a b s t r a c t

Objectives: The objective is of the study to evaluate the effect of Antihypertensive therapy in emergency depart- ment (ED) patients with markedly Elevated blood pressure (BP) but no signs/symptoms of acute Target organ damage (TOD).

Methods: This is a retrospective cohort study of ED patients age 18 years and older with an initial BP greater than or equal to 180/100 mm Hg and no acute TOD, who were discharged with a primary diagnosis of hypertension. Patients were divided based on receipt of antihypertensive therapy and outcomes (ED revisits and mortality) and were compared.

Results: Of 1016 patients, 435 (42.8%) received antihypertensive therapy, primarily (88.5%) oral clonidine. Aver- age age was 49.2 years, and 94.5% were African American. Treated patients more often had a history of hyperten- sion (93.1% vs 84.3%; difference = -8.8; 95% confidence interval [CI], -12.5 to -4.9) and had higher mean

initial systolic (202 vs 185 mm Hg; difference = 16.9; 95% CI, -19.7 to -14.1) and diastolic (115 vs 106 mm

Hg; difference = -8.6; 95% CI, -10.3 to -6.9) BP. Emergency department revisits at 24 hours (4.4% vs 2.4%;

difference = - 2.0; 95% CI, - 4.5 to 0.3) and 30 days (18.9% vs 15.2%; difference = - 3.7; 95% CI, - 8.5 to

0.9) and mortality at 30 days (0.2% vs 0.2%; difference = 0; 95% CI, - 1.1 to 0.8) and 1 year (2.1% vs 1.6%;

difference = - 0.5; 95% CI, - 2.5 to 1.2) were similar.

Conclusions: Revisits and mortality were similar for ED patients with markedly elevated BP but no acute TOD, whether they were treated with antihypertensive therapy, suggesting relative safety with either approach.

(C) 2015

Introduction

Background

Hypertension (HTN) is the most common risk factor for cardiovas- cular disease in the United States, affecting approximately 75 million individuals and, as such, is a frequently encountered condition among pa- tients treated in the emergency department (ED) [1,2]. Accordingly,

* Corresponding author. Department of Emergency Medicine, Wayne State University School of Medicine, 4201 St Antoine, Detroit, MI 48201. Tel.: +1 248 840 0169; fax: +1

313 993 7703.

E-mail address: [email protected] (P.D. Levy).

emergency physicians evaluate and treat a variety of Hypertensive patients ranging from those who have incidentally discovered elevations in their Blood pressure to those who are critically ill with acute Target organ damage [3,4]. Although there is relative uniformity in the approach to treatment of those with true Hypertensive emergencies, a clear consensus on the management of patients with severely elevated BP in the ED who lack clinical evidence of acute TOD does not exist.

Importance

Significant confusion exists regarding how and when to intervene for patients with markedly elevated BP (ie, >= 180/100 mm Hg) but no symptoms of acute TOD, leading to widely divergent practice patterns

http://dx.doi.org/10.1016/j.ajem.2015.05.036 0735-6757/(C) 2015

[5,6]. Much of this is driven by a scarcity of relevant outcome data, specifically as it pertains to the possible benefit or harm associated with acute BP reduction.

No

Yes

Yes

No

No

Yes

Index visit?

Not Treated

No specific intervention for acute Blood pressure reduction given

N = 581

Excluded from further analysis

N = 89

Excluded from further analysis

N = 204

Signs or symptoms attributed to potential acute hypertensive target organ damage by the treating clinician?

Excluded from further analysis

N = 489

Initial blood pressure ? 180/100 mmHg?

Primary emergency department discharge diagnosis of hypertension

N = 1798

Goals of this investigation

The goal of this investigation is to compare outcomes for ED patients with markedly elevated BP but no signs or symptoms of acute TOD based on whether they received antihypertensive therapy for BP reduction in the ED.

Methods

Study design and setting

This was an institutional review board-approved, retrospective cohort study of adult patients between the ages of 18 and 89 years who were discharged from the ED of an urban teaching hospital with a primary diagnosis of HTN between the dates of January 1, 2008, and December 31, 2008.

Study protocol

An initial search of ED billing company data yielded 1798 visits over the study period where HTN was the primary discharge diagnosis. For this analysis, we included only patients with an initial triage BP greater than or equal to 180/100 mm Hg (ie, those with markedly elevated BP) who had no documented signs or symptoms (eg, confusion or altered mental status, focal neurologic deficits, angina chest pain, cardiogenic dyspnea, visual field defects, or papilledema) attributed to potential acute hypertensive TOD by the treating attending emergency physician. For patients with repeat ED presentations over the study period, the first recorded ED visit was designated as the index visit.

The exposure of interest was ED treatment with oral clonidine or specific intravenous Antihypertensive medications (nitroglycerin, labetalol, metoprolol, hydralazine, nicardipine, and enalaprilat) for the expressed purpose of immediate BP reduction. Based upon this, 2 patient subgroups were defined: those who received ED treatment for acute BP control (“treated”) and those who did not (“not treated”). An overview of ascertainment methodology is provided in Fig. 1.

Measures

A dedicated group of specifically trained medical students performed all data extraction. Training included instruction on performance of key word visual searches, explanation of the study’s predefined data dictio- nary, and practice with the hospital’s electronic medical record system (FirstNet; Cerner, Kansas City, MO). In addition to triage information, all narrative components of the ED chart including chief complaint, history of present illness, review of systems, physical examination, evaluation and management services, disposition plan, final diagnosis, and teaching physician addendum (when present) were reviewed. Data extraction included documented reasons for ED presentation, demographic infor- mation, medical history, vital signs (at presentation, during the ED stay, and before discharge), results of ED work-up (if performed), ED treat- ment, and medications prescribed at discharge. As our intent was to evaluate outcomes related to treatment (vs no treatment) of patients with elevated BP and not to model predictors of the approach to patient care, provider-level data were not collected.

Collected data were entered into an Excel spreadsheet that contained

limits for plausible results and various checks for data integrity. An inde- pendent abstractor performed double data entry for a random sample of 100 charts, from which data were cross-checked for internal consistency. There was high agreement in chart abstraction (? = 0.93), and any discrepancy related to data abstraction was adjudicated by the senior study investigator (PDL). Specific outcomes of interest included all-

Treated

Received oral clonidine or intravenous antihypertensive medication

N = 435

Fig. 1. Patient ascertainment methodology.

cause return visits to the ED and all-cause hospital admissions within 24 hours and through 30 days, from the index visit going forward. At both time points, we further subdivided ED visits into those that were potentially attributable to complications of HTN treatment or nontreat- ment. The latter were determined by research personnel using standard- ized abstraction methods including comprehensive ED chart review, key word visual searches (eg, hypotension, HTN, hypertensive urgency, Hypertensive crisis, Hypertensive emergency, and elevated BP), and re- view of ED admitting or discharge diagnosis (eg, hypertensive encepha- lopathy, stroke, cerebrovascular accident, acute myocardial infarction, acute kidney injury, acute heart failure, and aortic dissection). All-cause mortality within 30 days and at 1 year was also assessed using the Social Security Death Index, a highly reliable source of mortality data [7]. All re- search personnel who collected outcome data were blinded to treatment subgroup and not involved in data abstraction for the index ED visit.

Data analysis

Sample size calculations were based on the following premises; the baseline proportion of patients who would return to the ED at 30 days for an HTN-related issue (the referred outcome) would range from 10% to 30%, and a difference in return proportion of 5% would be clini- cally important. Therefore, a range based on the above assumption would be more valuable. Given the above, it was determined that a sample size of 1000 patients would result in power of 0.9 if the

proportion of return visits was 10%, a power of 0.75 if return visits was 20%, and a power of 0.65 if return visit was 30%. This power analysis used an exact conditional distribution.

Univariate comparison of group differences with accompanying precision was performed. Continuous (means) and categorical (propor- tion) data for treated and nontreated patients were compared, and differences between groups with corresponding 95% confidence inter- vals (CIs) were derived. No formal adjustment for multiple outcomes was used.

Multivariable logistic regression was used to model the primary outcome of ED recidivism at 30 days. The main predictor of interest was if the patient was treated for HTN during the index ED visit (treatment group). Mandatory variables included in the model were age and the propensity for treatment. The latter was determined by deriving a propensity score (PS) that included the following set of factors: sex, medical history of HTN, initial systolic BP, headache on presentation, performance of an electrocardiogram as part of the index ED visit work-up, visit to the ED specifically for HTN, and baseline use of oral clonidine. Nearly all cases (96.9%) have sufficient data for in- clusion in PS derivation, and the PS was strongly predictive of treatment (c statistic = 0.84) with goodness of fit confirmed by the Hosmer- Lemeshow goodness-of-Fit test (P = .086). Propensity scores were reasonably well balanced within a quintile of PS between the treated and nontreated groups. Other covariates incorporated in the logistic re- gression model, including uninsured status and ED visit within 24 hours of discharge from the index ED visit, were selected for analysis, as they are plausible reasons for ED recidivism, independent of being treated for HTN in the ED. Because of the number of outcomes, we limited the number of additional variables included to 5 (total of 7). The model was examined for fit (Hosmer-Lemeshow goodness-of-fit test), predictability and for collinearity of the included predictor variables.

Because a return to the ED within 24 hours could be from a variety of issues including persistent HTN (noted in 52%), we performed a sensi- tivity analysis that excluded all patients who had returned to the study hospital ED within 24 hours. Finally, to account for treatment similarities by treating physician, we fit a general estimating equation (PROC GEE in SAS, Cary, NC) using the treating physician as the subject effect. A compound symmetry correlation matrix was chosen because it was a good fit given the data and model diagnostics along with the theoretical concept that treatment constructs would be similar within subject (physician). These and all other calculations were performed using SAS 9.3 or above.

Results

Characteristics of study subjects

A total of 1016 patients met eligibility criteria, 435 (42.8%) of whom received treatment for elevated BP. As shown in Table 1, patients in both groups (treated vs not treated) were largely African American and rela- tively young with a mean (SD) age of 49.8 (11.6) vs 48.6 (11.9) years, respectively. Treated patients were more likely to have a history of HTN (93.1% vs 84.3%; difference of -8.8; 95% CI, -12.5 to -4.9) and be on clonidine therapy at baseline (45.0% vs 7.5%; difference of

-36.9; 95% CI, -42.0 to -31.7). Those who were treated also presented with “elevated BP” as their chief complaint more often, were more likely to report a headache on arrival, and had higher mean initial systolic (202 vs 185 mm Hg; difference of - 16.9; 95% CI, - 19.7 to - 14.1) and

diastolic (115 vs 106 mm Hg; difference of - 8.6; 95% CI, - 10.3 to

-6.9) BP at triage.

Emergency department care

Of those who were treated in the ED with antihypertensive medica- tions, 385 (88.5%) received clonidine, 44 (10.1%) labetalol, 33 (7.6%)

hydralazine, 7 (1.6%) enalaprilat, and 2 (0.5%) metoprolol. Some

patients received more than 1 agent. In addition, 174 (40%) treated patients also received an oral dose of their baseline antihypertensive medication in the ED. As shown in Table 1, diagnostic evaluation was more common in treated patients, with laboratory tests, chest x-ray, and electrocardiography ordered more frequently in this group. Predis- charge mean BPs were slightly lower in the treated patients (systolic: 162 vs 166 mm Hg; difference of 3.2; 95% CI, 0.3-6.1), and an equal proportion of patients received an antihypertensive prescription upon discharge from the ED (72.1% vs 68.6%; difference of - 3.6%; 95% CI

-9.2 to 2.2).

Outcomes

There were similar overall rates of ED return visits between treated and not-treated cohorts (Table 2) within 24 hours (4.4% vs 2.4%; difference of -2.0; 95% CI, -4.5 to 0.3) and within 30 days (18.9 vs

15.2; difference of -3.7; 95% CI, -8.5 to 0.9). A total of 977 patients (96.2% of the entire cohort) had complete data for needed for inclusion in multivariable logistic regression modeling. After adjustment for PS, insurance status, medical history (chronic kidney disease, chronic heart failure, or diabetes mellitus), and age (Fig. 2), receiving acute treatment at the index ED visit was not associated with more frequent ED visits at 30 days (odds ratio [OR], 1.23; 95% CI, 0.66-2.31). The adjusted model itself was modestly predictive (c statistic = 0.65) and demon- strated good fit (Hosmer-Lemeshow, P = .53), and only a history of chronic heart failure was associated with higher odds of an ED revisit at 30 days (OR, 3.25; 95% CI, 1.42-7.43). Findings on sensitivity analysis that excluded patients with a 24-hour ED revisit (OR, 1.48; 95% CI, 0.76- 2.90) and with modeling using a general estimating equation that accounted for treatment clustering by physician (OR, 1.22; 95% CI, 0.61-2.42) were similar with no evidence of a relationship between treatment and 30-day ED revisit.

Although there was a trend toward more frequent return visits to the ED for issues related to HTN at 30 days among those who were treated, in general, complications were infrequent, hospital admission was uncom- mon, and few deaths occurred in either group with no difference in all- cause mortality at 30 days (0.2% vs 0.2%; difference of 0; 95% CI, -1.1 to 0.8) or 1 year (2.1% vs 1.6%; difference of -0.5; 95% CI, -2.5 to 1.2).

Discussion

Despite strong sentiment that acute antihypertensive therapy is not necessary in the setting of HTN without acute TOD [3,8], clinicians often feel inclined to do something when patients present with markedly elevated BP. However, as we demonstrate in this retrospective cohort study, acute BP reduction appears to provide no direct benefit to such patients. Moreover, the occurrence of adverse events was minimal, and Death rates were low in both treated and not-treated patients signaling an absence of harm without treatment. Our retrospective data are consistent with existing, albeit scant trial results, which support a lack of specific benefit with BP lowering in the setting of markedly elevated BP in the absence of acute hypertensive TOD. [3,9,10]

Similar to other studies on the treatment of HTN in the ED, we found that physicians were more likely to provide antihypertensive therapy to those with a history of HTN and higher BP upon arrival [11,12]. Although BP in the ED decreased significantly in both groups, the rela- tive difference in interim BP reduction was nearly double for patients who were given an antihypertensive agent, with a drop in systolic BP of 39.8 mm Hg (vs 19.7 mm Hg) and Diastolic BP of 21.1 mm Hg (vs 10.0 mm Hg). Blood pressure at discharge was also statistically lower in the treated group, although the observed difference is unlikely to be clinically relevant. Moreover, that both groups experienced a decrease in BP over time serves to highlight the concept of regression to the mean that has previously been observed in ED-based BP studies [13]. Clinicians should be mindful of this and prior real-world expe- rience with sublingual nifedipine, where detrimental consequences

Table 1

Baseline data

Variable Not treated,

Treated,

Difference (95% CI)

Table 1 (continued)

Variable Not treated,

n= 581

Treated, n = 435

Difference (95% CI)

n = 581

n = 435

Mean predischarge systolic

165.7 (22.5) 162.5 (23.7) 3.2 (0.3, 6.1)^a

Mean age in years (SD) 48.6 (11.9) 49.8 (11.6) -1.2 (-2.7,0.3)

Male sex, n (%) 366 (63.0) 239 (55.0) 8.1 (-2.0, 14.0)

African American race, n (%) 549 (94.5) 415 (95.4) -0.9 (-3.6, 2.0) Insurance status, n (%)

Prisoner 143 (24.8) 107 (24.8) 0.0 (-5.4, 5.3)

Uninsured 179 (31.0) 120 (27.8) 3.2 (-2.5, 8.8)

Medicaid 33 (5.7) 32 (7.4) -1.7 (-5.0, 1.4)

BP in mm Hg (SD)

Mean predischarge diastolic BP in mm Hg (SD)

Mean predischarge heart rate in beats per minute (SD)

Discharged on antihypertensive therapy, n (%)

96.1 (13.9) 93.6 (16.1) 2.5 (0.7, 4.4)^a

78.5 (16.8) 78.1 (15.6) 0.4 (-1.6, 2.4)

396 (68.6) 312 (72.1) -3.6 (-9.2, 2.2)

Medicare 67 (11.6) 52 (12.1) -0.4 (-4.5, 3.5)

Private insurance 167 (28.9) 120 (27.8) 1.1 (-4.5, 6.7) Baseline medications, n (%)

Diuretic 164 (29.1) 129 (30.1) -1.4 (-7.1, 4.2)

?-blocker 119 (21.1) 83 (19.4) 1.4 (-3.6, 6.3)

ACEi/ARB 152 (26.1) 106 (24.3) 1.8 (-3.7, 7.1)

Hydralazine 19 (3.4) 25 (5.8) -2.5 (-5.3, 0.0)

CCB 108 (19.2) 66 (15.4) 3.4 (-1.3, 8.0)

Clonidine 42 (7.5) 192 (45.0) -36.9 (-42.0, -31.7)^a

Compliant with medications 51 (9.2) 44 (10.6) -1.3 (-5.1, 2.2) Medical history, n (%)

HTN 490 (84.3) 405 (93.1) -8.8 (-12.5, -4.9)^a

CHF 24 (4.1) 25 (5.8) -1.6 (-4.5, 1.1)

DM 78 (13.4) 67 (15.4) -2.2 (-6.7, 2.1)

Hyperlipidemia 38 (6.5) 27 (6.2) 0.3 (-2.9, 3.3)

CHD 15 (2.6) 14 (3.2) -0.6 (-3.0, 1.5)

CVA/TIA 32 (5.5) 23 (5.3) 0.2 (-2.8, 3.0)

CKD 17 (2.9) 28 (6.4) -3.5 (-6.4, 0.9)

Social history, n (%)

alcohol consumption 177 (32.7) 125 (31.3) 1.7 (-4.0, 7.3)

cigarette smoking 255 (46.8) 213 (52.6) -5.1 (-11.2, 1.1)

cocaine use 25 (4.7) 45 (11.2) -6.0 (-9.5, -2.8)^a

Family history, n (%)

HTN 188 (39.6) 177 (49.0) -8.3 (-14.3, -2.4)^a

CHF 19 (4.2) 10 (3.0) 1.0 (-1.2, 3.0)

Clonidine 18 (3.1) 136 (31.5) -28.2 (-32.8, -23.6)^a

Diuretic 227 (39.3) 145 (33.5) -5.7 (-0.3, 11.6)

Beta-blocker 86 (14.9) 63 (14.6) 0.3 (-4.2, 4.6)

ACEi/ARB 116 (20.5) 81 (18.6) 1.3 (-3.6, 6.2)

Hydralazine 9 (1.5) 9 (2.1) -0.5 (-2.5, 1.2)

CCB 94 (16.3) 50 (11.6) 4.7 (3.4, 8.9)^a

Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; CCB, calcium channel blocker; CHF, congestive heart failure; DM, diabetes mellitus; CHD, coronary heart disease, CVA, cerebral vascular accident; TIA, transient ischemic attack; CKD, chronic kidney disease; BUN, blood urea nitrogen; ECG, electrocar- diogram; LVH, left ventricular hypertrophy.

^a P b .05. Difference column and asterisks are all for unadjusted comparisons.

were frequently encountered when a precipitous decline in BP was initiated in patients who do not have a true hypertensive emergency [14]. The latter is particularly relevant for individuals with long- standing, untreated HTN, as marked BP elevations may actually reflect baseline values rather than acute changes.

Thus, absent clinical or laboratory evidence of acute TOD and numerical BP values in such patients may be more likely to represent a “pseudohypertensive emergency.” The potential danger with initiat-

ED visit specifically for HTN,

n (%)

377 (64.9) 309 (71.0) -6.2 (-11.8, -0.3)^a

ing treatment in this group lies in the autoregulation of cerebral and

renal perfusion, which is shifted to accommodate chronic elevations

Symptoms present on ED arrival, n (%)

Headache 73 (12.6) 118 (27.1) -14.6 (-19.6, -9.6)^a

Palpitation 2 (0.3) 3 (0.7) -0.4 (-1.7, 0.7)

Blurred vision 16 (2.8) 17 (3.9) -1.1 (-3.7, 1.1)

Dizziness 32 (5.5) 28 (6.4) -.9 (-4.1, 2.0)

Chest pain 7 (1.2) 10 (2.3) -1.1 (-3.1, 0.5)

Shortness of breath 8 (1.4) 6 (1.4) 0.0 (-1.7, 1.5) Triage vitals, mean (SD)

in systemic pressure. In such patients, an abrupt decrease in BP can cause a rapid decline in perfusion to these vital structures [15]. Our study was not designed to capture all aspects of potential cerebral or renal injury and instead relied on fairly gross outcome measures. Despite the absence of harm to patients who received BP treatment in our cohort, this very real consideration should be kept in mind whenever

Systolic BP at triage in mm Hg (SD)

Diastolic BP at triage in mm Hg (SD)

Heart rate at triage in beats per minute (SD)

185.4 (20.6) 202.3 (25.4) -16.9 (-19.7, -14.1)^a

106.1 (12.5) 114.7 (15.2) -8.6 (-10.3, -6.9)^a

84.0 (16.1) 85.1 (15.7) -1.1 (-3.1, .9)

administering antihypertensive therapy to a patient without evidence of target-organ compromise. To that end, our group is currently testing the effect of rapid BP reduction on cerebral flow in hypertensive ED patients. As noted above, clonidine was widely used as a baseline antihyper- tensive for subjects overall and proportionally a more common home

Laboratory tests done, n (%) 202 (34.8) 237 (54.5) -19.7 (-25.7, -13.6)^a

medication in the treatment group. Clonidine is frequently used in

Mean serum BUN in mg/dL (SD)

Mean serum creatinine in mg/dL (SD)

15.5 (8.5) 19.5 (14.9) -4.0 (-5.6, -2.6)^a

1.1 (0.8) 1.6 (2.4) -0.5 (-0.7, -0.3)^a

urban communities such as ours, as it is perceived to be an inexpensive yet effective antihypertensive agent. Furthermore, clonidine is often preferred to manage elevated BP in the ED due to its easy dosing,

Chest x-ray done, n (%) 32 (5.6) 54 (12.5) -6.9 (-10.7, -3.4)^a

oral route of administration, low cost, and quick onset. This may be

Cardiomegaly n (% with chest X-ray)

5 (15.1) 14 (25.9) -2.4 (-4.5, -0.6)^a

appropriate for patients with severe HTN attributable to poor adher- ence who are experiencing “clonidine withdrawal”; however, when

ECG done, n (%) 73 (12.7) 101 (23.3) -10.7 (-15.5, -5.9)^a

LVH, n (% with ECG) 14 (13.7) 33 (32.7) -5.2 (-8.2, -2.5)^a

ED antihypertensive medication, n (%)

Clonidine 385 (88.5)

Labetalol 44 (10.1)

Hydralazine 33 (7.6)

Enalaprilat 7 (1.6)

Metoprolol 2 (0.5)

Nitroglycerin 2 (0.5)

compared with other oral agents, it has a suboptimal antihypertensive

Table 2

Outcome measures

Outcome measure Not Treated Difference treated (95% CI)

n = 581 n = 435

Oral dose baseline medication, n (%)

Oral dose nonbaseline

medication, n (%)

144 (24.8) 173 (40.0) -15.0 (-20.7, -9.2)^a

29 (5.0) 28 (6.4) -1.5 (-4.5, 1.4)

ED visit 24 h, n (%) 14 (2.4) 19 (4.4) -2.0 (-4.5, 0.3)

Hospital admission 24 h, n (%) 0 (0.0) 3 (0.7) -0.7 (-2.0, 0.1)

HTN-related complication at 24 h, n (%) 0 (0.0) 1 (0.2) -0.2 (-1.3, 0.5)

Posttreatment BP obtained, n (%) 247 (42.5) 355 (81.6) -39.1 (-44.3, -33.5)^a

ED visit within 30 d, n (%) 88 (15.2) 82 (18.9) -3.7 (-8.5, 0.9)

Mean posttreatment systolic BP in mm Hg (SD)

Mean posttreatment diastolic BP in mm Hg (SD)

166.0 (23.3) 163.1 (22.8) 2.9 (0.0, 5.8)

96.0 (13.1) 94.1 (15.7) 1.9 (0.1, 3.7)

Hospital admission within 30 d, n (%) 15 (2.6) 13 (3.0) -0.4 (-2.7, 1.6)

HTN-related complication within 30 d, n (%) 6 (1.0) 11 (2.5) -1.5 (-3.5, 0.1)

Death within 30 d, n (%) 1 (0.2) 1 (0.2) 0.0 (-1.1,0.7)

Death within 1 y, n (%) 9 (1.6) 9 (2.1) -0.5 (-2.5, 1.2)

Predischarge BP obtained, n (%) 291 (50.0) 356 (81.8) -31.8 (-37.0, -26.1)^a

Difference column is for unadjusted comparisons.

Fig. 2. Adjusted ORs for 30-day ED return visit.

profile with unpredictable effects on BP and cerebral perfusion [16]. Nearly one-third of those who received clonidine in the ED were given a prescription for it as an outpatient as well. Although abrupt discontinuation of clonidine is not recommended, maintenance of this therapy for noncompliant patients or de novo initiation is not ideal and may foster return visits to the ED with rebound HTN, a known consequence clonidine treatment cessation [17].

To ameliorate sequela of chronic HTN, initiation of proper antihyper- tensive therapy is of utmost importance. Whether this should fall on the emergency physician is a matter of debate and one of the primary questions that remains unresolved according to the most recent policy statement on asymptomatic HTN put forth by the American College of Emergency Physicians [18]. In another large retrospective cohort, less than 10% of those with an systolic BP greater than or equal to 160 or an di- astolic BP greater than or equal to 100 mm Hg received a prescription for an antihypertensive upon discharge, and less than 20% of those patients re- ceived a directED referral for their HTN. In the same study, when actual re- ferral frequency was compared with self-reported rates, the vast majority of physicians were found to be overestimating their referral practice [19]. Although we did not track outpatient referrals in our study, we did find that more than 70% of patients were given an antihypertensive prescrip- tion at discharge suggesting that ED clinicians can play a more active role in chronic disease management for asymptomatic HTN.

Given that uncontrolled HTN is the primary risk factor for cardiovas- cular disease worldwide [20], emergency physicians should be familiar with existing guidelines for BP therapy and have a low threshold to discharge patients on an appropriate antihypertensive agent or, at the least, facilitate follow-up with a provider who is willing and able to do so. Understanding the latter may be challenging in underresourced settings such as ours, but we believe the benefits of initiating outpatient antihypertensive therapy from the ED far outweigh the risks. Moreover, unlike acute, episodic BP reduction, which appears to provide little if any benefit, achievement of chronic BP control using oral antihyperten- sive therapy unequivocally contributes to improved patient outcomes as proven in multiple, well-designed trials [21].

Limitations

By only including individuals who were discharged with a primary diagnosis of HTN, this study likely excluded a number of ED patients with marked BP elevation for whom HTN was either a secondary diag- nosis or not diagnosed at all. Reviewing charts for consecutive patients may have allowed for greater capture of potentially eligible patients and reduced potential for selection bias. However, such an approach is labor intensive, and, given the low incidence of adverse events, it is

unlikely that it would have substantively altered study findings. In addi- tion, we acknowledge that there is inherent subjectivity in determining signs and symptoms attributed to acute hypertensive TOD, both during the clinical encounter and on chart review. Consequently, our findings are subject to recall and misclassification bias, as some patients may not recognize subtle symptoms that are indicative of HTN related com- plications, and some clinicians may have missed clinical evidence of acute TOD or underreported related features in their documentation. Despite this, all included patients were deemed safe for discharge by an attending emergency physician, and, given the low Postdischarge event rates, it is reasonable to presume that appropriate care was delivered for the vast majority of patients during the index ED visit.

This was a single-center study with skewED representation by African Americans, many of whom use the ED for chronic management of their HTN, and results may not be applicable to the broader popula- tion of ED patients. However, our findings should stand to represent other large, urban EDs, particularly those that treat a predominantly underserved, African American patient population. Moreover, because this was a retrospective cohort study, the accuracy and completeness of our data were entirely dependent on a review of the electronic medical record. As a result, potentially important components of management including comprehensive Screening tests for end-organ damage and repeat BP at ED discharge were not uniformly available, and standardized assessments of potential adverse effects related to treatment or nontreatment with antihypertensive therapy were not performed. To address this, a prospective registry that seeks to capture such information is currently in development. Lastly, there may have been significant underestimation of postdischarge outcomes especially return visits to the ED, as patients may have been seen at other nearby hospitals or health care systems with which we do not share electronic medical records. As insurance status did not differ between treated and not-treated cohorts, we do not have a reason to believe that outside ED visit rates would have differed between cohorts; however, absent patient-level tracking of ED recidivism and actual event cannot be known. Of note, by using the Social Security Death Index, we were able collect some follow-up outcome data that support the relative safety of either approach to management. Although missed deaths are possible using this database, its accuracy has been previously shown to be ap- proximately 95% [7], and it is unlikely that a difference in availability of morality data would exist between treated and not-treated patients.

Conclusions

In summary, in this single-center, retrospective cohort of predominant- ly African American ED patients with markedly elevated BP but no signs or symptoms of acute TOD, we found no evidence of benefit with treatment to acutely lower BP. That said, complications related to HTN were infre- quent, and the all-cause mortality rate was low with no differences be- tween groups providing a measure of reassurance to ED physicians who use either approach in the care of their acutely hypertensive patients.

References

1. Levy PD, Cline D. Asymptomatic hypertension in the emergency department: a mat- ter of critical public health importance. Acad Emerg Med 2009;16:1251-7.
2. Mozaffarian D, Benjamin EJ, Go AS, Arnett DK, Blaha MJ, Cushman M, et al. Heart dis- ease and stroke statistics — 2015 update: a report from the American Heart Associ- ation. Circulation 2015;131:e29-322.
3. Decker WW, Godwin SA, Hess EP, Lenamond CC, Jagoda AS. Clinical policy: critical issues in the evaluation and management of adult patients with asymptomatic hypertension in the emergency department. Ann Emerg Med 2006;47:237-49.
4. Shayne PH, Pitts SR. Severely increased blood pressure in the emergency depart- ment. Ann Emerg Med 2003;41:513-29.
5. Kinsella K, Baraff LJ. Initiation of therapy for asymptomatic hypertension in the emergency department. Ann Emerg Med 2009;54:791-2.
6. Shayne P. Against routine initiation of antihypertensive therapy in the emergency department. Ann Emerg Med 2009;54:792-3.
7. Huntington JT, Butterfield M, Fisher J, Torrent D, Bloomston M. The Social Security Death Index (SSDI) most accurately reflects true survival for older oncology patients. Am J Cancer Res 2013;3:518-22.
8. Gallagher EJ. Hypertensive urgencies: treating the mercury? Ann Emerg Med 2003; 41:530-1.
9. Veterans Administration Cooperative Group on antihypertensive agents, Effects of treatment on morbidity in hypertension: results in patients with diastolic blood pressures averaging 115 through 129 mm hg. JAMA 1967;202:1028-34.
10. Zeller KR, Von Kuhnert L, Matthews C. Rapid reduction of severe asymptomatic hypertension: a prospective, controlled trial. Arch Intern Med 1989;149: 2186-9.
11. Chiang WK, Jamshahi B. Asymptomatic hypertension in the ED. Am J Emerg Med 1998;16:701-4.
12. Tilman K, DeLashaw M, Lowe S, Springer S, Hundley S, Counselman FL. Recognizing asymptomatic elevated blood pressure in ED patients: how good (bad) are we? Am J Emerg Med 2007;25:313-7.
13. Pitts SR, Adams RP. Emergency department hypertension and regression to the mean. Ann Emerg Med 1998;31:214-8.
14. Grossman E, Messerli FH, Grodzicki T, Kowey P. Should a moratorium be placed on sublingual nifedipine capsules given for hypertensive emergencies and pseudoemergencies? JAMA 1996;276:1328-31.
15. Varon J, Marik P. Clinical review: the management of Hypertensive crises. Crit Care 2003;7:374-84.
16. Bertel O, Conen D, Radu E, Muller J, Lang C, Dubach U. Nifedipine in hypertensive emergencies. Br Med J (Clin Res Ed) 1983;286:19-21.
17. Geyskes GG, Boer P, Mees DEJ. Clonidine withdrawal. Mechanism and frequency of rebound hypertension. Br J Clin Pharmacol 1979;7:55-62.
18. Wolf SJ, Lo B, Shih RD, Smith MD, Fesmire FM. Clinical policy: critical issues in the evaluation and management of adult patients in the emergency department with asymptomatic elevated blood pressure. Ann Emerg Med 2013;62:59-68.
19. Baumann BM, Cline DM, Cienki JJ, Egging D, Lehrmann JF, Tanabe P. Provider self- report and practice: reassessment and referral of emergency department patients with elevated blood pressure. Am J Hypertens 2009;22:604-10.
20. World Health Organization. A global brief on hypertension: silent killer, global public health crisis. World Health Day 2013. www.who.int. [Accessed January 10, 2014].
21. James PA, Oparil S, Carter BL, Cushman WC, Dennison-Himmelfarb C, Handler J, et al. 2014 evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the eighth joint national com- mittee (JNC 8). JAMA 2014;311:507-20.