽ Sometimes 0 ⫽ Yes, regularly Where do you usually go for health care when you are sick? 0 ⫽ Mainly to a doctor’s office or clinic 1 ⫽ Mainly to an emergency room 1 ⫽ Sometimes to a doctor’s office and sometimes to an emergency room 1 ⫽ Other, specify: Everyday I follow the treatment program my doctor set up for me. 1 ⫽ Strongly disagree 1 ⫽ Disagree 0 ⫽ Agree 0 ⫽ Strongly agree What would you rather do? 1 ⫽ Eat, drink, and live my life the way I want and have poorer health in 5 years 0 ⫽ Give up eating, drinking, and living the way I want and have better health in 5 years Imagine you smoke a pack of cigarettes daily, what would you be likely to do? 0 ⫽ Participate in 12 weekly quit smoking classes, which cost $120 and meet near your home in the evening 0 ⫽ Participate in the quit smoking classes and take pills, which cost an extra $120 but which will ensure you really stay off cigarettes (class and pills total cost $240) 0 ⫽ Quit smoking on my own 1 ⫽ Cut back smoking on my own 1 ⫽ I probably would not quit or cut back until I feel a negative effect from smoking
related to patients time-value preferences. These were of the general form: “If you win the lottery today, which prize would you prefer? $1,000 cash paid today OR $1,500 paid in cash 5 years from today?”
Data Collection/Group Surveys Patients presenting for group sessions were given written survey forms to complete. All survey items were pretested, and the tool included selected and previously validated items from the Perceived Efficacy in Patient-Physician Interactions (PEPPI) questionnaire.26 Study personnel were available to provide assistance to individuals in interpreting and answering questions upon request. In addition to basic demographic information, survey topics included access to and preferences for primary care, quality of physician-patient interactions, knowledge of hypertension and its complications, and patient self-efficacy. Additional questions
Four follow-up survey items refined the patients’ timevalue preferences by presenting higher values for the hypothetical future payment to more accurately define individual discount rates. Finally, patients were asked a series of hypothetical questions regarding their actual or likely health behaviors. Topics included likelihood of checking BP at home, likelihood of altering diet and exercise patterns, likelihood of following a doctor’s treatment plans, likelihood of smoking cessation, and likelihood of seeking sick-day care outside of the primary care home. Relevant questions and the coding scheme for answers are shown in Table 1.
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
Statistical Analysis After survey completion, all demographic variables were entered into a secure database prior to analysis. Summary statistics were generated for demographic variables including age, gender, race, income level, and self-reported health status. From a theoretical perspective, we assumed that each respondent had an underlying discount rate that is not directly observable. However, the range in which the underlying discount rate must fall can be inferred. For example, if a person said they preferred $1,500 in 5 years over $1,000 today, then we would know that the rate at which this person discounts the future must be below the 8.5% annual interest rate implied by the $500 increase in value over 5 years; we assume that the discount rate is above 0%. Thus, while the persons rate of discounting is unobservable, it falls within the [0.0, 0.085] interval. Further, we assume that this discount rate is a function of the person’s characteristics, which can be used as explanatory variables in a regression model. Once we have estimated the regression model, we can use the parameters to predict the actual discount rate for each person. We used a generalized linear modeling technique called “interval regression,” or grouped regression, to estimate these parameters. (Note that this is a maximum likelihood model that uses upper and lower bounds on the implied discount rate to estimate the model; as such there is no “dependent variable” in the traditional sense of the word.) Respondents were then grouped into quintiles based on their estimated discount rates comparing the group with the highest rates of discounting future value to those in the lower four quintiles. The proportion of patients willing to forego individual health behaviors was then compared between groups using 2 tests. Additional dichotomous regression analyses using probit models adjusted for age, gender, race, income, health status, and insurance status were performed to model the incremental influence of a change in discount rate on the likelihood of forgoing specific health behaviors. All statistical calculations were performed using Stata version 10 (StataCorp, College Station, TX).
Results A total of 1,465 individuals were contacted via telephone to participate in group surveys; 744 (50.8%) agreed to participate with 422 attending (28.6%) and completing the full survey. Basic demographics for study participants are listed in Table 2. Overall, respondents were ⬃75% AfricanAmerican, 75% female, and 55% were younger than 60 years old. Approximately half had household incomes of less than $15,000 per year, and 70% had some form of public health insurance (ie, Medicare or Medicaid). Seventy-eight percent said their doctor answered their questions in a way they understood. Reported self-efficacy was high, but
37
Table 2 Demographic characteristics of respondents Variable
Total
%
Sample Race African-American White Hispanic Asian Native American Gender Male Female Age 20–30 31–40 41–50 51–60 61–70 71–80 81–90 School 1–8 9–12 13–16 17–20 Income ⬍$15,000 $15,000 – $30,000 $30,000 – $50,000 $50,000 – $75,000 ⬎$75,000 Health Insurance Medicare Medicaid Private US Government State Government None
n ⫽ 439 Total 303 112 11 2 2 Total 104 335 Total 13 38 72 119 123 60 9 Total 62 237 109 19 Total 217 117 51 26 11 Total 177 87 132 21 58 73
% of 430 70.47 26.05 2.56 0.47 0.47 % of 439 23.7 76.3 % of 434 3.1 8.83 16.59 27.67 28.6 13.95 2.07 % of 427 14.52 55.5 25.53 4.45 % of 422 51.42 27.73 12.08 6.16 2.61 % of 439 40.32 19.82 30.11 4.78 13.21 16.63
nonadherence to self-management and pill taking for their hypertension was admitted by one-third of patients. Over 70% reported taking some home remedy for hypertension. A total of 98% were willing to take more medications and/or make lifestyle changes if the link to better outcomes was clearly stated. The results of the interval regression model to estimate discount rates are presented in Table 3. The overall discount rate in our cohort was 0.438 or (43.8%) per year (standard deviation [SD], 0.07). Compared to the four lowest quintiles, patients in the highest quintile of discount rates tended to have lower likelihood of ever checking BP at home (42.5% vs. 47.6%; P ⫽ .36), of not using their physician’s office for sick care (16.5% vs. 27.6%; P ⫽ .01), and of not altering their diet and exercise habits in response to a diagnosis of hypertension (6.8% vs. 12.4%; P ⫽ .07). The respondents in the highest quintile discount rate group had discount rates between 50% and 57.2% per year. The proportion of patients likely to continue smoking was larger in
38
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
Table 3 Regression model estimating health discount rates Variable
Coefficient
Race (black ⫽ 1) Race (white ⫽ 1) Gender (male ⫽ 1) Age (40–69) Age (70⫹) Education (at least high school ⫽ 1) Income (⬍$15,000/y ⫽ 1) Health status (very good ⫽ 1) Year of data collection (y ⫽ 1) Constant Number of observations ⫽ 422
0.0776042 0.0626919 ⫺0.0669549 0.0176825 0.0569559 ⫺0.006622 0.0764141 ⫺0.1067717 0.0435879 0.3169918
P value .182 .301 .057 .695 .222 .856 .03 .009 .209 0
Wald chi-square ⫽ 22.26; Prob ⬎ chi-square ⫽ .0081
the higher discounting group though not statistically significant (2.5% vs. 3.2%; P ⫽ .72). There were no differences in the proportion of patients unlikely to follow physician’s care plans between groups. Subsequent dichotomous regression analyses used probit models adjusted for gender, age, race, income, health status, and insurance status. These marginal effects models suggested the incremental contribution of one additional percentage point in discount rate to the likelihood respondents would not engage in specific health behaviors. Presented in Table 4, the model predicted that increasing discount rate by one percentage point decreased the likelihood respondents would check their BP by 3.5% (P ⫽ .003). Presented in Table 5, the model predicted that increasing discount rate percentage point decreased the likelihood respondents would alter diet and exercise habits by 0.6% (P ⫽ .004). Presented in Table 6, the model predicted that increasing discount rate by one percentage point increased the likelihood respondents would not follow doctors’ treatment plans by 1.6% (P ⫽ .05). Table 4 Marginal effects model for likelihood of not checking BP at home Variable
dF/dx
P value
Discount Race (black) Race (white) Gender (male) Age (40–69) Age (70⫹) Education (at least high school) Income (⬍$15,000/y) Insurance (Medicare) Insurance (Medicaid) Insurance (private) Insurance (no insurance) Health status (very good) Number of observations ⫽ 422
0.03536 ⫺0.14619 ⫺0.16436 0.28357 ⫺0.07564 ⫺0.32705 0.04534 ⫺0.16936 0.022771 0.04733 ⫺0.0251 ⫺0.07439 0.29286
.003 .313 .191 .003 .350 .001 .430 .108 .746 .514 .696 .357 .028
BP, blood pressure.
Table 5 Marginal effects model for likelihood of not altering diet and exercise habits Variable
dF/dx
P value
Discount Race (black) Race (white) Gender (male) Age (40–69) Age (70⫹) Education (at least high school) Income (⬍$15,000/y) Insurance (Medicare) Insurance (Medicaid) Insurance (private) Insurance (no insurance) Health status (very good) Number of observations ⫽ 422
0.00651 0.38182 0.97125 0.14999 ⫺0.02319 ⫺0.04325 0.01392 ⫺0.00876 0.00262 0.01649 ⫺0.01418 ⫺0.02392 0.09467
.107 .069 .000 .085 .282 .134 .360 .758 .876 .437 .378 .088 .408
Discussion These preliminary data indicate that the degree to which individuals discount the future has a significant impact on their self-reported health behaviors. This insight is particularly important to make in a hypertensive population where control rates remain less than ideal. Significant research has been appropriately directed at addressing provider and health system-specific barriers to improvement in chronic disease care over recent years in part on the assumptions that patient-specific barriers are relatively more difficult to predict and relatively more intransigent. However, an economic framework for understanding individual time preferences offers new potential for understanding the variations observed in individual health behavior and compliance with hypertension care. It seems intuitive that discount rates should affect behavior. Support for this notion comes from Brown et al,27 who
Table 6 Marginal effects model for likelihood of not following physician treatment plan Variable
dF/dx
P value
Discount Race (black) Race (white) Gender (male) Age (40–69) Age (70⫹) Education (at least high school) Income (⬍$15,000/y) Insurance (Medicare) Insurance (Medicaid) Insurance (private) Insurance (no insurance) Health status (very good) Number of observations ⫽ 422
0.06709 ⫺0.33987 ⫺0.45551 0.38974 ⫺0.21869 ⫺0.78358 ⫺0.09759 ⫺0.64560 0.33589 0.11182 ⫺0.04824 0.45070 0.53239
.052 .401 .231 .199 .360 .021 .571 .047 .123 .600 .798 .054 .210
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
found that when compared to more “future-oriented” persons, those with a more “present-oriented” outlook perceived themselves to be less susceptible to the consequences of hypertension, believed more in home remedies, and were more skeptical of the benefits of prescribed medication. Importantly, the present study is the one of the first to apply this type of quantitative economic modeling in the analysis of health behaviors and lifestyles. Additional support for this concept comes from earlier work by our group with a large longitudinal database, the most recent wave of the Health and Retirement Survey (2004).28 In addition to core components of that survey including questions about healthrelated behaviors, basic socio-demographics, insurance characteristics, comorbidity, and patients’ recent (past two years) health screening activities, a selected group of 1,039 respondents (out of 20,129) aged 24 to 65 years old were surveyed via telephone with questions on time preferences. In this cohort, the interval regression model yielded a mean imputed annual discount rate of 0.26 (26%) per year (SD, 0.05), which is lower than our current findings. Compared to respondents with moderate-to-low imputed discount rates, respondents in the highest imputed discount rate percentile group had a statistically lower rate of mammogram use (0.669 vs. 0.802; P ⫽ .0003), Papanicolaou (PAP) smear use (0.689 vs. 0.764; P ⫽ .03), prostate exam use (0.520 vs. 0.717; P ⫽ .002), dental visits (0.488 vs. 0.722; P ⬍ .001), cholesterol testing (0.716 vs. 0.797; P ⫽ .014), flu shot usage (0.438 vs. 0.511; P ⫽ .06), rates of vigorous exercise (0.219 vs. 0.378; P ⬍ .001), and higher rates of smoking (0.294 vs. 0.221; P ⫽ .03). Taken together, these results argue for more in-depth research in this area. Overall, the rates of discount we measured were quite high, at nearly 44% per year. It is important to note that clinics for the EXCEED Study were selected from clinics in the Hypertension Initiative that were known to provide services for disproportionately more low income and minority patients. The intentional selection bias may have affected the findings. Notwithstanding, such rates of future depreciation far exceed rates that would be acceptable as interest rates for mortgage loans or as depreciation rates on vehicles. They are not excessive when compared to the annualized interest rates on many types of consumer credit such as some credit cards or “payday lending” outlets. Payday loans, which are banned in several U.S. states, are a form of predatory lending also called “cash advances” where the annualized interest rates can reach 400%.29 However, such high discount rates for health may begin to rationally explain the apparently short-sighted decisions made by hypertensive patients regarding their health behaviors. Furthermore, the marginal effect of a 10% increase in discount rate on the probability of foregoing health behaviors in our models appears to be at least moderate in size. For example, a 35% increase in the likelihood of foregoing home BP checks observed in our model was similar in magnitude to the 32% decrease in probability observed with
39
moving from the lower age range (40 to 69 years) to the higher age range (70 years and up). Similarly, the 16% increase in the likelihood of not following a doctor’s treatment plans observed in our model was similar in magnitude to the 15% decrease in probability observed with moving into the lower income group from the other income groups. This research has significant implications for clinical care and for policy making. First, if the estimated discount rates are accurate, there is little hope that some individuals will voluntarily undertake preventive activities that entail significant near-term costs in terms of time, money, or effort if the perceived benefits accrue many years in the future. In that case, even very large future benefits to the individual would be discounted heavily in present value terms, and clinical attempts to alter patients’ behavior by stressing the health benefits accruing 10, 15, or 20 years in the future would likely be unsuccessful. For policy makers, the observation that high rates of discount are associated with negative effects on health behavior implies that increased advocacy and education for prevention may not be universally effective tools in certain patient groups. It will be important to incorporate this new knowledge into existing theoretical frameworks for health promotion such as the Predisposing, Reinforcing, and Enabling Constructs in Educational Diagnosis and Evaluation–Policy, Regulatory, and Organizational Constructs in Educational and Environmental Development (PRECEDE-PROCEED) model of Green and Kreuter.30 This dominant model of health promotion during the past three decades outlines processes for the program planning and evaluation. Individual time preferences might logically be incorporated into this framework as one of the predisposing factors surveyed as educational and ecological inputs during phase 4 of the PRECEDE process. Such factors include a “person or population’s knowledge, attitudes, beliefs, values, and perceptions that facilitate or hinder motivation for change.” This concept is illustrated graphically in the Figure. Health discount rates are necessarily individual characteristics, and prior research suggests that these are relatively stable within individuals over time.23 That is, individual discount rates may be somewhat inflexible and thus, not as amenable to traditional educational efforts. However, knowledge of mean discount rates for specific populations may help inform the types of interventions best suited to populations with different levels of health discounting. For instance, programs that promote more aggressive provider-driven disease management for hypertension and other chronic diseases may be emphasized over education and self-management strategies in groups with high levels of discounting. An alternative or complementary strategy would be to identify and promote potentially favorable near-term benefits, eg, improved exercise tolerance or to link long-term benefits to a clear outcome, eg, participating in a child or grandchild’s graduation or marriage ceremony.
40
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
4.
5.
6. Figure. Conceptual framework for incorporating health discount rates into the PRECEDE-PROCEED model of health promotion planning. Adapted from Green LW, Kreuter M. Health Program Planning: An Educational and Ecologic Approach. 4th ed. New York, New York: McGraw Hill, 2005.30
7.
8. Our study has several limitations. First, the survey was limited to 422 individuals. While this limited number of patients was adequate for the number of covariates we evaluated, more complex modeling was not possible with this data set. Secondly, some elements of our survey tool were not validated prospectively. However, the questionnaires were pretested, and selected query items were drawn from a previously validated survey tool. Additionally, our measure of health behaviors was indirect, relying on patient self-report. It is possible that patients’ actual behaviors could differ from their survey responses, but this measure represents a necessary first step along this line of investigation. Future inquiry in this area will directly test this relationship by ensuring a larger sample size and controlling for multiple potential confounders. Such study will allow for further refinement of the valuation methodology in order to make it more clinically applicable. One can imagine that a parsimonious and prospectively validated survey tool for clinical and research use could be useful in promoting preventive health behaviors.
9.
10.
11.
12.
13.
14.
15.
References 1. Kannel WB, Wilson PW. Cardiovascular risk factors and hypertension. In: Izzo JL BH, editor. Hypertension Primer. 3rd ed. Philadelphia, Pennsylvania: Lippincott, Williams, and Williams, 2003:235–39. 2. Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of hypertension in the United States, 1988 –2000. JAMA 2003;290:199 –206. 3. Fields LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, Sorlie P. The burden of adult hypertension in the United
16.
17. 18.
States 1999 to 2000: a rising tide. Hypertension 2004;44:398 – 404. Ostchega Y, Yoon SS, Hughes J, Louis T. Hypertension awareness, treatment, and control-continued disparities in adults: United States, 2005–2006. NCHS Data Brief 2008;3:1– 8. Chobanian AV, Bakris GL, Black HR, Cushman WC, Izzo JL. Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure. National High Blood Pressure Education Program Coordinating Committee. Hypertension 2003;42:1206 –52. Healthy People 2010. U.S. Department of Health and Human Services, 2007. Available at: http://www. healthypeople.gov/. Accessed: April 30, 2007. Oliveria SA, Lapuerta P, McCarthy BD, L’Italien GJ, Berlowitz DR, Asch SM. Physician-related barriers to the effective management of uncontrolled hypertension. Arch Intern Med 2002;162:413–20. Phillips LS, Branch WT, Cook CB, Friedman RH, Glickman M, Kader B, et al. Clinical inertia. Ann Intern Med 2001;135:825–34. Berlowitz DR, Ash AS, Hickey EC, Doyle JP, El-Kebbi IM, Gallina DL, et al. Inadequate management of blood pressure in a hypertensive population. N Engl J Med 1998;339:1957– 63. Okonofua EC, Simpson KN, Jesri A, Rehman SU, Durkalski VL, Egan BM. Therapeutic inertia is an impediment to achieving the Healthy People 2010 blood pressure control goals. Hypertension 2006;47:345–51. Brand FN, Smith RT, Brand PA. Effect of economic barriers to medical care on patients’ noncompliance. Public Health Rep 1977;92:72– 8. Lurie N, Ward NB, Shapiro MF, Brook RH. Termination from Medi-Cal— does it affect health? N Engl J Med 1984;311:480 – 84. Shulman NB, Martinez B, Brogan D, Carr AA, Miles CG. Financial cost as an obstacle to hypertension therapy. Am J Public Health 1986;76:1105– 8. Kotchen JM, Shakoor-Abdullah B, Walker WE, Chelius TH, Hoffmann RG, Kotchen TA. Hypertension control and access to medical care in the inner city. Am J Public Health 1998;88:1696 –99. Hill MN, Bone LR, Kim MT, Miller DJ, Dennison CR, Levine DM. Barriers to hypertension care and control in young urban black men. Am J Hypertens 1999;12: 951–58. Lawrence W, Green MWK. Health Promotion Planning: An Educational and Environmental Approach. 2nd ed. Mountain View, California: Mayfield Publishing Co, 1991. Grossman M. On the concept of health capital and the demand for health. J Polit Econ 1972;80:223–55. Wagstaff A. The demand for health: some new empirical evidence. J Health Econ 1986;5:195–233.
R.N. Axon et al. / Journal of the American Society of Hypertension 3(1) (2009) 35– 41
19. Van Doorslaer E. Health, Knowledge, and the Demand for Medical Care: An Econometric Analysis. Maastricht, The Netherlands: van Gorcum and Comp, 1987. 20. Wagstaff A. The demand for health: an empirical reformulation of the Grossman model. Health Econ 1993; 2:189 –98. 21. Grossman M, Kaestner R. Effects of education on health. In: Behrman J, Stacey N, The Social Benefits of Education. Ann Arbor, Michigan: University of Michigan Press, 1997:69 –123. 22. Zweifel P, Breyer F. Health Economics. New York, New York: Oxford University Press, 1997. 23. Frederick S, Loewenstein G, O’Donoghue T. Time discounting and time preference: a critical review. J Econ Lit 2002;XL:351– 401. 24. Bradford W, Kleit A, Krausel-Wood M, Re R. Doublebounded dichotomous choice method to assess willingness to pay for telemedicine. J Telemed Telecare 2004; 10:325–30. 25. Lackland DT, Lin Y, Tilley BC, Egan BM. An assessment of racial differences in clinical practices for hy-
26.
27.
28.
29.
30.
41
pertension at primary care sites for medically underserved patients. J Clin Hypertens (Greenwich) 2004;6:26 –31; quiz 2–3. Maly RC, Frank JC, Marshall GN, DiMatteo MR, Reuben DB. Perceived efficacy in patient-physician interactions (PEPPI): validation of an instrument in older persons. J Am Geriatr Soc 1998;46:889 –94. Brown C, Segal R. Ethnic differences in temporal orientation and its implications for hypertension management. J Health Soc Behav 1996;37:350 – 61. Bradford W. The role of individual time preferences on patients health maintenance habits. In: Center for Health Economics and Policy (CHEPS) Working Paper. Charleston, South Carolina: The Medical University of South Carolina, 2007:26. Center for Responsible Lending: A Resource for Predatory Lending Opponents. 2008. Available at: http://www. responsiblelending.org/. Accessed: June 19, 2008. Green LW, Kreuter M. Health Promotion Planning: An Educational and Ecologic Approach. 4th ed. New York, New York: McGraw Hill, 2005.