Trends in Time to Confirmation and Recognition of New-Onset Hypertension, 2002–2006
Achieving full benefits of blood pressure control in populations requires prompt recognition of previously undetected hypertension. In 2003, the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure provided definitions of hypertension and recommended that single elevated readings be confirmed within 1 to 2 months. We sought to determine whether the time required to confirm and recognize (ie, diagnose and/or treat) new-onset hypertension decreased from 2002 to 2006 for adult members of 2 large integrated healthcare delivery systems, Kaiser Permanente Northern California and Colorado. Using electronically stored office blood pressure readings, physician diagnoses, and pharmacy prescriptions, we identified 200 587 patients with new-onset hypertension (2002–2006) marked by 2 consecutive elevated blood pressure readings in previously undiagnosed, untreated members. Mean confirmation intervals (time from the first to second consecutive elevated reading) declined steadily from 103 to 89 days during this period. For persons recognized within 12 months after confirmation, the mean interval to recognition declined from 78 to 61 days. However, only 33% of individuals were recognized within 12 months. One third were never recognized during observed follow-up. For these patients, most subsequent blood pressure recordings were not elevated. Higher initial blood pressure levels, history of previous cardiovascular disease, and older age were associated with shorter times to recognition. Times to confirmation and recognition of new-onset hypertension have become shorter in recent years, especially for patients with higher cardiovascular disease risk. Variability in office-based blood pressure readings suggests that further improvements in recognition and treatment may be achieved with more specific automated approaches to identifying hypertension.
In 2002, findings of the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial Study,1 reinforced by the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC VII),2 provided a simplified, effective algorithm for initiating and advancing therapy in most newly recognized patients with hypertension. Monitoring and public reporting of blood pressure control for patients with hypertension in commercial and Medicare patients also helped to enhance health plan and physician awareness of blood pressure control.3 Between 1999 and 2004, blood pressure control improved in US adults with hypertension.4–6
However, substantial numbers of patients remain unaware of their elevated blood pressure.5 The full benefits of cardiovascular disease risk reduction through hypertension control will require that all individuals who develop elevated blood pressure be detected and treated promptly. JNC VII recommends that, for most patients with a single elevated blood pressure, a diagnosis of hypertension should be confirmed at a follow-up visit within 1 to 2 months depending on the observed blood pressure levels. Confirmation is complicated in many individuals by great intraindividual variability in blood pressure levels over time7–9; this may contribute to physician reluctance to recognize or initiate treatment in hypertension.9
We sought to determine whether the times to confirmation and recognition of new-onset hypertension decreased from 2002 through 2006 for adult members of 2 large integrated health care delivery systems with new-onset hypertension. Demographic and clinical predictors of the lengths of these 2 intervals were also studied.
This report is derived from the Hypertension Registry of the Cardiovascular Research Network (CVRN). This study was conducted within the Cardiovascular Research Network, a consortium of research organizations affiliated with the HMO Research Network and sponsored by the National Heart, Lung, and Blood Institute. The registry includes all of the patients identified with hypertension between 2000 and 2007 at 3 large integrated healthcare delivery systems: HealthPartners of Minnesota, Kaiser Permanente Colorado, and Kaiser Permanente Northern California. For this report, only the latter 2 systems are included, because electronic capture of outpatient blood pressure data at HealthPartners was not complete until 2003.
We defined hypertension using criteria adapted from a previous study10 based on outpatient blood pressure readings, diagnostic codes from outpatient and hospital records, pharmacy prescriptions, and laboratory results. Patients entered the registry on the date they first met ≥1 of the following criteria: (1) 2 consecutive elevated blood pressure measurements (ie, ≥140 mm Hg systolic and/or 90 mm Hg diastolic or ≥130/80 mm Hg in the presence of diabetes mellitus or chronic kidney disease); (2) 2 diagnostic codes for hypertension (International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes: 401.x to 405.x) recorded on separate dates; (3) 1 diagnostic code for hypertension plus prescription for an antihypertensive medication; or (4) 1 elevated blood pressure measurement plus 1 diagnostic code for hypertension.
Single blood pressure readings are taken by nursing staff (eg, medical assistants) after patients have been seated for 5 minutes in the examination room. Aneroid sphygmomanometers (Welch-Allyn) are the preferred measurement method in Kaiser Permanente Colorado, whereas automated measurement devices (Mindray Medical) are recommended in Kaiser Permanente Northern California. However, there is some use of the other method in each region. Staff are certified in blood pressure measurement at the time of hiring and retrained annually. Equipment is monitored by nursing staff and recalibrated as needed by bioengineering services. We excluded blood pressure readings from emergency and urgent care settings in making initial diagnoses of hypertension, because urgent symptoms could affect blood pressure values and because of uncertainty that the above protocol was consistently followed in emergency situations.
This present report includes only those patients who appeared to have new-onset hypertension between 2002 and 2006. Patients classified as new-onset were required to have ≥12 months of continuous health plan enrollment before meeting any criterion for hypertension. Previous blood pressure values were examined back to 2000, the year blood pressures were first captured in electronic databases. Diagnostic and prescription databases were searched as far back as 1996 or to health plan enrollment if that occurred after 1996. In addition, we did not include the approximately one third of new registry members who entered based on criteria other than 2 consecutive elevated blood pressure readings (ie, criteria 2 to 4 above), because confirmation intervals could not be meaningfully calculated without 2 readings.
To validate electronic blood pressure values and classification as new-onset hypertension, we reviewed charts of 300 randomly selected patients (150 from each site) with apparent new-onset hypertension. The overall agreement on blood pressure values between electronic database and chart records was 98%. A previous diagnosis of hypertension and/or previous use of a medication for hypertension were found in 4% of records, indicating 96% accuracy of the study’s algorithm for excluding preexisting hypertension.
Definition of Confirmation and Recognition Intervals
The confirmation interval is defined as the length of time, in days, between the first and second of the 2 consecutive elevated blood pressures that led to inclusion in the hypertension registry. For analyses of confirmation intervals, patients were assigned to year of the first elevated blood pressure.
The recognition interval is defined as the length of time, in days, between the second elevated blood pressure (date of confirmation and registry entry) and the date on which a diagnosis of hypertension or an initial prescription for an antihypertensive agent was first recorded. Note that this interval ranges from 0 if a diagnosis was made or a prescription issued on the date of confirmation to “never” for patients who were never diagnosed or treated. In analyses of recognition intervals, patients were assigned to the year of the second elevated blood pressure (the confirmation date). Because more than half of confirmed patients were not diagnosed or treated within 12 months, we also evaluated recognition intervals for the subgroup who had 3 (rather than just 2) consecutive elevated blood pressures surrounding entry. This interval began on the date of the third consecutive elevated blood pressure reading.
Predictors of Confirmation and Recognition Intervals
Patient age and sex were available for all of the patients from membership databases. Race/ethnicity was available for 75% of members. In Northern California, race/ethnicity data were obtained primarily from member satisfaction surveys, hospital discharge records, and other research survey data sets. In Colorado, these data were obtained directly from the electronic medical chart, having been entered at the time of registration for outpatient visits. Socioeconomic status was estimated by geocoding patients’ residence address to the 2000 US Census data at the block group level, using the proportion of block group residents with at least a bachelor’s degree as a proxy for individual socioeconomic status. Body mass index (BMI) at registry entry was obtained from the electronic medical chart. BMI was available for 50% of registry members because it was not routinely captured until 2005 in Northern California. Cigarette smoking status (current: yes/no) was recorded at ambulatory visits by medical assistants.
Relevant comorbid diagnoses were sought in inpatient and ambulatory records for the 12 months before registry entry for each patient. Diabetes mellitus was defined by the following: (1) 2 outpatient diagnoses or 1 primary inpatient discharge diagnosis of diabetes mellitus (ICD-9-CM code 250.x); (2) prescription for any antidiabetic medication other than metformin or thiazolidinediones; or (3) prescription for metformin or a thiazolidinedione plus a diagnosis of diabetes mellitus; or (4) hemoglobin A1c value >7% or 2 fasting plasma glucose values ≥7 mmol/L (126 mg/dL) on separate dates. Chronic kidney disease was defined as 2 consecutive serum creatinine values that yield estimated glomerular filtration rates <1 mL/s (60 mL/min) when the Modification of Diet in Renal Disease equation11 is applied or by an ICD-9-CM diagnostic code for chronic kidney disease (ICD-9-CM codes 585.1 to 585.9).
Other comorbid conditions included ischemic heart disease (ICD-9-CM diagnosis codes 410 to 414 or coronary revascularization procedure codes 36.0 and 36.3); stroke (diagnosis codes 431, 433.x, 434.x, and 436.x); peripheral vascular disease (diagnosis codes 441.3 to 441.7, 443.9, 444.0, and 444.2); depression (diagnosis codes 296.2, 296.3, 298.0, 300.4, 301.1, 309.0, 309.1, and 311.x); and congestive heart failure (diagnosis codes 428.x). To adjust for possible changes in background visit rates over time, we calculated annual visit rates to medicine and family medicine departments for all of the adult patients using each of the 109 facilities in the 2 systems. Registry members were linked to the facility where they received most of their outpatient care.
Analyses were restricted to new-onset cases between 2002 and 2006. We excluded new entrants from 2000 to 2001, because the absence of recorded blood pressures before mid-2000 precluded distinction of new-onset from prevalent cases. We excluded persons first identified in 2007 because we had <12 months of follow-up to measure recognition. Confirmation and recognition intervals were examined by year. Cox proportional hazards models were used to test hypotheses that the length of confirmation intervals decreased over the study period after adjusting for secular changes in visit rates and to examine the role of other demographic and clinical factors as predictors of interval length. Because of decreasing follow-up in later years, we censored all of the observations at 12 months after the first elevated pressure. By definition, there was no loss-to-follow-up before confirmation; that is, all of the hypertension registry members were eventually confirmed.
Multivariable logistic regression, rather than proportional hazards, was used to predict recognition within 12 months, because <40% were recognized within 12 months, and of those that were, >50% were recognized on date of registry entry. Approximately 7% of subjects left the health plan, by disenrollment or death, within 1 year and were excluded from the logistic regression analyses. However, a Cox proportional hazards analysis was also constructed for comparison with the model predicting confirmation. Individuals with <1 year of follow-up were included in this model and censored at dropout.
Total numbers of new-onset cases declined annually from 2002 through 2007, suggesting that some cases captured in earlier years would have been identified earlier had blood pressure data been available before mid-2000. Among all of the subjects, mean age and mean systolic blood pressure at entry declined slightly in later years (Table 1), but other demographic and clinical characteristics changed minimally. More than 40% of registry entrants with a recorded BMI in each year met criteria for obesity (BMI >30).
Confirmation of Hypertension by Year
Most patients were confirmed within <12 months (Table 2), and this proportion increased steadily from 2002 through 2006, although the steeper increase for the final year (2006) is likely attributable in part to the fact that there was little observation time beyond 12 months. Among patients confirmed within 12 months, median interval length declined by 19 days. This decline persisted in multivariable analyses (Table 3). Other independent predictors of shorter time to confirmation included higher first systolic and diastolic blood pressures, presence of cardiovascular disease comorbidities, older age, female sex, and nonwhite race/ethnicity (especially black and Asian), but the strengths of these predictive associations were quite modest.
Recognition of Hypertension by Year
Approximately one third of all of the confirmed patients were recognized as having hypertension within 12 months (Table 4). This proportion increased only slightly (from 33.5% to 36.2%) between 2002 and 2006. Among those recognized within a year, the mean interval to recognition declined from 78 days in 2002 to 61 days, and the percentage of patients recognized on the date of confirmation (ie, interval=0) rose from 48.6% to 54.2%.
In the smaller group of patients who had 3 consecutive elevated good pressure readings initially (ie, the 2 required for confirmation and the subsequent reading), the proportion recognized as having hypertension within 1 year was somewhat greater (45% versus 35%). However, year-to-year declines in length of recognition intervals did not differ from those of the larger confirmed group (data not shown).
In logistic regression models, the modest increase in recognition over time was confirmed; however the increase appeared to be almost exclusively in the first 3 years, after which no further increase was seen (Table 5). Clinical and demographic characteristics were more strongly associated with likelihood of being recognized within 12 months than with being confirmed. Initial blood pressure levels (systolic and diastolic) and previous cardiovascular disease comorbidity were the strongest predictors of recognition. Older age, male sex, being Asian or black race/ethnicity, and obesity were also strong predictors of recognition. These multivariable findings were very similar in the proportional hazards model predicting time to recognition within 12 months (data not shown).
Blood Pressure Values After Confirmation
The low rates of recognition led us to look more closely at recorded blood pressure values after initial confirmation for those not recognized within 12 months. We selected all persons (N=61 227) who were confirmed between 2002 and 2004, had ≥3 years of additional follow-up, and ≥5 eligible blood pressure recordings during that 3-year period. For those patients (N=29 373) recognized at some point after 12 months, the average proportions of subsequent readings with an elevated systolic or diastolic pressure were 43.5% and 24.0%, respectively. The majority of subsequent blood pressure readings were elevated in 37.5% of these individuals. For those patients never recognized (N=32 454), only 26.0% and 15.6% of subsequent systolic and diastolic blood pressures, respectively, were elevated, and the majority of readings were elevated in only 14.5% of these individuals. Conversely, for 12.4% of those never recognized, no subsequent recordings (systolic or diastolic) were elevated.
Several previous studies have examined clinician behavior in patients with recognized but poorly controlled hypertension,12–16 but to our knowledge only 213,15 have explored physicians’ likelihood of diagnosing hypertension or initiating therapy. In a 1995 survey, Hyman and Pavlik13 found that <50% of physicians followed JNC V guidelines for treatment initiation in patients with systolic blood pressures of 140 to 160 mm Hg, and 33% did not treat elevations of diastolic blood pressure pharmacologically until they exceeded 94 mm Hg consistently. Asch et al15 found that 31% of persons meeting criteria of 3 elevated blood pressures between 1998 and 2000 were not recognized as hypertensive by their physician. This is the first report of more recent trends in times to confirmation and recognition of hypertension.
Among new-onset hypertensive persons who were confirmed within 12 months of an initial elevated blood pressure reading, the median interval between first and second (confirming) recordings declined from 68 days in 2002 to 49 days in 2006. This trend was not explained by temporal changes in overall visit rates. Consistent with JNC VII guidance, higher initial blood pressure levels were associated with shorter intervals to confirmation. For example, in 2006, persons with initial systolic (or diastolic) blood pressures of ≥160 mm Hg (or ≥100 mm Hg) had confirmation intervals of 28 and 31 days, respectively. The modest associations of demographic and clinical patient characteristics, including older age, nonwhite race/ethnicity, and previous cardiovascular disease diagnoses, with shorter confirmation intervals may also signal some physician attention to getting elevated readings confirmed.
Findings for recognition of hypertension (ie, diagnosis and/or treatment) are more complex and less reassuring. They illustrate that the limitations of office-based blood pressure measurements can create dilemmas for clinicians and healthcare systems in making initial diagnoses of hypertension. Only 33% of confirmed patients were recognized within 1 year. An additional 30% of confirmed patients were recognized after 1 year, and 37% were never recognized during follow-up. Higher levels of systolic and diastolic blood pressures at confirmation and preexisting cardiovascular disease were strong predictors of being recognized. Older age, nonwhite race/ethnicity, and the presence of obesity also increased the likelihood of recognition. These associations suggest that clinicians are making diagnoses and initiating treatment sooner in persons they judge to be at greater risk for actually having hypertension and for developing cardiovascular disease. The finding of shorter intervals for nonwhite patients is consistent with previous studies showing that physicians are more aggressive in intensifying antihypertensive therapy in nonwhite patients.16,17 Interestingly, one important cardiovascular disease risk factor, cigarette smoking, had no apparent impact on recognition of hypertension.
Factors that may have contributed to apparent failures in recognizing hypertension include physician awareness of the known variability in blood pressure measures within individuals7,8 and observation of such variation in some patients; suspicion of “white coat hypertension”18 coupled with a false belief that it is benign19; awareness of the lower average values and superior predictive ability of ambulatory blood pressure monitoring over office-based measures20,21 for cardiovascular events; and reluctance to label patients with milder blood pressure elevations as hypertensive. Importantly, only a minority (26%) of subsequent office-based blood pressure recordings in patients who went unrecognized were elevated despite the absence of antihypertensive therapy. For some of these patients, lifestyle changes, possibly encouraged by physicians, may have led to a return to normal blood pressures. Nevertheless, ≈15% of these patients continued to have elevated blood pressures. Epidemiological evidence suggests that these patients, especially those with other cardiovascular disease risk factors or signs of early target organ damage, would have benefited from earlier recognition.22,23 Newer evidence suggests further that greater visit-to-visit variation in blood pressure and episodic elevations of blood pressure insufficient to meet diagnostic criteria may nevertheless identify an increased risk for cardiovascular disease.24 Thus, failure to recognize hypertension in these patients cannot be easily dismissed as good clinical judgment.
Our findings also point out difficulties that would be encountered by healthcare systems in using office-based blood pressure readings from electronic health records to improve hypertension recognition and control. These difficulties may explain the paucity of existing hypertension registries compared with the large numbers of diabetes mellitus and cardiovascular disease registries.25,26 Nevertheless, they do suggest that electronic health records could offer opportunities for reducing simple clinician oversight in detecting undiagnosed hypertension and for development of more complex decision rules for when to diagnose and treat. For example, patients with ≥2 consecutive elevated blood pressure readings, or possibly the subset with additional cardiovascular disease risk factors, could be sought electronically. Alerts to physicians or to associated nonphysician staff before a next visit could help insure that these patients are evaluated or monitored more closely.
This study has several limitations. Despite systemwide efforts to ensure the accuracy of office-based blood pressure measures, some degree of mismeasurement undoubtedly occurred. Imprecision in measuring the initial blood pressures would have exaggerated the apparent tendency for subsequent readings to be normal. The study is also potentially affected by both left and right censoring. Blood pressures were not recorded until 2000. Although we did not begin assessing recognition until 2002, patients identified in earlier years may have been more likely to have had previous elevated readings, which could make physicians more likely to diagnose and treat. This bias should diminish the trends toward prompter recognition that we observed over time. Right censoring, because of health plan disenrollment, was addressed by limiting follow-up to 12 months for all of the patients. Only 7% of subjects left the plan within <12 months.
This study suggests that surveillance of recorded blood pressures in electronic health record databases could be useful for improving early detection and treatment of hypertension. However, the variability inherent in office-based blood pressure readings will require that more sophisticated approaches than the simple application of JNC VII recommendations used here be developed. These approaches could incorporate larger numbers of blood pressure recordings over longer periods of time or could take other cardiovascular disease risk factors into account.
In 2 large integrated delivery systems, time to confirmation and recognition of new-onset hypertension decreased between 2002 and 2006. Higher initial blood pressure levels and other indicators of cardiovascular disease risk were strongly associated with shorter times to recognition. However, for one third of persons who met the hypertension definition of 2 consecutive elevated blood pressure readings, neither a diagnosis of hypertension nor treatment was observed to occur. In most of these patients, subsequent blood pressure recordings did not suggest sustained hypertension.
We thank La Tanya D. Jackson for excellent editorial assistance.
Sources of Funding
This study was supported by the National Heart, Lung, and Blood Institute (award No. U19HL091179). J.A.S. is supported by the Office of Research in Women’s Health Building Interdisciplinary Careers in Women’s Health K12 Career Development Award (K12HD052163). P.M.H. is supported by a Veterans’ Affairs Research and Development Career Development Award (05-026-2).
P.M.H. serves as a consultant for Wellpoint, Inc.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
- Received March 18, 2010.
- Revision received April 1, 2010.
- Accepted July 27, 2010.
ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group, The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic: the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA. 2002; 288: 2981–2997.
Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ; National Heart, Lung, and Blood Institute Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure; National High Blood Pressure Education Program Coordinating Committee. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. Hypertension. 2003; 42: 1206–1252.
Eddy DM, Pawlson LG, Schaaf D, Peskin B, Shcheprov A, Dziuba J, Bowman J, Eng B. The potential effects of HEDIS performance measures on the quality of care. Health Aff (Millwood). 2008; 27: 1429–1441.
Cutler JA, Sorlie PD, Wolz M, Thom T, Fields LE, Roccella EJ. Trends in hypertension prevalence, awareness, treatment, and control rates in United States adults between 1988–1994 and 1999–2004. Hypertension. 2008; 52: 818–827.
Keenan K, Hayen A, Neal BC, Irwig L. Long term monitoring in patients receiving treatment to lower blood pressure: analysis of data from placebo controlled randomised controlled trial. BMJ. 2009; 338: b1492.
Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, Kusek JW, Van Lente F; Chronic Kidney Disease Epidemiology Collaboration. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006; 145: 247–254.
Traylor A, Subramanian U, Uratsu CS, Mangione CM, Selby JV, Schmittdiel JA. Patient race/ethnicity and patient-physician race/ethnicity concordance in the management of CVD risk factors for patients with diabetes. Diabetes Care. 2010; 33: 520–525.
Mancia G, Facchetti R, Bombelli M, Grassi G, Sega R. Long-term risk of mortality associated with selective and combined elevation in office, home, and ambulatory blood pressure. Hypertension. 2006; 47: 846–853.
Staessen JA, Thijs L, Fagard R, O'Brien ET, Clement D, de Leeuw PW, Mancia G, Nachev C, Palatini P, Parati G, Tuomilehto J, Webster J; for the Systolic Hypertension in Europe Trial Investigators. Predicting cardiovascular risk using conventional vs. ambulatory blood pressure in older patients with systolic hypertension. JAMA. 1999; 282: 539–546.
Banegas JR, Segura J, Sobrino J, Rodríguez-Artalejo F, de la Sierra A, de la Cruz JJ, Gorostidi M, Sarría A, Ruilope LM; for the Spanish Society of Hypertension Ambulatory Blood Pressure Monitoring Registry Investigators. Effectiveness of blood pressure control outside the medical setting. Hypertension. 2007; 49: 62–68.
Igho-Pemu P, Quarshie A, Oduwole A, Lapu-Bula R, Mayberry R, Obialo C, Gibbons GH, Ofili E. Morehouse Community Physicians Network (CPN) hypertension registry: patterns of care and opportunities for targeted medical education. Ethn Dis. 2005; 15 (suppl): S5– 120–123.