A Systematic Review of the Effects of Home Blood Pressure Monitoring on Medication Adherence

Abstract

Home blood pressure monitoring (HBPM) improves blood pressure control, but little is known about its effects on medication adherence. The authors conducted a systematic review of the published literature on the effects of HBPM on medication adherence. Of 440 abstracts and citations reviewed, 11 randomized control trials met predefined criteria. Six of the 11 randomized controlled trials reported statistically significant improvement in medication adherence; 84% of these were complex interventions involving the use of HBPM in combination with other adherence‐enhancing strategies such as patient counseling by nurses, pharmacists, or a telephone‐linked system; patient education; and the use of timed medication reminders. Interventions conducted in primary care settings were not effective compared with those that occurred in hospital‐based clinics or nonclinical settings. The data on the effects of HBPM on patients' medication‐taking behavior are mixed. Future studies should investigate the independent effects of HBPM in primary care practices where the majority of hypertensive patients receive their care.

Blood pressure (BP) control among hypertensive patients in the United States remains poor, with approximately 30% of patients categorized as controlled in spite of treatment. ¹ The effectiveness of self‐monitoring in improving BP control is well documented, ² and it has been suggested as a useful strategy to improve adherence to prescribed antihypertensive medications. ¹ There are, however, few data to support the effectiveness of home BP monitoring (HBPM) in improving medication adherence among hypertensive patients. Most of the data to date have focused on the impact of HBPM on physicians' prescribing behavior rather than on patients' medication‐taking behavior. ²

The aim of this systematic review was to assess the evidence from published randomized controlled trials (RCTs) of the effects of HBPM on adherence to prescribed antihypertensive medications among hypertensive patients.

Identification and Selection of Trials

To identify published trials that met the inclusion criteria we used standard Cochrane Collaboration systematic review techniques. ³ We searched The Cochrane Library, Medline, Embase, CINAHL, PsycInfo, the Web of Science (Social Science and Science Citation Indexes), and Dissertation Abstracts from inception to August 23, 2005. In addition, key articles from each search were run in PubMed using the “Related Articles” feature to identify further articles not retrieved using the devised search strategies. To check for articles that may have been missed, we also ran the most relevant records through the “Cited References” feature of The Web of Science and the “Related Articles” feature of PubMed. Finally, we screened all relevant references to identify any additional publications and contacted experts in the field about other relevant trials or unpublished materials. Reviews and meta‐analyses were included where relevant additional studies might be identified. The Dickersin filter for randomized controlled trials was applied to Medline and adapted for Embase. ⁴

The search strategy was devised by representing each concept with the relevant free text and thesaurus terms (where available). All concepts were then combined. The vocabulary used in the search strategies focused on terms related to hypertension, self‐monitoring of BP, and compliance. These search terms were similar to the ones used by Cappuccio et al. ² in their systematic review of HBPM and those used by McDonald et al. ⁵ in their systematic review of interventions to improve adherence to prescription medications. Studies were limited to RCTs, but without publication year or language restrictions.

Eligibility Criteria of Studies

We included RCTs in which the intervention included use of HBPM and those in which there was assessment of medication adherence as an outcome either through electronic means, self‐report, pill counts, or pharmacy refills. We excluded RCTs and studies in which medication adherence was not assessed as an outcome. The outcome of interest was the between‐group difference in medication adherence and the within‐group change in medication adherence from baseline to end of follow‐up between intervention and control arms. We assessed medication adherence as the proportion of people who were categorized as adherent (taking at least 80% of prescribed medications in the case of electronic monitoring systems, pharmacy refills, and pill counts; or self‐reported as nonadherent in the case of self‐report data).

Data Extraction

Each author assessed all retrieved lists of citations and abstracts independently. Each reviewer indicated whether a citation was potentially relevant (i.e., appearing to meet the inclusion criteria), was clearly not relevant, or gave insufficient information to make a judgment. We resolved differences about inclusion of studies and interpretation of data by discussion. We then obtained printed copies of all potentially relevant citations or full paper versions of those where insufficient information was available. Finally, we extracted data independently from text, tables, and graphs of the selected articles in duplicate on study design, methods, participant characteristics, study groups, and outcomes, using a structured data collection form.

RESULTS

The process of identifying studies eligible for inclusion in our review is summarized in the Figure. We retrieved and screened 440 citations and abstracts from our searches of all relevant sources, 34 of which were judged to merit examination of the full article. Seventeen of the latter met all predefined inclusion criteria. ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ We excluded six of these studies for one of the following reasons: medication adherence was not assessed, ⁹ , ²⁰ , ²¹ incomplete data on adherence outcome, ²¹ HBPM was not used as the intervention, ¹² , ¹⁷ or they were not RCTs. ¹⁰ Thus, only 11 RCTs were included in this systematic review. ⁶ , ⁸ , ¹¹ , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁸ , ¹⁹ , ²² , ²³

Figure.

Flow diagram showing citations retrieved from literature searches and number of trials included in analysis. RCT=randomized controlled trial; BP=blood pressure

The characteristics of the studies included in the review are presented in Table I. The trials were conducted in various settings, including shopping malls, ¹⁴ community/retirement centers, ¹¹ , ¹⁵ a work site, ¹³ and hospital‐based clinics/tertiary care centers. ¹⁸ , ¹⁹ , ²³ Only three studies were conducted in primary care practices, ⁶ , ¹⁶ , ²³ and study setting was not documented for two trials. ⁸ , ²² Eighty‐two percent of the trials were complex interventions where the effects of HBPM on medication adherence was tested in combination with other adherence‐enhancing strategies such as patient education; ⁸ , ²³ counseling on medication adherence by nurses, pharmacists, or through a telephone‐linked system; ¹¹ , ¹⁶ , ¹⁸ use of timed medication reminders (such as special vials); ¹⁵ monthly home visits; ¹⁴ and nurse case management. ¹⁹ The remaining two trials ⁶ , ²² were simple interventions (they used HBPM alone).

Table I.

Characteristics of Studies Included in the Systematic Review

			Completed Follow‐Up (%)
Study (year)	Duration of Intervention	N	Intervention	Control	Adherence Measure	Statistical Improvement in Adherence
Bailey et al. 6 (1999)	8 wk	62	97	97	Pill count	No
Binstock et al. 8 (1988)	1 yr	111	100	100	Self‐report	No
Friedman et al. 11 (1996)	6 mo	267	85	92	Pill count	Yes
Haynes et al. 13 (1976)	6 mo	38	100	95	Pill count	Yes
Girvin et al. 12 (2004)	6 mo	136	97	97	Pill count	No
McKenney et al. 15 (1992)	24 wk	67	94	97	Electronic monitoring device	Yes
Mehos et al. 16 (2000)	6 mo	36	98	97	Pharmacy records	No
Ogbuokiri 18 (1980)	5 mo	24	79*	Pill count	Yes
Rudd et al. 19 (2004)	6 mo	150	94	91	Electronic monitoring device	Yes
Vrijens and Goetghebeur 22 (1997)	6 wk	628	n/a	n/a	Electronic monitoring device	Yes
Zarnke et al. 23 (1997)	8 wk	31	100	98	Self‐report	No
n/a=not applicable; *lost to follow‐up not differentiated among conditions

The sample size of the interventions was varied, with a range of 24–628 participants. Four trials enrolled fewer than 50 participants; two included 51–100 patients, and five enrolled more than 100 patients. Except for the study by Vrijens and Goetghbeur, ²² follow‐up was adequate in all the trials (range, 79%–100%). The median duration of these studies was 6 months, with a range of 1.5–12 months. Adherence was assessed with objective electronic monitoring means in only three of the 11 studies. ¹⁵ , ¹⁹ , ²² Pill count was used to assess adherence in five studies, ⁶ , ¹¹ , ¹³ , ¹⁴ , ¹⁸ pharmacy refills were used in one study, and self‐report data were collected in two studies. ⁸ , ²³

The effects of the interventions on medication adherence are summarized in Table II. We did not perform a meta‐analysis because of the heterogeneity of the measures employed in reporting adherence, the varying definitions of what constitutes adherence, and the complex nature of the majority of the interventions. As such, we reported in Table II only whether statistically significant differences were noted in adherence outcomes between the intervention and usual care groups within each study.

Table II.

The Effects of the Interventions on Medication Adherence

				Statistical Significance of Change (yes/no)*
				Adherence
				Between	Within	BP
Study (year)	Group/Condition	n	Adherence**	Group	Group	Systolic	Diastolic
Bailey et al. 6 (1999)	Intervention: self‐measurement	31	88±5	No	n/a	No	No
	Usual care	29	94±4	–	n/a	Yes	No
Binstock et al. 8 (1988)†	1. Education alone	32	n/a	–	n/a	No	No
	2. HBPM	23	n/a	No	n/a	Yes	Yes
	3. Compliance contract	15	n/a	No	n/a	Yes	Yes
	4. Pill packs	30	n/a	No	n/a	Yes	Yes
	5. HBPM, contract, pill packs	11	n/a	No	n/a	Yes	Yes
Friedman et al. 11 (1996)†	Intervention: telephone‐linked computer system	133	93	Yes for NAB	Yes	Yes for NAB	Yes
	Usual care	134	94	No	Yes	No	No
Haynes et al. 13 (1976)†	Intervention: HBPM, pill charts	20	65.8±8.2	Yes	Yes	n/a	Yes
	Usual care	18	43.2±10.1	–	No	n/a	No
Girvin et al. 12 (2004)†	1. Self‐recording, monthly home visits	35	Total: 76.3±4.6	n/a	No	n/a	Yes for those who had trouble
		19	NAB: 75.7±5.7				remembering
	2. Self‐recording only	34	Total: 78.0±5.4	n/a	No	n/a	Yes for those who had trouble
		18	NAB: 75.7±5.7				remembering
	3. Monthly home visits only	33	Total: 68.3±5.9	n/a	No	n/a	Yes for those who had trouble
		16	NAB: 56.47±9.0				remembering
	4. Usual care	34	Total: 68.5±7.3	–	No	n/a	No
		13	NAB: 61.1±10.1
McKenney et al. 15 (1992)†	A. Standard medication vials	16	79±22.61	–	No	No	No
	B. Medication vials with time cap	17	93.6±8.67	Yes	Yes	Yes	Yes
	C. Vials with time cap, card for recording BP	17	98.7±11.28	Yes	Yes	Yes	Yes
	D. Vials with time cap, card for recording BP, BP cuff	17	110.2±7.02	Yes	Yes	Yes	Yes
Mehos et al. 16 (2000)†	Intervention: counseling, HBPM, diary	18	82	No	No	No	Yes
	Control: counseling on antihypertensive drug therapy and lifestyle modifications	18	89	–	No	No	No
Ogbuokiri 18 (1980)†	Intervention: counseling, HBPM	12	73	No	Yes	No	No
	Control: counseling on importance of compliance, explanation of HTN, informational pamphlet	12	59	–	No	No	No
Rudd et al. 19 (2004)†	Intervention: nurse management system, HBPM, tips for enhancing adherence, follow‐up phone calls	74	80.5±23.0	Yes	n/a	Yes	Yes
	Usual care	76	69.2±31.1	–	–	No	No
Vrijens and Goetghebeur 22 (1997)	Intervention: HBPM, diary	309	6.6 pills/wk	Yes	n/a	No	n/a
	Usual care	319	5.8 pills/wk	–	n/a	No	n/a
Zarnke et al. 23 (1997)†	Intervention: teaching session, HBPM, self‐adjusted drug therapy	20	0.05±0.2 (doses missed/subject/wk)	No	No	Yes	Yes
	Control: teaching session with nurse on nonpharmacologic treatment, educational book	11	0.2±0.42 (doses missed/subject/wk)	–	No	No	No
BP=blood pressure; n/a=not applicable; HBPM=home BP monitor; NAB=subjects who were nonadherent at baseline; HTN=hypertension; *all analyses adjusted for covariates; **data shown are mean % or mean %± SD unless otherwise noted; †complex interventions are interventions in which HBPM was used with other adherence‐enhancing strategies

Of the 11 RCTs included in our review, six (54%) reported statistically significant improvement in medication adherence attributed to the intervention. ¹¹ , ¹³ , ¹⁵ , ¹⁸ , ¹⁹ , ²² Five of these six studies were complex interventions. ¹¹ , ¹⁵ , ¹⁸ , ¹⁹ , ²² The most complex of these was the study by McKenney et al., ¹⁵ in which 70 patients were randomly assigned to one of four groups with different adherence‐enhancing strategies, including HBPM. Group A served as the control group and received medication in standard vials; group B received medication in vials with the timepiece cap that served as a reminder for patients to take their medications; group C and group D received vials with the timepiece cap and pocket‐size cards for recording their BP readings at each clinic visit; while group D patients were also issued BP cuffs and instructed to perform BP measurements at home and to chart their results on the cards. The mean compliance for the intervention group was significantly higher than that of the control group (95.1% vs. 76.8%; p=0.0002) and this difference was greater as the number of compliance techniques increased. ¹⁵ The remaining four complex interventions with positive findings employed HBPM plus additional patient counseling on recognition of potential drug side effects, provision of tips for enhancing drug adherence, and reinforcement of positive behavior either by nurse managers, ¹⁹ trained pharmacists, ¹⁸ trained lay health workers, ¹³ or a telephone‐linked system. ¹¹ The differences in adherence rates between intervention and control groups were similar for all five studies.

Only one of the six studies that reported positive findings tested the independent effect of HBPM on medication adherence. In this study, Vrijens and Goetghbeur ²² randomized 628 patients who were seen in a research center to either HBPM alone or usual care and adherence was assessed with an electronic monitoring system for a period of 6 weeks. At baseline, the medication adherence rates were the same for both conditions, with a similar compliance behavior over the prior 6 days. Over time, the probability decreased in both groups, with the decrease being significantly greater in the control group as compared with the intervention group (p<0.05). At the 6‐week follow‐up, the intervention group was taking 6.6 pills per week on average compared with 5.8 pills per week in the control group. ²² This study is one of two studies that compared the independent effects of HBPM alone on medication adherence. The other study, by Bailey et al., ⁶ randomly assigned 60 patients from a primary care practice to HBPM alone vs. usual care. Adherence was assessed with pill counts. At the end of 8 weeks, the mean levels of adherence were comparable for both groups (94%±4% vs. 88%±5%, respectively; p>0.05).

When we assessed the studies based on the settings where they were conducted, we found that all three studies performed in primary care settings reported negative findings ⁶ , ¹⁶ , ²³ (Table II). Finally, when we examined the effects of HBPM on both medication adherence and BP control among the 11 RCTs reviewed, only four trials reported statistically significant improvement in both outcomes. ¹¹ , ¹³ , ¹⁵ , ¹⁹

DISCUSSION

We reviewed 11 RCTs that tested the effects of HBPM on medication adherence in various settings, including nonclinical sites. Nine of the 11 RCTs (82%) were complex interventions and two were simple interventions. Fifty‐four percent of the RCTs reviewed (six of 11) reported statistically significant improvement in medication adherence attributed to the intervention. Almost all the effective interventions were complex, including combination of HBPM with use of nurse case managers, patient counseling on recognition of side effects, use of special timed medication vials as reminders, and reinforcement of positive behavior.

When we compared the rate of positive findings according to whether the intervention was complex or simple, we found that the rates were similar for the two types of interventions (55% vs. 50%, respectively). The reported magnitude of the differences in medication adherence between intervention and usual care, however, was larger for the complex interventions compared with the simple interventions. This finding is consistent with the reviews of adherence‐enhancing interventions to date, which suggest that complex interventions are more effective and have longer‐term effects than interventions that are based on single strategies. ⁵ , ²⁴

When we categorized the effects based on the type of medication adherence measure employed in each of the individual studies, we found that all three studies that employed objective electronic assessment of medication adherence ¹⁵ , ¹⁹ , ²² and 60% (three of five trials) of those that utilized pill counts reported significant improvement in medication adherence ¹¹ , ¹³ , ¹⁸ ; whereas those studies that utilized self‐report measures or pharmacy refill data reported negative findings. ⁶ , ⁸ , ¹⁴ , ¹⁶ , ²³ The reason for this may be related to too few studies that evaluated self‐report (only two trials) or pharmacy refill data (one trial) or the small sample size of both studies. Finally, all three studies performed in primary care settings reported negative findings, and the reason for this may be related to the small sample size of these studies (range of 31–62 study participants).

The existing literature testing the effects of HBPM on medication adherence to date has several limitations. First, most of the interventions were small, with a majority of them having fewer than 70 patients (six of 11), thus lacking the power to detect clinically important effects. A rule of thumb suggested by Haynes et al. ²⁵ to achieve a satisfactory level of power in any adherence research study comparing a single intervention and control group is to include 60 participants per study arm to detect a 25% difference in the rate of adherence improvement. Only four of the 11 trials met this standard. Second, the reported adherence measure varied, with only about 27% reporting adherence with the objective electronic assessments. ² , ¹⁵ , ¹⁹ , ²² Third, the complex nature of 82% of the interventions reported (nine of 11) did not allow us to assess the independent effect of HBPM on medication adherence. Finally, of the interventions tested, only three were conducted in primary care practices where the majority of hypertensive patients receive their care. Although we believe that our search strategy was quite extensive, we may have missed some trials that meet our predefined criteria.

IMPLICATIONS

The use of HBPM is widely accepted and recommended by several guidelines for management of hypertension, including the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) ¹ and the British Hypertension Society (BHS IV), ²⁶ probably based on the evidence of its comparability to ambulatory BP monitoring in predicting cardiovascular risk ²⁷ and its effectiveness in improving BP control. ² Despite this endorsement of HBPM as a potential adjunct to the care of hypertensive patients, the mechanisms of the effects of HBPM on BP control have not been well studied. Although the effects of HBPM on patients' medication‐taking behavior are asserted, ¹ few studies have addressed this question. This is the first study to our knowledge to systematically review the evidence of the effects of HBPM on patients' medication‐taking behavior. Our findings indicate that the effect of HBPM on medication adherence is mixed at best, with only 54% of the trials reviewed reporting positive findings, and these effects are greatest when HBPM is combined with other adherence‐enhancing strategies.

CONCLUSIONS

We conclude that the data on the effects of HBPM on patients' medication‐taking behavior are mixed, given that only a little over half of the studies (54%) reviewed reported a statistically significant improvement in medication adherence between the interventions employed and usual care. The reported improvement in adherence between the intervention groups and the usual care groups were greater in the trials that tested HBPM along with other adherence‐enhancing strategies such as patient counseling, patient reminders, and the use of nurse case mangers. Based on our findings, and the widespread use of HBPM, its reliability, feasibility, and acceptability by both patients and their providers, ²⁶ it could be considered as a useful adherence‐enhancing strategy in combination with other strategies such as patient counseling. Future studies are needed to investigate the effectiveness of HBPM in primary care settings, as well as to test the role of medication adherence as the mediating mechanism of the effect of HBPM on BP control.

Acknowledgment and disclosure: We would like to acknowledge Ms. Louise Falzon for her well appreciated assistance with the literature search for this systematic review. Preparation of this manuscript was supported by grants R01 HL 69408, R01 HL078566, and R24 HL076857 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD.