A randomised controlled trial of providing personalised cardiovascular risk information to modify health behaviour

Abstract

We conducted a feasibility study of a web-based intervention, which provided personalized cardiovascular disease (CVD) risk information, behavioural risk reduction strategies and educational resources. Participants were block-randomized to the 3-month intervention (n = 47) or to usual care (n = 49). Participants in the intervention group were presented with their CVD risk based on the Framingham risk score, and in three subsequent online encounters could select two behavioural/lifestyle modules, giving them an opportunity to complete six modules over the course of the study. Because it was self-guided, participants had differing levels of engagement with intervention materials. Most intervention group participants (77%, n = 36) completed all modules. After 3 months there were no significant differences between the intervention and usual care groups for systolic blood pressure, body-mass index, CVD risk, smoking cessation or medication non-adherence. The study suggests that modest clinical improvements can be achieved by interventions that are entirely web-administered. However, web-based interventions do not replace the need for human interaction to communicate CVD risk and assist with decision-making.

Introduction

Cardiovascular disease (CVD) is the leading cause of death in the US.¹ Despite widely recommended lifestyle goals and the availability of effective treatment, many people who have developed or are at risk of CVD, fail to meet treatment goals.^2,3 A person’s perceived risk of stroke or heart attack is an important factor in their adherence to CVD risk reduction guidelines.^4,5 However, people often underestimate their CVD susceptibility.^2,6,7

According to the Health Belief Model,⁸ a person’s beliefs about their disease risk increases the likelihood of behaviour change. Providing tailored CVD risk information may cause patients to initiate and maintain behaviour change, thus improving risk reduction. Furthermore, by providing CVD risk information online, users can access information at their convenience.⁹ If proven effective, an Internet-delivered intervention could be scaled up for broad dissemination.

We have conducted a feasibility study of a web-based intervention, which provided personalized CVD risk information, behavioural risk reduction strategies and educational resources.

Methods

Participants were identified through an electronic medical record query. The inclusion criteria were: (1) affiliated primary care clinic patient for at least one year with one or more visits in the previous year; (2) diagnosed with CVD or a CVD-risk equivalent, e.g. diabetes; (3) having at least one modifiable outcome, e.g. hypertension or actively smoking. Patients were excluded if they: (1) had metastatic cancer, dementia, psychosis or end-stage renal disease; (2) lacked Internet access; (3) received nursing services; (4) were unable to read English; (5) were participating in another CVD study or a household member was a participant; (6) received or were a candidate for a heart transplant; (7) were hospitalized for cardiac-related illnesses in the previous three months; (8) or their arm circumference exceeded 50 cm. Potential participants were sent a recruitment letter approximately two weeks before a primary care clinic visit. A research staff member contacted them via telephone and arranged to meet in person and explain the study.

Randomization

After providing consent, participants were block-randomized to the 3-month intervention or to usual care. Randomization assignments were placed in sealed, consecutively numbered envelopes. The staff involved in the randomization were blinded to the block size.

Study design

There was one face-to-face meeting during the study period to obtain consent and provide the initial educational information. CVD risk and treatment adherence outcomes were assessed face-to-face at baseline and after 3-months.^10–18 The study was approved by the appropriate ethics committee.

Intervention group

At baseline, participants in the intervention group were presented with their CVD risk based on the Framingham risk score.¹⁹ All other interactions were conducted electronically.

In the first online encounter, participants used a web-based Framingham risk calculator. They adjusted their own risk scores and indicated areas they were willing to modify. For example, participants could indicate their readiness to change one aspect of their CVD risk (e.g. smoking cessation), but could remain tentative about altering another (e.g. exercise). Tailored educational information was provided based on participants’ readiness to change. For example, if participants were ready to make dietary changes, they were given literature about balancing caloric expenditure and using the glycaemic index to identify healthy food choices and control hunger. This educational material was provided electronically in PDF format and included information about diet (e.g. glycaemic index); physical activity (e.g. moving more) and smoking (e.g. education and no-smoking reminders).

Subsequent follow-up encounters were not scheduled. Patients were able to login at their own convenience. However, patients were sent an email message each month to remind them to login to complete the encounter. If a patient failed to login one week after the initial reminder, then a follow-up email message was sent to the patient. The follow-up message informed the patient that they had two weeks to complete the online encounter and that only seven days remained.

At each online encounter, participants selected two behavioural/lifestyle modules, giving them an opportunity to complete six modules over the course of the study (Table 1). After selecting a module, the subsequent follow-up encounters (online) were used to reinforce the previous interaction and facilitate maintenance or, if necessary, revise health behaviour goals. The modules provided evidence-based recommendations regarding lifestyle behaviours and participants were advised how to achieve their goals with respect to these behaviours. The intervention components were designed to be culturally sensitive. The behavioural modules covered diet, exercise, smoking, alcohol, patient-provider relationships and medication management (Table 2). All these measures have been used in previous clinical trials and are easy to implement, reliable and sensitive to change.^20,21 Each module asked the patient about their current beliefs and health practices. Based on the patient’s responses to a series of questions, there was tailored feedback to reinforce behaviour change. In addition, information on CVD medication management and side effects were provided with each encounter.

Table 1.

Intervention content. Participants selected two modules each month.

Modules	Health behaviours	Goal
Month 1
Medication adherence	Compliance
Diet – heart healthy	Actively trying to improve diet	Maintain healthy eating
Physical activity	Not exercising	Develop plan for regular exercise
Month 2
Medication adherence	Compliance
Diet – low sodium	Actively trying to improve diet	Maintain healthy eating and reduce salt intake
Risk factor knowledge	Blood pressure, smoking status, cholesterol level	Identify risk factors
Month 3
Medication adherence	Compliance
Diet – portion control	Actively trying to improve diet	Maintain healthy eating, reduce salt intake, measure intake
Smoking cessation	Smoker	Stop smoking

Table 2.

Baseline characteristics. Unless otherwise noted, values shown are number (%).

Total (n = 96)	Intervention (n = 47)	Usual care (n = 49)
Mean age, years (SD)	63.1 (12.2)	63.5 (11.6)	62.7 (12.8)
Race
Caucasian	62 (65)	30 (64)	32 (65)
African American	32 (33)	15 (32)	17 (35)
Other	1 (1)	1 (2)	0
Male	32 (33)	14 (30)	18 (37)
Partnered (married or living together)	60 (63)	30 (64)	30 (61)
Completed more than 12 years of school	90 (94)	44 (94)	46 (94)
Low literacy level (REALM score <61)	8 (8)	4 (9)	4 (8)
Inadequate income[a]	15 (16)	7 (15)	8 (16)
Moderate physical activity
Meets recommendations (≥30 min/day for ≥5 days/week)	33 (34)	17 (36)	16 (33)
Some, but insufficient (<30 min/day or <5 days/week)	51 (53)	22 (47)	29 (59)
None	9 (9)	6 (13)	3 (6)
Vigorous physical activity
Meets recommendations (≥20 min/day for ≥3 days/week)	22 (23)	12 (26)	10 (20)
Some, but insufficient (<20 min/day or <3 days/week)	23 (24)	11 (23)	12 (25)
None	50 (52.1)	23 (49)	27 (55)
Body Mass Index, kg/m² (SD)	31.9 (7.7)	31.5 (8.8)	32.2 (6.4)
Current smoker	16 (17)	8 (17)	8 (16)
Medication non-adherence	44 (46)	22 (47)	22 (45)
Has diabetes	28 (29)	17 (36)	11 (22)
Has hyperlipidaemia	75 (78)	39 (83)	36 (74)
Treated for high systolic blood pressure	82 (85)	41 (87)	41 (84)

[a] Inadequate income was defined as a participant reporting: (1) difficulty paying bills no matter what was done; or (2) having money to pay the bills only because they cut back on things

Control group

Participants in the usual care group received general, printed educational CVD information and additional information at their providers’ discretion. If requested, participants were given intervention materials at the study’s conclusion. All material provided was at a sixth grade reading level.

Measures

We compared predicted mean differences in clinical outcomes between the intervention group and the usual care group after three months. The outcomes were the 10-year CVD risk Framingham score, body mass index (BMI), smoking status, systolic blood pressure (SBP) and self-reported medication adherence.

Sociodemographic factors were assessed at baseline. Health literacy was evaluated using the Rapid Estimate of Adult Literacy in Medicine (REALM) test.^22,23 Physical activity was measured according to adherence with recommendations for both moderate and vigorous activity.²⁴ Medication adherence was evaluated using the four-item Morisky self-report scale.^25,26

Statistical analyses

We analysed sociodemographic variables using descriptive statistics. We used mixed linear models to test between-group differences in predicted means at 3-months. For continuous outcomes we used general linear mixed models and for categorical variables we used generalized estimating equations. Because of baseline differences in participant characteristics, models were constrained to make the groups similar for analysis. Unconstrained models were initially run for each measure and the mean baseline measure was compared between the intervention and control groups. When non-significant, the models were then constrained to make both groups equal at baseline (i.e. they had the same intercept). We calculated effect sizes using the Cox index (binary variables) and Cohen’s d (continuous variables).^27,28 We established thresholds for clinically important effect sizes of moderate (0.50–0.79) and large (>0.79). We conducted a sensitivity analysis comparing patients in the two groups who completed all three monthly encounters.

Analyses were conducted using a standard package (SAS version 9.2, SAS Institute Inc., Cary, NC, USA). The analysis was based on intention to treat and included all intervention-group participants.

Results

There were 96 participants, of whom 47 were randomized to the intervention and 49 to usual care (Table 2). The mean age was 63 years. Most participants were obese, identified as White (65%), female (67%), partnered (63%) and had completed a high school education (94%). Fewer than half (46%) reported medication non-adherence. Approximately one-third (29%) had diabetes and most had a diagnosis of hyperlipidaemia (78%) or hypertension (85%). At baseline, approximately 17% of participants were current smokers.

Because it was self-guided, participants had differing levels of engagement with intervention materials. Most intervention group participants (77%, n = 36) completed all modules. Some participants completed only one (11%, n = 5) or two (4%, n = 2) modules. However, 9% (n = 4) failed to log into the web system even once. In the sensitivity analyses, no significant differences in clinical outcomes were identified between participants in the intervention group who completed three modules versus those in the usual care group.

The predicted mean difference between groups at 3-months was 0.5 and −0.1 mm Hg for systolic and diastolic BP, respectively (Table 3). None of the predicted mean differences were significant.

Table 3.

Predicted outcomes at baseline and at 3 months.

Intervention group Predicted mean (SE)	Usual care group Predicted mean (SE)	Between-group predicted mean difference (95% CI)	P-value
Systolic BP (mm Hg)[a]
Baseline	128.2 (1.7)	128.2 (1.7)	0.5 (−5.4, 6.4)	0.87
Month 3	125.1 (2.3)	124.6 (2.2)	0.5 (−5.4, 6.4)	0.87
Diastolic BP (mm Hg)[a]
Baseline	75.4 (1.2)	75.4 (1.2)	−0.1 (−4.1, 3.9)	0.97
Month 3	73.4 (1.6)	73.5 (1.5)	−0.1 (−4.1, 3.9)	0.97
Body Mass Index (kg/m²)
Baseline	31.9 (0.8)	31.9 (0.8)	0.1 (−0.4, 0.7)	0.60
Month 3	32.1 (0.8)	31.9 (0.8)	0.1 (−0.4, 0.7)	0.60
CVD risk (10-year risk)
Baseline	0.3 (0.0)	0.3 (0.0)	−0.0 (−0.0, 0.0)	0.81
Month 3	0.2 (0.0)	0.3 (0.0)
	Predicted probability (95% CI)	Predicted probability (95% CI)	Odds ratio (95% CI)	P-value
Current smoker (0 = no, 1 = yes)
Baseline	0.17 (0.10, 0.26)	0.17 (0.10, 0.26)	0.84 (0.60, 1.17)	0.30
Month 3	0.14 (0.08, 0.24)	0.17 (0.10, 0.25)	0.84 (0.60, 1.17)	0.30
Medication non-adherence (1 = non-adherent, 0 = adherent)
Baseline	0.48 (0.38, 0.58)	0.48 (0.38, 0.58)	0.79 (0.38, 1.62)	0.51
Month 3	0.43 (0.30, 0.57)	0.49 (0.36, 0.63)	0.79 (0.38, 1.62)	0.51

[a] average of three blood pressure values

Discussion

The intervention was completely self-guided and did not involve social support from friends, family or online social media, or include human interaction. While a web-based forum is convenient for patients, it may be less effective without the guidance and accountability from clinician interaction. The minor clinical improvements observed in the study might have been greater if there had been interaction with a clinician. In the comments made after the study, participants expressed a desire for contact with someone associated with the intervention about CVD.

The success of e-health and web-based interventions varies and there have been few formal evaluations which have elucidated the reasons why some interventions are more effective than others.^29,30 Evidence suggests that web-administered interventions are more effective than usual care alone, but that multi-component interventions involving both web- and non-web-based approaches are superior.²⁹ It has been suggested that human interaction facilitates patient engagement and can overcome attrition.³¹

In order to develop successful e-health interventions, Morrison and colleagues suggested that four interactive design feature must be considered. These were social context and support, contact with the intervention, tailoring and self-management.³⁰ Our study did not evaluate the participants’ perception of the intervention’s design.

Web-based interventions appear most effective for people who are already motivated for the targeted behaviour.²⁹ Our objective was to convey CVD risk information to a population who may not have identified CVD as a personal risk. Lessons learned from this study may therefore be useful to future web-administered projects involving either self-guided or clinician-driven modules.

Modified intervention materials based on the present project are under development in several health care systems. A nurse-administered version of the risk information and self-management intervention is being implemented in a pilot study in the UK. A version of the program is being used in North Carolina Medicaid populations. Implementation of the program in three Veterans Affairs Medical Centers is being considered. The inclusion of human interaction, in addition to web-administered education, is expected to motivate participants to engage in self-management.

The present study suggests that modest clinical improvements can be achieved by interventions that are entirely web-administered. However to affect meaningful behavioural change, future studies should include interaction with a pharmacist or other clinician, perhaps by online messaging, email reminders or motivational messages, to communicate CVD risk factors and facilitate informed decision-making. Web-based interventions do not replace the need for human interaction to communicate CVD risk and assist with decision-making.

Footnotes

Acknowledgements

The study was conducted with financial support from the Informed Medical Decisions Foundation (formerly known as the Foundation for Informed Medical Decision Making), grant number 0170-1.

References

US Burden of Disease Collaborators. The state of US health, 1990-2010: burden of diseases, injuries, and risk factors. JAMA 2013; 310: 591–608.

Alwan

William

Viswanathan

Paccaud

Bovet

. Perception of cardiovascular risk and comparison with actual cardiovascular risk. Eur J Cardiovasc Prev Rehabil 2009; 16: 556–61.

Bosworth

. How can innovative uses of technology be harnessed to improve medication adherence? Expert Rev Pharmacoecon Outcomes Res 2012; 12: 133–5.

Mosca

Mochari

Christian

. National study of women’s awareness, preventive action, and barriers to cardiovascular health. Circulation 2006; 113: 525–34.

Won

Hong

Hwang

. Actual cardiovascular disease risk and related factors: a cross-sectional study of Korean blue collar workers employed by small businesses. Workplace Health Saf 2013; 61: 163–71.

Gholizadeh

Davidson

Salamonson

Worrall-Carter

. Theoretical considerations in reducing risk for cardiovascular disease: implications for nursing practice. J Clin Nurs 2010; 19: 2137–45.

Kling

Miller

Mankad

. Go Red for Women cardiovascular health-screening evaluation: the dichotomy between awareness and perception of cardiovascular risk in the community. J Womens Health (Larchmt) 2013; 22: 210–8.

Janz

. The Health Belief Model in understanding cardiovascular risk factor reduction behaviors. Cardiovasc Nurs 1988; 24: 39–41.

Griffiths

Lindenmeyer

Powell

Lowe

Thorogood

. Why are health care interventions delivered over the internet? A systematic review of the published literature. J Med Internet Res 2006; 8: e10–e10.

10.

Charles

Good

Hanusa

Chang

Whittle

. Racial differences in adherence to cardiac medications. J Natl Med Assoc 2003; 95: 17–27.

11.

Gazmararian

Kripalani

Miller

Echt

Ren

Rask

. Factors associated with medication refill adherence in cardiovascular-related diseases: a focus on health literacy. J Gen Intern Med 2006; 21: 1215–21.

12.

Gehi

Haas

Pipkin

Whooley

. Depression and medication adherence in outpatients with coronary heart disease: findings from the Heart and Soul Study. Arch Intern Med 2005; 165: 2508–13.

13.

Mochari

Ferris

Adigopula

Henry

Mosca

. Cardiovascular disease knowledge, medication adherence, and barriers to preventive action in a minority population. Prev Cardiol 2007; 10: 190–5.

14.

World Health Organization. Adherence to Long-Term Therapy: Evidence for Action. See http://www.who.int/chp/knowledge/publications/adherence_report/en/ (last checked 27 December 2013).

15.

Burt

Paulose-Ram

Dillon

. Gender differences in hypertension treatment, drug utilization patterns, and blood pressure control among US adults with hypertension: data from the National Health and Nutrition Examination Survey 1999-2004. Am J Hypertens 2008; 21: 789–98.

16.

Park

Kim

Jang

Koh

. Predictors of adherence to medication in older Korean patients with hypertension. Eur J Cardiovasc Nurs 2013; 12: 17–24.

17.

Trivedi

Ayotte

Edelman

Bosworth

. The association of emotional well-being and marital status with treatment adherence among patients with hypertension. J Behav Med 2008; 31: 489–97.

18.

Bryson

Rumsfeld

. Medication adherence: its importance in cardiovascular outcomes. Circulation 2009; 119: 3028–35.

19.

D’Agostino

Sr Vasan

Pencina

. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation 2008; 117: 743–53.

20.

Bosworth

Olsen

McCant

. Hypertension Intervention Nurse Telemedicine Study (HINTS): testing a multifactorial tailored behavioral/educational and a medication management intervention for blood pressure control. Am Heart J 2007; 153: 918–24.

21.

Bosworth

Olsen

Goldstein

. The veterans’ study to improve the control of hypertension (V-STITCH): design and methodology. Contemp Clin Trials 2005; 26: 155–68.

22.

DeWalt

Hink

. Health literacy and child health outcomes: a systematic review of the literature. Pediatrics 2009; 124(Suppl. 3): S265–74.

23.

Dewalt

Berkman

Sheridan

Lohr

Pignone

. Literacy and health outcomes: a systematic review of the literature. J Gen Intern Med 2004; 19: 1228–39.

24.

Stewart

Mills

King

Haskell

Gillis

Ritter

. CHAMPS physical activity questionnaire for older adults: outcomes for interventions. Med Sci Sports Exerc 2001; 33: 1126–41.

25.

Morisky

Ang

Krousel-Wood

Ward

. Predictive validity of a medication adherence measure in an outpatient setting. J Clin Hypertens (Greenwich) 2008; 10: 348–54.

26.

Morisky

Green

Levine

. Concurrent and predictive validity of a self-reported measure of medication adherence. Med Care 1986; 24: 67–74.

27.

Sánchez-Meca

Marín-Martínez

Chacón-Moscoso

. Effect-size indices for dichotomized outcomes in meta-analysis. Psychol Methods 2003; 8: 448–67.

28.

Cohen

. Statistical Power Analysis for the Behavioral Sciences, 2nd edn. New York: Lawrence Erlbaum Associates, 1988.

29.

Hutton

Wilson

Apelberg

. A systematic review of randomized controlled trials: web-based interventions for smoking cessation among adolescents, college students, and adults. Nicotine Tob Res 2011; 13: 227–38.

30.

Morrison

Yardley

Powell

Michie

. What design features are used in effective e-health interventions? A review using techniques from Critical Interpretive Synthesis. Telemed J E Health 2012; 18: 137–44.

31.

Rosser

Vowles

Keogh

Eccleston

Mountain

. Technologically-assisted behaviour change: a systematic review of studies of novel technologies for the management of chronic illness. J Telemed Telecare 2009; 15: 327–38.