Blood Pressure Control Among US VeteransClinical Perspective
A Large Multiyear Analysis of Blood Pressure Data From the Veterans Administration Health Data Repository
Background—Hypertension treatment and control remain low worldwide. Strategies to improve blood pressure control have been implemented in the United States and around the world for several years. This study was designed to assess improvement in blood pressure control over a 10-year period in a large cohort of patients in the Department of Veterans Affairs.
Methods and Results—A cohort of 582 881 hypertensive patients and 260 924 normotensive individuals treated in 15 Department of Veterans Affairs medical centers between 2000 and 2010 were examined. Strategies used system-wide included blood pressure control as a performance measure, automatic notification to healthcare providers, electronic reminders, and a systematic revisit schedule. The main outcome measure was the percentage of hypertensive patients whose hypertension was controlled and the level of blood pressure each month. In the hypertensive cohort (mean age 62.9±13.4 years, 96.0% male), 52.3% of patients were white, 25.1% were black, and 21.1% were Hispanic. Blood pressure control rates improved from 45.7% in September 2000 to 76.3% in August 2010. Improvements were similar across ethnic, racial, age, and sex groups. Average systolic/diastolic blood pressure decreased from 142.6/77.1 mm Hg in 2000 to 131.2/74.8 mm Hg in 2010, a decrease of 11.3/2.3 mm Hg (P<0.0001 for both). Systolic and diastolic blood pressures were lower in summer than in winter, and this trend continued through 2010. On average, control rates increased by 3.0% per year and were 6.8% higher in summer than in winter.
Conclusions—High rates of blood pressure control can be achieved in all age and ethnic groups and in both sexes.
Improvement in hypertension treatment and control has become a top priority for many institutions and healthcare organizations.1–6 In the United States, after the landmark studies from the Department of Veterans Affairs (VA) published in the late 1960s and early 1970s,7,8 numerous institution and community-based programs were established to improve awareness, treatment, and control of hypertension. Since then, reports have indicated steady improvement in control rates across the nation. In the VA, hypertension screening and treatment clinics were established, and organized interventions have been implemented system-wide. Numerous approaches have been used to address patient, provider, and healthcare system barriers to facilitate better blood pressure (BP) control. Interventions targeted improvement in patient involvement and compliance, reversal of physician inertia, improvement in lifestyle modifications, and pharmacological therapy. In January 2000, the Department of Health and Human Services launched the Healthy People 2010 initiative, a national health promotion and disease prevention initiative that included aggressive goals for BP control.9–11 Despite these efforts, the number of patients whose hypertension is treated and controlled in the United States and around the world remains low.
Clinical Perspective on p 2468
In the present study, we assessed improvement in BP control in a large population followed up in the VA system. Data were obtained from the VA electronic health records known as VistA (Veterans health Information Systems and Technology Architecture). The study protocol was approved by the VA Institutional Review Board.
The VA's electronic health record system was first deployed in 1982 as the Decentralized Hospital Computer Program. The Computerized Patient Record System, with a user-friendly graphical user interface, was implemented system-wide in 1997. The system is searchable and allows individual medical centers to control the care of the entire patient population. Performance measures to monitor and control quality of care were integrated into the Computerized Patient Record System in 1998. A vital sign package that included BP, pulse rate, respirations, weight, and height was incorporated in the system, and data could be entered directly into the database at the time of each visit. Built-in reminders notified the provider when the last BP was >140/90 mm Hg. Reminders also prompted physicians to treat elevated BP with a combination of medication and lifestyle modification. Patients with elevated BP were seen at frequent intervals until BP was controlled, then follow-up visits were scheduled at 3-month intervals. Patients with resistant or difficult to control hypertension were referred to specialized hypertension clinics. The ability to assess BP control was enhanced in 2000 by a combination of the databases housed at individual VA medical centers into a system-wide database, the Health Data Repository.
All veteran patients receiving health care in VA medical centers in 15 cities (Anchorage, AK; Baltimore, MD; Boston, MA; Chicago, IL; Washington, DC; Fargo, ND; Honolulu, HI; Houston, TX; Los Angeles, CA; Miami, FL; Minneapolis, MN; New York, NY; Philadelphia, PA; Portland, OR; and San Juan, Puerto Rico) with BP readings in the VA Vital Signs Database during the period of October 1999 to August 2010 were included. BP readings before September 2000 were only used to identify patients with hypertension, whereas readings after September 2000 were used to assess BP level and hypertension control rates. Medical centers were chosen to represent a range of geographic areas across the United States with varied climates. Individual patient data files were deidentified and downloaded for analysis on a personal computer running SAS software (version 9.2, Cary, NC).
BP measurements were taken by trained healthcare professionals (nurses, physician assistants, and attending physicians) who were instructed to follow standard procedures for BP measurement. Patients were seated in a comfortable position with the back supported and legs not crossed, and BP was taken after 3 to 5 minutes of rest. BP was taken at least twice if elevated and was entered into the vital sign package. In some centers, it was mandated that BP be rechecked by the attending physician if it was found to be elevated by the clinic nurse. Standard cuffs were used in most patients, but large cuffs were available for patients with large arm circumference. All BP readings, age, race, sex, height, and weight were recorded. Records were reviewed anonymously. An elevated BP reading was defined as a systolic BP >140 mm Hg or diastolic BP >90 mm Hg. Hypertensive patients were defined as having ≥3 days with elevated BP readings, using the lowest reading of the day. Normotensive patients were defined as those patients with ≥3 days of BP readings but no days with BP elevated. Patients with ≥3 readings but only 1 or 2 days with elevated BPs were not classified, and patients with <3 readings were excluded from the present analysis. All months with BP readings were examined for normotensive patients. Only months with readings after the patient was identified as being hypertensive (ie, after the patient's third day with elevated readings) were used for patients with hypertension. Age at entry into the data set was used, ie, on the first day with BP (for normotensive subjects) or on the date of the third elevated reading (for patients with hypertension).
BP readings were dropped if any of the following was true: Missing systolic or diastolic, systolic less than diastolic, systolic >300 mm Hg, systolic <60 mm Hg, diastolic <30 mm Hg, or diastolic >180 mm Hg. For each day, a subject was considered to be normotensive if systolic BP was ≤140 mm Hg and diastolic BP was ≤90 mm Hg. Computation of yearly and seasonal changes in the percentage of hypertensive subjects whose BP returned to normal was based on monthly readings. For every subject, the last reading of each month was used to represent that calendar month. Months with no readings for a given patient were not imputed for that patient. The percentage of patients who were normotensive for each month was calculated based on all patients with readings for that month. Yearly change in the percentage of patients who were normotensive was calculated for the period from September 2000 to August 2010. The confidence interval (CI) for this change was computed with the mean of each month's change between those years (eg, December 2009 minus December 2000, January 2010 minus January 2001). Seasonal variability was quantified by calculating the mean peak-to-trough-to-peak distance for the last 10 years of the study period with the CI estimated using an N of 10 years. The percentage controlled per month was then examined by sex, race (black, Hispanic, white, other), city, and age group (<55, 55–70, >70–80, >80 years). Age was defined at the first BP reading for those with normal BP and at the first BP reading after being identified as hypertensive (ie, the first reading after the third elevated reading) for patients with hypertension. Using the monthly BP values as defined above, we also examined the mean systolic and diastolic BP per month in each group.
Univariate comparisons of year effects were made between subgroups of patients based on age, race, sex, and latitude by use of 2-tailed independent-group t tests. The 95% CIs were calculated with the formula mean±1.96 (SD/√n).
Multivariable modeling techniques were used to examine whether patient age, sex, race, city, or time in treatment had independent effects on BP control rate. For this analysis, the percentage of months controlled in each calendar year was calculated for each hypertensive subject. Multiple regression (using a general linear model as implemented in the GLM procedure in SAS) was used to examine prediction models for the percentage of months when BP was controlled during the last calendar year available for each subject. Predictors included age group (<55, 55 to <70, 70 to <80, and ≥80 years), sex, race, city, and number of months in the data set after the subject was identified as hypertensive. Adjusted final percent elevated (the percentage of months with elevated BP in the final year, adjusted for the other covariates), was obtained by use of the least squares means option in the SAS GLM procedure. In large data sets, changes over time can occur because of different patients being in the cohort at different time points. To determine the significance of within-subject changes over time, repeated-measures ANOVA (with the GLM procedure in SAS) was used to examine changes over time in the percentage of months when BP was controlled. Change from the calendar year in which each subject was identified as being hypertensive to their last calendar year in the data set was examined first. Within-subject changes for subjects in the data set during calendar year 2001 compared with 2009, as well as 2001 versus 2005, were also examined. P<0.05 was considered significant. Data analysis was performed by one of the authors (R.L.A.) using SAS (version 9.2, Cary, NC).
The overall sample included 1.54 million subjects and 53.3 million BP readings from September 2000 through August 2010. Of these, 582 881 patients were identified as hypertensive and 260 924 as normotensive. The remaining patients either had <3 readings, too few to be classified (383 439), or had ≤2 elevated readings (279 191), not enough to qualify them as clearly hypertensive, and were excluded from the present study. At the time point of the month when the patient was identified as having hypertension (the third elevated reading), the mean age was 61.9±13.4 years, and 96.0% were male. Race was not specified in 50.8% of the population. Of those declaring their race, 52.3% were white, 25.1% were black, 21.1% were Hispanic, and 1.5% were classified as other (Asian, Native American). The average number of days with BP readings was 44.7.0±69.8 per subject. Compared with patients with normal BP, patients with hypertension were significantly older, were more likely to be male and black, and had more BP readings (all P<0.0001; Table). There were small increases across years in the proportion of females and younger patients, and there were small changes in the race mix (Table).
BP control rates improved steadily from the year 2000 to 2010 (Figure 1). Among patients with hypertension who had readings in November 2000, 43% had their hypertension controlled (BP <140/90 mm Hg). During the period from September 2000 to August 2001, control rates averaged 46.8%, and during the period from September 2009 to August 2010, they averaged 74.2%. This translates into a yearly improvement in control rates of 3.0% (95% CI, 2.9%–3.2%). In August 2010, 76.3% of patients with hypertension had their hypertension controlled. Improvement in control rates was greater in the earlier period (2000 to 2007; annual improvement 3.7%; 95% CI, 3.1%–4.3%; P<0.0001) than in the latest period (2007–2010; annual improvement 2.0%; 95% CI, 1.9%–2.2%; P<0.0001). Control rates were higher in the summer and lower in the winter. Mean summer versus winter control rates differed by 6.8.% (Figure 1).
Improvement in control rates occurred across all racial/ethnic groups (Figure 2). Initially, BP control was better among white patients, intermediate among blacks, and worse among Hispanic and unspecified groups (difference between all groups, P<0.05). Thereafter, steady improvement in BP control over time was noted in all ethnic groups. After 2004, improvement in control rates accelerated among Hispanics and reached the same levels of control as in whites by the year 2007, then became slightly better by the year 2010. Improvement among blacks was steady but slower and did not reach the same level of control as in the other 2 groups by the year 2010 (P<0.0001 versus whites and Hispanics; Figure 2). The unspecified group followed the same pattern as the black group from 2005 to 2010.
The association of age with BP control rates was also examined (Figure 3). Hypertension in younger patients was better controlled for the first 3 years, but after the year 2004, control rates became similar among all age groups, and curves appeared almost superimposed until 2006. In the last 3 years of follow-up, the middle and older age groups demonstrated better control rates.
The association of sex with control rates was also examined. Although the female hypertensive population (n=23 501) was much smaller than the male hypertensive population (n=559 350), in the year 2000, female patients showed better control rates than male patients (46.7% versus 49.5%, P<0.001). After the year 2004, however, control rates were similar between the 2 sexes until the end of the follow-up period.
Mean changes in systolic and diastolic BP were examined for both the hypertensive and normotensive cohorts (Figure 4). In hypertensive patients, average systolic and diastolic BP decreased (year effect) from 142.6/77.1 mm Hg in 2000 to 131.2/74.8 mm Hg in 2010, a decrease of 11.3/2.3 mm Hg (P<0.0001 for both systolic and diastolic BP). Among normotensive patients, BP increased slightly from 2000 to 2010 (P<0.0001 for both systolic and diastolic BP).
For patients with hypertension who had BP readings during calendar year 2000, a mean of 51.5% (95% CI, 51.3%–51.6%) of months were controlled. This increased to 74.1% (95% CI, 74.0%–74.2%, P<0.001) of months for patients who had BP readings in 2010. Independent predictors of the percentage of months controlled in the final year included city, age group, race, duration of treatment (all P<0.0001), and sex (P=0.002). After accounting for covariates, patients <55, 55 to <70, 70 to <80, and ≥80 years old had adjusted final percent elevated (the percentage of months with elevated BP in the final year) of 32.6%, 29.7%, 29.5%, and 29.3%, respectively. Duration of treatment (months in the data set) was significantly associated with an improved rate of control in the final year, after accounting for other covariates. The adjusted final percent elevated for blacks, Hispanics, and whites was 33.9%, 30.3%, and 28.3%, respectively. The adjusted final percent elevated was 29.9% for females and 30.7% for males.
To evaluate within-subject change over time, the first repeated-measures analysis examined change from each subject's year of being identified as hypertensive compared with their last year in the data set. The time effect was significant (P<0.0001), which indicates that BP control rates improved within patients. On average, the proportion of months with elevated BP changed from 0.53±0.33 in the first year to 0.28±0.36 in the patient's last year (online-only Data Supplement Table I). The second repeated-measures analysis examined all subjects who were present in the data set during calendar year 2001 versus 2009. The proportion of months with elevated BP changed from 0.46±0.39 in 2001 to 0.25±0.31 in 2009, a significant improvement (P<0.0001; online-only Data Supplement Table II). Finally, the last repeated-measures analysis included all hypertensive patients present during 2001 and 2005 and again found the improvement over time to be significant (P<0.0001; proportion elevated changed from 0.47±0.39 in 2001 to 0.33±0.35 in 2005; online-only Data Supplement Table III).
The main finding of the present study was a steady improvement in BP control among all groups of veterans. Overall control rates among patients with hypertension improved from 43.0% in the year 2000 to 76.6% in 2010. Average BP decreased by 11.3/2.3 mm Hg in the same period of time. Similar improvements were noted among all age and ethnic groups and in both sexes. The rate of improvement in BP control was better from 2000 to 2007 than from 2007 to 2010. Although no single intervention can be credited for this impressive improvement in control rates, a number of interventions probably contributed to it substantially. The VA implemented many changes to improve care in the last decade. Among the changes that influence hypertension control, a prominent role is attributable to electronic health records, the ability to provide feedback to patients and healthcare professionals regarding patient appointments, prescription refills, regimen adjustment, and the need to meet existing national performance measures. Automatic reminders in the electronic health record warned all providers when BP was elevated. Another important variable was the ability to provide frequent appointments to patients until BP was under control. This improved patient involvement, led to adherence to medication, and ensured patient compliance. Furthermore, patient involvement was enhanced by providing (free of charge) BP monitors for home measurements, which enabled patients to store and transmit values electronically, and by establishing virtual communication with providers.
Because the group of patients followed up in VA medical centers receive care for free or for very little cost, there were no major issues with access to health care or cost of therapy. Usual and important issues in healthcare delivery, such as poor adherence to medications, clinician inertia, clinician and patient education, and unhealthy lifestyles, were aggressively addressed.
National trends to use nonphysician healthcare professionals such as nurses, physician assistants, nutritionists, and pharmacists as part of the management team were also pursued in many VA medical centers. Control rates achieved in this cohort were better than national data published by Egan et al12 and the National Health and Nutrition Examination Survey 2007 to 2008.12,13 In the most recent National Health and Nutrition Survey,13 data indicate improvement in control rate from 35% to 50%. The 50% control rate is particularly important and meets the Healthy People 2010 objective. In the present study, control rates reached >76% during the summer of 2010. The difference in control rates can be explained in part by the number of patients treated rather than the ability to reduce BP to normal levels in individual patients. In the National Health and Nutrition Examination Survey, even under the best circumstances, only 73% of patients received antihypertensive medication. Data from the Washington, DC, VA medical center indicate that >90% of patients with hypertension are under treatment (data not shown). Thus, identification and treatment of patients with hypertension in the community may be the most efficient way to improve control rates. In the present study, BP improvement continued throughout the study period (2000–2010). Control rates improved in all age groups, including the old and very old (octogenarians), and in all ethnic groups. Blacks lagged slightly behind. In the study by Egan et al,12 the highest percentage of patients treated and not controlled was in the age group ≥60 years. BP control in the older groups, especially among octogenarians, is of particular interest. This subgroup is thought of as difficult to control and more prone to adverse events; however, this subgroup achieved control rates similar to the younger subgroups in the present study. Recent data from the Hypertension in the Very Elderly Trial (HYVET)14 suggest that BP control in this population can result in substantial benefits from prevention of cardiovascular complications. An analysis of mortality data from the present study cohort also suggests continuous improvement in all-cause mortality in all age groups, including the octogenarians (data not shown).15 The elimination of difference in control rates in the Hispanic versus white racial groups is similarly encouraging. BP control improved among blacks but did not achieve the same rate of control as in whites. The reason for this difference is not clear.
Antihypertensive drug use for the 15 hospitals in the present study was not available in the database at the time of this writing; however, data previously published from the national VA pharmacy repository,16 which includes the 15 VA hospitals in the present study, indicate the following changes in drug use from 2000 to 2006: Use of angiotensin-converting enzyme inhibitors increased from 56% in 2000 to 64% in 2006; use of β-blockers increased from 38% to 50%; use of thiazide diuretics increased from 33% to 42%; and use of calcium channel blockers decreased from 43% to 38%. Use of α-blockers and loop diuretics remained unchanged. Use of angiotensin receptor blockers increased from 3% to 12%. The percentage of patients using 3 drugs increased from 23% to 26%, and the percentage of patients using ≥4 drugs increased from 13% to 16%.
Specific information on adherence and compliance to medication was not available for the 15 hospitals included in the present study. During visits, compliance was monitored by verbally asking patients, “Do you take your pills?” Pill count was not performed, and no other objective method of compliance monitoring was used.
Findings from the present study indicate a steady improvement in BP control from 2000 to 2010 irrespective of age, sex, ethnicity, or climate.
None. The principal author of this paper had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.111.029983/-/DC1.
- Received March 3, 2011.
- Accepted March 26, 2012.
- © 2012 American Heart Association, Inc.
- Chobanian AV,
- Bakris GL,
- Black HR,
- Cushman WC,
- Green LA,
- Izzo JL Jr.,
- Jones DW,
- Materson BJ,
- Oparil S,
- Wright JT Jr.,
- Roccella EJ
- Ong KL,
- Cheung BMY,
- Man YB,
- Lau CP,
- Lam KSL
- Falaschetti E,
- Chaudhury M,
- Mindell J,
- Poulter N
- Campbell NR,
- Brant R,
- Johansen H,
- Walker RL,
- Wielgosz A,
- Onysko J,
- Gao R-N,
- Sambell C,
- Phillips S,
- McAlister FA
- Califf RM
National Center for Health Statistics. Healthy People 2000 Final Review. Hyattsville, MD: Public Health Service; 2001.
Healthy People 2010 Web site. Heart disease and stroke. http://www.healthypeople.gov/data/midcourse/default.htm. Accessed February 9, 2011.
- Beckett NS,
- Peters R,
- Fletcher AE,
- Staessen JA,
- Liu L,
- Dumitrascu D,
- Stoyanovsky V,
- Antikainen RL,
- Anderson C,
- Belhani A,
- Forette F,
- Rajkumar C,
- Thijs L,
- Banya W,
- Bulpitt CJ
- Fletcher RD,
- Amdur RL,
- Jones RE,
- Papademetriou CFV
- Furmaga EM,
- Cunningham FE,
- Cushman WC,
- Dong D,
- Jiang R,
- Basile J,
- Katz LA,
- Rutan GH,
- Berlowitz DR,
- Papademetriou V,
- Glassman PA
Despite the proven benefits of treatment of hypertension, control rates remained low in this country and around the world for many years. After the publication of the Veterans Affairs cooperative studies in the late 1960s and early 1970s that demonstrated a substantial reduction of cardiovascular complications with treatment of hypertension, efforts were put forth to identify and treat as many patients as possible. Concerted efforts and organized interventions have been orchestrated by many healthcare organizations to improve blood pressure control. The Healthy People 2010 initiative set and achieved a goal of ≥50% control of patients with hypertension. The Department of Veterans Affairs implemented a series of improvements in healthcare delivery in the late 1990s that included implementation of electronic medical records. The present study was designed to assess improvement of blood pressure control in 15 Veterans Administration Medical Centers across the United States from 2000 to 2010. Organized interventions that used the electronic medical record system, provider reminders, and enhancement of patient-provider interaction made it possible to improve blood pressure control steadily in the last 10 years. Blood pressure control rates improved from 46% in 2000 to 76% in 2010, a remarkable achievement. In some hospitals, control rates were as high as 85%. This improvement was realized across all ethnic and age groups and among both sexes. Preliminary analysis of results indicates substantial benefits on health outcomes. This study also found substantial seasonal differences in blood pressure control rates that may affect clinical outcomes.