Chronic Disease Diagnosis as a Teachable Moment for Health Behavior Changes Among Middle-Aged and Older Adults

Abstract

Objective: To examine the impact of a new chronic disease diagnosis on substance use (i.e., smoking and drinking), utilization of preventive medical procedures, and physical activity among middle-aged and older adults. Method: Individual-level data came from 1996 to 2010 waves of the U.S. Health and Retirement Study. Disease diagnosis was ascertained from self-reports of physician-diagnosed diseases. A case-control difference-in-differences approach estimated in logistic regression was applied to test study hypotheses. Results: After a diagnosis of chronic disease, participants decreased substance use and increased utilization of preventive medical procedures. Physical activity declined after a diagnosis of lung disease, cancer, and stroke. Discussion: Chronic disease diagnosis may be an important teachable moment that can motivate individuals to adopt multiple risk-reducing health behaviors. Future research needs to elucidate the mechanisms through which disease diagnosis affects behaviors and test the modifying effect of time since diagnosis on intervention effectiveness.

Keywords

Health and Retirement Study teachable moment health behavior physical activity

The elderly population is growing at exponential rates and enjoying a much longer life expectancy (Ortman, Velkoff, & Hogan, 2014). One consequence of population aging is the rising epidemic of chronic disease. Currently, about 80% of adults aged 65 years and older in the United States have one chronic disease and half have at least two (He, Sengupta, Velkoff, & DeBarros, 2005). People with multiple chronic conditions are the major driving force for the increase in health care spending. A recent analysis of the Medicare and Medicaid claims data showed that the nearly one third of beneficiaries with zero or one chronic condition accounted for only 7% of Medicare spending, whereas 14% with six or more chronic conditions accounted for almost half (46%) of Medicare spending (Centers for Medicare and Medicaid Services, 2012). The growing burden of chronic diseases in our aging society highlights the importance of promoting health behaviors among middle-aged and older adults. Type 2 diabetes, heart disease, stroke, cancer, and chronic lung disease are potentially preventable and certainly manageable conditions (He et al., 2005). Health behaviors of smoking cessation, moderate drinking, regular exercise, and being up to date with recommended preventive medical procedures (e.g., influenza vaccination and cancer screening) have been demonstrated to play a critical role in healthy aging (Sabia et al., 2012). Engaging in these health behaviors has established benefits, from preventing morbidity among healthy individuals (Akesson, Larsson, Discacciati, & Wolk, 2014) to reducing additional morbidity and functional limitations among those with chronic diseases (Rejeski et al., 2012).

However, unhealthy behaviors prevail in the U.S. populations of middle-aged and older adults. One fifth of middle-aged adults and nearly 10% of older adults are current smokers; and nearly one fifth of middle-aged adults have five or more drinks in a day at least once in the past year (Schiller, Ward, & Freeman, 2014). Moreover, older adults are the least physically active of any age group. The rates of middle-aged and older adults being up to date with recommended preventive medical procedures (Schiller et al., 2014) are not meeting the 2020 Healthy People goals (U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion, 2015). Developing strategies to promote healthy behaviors in midlife and older adulthood is thus crucial to promote healthy aging and to reduce individual suffering and medical costs.

Behavioral theories and interventions emphasize the importance of cues in promoting motivation for behavior change (Hochbaum, 1958; Rosenstock, 1974). A specific type of cue, labeled as “teachable moment” (McBride, Emmons, & Lipkus, 2003), describes a naturally occurring health event that could motivate individuals to spontaneously adopt risk-reducing health behaviors. While chronic illnesses are a major threat to health and quality of life, disease diagnosis itself could serve as a critical teachable moment for patients to initiate behavior change. If this hypothesis holds, health promotion programs may take advantage of the timing of disease diagnosis to enhance effectiveness. However, a comprehensive literature review revealed that population-based studies testing this hypothesis are limited. Few studies incorporated comparisons with healthy controls into their design to estimate the net changes in health behaviors attributable to a chronic disease diagnosis (Bidstrup et al., 2013; Karlsen et al., 2012; Newsom et al., 2012; Williams, Steptoe, & Wardle, 2013). The majority of studies either focused on a single disease (predominately cancer; Bidstrup et al., 2013; Hawkes, Lynch, Youlden, Owen, & Aitken, 2008; Karlsen et al., 2012; Williams et al., 2013) or grouped varying conditions together (Falba, 2005; Keenan, 2009; Margolis, 2013). Furthermore, most studies focused on a single health behavior outcome (Platt, Sloan, & Costanzo, 2010; Schneider et al., 2014), particularly smoking (Falba, 2005; Keenan, 2009; Twardella et al., 2006).

The present study examined the effect of a new chronic disease diagnosis on health behaviors over a 14-year period in a nationally representative sample of middle-aged and older adults in the United States. I examined a number of chronic diseases, including diabetes, heart disease, stroke, cancer, and chronic lung disease, and tested changes in multiple health behaviors, including substance use habits (i.e., smoking and drinking), utilization of preventive medical procedures (i.e., influenza vaccination, test for cholesterol, prostate exam, and mammogram), and patterns of physical activity.

Method

Data Source

This study analyzed eight waves (1996-2010) of individual-level data from the U.S. Health and Retirement Study (HRS), a longitudinal panel study of changes in health and labor force participation in later life. The HRS is sponsored by the National Institute on Aging (Grant NIA U01AG009740) and conducted by the University of Michigan. The HRS conducts biannual interviews with age-eligible members (from specific birth cohorts, generally those older than 50 years of age at the time of the first interview) from eligible household units, along with their non-age-eligible spouses. Eligible household units were selected using a multistage area probability sampling design, with oversamples of African Americans, Hispanics, and Floridians. The methodology of the HRS includes the use of “proxy” respondents when the original respondent could not complete the interview. Detailed information on the survey design, questionnaires, and relevant data are available at the HRS web portal (http://hrsonline.isr.umich.edu).

Participants

This study included four birth cohorts—the original HRS cohort, the asset and health dynamics among the oldest old (AHEAD) cohort, the children of depression (CODA) cohort, and the war baby (WB) cohort. Because birth cohorts entered the HRS at different years, all data points were centered at the time each cohort was selected, referred to as baseline hereafter. Among the 22,840 participants interviewed at baseline, I excluded respondents younger than 50 years of age (n = 905); older than 80 years of age (n = 2,317); proxies (n = 1,401); persons with missing data on chronic disease diagnosis (n = 17), and those with an existing diagnosis of diabetes, heart disease, stroke, cancer, or chronic lung disease (n = 6,761). The final study sample consisted of 11,439 participants who were free of the five chronic diseases at baseline and followed for an average of 12.4 years (6.2 waves). I excluded participants with an existing chronic disease diagnosis at baseline to reduce the potential confounding effects of a prior diagnosis on health behaviors.

Measures

Chronic disease diagnosis

During each interview, the respondents were asked, “Has a doctor ever told you that you had . . .” for the following list of conditions: (a) “diabetes or high blood sugar?” (b) “a heart attack, coronary heart disease, angina, congestive heart failure, or other heart problems?” (c) “stroke?” (d) “cancer or a malignant tumor, excluding minor skin cancer?” and (e) “chronic lung disease, such as chronic bronchitis or emphysema?” A diagnosis was ascertained from answers of “yes” to these questions, respectively. I determined the time of diagnosis by the study wave when each condition was first reported.

Substance use habits

Current smoking was ascertained from answers of “yes” to the question, “Do you smoke cigarettes now?” The HRS asked several questions about drinking habits, including (a) “Do you ever drink any alcoholic beverages such as beer, wine, or liquor?” (b) “In the last three months, on average, how many days per week have you had any alcohol to drink?” (c) “In the last three months, on the days you drink, about how many drinks do you have?” and (d) “In the last three months, on how many days have you had four or more drinks on one occasion?” I used answers from the first three questions to calculate average drink per day and the fourth question for incidence of binge drinking (defined as ≥4 drinks on any occasion). Based on this information and the 2010 Dietary Guidelines for Americans (U.S. Department of Agriculture & U.S. Department of Health and Human Services, 2010), I created a single indicator of excessive drinking, defined as >1 drink per day on average or any binge drinking in the past 3 months for women, and >2 drinks per day on average or any binge drinking in the past 3 months for men.

Utilization of preventive medical procedures

Utilization of preventive medical procedures was assessed consistently but not repeatedly at every interview. At baseline, participants were asked if they had the following medical procedures in the last 2 years, including (a) a influenza vaccination, (b) a blood test for cholesterol, (c) an examination of prostate to screen for cancer (for men only), and (d) a mammogram or X-ray of the breast to screen for cancer (for women only). The same questions were asked again in the odd waves only. To facilitate analysis, I carried forward prior wave values when appropriate (e.g., Wave 3 values forward to Wave 4) as recommended by Chien et al. (2013).

Patterns of physical activity

Being physically active was ascertained from answers of “yes” to the question, “On average over the last 12 months have you participated in vigorous physical activity or exercise three times a week or more?” The HRS initially defined vigorous physical activity as things such as sports, heavy housework, or a job that involves physical labor. After the 2002 survey, the wording and response categories of physical activity questions changed, and activity due to work was excluded; therefore, I used data points from 1996 through 2002 only to model physical activity to avoid bias due to measurement change.

Individual characteristics

I adjusted the following individual characteristics at baseline in multivariate analysis: age, sex, race/ethnicity, education, marital status, employment status, household net wealth (coded into quartiles due to skewness), any difficulty in performing activities of daily living (ADLs; tasks, including bathing, eating, dressing, walking across a room, and getting in or out of bed), and self-reports of physician-diagnosed hypertension.

Statistical Analysis

I utilized a case-control difference-in-differences (DD) approach (Angrist & Pischke, 2009), a commonly used econometric method, to estimate the effect of a new diagnosis on participants’ health behaviors. The DD approach estimates the unbiased effect of a chronic disease diagnosis by distinguishing the secular trends in engagement in a health behavior (e.g., decline physical activity due to physiological changes associated with aging). The key assumption is that if people with a chronic disease had not been diagnosed with the disease, they would have showed the same amount of change as their disease-free counterparts in a health behavior during the aging process. The independent effect of a chronic disease diagnosis can therefore be estimated by netting out the secular trends observed among disease-free individuals from total changes observed among those with a chronic disease diagnosis. The DD estimator requires observations on cases before and after the diagnosis and simultaneous data on a control group. I used a non-parametric nearest-neighbor matching method to match individuals with the specific chronic disease diagnosis (“cases”) to those free of all five chronic diseases throughout the study period (“controls”). The primary purpose of matching was to assign artificial “pre- and post-diagnosis periods” for “controls” to facilitate comparisons. I chose non-parametric matching because it works well when there are relatively few covariates and is robust against specification error (Stuart, 2010). I applied one-to-one matching without replacement to ensure group balance and included study cohorts and total waves of follow-up as matching variables. Following balance check, I estimated the DD models in Stata 11.0 SE (StataCorp, College Station, TX) using logit command with clustered robust standard errors to account for correlation within subjects. The relevant post-estimation procedure (margins command with dydx function) in Stata was used to calculate the net changes in the probability of engagement in a health behavior attributable to a chronic disease diagnosis.

Results

Table 1 presents the sample characteristics. Participants were 61.4 years old on average (linearized standard error [SE] = 0.15). The majority was female (55.5%), non-Hispanic White (82.2%), married (68.8%), and not employed (50.3%). The average amount of household wealth was US$311,698 (SE = 12,422) in 1996 dollars. The prevalence of ADLs limitations was 7.7%, and nearly one third (31.4%) of the participants had hypertension at baseline. During the follow-up, one fifth (20.7%) reported a heart disease diagnosis, followed by diabetes (15.4%), cancer (13.5%), chronic lung disease (9.0%), and stroke (7.8%).

Table 1.

Characteristics of the Study Sample From the HRS.

Baseline characteristics	Unweighted number (N = 11,439)	Weighted
HRS cohorts (%)
AHEAD	2,156	15.59 [14.59, 16.66]
CODA	1,159	11.26 [10.13, 12.51]
HRS	6,554	42.27 [40.99, 43.55]
WB	1,570	30.87 [29.25, 32.55]
Age in years	—	61.39 [61.09, 61.69]
Sex (%)
Female	6,868	55.49 [54.38, 56.60]
Male	4,571	44.51 [43.40, 45.62]
Race/ethnicity (%)
White, non-Hispanic	8,779	82.24 [79.73, 84.50]
African American, non-Hispanic	1,538	9.11 [8.10, 10.24]
Other race/multi-race, non-Hispanic	952	1.76 [1.24, 2.49]
Hispanic	170	6.88 [5.03, 9.35]
Education attainment (%)
Less than high school	2,602	19.21 [17.70, 20.82]
High school	4,275	36.84 [35.43, 38.27]
Some college	2,312	21.17 [20.05, 22.32]
College graduate or higher	2,249	22.78 [20.85, 24.84]
Marital status (%)
Married/partnered	8,371	68.80 [67.15, 70.41]
Divorced/separated/widowed	2,678	27.01 [25.50, 28.58]
Never married	361	4.18 [3.68, 4.76]
Employment status (%)
Work full-time or part-time	5,037	49.67 [48.26, 51.08]
Unemployed/not in labor force/retired/other	6,402	50.33 [48.92, 51.74]
Household wealth (US$)	—	311,698 [286,771, 336,625]
Had difficulty in activities of daily living (%)	917	7.72 [7.08, 8.42]
Hypertension diagnosis (%)	3,825	31.40 [30.26, 32.56]
Chronic disease diagnosis during follow-up (%)
Diabetes	1,831	15.37 [14.51, 16.26]
Heart disease	2,454	20.71 [19.93, 21.52]
Stroke	988	7.84 [7.32, 8.40]
Cancer	1,607	13.54 [12.80, 14.31]
Chronic lung disease	1,026	8.97 [8.33, 9.66]

Note. HRS sampling design was incorporated in estimating. 95% confidence intervals in brackets. HRS = Health and Retirement Study; AHEAD = Asset and Health Dynamics Among the Oldest Old; CODA = Children of Depression; WB = War Baby.

Table 2 presents the average prevalence of engagement in a health behavior before and after a new chronic disease diagnosis among participants with chronic disease. Consistent patterns of change emerged across disease diagnoses. Overall, the rates of substance use declined and utilization of preventive medical procedures increased during the post-diagnosis period. However, the percentage of participants engaging in vigorous physical activity decreased following a disease diagnosis.

Table 2.

Average Prevalence of Health Behaviors Before and After a Chronic Disease Diagnosis Among Persons With Chronic Disease.

Chronic disease diagnosis	Health behaviors (%)
Chronic disease diagnosis	Smoking	Excessive drinking	Vigorous physical activity	Influenza vaccination	Cholesterol test	Prostate exam (for male only)	Mammography (for female only)
Diabetes
Pre-diagnosis	16.97	10.66	42.29	50.19	74.17	70.00	71.44
	[14.72, 19.49]	[9.07, 12.48]	[40.07, 44.54]	[47.31, 53.07]	[72.13, 76.11]	[65.65, 74.02]	[68.54, 74.17]
Post-diagnosis	10.37	8.51	39.38	67.31	89.13	78.13	71.75
	[8.18, 13.07]	[6.89, 10.48]	[35.44, 43.46]	[55.28, 60.44]	[87.42, 90.63]	[74.13, 81.66]	[67.95, 75.26]
Difference	−6.60***	−2.15*	−2.91	17.12***	14.96***	8.13***	0.31
Heart disease
Pre-diagnosis	17.12	13.22	48.37	57.52	76.08	74.09	73.55
	[14.94, 19.54]	[11.25, 15.48]	[45.98, 50.77]	[54.56, 60.43]	[74.32, 77.75]	[71.03, 76.93]	[71.02, 75.94]
Post-diagnosis	10.03	9.45	39.57	73.68	89.67	77.25	71.56
	[8.31, 12.06]	[8.13, 10.96]	[35.59, 43.69]	[70.80, 76.38]	[88.02, 91.11]	[74.50, 79.78]	[68.65, 74.29]
Difference	−7.09***	−3.77***	−8.80***	16.16***	13.59***	3.16*	−1.99
Stroke
Pre-diagnosis	18.94	12.60	47.08	60.12	74.93	69.86	70.50
	[15.72, 22.65]	[10.69, 14.80]	[43.18, 51.02]	[56.75, 63.40]	[71.88, 77.74)	[64.78, 74.49]	[65.15, 75.34]
Post-diagnosis	14.40	8.10	21.23	71.11	86.05	74.86	63.32
	[11.30, 18.18]	[6.44, 10.16]	[17.26, 25.82]	[68.18, 73.87]	[83.55, 88.22]	[70.82, 78.52]	[58.43, 67.96]
Difference	−4.54*	−4.50**	−25.85***	10.99***	11.12***	5.00	−7.18**
Cancer
Pre-diagnosis	18.90	15.63	51.53	51.53	73.92	69.86	73.4
	[15.84, 22.40]	[13.39, 18.17]	[48.90, 54.15)	[48.29, 54.75)	[71.13, 76.53)	[65.29, 74.07]	[69.66, 76.84]
Post-diagnosis	9.14[6.94, 11.94]	11.04[9.58, 12.69]	39.54[35.33, 43.91]	71.19[68.36, 73.87]	86.06[83.53, 88.26]	80.79[77.83, 83.44]	79.38[76.23, 82.22]
Difference	−9.76***	−4.59***	−11.99***	19.66***	12.14***	10.93***	5.98**
Chronic lung disease
Pre-diagnosis	40.31	15.29	44.99	54.09	72.47	63.62	71.84
	[36.61, 44.12]	[13.48, 17.29]	[42.09, 47.93]	[50.64, 57.50]	[69.13, 75.57]	[58.74, 68.23]	[67.94, 75.44]
Post-diagnosis	26.90	9.80	31.41	71.91	84.84	71.24	68.44
	[23.19, 30.97]	[8.20, 11.66]	[25.82, 37.59]	[67.97, 75.54]	[82.51, 86.91]	[64.49, 77.16]	[64.38, 72.23]
Difference	−13.41***	−5.49***	−13.58***	17.82***	12.37***	7.62*	−3.4

Note. HRS sampling design was incorporated in estimating. 95% confidence intervals in brackets.

p < .05. **p < .01. ***p < .001.

Table 3 shows the effect of a chronic disease diagnosis on health behaviors from logistic regression after adjusting for individual characteristics. The DD estimators, corresponding to rows “Diagnosis × Time,” represented the independent effect of a new chronic disease diagnosis on the odds of engagement in a particular behavior. An odds ratio less than 1 denotes reduced likelihood in the occurrence of the behavior outcome post diagnosis, whereas an odds ratio greater than 1 denotes increased likelihood of the behavior outcome post diagnosis. Because interaction terms in logistic regression can be difficult to interpret, I used relevant post-estimation procedure to calculate the independent effect of a chronic disease in probability terms.

Table 3.

The Effect of a Chronic Disease Diagnosis on the Likelihood of Behavior Change Estimated in Difference-in-Differences Model.

Chronic disease diagnosis	Smoking	Excessive drinking	Vigorous physical activity	Influenza vaccination	Cholesterol test	Prostate exam (for male only)	Mammography (for female only)
Diabetes
Diagnosis	1.08	0.66	0.70	1.12	1.15	1.17	0.90
	[0.89, 1.32]	[0.54, 0.79]***	[0.63, 0.78]***	[0.99, 1.28]	[1.02, 1.31]*	[0.96, 1.41]	[0.75, 1.07]
Time	1.32	1.26	0.85	1.12	1.36	1.04	1.08
	[1.12, 1.56]***	[1.06, 1.49]**	[0.74, 0.98]*	[1.00, 1.25]	[1.20, 1.54]***	[0.86, 1.27]	[0.93, 1.26]
Diagnosis × Time	0.61	0.72	1.06	1.36	1.91	1.31	1.22
	[0.49, 0.75]***	[0.57, 0.90]**	[0.86, 1.30]	[1.17, 1.58]***	[1.60, 2.28]***	[1.01, 1.68]*	[1.00, 1.49]
Heart disease
Diagnosis	1.36	0.94	0.93	1.24	1.22	1.43	1.08
	[1.14, 1.62]**	[0.79, 1.11]	[0.81, 1.02]	[1.11, 1.38]***	[1.09, 1.36]**	[1.20, 1.70]***	[0.93, 1.25]
Time	1.10	1.07	0.87	1.14	1.26	1.10	1.06
	[0.94, 1.29]	[0.92, 1.24]	[0.76, 0.99]*	[1.03, 1.26]*	[1.13, 1.40]***	[0.92, 1.30]	[0.94, 1.20]
Diagnosis × Time	0.72	0.74	0.89	1.29	1.98	0.91	1.08
	[0.59, 0.88]**	[0.61, 0.89]**	[0.74, 1.06]	[1.13, 1.48]***	[1.69, 2.32]***	[0.73, 1.13]	[0.91, 1.29]
Stroke
Diagnosis	1.60	1.14	0.97	1.05	1.09	1.05	0.96
	[1.20, 2.14]**	[0.87, 1.48]	[0.84, 1.12]	[0.88, 1.25]	[0.91, 1.29]	[0.80, 1.39]	[0.77, 1.21]
Time	1.55	1.20	0.82	1.25	1.24	0.86	1.04
	[1.20, 2.02]**	[0.97, 1.49]	[0.66, 1.01]	[1.07, 1.47]**	[1.04, 1.48]*	[0.64, 1.15]	[0.85, 1.26]
Diagnosis × Time	0.66	0.60	0.49	0.99	1.43	1.24	0.82
	[0.48, 0.91]*	[0.44, 0.83]**	[0.36, 0.68]***	[0.80, 1.22]	[1.12, 1.82]**	[0.87, 1.77]	[0.63, 1.06]
Cancer
Diagnosis	1.50	1.10	0.88	1.04	1.09	1.17	1.03
	[1.22, 1.85]***	[0.90, 1.33]	[0.79, 0.99]*	[0.91, 1.19]	[0.95, 1.26]	[0.95, 1.43]	[0.85, 1.26]
Time	1.12	1.22	1.09	1.12	1.17	0.98	1.07
	[0.92, 1.35]	[1.02, 1.46]*	[0.93, 1.27]	[0.99, 1.27]	[1.03, 1.34]*	[0.80, 1.20]	[0.91, 1.26]
Diagnosis × Time	0.50	0.72	0.66	1.44	1.63	1.80	1.71
	[0.39, 0.64]***	[0.58, 0.90]**	[0.53, 0.81]***	[1.23, 1.69]***	[1.35, 1.96]***	[1.38, 2.35]***	[1.34, 2.18]***
Chronic lung disease
Diagnosis	4.52	1.47	0.77	1.18	1.11	1.19	1.00
	[3.55, 5.75]***	[1.14, 1.90]**	[0.66, 0.89]***	[1.00, 1.41]	[0.93, 1.32]	[0.90, 1.58]	[0.80, 1.26]
Time	1.29	1.13	0.88	1.01	1.13	1.07	0.91
	[1.03, 1.62]*	[0.91, 1.42]	[0.72, 1.09]	[0.87, 1.17]	[0.96, 1.34]	[0.82, 1.40]	[0.75, 1.09]
Diagnosis × Time	0.61	0.65	0.69	1.70	1.75	0.97	1.07
	[0.47, 0.79]***	[0.50, 0.86]**	[0.52, 0.93]*	[1.39, 2.08]***	[1.39, 2.20]***	[0.69, 1.37]	[0.82, 1.38]

Note. Odds ratios and 95% confidence intervals presented in brackets. Persons with a chronic disease were matched to those free of all five chronic diseases in the study period. Baseline characteristics adjusted for age, sex, ethnicity, education level, marital status, employment status, household wealth, activities of daily living, and hypertension.

p < .05. **p < .01. ***p < .001.

Table 4 shows the net changes in the engagement of a health behavior resulting from a chronic disease diagnosis in probability terms. A negative value indicates a decrease in engagement in a behavior, whereas a positive value suggests an increase.

Table 4.

Net Changes in the Probability of Engaging in a Health Behavior Attributable to a Chronic Disease Diagnosis.

Health behaviors	Chronic disease diagnosis
Health behaviors	Diabetes	Heart disease	Stroke	Cancer	Chronic lung disease
Smoking	−.047 (−.065, −.029)***	−.030 (−.047, −.013)**	−.038 (−.065, −.011)**	−.064 (−.084, −.044)***	−.067 (−.100, −.034)***
Excessive drinking	−.026 (−.043, −.010)**	−.025 (−.040, −.010)**	−.036 (−.056, −.016)***	−.030 (−.049, −.011)**	−.035 (−.057, −.014)**
Vigorous physical activity	.013 (−.036, .062)	−.028 (−.070, .015)	−.157 (−.223, −.092)***	−.099 (−.148, −.049)***	−.084 (−.149, −.019)*
Influenza vaccination	.065 (.034, .097)***	.053 (.025, .080)***	−.003 (−.048, .042)	.078 (.044, .112)***	.108 (.068, .148)***
Cholesterol test	.090 (.068, .111)***	.094 (.075, .113)***	.054 (.020, .089)**	.074 (.048, .101)***	.083 (.052, .114)***
Prostate exam (for men only)	.047 (.003, .092)*	−.018 (−.059, .023)	.042 (−.025, .109)	.100 (.058, .142)***	−.006 (−.073, .061)
Mammography (for women only)	.036 (.000, .071)	.014 (−.018, .046)	−.039 (−.091, .013)	.085 (.050, .121)***	.012 (−.036, .059)

Note. 95% confidence intervals in parentheses. Changes in probability were obtained using margins command with dydx function in Stata based on results from the case-control difference-in-differences estimator estimated in logistic regression. Analyses adjusted for baseline characteristics, including age, sex, ethnicity, education level, marital status, employment status, household wealth, limitations in activities of daily living, and hypertension.

p < .05. **p < .01. ***p < .001.

For example, the estimated change in the probability of smoking after a diabetes diagnosis was −0.047, which we could interpret as a diabetes diagnosis was associated with 47 per 1,000 persons quitting smoking. Similarly, 30 to 67 per 1,000 persons quit smoking after other diagnoses with the largest change associated with lung disease, and about 30 per 1,000 persons stopped excessive drinking. A chronic disease diagnosis was associated with a net increase of 53 to 108 per 1,000 persons receiving influenza vaccination, with the largest change associated with chronic lung disease; and 54 to 94 per 1,000 persons receiving a cholesterol test, with the largest changes associated with diabetes and heart disease. A cancer diagnosis was associated with an increase of 100 per 1,000 men receiving a prostate exam and 85 per 1,000 women receiving a mammogram. The utilization of prostate exams did not change after a diagnosis of heart disease, stroke, or chronic lung disease. Chronic disease diagnoses other than cancer were not associated with utilization of mammograms among women. Diagnoses of lung disease, cancer, and stroke were associated with 84 to 157 per 1,000 persons ceasing regular vigorous physical activity, whereas physical activity level did not change significantly after a diagnosis of diabetes and heart disease.

Discussion

The present study adds to the literature by examining changes in multiple health behaviors after the diagnosis of a chronic disease in a population-based panel survey. Two major themes emerged. First, older adults tended to reduce substance use and increase utilization of preventive medical procedures. Previous studies have consistently documented smoking cessation following a chronic disease diagnosis (Falba, 2005; Keenan, 2009; Newsom et al., 2012; Twardella et al., 2006; van Gool, Kempen, Penninx, Deeg, & van Eijk, 2007). Although evidence on drinking is somewhat mixed, several population-based studies reported reduction in alcohol consumption after a chronic disease diagnosis (Newsom et al., 2012; Platt et al., 2010; van Gool et al., 2007). The null findings in drinking outcomes appear to be largely driven by studies focusing on certain types of cancers (Bidstrup et al., 2013; Hawkes et al., 2008; Williams et al., 2013) for which drinking may not be widely perceived as a strong risk factor. Although I did not expect to find physical activity decline after a disease diagnosis, this finding is consistent with previous studies (Hawkes et al., 2008; Newsom et al., 2012; van Gool et al., 2007). Significant barriers to physical activity may arise after the onset of a chronic disease, such as functional limitations, fear of injury, and misconceptions regarding the role of exercise in certain chronic conditions (Rimmer, Riley, Wang, Rauworth, & Jurkowski, 2004). However, the physical activity measures used here did not assess light or moderate forms of activities, for which changes might have occurred.

Second, it appears that changes are the most substantial when people perceive the behavior examined as an important risk factor for a specific disease. These findings are in line with the mechanisms of teachable moment delineated by McBride et al. (2003), who emphasized the importance of the cognitive processes in motivating behavior change. A chronic disease diagnosis is significant enough to be a teachable moment by increasing perceptions of vulnerability, eliciting strong emotional responses, and altering perceived norms or self-concepts. These cognitive responses may differ by disease diagnoses, resulting in different magnitudes of behavior changes across diagnoses. For example, cognitive responses to a cancer diagnosis might be stronger than a diagnosis of diabetes because cancer has been described as the most feared of modern diseases in the mass media (Clarke & Everest, 2006). The effects of diagnosis may differ according to the behavior outcomes due to differences in perceptions of risk and benefits associated with a particular behavior and disease. For example, benefits of smoking cessation in patients with lung disease are widely recognized whereas the relationship between alcohol consumption and exacerbation of lung disease is less clear (Boe, Vandivier, Burnham, & Moss, 2009). Thus, I expect that a chronic lung disease diagnosis will elicit a greater reduction in smoking, whereas its effect on drinking is likely to be much smaller. Similarly, increase in the receipt of cholesterol test was greatest for diabetes and heart disease, which is expected given that dyslipidemia is closely associated with these conditions (O’Brien, Nguyen, & Zimmerman, 1998), and that periodic screening for dyslipidemia is recommended in widely regarded guidelines for managing diabetes and heart disease (American Diabetes Association, 2015).

Considerable changes in several health behaviors occurred after a chronic disease diagnosis, suggesting that the time of a chronic disease diagnosis can be an opportunity for multiple health behavior change (MHBC). MHBC focuses on promoting multiple health behaviors either simultaneously or sequentially within a limited time period to improve efficiency because health risk behaviors often cluster (Prochaska, Spring, & Nigg, 2008). Although many MHBC intervention studies conducted in the past met with limited success (Prochaska et al., 2008), more positive results have been reported in recent literature (Emmons et al., 2014; Green, Hayman, & Cooley, 2015). In a recent systematic review of MHBC interventions for cancer survivors, Green et al. (2015) found that in-person interventions consistently produced changes in diet, exercise, and smoking, and interventions using phone or mail contact with longer durations also showed promising results. This study suggests that MHBC interventions could take advantage of the timing of a chronic disease diagnosis to deliver behavioral interventions to improve effectiveness. Future studies should explicitly examine whether time since a chronic disease diagnosis modifies the impact of behavioral interventions.

Although this study found significant changes in multiple health behaviors, some changes were small and might be of little substantive significance, highlighting the need to implement effective behavioral interventions for older adults with chronic diseases. Under the Affordable Care Act, Medicare established an annual wellness visit program that includes a health risk assessment and a customized wellness or personal prevention plan without cost sharing (Koh & Sebelius, 2010). This study suggests that it may be useful to consider time since diagnosis in the development of personal prevention plan. Targeting middle-aged and older adults with newly diagnosed chronic diseases may augment the effectiveness of health behavior interventions because they may be particularly receptive to change. It may be important for health care providers to pay special attention to health behaviors that may not be perceived as an immediate risk factor for the relevant condition. These older adults can be conveniently recruited at clinics, hospitals, and other health care settings as they attend appointments for their new diagnosis. It might be beneficial to provide ongoing behavioral counseling during the next couple of office visits, because this study suggests that a sizable subset of individuals changed their behaviors within the 2 years after their diagnoses. This is important, as current evidence suggests that adults with chronic diseases are not adequately advised by health care providers to engage in health behaviors (Lobo, Loeb, Ghushchyan, Schauer, & Huebschmann, 2012; Xiang, Hernandez, & Larrison, 2015).

In addition to traditional in-office counseling, the dramatic growth of electronic health record, mobile technology, and online social networks for health care delivery opens up new opportunities for implementing behavioral interventions with older patients. The wide adoption of electronic health record, particularly those that capture patient-reported measures such as health behaviors, enables providers to make informed and efficient decisions about health promotion at point of care (Glasgow, Kaplan, Ockene, Fisher, & Emmons, 2012). Technology-based systems can automatically tailor health messages based on participant characteristics or responses to assessments (Heron & Smyth, 2010), which has been shown to be more effective in helping some patients initiate health behavior changes (Noar, Benac, & Harris, 2007). Mobile technology, in particular, can be switched on and remain with the owner, making it possible for individually tailored interventions to be delivered during “real time” and in “real world” when they are most needed (Heron & Smyth, 2010). Although older adults are less likely to use technology in general and search online for health information than their younger counterparts (Miller & Bell, 2011), studies have shown that older adults are interested and willing to use technology in assisting their disease management (Parker, Jessel, Richardson, & Reid, 2013). Future studies should continue to develop and evaluate technology-based behavioral interventions that are age- and function appropriate for older adults with chronic disease.

At the same time, it is important to consider social support, environmental, and structural factors that influence health behaviors in older life. The recent FrameWorks Institute report on aging pointed that the dominate pattern of public understanding is the deep assumption that individuals are exclusively responsible for how they age without reference to social determinants of successful aging (O’Neil & Haydon, 2015). Behavioral interventions that exclusively focus on individual responsibility without considering social and environmental barriers to healthy lifestyles are likely to have limited success. This may be particularly true for health promotion in older adults with chronic diseases with or at risk of mobility limitation because they are more reliant on the resources in their social support system and environment. Significant gaps remain, and additional research is needed to develop multilevel ecological health promotion approach for older adults with chronic disease (Prohaska et al., 2006).

Limitations

Self-reports of physician diagnosis are prone to recall errors and systematically exclude people with undiagnosed or underdiagnosed illnesses. Self-reports of health behaviors are particularly subject to social desirability bias. Furthermore, the findings should be interpreted as correlations rather than causations. It was not possible to pinpoint the time of behavior change because interviews were conducted every 2 years, and responses were based on recall of events since the last interview. It remains unclear whether it was the diagnosis itself or services received around the time of diagnosis that contributed to behavior change. In addition, intrapersonal variations in responses to a chronic disease diagnosis were not explored in this study. Individual characteristics such as sociodemographics and mental health status and health service factors such as location of care and types of providers may influence changes in health behaviors after a chronic disease diagnosis. Finally, the study sample consisted of individuals who were relatively healthy in their midlife because those with an existing chronic disease diagnosis at baseline were excluded from the analyses.

Conclusion

A new diagnosis may serve as a window of opportunity for multiple health behavioral change simultaneously. If interventions cannot be delivered conveniently at the time of diagnosis, it might be beneficial to provide ongoing health counseling during the next couple of office visits, as this study suggests that a sizable subset of individuals changed behaviors within the 2 years after their diagnoses. In addition, it may be beneficial for health care providers to pay special attention to counsel health risk behaviors that may not be widely perceived as immediate risk factors for a certain disease. Finally, this study highlights the need to develop and implement effective physical activity promotion interventions for older adults with chronic disease.

Footnotes

Acknowledgements

The author would like to thank her dissertation committee (Chris Larrison, PhD, Ruopeng An, PhD, Min Zhan, PhD, and Sarah Gehlert, PhD) for their guidance for her dissertation. She would also like to extend her appreciation to Dr. Allen Heinemann for his feedback on this article.

Author’s Note

This study was developed from the doctoral dissertation of the author (available upon request from ).

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The writing of this article was done while the author was a postdoctoral fellow under a grant from the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR), located in the Administration for Community Living (ACL) in the U.S. Department of Health and Human Services, Grant H133P130013 (PI: Allen Heinemann, PhD). However, those contents do not necessarily represent the policy of the ACL, and we should not assume endorsement by the Federal Government.

References

Akesson

Larsson

S. C.

Discacciati

Wolk

(2014). Low-risk diet and lifestyle habits in the primary prevention of myocardial infarction in men: A population-based prospective cohort study. Journal of the American College of Cardiology, 64, 1299-1306. doi:10.1016/j.jacc.2014.06.1190

American Diabetes Association. (2015). Standards of medical care in diabetes—2015. Diabetes Care, 38(Suppl. 1), S1-S93. doi:10.2337/dc15-S001

Angrist

J. D.

Pischke

J. S.

(2009). Mostly harmless econometrics: An empiricist’s companion. Princeton, NJ: Princeton University Press.

Bidstrup

P. E.

Dalton

S. O.

Christensen

Tjonneland

Larsen

S. B.

Karlsen

. . . Johansen

(2013). Changes in body mass index and alcohol and tobacco consumption among breast cancer survivors and cancer-free women: A prospective study in the Danish Diet, Cancer and Health Cohort. Acta Oncologica (Stockholm, Sweden), 52, 327-335. doi:10.3109/0284186X.2012.746466

Boe

D. M.

Vandivier

R. W.

Burnham

E. L.

Moss

(2009). Alcohol abuse and pulmonary disease. Journal of Leukocyte Biology, 86, 1097-1104. doi:10.1189/jlb.0209087

Centers for Medicare and Medicaid Services. (2012). Chronic conditions among Medicare beneficiaries (Chart book: 2010 ed.). Baltimore, MD: Author.

Chien

Campbell

Hayden

Hurd

Main

Mallett

. . . St.Clair

(2013). RAND HRS data documentation, version M. Retrieved from http://hrsonline.isr.umich.edu/modules/meta/rand/randhrsm/randhrsM.pdf

Clarke

J. N.

Everest

M. M.

(2006). Cancer in the mass print media: Fear, uncertainty and the medical model. Social Science & Medicine, 62, 2591-2600. doi:10.1016/j.socscimed.2005.11.021

Emmons

K. M.

Puleo

Greaney

M. L.

Gillman

M. W.

Bennett

G. G.

Haines

. . . Viswanath

(2014). A randomized comparative effectiveness study of Healthy Directions 2—A multiple risk behavior intervention for primary care. Preventive Medicine, 64, 96-102. doi:10.1016/j.ypmed.2014.03.011

10.

Falba

(2005). Health events and the smoking cessation of middle aged Americans. Journal of Behavioral Medicine, 28, 21-33.

11.

Glasgow

R. E.

Kaplan

R. M.

Ockene

J. K.

Fisher

E. B.

Emmons

K. M.

(2012). Patient-reported measures of psychosocial issues and health behavior should be added to electronic health records. Health Affairs, 31, 497-504. doi:10.1377/hlthaff.2010.1295

12.

Green

A. C.

Hayman

L. L.

Cooley

M. E.

(2015). Multiple health behavior change in adults with or at risk for cancer: A systematic review. American Journal of Health Behavior, 39, 380-394. doi:10.5993/AJHB.39.3.11

13.

Hawkes

A. L.

Lynch

B. M.

Youlden

D. R.

Owen

Aitken

J. F.

(2008). Health behaviors of Australian colorectal cancer survivors, compared with noncancer population controls. Supportive Care in Cancer, 16, 1097-1104. doi:10.1007/s00520-008-0421-5

14.

Sengupta

Velkoff

V. A.

DeBarros

K. A.

(2005). Census bureau, current population reports, 65+ in the United States: 2005. Retrieved from http://www.census.gov/prod/2006pubs/p23-209.pdf

15.

Heron

K. E.

Smyth

J. M.

(2010). Ecological momentary interventions: Incorporating mobile technology into psychological and health behaviour treatments. British Journal of Health Psychology, 15, 1-39. doi:10.1348/135910709X466063

16.

Hochbaum

G. M.

(1958). Public participation in medical screening programs: A socio-psychological study. Washington, DC: U.S. Department of Health, Education, and Welfare, Public Health Service, Bureau of State Services, Division of Special Health Services, Tuberculosis Program.

17.

Karlsen

R. V.

Bidstrup

P. E.

Christensen

Larsen

S. B.

Tjonneland

Dalton

S. O.

Johansen

(2012). Men with cancer change their health behaviour: A prospective study from the Danish Diet, Cancer and Health Study. British Journal of Cancer, 107, 201-206. doi:10.1038/bjc.2012.238

18.

Keenan

P. S.

(2009). Smoking and weight change after new health diagnoses in older adults. Archives of Internal Medicine, 169, 237-242. doi:10.1001/archinternmed.2008.557

19.

Koh

H. K.

Sebelius

K. G.

(2010). Promoting prevention through the Affordable Care Act. The New England Journal of Medicine, 363, 1296-1299. doi:10.1056/NEJMp1008560

20.

Lobo

I. E.

Loeb

D. F.

Ghushchyan

Schauer

I. E.

Huebschmann

A. G.

(2012). Missed opportunities for providing low-fat dietary advice to people with diabetes. Preventing Chronic Disease, 9, Article E161. doi:10.5888/pcd9.120086

21.

Margolis

(2013). Educational differences in healthy behavior changes and adherence among middle-aged Americans. Journal of Health and Social Behavior, 54, 353-368. doi:10.1177/0022146513489312

22.

McBride

C. M.

Emmons

K. M.

Lipkus

I. M.

(2003). Understanding the potential of teachable moments: The case of smoking cessation. Health Education Research, 18, 156-170. doi:10.1093/her/18.2.156

23.

Miller

L. M.

Bell

R. A.

(2011). Online health information seeking: The influence of age, information trustworthiness, and search challenges. Journal of Aging and Health, 24, 525-541. doi:10.1177/0898264311428167

24.

Newsom

J. T.

Huguet

McCarthy

M. J.

Ramage-Morin

Kaplan

M. S.

Bernier

. . . Oderkirk

(2012). Health behavior change following chronic illness in middle and later life. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 67, 279-288. doi:10.1093/geronb/gbr103

25.

Noar

S. M.

Benac

C. N.

Harris

M. S.

(2007). Does tailoring matter? Meta-analytic review of tailored print health behavior change interventions. Psychological Bulletin, 133, 673-693. doi:10.1037/0033-2909.133.4.673

26.

O’Brien

Nguyen

T. T.

Zimmerman

B. R.

(1998). Hyperlipidemia and diabetes mellitus. Mayo Clinic Proceedings, 73, 969-976. doi:10.4065/73.10.969

27.

O’Neil

Haydon

(2015). Aging, agency, and attribution of responsibility: Shifting public discourse about older adults. FrameWorks Institute. Retrieved from http://www.frameworksinstitute.org/assets/files/aging_ffa_final_090215.pdf

28.

Ortman

J. M.

Velkoff

V. A.

Hogan

(2014). An aging nation: The older population in the United States—Population estimates and projections (Current Population Reports No. P25-1140). Retrieved from http://www.census.gov/prod/2014pubs/p25-1140.pdf

29.

Parker

S. J.

Jessel

Richardson

J. E.

Reid

M. C.

(2013). Older adults are mobile too! Identifying the barriers and facilitators to older adults’ use of mHealth for pain management. BMC Geriatrics, 13, Article 43. doi:10.1186/1471-2318-13-43

30.

Platt

Sloan

F. A.

Costanzo

(2010). Alcohol-consumption trajectories and associated characteristics among adults older than age 50. Journal of Studies on Alcohol and Drugs, 71, 169-179.

31.

Prochaska

J. J.

Spring

Nigg

C. R.

(2008). Multiple health behavior change research: An introduction and overview. Preventive Medicine, 46, 181-188. doi:10.1016/j.ypmed.2008.02.001

32.

Prohaska

Belanskly

Belza

Buchner

Marshall

McTigue

. . . Wilcox

(2006). Physical activity, public health, and aging: Critical issues and research priorities. Journal of Gerontology, Series B: Psychological Sciences & Social Sciences, 61(5), S267-S273.

33.

Rejeski

W. J.

E. H.

Bertoni

A. G.

Bray

G. A.

Evans

Gregg

E. W.

. . . Look AHEAD Research Group. (2012). Lifestyle change and mobility in obese adults with type 2 diabetes. The New England Journal of Medicine, 366, 1209-1217. doi:10.1056/NEJMoa1110294

34.

Rimmer

J. H.

Riley

Wang

Rauworth

Jurkowski

(2004). Physical activity participation among persons with disabilities: Barriers and facilitators. American Journal of Preventive Medicine, 26, 419-425. doi:10.1016/j.amepre.2004.02.002

35.

Rosenstock

I. M.

(1974). Historical origins of the health belief model. Health Education Monographs, 2, 328-335.

36.

Sabia

Singh-Manoux

Hagger-Johnson

Cambois

Brunner

E. J.

Kivimaki

(2012). Influence of individual and combined healthy behaviours on successful aging. Canadian Medical Association Journal, 184, 1985-1992. doi:10.1503/cmaj.121080

37.

Schiller

J. S.

Ward

B. W.

Freeman

(2014). Early release of selected estimates based on data from the 2013 National Health Interview Survey. National Center for Health Statistics. Retrieved from http://www.cdc.gov/nchs/data/nhis/earlyrelease/earlyrelease201406.pdf

38.

Schneider

K. L.

Andrews

Hovey

K. M.

Seguin

R. A.

Manini

Lamonte

M. J.

. . . Pagoto

S. L.

(2014). Change in physical activity after a diabetes diagnosis: Opportunity for intervention. Medicine & Science in Sports & Exercise, 46, 84-91. doi:10.1249/MSS.0b013e3182a33010

39.

Stuart

E. A.

(2010). Matching methods for causal inference: A review and a look forward. Statistical Science, 25, 1-21. doi:10.1214/09-STS313

40.

Twardella

Loew

Rothenbacher

Stegmaier

Ziegler

Brenner

(2006). The diagnosis of a smoking-related disease is a prominent trigger for smoking cessation in a retrospective cohort study. Journal of Clinical Epidemiology, 59, 82-89. doi:10.1016/j.jclinepi.2005.05.003

41.

U.S. Department of Agriculture & U.S. Department of Health and Human Services. (2010). Dietary guidelines for Americans (7th ed.). Washington, DC: U.S. Government Printing Office. Retrieved from http://www.health.gov/dietaryguidelines/dga2010/dietaryguidelines2010.pdf

42.

U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion. (2015). Healthy people 2020. Washington, DC: Author. Retrieved from http://www.healthypeople.gov/2020/default

43.

van Gool

C. H.

Kempen

G. I.

Penninx

B. W.

Deeg

D. J.

van Eijk

J. T

. (2007). Chronic disease and lifestyle transitions: Results from the Longitudinal Aging Study Amsterdam. Journal of Aging and Health, 19, 416-438. doi:10.1177/0898264307300189

44.

Williams

Steptoe

Wardle

(2013). Is a cancer diagnosis a trigger for health behaviour change? Findings from a prospective, population-based study. British Journal of Cancer, 108, 2407-2412. doi:10.1038/bjc.2013.254

45.

Xiang

Hernandez

Larrison

C. R.

(2015). Provider advice on exercise and diet among adults with comorbid serious psychological distress and diabetes or diabetes risk factors. The Diabetes Educator, 41, 185-193. doi:10.1177/0145721714567234