Testing For Primary Hyperaldosteronism and MRA TX Annals 2020
Testing For Primary Hyperaldosteronism and MRA TX Annals 2020
Testing For Primary Hyperaldosteronism and MRA TX Annals 2020
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.
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
Figure 1. Cohort derivation and primary aldosteronism testing status, initial testing results, MRA use, and adrenalectomy status
among participants who met inclusion criteria.
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
Figure 2. Proportion of patients with apparent treatment-resistant hypertension tested for primary aldosteronism, by medical center
(left) and index year (right).
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.
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
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)
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.