Association of Cardiovascular Outcomes With Masked Hypertension Defined by Home Blood Pressure Monitoring in a Japanese General Practice Population

Key Points

Question  Is masked hypertension defined by home blood pressure monitoring associated with stroke or coronary heart disease events in a clinical setting?

Findings  This cohort study of 4261 outpatients observed that the group with masked hypertension had a greater risk for stroke compared with the group with controlled blood pressure levels, independent of traditional cardiovascular risk factors, urine albumin to creatinine ratio, and circulating B-type natriuretic peptide levels. Conversely, masked hypertension showed no association with coronary heart disease risk.

Meaning  Finding that masked hypertension is associated with an increased risk for stroke may improve the assessment of blood pressure–related risks and identify interventions for cardiovascular disease.

Importance  The clinical outcomes associated with masked hypertension defined by home blood pressure monitoring (HBPM) in clinical settings remain uncertain.

Objective  To assess the association between masked hypertension and cardiovascular disease events in clinical settings.

Design, Setting, and Participants  This observational cohort study used data from 4261 outpatients treated at 71 primary practices or university hospitals throughout Japan who were enrolled in the Japan Morning Surge–Home Blood Pressure study between January 1, 2005, and December 31, 2012. Participants had a history of or risk factors for cardiovascular disease and were followed up through March 31, 2015. Participants underwent clinic blood pressure (BP) measurements on 2 occasions as well as HBPM measurements in the morning and evening for a 14-day period. Urine albumin to creatinine ratio and circulating brain (or B-type) natriuretic peptide levels were quantified at baseline as a marker of cardiovascular end-organ damage. Data were analyzed from July 1, 2017, to October 31, 2017.

Exposures  Participants were categorized into 4 BP groups: (1) masked hypertension—hypertensive home BP levels (systolic, ≥135 mm Hg; diastolic, ≥85 mm Hg) and nonhypertensive clinic BP levels (systolic, <140 mm Hg; diastolic, <90 mm Hg); (2) white-coat hypertension—nonhypertensive home BP levels (systolic, <135 mm Hg; diastolic, <85 mm Hg) and hypertensive clinic BP levels (systolic, ≥140 mm Hg; diastolic, ≥90 mm Hg); (3) sustained hypertension—hypertensive home and clinic BP levels; and (4) controlled BP—nonhypertensive home and clinic BP levels.

Main Outcomes and Measures  Incident stroke and coronary heart disease.

Results  Of the 4261 participants, 2266 (53.2%) were women, 3374 (79.2%) were taking antihypertensive medication, and the mean (SD) age was 64.9 (10.9) years. During a median (interquartile range) follow-up of 3.9 (2.4-4.6) years, 74 stroke (4.4 per 1000 person-years) and 77 coronary heart disease (4.6 per 1000 person-years) events occurred. The masked hypertension group had a greater risk for stroke compared with the controlled BP group (hazard ratio, 2.77; 95% CI, 1.20-6.37), independent of traditional cardiovascular risk factors, urine albumin to creatinine ratio, and circulating B-type natriuretic peptide levels. Conversely, masked hypertension yielded no association with coronary heart disease risk.

Conclusions and Relevance  In the Japanese general practice population, masked hypertension defined by HBPM may be associated with an increased risk for stroke events. Use of HBPM may improve the assessment of BP-related risks and identify new therapeutic interventions aimed at preventing cardiovascular disease events.

Clinic blood pressure (BP) may not accurately reflect the BP levels that patients experience in their daily environment outside the clinic (eg, at home, at work, and during sleep).^1-4 Of importance, 30% to 50% of patients with hypertension whose BP levels appear well controlled when measured in a clinical setting have high BP outside the physician’s office.^5-8 This phenomenon of hypertension in which the diagnosis is missed in a clinical setting is known as masked hypertension.^9,10 Masked hypertension confers a 2-fold higher risk of cardiovascular disease (CVD) compared with normotension both in and out of the clinic.^11-13

For out-of-clinic BP measurements, 2 techniques have been developed: ambulatory BP monitoring (ABPM) and home BP monitoring (HBPM).^2-4 The standard method for diagnosing masked hypertension is ABPM. However, ABPM has constraints, including the limited availability and affordability of monitoring devices and low tolerability by examinees as a result of sleep disturbances, discomfort, and restrictions in daily activities.^2,14 Compared with ABPM, HBPM is more acceptable and better tolerated by examinees.¹⁴ Increasing evidence shows that, in community-based populations, adverse cardiovascular outcomes are associated with masked hypertension defined by HBPM.^12,15 However, little is known regarding the outcomes associated with masked hypertension in clinical settings. Masked hypertension is associated with cardiovascular end-organ damage (ie, albuminuria and cardiac hypertrophy).^12,16-19 Therefore, masked hypertension may confer an increased risk of CVD events only because it reflects the presence or severity of cardiovascular end-organ damage. However, few studies have investigated whether the CVD risk associated with masked hypertension is independent of the extent of cardiovascular end-organ damage.

We are uniquely positioned to address this gap in knowledge by using data from the Japan Morning Surge–Home Blood Pressure (J-HOP) study.²⁰ The J-HOP study enrolled outpatients with history of or risk factors for CVD and conducted HBPM in the morning and evening during a 14-day period. We examined the following: (1) the association between masked hypertension and CVD events and (2) whether the association is independent of the extent of cardiovascular end-organ damage, including the urine albumin to creatinine (Cr) ratio (UACR) and the circulating brain (or B-type) natriuretic peptide (BNP) level.^21,22

The rationale, design, and procedures of the J-HOP study have been published previously (eMethods in the Supplement).²⁰ Briefly, 4310 outpatients with history of or risk factors for CVD were enrolled between January 1, 2005, and December 31, 2012, from 71 primary practices or university hospitals throughout Japan. The participants were followed up through March 31, 2015. The institutional review board of Jichi Medical University School of Medicine (Shimotsuke, Japan) approved the methods of the J-HOP study. All of the patients provided written informed consent to participate and to have their data published. The present study was conducted to fulfill one of the overall aims of the J-HOP study and used the preexisting data set from the study. Thus, the institutional review board of Jichi Medical University School of Medicine did not require a separate approval for this analysis.

The J-HOP study methods are described in the eMethods in the Supplement. Three clinic BP readings were taken at 15-second intervals on 2 different occasions, and the mean of the 6 readings was defined as the clinic BP level. Self-measured home BP levels were obtained according to the Japanese BP guideline.²³ Three home BP readings were taken at 15-second intervals, with patients in a seated position in both the morning (within 1 hour of waking and before taking antihypertensive medication) and the evening (before going to bed) for 14 consecutive days. The first day’s home BP measurements were excluded, and the mean (SD) of the remaining morning (35.1 [7.2] readings) and evening (33.3 [8.2] readings) BP levels were calculated separately. The clinic and home BP levels were obtained using the same validated, automatic, and oscillometric device (HEM-5001; Omron Healthcare).²⁴ To prevent reporting bias, BP data were automatically stored in the memory of the device and were downloaded to a computer by a physician or nurse during the clinic visits.

Laboratory methods are described in the eMethods in the Supplement. The spot UACR and plasma BNP levels were assessed at baseline. Diabetes mellitus was defined as self-reported history of a physician’s diagnosis, diabetes medication use, or a fasting blood glucose level of 126 mg/dL or higher or a nonfasting glucose level of 200 mg/dL or higher (to convert glucose levels to millimoles per liter, multiply by 0.0555). History of CVD events, including angina pectoris, myocardial infarction, and stroke, was ascertained at baseline. The characteristic attitude and symptoms of depression in 4001 participants were evaluated using the Beck Depression Inventory (BDI) test,²⁵ and sleep duration in 3772 participants was evaluated using a questionnaire.

A previous report on the J-HOP study²⁰ indicated that the CVD risk associated with home BP levels differed according to whether the home BP readings were obtained in the morning or evening. Morning home BP levels were associated with adverse cardiovascular outcomes, whereas evening home BP levels were not. Therefore, only morning home BP levels were used to define the 4 BP groups (masked hypertension group, white-coat hypertension group, sustained hypertension group, and controlled BP group). Masked hypertension was defined as hypertensive home BP levels (systolic BP [SBP], ≥135 mm Hg; diastolic BP [DBP], ≥85 mm Hg) and nonhypertensive clinic BP levels (SBP, <140 mm Hg; DBP, <90 mm Hg), white-coat hypertension was defined as nonhypertensive home BP levels (SBP, <135 mm Hg; DBP, <85 mm Hg) and hypertensive clinic BP levels (SBP, ≥140 mm Hg; DBP, ≥90 mm Hg), sustained hypertension was defined as hypertensive home and clinic BP levels, and controlled BP was defined as nonhypertensive home and clinic BP levels. In a secondary analysis, we used morning and evening home BP levels to define the 4 BP groups.

Each participant’s vital status was ascertained through March 31, 2015. Incident CVD events during follow-up, including coronary heart disease (CHD) (fatal and nonfatal coronary artery disease and sudden death within 24 hours of the abrupt onset of symptoms) and stroke (fatal and nonfatal), were assessed as outcomes. The J-HOP study²⁰ reported that a higher home BP level was associated with an increased risk for stroke but not for CHD. Therefore, stroke and CHD were evaluated separately. If events occurred on 2 or more occasions, the first occurrence was included in the analysis. Cardiovascular disease events were ascertained by ongoing reports from a general physician at each participating institution. Follow-up time was censored on the data of event ascertainment. Participants who did not experience CVD events were censored at the final study visit. Additional details are given in the eMethods of the Supplement.

Data were analyzed from July 1, 2017, to October 31, 2017. Descriptive statistics are presented as means (SDs), proportions, and medians (interquartile ranges [IQRs]) where appropriate. The characteristics of the 4261 J-HOP study participants who were included in the present study were compared with the characteristics of 49 individuals who were not included using the unpaired, 2-tailed t test or Fisher exact test. Demographic variables and clinical and behavioral characteristics were compared across the 4 BP groups using 1-way analysis of variance (for continuous variables) or Fisher exact test (for categorical variables). The Holm correction²⁶ was used for multiple comparisons. Kruskal-Wallis tests were used to compare differences in UACR and circulating BNP levels across the 4 BP groups, with the Steel-Dwass test used for post hoc multiple comparisons. The Kaplan-Meier cumulative incidence of stroke or CHD events by the 4 BP groups was calculated. With the use of Cox proportional hazards regression models, the hazard ratios (HRs) and 95% CIs of stroke or CHD events associated with masked hypertension, white-coat hypertension, and sustained hypertension (vs the controlled BP group) were calculated. The proportionality assumption for the Cox proportional hazards regression analyses was confirmed graphically and via the inclusion of a time × BP interaction. Hazard ratios were calculated in an unadjusted model (model 1), after adjustment for covariates (model 2), and after additional adjustment for UACR and BNP levels (model 3). Model 2 included adjustment for demographic variables (age and sex) and clinical and behavioral characteristics at baseline (body mass index [calculated as weight in kilograms divided by height in meters squared]; smoking status; prevalent diabetes status; prevalent angina pectoris, myocardial infarction, or stroke; total cholesterol level; high-density lipoprotein cholesterol level; and statin or antihypertensive medication use). These covariates were selected a priori because they have been reported to show correlations with both BP level²⁷ and CVD risk²⁸ and could potentially confound the association between BP level and CVD risk. For models 2 and 3, we adjusted for a composite risk score rather than individual covariates because of concerns of overfitting the models as a result of the limited number of outcomes among J-HOP participants. The composite risk score was created in the overall J-HOP population by determining the 4-year predicted probabilities for CVD using Cox proportional hazards regression models, including the covariates used for models 2 and 3. Sleep duration and BDI test score were not included in the composite risk score because of their missing values. In a sensitivity analysis, we adjusted for sleep duration and BDI test scores.

We conducted sensitivity analyses by (1) using the mean morning and evening home BP levels to define the 4 BP groups and (2) considering home and clinic BP levels (as a continuous variable) jointly in stroke or CHD prediction models. This approach allowed us to assess whether home BP levels have a stronger association with stroke or CHD events compared with clinic BP levels. Two-sided P < .05 was statistically significant. All statistical analyses were performed with R, version 3.3.1 (The R Foundation for Statistical Computing) and SAS, version 9.4 (SAS Institute Inc).

Among the 4310 participants, 32 participants who were lost to follow-up and 17 patients who had missing UACR or BNP data were excluded, leaving a sample of 4261 participants for analysis (Table 1). The included participants (of whom 2266 [53.2%] were women, 3374 [79.2%] were taking antihypertensive medication, and the mean [SD] age was 64.9 [10.9] years) showed a lower prevalence of male sex and higher BNP levels compared with those not included in this study (n = 49) (eTable 1 in the Supplement). Table 1 provides demographic variables and clinical and behavioral characteristics overall and by the 4 BP groups. In the overall J-HOP population, the mean (SD) clinic SBP level was 141.3 (16.4) mm Hg and DBP level was 81.2 (10.6) mm Hg, whereas the home SBP level was 138.4 (15.8) mm Hg and DBP level was 79.1 (10.0) mm Hg. The prevalence of masked hypertension was 19.0% among the overall population and 40.8% among participants with a normal clinic BP level (Figure 1). Age (mean [SD], 65.8 [10.7] years vs 63.1 [10.5] years), body mass index (mean [SD], 24.5 [3.5] vs 24.0 [3.4]), diabetes status (239 of 810 [29.5%] vs 259 of 1177 [22.0%]), and statin use (201 of 810 [24.8%] vs 320 of 1177 [27.2%]) were higher in the masked hypertension group compared with the controlled BP group. The UACR and BNP level in the masked hypertension group (UACR: median [IQR], 13.4 [7.7-27.3] mg/g Cr; BNP: median [IQR], 19.7 [9.8-40.8] pg/mL) were greater compared with the levels in the controlled BP group (UACR: median [IQR], 9.3 [6.1-18.4] mg/g Cr; BNP: median [IQR], 15.8 [7.9-32.4] pg/mL) and white-coat hypertension group (UACR: median [IQR], 12.0 [7.1-24.8] mg/g Cr; BNP: median [IQR], 17.7 [9.2-32.7] pg/mL) but were lower than the levels in the sustained hypertension group (UACR: median [IQR], 17.5 [9.0-46.8] mg/g Cr; BNP: median [IQR], 20.4 [10.3-43.3] pg/mL) (to convert BNP levels to nanograms per liter, multiply by 1.0).

The incidence rate of stroke or CHD events and the Kaplan-Meier cumulative incidence were determined by the 4 BP groups (Table 2 and Figure 2). During a median (IQR) follow-up of 3.9 (2.4-4.6) years (16 875 person-years), 74 stroke (4.4 per 1000 person-years) and 77 CHD (4.6 per 1000 person-years) events occurred. The masked hypertension group had a higher incidence rate of stroke events compared with the controlled BP group (5.4 per 1000 person-years; 95% CI, 3.4-8.5 vs 1.7 per 1000 person-years; 95% CI, 0.9-3.4). Conversely, the incidence rate of CHD was similar across the groups.

With adjustments for covariates, results from the Cox proportional hazards regression models suggested that both the masked (HR, 2.66; 95% CI, 1.15-6.13) and sustained (HR, 3.43; 95% CI, 1.61-7.30) hypertension groups had a greater risk for stroke compared with the controlled BP group (model 2; Table 2). We further adjusted for UACR and circulating BNP levels, and the results were similar (model 3; masked hypertension group HR, 2.77; 95% CI, 1.20-6.37 vs sustained hypertension group HR, 3.34; 95% CI, 1.56-7.13). Masked and sustained hypertension yielded no association with CHD events (Table 2). White-coat hypertension yielded no association with stroke or CHD events.

When the 4 BP groups were defined by the mean morning and evening home BP levels, the results were similar (eTable 2 in the Supplement). We adjusted for sleep duration (eTable 3 in the Supplement) and BDI test scores (eTable 4 in the Supplement), and the results were similar.

When clinic and home SBP levels (as continuous variables) were analyzed jointly, only home SBP levels (HR, 1.52; 95% CI, 1.22-1.89) were associated with stroke risk (Table 3). Neither clinic nor home SBP and DBP levels were associated with CHD risk.

In this nationwide, practice-based study of 4261 Japanese with history of or risk factors for CVD, we demonstrated that masked hypertension defined by HBPM (vs nonhypertensive BP level in the clinic and home) was associated with (1) greater cardiovascular end-organ damage, including UACR and circulating BNP levels, and (2) an increased risk for stroke events independent of the extent of cardiovascular end-organ damage. The magnitudes of cardiovascular end-organ damage and stroke risk in the masked hypertension group were intermediate between those of the controlled BP and sustained hypertension groups. Masked hypertension was not associated with CHD risk. White-coat hypertension yielded no association with stroke or CHD events.

The Self-measurement of Blood Pressure at Home in the Elderly: Assessment and Follow-up (SHEAF) study recruited Europeans 60 years or older who had been treated with antihypertensive medication in a clinical setting (4939 participants; mean age, 70 years; 2413 men [48.9%]).¹¹ Home BP readings were obtained in the morning and evening for 3 to 4 days. During a mean follow-up of 3.2 years, masked hypertension was associated with an increased risk for composite CVD events (HR, 2.06; 95% CI, 1.22-3.47) compared with nonhypertensive BP level in the clinic and home. Conversely, white-coat hypertension was not associated with CVD risk. The International Database of Home Blood Pressure in Relation to Cardiovascular Outcome (IDHOCO) study incorporated 5 community-based studies from Japan,^29,30 Greece,³¹ Finland,³² and Uruguay (Montevideo)³³ (6458 participants; 1453 [22.5%] received antihypertensive medication). In the IDHOCO study, home BP readings were obtained in the morning or evening for 1 to 26 days (≥7 days in 84% of the IDHOCO participants).

Investigators conducted stratified analysis by the presence or absence of antihypertensive medication use and reported that masked hypertension conferred greater CVD risk compared with nonhypertensive BP level in the clinic and home (HR, 1.76; 95% CI, 1.23-2.53) among participants taking antihypertensive medication (n = 1451).¹⁵ Conversely, white-coat hypertension was not associated with CVD risk in IDHOCO participants taking antihypertensive medication.

In the present study, we extend these findings by demonstrating that, among middle-aged and older Japanese being treated for hypertension and/or other cardiovascular risk factors, (1) the masked hypertension group had a higher risk for stroke events compared with the controlled BP group (ie, nonhypertensive BP level in the clinic and home) and a similar risk as the sustained hypertension group, and (2) the white-coat hypertension group had a risk for stroke or CHD events similar to that of the controlled BP group. Analyses including home and clinic SBP levels jointly (as a continuous variable) supported the conclusion that home SBP levels were associated with stroke risk, whereas clinic SBP levels were not.

Cross-sectional association between masked hypertension defined by HBPM and cardiovascular end-organ damage, including albuminuria and cardiac hypertrophy, has been demonstrated.^12,18 Baseline cardiovascular end-organ damage was not assessed in the SHEAF¹¹ and IDHOCO¹⁵ studies. Therefore, whether masked hypertension is a potential contributor to adverse cardiovascular events or merely an epiphenomenon of cardiovascular damage remains to be determined. Modeling adjustments for UACR and circulating BNP levels did not materially change our present findings, but possible residual confounders may have affected the masked hypertension–stroke risk association. Sodium intake, lower socioeconomic status, job strain, psychosocial stress (eg, anxiety disorders and anger), and greater BP level variability could be associated with both masked hypertension and incident stroke events.^34-40 Whether the increased stroke risk of masked hypertension was solely attributable to home BP level elevation is unclear. Unraveling the mechanistic links between masked hypertension and adverse outcomes will require further investigation.

In the present study, masked hypertension conferred no CHD risk. This finding is consistent with a previous report from the J-HOP study indicating that home SBP and DBP levels were not associated with CHD events.²⁰ The previous masked hypertension studies using HBPM did not evaluate risk associations by distinct cardiovascular outcomes (ie, CHD and stroke, separately).^11,12,15 However, the IDHOCO study reported that home SBP levels were associated with an increased risk for stroke but not CHD events in participants taking antihypertensive medications.⁴¹ Conversely, higher home SBP levels were associated with CHD risk among IDHOCO participants not taking antihypertensive medication.⁴¹ Because of the limited number of participants not taking antihypertensive medication in the J-HOP study, we were not able to discern whether the association between masked hypertension and CHD risk differs by the presence or absence of antihypertensive medication use.

The Community Preventive Services Task Force recommends the use of HBPM for confirmation and treatment of hypertension.⁴² The 2017 American College of Cardiology/American Heart Association “Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults” recommends screening for masked hypertension using ABPM or HBPM to guide the initiation and intensification of antihypertensive medication.⁴³ Our findings support the application of HBPM to identify adults with masked hypertension who are at high risk for stroke events. However, evidence is sparse regarding whether nonpharmacological and pharmacological interventions in adults with masked hypertension reduce CVD events. Further studies are thus warranted to assess whether reductions in home BP level can help to prevent CVD events among adults with masked hypertension.⁴⁴

The strengths of this study include its nationwide scope, the application of standard clinic and home BP measurements, and the high patient retention rate. However, the study has limitations. First, because this was an observational study, we were unable to determine causality in the findings. Second, 79.2% of our recruited patients were taking antihypertensive medication, and we were unable to assess the association between outcomes and the changes in BP phenotypes and the use of antihypertensive medication during follow-up. Hypertension treatment based on clinic and home BP levels is recommended for all Japanese physicians.^23,45,46 These factors could potentially dilute any true association between each BP phenotype and stroke or CHD risk. Third, we defined masked hypertension using home BP levels in the morning. In this study, CVD risk associated with masked hypertension appeared to be attenuated when we defined masked hypertension using the mean morning and evening home BP levels. We did not include midday BP level in defining masked hypertension. It remains uncertain whether clinical outcome associated with masked hypertension differs when defined using morning, midday, and evening home BP measurements vs morning home BP level alone. How HBPM should be performed to better identify the masked hypertension group at high risk for CVD events merits further investigation. Fourth, we evaluated UACR and circulating BNP levels as a marker of cardiovascular end-organ damage. However, UACR and circulating BNP levels may not accurately reflect the extent of cardiovascular end-organ damage. Further studies are warranted to investigate whether the association between masked hypertension and stroke risk is independent of rigorous cardiovascular end-organ damage markers, including brachial-ankle pulse wave velocity and left ventricular mass index on echocardiography. Fifth, our findings may not be generalizable to other racial/ethnic groups.

An increased stroke risk may be associated with masked hypertension defined by HBPM in the Japanese general practice population. Masked hypertension could be missed if out-of-clinic BP level is not evaluated. Home BP monitoring may emerge as a complementary prevention and treatment strategy in clinical practice and could help reduce the public health burden of CVD. Further studies are warranted to assess whether reductions in home BP level can help prevent CVD events among individuals with masked hypertension. This hypothesis will need to be confirmed in interventional trials, with consideration of all of the complex issues at play, including cost-effectiveness and patient perspectives.

Accepted for Publication: March 30, 2018.

Correction: This article was corrected on June 27, 2018, to fix incorrect units of measure in Table 2 and corresponding text in the Results section.

Corresponding Author: Kazuomi Kario, MD, PhD, Division of Cardiovascular Medicine, Department of Medicine, Jichi Medical University School of Medicine, 3311-1, Yakushiji, Shimotsuke, Tochigi 329-0498, Japan (kkario@jichi.ac.jp).

Published Online: May 23, 2018. doi:10.1001/jamacardio.2018.1233

Author Contributions: Drs Fujiwara and Kario had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Fujiwara, Yano, Hoshide, Kario.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Fujiwara, Yano, Hoshide, Kario.

Critical revision of the manuscript for important intellectual content: Fujiwara, Yano, Kanegae, Kario.

Statistical analysis: Fujiwara, Hoshide, Kanegae.

Obtained funding: Kario.

Administrative, technical, or material support: Hoshide, Kario.

Study supervision: Yano, Hoshide, Kario.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This study was funded in part by the 21st Century Center of Excellence Project, Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan; a grant from the Foundation for Development of the Community (Tochigi); a grant from Omron Healthcare Co, Ltd; grant 21390247 of Grant-in-Aid for Scientific Research (B) from MEXT, 2009-2013; the MEXT-Supported Program for the Strategic Research Foundation at Private Universities, 2011-2015; grant S1101022 from the Cooperative Basic and Clinical Research on Circadian Medicine (Dr Kario); and award number P20GM104357 from the National Institute of General Medical Sciences, National Institutes of Health (Dr Yano).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed herein are those of the authors and do not reflect the official policy or position of the National Institutes of Health.