Home blood pressure telemonitoring figures among the possible solutions that could help improve blood pressure control of hypertensive patients. To summarize the effectiveness of home blood pressure telemonitoring on blood pressure control from randomized, controlled studies.
Electronic databases were searched for publications in English, reporting on randomized trials of home blood pressure telemonitoring vs. usual care. Outcome measures were office or ambulatory blood pressure changes, rate of blood pressure control, and number of antihypertensive drugs used by patients. A random effects model was applied.
Twelve studies met inclusion criteria. A high level of heterogeneity was found among studies for all the variables explored. Office blood pressure was reduced significantly more in patients randomized to home telemonitoring (systolic: 5.64 (95% confidence interval: 7.92, 3.36) mm Hg; diastolic: 2.78 (3.93, 1.62) mm Hg; 11 comparisons, n = 4,389). The effect on ambulatory blood pressure was smaller than on office blood pressure (systolic: 2.28 (4.32, 0.24); diastolic: 1.38 (3.55, +0.79) mm Hg; 3 comparisons, n = 655). The relative risk of blood pressure normalization (<140/90 mm Hg nondiabetics and <130/80 mm Hg diabetics) in the telemonitoring vs. the usual care group was 1.31 (1.06, 1.62) (5 comparisons, n = 2,432 subjects). Use of telemonitoring was associated with a significantly increased use of antihypertensive medications (+0.22 (+0.02, +0.43), 5 comparisons, n = 1,991).
Home blood pressure telemonitoring may represent a useful tool to improve blood pressure control. However, heterogeneity of published studies suggests that well designed, large-scale, randomized, controlled studies are still needed to demonstrate the clinical usefulness of this technique.
American Journal of Hypertension, advance online publication 9 June 2011;doi:10.1038/ajh.2011.100
General population and community-based studies performed worldwide have shown that blood pressure control is achieved in not more than 50% of patients taking antihypertensive drug treatment.1,–3 Unfortunately, inadequate blood pressure control and poor follow-up of hypertension are associated with increased cardiovascular morbidity and mortality.4,–6
In recent years, extensive research has shown that home blood pressure monitoring may provide a reliable assessment of individual patient's blood pressure and thereby may contribute to a more appropriate diagnosis and treatment of hypertension, and a more prompt and effective prevention of adverse cardiovascular outcomes.7,,–10 Regular self-monitoring of blood pressure at home is associated with a better control of blood pressure, as compared to conventional or usual care,11,12,13 and may increase adherence to antihypertensive drug treatment, reduce clinical inertia of physicians, and promote a more appropriate use of antihypertensive medications.7,8,10,14,15
However, despite its increasing use in daily practice and the growing awareness in the scientific community about its positive impact on the diagnostic and therapeutic management of hypertensive patients, the potential advantages offered by home blood pressure monitoring have not yet been fully exploited. This occurs because the information offered by this clinical tool is often incomplete and imprecise, relying on handwritten and frequently inaccurate or illegible logbooks provided to the physician.16 Although a practical and cheap solution to this problem might be the use of devices with a storage memory, this does not eliminate the possibility of patients adhering poorly to measurement schedules.17
Systems for remote monitoring of blood pressure at home may help overcome such difficulties.18 However, implementation of such a telemedicine solution in the clinical practice is made difficult by the high cost and by the lack of definitive scientific evidence on its clinical value. Differences in study designs, technologies, and blood pressure monitoring programs, and heterogeneous patients' selection criteria have made interpretation of single studies difficult.18
At present, a few published reviews have suggested that blood pressure telemonitoring might improve blood pressure control, but no systematic and quantitative analysis has yet been performed.18,,,–22 The objective of the present meta-analysis was thus to summarize the evidence from randomized controlled trials of the effectiveness of home blood pressure telemonitoring as compared to usual care. The meta-analysis was carried-out following the PRISMA proposal guidelines for reporting.23
Methods
Search methodology. In order to locate eligible studies, electronic databases (PubMed, Embase, Cochrane database of systematic reviews) were scanned for articles published in English language from inception to October 2010. Databases were searched using a strategy combining different MeSH or free text terms (see Appendix). Previous systematic reviews and meta-analyses retrieved during the search were searched and the reference lists of these studies were checked for possible additional publications. Two reviewers (S.O. and A.G.) assessed lists of resulting abstracts, independently. The selection of the relevant articles was based on the information obtained from those abstracts, which gave some expectation that eligibility criteria would be met. After exclusion of not relevant articles, original publications of potentially eligible studies were retrieved and reviewed in detail to assess eligibility.
Data selection and extraction. Established criteria for inclusion of a given study in the meta-analysis were the availability of an intervention group based on home blood pressure telemonitoring and of a control group based on usual care not including home blood pressure monitoring. Home blood pressure telemonitoring had to be based on the use of an electronic automated blood pressure monitor storing blood pressure values obtained at the patient's home and transferring them to a remote computer through a telephone line (wired or wireless), a modem or an Internet connection (web).18 Outcome measures different from home blood pressure, namely office and/or ambulatory blood pressure, had to be available, independently from the overall number of subjects included in each individual study. Only randomized controlled studies could be included in the analysis. For further details on inclusion and exclusion criteria see Appendix.
The two reviewers independently extracted data from selected papers and data were entered in an electronic spreadsheet. In case of differences in selection or inclusion of studies and interpretation of data a consensus was reached between the reviewers, after discussion. The following general study characteristics were considered: study design, sample size, study duration, age and gender, methodology of home blood pressure measurement, telemonitoring technique, type of usual care, additional interventions (e.g., educational intervention, nurse or pharmacist led care, appointment reminder systems, phone calls, etc.), study outcome measures and methods used for their adjustment.
Apart from following strict inclusion and exclusion criteria no specific study quality analysis was applied, but according to the PRISMA indications,23 statistical techniques, such as sensitivity analyses, were used to assess the impact of exclusion or inclusion of certain studies on the effect size (see below).
Where data were missing from original publications, attempts were made to retrieve them after contacting the corresponding author. Otherwise, missing data were replaced by statistical methods (see below). Where a study foresaw more than one outcome time point, data concerning the longest follow-up were extracted.
Outcome measures. The measures assessed for both the intervention and control group were: (i) change in mean office or ambulatory systolic and diastolic blood pressure from baseline to the final follow-up time point; (ii) rate of control of office or ambulatory blood pressure at study end, according to thresholds specified in the original study; (iii) number of antihypertensive drugs used at the end of the study. When available from each single study, data adjusted for confounding factors were employed.
Statistical analysis. The meta-analysis was performed according to a random effect model. Weighted mean differences together with their 95% confidence interval between intervention and control group were calculated for the overall mean change in systolic and diastolic blood pressure and for the average number of drugs used by patients at study end. To compare the percentage of patients achieving blood pressure control at follow-up between groups, relative risk and 95% confidence intervals were estimated. Analysis of blood pressure changes and achieved targets was separately performed on office and ambulatory measures.
In some studies, baseline-end of study difference in blood pressure and corresponding standard deviation were missing. In these cases, difference was computed by subtracting the average value at the end of follow-up from that available at baseline. When not available, standard deviation of the difference was estimated by applying the elementary theory of differences of correlated variables. The correlation between baseline and final value was estimated from studies where all three standard deviations were reported and then used in combination with the latter two standard deviations to estimate the standard deviation of change when not available.24 When either of the latter two standard deviations was missing then an average value from the other studies was input.25
Clinical heterogeneity was assessed across studies using a χ2 test for systematic variation and I2 (ref.26). A series of sensitivity analyses was performed to assess the impact of each study on the overall outcomes with recalculation of both the weighted mean differences as each study was removed one at a time from the analysis. In addition to sensitivity analysis, the likely presence or absence of a publication bias was examined graphically by producing funnel plots of effect estimates against precision (1/s.e.).27 Sensitivity analyses and funnel plots helped in providing a visual feedback of any potential bias due to missing publications and in giving a feedback on possible undue influence of some of the studies included in the analysis.
A P value <0.05 was set as the minimum level of statistical significance throughout the text analysis.
Results
A total of 513 references were identified from the electronic search (Figure 1). Of these, 482 were excluded because articles were not relevant to the analysis, no home monitoring was available, or the publication reported on trial design only or was a meta-analysis or a systematic review. A total of 31 original publications were retrieved and reviewed for potential inclusion in the meta-analysis: 12 unique randomized controlled trials met the inclusion criteria and were thus included in the meta-analysis.28,,,,,,,,,,–39
Flow diagram of the process of selection of the studies to be included in the meta-analysis.
A summary of the characteristics of the included studies is reported in Table 1. A total of 10 studies (11 comparisons) used office blood pressure as primary or secondary outcome measure, with an intervention group of 2,224 subjects and a usual care group of 2,165 subjects. Three studies considered ambulatory blood pressure as main study endpoint (360 subjects for the intervention and 295 for the control group). The overall median length of follow-up was 24 weeks (range 8–240 weeks).
Characteristics of the studies included in the meta-analysis
Table 2 shows summary data for age, gender and blood pressure values separately for the group of studies based on office blood pressure and for the studies based on ambulatory blood pressure. No statistically significant differences were observed between the intervention and usual care group for all the variables explored.
Age, gender distribution, office and ambulatory blood pressure in the patients of the meta-analysis
Publication bias for office blood pressure changes, office blood pressure normalization, and for the number of medications, assessed by funnel plots, revealed a good symmetry of data distribution with possible existence of unpublished studies, but these seemed to be limited to small sample size trials, with a likely marginal effect on the overall meta-analysis results (Figure 2). Publication bias was not assessed for ambulatory blood pressure changes and normalization, due to the limited number of comparisons available.
Funnel plots for office systolic (SBP) and diastolic blood pressure (DBP) changes, for relative risk of office blood pressure (BP) normalization and for number of antihypertensive medications used at the end of follow-up. md, mean difference.
Systolic blood pressure changes
Pooled data from randomized controlled trials which reported on differences in mean office systolic blood pressure showed a significantly (P < 0.0001) larger reduction for this outcome in the telemonitoring group as compared to usual care (weighted mean difference: 5.64 (95% confidence interval: 7.92, 3.36) mmHg) (Figure 3). However, a significantly high level of between-studies heterogeneity was detected (I2 = 65.8%; P < 0.01). A sensitivity analysis, excluding those studies with potential bias, did not significantly alter the study results, and the range of weighted mean difference (from 6.19 to 4.89mmHg) was very similar to the overall random effect.
Weighted mean difference (95% confidence interval (CI)) for office systolic blood pressure (SBP) changes in patients randomized to telemonitoring compared with patients followed by their healthcare givers in the usual clinical setting.
In the three studies assessing ambulatory blood pressure, the reduction in this parameter was smaller than that for office blood pressure, though still significantly (P < 0.05) greater in the telemonitoring group (2.28 (4.32, 0.24) mmHg). Some homogeneity was found among studies based on ambulatory blood pressure, though such a test has a limited power with only three comparisons.
Diastolic blood pressure changes
Office diastolic blood pressure was reduced significantly (P < 0.0001) more in patients randomized to telemonitoring than in those receiving usual care (weighted mean difference: 2.78 (3.93, 1.62) mmHg) (Figure 4). Also in this case there was a significant heterogeneity among studies (I2 = 56.6%; P < 0.05). Removal of studies in the sensitivity analysis did not reveal any significant influence of a given study (range of weighted mean differences: 3.08–2.52mmHg).
Weighted mean difference (95% confidence interval (CI)) for office diastolic blood pressure (DBP) changes in patients randomized to telemonitoring compared with patients followed by their healthcare givers in the usual clinical setting.
A lower and non significant (P = 0.21) effect of telemonitoring was observed when the three studies based on ambulatory monitoring were analyzed (1.38 (3.55, +0.79) mmHg): no heterogeneity was found among these studies.
Blood pressure normalization
In the randomized controlled trials that reported on office blood pressure normalization, there was a significant (P < 0.05) improvement in blood pressure control in the telemonitoring group as compared to the usual care group (relative risk 1.31 (1.06, 1.62)), with a high level of heterogeneity (I2 = 77.9%; P < 0.01) (Figure 5). The relative risk resulting after excluding single studies in the sensitivity analysis ranged between 1.20 and 1.41.
Relative risk (RR) of office blood pressure (BP) below target (i.e., normalization) in patients telemonitoring their BP at home compared with patients followed by their healthcare givers in the usual clinical setting. CI, confidence interval.
Analysis of patients with controlled ambulatory blood pressure was not done because this outcome measure was available for two studies only.
Number of antihypertensive drugs
Use of telemonitoring was associated with a modest, but statistically significant (P < 0.001), increase in the number of antihypertensive drugs taken by patients at the end of follow-up (weighted mean difference: +0.22 (+0.02, +0.43)) (Figure 6). Heterogeneity among the five comparisons was again high as for the other studied measures (I2 = 79.1%; P < 0.001). The increase in the use of medications did not substantially vary after removal of each single study from the overall analysis (interval 0.18–0.32 in the sensitivity analysis).
Weighted mean difference (95% confidence interval (CI)) in the number of antihypertensive medications used by patients randomized to telemonitoring as respect to those used by patients followed by their healthcare givers in the usual clinical setting.
Discussion
Results of the present meta-analysis of randomized controlled trials show that use of home telemonitoring is associated with a significantly larger office systolic and diastolic blood pressure reduction than usual care. In the intervention group, a significantly greater chance of achieving office blood pressure control and a significant increase in the use of antihypertensive medications at the end of follow-up were also observed.
The small office blood pressure differences between the two groups of our meta-analysis (about 6mmHg for systolic and 3mmHg for diastolic blood pressure) are not clinically irrelevant, suggesting that telemonitoring might contribute to reducing adverse cardiovascular events. As a matter of fact, large meta-analyses of prospective, observational or randomized placebo controlled trials, or antihypertensive drug trials, report that a 2–3mmHg reduction in systolic or a 4–5mmHg reduction in diastolic blood pressure may be associated with a 10% reduction in cardiovascular mortality and may reduce by 20–30% the risk of major cardiovascular events.40,,–43
In the recent past several large, randomized, controlled trials and subsequent systematic reviews and meta-analyses11,–13 have addressed the impact of home blood pressure monitoring on blood pressure control, clearly demonstrating that this technique may be useful to improve blood pressure control of hypertension. Higher blood pressure reductions (by 2.2–3.8mmHg for systolic and by 1.5–1.9mmHg for diastolic blood pressure) and a 3–10% increased chance of meeting office blood pressure targets, were observed in patients making use of self blood pressure monitoring at home. Some reviews summarized the results of randomized and controlled or observational and uncontrolled studies assessing the effect of telemedicine in different chronic illnesses, including hypertension. However, to our knowledge, no systematic and quantitative analysis has yet been done on randomized controlled studies based solely on self blood pressure telemonitoring. This is probably due to the fact that only in the last 3 years a consistent number of large trials with such a design has been published on this topic. Comparison of results of our meta-analysis with those of previous reviews based on studies on home blood pressure monitoring supports the idea that the degree of blood pressure control achieved with telemonitoring is not only better than that observed in patients under usual care, but might also overcome that of simple home blood pressure monitoring.
Unfortunately our analysis shares all the limits of the meta-analytical approach. This meta-analysis was characterized by large heterogeneity across studies: in particular, one study showed a large effect size on blood pressure changes, though the weight of this study on the overall results was of 1% only. This heterogeneity might be explained by differences in clinical settings, telemedicine technologies, timing of self-monitoring and number of readings, inclusion criteria and feature of the comparative group. Despite the use of a random effect model, we cannot exclude the likelihood of publication bias, the possible existence of studies with negative results or the relatively poor scientific validity of some studies included in the meta-analysis.
We could not assess the effect of telemonitoring on ambulatory blood pressure because of the limited number of studies available. However, in the three studies assessing intervention on ambulatory blood pressure, the overall effect size was smaller than that of studies based on office blood pressure. This might be explained by the fact that usually ambulatory blood pressure changes following intervention are smaller than those seen with office blood pressure, due to the lack of the white coat and placebo effect.44,45 Probably studies using office blood pressure as the primary endpoint might overestimate the effect of the intervention and more studies based on ambulatory blood pressure monitoring might be needed in the future.
The analysis of patients achieving good blood control was done on a limited number of studies, with the reference threshold for office blood pressure normalization set to 140/90mmHg for nondiabetics and to 130/80mmHg for diabetics and patients with chronic kidney disease. Thus some heterogeneity in the methodology and the limited sample of studies used to assess this endpoint may make extrapolation of results to clinical practice difficult.
Not all studies included hypertensive patients. Goulis et al.31 studied obese subjects and Shea et al.36 diabetic subjects: both authors did not report on the proportion of hypertensive patients, though average baseline office systolic and diastolic blood pressure values were above normal, suggesting that also hypertensive patients might have been included in these studies. In any case the effect sizes observed for these two studies were not different from those of the other studies including only hypertensive patients.
Finally, we were unable to examine two additional important issues related to home blood pressure telemonitoring, namely its impact on patient's quality of life and the ratio between the clinical benefits and costs of telemedicine. Unfortunately, such information was available from only a couple of studies. In one study,35 quality of life tended to be higher and costs lower in the group teletransmitting blood pressure data. In another study no improvement in quality of life was observed.31 In two separate publications of the main study included in our meta-analysis,33 Madsen et al. observed lower medication and consultation costs in the intervention group, which were however offset by the cost of the telemonitoring equipment,46 and an improvement in health related quality of life during telemonitoring.47
Our meta-analysis suggests that home blood pressure telemonitoring is associated with a better blood pressure control than usual care management. This technique might be considered a useful, though adjunctive, practice to more effectively manage hypertensive patients in the clinical practice, probably on selected patients, such as those at high-risk (where tighter blood pressure control has to be achieved) or patients with a reduced adherence to treatment schedule. However, a significant heterogeneity of the studies included in our meta-analysis suggests that large-scale, well-designed, randomized, controlled studies, based on easy-to-use technologies and standardized procedures, are needed to clearly more convincingly show the superiority and clinical usefulness of home blood pressure telemonitoring as compared to conventional patient management. Further work should explore which kind of telemonitoring techniques and combined intervention might optimize the impact on reducing blood pressure and helping patients in reaching target levels. Planning and execution of studies evaluating possible advantages or adds-on of home telemonitoring as compared to conventional home blood pressure monitoring should also be encouraged. Finally, future investigations should be done considering carefully the design of the intervention and the use of more objective outcomes (e.g., ambulatory blood pressure monitoring) that are less likely to be biased by the operator and the condition of blood pressure measurement. Also the actual impact of home blood pressure telemonitoring on patient's quality of life, the cost of disease management, and cardiovascular morbidity and mortality outcomes needs to be explored in future studies.
Appendix: Additional Methodological Information
Search strategy
The search strategy was based on the use of selected Medical Subject Headings (MeSH) and free text terms. The following terms were used:
randomized controlled trial (MeSH)
blood pressure (MeSH)
self (MeSH)
home (MeSH)
blood pressure (MeSH)
monitoring (MeSH)
blood pressure monitoring (MeSH)
blood pressure measurement (free text terms)
telemedicine (MeSH)
telemonitoring (free text term)
telecare (free text term)
telehealth (free text term)
telehomecare (free text term)
These terms were combined or used individually in order to select potential papers to be included in the meta-analysis. The search strings employed were:
randomized controlled trial AND blood pressure AND self
home AND blood pressure
home AND monitoring AND blood pressure monitoring
blood pressure measurement AND telemedicine
telemonitoring
home AND telecare
telehealth
telehomecare
Selection criteria
Identification of studies to be included in the meta-analyses was based on pre-defined selection criteria.
Articles were included if:
The study design was a randomized controlled trial
An intervention group based on home blood pressure telemonitoring was available. Home blood pressure telemonitoring had to be based on the use of an electronic automated blood pressure monitor storing blood pressure values obtained at the patient's home and transferring them to a remote computer through a telephone line (wired or wireless), a modem or an Internet connection (web). At least one self blood pressure measurement had to be available for each patient in the intervention group
A control group based on usual care not including home blood pressure monitoring was available
Outcome measures such as office and/or ambulatory blood pressure, had to be available, independently from the overall number of subjects included in each individual study, for both the intervention and control group. Such measures had to include absolute values at baseline and at the end of follow-up or changes from baseline to study end. Additional information could be represented by the proportion of patients achieving the blood pressure targets
Possibly secondary additional endpoints such as the number of antihypertensive medications, quality of life indices or costs of services were available
Published in the English language
Appearing in peer-review journals
Articles were excluded if:
The study was not a randomized controlled trial
Outcome measures on intervention and control groups were not available or incomplete or missing data could not be retrieved by the authors
Conference and poster abstracts were not considered in this review.
Disclosure:
Both S.O. and A.G. have occasionally received consulting fees or travel grants from manufacturers and providers of telemedicine services (Tensiomed and Biotechmed). No funding has been received to undertake this study.
