This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Web-based interventions can improve single cardiovascular risk factors in adult populations. In view of global aging and the associated increasing burden of cardiovascular disease, older people form an important target population as well.
In this systematic review and meta-analysis, we evaluated whether Web-based interventions for cardiovascular risk factor management reduce the risk of cardiovascular disease in older people.
Embase, Medline, Cochrane and CINAHL were systematically searched from January 1995 to November 2014. Search terms included cardiovascular risk factors and diseases (specified), Web-based interventions (and synonyms) and randomized controlled trial. Two authors independently performed study selection, data-extraction and risk of bias assessment. In a meta-analysis, outcomes regarding treatment effects on cardiovascular risk factors (blood pressure, glycated hemoglobin A1c (HbA1C), low-density lipoprotein (LDL) cholesterol, smoking status, weight and physical inactivity) and incident cardiovascular disease were pooled with random effects models.
A total of 57 studies (N=19,862) fulfilled eligibility criteria and 47 studies contributed to the meta-analysis. A significant reduction in systolic blood pressure (mean difference –2.66 mmHg, 95% CI –3.81 to –1.52), diastolic blood pressure (mean difference –1.26 mmHg, 95% CI –1.92 to –0.60), HbA1c level (mean difference –0.13%, 95% CI –0.22 to –0.05), LDL cholesterol level (mean difference –2.18 mg/dL, 95% CI –3.96 to –0.41), weight (mean difference –1.34 kg, 95% CI –1.91 to –0.77), and an increase of physical activity (standardized mean difference 0.25, 95% CI 0.10-0.39) in the Web-based intervention group was found. The observed effects were more pronounced in studies with short (<12 months) follow-up and studies that combined the Internet application with human support (blended care). No difference in incident cardiovascular disease was found between groups (6 studies).
Web-based interventions have the potential to improve the cardiovascular risk profile of older people, but the effects are modest and decline with time. Currently, there is insufficient evidence for an effect on incident cardiovascular disease. A focus on long-term effects, clinical endpoints, and strategies to increase sustainability of treatment effects is recommended for future studies.
The field of eHealth is expanding the potential of contemporary medicine [
Older people form an important target population because cardiovascular risk reduction appears effective until old age [
In this systematic review and meta-analysis, we aim to answer the question whether Web-based interventions for cardiovascular risk factor management reduce cardiovascular risk and disease in older people.
We performed a systematic literature search for randomized controlled trials (RCT) on Web-based interventions in older people targeting one or more cardiovascular risk factors and/or disease. Methods were predefined in a research protocol using the PRISMA checklist and the Systematic Reviews Guidelines of the Center of Reviews and Dissemination (
Main outcomes of interest were incident cardiovascular disease (myocardial infarction, angina pectoris, heart failure, stroke or transient ischemic attack, and peripheral arterial disease), cardiovascular mortality and overall mortality, and changes in cardiovascular risk factors including blood pressure (BP), glycated hemoglobin A1c (HbA1c), low-density lipoprotein (LDL) cholesterol, smoking status, weight, level of physical exercise, or a composite cardiovascular risk score.
We performed a comprehensive literature search in the EMBASE, Medline, CINAHL, and Cochrane databases from 1995 onward (because the Internet was not widely available before then). Key search terms were cardiovascular risk factors and diseases (separate diseases and risk factors specified), terms related to aspects of cardiovascular risk management (eg, diet, exercise, BP control), Web-based interventions (including all definitions and synonyms), and RCT/review/meta-analysis. The search was last updated on November 3, 2014 by CRB. The comprehensive search strategy is provided in
Two reviewers (BS and CRB) extracted data using a predefined data extraction form (
For categorical variables, we calculated odds ratios with 95% confidence intervals. We estimated pooled odds ratios with Mantel-Haenszel random-effects models. For continuous outcomes, mean differences or standardized mean differences (Hedges’
For studies with multiple arms, we included only one intervention arm in the meta-analysis in order not to create “unit-of-analysis” error by double counting the control group. Where possible, we selected the Internet-only intervention arm. No data were imputed.
We estimated pooled effects for all single cardiovascular risk factors. To address the overall question of efficacy of Web-based interventions for cardiovascular risk factor management, we evaluated the effect on cardiovascular composite scores, clinical outcomes (cardiovascular morbidity and mortality), and pooled the standardized primary outcomes of all studies. We used the primary outcomes as defined by the authors of the studies.
Funnel plots were inspected to assess for potential publication bias. Statistical heterogeneity was assessed using Q and I2 tests. We explored reasons for heterogeneity by jackknife analysis and subgroup analyses. We assessed the following factors in subgroup analyses: study duration (predefined, short term [<12 months] versus long term [≥12 months]), type of cardiovascular prevention (primary versus secondary) [
The search yielded 5251 papers after removal of duplicates. We did not identify additional studies by searching reference lists. After screening of titles and abstracts, 462 papers remained. Review of these full texts resulted in 57 RCTs (corresponding with 84 papers) that fulfilled the selection criteria and were included in the systematic review. We contacted 16 authors to request additional data: nine authors responded and three authors complied with our request. Out of this final selection, 47 studies could be included in the meta-analysis (see
Prisma flowchart illustrating literature search.
The 57 RCTs included 19,862 individuals (
Methodological quality of the included studies varied (
Of the 57 studies included in the systemic review, 47 studies [
The pooled analysis showed a significant reduction in both systolic and diastolic BP favoring the intervention (26 studies; n=7720;
Characteristics of the studies included for the systematic review: interventions targeting diabetes.a
Study | Setting and study length | Participants | Age (years), mean (SD) | Sex (% female) | Intervention | Control | Primary; secondary outcomes |
Bond 2010 [ |
2-arm RCT; USA; 6 m | 62 people with DM via university/veteran clinic | 67.2 (6.0) | 45 | Website: education, self-monitoring (glucose, exercise, weight, BP, medication), forum; nurse support (email, chat) | Standard diabetes care | HbA1c, BP, weight, total cholesterol, HDL cholesterol |
IDEATEL 2000-2010 [ |
2-arm RCT; USA; 60 m | 1665 Medicare recipients with DM | 70.9 (6.7) | 63 | Online home telemedicine unit: nurse support (video chat), Web portal for self-monitoring (glucose, BP), education | Standard diabetes care | HbA1c, systolic BP, diastolic BP, total cholesterol, LDL cholesterol |
D-net 2001 [ |
4-arm RCT; USA; 10 m | 320 people with DM2, Internet, from 16 GPs | 59 (9.2) | 53 | Website: (1) Self-management (glucose), coach support; (2) education, forum; (3) 1 and 2 combinedb | (4) Information on medical and lifestyle aspects of diabetes | Not defined; behavioral, biological, and psychosocial outcomes |
My path 2010 [ |
3-arm RCT; USA; 12 m | 463 Medicare recipients with DM2, BMI ≥25 kg/m2 or ≥1 CV risk factor, Internet | 58.4 (9.2) | 50 | (1) Website for computer-assisted self-management(CASM): goal setting, monitoring (HbA1c, BP, cholesterol), forum, education;b,c (2) CASM+ social support (coach, group sessions)b,c | (3) Computer-based health risk appraisal, no key features of CASM | Behavior changes in diet, physical activity, medication adherence |
My care team 2005 [ |
2-arm RCT; USA; 12 m | 104 people with DM, HbA1c ≥9.0% via veteran clinic | 63.5 (7.0) | 0.5 | Website: self-management (glucose, BP), education, reminders (phone); care manager support | DM self-management training, usual care | HbA1c and BP at 3, 6, 9, and 12 m |
Mobile DM 2011 [ |
4-arm cluster RCT; USA; 12 m | 26 physician practices with 163 people with DM and HbA1c ≥7.5% | 52.8 (8.1) | 50 | (2) Self-management via website + mobile phone, patient informs doctor;b (3) 2 + doctor access to data; (4) 3 + advice from doctorc | (1) Care as usual | Change in HbA1c over 1 year |
Avdal 2011 [ |
2-arm RCT; Turkey; 6 m | 122 people with DM2, Internet from clinic | 51 (7.3) | 51 | Website: review risk profile, messaging to researcher, daily glucose monitoring | Education and usual care | HbA1c, attendance rates at outpatient clinic |
Cho 2006 [ |
2-arm RCT; South Korea; 30 m | 80 people with DM, Internet from clinic | 53 (9) | 39 | Website: monitoring (glucose, medication, BP, weight, lifestyle), nurse feedback, medication alterations | Conventional note-keeping record system | HbA1c and HbA1c fluctuation index |
Lorig 2010 [ |
3-arm RCT; USA; 6 m | 761 people with DM2, Internet | 54.3 (9.9) | 73 | Self-management website with peer support: lessons, action plans, bulletin board, messaging | Care as usual | HbA1C level at 6 and 18 months |
Grant 2008 [ |
2-arm cluster RCT; USA; 12 m | 244 people with DM, HbA1c >7.0% from 11 primary clinics | 56.1 (11.6) | 49 | Online personal health record: education, diabetes care plan, agenda, messaging, prescription refills | Access to general website Patient Gateway | Changes in HbA1c, BP, and LDL cholesterol |
McMahon 2012 [ |
3-arm RCT; USA; 12 m | 151 people with DM, HbA1c>8.5% from veteran health services | 60.2 (10.8) | 5 | (1) Self-monitoring via phone (BP, glucose); (2) website: self-monitoring (BP, glucose), education, support by care managersb,c | (3) Website with links to other DM websites; usual care | Change in HbA1c and BP over time |
Ralston 2009 [ |
2-arm RCT; USA; 12 m | 83 people with DM2, HbA1c≥7.0% and Internet from clinic: 65% with 2 CV risk factors | 57.3 (—) | 52 | Electronic medical record: self-monitoring (glucose, exercise, diet, medication), support by care manager, usual care visits | Usual care visits | Change in HbA1c |
Kwon 2004 [ |
2-arm RCT; South Korea; 3 m | 110 people with DM2, Internet from clinic: 27% hypertension | 54.1 (9.1) | 33 | Website: self-monitoring (glucose), reminders, professor/nurse/dietician-support | Monthly visit to diabetes specialist | HbA1c |
EMPOWER-D 2013 [ |
2-arm RCT; USA; 12 m | 415 people with DM and HbA1c ≥7.5% from clinic | 53.7 (10.2) | 40 | Online health record: risk estimation, self-monitoring (glucose, diet, exercise, BP), nurse support, own doctor informed | Usual care | HbA1c at 12 m |
REDEEM 2013 [ |
3-arm RCT; USA; 12 m | 392 people with DM2, Internet from community centers | 56.1 (9.6) | 54 | (1) CASM website: goal setting; self-monitoring (HbA1c, BP, cholesterol); 8 phone calls;b (2) Computer-assisted self-management + problem solving treatment (CASP): CASM + 8 sessions problem solving | Computer health risk appraisal, education, same phone calls as intervention | Diabetes distress; HbA1c, physical activity, medication compliance |
a Abbreviations: BP: blood pressure; CASM: computer-assisted self-management; CASP: computer-assisted self-management + problem solving treatment; CV: cardiovascular; DM: diabetes mellitus; DM2: type 2 diabetes mellitus; GP: general practitioner; HbA1c; glycated hemoglobin A1c; HDL: high-density lipoprotein; LDL: low-density lipoprotein.
b For studies with more than 2 arms, this arm was used for all analyses.
c For studies with more than 2 arms, this arm was used for the subgroup analysis on blended interventions.
Effect on systolic blood pressure (26 studies).
Characteristics of the studies included for the systematic review: interventions targeting blood pressure.a
Study | Setting and study length | Participants | Age (years), mean (SD) | Sex (% female) | Intervention | Control | Primary; secondary outcomes |
e-BP 2008 [ |
3-arm RCT; USA; 12 m | 778 people with Internet, hypertension, from GPs: 61.1% obese | 59.1 (8.5) | 52 | (1) Website: BP self-monitoring;b (2) 1 + pharmacist supportc | General website: personal medical record | Change in diastolic, systolic and mean BP |
Nolan 2012 [ |
2-arm RCT; Canada; 4 m | 387 people with hypertension via website: 41% obese | 56.5 (7.4) | 59 | BP action plan website: assessing motivational readiness, advice, feedback, education | E-newsletters | Change in diastolic and systolic BP, and pulse pressure |
Bove 2013 [ |
2-arm RCT; USA; 6 m | 241 people with elevated BP from 2 clinics | 59.6 (13.6) | 65 | Website + telephone system: education, self-monitoring (BP, weight, exercise), online nurse support, doctor informed | Provision of data from initial assessment, usual care | Proportion of participants with controlled BP at 6 m |
Madsen 2008 [ |
2-arm RCT; Denmark; 6 m | 236 people with hypertension from 10 GPs | 55.9 (11.7) | 50 | Website: self-monitoring (BP), feedback from own doctor by email | Usual care | Change in ambulatory systolic BP -at 6 m |
Magid 2013 [ |
2-arm RCT; USA; 6 m | 348 people with hypertension from 10 clinics | 60 (11) | 40 | Written educational material, website: self-monitoring (BP), pharmacist support, doctor informed, reminders | Written education material, usual care | Proportion of participants with controlled BP at 6 m |
McKinstry 2013 [ |
2-arm RCT; Scotland; 6 m | 401 people with hypertension from 20 GPs | 60.7 (11.2) | 40 | Telemonitoring unit + website: self-monitoring (BP), feedback from own doctor | Usual care | Mean ambulatory BP at 6 m |
Thiboutot 2013 [ |
2-arm cluster RCT; USA; 12 m | 500 patients with elevated BP from 54 GPs | 60.5 (11.9) | 58 | Website: self-monitoring (BP, medication), feedback, reminders | Different prevention website (eg, breast screening) | BP control at 12 m |
a Abbreviations: BP: blood pressure; GP: general practitioner.
b For studies with more than 2 arms, this arm was used for all analyses.
c For studies with more than 2 arms, this arm was used for the subgroup analysis on blended interventions.
Characteristics of the studies included for the systematic review: interventions targeting weight loss and weight loss maintenance.a
Study | Setting and study length | Participants | Age (years), mean (SD) | Sex (% female) | Intervention | Control | Primary; secondary outcomes | |
|
|
|
|
|
|
|
|
|
|
Appel 2011 [ |
3-arm RCT; USA; 24 m | 415 people with obesity, ≥1 CV risk factor, Internet from 6 primary clinics | 54 (10.2) | 64 | (1) Website + mobile coach support: education, self-monitoring (weight, diet, exercise), reminders, doctor informed;b,c (2) 1 + in-person support | 1 (or 2) meetings with coach; brochure with websites for weight loss | Change in weight from baseline to 24 m |
|
Bennett 2012 [ |
2-arm RCT; USA; 24 m | 365 obese people with hypertension from 3 clinics | 54.6 (10.9) | 69 | Website/interactive voice response system: self-monitoring weight, setting, coach support (phone), group sessions, education | Self-help booklet | Change in weight at 24 m |
|
Bennett 2010 [ |
2-arm RCT; USA; 3 m | 101 obese people with hypertension, Internet from clinic | 54.4 (8.1) | 48 | Website: goal setting, self-monitoring, behavioral skills education, forum, coach support (online, phone, face-to-face) | Folder on healthy weight, usual care | Change in weight at 12 weeks |
|
Kraschnewsky 2011 [ |
2-arm RCT; USA; 3 m | 100 overweight people, Internet via flyers/Internet | 50.3 (10.9) | 70 | Website: target body weight, monitoring, behavioral tips, videos, weight loss plan, tailored feedback, reminders | Wait list, people got access to website after 12 weeks | Weight loss |
|
Webberd 2008 [ |
2-arm RCT; USA; 4 m | 66 women, BMI 25-40, Internet from advertisements | 50.0 (9.9) | 100 | Website: weight loss tips, lessons, message board, self-monitoring (weight, diet), chat sessions | All features of intervention except for online chat sessions | Not defined; weight, BMI, diet, exercise |
|
E-LITE 2013 [ |
3-arm RCT; USA; 15 m | 241 people with a BMI ≥25, metabolic syndrome from 1 clinic | 52.9 (10.6) | 47 | (1) Website + 12 lifestyle classes;c (2) website: self-monitoring (weight, exercise), messaging, DVD with lifestyle classesb | Usual care | Change in BMI from baseline to 15 m |
|
POWER 2014 [ |
4-arm RCT; UK; 12 m | 179 people with BMI ≥30 kg/m2 or ≥28 kg/m2 + CV risk factors from 5 GPs | 51.2 (13.1) | 66 | (1) Website: 12 self-management sessions monitoring (weight), nurse support (email);b,c (2) 1 + 3 nurse contacts; (3) 1 + 7 nurse contacts | Usual care | Weight at 12 m |
|
|
|
|
|
|
|
|
|
|
Stop Regain 2008 [ |
3-arm RCT; USA; 18 m | 314 people with 10% weight loss in 2 years, via advertisements | 51 (10) | 81 | (1) Website: self-monitoring, email counseling, experts chat;b (2) face-to-face: self-monitoring via phone, weekly group sessions | (3) Newsletters | Weight gain at 18 m |
|
WLM 2008 [ |
2-phase 3-arm RCT; USA; 30 m | 1032 people with ≥4 kg previous weight loss, hypertension, Internet via university/ medicare | 55.6 (8.7) | 63 | (1) Website: goal setting, action plans, self-monitoring (weight, PA, diet), education, bulletin board, reminders, support (email/phone);b (2) personal contact (phone +face-to-face) | Printed lifestyle guidelines, 1 visit with coach | Change in weight |
a Abbreviations: BMI: body mass index; CV: cardiovascular; GP: general practitioner; PA: physical activity.
b For studies with more than 2 arms, this arm was used for all analyses.
c For studies with more than 2 arms, this arm was used for the subgroup analysis on blended interventions.
d Control arm consists of same interactive Internet platform as intervention arm.
Characteristics of the studies included for the systematic review: interventions targeting physical activity and cholesterol.a
Study | Setting and study length | Participants | Age (years), mean (SD) | Sex (% female) | Intervention | Control | Primary; secondary outcomes | |
|
|
|
|
|
|
|
|
|
|
Richardsonb 2010 [ |
2-arm RCT; USA; 4 m | 324 patients from clinic: 12% CHD, 20% DM2, 62% BMI >30 | 52.0 (11.4) | 65 | Website as control + online community forum | Website: pedometer, tailored feedback | Change in average daily step count, patient attrition |
|
Reid 2011 [ |
2-arm RCT; Canada; 12 m | 223 patients with a recent CHD event, Internet via 2 cardiac centers | 56.4 (9.0) | 16 | Website: tutorials, exercise plans, self-monitoring, specialist support | Usual care, education booklet | Mean steps per day |
|
Ferney 2009 [ |
2-arm RCT; Australia; 6 m | 106 inactive residents: 58% overweight | 52.0 (4.6) | 72 | Website: behavioral strategies, goal setting, self-monitoring, advice, bulletin board, news | Website with minimal interactivity | Not defined; physical activity, website use |
|
Active after 55 2013 [ |
2-arm RCT; USA; 3 m | 405 sedentary people with Internet via senior centers/websites | 60.3 (4.9) | 69 | Website: education, goal setting, exercise planning, 11 online exercise lessons, self-monitoring, reminders | No access to the intervention | Not defined; physical activity, BMI |
|
HEART 2014 [ |
2-arm RCT; New Zealand; 6 m | 171 people with stable CHD, Internet from 2 hospitals | 60.2 (9.2) | 19 | Exercise prescription, behavioral strategies, Website: videos, self-monitoring (exercise), education, reminders | Usual care | Change in peak oxygen uptake from baseline to 6 m |
|
Philips Direct Life 2013 [ |
2-arm RCT; Netherlands; 3 m | 235 inactive people with Internet through local media | 64.8 (2.9) | 41 | Website: goal setting, self-monitoring (exercise), e-coach feedback | Waitlist control | Change in physical activity |
|
Suboc 2014 [ |
3-arm RCT; USA; 3 m | 114 sedentary people through media and Internet | 63.0 (7.0) | 34 | (1) Pedometer; (2) website + pedometer: exercise strategies, goal setting, self-monitoring (exercise) feedback, forumc | No intervention | Endothelial function; vascular stiffness, step count, exercise |
|
Peels 2013 [ |
5-arm cluster RCT; Netherlands; 12 m | 2140 people from 6 municipal regions, ±50% overweight | 63.2 (8.4) | 51 | (1) Printed feedback report; (2) 1 + local exercise tips; (3) Web-based feedback report; (4) 3 + local exercise tipsc | Waitlist control | Physical activity |
|
|
|
|
|
|
|
|
|
|
Blochb 2006 [ |
3-arm RCT; USA; 6 m | 171 employees with increased cholesterol, DM or CHD | 54.8 (9.4) | — | (1) Website + financial reward; (2) website + 4 classes, nurse support (phone) | Website, 10-year CVD score, monitoring, goals, tailored info | LDL cholesterol change at 6 m |
|
Live well 2013 [ |
2-arm RCT; USA; 3 m | 61 people with LDL cholesterol ≥3.37 mmol/L, Internet from primary clinics | 52.0 (12.8) | 75 | Web-based rate-your-plate assessment, written educational material, Website: goal setting, self-monitoring, reminders | Web-based rate-your-plate assessment | Not defined; cholesterol, weight, Framingham risk score |
a Abbreviations: BMI: body mass index; CHD: coronary heart disease; CVD: cardiovascular disease; DM: diabetes mellitus; DM2: type 2 diabetes mellitus; LDL: low-density lipoprotein.
b Control arm consists of same interactive Internet platform as intervention arm.
c For studies with more than 2 arms, this arm was used for all analyses.
Characteristics of the studies included for the systematic review: interventions targeting multiple risk factors.a
Study | Setting and study length | Participants | Age (years), mean (SD) | Sex (% female) | Intervention | Control | Primary; secondary outcomes |
Lindsay 2008 [ |
2-arm RCT; UK; 6 m | 108 heart patients living in deprived areas | 62.9 (6.0) | 33 | eHealth portal: glossary, education, local community links, discussion forum | No access to the eHealth portal | Not defined; behavior change (exercise, smoking, diet) |
Heartcare II 2010 [ |
2-arm cluster RCT; USA; 30 m | 282 patients with chronic heart disease needing nursing care | 64.0 (12.7) | 39 | Personal health record: education, monitoring, communication, goal setting, email, bulletin board | Usual care as the home care agencies use to provide | Satisfaction with nursing care |
Hughes 2011 [ |
3-arm RCT; USA; 12 m | 423 senior university employees with Internet, 32% overweight, 46% obese | 51.0 (7.0) | 82 | (1) Coach for Web-based risk assessment, lifestyle plan, email, phone or in-person contact;c (2) website: risk profile assessment, advice, goal setting, action planningb | Printed list of health promotion programs | Not defined; diet, exercise, weight |
Southard 2003 [ |
2-arm RCT; USA; 6 m | 104 patients with CHD or heart failure from 10 hospitals, 200 GPs, adverts | 62.3 (10.6) | 25 | Website + nurse: education, self-monitoring, discussion group, links contact (email, phone or mail), dietician | Usual care | Not defined; weight, exercise, BP, lipid profile, new CV events |
Winett 2007 [ |
3-arm cluster RCT; USA; 16 m | 14 churches with 1071 members: 57% overweight, 60% sedentary | 51.4 (15.7) | 67 | (1) Website: education, goal setting, pedometer;b (2) 1 + pulpit supportc | Waitlist condition | Nutrition improvement, physical activity |
Vernooij 2012 [ |
2-arm RCT; Netherlands; 12 m | 330 patients with CVD, 2 risk factors, Internet via 2 hospitals | 59.9 (8.4) | 25 | Website: risk profile, self-monitoring (BP, cholesterol), treatment goal, nurse support, news, medication changes | Usual care by specialist or GP, receiving baseline risk profile | Relative change in Framingham heart risk score after 1 year |
Verheijden 2004 [ |
2-arm RCT; Canada; 8 m | 146 people with increased CV risk, Internet from 14 GPs | 63.0 (10.5) | 45 | Website: tailored information, diet tool, bulletin board | Usual care | Not defined; BMI, BP, lipid profile |
Ross 2004 [ |
2-arm RCT; USA; 12 m | 107 patients with heart failure, Internet via clinic | 56.0 (-) | 23 | Online medical record (clinical notes, laboratory reports, test results), education, nurse support | Usual care | Change in self-efficacy domain |
Bove 2011 [ |
2-arm RCT; USA; 12 m | 465 people with CVD risk >10% via community, clinics, churches | 61.0 (10.0) | 46 | Online telemedicine system: laboratory and medication review, self-monitoring (BP, weight, pedometer), feedback, education, own doctor involved | 4-months meetings with nurse: review data from logbooks | Reduction in Framingham 10-year CVD risk score |
Keyserling 2014 [ |
2-arm RCT; USA; 12 m | 385 people with CHD risk score ≥10% but no CVD from 5 GPs | 62.0 (7.8) | 48 | Website: CHD risk calculator, advice, education, action planning, goal setting. | Same CHD risk calculator, but in-person and by phone | Framingham 10-year CHD risk score at 4 and 12 m |
Zullig 2014 [ |
2-arm RCT; USA; 3 m | 96 people with CVD or DM from primary clinics | 36.1 (12.2) | 67 | CVD risk assessment, website: 6 modules with risk assessments, goal setting, education | Printed information on CVD | Not defined; Framingham 10-year CVD risk score, BMI, smoking status, systolic BP |
Activate your Heart 2014 [ |
2-arm RCT; UK; 6 m | 95 people with stable angina, Internet from 9 GPs | 66.2 (9.2) | 25 | Website: CVD risk assessment, education, goal setting, self-monitoring, email/chat with experts | Usual care with GP | Change in step count at 6 weeks and 6 m |
e-Care 2014 [ |
2-arm RCT USA 6m | 101 people with BMI >26, elevated BP via electronic health records | 56.9 (7.0) | 42 | Website + dietician: CVD risk assessment, goal setting, action planning, self-monitoring (weight, BP, physical activity, diet) | Usual care, printed report for patient and doctor | Change in systolic BP, weight and 10-year CVD risk score |
Greene 2012 [ |
2-arm RCT; USA; 6 m | 513 employees + families 45% overweight and 48% obese | 60% older than 50 years | 79 | Printed lifestyle guide, website: online social network, self-monitoring (weight, exercise), goal setting, feedback | Printed lifestyle guide | Not defined; physical activity, weight, lipid profile |
Holbrook 2009 [ |
2-arm cluster RCT; Canada; 12 m | 46 GPs with 511 people with DM, ≥1 CV risk factor | 60.7 (12.5) | 49 | Personal Web-based profile overview for DM/CVRM care, automated telephone reminders, summary for doctor, doctor involved | Usual care | Composite score for process of care |
Diabetes in Check 2014 [ |
2-arm RCT; Australia; 9 m | 436 people with DM, Internet via DM network | 58.2 (10.3) | 48 | Website: self-monitoring (exercise) goal setting, education, discussion board | General website with home page and contacts page only | Not defined; physical activity, BMI |
a Abbreviations: BMI: body mass index; BP: blood pressure; CHD: coronary heart disease; CV: cardiovascular; CVD: cardiovascular disease; CVRM: cardiovascular risk management; DM: diabetes mellitus; GP: general practitioner.
b For studies with more than 2 arms, this arm was used for all analyses.
c For studies with more than 2 arms, this arm was used for the subgroup analysis on blended interventions.
Effect on diastolic blood pressure (26 studies).
A significant reduction in HbA1c level favoring the intervention among patients with type 2 diabetes mellitus was found (21 studies; n=6518;
Effect on glycated hemoglobin (21 studies).
Fifteen studies tested interventions for weight loss and two studies tested interventions for maintenance of weight loss. The pooled analysis (17 studies; n=3713;
Effect on weight (17 studies).
A small but significant reduction in LDL cholesterol favoring the intervention was found (17 studies; n=5035;
Effect on low-density lipoprotein cholesterol (17 studies).
Fourteen studies (n=4444;
Nine studies (n=2321;
Finally, we pooled the primary outcomes of the 37 studies (n=11,021;
Effect on physical activity (14 studies).
Effect on cardiovascular composite scores (9 studies).
General effect on primary outcomes (37 studies).
Six studies (n=1904; 1 short-term and 5 long-term studies) reported on cardiovascular event rates. The mean length of the studies was 13 months (range 6-24 months). The pooled analysis showed no difference in rate between groups (pooled OR 0.75, 95% CI 0.39-1.42; I2=27%;
Effect on cardiovascular event rates (6 studies).
Association between study duration and effect size (Hedges' g). One outlier study (Ideatel) was removed from analysis.
Results are summarized in
Subgroup analyses within the analysis of standardized primary outcomes.
Subgroup | N of studies | Hedges’ |
95% CI | I2 | |
|
|
|
|
|
|
|
Short (<12 months) | 15 | –0.43 | –0.57, –0.29 | 69% |
|
Long (≥12 months) | 22 | –0.12 | –0.19, –0.06 | 41% |
|
|
|
|
|
|
|
Primary (including diabetes control) | 33 | –0.25 | –0.32, –0.17 | 72% |
|
Secondary | 4 | –0.20 | –0.34, –0.06 | 0% |
|
|
|
|
|
|
|
Not all older than 50 years | 29 | –0.23 | –0.33, –0.14 | 72% |
|
All older than 50 years | 4 | –0.30 | –0.51, –0.09 | 80% |
Internet only vs controlc | 14 | –0.15 | –0.23, –0.07 | 40% | |
Blended vs controld | 26 | –0.33 | –0.43, –0.22 | 79% |
a Subgroup analysis performed in the sample of studies that was used for the analysis of primary outcomes.
b Subgroup analysis performed on the sample of studies that targeted primary prevention (including diabetes control).
c Subgroup analysis performed on the sample of studies that evaluated an Internet-only intervention. In case a study tested multiple arms, the appropriate arm was included in the analysis.
d Subgroup analysis performed on the sample of studies that evaluated a blended intervention. In case a study tested multiple arms, the appropriate arm was included in the analysis.
Because of the fairly consistent finding that treatment effects were higher in short-term studies than in long-term studies, we performed a mixed effects meta-regression to explore the association between study duration and effect size. The effect size seemed to become smaller in studies with longer follow-up, although the association was not significant (Hedges’
We performed sensitivity analyses for each of the six domains of bias assessed with the adapted Cochrane Risk of Bias Tool by comparing the standardized primary outcomes of the studies with the low risk and unknown/high risk of bias (
In this systematic review and meta-analysis, we found for people with elevated cardiovascular risk, Web-based interventions lead to improvement of systolic and diastolic BP, HbA1c, weight, LDL cholesterol, physical activity levels, and cardiovascular risk composite scores. Only seven studies included participants all aged 50 years or older. Therefore, our conclusions apply for the population in middle age and beyond. Effects were more pronounced over the short term (study duration <12 months) and in studies that tested a blended intervention (combination of an Internet application and human support). We found no evidence for an effect on incident cardiovascular disease.
Our findings on single cardiovascular risk factors are consistent with conclusions of other meta-analyses in younger adult populations [
We found that the beneficial effects of Web-based interventions decline over time and effects are larger when interventions are combined with human support. Decreasing adherence over time was reported in several studies included in our meta-analysis and could be an important contributor to the decreasing effect over time. We were unable to formally test this because information on adherence and engagement was only reported by 22 studies and definitions varied widely. The identified effect moderators are not specific to Web-based interventions for cardiovascular risk factors [
Our results do not show a beneficial effect of Web-based interventions on incident cardiovascular disease. Although the declining effect over time could play a role, more likely explanations for these findings are the limited follow-up time of the studies to detect these outcomes (mean length of the studies was 13 months) and the fact that these outcomes were not the primary focus of these studies. Because of the latter, data collection may not have been systematic and adjudication of the data by an independent committee may be lacking. Therefore, we cannot draw strong conclusions from these findings.
The results of this study should be interpreted with caution because of several limitations. The methodological quality of the studies was fair, but none of the studies was double blind, rendering them prone to performance bias. Only 20 studies had a blinded outcome assessment, so detection bias may also be present. Because the sensitivity analyses for the risk-of-bias assessment did not reveal significant differences between the low risk and unknown/high risk-of-bias subgroups, except for the domain of random sequence generation, we think that our findings have not been largely affected by these potential sources of bias. Another limitation is the substantial heterogeneity in several of the meta-analyses that is, in part, explained by two effect modifiers: study duration and intervention type. Patient groups with a higher burden have a larger window of opportunity for improvement potentially resulting in larger intervention effects [
Strengths of our study are the comprehensive search strategy, the quantitative meta-analysis, and the assessment of the effect of Web-based interventions for all cardiovascular risk factors using both intermediate and clinical outcomes. Our search strategy was comprehensive because we used a broad definition of Web-based interventions and only excluded telemedicine and mobile phone interventions. It was not always possible to set Web-based interventions apart from telemedicine and mobile phone interventions. As long as the Web-based program was the main component of the intervention, we judged the study eligible for our systematic review. By pooling the effect sizes on all different cardiovascular risk factors, we aimed to assess the overall effect of an Internet-based approach for people with increased risk of cardiovascular disease. This approach provides insight into the overall potential of Internet-based interventions in this field. Although basic computer literacy as an inclusion criterion probably led to selection of participants with a relatively high socioeconomic status, several studies included in the meta-analysis focused on people from medically underserved areas. Therefore, the external validity of the results might be acceptable and may be generalizable to middle-aged to older primary care populations with an increased risk of cardiovascular disease.
Our results show that Web-based interventions can be effective in improving the cardiovascular risk factor profile of middle-aged and older people, but effects are modest and can only have clinical relevance on the population level if sustained over time. Considering the current interest and focus on eHealth by policy makers, funding agencies, and a myriad of research and patient organizations [
Research protocol.
Comprehensive search strategy.
Data-extraction form.
Summary of the risk of bias assessment.
Funnel plots.
Subgroup analysis: study-duration.
Risk of bias assessment: sensitivity analyses for the domains of risk of bias.
blood pressure
glycated hemoglobin A1c
interquartile range
low-density lipoprotein
physical activity
randomized controlled trial
We thank R Spijker, medical librarian at the Academic Medical Centre of Amsterdam, for his assistance in designing and conducting the search and M Siervo, lecturer in Nutrition and Ageing at the Institute for Ageing and Health, Newcastle University, for his advice on our statistical analysis plan. We also thank the study authors (LA Volk, CC Quinn, and JD Ralston) who provided additional data. The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 305654. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
CRB had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. ER, EMvC, WvG, CB, MK, SA, and HS conceived the study. CRB, BS, and ER wrote the study protocol and analysis plan. CRB and BS collected and extracted the data, with support from ER and EMvC. CRB performed the statistical analysis, supported by BS, ER, WB, and WvG. CRB and ER drafted the manuscript and all authors critically revised the manuscript for important intellectual content.
None declared.