The Changes in Home versus Clinic Blood Pressure with Antihypertensive Treatments: a Meta-analysis

Abstract

Home blood pressure (HBP) monitoring is recommended for assessing the effects of antihypertensive treatment, but it is not clear how the treatment-induced changes in HBP compare with the changes in clinic BP (CBP). We searched PubMed using the terms “home or self-measured blood pressure,” and selected articles in which the changes in CBP and HBP (using the upper arm oscillometric method) induced by antihypertensive drugs were presented. We performed a systematic review of 30 articles published before March 2008 that included a total of 6794 subjects. As there was significant heterogeneity in most of the outcomes, a random effects model was used for the meta-analyses. The mean changes (± standard error) in CBP and HBP (systolic/diastolic) were -15.2±0.03/-10.3±0.03 mmHg and -12.2±0.04/-8.0±0.04 mmHg respectively, although there were wide varieties of differences in the reduction between HBP and CBP. The reductions in CBP were correlated with those of HBP (systolic BP; r=0.66, B=0.48, diastolic BP; r=0.71, B=0.52, P<0.001). In 7 studies that also included 24-hour BP monitoring, the reduction of HBP was greater than that of 24-hr BP in systolic (HBP; -12.6±0.06 mmHg, 24-hr BP; -11.9±0.04 mmHg, P<0.001). In 5 studies that included daytime and nighttime systolic BP separately, HBP decreased 15% more than daytime ambulatory BP and 30% more than nighttime ambulatory BP. In conclusion, HBP falls about 20% less than CBP with antihypertensive treatments. Daytime systolic BP falls 15% less and nighttime systolic BP falls 30% less than home systolic BP.

Introduction

The reduction of blood pressure (BP) levels by antihypertensive drug treatment is associated with decreased risks of cardiovascular events independently of the antihypertensive drug classes used to cause the reduction.¹ In most of the studies investigating this effect BP has been evaluated using mercury or aneroid sphygmomanometers in a clinic or office setting. There is now increasing evidence that BP measurements made in “out-of-office” settings (using both home and ambulatory monitoring) are better predictors of cardiovascular events.² The most widely used measure in these studies has been the twenty-four hour average ambulatory BP, which has been demonstrated to be a better predictor of cardiovascular events in patients taking antihypertensive medication.³ In a meta-analysis by Mancia and Parati,⁴ it was reported that the reduction of 24-hr average BP with antihypertensive medications (14.6 / 9.2 mmHg) was about 60 % of the reduction measured by clinic BP (24.9 / 14.5 mmHg).

Recently, devices for the self-measurement of home BP have become widely available, and their use for the evaluation of the effects of antihypertensive treatment has been endorsed by national and international guidelines,^5-7 which recommend that BP should be measured both in the morning and at night. Like 24-hr ambulatory BP, home BPs also have greater predictive value for cardiovascular events than clinic BPs.^{8, 9} Home BP monitoring is less expensive and more suited to long-term repeated use than 24-hr ambulatory BP monitoring,¹⁰ and can assess the effectiveness of antihypertensive medication on BP control at different times of the day. Verberk et al.¹¹ reported in the meta-analysis of home BP monitoring that the average reductions of clinic BP (20.1/13.6 mmHg) produced by antihypertensive medications were larger than the reductions of home BP (13.9/9.1 mmHg), but the number of studies available for this analysis was small (6 articles).

The purpose of this study was to compare, in a meta-analysis, the changes in home and clinic BP produced by different antihypertensive drugs, including the changes in both morning and evening BP levels.

Methods

Identification of papers

We performed a systematic review of home BP monitoring in PubMed at the end of December 2007. We used the key words “home blood pressure,” and “self-measured blood pressure” to search for candidate articles. Additional papers were collected from reference lists of the identified articles and reviews, and a follow-up search for more recent publications (through March 2008). There were no available articles in the Cochrane library.

Paper selection

A flow chart summarizing the paper selection process is shown in Figure 1.^12-41 The full text of those papers identified as potentially relevant on the basis of their titles and abstracts was reviewed by 2 independent investigators (JI and DJC). The criteria for papers to be included in the present meta-analysis were as follows: (1) study evaluated the effect(s) of at least one antihypertensive drug therapy, (2) results published for both home BP and clinic BP, pre- and post-treatment, (3) home BP measured using a semiautomatic or automatic upper-arm cuff oscillometric device, and (4) home BP measured in the morning and/or evening according to the guidelines.^5-7 Studies were excluded from our analysis if they included: (1) pregnant women, (2) children, (3) patients on hemodialysis, and (4) patients with severe arrhythmia. We included 2 articles in which home BPs were measured only in the morning^{31, 36} (the morning home BPs were used as the average home BPs in these studies) and one article where BP was measured in the morning, at noon, and in the evening.¹⁶ We included some articles in which the method of clinic BP measurements was not clearly described,^{15, 19, 24, 30, 31, 33, 35-37} and/or in which clinic BP was measured only once per clinic visit,³¹ but we repeated our analyses excluding these articles. Rater differences in the selection of articles were discussed, and the last author (TGP) resolved any remaining discrepancies concerning article eligibility.

Figure 1. Flow chart of paper selection.

Data extraction

The averages of age, percentage of males, body mass index, antihypertensive medication use, dose and timing of antihypertensive drugs, and the methods of clinic and home BP measurements for each study were all abstracted as background data. The levels and changes of the clinic and home BPs and their standard deviations (SD) during the antihypertensive treatment periods were also collected. One article⁴⁰ published data only for mean BP and pulse pressure, but the systolic and diastolic BP could be obtained from the original database, because the first author of this meta-analysis was a co-author of the article. In 7 articles,^{18, 20, 22, 25, 26, 32, 39} 24-hr ambulatory BP data were also available, which we included in our analysis.

Data synthesis

The changes in clinic and home BPs were extracted for each of the eligible studies, excluding data for the placebo and control groups,^{14, 40} and also excluding 9 articles in which the method of clinic BP measurements were not described.^{15, 19, 24, 30, 31, 33, 35-37} The exclusion criteria of the 9 articles were as follows: 1) the technique and/or device of clinic BP measurements were not written, 2) the clinic BP was measured only once, and 3) the clinic BP was not measured by physicians or nurses. There were 13 articles in which home BP data were described separately for the morning and evening BP.^{12-15, 18, 20, 23, 25, 30, 37, 38, 40, 41} We also evaluated the changes in clinic, morning, and evening BP depending on whether the antihypertensive medication was taken in the morning^{12-15, 18, 20, 23, 25, 30, 41} or in the evening.^{12, 37, 38, 40} The effects of different antihypertensive drug classes on the changes in clinic and home BP (using morning dosing) were evaluated for calcium channel blockers (6 studies),^{12, 16, 17, 25, 35, 37} angiotensin converting enzyme inhibitors (5 studies),^{13, 14, 18, 21, 22} and angiotensin II receptor blockers (7 studies).^{18, 20-22, 24, 33, 41} Finally, we were able to compare the reductions of home BP with those of 24-hr BP in 7 articles^{18, 20, 22, 25, 26, 32, 39}, and with daytime BP and nighttime BP in 5 articles. ^{22, 25, 26, 32, 39}

Statistical analysis

The changes in BP measured before and after antihypertensive treatment periods are shown as mean ± standard error (SE), and the BPs measured at baseline are shown as mean ± SD. In the data sets in which standard deviations after the treatment periods were unavailable, it was assumed that the SDs of the BP before and after treatment were the same. As most of the accepted articles did not show SDs of the changes in clinic and home BP, the SDs were calculated using the formula⁴²: [(SD at baseline)² + (SD at follow-up)² − 2 r (SD at baseline) (SD at follow-up)]^(0.5) . The r values of changes in clinic and home BPs were estimated from a study in which all variables were available (i.e. clinic systolic BP 0.50, clinic diastolic BP 0.32, home systolic BP 0.56, home diastolic BP 0.75, morning systolic BP 0.61, morning diastolic BP 0.78, evening systolic BP 0.54, evening diastolic BP 0.72).⁴⁰ Before we evaluated the changes in clinic and home BP, we performed a heterogeneity test. The estimates of change in clinic and home BP exhibited significant heterogeneity across articles, and data synthesis was therefore performed using an unstandardized random effects model. Simple correlations between the reductions of BP were evaluated using Pearson's test weighted by the number of subjects enrolled in the studies. The differences in the change in clinic and home BP were calculated in all data sets, and meta-analyzed using a random effects model, except for the differences between home and 24-hr or daytime diastolic BP. These differences were analyzed using a fixed effects model because there was no significant heterogeneity. Additionally, the treatment induced BP changes corrected for the level of BP were calculated using the Oldham's correction⁴³ (corrected BP change = actual change divided by the average of pre- and post-treatment BPs). The mean pre- and post-treatment BPs were meta-analyzed using a random effects model. All statistical analyses were performed using Excel-based (Microsoft Corp. USA) meta-analysis worksheets (developed by JES and JI) and SPSS (version 14.0, SPSS Inc. Chicago, Illinois, USA).

Results

The methods of clinic and home BP measurements in the selected studies are shown in Table 1.^12-41 The baseline characteristics in the selected studies are shown in the Supplemental Table (S1) (please see http://hyper.ahajournals.org). The baseline levels and treatment-induced reductions in clinic and home BP, and the antihypertensive medications used, are shown in Table 2. ^12-41 The baseline levels and treatment-induced BP reductions including in morning and evening and 24-hr BP, and the details of antihypertensive medications in the selected studies are shown in the Supplemental Table (S2) (please see http://hyper.ahajournals.org). The changes in clinic and home BP are shown in Figure 2. The reductions of home BP (-12.2/-8.0 mmHg) were 20% less for systolic and 22% less for diastolic BP than the reductions of clinic BP (-15.2/-10.3 mmHg, P<0.001, N=6794). When we excluded data sets in which the methods of clinic BP measurement were not clearly described,^{15, 19, 24, 30, 31, 33, 35, 36} the reductions of clinic and home BP were similar: -14.7 ± 0.04 / -10.7 ± 0.03 mmHg and -11.8 ± 0.04 / -8.1 ± 0.05 mmHg (P<0.001 for the both differences, N=6322). Scatter plots of the study-specific average changes in clinic and home systolic and diastolic BPs, and their correlations (weighted by the numbers of enrolled subjects) are shown in Figure 3. Additionally, scatter plots of study-specific average baseline BP levels and the BP reduction, and the correlations weighted by numbers of enrolled subjects are shown in Figure 4. The treatment induced BP changes corrected for the level of BP by using the Oldham's correction were 10.2 for clinic systolic BP, 11.5 % for clinic diastolic BP, 10.7 % for home systolic BP, and 9.2 % for home diastolic BP.

Table 1. Blood pressure measurement methods of 30 collected papers.

		Clinic BP	HBP measurements
First author	N. Sub.	Clinic BP devices	Home BP devices	N. HBP	Time	Dur.
Mengden T12	19	Sphg.	Visomat OZ1	?	ME	2 w
Chatellier G13	816	Sphyg.	UA751, A&D	3	ME	4 d
Vaur L14	24	Mercury sphyg.	A&D UA751	3	ME	7 d
Bailey B15	31	Sphyg.	Omron HEM 706	?	ME	?
Fernandez-G16	1395	Omron HEM-705CP	Omron HEM-705CP	2	ME, noon	?
Leeman MJ17	760	Mercury or Aneroid Sphyg.	Omron HEM-705CP	2	ME	8 d
Ragot S18	199	Mercury sphyg.	UA751, A&D	3	ME	7 d
Broege PA19	20	Sphyg.	Omron HEM-702	3	ME	?
Mancia G20	215	Sphyg,	Pressolink-T	3	ME	7 d
	211	Sphyg,	Pressolink-T	3	ME	7d
Ragot S21	217	Omron HEM-705CP	Omron HEM-705CP	3	ME	7 d
	218	Omron HEM-705CP	Omron HEM-705CP	3	ME	7 d
Stergiou GS22	33	Sphyg.	Omron HEM-705CP	2	ME	5 d
Suzuki H23	113	Sphyg.	Omron HEM 401C	1	ME	2 d
Mengden T24	53	Usual device	TensioPhone 2	≥1	ME	?
Mion D25	65	Omron IC	Omron IC	3	ME	7 d
Staessen JA26	197	Shyg. ?	Omron HEM-705CP	3	ME	7 d
	203	Shyg. ?	Omron HEM-705CP	3	ME	7 d
Girerd X27	171	TensioPhone	TensioPhone	3	ME	4 d
Halme L28	113	Omron M4	Omron M4	2	ME	7 d
Borie G29	227	Sphyg. or Electronical device	Tensio Day Monitor	3	ME	5 d
	222	Sphyg. or Electronical device	Tensio Day Monitor	3	ME	5 d
Hashimoto J30	76	NA	Omron HEM701C, 703C, 747IC, 747IC-N	1	ME	5 d
Kamoi K31	50	FT-200	Semiautomatic arm-cuff oscillometric	1	M	?
Niiranen TJ32	56	Sphyg.	Omron M4, HEM-722C	2	ME	7 d
Ewald S33	53	NA	TensioPhone 2	≥1	ME	?
Karotsis AK34	185	A&D767	A&D767	?	ME	3 d
Yamagishi T35	54	NA	Omron HEM401C	1	ME	?
	54	NA	Omron HEM401C	1	ME	?
Yamagishi T36	43	NA	Automatic arm-cuff device	1	M	?
	15	NA	Automatic arm-cuff device	1	M	?
Ashizawa N37	23	NA	Automatic arm-cuff device	?	ME	?
Ikeda T38	49	Omron, HEM907	Omron HEM-737	2	ME	?
Verberk WJ39	214	Omron, HEM705CP	Omron, HEM705CP	3	ME	7 d
Eguchi K40	308	Omron, HEM705IT	Omron, HEM705IT	2	ME	3 d
Nakamoto H41	21	Omron, HEM705IT	Omron, HEM705IT	≥1	ME	?
	19	Omron, HEM705IT	Omron, HEM705IT	≥1	ME	Tel

NA indicates not applicable; N. Sub., Number of enrolled subjects; Sphyg., sphygmomanometer; N. HBP, number of home blood pressure measurements in one occasions; ME, morning and evening; M, morning; w, weeks; d, days; Tel, telemedicine. In the paper of Girerd X et al. home blood pressure was measured in the standing position.

Table 2. Changes in blood pressures and baseline blood pressure levels.

		Anti-HTN treatments	Reduction of systolic / diastolic BPs (mmHg)		Baseline level of systolic / diastolic BPs (mmHg)
First author	N	Drug name	CBP	HBP	CBP	HBP
Mengden T12	19	Amlodipine	13 / 8	11.5 / 6.5	158 / 103	151 / 97.5
	19	Amlodipine	12 / 5	7 / 4	158 / 103	151 / 97.5
Chatellier G13	816	Trandpril	21.8 / 15.3	13.8 / 8.7	166.4 / 101.4	153.2 / 93.8
Vaur L14	24	Trandpril	10.2 / 8.3	10.7 / 5.8	163 / 101	146 / 94
Bailey B15	31	Perindopril or Indapamide	8 / 4	8 / 4	156 / 93	146 / 88.5
Fernandez-G16	1395	Verapamil SR	20.1 / 13.0	16.8 / 11.0	165.9 / 97.8	158.7 / 93.3
Leeman MJ17	760	Diltiazem SR	20 / 13	11 / 8	162 / 99	155 / 95
Ragot S18	105	Losartan	17.5 / 14.3	8.8 / 4.3	160.0 / 99.7	144.0 / 88.6
	94	Trandpril	16.9 / 12.7	9.6 / 5.2	160.2 / 99.8	144.3 / 90.5
Broege PA19	20	Cholrthalidone Prinivil *3	-3 / -5	2 / 0	153 / 78	140 / 77
Mancia G20	215	Valsartan	10 / 7.3	8.2 / 4.9	158 / 100.8	149.5 / 95.8
	211	Irbesartan	16.2 / 10.5	11.1 / 6.9	159.3 / 100.7	140.0 / 95.5
Ragot S21	217	Telmisartan	13.9 / 8.9	12.3 / 6.6	157.9 / 97.6	151.6 / 94.2
	218	Perindopril	10.9 / 6.6	8.9 / 5.2	158.9 / 97.9	152.2 / 94.8
Stergiou GS22	33	Lisinopril	15 / 10.5	17 / 10.1	144 / 95.6	145 / 93.2
	33	Losartan	8.3 / 5.9	8.3 / 4.5	144 / 95.6	145 / 93.2
Suzuki H23	113	Amlodipine Benazepril	19 / 20	20.5 / 16	148 / 89.0	155 / 91.0
Mengden T24	53	Olmesartan	27.6 / 16.0	9.2 / 6.7	162.6 / 96.3	149.8 / 85.7
Mion D25	65	Amlodipine	2 / 5	6.5 / 1	164 / 99	160 / 93.5
Staessen JA26	197	Lisinopril, HCZ, Amlodipine, or Prazosin *1	22 / 14	16 / 10.2	159.1 / 101.5	146.4 / 92.2
	203	Lisinopril, HCZ, Amlodipine, or Prazosin *2	15.3 / 10.5	11.1 / 7.3	160.8 / 101.8	146.8 / 92.0
Girerd X27	171	Valsartan HCZ	6 / 4	6 / 4	157 / 91	152 / 87
Halme L28	113	Any	12.7 / 7.1	7.8 / 3.1	159.5 / 94.1	143.1 / 85.3
Borie G29	227	Valsartan HCZ	11.8 / 6.9	10.6 / 7.4	153 / 90	149 / 89
	222	Irbesartan HCZ	15 / 8.6	13 / 9.5	153 / 91	149 / 90
Hashimoto J30	76	Indapamide	9 / 5	12 / 6	145 / 86	142.5 / 83
Kamoi K31	50	Doxazosin	9 / 4	18 / 5	151 / 87	164 / 85
Niiranen TJ32	56	Candesartan HCZ *3 Felodipine *3	18.5 / 10.3	17.1 / 10.0	149.3 / 94.4	143.4 / 90.5
Ewald S33	53	Olmesartan	29.1 / 16.5	9.2 / 6.7	162.6 / 96.3	147.8 / 85.7
Karotsis AK34	185	Chlorthalidone Felodipine, Lisinopril, or Valsartan	21.6 / 7.7	14.7 / 5.7	158.6 / 86.1	150.3 / 83.0
Yamagishi T35	54	Azelnidipine	28.2 / 17.2	23.5 / 14.2	165.5 / 99.1	160.1 / 97.5
	54	Amlodipine	23.2 / 14.1	20.5 / 12.2	157.3 / 96.0	154.8 / 95.4
Yamagishi T36	43	Cilnidipine	15.5 / 9.1	11.9 / 6.7	146.3 / 87.5	146.2 / 88.7
	15	Cilnidipine	29.9 / 19.0	19 / 12.1	163.7 / 99.5	150.3 / 95.1
Ashizawa N37	23	Cilnidipine	6.9 / 1.9	11.75 / 3.9	144.3 / 84.1	143.1 / 81.6
Ikeda T38	49	Doxazosin	1.6 / 2.5	7.2 / 4.4	137.9 / 87.7	137.3 / 85.7
Verberk WJ39	214	NA	22 / 11.7	21.8 / 11.9	166.2 / 97.1	156.1 / 92.8
Eguchi K40	308	Doxazosin, Atenolol *3	12.4 / 9.8	10.3 / 7.1	157 / 83	146 / 79
Nakamoto H41	21	Telmisartan	19 / 9	10 / 10	167 / 97	144 / 93
	19	Amlodipine	14 / 10	8 / 6	164 / 95	143 / 91

CBP indicates clinic blood pressure; HBP, home blood pressure; HCZ, Hydrochlorothiazide; Anti-HTN, anti-hypertensive.

^*1was treated based on clinic blood pressure level, and

^*2was treated based on home blood pressure level.

^*3indicate added antihypertensive drugs when the subjects did not achieved target blood pressure levels.

Figure 2. Changes in home and clinic blood pressures.

Data were collected from 30 articles (6794 subjects), and analyzed using a random effects model. Data are shown as mean (standard error). CSBP indicates clinic systolic blood pressure; CDBP, clinic diastolic blood pressure; HSBP, home systolic blood pressure; HDBP, home diastolic blood pressure.

Figure 3. Scatterplot of changes in between clinic and home blood pressures in all studies.

Data were collected from 30 articles (6794 subjects). Regression line, correlation coefficients (r), and regression coefficients (B) were analyzed weighted for number of enrolled subjects. CSBP indicates clinic systolic blood pressure; CDBP, clinic diastolic blood pressure; HSBP, home systolic blood pressure; HDBP, home diastolic blood pressure.

Figure 4. Relationship between baseline blood pressure levels and the reduction.

Data were collected from 30 articles (6794 subjects). Regression line, correlation coefficients (r), and regression coefficients (B) were analyzed weighted for number of enrolled subjects. CSBP indicates clinic systolic blood pressure; CDBP, clinic diastolic blood pressure; HSBP, home systolic blood pressure; HDBP, home diastolic blood pressure.

In the 7 articles in which 24-hr average BPs were available (N=1246),^{18, 20, 22, 25, 26, 32, 39} the changes in clinic, home, and 24-hr ambulatory BPs were -15.2 ± 0.07 / -10.2 ± 0.10 mmHg, -12.6 ± 0.06 / -8.2 ± 0.07 mmHg, -11.9 ± 0.04 / -8.5 ± 0.02 mmHg, respectively, and the differences between home and 24-hr ambulatory systolic BP was significant (1.2 mmHg, 95%CI 0.8-1.5 mmHg, P<0.001). Home systolic BP decreased 6% more than the 24-hr average systolic BP. The reductions of home diastolic BP and 24-hr diastolic BP (8.2 and 8.5 mmHg) were similar, although the difference was statistically significant (0.4 mmHg, 95%CI 0.2 to 0.7 mmHg, P<0.001).

In the 5 articles in which both daytime BP and nighttime BP were separately available (N=801) (excluding 2 articles in which only 24-hr average BP data was available),^{22, 25, 26, 32, 39} the treatment-induced changes are shown in Figure 5. Thus home systolic BP decreased 15% more than daytime ambulatory systolic BP, and 30% more than nighttime ambulatory systolic BP. The differences in home and daytime BP reductions were 1.6 mmHg (95%CI 1.1 to 2.2 mmHg, P<0.001) for systolic and 0.2 mmHg (95%CI -0.4 to 0.8 mmHg, P=0.55) for diastolic, and that in home and nighttime BP reductions were 3.8 mmHg (95%CI 3.3 to 4.4 mmHg, P<0.001) for systolic and 1.2 mmHg (95%CI 0.6 to 1.8 mmHg, P<0.001) for diastolic. Additionally, the treatment induced BP changes corrected for the level of BPs by using the Oldham's correction were 10.3 % for clinic systolic BP, 10.9 % for clinic diastolic BP, 10.1 % for home systolic BP, 10.4 % for home diastolic BP, 8.5 % for daytime systolic BP, 9.0 % for daytime diastolic BP, 8.1 % for nighttime systolic BP, and 9.0 % for nighttime diastolic BP.

Figure 5. Changes in home and daytime and nighttime blood pressures.

Data were collected from 5 articles (801 subjects), and analyzed using a random effects model except for the difference between home DBP and daytime DBP. Data are shown as mean (standard error). CSBP indicates clinic systolic blood pressure; CDBP, clinic diastolic blood pressure; HSBP, home systolic blood pressure; HDBP, home diastolic blood pressure. The difference between home BP and daytime BP are P<0.001 for systolic and P=0.59 for diastolic, and the difference between home BP and nighttime BP are both P<0.001

In the 10 articles in which antihypertensive drugs were taken in the morning (N=1809),^{12-15, 18, 20, 23, 25, 30, 41} the changes in clinic, morning and evening BP were -13.0 ± 0.04 / -11.2 ± 0.04 mmHg, -10.8 ± 0.08 / -6.1 ± 0.09 mmHg, -11.7 ± 0.06 / -7.4 ± 0.06 mmHg. The evening BP was reduced marginally more than morning BP [the difference: 1.1 mmHg (95%CI 0.9 to 1.2 mmHg, P<0.001) for systolic and 1.5 mmHg (95%CI 1.3 to 1.6 mmHg, P<0.001) for diastolic]. In the 4 articles in which antihypertensive drugs were taken in the evening (N=399),^{12, 37, 38, 40} the changes in clinic, morning, and evening BP were -8.2 ± 0.07 / -5.0 ± 0.10 mmHg, -12.5 ± 0.17 / -7.5 ± 0.20 mmHg, -5.6 ± 0.12 / -3.8 ± 0.15 mmHg. The morning BP reduced more than evening BP [the difference: 6.8 mmHg (95%CI 6.57 to 6.95 mmHg, P<0.001) for systolic and 3.6 mmHg (95%CI 3.4 to 3.9 mmHg, P<0.001) for diastolic].

In 6 articles ^{12,16,17, 25,35,41}, the effects of calcium channel blockers (taken in the morning) on the reduction of CBP and HBP were 17.5 ± 0.10 / 12.4 ± 0.13 mmHg and 14.1 ± 0.07 / 9.1 ± 0.08 mmHg, and the differences between clinic and home BP were 3.3 mmHg (95%CI 3.1 to 3.5 mmHg, P<0.001) for systolic and 3.0 mmHg (95%CI 2.8 to 3.3 mmHg). The BP reductions corrected for the level of BPs by using the Oldham's correction were 11.4 % for clinic systolic BP, 13.5 % for clinic diastolic BP, 9.5 % for home systolic BP, and 10.2 % for home diastolic BP. In 5 articles,^{13, 14, 18, 21, 22} the effects of angiotensin converting enzyme inhibitors on the reduction of clinic BP and home BP were 15.1 ± 0.07 / 10.9 ± 0.08 mmHg and 11.9 ± 0.15 / 6.8 ± 0.14 mmHg, and the difference between clinic and home BP were 3.2 mmHg (95%CI 3.0 to 3.4 mmHg) for systolic and 4.3 mmHg (95%CI 3.7 to 4.9 mmHg) for diastolic. The BP reductions corrected for the level of BPs by using the Oldham's correction were 9.9 % for clinic systolic BP, 11.4 % for clinic diastolic BP, 8.4 % for home systolic BP, and 7.6 % for home diastolic BP. In 7 articles,^{18, 20, 22, 24, 33, 41} the effect of angiotensin II receptor blockers on reduction of CBP and HBP were 17.6 ± 0.07 / 10.5 ± 0.11 mmHg and 9.9 ± 0.23 / 6.5 ± 0.14 mmHg, and the difference between clinic and home BP were 8.0 mmHg (95%CI 7.8 to 8.2 mmHg) for systolic and 3.2 mmHg (95%CI 2.7 to 3.8 mmHg) for diastolic. The BP reductions corrected for the level of BPs by using the Oldham's correction were 14.3 % for clinic systolic BP, 11.2 % for clinic diastolic BP, 6.9 % for home systolic BP, and 7.3 % for home diastolic BP. In 2 articles in which there were placebo or control subjects,^{14, 40} the changes in clinic and home BP were -4.5 ± 0.89 / -4.8 ± 0.17 mmHg and -2.1 ± 0.32 / -1.9 ± 0.37 mmHg, respectively (the difference; both P<0.001).

Discussion

The main findings of the present meta-analyses are as follows: (1) the changes produced by antihypertensive drug treatments in home BP were 20% smaller than those of clinic BP, and the changes in clinic BP were linearly related to those of home BP; (2) the difference in the BP reduction between clinic and home BP were attributable to the difference in the baseline BP levels; (3) the changes in home SBP were intermediate between the changes of clinic and ambulatory SBPs (including 24-hr SBP, daytime SBP, and nighttime SBP); and (4) the differing effects on clinic and home BP were similar for calcium channel blockers, angiotensin converting enzyme inhibitors, and angiotensin II receptor blockers, and also for placebo or control groups.

The changes in clinic and home BP (systolic / diastolic) produced by antihypertensive drug treatments were -15.2 / -10.3 mmHg and -12.2 / -8.0 mmHg, respectively, and these reductions of BP were smaller than the values shown in the Verberk's previous meta-analysis¹¹. The changes in home and clinic BPs were linearly related to each other in the present meta-analysis, such that the effect on home BP was typically about 80% of the effect on clinic BP. Mancia and Parati⁴ reported that the reductions of clinic BP and 24-hr BP with antihypertensive treatments were -24.9 / -14.5 mmHg and -14.6 / -9.2 mmHg, respectively, and that the effect on 24-hr ambulatory BP was about 60% of the effect on clinic BP. Taken together, these results indicate that the reduction of home BP is about 20 % smaller, and that of 24-hr BP 40% smaller than the reduction of clinic BP produced by antihypertensive drugs.

The greater reductions of clinic BP than home BP with antihypertensive treatment were attributable to the difference in BP levels at baseline (usually clinic BP > home BP). This indicated that the lower the baseline BP levels are, the smaller the difference in the BP reduction between clinic BP and home BP will be.

In our meta-analysis of 7 studies which included 24-hr BP data as well as home BP,^{18, 20, 22, 25, 26, 32, 39} the reduction of 24-hr systolic BP was indeed less than the reduction of home SBP by 6%. When we analyzed the data in 5 studies in which both daytime and nighttime SBP were available, ^{22, 25, 26, 32, 39} the reductions of 24-hr systolic BP, daytime systolic BP and nighttime systolic BP were about 15%, 15% and 30% smaller than that of home systolic BP. In the meta-analysis of Mancia and Parati,⁴ it was also reported that nighttime BP was reduced less than daytime BP by antihypertensive drugs taken in the morning. Our meta-analysis indicated that the reduction of daytime BP was smaller than that of home BP. The difficulties in reducing nighttime BP might derive from morning dosing of antihypertensive drugs. To obtain nighttime BP control, Kario et al.⁴⁴ previously reported that nighttime dosing of doxazosin, an alpha-adrenergic blocker, is effective. Additionally, Hermida et al.^45-47 have reported that angiotensin II receptor blockers taken at bedtime reduced nighttime BP significantly more than when taken in the morning.

It is not clear why home BP exhibits a smaller reduction than clinic BP and a greater reduction than 24-hr BP following antihypertensive drug treatment. The baseline levels of clinic BP were higher than home BP, and there might be a regression-to-the-mean effect. The regression line of baseline BP levels and the BP reduction were similar between clinic BP and home BP. Even in placebo or control groups, the reductions of clinic BP were larger than home BP. Additionally, there might be a white coat effect (transient BP elevation in the physician's office) which is generally recognized to be reduced by both active and placebo treatment.⁴⁸ Another factor affecting the differences in the reductions of clinic and home BP seems to be the timing of the administration of antihypertensive medications. Antihypertensive drugs taken in the evening reduced morning BP greater than clinic BP.

In some trials the BP decline was larger at home than in the office; in other trials this difference was negligible, and in some the office BP decline was 3 times larger. These differences in the reductions of CBP and HBP might be explained by inaccuracy of CBP measurements, because some studies did not describe the technique of CBP measurements (i.e. the BP devices and number of measurements). Therefore, we analyzed the reductions of CBP and HBP after excluding data sets in which the methods of clinic BP measurement were not clearly described; however, the reductions of CBP and HBP were similar to the results of the whole data set.

Perspectives.

The reduction of home BP produced by antihypertensive drug treatment is about 80% of the magnitude of the reduction of clinic BP.