PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Am J Hypertens. Author manuscript; available in PMC 2013 March 18.
Published in final edited form as:
PMCID: PMC3600431
NIHMSID: NIHMS448840

Underdiagnosis of hypertension using electronic health records

D. Banerjee, MD, MS,1 S. Chung, PhD,2 E. C. Wong, MS,2 E. J. Wang, MS,2 R. S. Stafford, MD, PhD,3 and L. P. Palaniappan, MD, MS2,3

Abstract

BACKGROUND

Hypertension is highly prevalent and contributes to cardiovascular morbidity and mortality. Appropriate identification of hypertension is fundamental for its management. The rates of appropriate hypertension diagnosis in outpatient settings using an electronic health record (EHR) have not been well studied. We sought to identify prevalent and incident hypertension cases in a large outpatient healthcare system, examine the diagnosis rates of prevalent and incident hypertension, and identify clinical and demographic factors associated with appropriate hypertension diagnosis.

METHODS

We analyzed a three-year, cross-sectional sample of 251,590 patients aged ≥18 years using patient EHRs. Underlying hypertension was defined as two or more abnormal blood pressure (ABP) readings ≥140/90 mmHg and/or pharmaceutical treatment. Appropriate hypertension diagnosis was defined by the reporting of ICD-9 codes (401.0 – 401.9). Factors associated with hypertension diagnosis were assessed through multivariate analyses of patient clinical and demographic characteristics.

RESULTS

The prevalence of hypertension was 28.7%, and the diagnosis rate was 62.9%. The incidence of hypertension was 13.3%, with a diagnosis rate of 19.9%. Predictors of diagnosis for prevalent hypertension included older age, Asian, African American, higher BMI, and increased number of ABP readings. Predictors for incident hypertension diagnosis were similar. Patients with underlying hypertension were more likely to be treated when they had a hypertension diagnosis in the EHR (92.6%) than if they did not (15.8%, p < 0.0001).

CONCLUSIONS

Outpatient EHR diagnosis rates are suboptimal, yet EHR diagnosis of hypertension is strongly associated with treatment. Targeted efforts to improve diagnosis should be a priority.

Keywords: Hypertension, prevalence, incidence, diagnosis

Introduction

Hypertension is rising in prevalence and contributes strongly to cardiovascular morbidity and mortality sequelae that can be reduced with appropriate treatment.13 Despite the Joint National Committee (JNC) VII recommendations highlighting the importance of screening for and treating abnormal blood pressure (ABP) levels, identification and treatment of hypertension remain suboptimal.4,5 Underdiagnosis of hypertension even in the presence of high blood pressure readings may lead to omission of recommended therapies, with potentially devastating consequences.6,7 The electronic health record (EHR) allows for the rapid retrieval of information helpful in identifying hypertension (e.g., blood pressure [BP] readings and EHR diagnoses of hypertension), but the accuracy of identifying adults with hypertension in the context of EHR use is not known.

Previous investigators have noted varying rates of chronic disease identification in hospital settings, ranging from a sensitivity of 35% for acute stroke diagnosis to 82% for the diagnosis of both type 2 diabetes mellitus and hypertension.810 It is unknown whether the high inpatient hypertension diagnosis rate in adults translates to non-hospital settings. A recent pediatric outpatient investigation revealed that only 26% of children with hypertension based on ABP had this diagnosis entered in the EHR.11 There is little information regarding diagnosis rates of hypertension using an EHR in adult outpatients. To our knowledge, past research has not analyzed outpatient diagnosis rates of incident, rather than prevalent, hypertension. If diagnosis rates of incident hypertension are low, improving diagnosis of incident hypertension would lead to earlier identification of at-risk patients, and possibly greater therapeutic benefit. Furthermore, if EHR diagnosis rates of hypertension are low, it may be that more sensitive data (i.e., recorded BP) should be used for the identification of hypertension rather than a physician reported EHR diagnosis alone.

We sought to describe the diagnosis rates, in an EHR, of prevalent and incident hypertension in adult outpatients. We defined underlying hypertension as two separate ABP values and/or use of antihypertensive medication. Additionally, we examined clinical and demographic predictors of both prevalent and incident hypertension diagnosis.

Methods

Study Setting

The Palo Alto Medical Foundation (PAMF) is an outpatient, multi-specialty series of clinics in the San Francisco Bay Area with nearly 1,000 physicians, serving more than 600,000 active patients, with a broad racial/ethnic distribution.12 Clinical and service use data for active patients between 2006 and 2008 were extracted from the EpicCare EHR system (Epic Systems Corp, Madison, WI). All datasets analyzed were HIPAA de-identified. No patients were directly contacted for the study, which received approval from the PAMF IRB.

Measurement of Study Parameters

Blood pressure (BP) was measured according to American Heart Association guidelines13 by trained health professionals, either medical assistants or nurses, as part of routine clinical care. These guidelines suggest that patients be seated for five minutes prior to BP being taken using automatic BP cuffs, and multiple cuff sizes were available for measurement. BP values were then entered into the EHR by the measuring provider. We excluded BP values obtained from urgent care visits in our analysis to minimize the possibility of obtaining transiently elevated BP readings associated with acute medical conditions.14 BP was measured at 98% of visits in this cohort.

Study Sample

The overall sample (N=251,590) included adult patients (≥ 18 years) who had at least two clinic visits between January 1, 2006 and December 31, 2008. Patients who had at least two readings of BP ≥140/90 mmHg, and/or antihypertensive medication prescription either before or during the study period were included in the prevalent hypertension cohort (N=72,112). Patients with no history of hypertension (indicated by the absence of BP ≥140/90 mmHg, antihypertensive medication prescription and ICD-9 codes [401.0–401.9]) prior to baseline for at least one year, were included in the incident hypertension cohort (N=43,845). Incident hypertension was assessed in patients who had at least two readings of BP ≥140/90 mmHg, and/or antihypertensive medication prescription during the study period. Appropriate hypertension diagnosis was defined by physician-recorded ICD-9 codes (401.0–401.9) for both the prevalent and incident cohorts.

Definition of Underlying HTN

Underlying hypertension was defined by ABP or antihypertensive medication prescription. ABP was defined as two or more readings of BP ≥140/90 mmHg, based upon the JNC VII guidelines for the diagnosis of hypertension. We also considered antihypertensive medication prescription in defining underlying hypertension since BP values may be in the normal range with the medical treatment of hypertension. Antihypertensive medications were adjudicated by a committee of three internal medicine physicians, and included but were not limited to diuretics (including loop diuretics), calcium channel blockers, beta blockers, angiotensin converting enzyme inhibitors, and angiotensin receptor antagonists.

Electronic Health Record (EHR)

Palo Alto Medical Foundation uses the EpicCare EHR system (Epic Systems Corp, Madison, WI). All physicians have access to the EHR when examining patients. The EHR carries a record of a patient’s prior visits, including their blood pressures, as well as their current medication list and comorbidities. Problem lists are internal, physician-maintained EHR lists of ICD-9 codes, consisting of current medical conditions, available for internal viewing across multiple visits, but not externally reported for medical billing purposes. (Figure, EPIC screenshot) In contrast, visit diagnoses are physician-recorded ICD-9 codes designating the reason for a patient’s visit, with at least one visit diagnosis available from each visit for the purpose of generating a medical bill. Visit diagnoses are commonly analyzed in large administrative datasets,16 whereas problem lists are less often available for analyses. In this EPIC system, there is no limit on the number of codes that can be entered in the problem list. We noted a range of 1 to 27 ICD-9 codes on the problem list. In our analysis, we examined all diagnoses on the problem list rather than restricting the analysis to the primary diagnosis. An EHR diagnosis is needed to prescribe medications for HTN, although clinicians can view prior medications and blood pressures to aid in patient management.

Appropriate Diagnosis of Hypertension

Appropriate diagnosis of hypertension was defined as an ICD-9 code of 401.1 – 401.9 on the EHR problem list15 or visit diagnosis at any time during the three-year study period. Very few patients (1.7%; N=796) had an ICD-9 code for hypertension, without evidence of hypertension (two readings of BP ≥140/90 mmHg, and/or antihypertensive medication prescription) in our dataset.

Diagnosis Rates

For both prevalent and incident hypertension, the diagnosis rate was calculated as the ratio of patients with underlying hypertension with a hypertension diagnosis (numerator) to all patients with underlying hypertension (denominator) in the EHR. For example, the diagnosis rate of hypertension (BP ≥ 140/90 mmHg) was calculated as: (number of patients with a hypertension ICD-9 code)/(number of patients with either two or more readings of BP ≥ 140/90 mmHg and/or antihypertensive medication prescription). We computed diagnosis rates in two different patient populations: (1) all patients in the study sample, to assess prevalent hypertension diagnosis rate; and (2) patients with no history of hypertension (indicated by the absence of ABP, antihypertensive medication prescription and ICD-9 codes [401.0–401.9]) prior to baseline, to assess diagnosis rates of incident hypertension.

Predictors of Hypertension Diagnosis

We used logistic regression to model the likelihood of an appropriate hypertension diagnosis in those with underlying hypertension. Predictors of appropriate hypertension diagnosis examined were patient age, sex, patient reported race/ethnicity, baseline BMI, number of BP readings ≥140/90 mmHg, and number of BP readings ≥160/100 mmHg, all measured during the three-year study period.

Statistical Analysis

Demographic and clinical characteristics of patients with underlying hypertension who did and did not have an appropriate diagnosis of hypertension were compared using t-tests (for continuous variables) or Chi-square test (for dichotomous variables). Multivariate analysis was performed using logistic regression with diagnosis of hypertension used as the binary outcome measure. All analyses were performed using Stata version 10.0 (StataCorp, College Station, TX).

Results

Patient Characteristics

The overall sample of patients with and without underlying hypertension (N=251,590) was 59.6% female, had a mean age of 44 ± 16 years, and a mean BMI of 26.4 kg/m2 ± 5.5. BP was measured a mean of six times per patient during the study period (Table 1).

Table 1
Summary Statistics of Patient Characteristics. Statistics represent mean (SD) for continuous variables and frequency % for dichotomous variables.

Prevalence, Incidence and Diagnosis Rates of Hypertension

The prevalence of underlying hypertension (BP ≥140/90 mmHg) among all patients (N=251,590) was 28.7%. When using JNC VII Stage II criteria (BP ≥160/100 mmHg), the prevalence of hypertension was 22.3%. Among patients for whom there was no evidence of hypertension prior to baseline (N=43,845), 13.3% developed incident hypertension over the three-year study period (Table 2). When restricted to a one-year window (2006 calendar year), the incidence was 5.0%.

Table 2
Prevalence and 3 year Incidence of Hypertension

The appropriate diagnosis rates for hypertension (BP ≥140/90 mmHg) were low for both those with prevalent (62.9%) and incident (19.9%) hypertension. For patients with hypertension and an EHR diagnosis of hypertension, the diagnosis was more likely to be coded as a visit diagnosis in the EHR (97.4%) than as an entry on the problem list (58.1%).

Determinants of Appropriate Hypertension Diagnosis

Among patients with prevalent hypertension (N=72,112), we compared the cohort with a diagnosis of hypertension (62.9%) to those without a diagnosis of hypertension. In bivariate analyses, those with an appropriate hypertension diagnosis were likely to be older, male, Asian, African American, have an antihypertensive medication prescribed, increased number of ABP readings, and higher BMI, (P<0.0001 for all factors, Table 3). Baseline systolic BP, diastolic BP and highest systolic BP were initially considered but were not found to be independent predictors of hypertension diagnosis, nor did they improve model fit statistics in multivariate analysis, and thus were excluded from the final model.

Table 3
Likelihood of Having a Hypertension Diagnosis. Outcome variable = having hypertension diagnosis in either problem list or visit diagnosis (0/1) Statistics represent Odds Ratio (99% Confidence Interval)

Demographic predictors of appropriate diagnosis of incident hypertension included age, Asian and African American race/ethnicity (referent category: Non-Hispanic Whites) (Table 3). Positive clinical predictors included number of ABP readings ≥ 140/90 mmHg, number of ABP readings ≥ 160/100 mmHg, and baseline BMI. Predictors of hypertension diagnosis in those with prevalent hypertension were similar to those with incident hypertension except that females were less likely to be diagnosed than males and frequency of BP readings ≥ 140/90 mmHg was not a significant predictor (while readings ≥ 160/100 remained a predictor).

Association Between Appropriate Diagnosis and Treatment

An appropriate diagnosis of hypertension was strongly associated with treatment (Figure 1). Among patients with ABP (BP ≥ 140/90 mmHg; N=47,089) noted in the EHR, those with an appropriate diagnosis were more likely to be treated with an antihypertensive medication than patients without a diagnosis (92.6% vs. 15.8%, p < 0.0001). Most (73%) patients taking an antihypertensive medication had an appropriate hypertension diagnosis with many patients lacking a hypertension diagnosis having other conditions for which medications used for hypertension might be used.

Figure 1
Relationship Between EHR Diagnosis of Hypertension (HTN) and Treatment in Patients with Abnormal Blood Pressure. (ABP; BP ≥140/90 mmHg).

To analyze the differential association of ABP and antihypertensive medication treatment with a diagnosis of hypertension, we compared diagnosis rates in three groups of patients with incident hypertension: (1) non-treated hypertension: patients with ABP without antihypertensive treatment, (2) controlled hypertension: patients with antihypertensive treatment with normal BP, and (3) treated, uncontrolled hypertension: patients with both antihypertensive treatment and ABP. There was a striking increase in diagnosis rates among the groups, with the highest rate (76.0%) of appropriate diagnosis in those patients on antihypertensive medication treatment, but still with abnormally high BP readings (Figure 2).

Figure 2
Diagnosis Rates of Incident Hypertension (HTN) (Blood Pressure [BP] cutoff 140/90).

Discussion

In this investigation, we expanded on previous analyses of hypertension diagnosis by (1) distinguishing between prevalent and incident hypertension in adult outpatients, and (2) examining clinical and demographic predictors of prevalent and incident hypertension diagnosis. The prevalence and incidence of hypertension in our cohort were similar to those found in nationally representative data,17 reinforcing the external validity of our findings. We observed that both prevalent and incident hypertension are underdiagnosed in an EHR in adult outpatients. Of patients with prevalent hypertension, 62.9% received an appropriate diagnosis of hypertension recorded in the EHR. Only 19.9% of patients with incident hypertension received an EHR diagnosis of hypertension. EHR diagnosis of hypertension was strongly associated with treatment, and this relationship was consistent across demographic categories and different ABP cutoffs (results not shown). While we could not ascertain whether diagnosis preceded treatment (rather than occurring simultaneously as a means of justifying and billing for treatment), the fact that diagnosis and treatment were coupled suggests that if diagnosis rates of hypertension improved, treatment rates would also improve. Further studies could explore the temporal relationship between hypertension diagnosis and treatment.

We also noted significant gender and racial/ethnic differences in the diagnosis of hypertension in our multivariate analysis, which could help to direct efforts to reduce disparities in hypertension diagnosis and treatment. Men with hypertension were more likely to be diagnosed and treated, which may partially explain the sex differences in CVD outcomes.18 Interestingly, racial/ethnic minorities (Asians and African Americans) were more likely to be diagnosed in this cohort, perhaps due to heightened physician awareness of hypertension in these populations. Hispanic/Latinos exhibited a non-significant finding of higher diagnosis rates of hypertension compared with Non-Hispanic Whites, similar to other racial/ethnic minorities. One possible strategy to improve hypertension diagnosis would be to direct efforts at those populations where physicians have the greatest likelihood of failing to diagnose hypertension, such as patients who are younger with normal BMIs.

The fact that increased number of ABP values is associated with higher diagnosis rates of incident hypertension in the EHR is unsurprising. BP levels vary throughout the day and can be affected by many stimuli, including anxiety and pain, such that consistently or highly abnormal values provide physicians with more confidence in making a diagnosis of hypertension. Furthermore, labeling a patient with a diagnosis of hypertension may have profound consequences, including an increase in health insurance premiums and prescription of medications with adverse side effects, which could make physicians more reluctant to label their patients as hypertensive if situational causes cannot be excluded. An increased number of ABP values were associated with diagnosis, and increasing the frequency of follow-up BP checks may be a possible intervention to increase diagnosis rates. While lifestyle measures, such as lowering dietary sodium intake, may be employed as initial approaches to treat hypertension, serially elevated BP levels indicate the need for a pharmacologic approach for adequate control and prevention of morbidity and mortality.3

A higher number of ABP values was not associated with higher diagnosis rates of prevalent HTN in our study. It may be that there is a ‘window of opportunity’ for the diagnosis of certain chronic conditions, including HTN. When providers first see patients, they may be more likely to diagnose patients with these conditions, particularly when there are a number of abnormal findings (ie ABP). However, when a patient has been established in a provider’s practice, there may be ‘provider fatigue’, so that further abnormal values do not trigger provider diagnosis of that condition. This may explain the discrepancy in our regression analysis of ABP as a significant predictor of hypertension diagnosis only in the incident cohort, but not in the prevalent cohort.

The greatest impact of higher diagnosis rates of hypertension would be to expand the cohort of patients considered for therapy.5 Our study demonstrates a gap between prevalent and EHR diagnosed hypertension. By increasing the diagnosis of hypertension through accurate EHR coding, more patients at risk could be identified and managed appropriately. Thus, physician education may be instrumental in effecting improved hypertension diagnosis. Based on our data, however, relying on physician initiated EHR diagnoses of hypertension alone will not be sufficient to improve the identification and treatment of hypertension. EHR-based efforts aimed at better identification may be more useful. Measures to improve underdiagnosis of HTN in the EHR should include physician education regarding the importance and implementation of HTN diagnosis in the EHR while also improving the functionality of the EHR to allow for easier identification of patients with HTN. The latter could include easier extraction of a patient’s BP record, such as one touch graphical representation of BP over a period of time, automatic display of a patient’s last 3 BP, and/or electronic alerts for ABP to both providers and patients. For example, PAMF has recently made a concerted effort to better identify patients with type 2 diabetes mellitus and improve diabetes-related care. One measure employed was to automatically populate type 2 diabetes mellitus in the EHR of patients with an abnormal hemoglobin A1c value (≥ 8%) and send electronic reminders to patients and providers regarding diabetic appropriate screening (e.g., foot exams, eye exams, urine microalbumin). EHR-based reminders have been used in other settings with success, such as improving the use of beta-blockers in patients with systolic cardiac dysfunction by adding a reminder to echocardiogram reports.19

Similar measures could lead to substantial improvements in the appropriate diagnosis of hypertension. EHR-based efforts could include electronic alerts for ABP in the EHR and automatic uploading of a hypertension diagnosis to the problem list based on serial ABP values. Future studies would be needed to determine whether these interventions would ultimately improve hypertension treatment and control.

Limitations

Though we excluded urgent care visits from our analysis to try to eliminate patient visits for anxiety and pain, patients do see their physicians on a non-urgent basis for conditions that may transiently elevate BP. That a significant proportion of patients with ≥ 2 ABP (where it would be highly unlikely for all readings to be due to pain or anxiety) were not diagnosed with hypertension suggests that the ‘acute pain’ hypothesis14 cannot entirely explain the underdiagnosis of hypertension in our cohort.

Physicians may be reluctant to label patients as hypertensive whose BP are minimally elevated. Furthermore, the phenomenon of ‘white coat hypertension,’ where patients exhibit higher BP in clinical settings than at home, may bias physicians toward delaying or withholding a diagnosis of hypertension. However, white coat hypertension is not a benign condition as it portends higher risk of cardiovascular events compared to those without it.20 While a medication is not necessarily recommended by JNC VII guidelines for patients diagnosed with hypertension, patients with 2 ABP should be diagnosed as hypertensive as they are at risk for future cardiovascular complications. In fact, JNC recognizes that lifestyle modification should begin at even lower levels of blood pressure, and the risk of CVD doubles with each increase in SBP of 20 mm Hg. Thus we think it is important that even patients at the ‘lower range’ of hypertension be labeled as hypertensive in order to target these individuals for risk modification.

We did not examine a causal relationship between EHR coding of hypertension and pharmacological treatment of hypertension. Comparison of coded vs. not-coded hypertension cases and differential treatment patterns would be necessary to fully understand the coding-treatment relationship. Whether more accurate diagnosis of hypertension leads to better cardiovascular outcomes is not addressed in this paper. Finally, we did not address physician characteristics (e.g., age, gender, and specialty) that predict diagnosis rates of hypertension, which is crucial since physicians are uniquely situated at the intersection of diagnosis and treatment.

We acknowledge the possibility the deficits in hypertension diagnosis we identified could be in part be due to administrative misclassification (ie coding errors or missed fields), which would not improve HTN care. Even if this were the case, improving HTN diagnosis would be a worthwhile goal in order to improve recording processes. Also, unlike previous studies that utilized claims/billing information to assess physician diagnosis of chronic medical conditions, our study used both data within the EHR to assess the appropriate diagnosis of hypertension as well as billing data (visit diagnoses). We were thus not vulnerable to the errors inherent in using claims data, such as being restricted to 1 or 2 billing codes, which would underestimate HTN diagnosis rates).

Conclusion

In summary, we found that prevalent and incident hypertension were greatly underdiagnosed in the EHR of an adult outpatient cohort, and that multiple patient factors, including gender, race/ethnicity and number of ABPs, predicted diagnosis. Diagnosis of hypertension was strongly associated with its treatment. Thus, approaches to improve hypertension diagnosis through clinical reminders (e.g., an automated EHR alert) and attention to factors that predict diagnosis could improve the appropriate treatment of patients with hypertension.

Acknowledgments

Funding Sources: Dr. Dipanjan Banerjee was supported by an institutional National Research Service Award (5T32-HL07034). Dr. Randall S. Stafford’s contribution to this work was supported by a mid-career development award from the National Heart, Lung and Blood Institute (K24-HL086703). Dr. Latha P. Palaniappan, Dr. Sukyung Chung, Eric C. Wong, and Elsie J. Wang were supported by an award from the National Institute of Diabetes and Digestive and Kidney Diseases (R01-DK081371).

Footnotes

Disclosure: There are no potential conflicts of interest for all authors.

References

1. Hypertension Detection and Follow-up Program Cooperative Group. The effect of treatment on mortality in “mild” hypertension: results of the hypertension detection and follow-up program. N Engl J Med. 1982 Oct 14;307(16):976–980. [PubMed]
2. Dahlof B, Devereux R, de Faire U, Fyhrquist F, Hedner T, Ibsen H, Julius S, Kjeldsen S, Kristianson K, Lederballe-Pedersen O, Lindholm LH, Nieminen MS, Omvik P, Oparil S, Wedel H. The Losartan Intervention For Endpoint reduction (LIFE) in Hypertension study: rationale, design, and methods. The LIFE Study Group. Am J Hypertens. 1997 Jul;10(7 Pt 1):705–713. [PubMed]
3. Messerli FH, Williams B, Ritz E. Essential hypertension. Lancet. 2007 Aug 18;370(9587):591–603. [PubMed]
4. Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of hypertension in the United States, 1988–2000. JAMA. 2003 Jul 9;290(2):199–206. [PubMed]
5. Andrade SE, Gurwitz JH, Field TS, Kelleher M, Majumdar SR, Reed G, Black R. Hypertension management: the care gap between clinical guidelines and clinical practice. Am J Manag Care. 2004 Jul;10(7 Pt 2):481–486. [PubMed]
6. Trilling JS, Froom J. The urgent need to improve hypertension care. Arch Fam Med. 2000 Sep-Oct;9(9):794–801. [PubMed]
7. Hansen ML, Gunn PW, Kaelber DC. Underdiagnosis of hypertension in children and adolescents. JAMA. 2007 Aug 22;298(8):874–879. [PubMed]
8. Birman-Deych E, Waterman AD, Yan Y, Nilasena DS, Radford MJ, Gage BF. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care. 2005 May;43(5):480–485. [PubMed]
9. Kokotailo RA, Hill MD. Coding of stroke and stroke risk factors using international classification of diseases, revisions 9 and 10. Stroke. 2005 Aug;36(8):1776–1781. [PubMed]
10. Rhodes ET, Laffel LM, Gonzalez TV, Ludwig DS. Accuracy of administrative coding for type 2 diabetes in children, adolescents, and young adults. Diabetes Care. 2007 Jan;30(1):141–143. [PubMed]
11. Woo JG, Zeller MH, Wilson K, Inge T. Obesity identified by discharge ICD-9 codes underestimates the true prevalence of obesity in hospitalized children. J Pediatr. 2009 Mar;154(3):327–331. [PubMed]
12. Wong EC, Palaniappan LP, Lauderdale DS. Using Name Lists to Infer Asian Racial/Ethnic Subgroups in the Healthcare Setting. Med Care. 2010;48(6):540–546. [PMC free article] [PubMed]
13. Pickering TG, Hall JE, Appel LJ, Falkner BE, Graves J, Hill MN, Jones DW, Kurtz T, Sheps SG, Roccella EJ. Recommendations for blood pressure measurement in humans and experimental animals: Part 1: blood pressure measurement in humans: a statement for professionals from the Subcommittee of Professional and Public Education of the American Heart Association Council on High Blood Pressure Research. Hypertension. 2005 Jan;45(1):142–161. [PubMed]
14. Goetzl FR, Bien CW, Lu G. Changes in blood pressure in response to presumably painful stimuli. J Appl Physiol. 1951 Sep;4(3):161–170. [PubMed]
15. [Accessed September 10, 2009];Epic Electronic Health Record Software Epic Web Site. http://www.epic.com.
16. Engel-Nitz NM, Martin S, Sun P, Buesching D, Fonseca V. Cardiovascular events and insulin therapy: a retrospective cohort analysis. Diabetes Res Clin Pract. 2008 Jul;81(1):97–104. [PubMed]
17. Ong KL, Cheung BM, Man YB, Lau CP, Lam KS. Prevalence, awareness, treatment, and control of hypertension among United States adults 1999–2004. Hypertension. 2007 Jan;49(1):69–75. [PubMed]
18. Polk DM, Naqvi TZ. Cardiovascular disease in women: sex differences in presentation, risk factors, and evaluation. Curr Cardiol Rep. 2005 May;7(3):166–172. [PubMed]
19. Heidenreich PA, Gholami P, Sahay A, Massie B, Goldstein MK. Clinical reminders attached to echocardiography reports of patients with reduced left ventricular ejection fraction increase use of beta-blockers: a randomized trial. Circulation. 2007 Jun 5;115(22):2829–2834. [PubMed]
20. Gustavsen PH, Hoegholm A, Bang LE, Kristensen KS. White coat hypertension is a cardiovascular risk factor: A 10-year follow-up study. J Hum Hypertens. 2003;17:811–817. [PubMed]