Annals of Internal Medicine ORIGINAL RESEARCH

Testing for Primary Aldosteronism and Mineralocorticoid Receptor

Antagonist Use Among U.S. Veterans
A Retrospective Cohort Study
Jordana B. Cohen, MD, MSCE; Debbie L. Cohen, MD; Daniel S. Herman, MD, PhD; John T. Leppert, MD, MS;
James Brian Byrd, MD, MS*; and Vivek Bhalla, MD*

Background: Primary aldosteronism is a common cause of endocrinologist (HR, 2.48 [CI, 1.69 to 3.63]) was associated
treatment-resistant hypertension. However, evidence from with a higher likelihood of testing compared with primary
local health systems suggests low rates of testing for primary care. Testing was associated with a 4-fold higher likelihood
aldosteronism. of initiating MRA therapy (HR, 4.10 [CI, 3.68 to 4.55]) and
with better BP control over time.
Objective: To evaluate testing rates for primary aldosteron-
ism and evidence-based hypertension management in patients Limitations: Predominantly male cohort, retrospective design,
with treatment-resistant hypertension. susceptibility of office BPs to misclassification, and lack of con-
firmatory testing for primary aldosteronism.
Design: Retrospective cohort study.
Conclusion: In a nationally distributed cohort of veterans
Setting: U.S. Veterans Health Administration. with apparent treatment-resistant hypertension, testing for
primary aldosteronism was rare and was associated with
Participants: Veterans with apparent treatment-resistant
higher rates of evidence-based treatment with MRAs and
hypertension (n = 269 010) from 2000 to 2017, defined as ei-
better longitudinal BP control. The findings reinforce prior
ther 2 blood pressures (BPs) of at least 140 mm Hg (systolic)
observations of low adherence to guideline-recommended
or 90 mm Hg (diastolic) at least 1 month apart during use of
practices in smaller health systems and underscore the
3 antihypertensive agents (including a diuretic), or hyperten-
urgent need for improved management of patients with
sion requiring 4 antihypertensive classes.
treatment-resistant hypertension.
Measurements: Rates of primary aldosteronism testing (plasma
Primary Funding Source: National Institutes of Health.
aldosterone–renin) and the association of testing with evidence-
based treatment using a mineralocorticoid receptor antagonist
(MRA) and with longitudinal systolic BP.
Ann Intern Med. doi:10.7326/M20-4873 Annals.org
Results: 4277 (1.6%) patients who were tested for primary For author, article, and disclosure information, see end of text.
aldosteronism were identified. An index visit with a nephrol- This article was published at Annals.org on 29 December 2020.
ogist (hazard ratio [HR], 2.05 [95% CI, 1.66 to 2.52]) or an * Drs. Byrd and Bhalla contributed equally to this work.

H ypertension affects 46% of the adult population in

the United States and is a leading risk factor for dis-
ability, cardiovascular morbidity, and mortality (1, 2).
treated with MRAs or surgery (if lateralization is present)
(14). Accordingly, several national and international
medical societies, including the American College of
Although treatment reduces morbidity and mortality, Cardiology/American Heart Association (2) and the
approximately 17% to 20% of patients using antihyper- Endocrine Society (14), recommend assessing for primary
tensive medications have apparent treatment-resistant aldosteronism by measuring plasma aldosterone–renin
hypertension (3), defined as inadequately controlled activity or concentration in patients with treatment-resist-
blood pressure (BP) with 3 antihypertensive medications, ant hypertension and those who have hypertension with
including a diuretic, or a requirement for at least 4 antihy- hypokalemia.
pertensive medications to achieve adequate control. Studies of health systems in California (15), Illinois
Compared with patients who require fewer antihyperten- (16), and New York (17) found that testing rates for pri-
sive agents, those with apparent treatment-resistant mary aldosteronism were less than 3% among patients
hypertension are at increased risk for cardiovascular and for whom it is recommended. However, no similar study
all-cause mortality, independent of BP control (4). High- has been performed on a large scale, and whether test-
quality evidence supports the use of mineralocorticoid ing rates are low in a large, highly integrated health care
receptor antagonist (MRA) therapy for management of system is unknown. We aimed to evaluate the frequency
treatment-resistant hypertension (5, 6). of testing for primary aldosteronism in U.S. veterans with
Primary aldosteronism is a common cause of second-
ary hypertension and is highly prevalent among patients
with treatment-resistant hypertension (7). Primary aldos- See also:
teronism is associated with a 4- to 12-fold increased risk
for adverse cardiovascular events (including atheroscler- Web-Only
otic cardiovascular disease and arrhythmias) compared Supplement
with primary hypertension (8–13) and can be effectively
Annals.org Annals of Internal Medicine © 2020 American College of Physicians 1

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

ORIGINAL RESEARCH Testing for Primary Aldosteronism in U.S. Veterans

incident apparent treatment-resistant hypertension and and select internal medicine subspecialty practices. For
factors associated with testing. We also sought to assess baseline BMI and laboratory values, we recorded values
whether testing was associated with evidence-based closest to the index date, limited to those collected in the
treatment for apparent treatment-resistant hypertension year before the index date. We also reported the mini-
with MRA therapy and with differences in longitudinal BP mum serum potassium level between 1 January 2000
control. and the index date. Adherence was assessed using VHA
pharmacy fill data and was calculated as the proportion
of days covered by antihypertensive medication fills in
METHODS the year before cohort entry (25). Provider and VHA
Detailed information is provided in the Supplemental medical center were defined by the clinic where the
Methods section of the Supplement (available at Annals patient was seen on the index date.
.org).
End Points
Data Source The primary end point was testing for primary aldos-
We used national Veterans Health Administration teronism, defined as concomitant measurement of blood
(VHA) data from the VHA Corporate Data Warehouse, aldosterone concentration and either plasma renin activ-
which contains detailed diagnostic codes, laboratory ity or plasma renin concentration (8, 26). Secondary end
results, vital signs, and pharmacy fill records on approxi- points were initiation of MRA treatment and change in
mately 9 million veterans followed by the VHA nationally SBP over time.
(18). The Institutional Review Board at the Corporal
Michael J. Crescenz VA Medical Center in Philadelphia, Censoring
Pennsylvania, approved the study. We followed patients until they achieved the end point,
were lost to follow-up, or died, whichever occurred first.
Patient Population We defined loss to follow-up as no VHA encounters or
The study cohort was derived from the Antihyperten- pharmacy fills for 2 years.
sives in Obesity Management Cohort, an observational
cohort study that includes more than 1 million veterans with Statistical Analysis
incident hypertension (defined as ≥1 outpatient billing code We present demographic, clinical, laboratory, pro-
for essential hypertension and ≥2 consecutive antihyperten- vider, and center characteristics across patients by pri-
sive agent pharmacy fills ≤1 year apart) from 1 January 2000 mary aldosteronism testing status on or after the index
to 31 December 2017 (followed through 31 July 2019) and date. Continuous variables were compared using the
evidence of receipt of primary care in the VHA (19, 20). Kruskal–Wallis test and are reported as medians and
Participants were excluded if they had end-stage kidney dis- interquartile ranges (IQRs), and categorical and binary
ease or no body mass index (BMI) documented in the year variables were compared using the v2 test and are
before incident hypertension. reported as numbers and percentages.
For the current study, we restricted the cohort to We performed multivariable mixed-effects survival
Veterans with incident apparent treatment-resistant modeling to evaluate factors associated with time to pri-
hypertension, defined as either 2 successive BPs of at mary aldosteronism testing on or after incident apparent
least 140 mm Hg (systolic) or 90 mm Hg (diastolic) at treatment-resistant hypertension, as well as the associa-
least 1 month apart during use of 3 antihypertensive tion of time-updated primary aldosteronism testing with
agents (including a diuretic), or receipt of 4 antihyperten- initiation of MRA treatment. The analyses applied a ran-
sive classes (8, 15). We excluded patients who had test- dom slope and intercept model to account for medical
ing for primary aldosteronism or initiated MRA treatment center and provider random effects, with an independ-
ent covariance structure (27). We selected covariates a
before meeting criteria for apparent treatment-resist- priori based on factors known or suspected to potentially
ant hypertension and who had chronic kidney disease influence management of apparent treatment-resistant
stage 4 or 5 or end-stage kidney disease on or before hypertension, including patient-level factors (age, sex,
meeting criteria for apparent treatment-resistant race, BMI, SBP, DBP, estimated glomerular filtration rate
hypertension. The index date was defined as the first [eGFR], minimum potassium level, diabetes mellitus,
date that a veteran met criteria for apparent treatment- heart failure, arrhythmia, atherosclerotic cardiovascular
resistant hypertension. disease, stroke, smoking history, antihypertensive class,
number of antihypertensive agents, cancer, dementia,
Covariates alcohol misuse, and adherence), provider-level factors
We determined baseline covariates using previously (specialty and number of patients with treatment-resist-
ant hypertension), and center-level factors (number of
published algorithms (see the Supplemental Methods
patients seen annually, number of patients with treat-
section of the Supplement for details) (21–24). Baseline ment-resistant hypertension, rural location, and aca-
comorbidities were defined as those diagnosed on or demic affiliation) (8, 15). Continuous covariates were
anytime before the index date. We used VHA pharmacy assessed for nonlinear relationships with the outcome
fill data to determine antihypertensive and statin use. and incorporated using restricted cubic splines where
Baseline systolic BP (SBP) and diastolic BP (DBP) were appropriate.
defined as the mean values during the year before the For the analyses evaluating primary aldosteronism
index date, restricted to BPs measured in primary care testing and initiation of MRA therapy, incidence rates
2 Annals of Internal Medicine Annals.org

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

Testing for Primary Aldosteronism in U.S. Veterans ORIGINAL RESEARCH
were calculated per 1000 person-years of follow-up. likely to influence treatment-resistant hypertension man-
Restricted mean survival time was used to estimate the agement (8, 15).
difference in average survival time before initiation of We performed all analyses using Stata, version 16.0
therapy across testing status groups at 5 years of follow- (StataCorp), specifically the mestreg (31), strmst2 (32),
up (28, 29). We assessed for effect modification by index and mixed (33) packages.
year and hypokalemia (minimum prior potassium level In the year before incident treatment-resistant hyper-
≤3.5 or >3.5 mmol/L) and stratified the analyses by hypo- tension, BMI was missing in 304 (0.1%) patients, and
kalemia status. In sensitivity analyses, we censored eGFR and potassium level were missing in 11 400 (4%)
patients at the time of adrenalectomy or at the time of patients. We addressed missing data by multiple imputa-
testing for those with biochemical evidence of primary tion with chained equations, using 10 imputed data sets
aldosteronism (defined as plasma aldosterone concentra- to iterate each analysis (34).
tion ≥15 ng/dL and aldosterone–renin activity ratio ≥30 or
Role of the Funding Source
aldosterone–renin concentration ratio ≥4.8, along with
This research was supported in part by a grant from
suppressed renin) because many of these patients benefit
the National Institutes of Health. The funding source had
from adrenalectomy rather than MRA treatment and adre-
no role in study design or implementation. The interpre-
nalectomies may have been performed outside the VHA
tation and reporting of these data are the responsibility
(8, 14).
of the authors and should in no way be seen as an official
In secondary analyses, we performed linear mixed-
policy of or interpretation by the National Institutes of
effects modeling to evaluate the association of testing
Health.
for primary aldosteronism with BP over time, with ran-
dom intercepts for medical center, provider, and patient.
Sensitivity analyses applied marginal structural modeling RESULTS
with stabilized inverse probability weighting to account Baseline Characteristics
for time-updated confounding by MRA use, with weight A total of 279 376 veterans met criteria for incident
trimming at the first and 99th percentiles (30). We used apparent treatment-resistant hypertension during follow-
the same covariates for the secondary analyses as in the up. After exclusion of patients with advanced chronic
primary analyses because of overlapping factors that are kidney disease (n = 8358) and those who had testing for

Figure 1. Cohort derivation and primary aldosteronism testing status, initial testing results, MRA use, and adrenalectomy status
among participants who met inclusion criteria.

Incident hypertension between

1 January 2000 and 31 December 2017
(n = 1 157 629)
No resistant
hypertension
(n = 878 253)
Incident apparent treatment-resistant
hypertension (n = 279 376)
Chronic kidney disease
stage 4 or 5 or receiving
dialysis (n = 8358)
Resistant hypertension with mild or no
chronic kidney disease (n = 271 018) Tested for primary
aldosteronism or
received an MRA before
developing resistant
Primary cohort (n = 269 010) hypertension (n = 2008)

Tested for primary Not tested for primary

aldosteronism (n = 4277) aldosteronism (n = 264 733)

No evidence of Evidence of primary

primary aldosteronism aldosteronism on
on initial testing initial testing
(n = 3760) (n = 517)

Initiated MRA therapy Adrenalectomy

Initiated MRA Initiated MRA Adrenalectomy (n = 35 570) (n = 37)
therapy (n = 1053) therapy (n = 233) (n = 65)

MRA = mineralocorticoid receptor antagonist.

Annals.org Annals of Internal Medicine 3

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

ORIGINAL RESEARCH Testing for Primary Aldosteronism in U.S. Veterans

Table 1. Baseline Patient, Provider, and Center Characteristics, by Primary Aldosteronism Testing Status*
Characteristic Overall Cohort (n = 269 010) No Testing (n = 264 733) Testing (n = 4277)

Patient level
Median age (IQR), y 65 (58–72) 65 (58–72) 59 (52–66)
Female, n (%) 11 009 (4) 10 675 (4) 334 (8)
Black non-Hispanic, n (%) 50 883 (19) 49 517 (19) 1366 (32)
Obese, n (%) 135 551 (50) 133 164 (50) 2387 (56)
Median systolic blood pressure (IQR), mm Hg 140 (132–150) 140 (132–150) 145 (136–156)
Median diastolic blood pressure (IQR), mm Hg 79 (72–86) 79 (72–86) 84 (76–91)
Hypokalemia, n (%) 8313 (3) 7878 (3) 435 (10)
Diabetes, n (%) 106 617 (40) 104 892 (40) 1725 (40)
Obstructive sleep apnea, n (%) 28 342 (11) 27 918 (11) 424 (10)
Heart failure, n (%) 37 678 (14) 37 297 (14) 381 (9)
Arrhythmia, n (%) 66 023 (25) 65 268 (25) 755 (18)
Atherosclerotic cardiovascular disease, n (%) 64 479 (24) 63 677 (24) 802 (19)
Antihypertensive class, n (%)
ACEI or ARB 218 059 (81) 214 571 (81) 3488 (82)
dCCB 105 656 (39) 103 562 (39) 2094 (49)
Thiazide or thiazide-like diuretic 159 219 (59) 156 306 (59) 2913 (68)
b-Blocker 173 369 (64) 170 779 (65) 2590 (61)
Other 176 815 (66) 174 380 (66) 2435 (57)
Number of antihypertensive classes, n (%)
3 218 523 (81) 215 215 (81) 3308 (77)
4 48 337 (18) 47 450 (18) 887 (21)
≥5 2372 (1) 2290 (1) 82 (2)
Median adherence (IQR), % 89 (68–98) 89 (68–98) 88 (63–99)

Provider level
Median number of patients with resistant hypertension (IQR) 16 (7–31) 16 (7–31) 16 (6–32)
Index visit provider specialty, n (%)
Primary care 237 369 (88) 233 580 (88) 3789 (89)
Nephrology 2920 (1) 2800 (1) 120 (3)
Endocrinology 912 (<1) 876 (<1) 36 (<1)
Cardiology 28 031 (10) 27 699 (10) 332 (8)

Center level
Median number of patients seen annually (IQR) 55 253 (37 221–87 936) 55 253 (37 221–87 936) 60 283 (39 914–96 306)
Median number of patients with resistant hypertension (IQR) 2374 (1700–3716) 2374 (1700–3716) 2412 (1735–4405)
Rural location, n (%) 17 796 (7) 17 683 (7) 113 (3)
Academic affiliation, n (%) 173 969 (65) 170 793 (64) 3176 (74)
ACEI = angiotensin-converting enzyme inhibitor; ARB = angiotensin-receptor blocker; dCCB = dihydropyridine calcium-channel blocker;
IQR = interquartile range.
* All characteristics were measured on or before the date of incident resistant hypertension. Primary aldosteronism testing occurred on or
after the date of incident resistant hypertension. Systolic and diastolic blood pressures were calculated as the mean value in the year
before baseline. Laboratory values were the closest value obtained within 1 y before baseline. Hypokalemia was defined as a minimum po-
tassium level ≤3.5 mmol/L before the index date. Medications are listed if the patient had an active fill of the medication on or before base-
line based on the number of pills prescribed. A more extensive list of baseline characteristics is provided in Supplement Table 1 (available
at Annals.org).

primary aldosteronism (n = 1786) or received an MRA for primary aldosteronism. Figure 2 (left) shows the pro-
(n = 222) before meeting criteria for apparent treatment- portion of patients with apparent treatment-resistant
resistant hypertension, 269 010 patients met inclusion hypertension who were tested for primary aldosteronism
criteria (Figure 1). In the overall cohort, the median across each VHA medical center (n = 130). Testing rates
age was 65 years (IQR, 58 to 72 years), 4% were women, ranged from 0% to 6%, and the number of patients with
and 19% were Black Non-Hispanic race (Table 1; apparent treatment-resistant hypertension was not corre-
Supplement Table 1, available at Annals.org). There lated with testing rates across medical centers (r = 0.17).
were 517 patients who had biochemical evidence sug- Figure 2 (right) shows testing rates by index year. Testing
gesting primary aldosteronism based on their initial rates ranged from 1% to 2% per year, with slightly lower
testing.
rates among patients who met criteria for apparent treat-
ment-resistant hypertension in more recent years and
Factors Associated With Testing for Primary thus had shorter follow-up.
Aldosteronism We performed mixed-effects modeling to evaluate
After a median follow-up of 3.3 years (IQR, 1.0 to 6.7 factors associated with primary aldosteronism testing,
years) after meeting criteria for apparent treatment- accounting for differences in testing patterns across pro-
resistant hypertension, 4277 (1.6%) patients had testing viders and centers (Appendix Figure, available at Annals
4 Annals of Internal Medicine Annals.org

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

Testing for Primary Aldosteronism in U.S. Veterans ORIGINAL RESEARCH
.org). At the patient level, several factors, including hypo- survival time (CI, 0.9 to 1.2 years) before MRA therapy ini-
kalemia (standardized hazard ratio [HR], 1.93 [95% CI, tiation compared with those who were not tested. In
1.80 to 2.07]) and higher SBP (standardized HR, 1.43 [CI, models adjusted for patient-, provider-, and center-level
1.37 to 1.49]), were associated with a higher likelihood of factors, testing for primary aldosteronism was associated
undergoing testing (Supplement Table 2, available at with a 4-fold higher likelihood of subsequently starting
Annals.org). At the provider level, index visits with a MRA therapy compared with no testing (HR, 4.10 [CI,
nephrologist (HR, 2.05 [CI, 1.66 to 2.52]) or an endocri- 3.68 to 4.55]). There was effect modification by hypokale-
mia status but not index year. Patients with a history of
nologist (HR, 2.48 [CI, 1.69 to 3.63]), but not a cardiolo-
hypokalemia who underwent testing were more likely to
gist, were associated with a higher likelihood of testing
be treated with an MRA (HR, 7.11 [CI, 6.25 to 8.10]) than
compared with primary care. At the center level, rural those without hypokalemia (HR, 4.21 [CI, 3.59 to 4.94]).
location was associated with a lower likelihood of testing Sensitivity analyses that censored patients at the time of
than nonrural location (HR, 0.53 [CI, 0.31 to 0.91]). The adrenalectomy or at the time of demonstrating biochem-
number of patients with treatment-resistant hypertension ical evidence of primary aldosteronism showed similar
seen by a provider or center, overall center volume, and results.
center academic affiliation were not meaningfully associ-
ated with testing. Testing for Primary Aldosteronism and
BP Over Time
Testing for Primary Aldosteronism and Using linear mixed-effects modeling, we evaluated
Likelihood of Initiating MRA Therapy the association of testing for primary aldosteronism and
We performed mixed-effects survival modeling to longitudinal BPs (Table 3). These analyses accounted for
evaluate the association of testing for primary aldoster- differences in patterns of BP control across patients, pro-
onism with initiation of MRA therapy on or after meeting viders, and centers. Patients had a median of 17 BPs
criteria for apparent treatment-resistant hypertension (IQR, 8 to 31) measured over 4.2 years (IQR, 1.7 to 7.8
(Table 2). Similar to the prior analyses, these analyses years) of follow-up and had a mean SBP of 136.7 mm Hg
accounted for differences in testing patterns across pro- (CI, 136.4 to 136.9 mm Hg) during follow-up. Although
viders and centers. At 5 years of follow-up, patients who patients who underwent testing had higher baseline SBP
underwent testing had an average 1.1-year shorter (mean, 146.7 vs. 140.7 mm Hg), testing was associated

Figure 2. Proportion of patients with apparent treatment-resistant hypertension tested for primary aldosteronism, by medical center
(left) and index year (right).

7000 70% 20 000 70%

Number not tested

Number tested 18 000

Proportion of Patients With Treatment-Resistant Hypertension

6000 Proportion tested 60% 60%
Proportion of Patients With Treatment-Resistant Hypertension

16 000
Patients With Treatment-Resistant Hypertension, n
Patients With Treatment-Resistant Hypertension, n

50%
5000 50% 14 000

12 000
40%
4000 40%
10 000

30%
3000 30% 8000

6000
20%
2000 20%
4000

10%
1000 10% 2000

0 0%
20 1
02

20 3
20 4
20 5
06

20 7
20 8
20 9
20 0
11

20 2
13

20 4
20 5
20 6
17
0

0
0
0

0
0
0
1

1
1
1

0 0%
20

20

20

20

20

The columns depict the total number of patients with treatment-resistant hypertension (light green) and the total number with treatment-resistant hyper-
tension who were tested for primary aldosteronism (dark green), quantified on the left y-axis. The solid black line depicts the proportion of patients with
treatment-resistant hypertension who were tested for primary aldosteronism, quantified on the right y-axis.

Annals.org Annals of Internal Medicine 5

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

ORIGINAL RESEARCH Testing for Primary Aldosteronism in U.S. Veterans

Table 2. Likelihood of Initiating MRA Therapy on or After Testing for Primary Aldosteronism Versus No Testing*
Incidence per 1000 Average Survival Time Before
Group Adjusted Hazard
Person-Years Starting MRA Therapy (95% CI), y†
Ratio (95% CI)‡
No Testing Testing No Testing Testing Difference
Overall cohort 35 57 4.5 (4.4–4.5) 3.4 (3.2–3.6) 1.1 (0.9–1.2) 4.10 (3.68–4.55)
Patients with hypokalemia§ 43 82 4.3 (4.2–4.3) 3.0 (2.8–3.3) 1.3 (1.0–1.5) 7.11 (6.25–8.10)
Patients without hypokalemia§ 29 44 4.5 (4.5–4.5) 3.8 (3.5–4.0) 0.7 (0.5–1.0) 4.21 (3.59–4.94)
Censored at positive test result or 35 52 4.5 (4.4–4.5) 3.6 (3.4–3.8) 0.9 (0.7–1.1) 3.75 (3.34–4.20)
adrenalectomy
MRA = mineralocorticoid receptor antagonist.
* All adjusted analyses included baseline age, sex, race/ethnicity, body mass index, systolic and diastolic blood pressures (mean in previ-
ous year), estimated glomerular filtration rate, minimum potassium level, diabetes mellitus, heart failure, arrhythmia, atherosclerotic cardio-
vascular disease, stroke, smoking history, angiotensin-converting enzyme inhibitor or angiotensin-receptor blocker use, dihydropyridine
calcium-channel blocker use, thiazide/thiazide-like diuretic use, b-blocker use, other antihypertensive use, cancer, dementia, alcohol mis-
use, adherence, provider- and center-level number of patients with resistant hypertension, provider specialty, annual center volume, rural
center location, and center academic affiliation. The survival models used mixed-effects modeling with random intercepts for provider and
center.
† Estimated using restricted mean survival time at 5 y.
‡ Estimated using time-varying survival models evaluating the association of primary aldosteronism testing with initiation of MRA therapy.
§ Hypokalemia was defined as a minimum potassium level ≤3.5 mmol/L before the index date.

with an average 1.32–mm Hg (CI, 1.50 to 1.14 mm Recent data showed that the incidence of primary aldos-
Hg) lower SBP over time in unadjusted analyses (mean teronism in persons evaluated with confirmatory testing
decrease in SBP, 10.7 vs. 3.3 mm Hg). In analyses (24-hour urinary aldosterone levels after sodium loading)
adjusted for patient-, provider-, and center-level covari- is 11% in patients with normotension and 20% to 22% in
ates (including baseline BP), compared with no testing, those with apparent treatment-resistant hypertension (7,
testing for primary aldosteronism was associated with an 35). We observed that hypokalemia was associated with
average 1.47–mm Hg (CI, 1.64 to 1.29 mm Hg) lower
a higher likelihood of testing for primary aldosteronism.
SBP over time. The results were similar after adjustment
However, testing was infrequent even among patients
for MRA use.
with hypokalemia (5% testing rate [Table 1]), for whom
testing has been recommended in guidelines for deca-
DISCUSSION des (36). Moreover, most patients with primary aldoster-
This is, to our knowledge, the first large-scale multi- onism have a normal serum potassium level, such that a
center study of testing practices for primary aldosteron- large proportion fall outside the classic description of the
ism. We observed that fewer than 2% of patients with disease. Although a reliable testing approach to identify
incident apparent treatment-resistant hypertension under-
normotensive patients with primary aldosteronism is not
went guideline-recommended testing for primary aldos-
teronism. Testing rates ranged from 0% to 6% across available, testing of patients with apparent treatment-
medical centers and did not correlate to population size resistant hypertension is recommended by guidelines
of patients with apparent treatment-resistant hyperten- and is achievable in their routine care (26).
sion. Testing rates also did not change meaningfully over We considered several reasons for the low rates of
nearly 2 decades of follow-up despite an increasing testing. Various barriers to testing for primary aldoster-
number of guidelines recommending testing for primary onism have been identified via focus group from the per-
aldosteronism in this population. Our finding of infre- spective of German primary care clinicians (37). For
quent testing among patients with apparent treatment- example, some clinicians preferred an empirical trial of
resistant hypertension accords with prior studies in spironolactone without making a diagnosis, and others
smaller health systems (15–17). We also found that con- noted that stopping b-blockers before testing was
sultation with a nephrologist or an endocrinologist and impractical. However, we observed relatively low rates of
nonrural center location were independently associated MRA therapy initiation after patients met criteria for
with a higher likelihood of testing. In addition, we
apparent treatment-resistant hypertension, particularly
observed that testing for primary aldosteronism was
among those who were not tested for primary aldos-
associated with a substantially higher likelihood of initia-
tion of evidence-based MRA therapy for management of teronism. Only 13% of patients with incident apparent
apparent treatment-resistant hypertension (even in the treatment-resistant hypertension ultimately started MRA
absence of biochemical evidence of primary aldosteron- therapy, even though MRAs are recommended in approxi-
ism) and with greater improvement in BP over time. mately 70% of patients with treatment-resistant hyperten-
Although this may be the first large-scale study of sion (8). Our results show that empirical MRA therapy in
testing practices for primary aldosteronism in patients patients with apparent treatment-resistant hypertension is
with apparent treatment-resistant hypertension, many widely underused, particularly among those who are not
studies have established that the disease is not rare. tested for primary aldosteronism. In addition, testing for
6 Annals of Internal Medicine Annals.org

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

Testing for Primary Aldosteronism in U.S. Veterans ORIGINAL RESEARCH
primary aldosteronism can proceed without stopping med- Our study has several strengths. Because of the
ication use under most circumstances (26). detailed information in the VHA electronic health record,
Generally, failure to test patients with apparent treat- we were able to carefully identify patients with new-onset
ment-resistant hypertension for primary aldosteronism apparent treatment-resistant hypertension who were fol-
may reflect a lack of familiarity with this common and lowed at the VHA for their primary care since the time of
treatable condition or a broader propensity for treatment incident hypertension diagnosis, thus minimizing mis-
inertia in this patient population (38). We found no rela- classification of testing rates (for example, due to long-
tionship between center or provider volume of patients standing apparent treatment-resistant hypertension with
with apparent treatment-resistant hypertension and likeli- prior work-up or outmigration). The highly integrated na-
hood of testing for primary aldosteronism. Nonetheless, ture of the care in the VHA makes it unlikely that the test-
we did see substantial variation in testing practices across ing rates are lower than in health systems with more
centers and providers. We observed higher rates of test- fragmented care delivery systems. We also had access to
ing among patients seen by endocrinologists and neph-
pharmacy fill data, which facilitated classification of treat-
rologists (who typically perform confirmatory testing for
ment-resistant hypertension based on medications that
and oversee management of primary aldosteronism)
were more likely administered than if they had been
compared with those seen by primary care providers or
cardiologists. We also saw higher rates of evidence-based obtained from a standard electronic health record, as
MRA treatment and better BP control over time among well as determination of adherence based on the pro-
those who had testing, despite higher baseline BPs. portion of days with medications filled.
These findings support the hypothesis that testing coin- There are also important limitations. Because of the
cides with familiarity with primary aldosteronism and com- large proportion of men in the VHA, the findings may not
plex hypertension management. Further investigation is be fully generalizable to female patients, although the
needed into barriers to testing for primary aldosteronism cohort included more than 11 000 women with apparent
and ways to better implement best practices and guide- treatment-resistant hypertension. Because this is a retro-
lines among providers and medical centers that care for spective cohort study, it is prone to unmeasured con-
patients with treatment-resistant hypertension. founding. We were unable to measure serum antihyper-
Testing for primary aldosteronism in patients with tensive medication levels or directly observe therapy to
apparent treatment-resistant hypertension is cost- confirm treatment resistance. In addition, BPs in the elec-
effective and may substantially improve long-term out- tronic health record are susceptible to misclassification of
comes (39). In patients with positive results who undergo hypertension, in contrast to, for example, out-of-office BP
adrenalectomy for an aldosterone-secreting adenoma, monitoring or research-quality office BPs (44, 45). To mit-
compared with usual care, adrenalectomy is associated igate this, we restricted BP measurements to those
with lower risk for all-cause mortality (HR, 0.23 [CI, 0.13
obtained in practices focused on hypertension manage-
to 0.26]) (40), atrial fibrillation (HR, 0.55 [CI, 0.32 to 0.93])
ment, including primary care and select internal medi-
(41), and chronic kidney disease (difference in cumula-
tive incidence, 6.5 [CI, 2.7 to 10.4]) (42) and an improve- cine subspecialty practices. Finally, in the absence of
ment in quality-adjusted life-years (39). In addition to individual-level chart reviews, we were not able to deter-
detecting primary aldosteronism, testing can identify mine whether confirmatory testing (such as salt loading, flu-
patients with suppressed renin and normal aldosterone drocortisone suppression test, or captopril challenge test)
levels who may particularly benefit from treatment with an was performed for primary aldosteronism or the results of
MRA or another potassium-sparing diuretic (5, 43). Testing confirmatory testing, and adrenalectomies are suspected
can also help to identify apparent mineralocorticoid excess to have been undercaptured (8, 14, 26).
syndrome and Liddle syndrome (in which patients tend to In conclusion, our data provide evidence of consid-
have low renin and aldosterone levels) (8). erable underperformance of guideline-recommended

Table 3. Difference in SBP Over Time With Primary Aldosteronism Testing Versus No Testing*
Variable Average SBP During Difference in SBP
Follow-up (95% CI), mm Hg (95% CI), mm Hg
Unadjusted 136.7 (136.4 to 136.9) 1.32 ( 1.50 to 1.14)
Adjusted for baseline covariates 136.6 (136.4 to 136.7) 1.47 ( 1.64 to 1.29)
Adjusted for MRA use† 136.7 (136.4 to 137.0) 1.30 ( 1.93 to 0.66)

MRA = mineralocorticoid receptor antagonist; SBP = systolic blood pressure.

*All analyses used linear mixed-effects modeling with random intercepts for patient, provider, and center. Adjusted analyses included
baseline age, sex, race/ethnicity, body mass index, systolic and diastolic blood pressures (mean in previous year), estimated glomerular fil-
tration rate, minimum potassium level, diabetes mellitus, heart failure, arrhythmia, atherosclerotic cardiovascular disease, stroke, smoking
history, angiotensin-converting enzyme inhibitor or angiotensin-receptor blocker use, dihydropyridine calcium-channel blocker use,
thiazide/thiazide-like diuretic use, b-blocker use, other antihypertensive use, cancer, dementia, alcohol misuse, adherence, provider- and
center-level number of patients with resistant hypertension, provider specialty, annual center volume, rural center location, and center aca-
demic affiliation.
† These analyses used inverse probability weights to account for time-updated confounding by mineralocorticoid receptor antagonist use.

Annals.org Annals of Internal Medicine 7

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

ORIGINAL RESEARCH Testing for Primary Aldosteronism in U.S. Veterans

testing for primary aldosteronism in a national cohort of References

veterans with apparent treatment-resistant hypertension, 1. GBD 2017 Risk Factor Collaborators. Global, regional, and
bolstering similar observations in smaller health systems national comparative risk assessment of 84 behavioural, environ-
(15–17). We also showed that testing practices are mental and occupational, and metabolic risks or clusters of risks for
195 countries and territories, 1990–2017: a systematic analysis for
strongly associated with evidence-based treatment of
the Global Burden of Disease Study 2017. Lancet. 2018;392:1923-
apparent treatment-resistant hypertension with MRAs 94. [PMID: 30496105] doi:10.1016/S0140-6736(18)32225-6
and better BP control over time. The reasons for these 2. Whelton PK, Carey RM, Aronow WS, et al. 2017 ACC/AHA/
widespread missed opportunities for testing and treat- AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for
ment are unclear. The consequences of undertesting for the prevention, detection, evaluation, and management of high
primary aldosteronism and underuse of MRAs in patients blood pressure in adults: a report of the American College of
Cardiology/American Heart Association Task Force on Clinical Practice
with apparent treatment-resistant hypertension may be Guidelines. Hypertension. 2018;71:e13-e115. [PMID: 29133356]
substantial, potentially increasing morbidity and mortal- doi:10.1161/HYP.0000000000000065
ity. We identified several factors associated with under- 3. Carey RM, Sakhuja S, Calhoun DA, et al. Prevalence of apparent
testing. Given the unique infrastructure of the VHA, our treatment-resistant hypertension in the United States. Hypertension.
findings suggest an opportunity to introduce innovative 2019;73:424-31. [PMID: 30580690] doi:10.1161/HYPERTENSIONAHA
.118.12191
practices to meaningfully improve education of pro-
4. van der Sande NGC, de Beus E, Bots ML, et al; SMART study
viders and increase testing to enhance management in group. Apparent resistant hypertension and the risk of vascular
this high-risk patient population. Future studies should events and mortality in patients with manifest vascular disease.
evaluate implementation of tools to identify patients via J Hypertens. 2018;36:143-50. [PMID: 28763342] doi:10.1097
the electronic health record and alert providers to their /HJH.0000000000001494
5. Williams B, MacDonald TM, Morant S, et al; British Hypertension
appropriateness for testing.
Society's PATHWAY Studies Group. Spironolactone versus pla-
cebo, bisoprolol, and doxazosin to determine the optimal treat-
From Perelman School of Medicine, University of Pennsylvania, and ment for drug-resistant hypertension (PATHWAY-2): a randomised,
Corporal Michael J. Crescenz VA Medical Center, Philadelphia, double-blind, crossover trial. Lancet. 2015;386:2059-68. [PMID:
Pennsylvania (J.B.C.); Perelman School of Medicine, University of 26414968] doi:10.1016/S0140-6736(15)00257-3
Pennsylvania, Philadelphia, Pennsylvania (D.L.C., D.S.H.); Stanford 6. Zhao D, Liu H, Dong P, et al. A meta-analysis of add-on use of spi-
University School of Medicine, Stanford, California, and Veterans ronolactone in patients with resistant hypertension. Int J Cardiol.
Affairs Palo Alto Health Care System, Palo Alto, California (J.T.L.); 2017;233:113-7. [PMID: 28089457] doi:10.1016/j.ijcard.2016.12.158
University of Michigan Medical School, Ann Arbor, Michigan 7. Brown JM, Siddiqui M, Calhoun DA, et al. The unrecognized
(J.B.B.); and Stanford University School of Medicine, Stanford, prevalence of primary aldosteronism: a cross-sectional study. Ann
California (V.B.). Intern Med. 2020;173:10-20. doi:10.7326/M20-0065
8. Carey RM, Calhoun DA, Bakris GL, et al. Resistant hypertension:
detection, evaluation, and management: a scientific statement from the
Disclaimer: Department of Veterans Affairs Health Services
American Heart Association. Hypertension. 2018;72:e53-e90. [PMID:
Research and Development Service work was supported using 30354828] doi:10.1161/HYP.0000000000000084
resources and facilities at the VA Informatics and Computing 9. Smith SM, Huo T, Delia Johnson B, et al. Cardiovascular and
Infrastructure, VA HSR RES 13-457. The contents of this article mortality risk of apparent resistant hypertension in women with sus-
do not represent the views of the U.S. Department of Veterans pected myocardial ischemia: a report from the NHLBI-sponsored
Affairs or the U.S. government. WISE Study. J Am Heart Assoc. 2014;3:e000660. [PMID: 24584740]
doi:10.1161/JAHA.113.000660
Grant Support: Drs. J. Cohen and Byrd were supported by 10. de la Sierra A, Segura J, Banegas JR, et al. Clinical features of
grants from the National Institutes of Health, National Heart, 8295 patients with resistant hypertension classified on the basis of
Lung, and Blood Institute (K23-HL133843 and K23-HL128909, ambulatory blood pressure monitoring. Hypertension. 2011;57:898-
respectively). 902. [PMID: 21444835] doi:10.1161/HYPERTENSIONAHA.110.168948
11. Daugherty SL, Powers JD, Magid DJ, et al. Incidence and
prognosis of resistant hypertension in hypertensive patients.
Disclosures: Disclosures can be viewed at www.acponline.org
Circulation. 2012;125:1635-42. [PMID: 22379110] doi:10.1161
/authors/icmje/ConflictOfInterestForms.do?msNum=M20-4873. /CIRCULATIONAHA.111.068064
12. Egan BM, Zhao Y, Axon RN, et al. Uncontrolled and apparent
Reproducible Research Statement: Study protocol and statisti- treatment resistant hypertension in the United States, 1988 to 2008.
cal code: Available from Dr. Jordana Cohen (e-mail, jco Circulation. 2011;124:1046-58. [PMID: 21824920] doi:10.1161
@pennmedicine.upenn.edu). Data set: Available with appropri- /CIRCULATIONAHA.111.030189
ate approvals from Dr. Jordana Cohen. 13. Persell SD. Prevalence of resistant hypertension in the United
States, 2003–2008. Hypertension. 2011;57:1076-80. [PMID: 21502568]
doi:10.1161/HYPERTENSIONAHA.111.170308
Corresponding Author: J. B. Cohen, MD, MSCE, Perelman
14. Funder JW, Carey RM, Mantero F, et al. The management of
School of Medicine, University of Pennsylvania, 423 Guardian
primary aldosteronism: case detection, diagnosis, and treatment:
Drive, Blockley 831, Philadelphia, PA 19104; e-mail, jco@
an Endocrine Society clinical practice guideline. J Clin Endocrinol
pennmedicine.upenn.edu; Twitter: @jordy_bc. Metab. 2016;101:1889-916. [PMID: 26934393] doi:10.1210/jc.2015
-4061
Current author addresses and author contributions are avail- 15. Jaffe G, Gray Z, Krishnan G, et al. Screening rates for primary
able at Annals.org. aldosteronism in resistant hypertension: a cohort study. Hypertension.

8 Annals of Internal Medicine Annals.org

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

Testing for Primary Aldosteronism in U.S. Veterans ORIGINAL RESEARCH
2020;75:650-9. [PMID: 32008436] doi:10.1161/HYPERTENSIONAHA. 32. Cronin A, Tian L, Uno H. strmst2 and strmst2pw: New
119.14359 commands to compare survival curves using the restricted mean
16. Ruhle BC, White MG, Alsafran S, et al. Keeping primary aldoster- survival time. Stata J. 2016;16:702-16. doi:10.1177/1536867X160
onism in mind: deficiencies in screening at-risk hypertensives. Surgery. 1600310
2019;165:221-7. [PMID: 30415872] doi:10.1016/j.surg.2018.05.085 33. Rabe-Hesketh S, Skrondal A. Multilevel and Longitudinal
17. Sivarajah M, Beninato T, Fahey TJ 3rd. Adherence to consensus Modeling Using Stata, Third Edition. Stata Pr; 2012.
guidelines for screening of primary aldosteronism in an urban 34. White IR, Royston P, Wood AM. Multiple imputation using
healthcare system. Surgery. 2020;167:211-5. [PMID: 31564486] chained equations: issues and guidance for practice. Stat Med.
doi:10.1016/j.surg.2019.05.087 2011;30:377-99. [PMID: 21225900] doi:10.1002/sim.4067
18. Fihn SD, Francis J, Clancy C, et al. Insights from advanced ana- 35. Calhoun DA, Nishizaka MK, Zaman MA, et al. Hyperaldoster-
lytics at the Veterans Health Administration. Health Aff (Millwood). onism among black and white subjects with resistant hypertension.
2014;33:1203-11. [PMID: 25006147] doi:10.1377/hlthaff.2014.0054
Hypertension. 2002;40:892-6. [PMID: 12468575]
19. Suarez J, Cohen JB, Potluri V, et al. Racial disparities in nephrol-
36. Chobanian AV, Bakris GL, Black HR, et al; National Heart, Lung,
ogy consultation and disease progression among veterans with
and Blood Institute Joint National Committee on Prevention,
CKD: an observational cohort study. J Am Soc Nephrol. 2018;29:
Detection, Evaluation, and Treatment of High Blood Pressure. The
2563-73. [PMID: 30120108] doi:10.1681/ASN.2018040344
Seventh Report of the Joint National Committee on Prevention,
20. Serper M, Weinberg EM, Cohen JB, et al. Mortality and hepatic
Detection, Evaluation, and Treatment of High Blood Pressure: the
decompensation in patients with cirrhosis and atrial fibrillation
treated with anticoagulation. Hepatology. 2020. [PMID: 32267547] JNC 7 report. JAMA. 2003;289:2560-72. [PMID: 12748199]
doi:10.1002/hep.31264 37. Hannemann A, Bidlingmaier M, Friedrich N, et al. Screening for
21. Wilchesky M, Tamblyn RM, Huang A. Validation of diagnostic primary aldosteronism in hypertensive subjects: results from two
codes within medical services claims. J Clin Epidemiol. 2004;57: German epidemiological studies. Eur J Endocrinol. 2012;167:7-15.
131-41. [PMID: 15125622] [PMID: 22495491] doi:10.1530/EJE-11-1013
22. Bullano MF, Kamat S, Willey VJ, et al. Agreement between 38. Harle CA, Harman JS, Yang S. Physician and patient characteris-
administrative claims and the medical record in identifying patients tics associated with clinical inertia in blood pressure control. J Clin
with a diagnosis of hypertension. Med Care. 2006;44:486-90. Hypertens (Greenwich). 2013;15:820-4. [PMID: 24283597] doi:10
[PMID: 16641668] .1111/jch.12179
23. Quan H, Li B, Saunders LD, et al; IMECCHI Investigators. 39. Lubitz CC, Economopoulos KP, Sy S, et al. Cost-effectiveness of
Assessing validity of ICD-9-CM and ICD-10 administrative data in re- screening for primary aldosteronism and subtype diagnosis in the re-
cording clinical conditions in a unique dually coded database. sistant hypertensive patients. Circ Cardiovasc Qual Outcomes.
Health Serv Res. 2008;43:1424-41. [PMID: 18756617] 2015;8:621-30. [PMID: 26555126] doi:10.1161/CIRCOUTCOMES.
24. Suchard MA, Schuemie MJ, Krumholz HM, et al. Comprehen- 115.002002
sive comparative effectiveness and safety of first-line antihyperten- 40. Wu VC, Wang SM, Chang CH, et al. Long term outcome of
sive drug classes: a systematic, multinational, large-scale analysis. aldosteronism after target treatments. Sci Rep. 2016;6:32103.
Lancet. 2019;394:1816-26. [PMID: 31668726] doi:10.1016/S0140 [PMID: 27586402] doi:10.1038/srep32103
-6736(19)32317-7 41. Rossi GP, Maiolino G, Flego A, et al; PAPY Study Investigators.
25. Sattler EL, Lee JS, Perri M 3rd. Medication (re)fill adherence Adrenalectomy lowers incident atrial fibrillation in primary aldoster-
measures derived from pharmacy claims data in older Americans: a onism patients at long term. Hypertension. 2018;71:585-91. [PMID:
review of the literature. Drugs Aging. 2013;30:383-99. [PMID: 29483224] doi:10.1161/HYPERTENSIONAHA.117.10596
23553512] doi:10.1007/s40266-013-0074-z 42. Hundemer GL, Curhan GC, Yozamp N, et al. Renal outcomes
26. Byrd JB, Turcu AF, Auchus RJ. Primary aldosteronism: practical
in medically and surgically treated primary aldosteronism.
approach to diagnosis and management. Circulation. 2018;138:823-
Hypertension. 2018;72:658-66. [PMID: 29987110] doi:10.1161
35. [PMID: 30359120] doi:10.1161/CIRCULATIONAHA.118.033597
/HYPERTENSIONAHA.118.11568
27. Littell RC, Pendergast J, Natarajan R. Modelling covariance
43. Williams B, MacDonald TM, Morant SV, et al; British
structure in the analysis of repeated measures data. Stat Med.
Hypertension Society programme of Prevention And Treatment of
2000;19:1793-819. [PMID: 10861779]
Hypertension With Algorithm based Therapy (PATHWAY) Study
28. Royston P, Parmar MK. Restricted mean survival time: an alter-
native to the hazard ratio for the design and analysis of randomized Group. Endocrine and haemodynamic changes in resistant hyper-
trials with a time-to-event outcome. BMC Med Res Methodol. tension, and blood pressure responses to spironolactone or amilor-
2013;13:152. [PMID: 24314264] doi:10.1186/1471-2288-13-152 ide: the PATHWAY-2 mechanisms substudies. Lancet Diabetes
29. Uno H, Wittes J, Fu H, et al. Alternatives to hazard ratios for Endocrinol. 2018;6:464-75. [PMID: 29655877] doi:10.1016/S2213
comparing the efficacy or safety of therapies in noninferiority stud- -8587(18)30071-8
ies. Ann Intern Med. 2015;163:127-34. doi:10.7326/M14-1741 44. Cohen JB, Lotito MJ, Trivedi UK, et al. Cardiovascular events
30. Cole SR, Hernán MA. Constructing inverse probability weights and mortality in white coat hypertension: a systematic review and
for marginal structural models. Am J Epidemiol. 2008;168:656-64. meta-analysis. Ann Intern Med. 2019;170:853-62. doi:10.7326
[PMID: 18682488] doi:10.1093/aje/kwn164 /M19-0223
31. Crowther MJ. Multilevel mixed-effects parametric survival 45. Myers MG. The great myth of office blood pressure measure-
analysis: estimation, simulation, and application. Stata J. 2019;19: ment. J Hypertens. 2012;30:1894-8. [PMID: 22871894] doi:10.1097
931-49. doi:10.1177/1536867X19893639 /HJH.0b013e3283577b05

Annals.org Annals of Internal Medicine 9

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.

Current Author Addresses: Dr. J. Cohen: Perelman School of Author Contributions: Conception and design: J.B. Cohen, D.S.
Medicine, University of Pennsylvania, 423 Guardian Drive, Herman, J.T. Leppert, J.B. Byrd, V. Bhalla.
Blockley 831, Philadelphia, PA 19104. Analysis and interpretation of the data: J.B. Cohen, D.L. Cohen,
Dr. D. Cohen: Hospital of the University of Pennsylvania, 3400 J.T. Leppert, J.B. Byrd, V. Bhalla.
Spruce Street, Founders 1 Renal Division, Philadelphia, PA Drafting of the article: J.B. Cohen, D.L. Cohen.
19104. Critical revision of the article for important intellectual content:
Dr. Herman: University of Pennsylvania, 3700 Hamilton Walk, J.B. Cohen, D.L. Cohen, D.S. Herman, J.T. Leppert, J.B. Byrd, V.
Richards B202, Philadelphia, PA 19104. Bhalla.
Dr. Leppert: Stanford University School of Medicine, 300 Final approval of the article: J.B. Cohen, D.L. Cohen, D.S.
Pasteur Drive, Stanford, CA 94305. Herman, J.T. Leppert, J.B. Byrd, V. Bhalla.
Dr. Byrd: University of Michigan Medicine, Domino's Farms, 24 Obtaining of funding: J.B. Cohen.
Frank Lloyd Wright Drive, Ann Arbor, MI 48105. Administrative, technical, or logistic support: J.T. Leppert.
Dr. Bhalla: Stanford University School of Medicine, 777 Welch Collection and assembly of data: J.B. Cohen.
Road, Suite DE, Stanford, CA 94305.

Appendix Figure. Association of patient-, provider-, and center-level factors with testing for primary aldosteronism.

Patient-Level Factors
0.63
Older age
1.43
Higher systolic blood pressure
Hypokalemia 1.93
Black race 1.26
Provider-Level Factors
Higher volume of patients 0.92
with resistant hypertension
Specialty
Primary care (reference)
2.05
Nephrology
2.48
Endocrinology
Cardiology 0.95
Center-Level Factors
Higher volume of patients 1.05
Higher volume of patients 0.99
with resistant hypertension
0.53
Rural location
1.24
Academic affiliation

0 1 2 3 4
Hazard Ratio (95% CI)

The circles represent multivariable-adjusted hazard ratios, and the bars represent 95% CIs. The corresponding values and additional patient-level fac-
tors are tabulated in Supplement Table 2 (available at Annals.org). For continuous variables, the hazard ratios were calculated on the basis of each SD
change in the variable. All variables were measured at baseline, the date of incident treatment-resistant hypertension. The analyses were adjusted for
age, sex, race/ethnicity, body mass index, systolic and diastolic blood pressures (mean in previous year), estimated glomerular filtration rate, hypokale-
mia (defined as a minimum potassium level ≤3.5 mmol/L before the index date), diabetes mellitus, heart failure, arrhythmia, atherosclerotic cardiovascu-
lar disease, stroke, smoking history, angiotensin-converting enzyme inhibitor or angiotensin-receptor blocker use, dihydropyridine calcium-channel
blocker use, thiazide/thiazide-like diuretic use, b-blocker use, number of antihypertensive agents, cancer, dementia, alcohol misuse, adherence, pro-
vider- and center-level number of patients with treatment-resistant hypertension, provider specialty, annual center volume, rural center location, and
center academic affiliation. Hazard ratios were estimated using mixed-effects survival modeling with random intercepts for provider and center.

10 Annals of Internal Medicine Annals.org

Downloaded from https://annals.org by Tel Aviv University on 12/28/2020.