A pharmacy database was used to identify patients with resistant hypertension who could then be tested for the presence of primary hyperaldosteronism. Inclusion criteria were: (1) resistant hypertension defined as uncontrolled hypertension and use of 3 antihypertensive medication classes or ≥4 antihypertensive classes regardless of blood pressure; (2) low or normal potassium levels (≤4.9 mEq/L); and (3) continuous health care from October 1, 2008, to February 28, 2009. Exclusion criteria were: (1) past or current use of an aldosterone antagonist, or (2) a medication possession ratio (adherence) <80% for any antihypertensive drug. Hyperaldosteronism was classified as an aldosterone/renin ratio (ARR) ≥30. Using the computer, 746 patients were identified who met criteria. After manual chart review to verify inclusion and exclusion criteria, 333 patients remained. Of 184 individuals in whom an ARR was obtained, 39 (21.2%) had a ratio of ≥30. A computer database is useful to identify patients with resistant hypertension and those who may have primary aldosteronism.
Primary aldosteronism (PA) is a condition in which the hormone aldosterone is released in excess by the adrenal cortex.1 Causes of PA include an adrenal adenoma, unilateral or bilateral adrenal hyperplasia, or in rare cases an inherited glucocorticoid-remediable form of PA.2 PA can lead to hypertension and put a patient at increased risk for cardiovascular events such as myocardial infarction, cerebral hemorrhage, and excess left ventricular hypertrophy,3 more so than in age- and sex-matched patients with essential hypertension.4,5 It can also lead to hypertension that is resistant to treatment with many antihypertensive drugs. Therefore, the need to identify patients with this condition is very important. The Endocrine Society Clinical Practice Guideline (ESCPG) states that there are specific treatments available that improve the impact of this condition on clinical outcomes.5
The prevalence of PA has long been debated. Previously, it was suggested that PA was rare, occurring in <2% of hypertensive patients.6 However, in 1954, Dr Jerome Conn estimated it to be present in up to 20% of all antihypertensive patients.3 The ESCPG states that based on cross-sectional and prospective studies, PA is seen in >10% of hypertensive patients in both general and specialty settings.5
The screening test for PA introduced by Dunn and Espiner7 in 1976 is the aldosterone/renin ratio (ARR). Patients with PA are known to have a high level of plasma aldosterone along with expanded extracellular plasma volume with a resulting low level of plasma renin activity. Because of this, Dunn and Espiner reasoned that the expected ARR would be high in patients with PA vs those with essential hypertension. Since then, ARR has been used as the screening test of choice for PA.8
There are more than 60 million patients in the United States with hypertension,9 and many remain uncontrolled despite drug therapy. Consequently, we studied the value of using a computerized database to identify patients with uncontrolled hypertension who could then be screened for hyperaldosteronism.
We screened for resistant hypertension in nearly 70,000 veterans cared for in a Veterans Affairs (VA) Health Care System in Northern California from October 2008 to February 2009 using the VA computer database. Patients were identified as having resistant hypertension if they had a diagnosis of hypertension, were taking 3 antihypertensive agents and had uncontrolled hypertension (≥140/90 mm Hg), or were taking ≥4 antihypertensive agents with or without controlled hypertension. In addition, patients were included if they had low or normal potassium levels (≤4.9 mEq/L). Patients were excluded if they had a history of spironolactone or eplerenone use or were <80% adherent with antihypertensive medication as measured by medication possession ratios (MPRs). The MPR was calculated by the total days supply dispensed divided by the dispensing period in days. Further exclusion of patients occurred after verifying the computer data analysis through chart review, which occurred up to 2 months after the computer program was run. Further screening was performed to ensure that the most up-to-date data were being evaluated when sending request for laboratory testing, since medications, adherence, or blood pressure (BP) control might have changed between the date of the computer search and the date of chart review. Prescription data were captured on February 24, 2009. BP included the most recent systolic and diastolic values, searched on March 3, 2009, going back to May 30, 2008. Individual provider permission was requested for patients who met computer-based criteria for resistant hypertension. After obtaining individual provider permission, plasma for aldosterone and plasma renin determinations were obtained in the morning. Aldosterone/renin ratios (ARRs) were calculated from these values using the following formula:
Individuals with an ARR of ≥30 were suspected of having primary aldosteronism, while an ARR of <30 was considered normal.1,5 The study was approved by the Human Studies Subcommittee, Research Service, VA Northern California Health Care System. Individual consent was not required as analysis was done by group deindentified data.
Blood was obtained from patients in the sitting position and specimens were maintained at room temperature. Plasma renin activity was measured with an immunometric assay and plasma aldosterone with a radioimmunoassay. The renin assay was sufficiently sensitive to measure levels as low as 0.15 ng/mL·h.
Computer Data Analysis
Data from the patient electronic medical records including patient demographics, prescriptions, laboratory results, and diagnostic codes are transferred from the VA’s mainframe electronic medical record into Microsoft SQL data tables monthly. The data used in this study were organized according to inclusion/exclusion criteria using SQL and Access queries.
SigmaStat version 3.2 (Systat Software, Inc, San Jose, CA) was used for computer statistical analysis. Analysis compared patients with an elevated ARR with those with normal ratios. Student t test and Mann-Whitney rank sum test were used for continuous variables such as age and laboratory tests. Chi-square test was used to analyze the proportional use of antihypertensive drug classes. Statistical significance was defined as a P value <.05.
There were 5516 patients retrieved from the VA computer database who met our criteria for resistant hypertension. After matching these data with patient identifiers for hypertension International Classification of Diseases, Ninth Revision (ICD-9) code and elevated potassium and excluding individuals with hyperkalemia, 2162 patients remained. After excluding patients with a history of spironolactone or eplerenone use, patients taking 3 antihypertensive agents with normal BP, and patients with an MPR of <80%, 746 patients remained. Prior to obtaining aldosterone and renin values, charts were manually reviewed for data subsequent to the computer database review to validate adherence by reviewing recent pharmacy prescription-filling activity. After this review, 333 patients with resistant hypertension remained. The providers of these patients were contacted by letter requesting permission to order laboratory tests. Plasma aldosterone and renin values were then ordered in 207 patients who continued to meet criteria and in whom provider permission was obtained.
Aldosterone and renin values were obtained in 184 patients, and of these 39 (21.2%) were identified as having primary aldosteronism defined as an elevated ARR of ≥30 (Figure). In these patients, the average aldosterone and plasma renin levels were 17.6 (standard deviation, 9.8; range, 5.9–62.1) and 0.283 (standard deviation, 0.251; range, 0.15–1.61), respectively. Baseline age, systolic BP, diastolic BP, and serum creatinine were similar between the group with elevated ARR and the group with normal ARR (Table I). Baseline potassium level was statistically but not clinically significantly elevated in the group without PA (4.0 vs 4.2, respectively; P=.023). Of 13 patients with potassium levels <3.5, five (38.5%) screened positive for hyperaldosteronism. Of 109 patients with a potassium level >4.0, seventeen (15.6%) screened positive for hyperaldosteronism. As expected, patients with lower potassium levels were more likely to screen positive with an elevated ARR. Baseline antihypertensive agents used in the 184 patients who had their ARRs obtained were similar (Table II).
Table I. Baseline Patient Population Characteristics According to ARR
ARR ≥30 (n=39)
ARR <30 (n=145)
Abbreviations: ARR, aldosterone/renin ratio; DBP, diastolic blood pressure; NS, not significant; SBP, systolic blood pressure; SCr, serum creatinine concentration. Values are mean ± standard deviation.
This study was designed to determine whether a computer database can be used to screen for patients with resistant hypertension who might have undiagnosed PA. The prevalence of PA in patients with resistant hypertension in our study was 21.2%. This is similar to the 10% to 20% reported by authors from previous studies.3,5,10,11 Although our study eliminated many potential patients, a strength of our study was the requirement that patients had an MPR of ≥80%. Resistant hypertension cannot be diagnosed if patients are not taking their medication.
There is no universally agreed upon value of the ARR for the diagnosis of PA. In prior reports, it has ranged from 20 to 50.5 We used an ARR of ≥30 to identify PA because this is the most commonly adopted cutoff point based on recommendations from the ESCPG.5 We did not use aldosterone values in addition to ARR since some individuals with primary aldosteronism may present with normal aldosterone levels.12,13
Providers of patients with elevated ARRs were sent an algorithm for further workup that might include computed tomography.14,15 It was left to the patients’ physicians whether to perform confirmatory tests or to pursue further evaluation for surgically correctable lesions. However, since the VA patient population is older (>65 years), some patients might not be candidates for surgery if an abnormality was detected.5 Additionally, many patients in our population are individuals with comorbidities, and some confirmatory tests might worsen these conditions; for example, the oral sodium loading test may exacerbate congestive heart failure. These tests may also be unreliable,16 and the time spent to perform these tests might delay the treatment of PA.17
There has been controversy over the diagnostic accuracy of ARR in screening for PA, particularly in patients taking certain antihypertensive medications. In a prospective cohort study of 118 individuals with a previous diagnosis of essential hypertension, patients obtained an ARR while they were still either receiving antihypertensive drug therapy, after 2 weeks of no antihypertensive medication use, after 4 days of dietary sodium loading, or after patients were given acute furosemide diuresis. Primary aldosteronism was diagnosed by 24-hour urine aldosterone excretion and plasma renin activity on the fourth day of dietary sodium loading. There was no significant difference across all 4 groups in the diagnostic accuracy of the ARR (P=.72). The authors concluded that although being on antihypertensive therapy can influence optimal cutoff values of the ARR (eg, 12.4 mL/dL/h on antihypertensive therapy compared with 14.9 mL/dL/h off antihypertensive therapy), the diagnostic accuracy of the ARR was not significantly affected by antihypertensive medication (P=.49). In addition, the sodium diet did not have a significant effect on ARR and, therefore, a controlled sodium diet before obtaining ARR was not felt to be necessary.8
Another study also examined the effects of antihypertensive drug therapy on ARR in PA. A total of 230 individuals were taking either amlodipine 10 mg/d, fosinopril 20 mg/d, atenolol 100 mg/d, doxazosin 8 mg/d, or irbesartan 300 mg/d for 2 months. The ARR cutoff was >50 to better improve the specificity of screening. Atenolol significantly increased ARR compared with the other medications (P<.0001) and can possibly cause a false-positive ARR. In addition, patients taking irbesartan required a confirmatory test because of the high rate of a false-negative diagnosis (23.5%). Only one individual from the amlodipine group had a false-negative ARR (1.8%). Doxazosin and fosinopril had no significant affect on ARR and, therefore, can be used in patients while screening for PA.18
The ESCPG advises that restricting dietary salt to obtain ARR is not necessary. However, it recommends that patients with mild to moderate hypertension have a washout period of all antihypertensive medications that may interfere with ARR determination. It also recommends that individuals with severe hypertension substitute other medications with minimal effect on the ARR such as verapamil, hydralazine, and α-adrenergic blockers before screening.5 In our observational study, it was not possible to do a washout, but we do not believe concurrent use of antihypertensive medications had a significant effect on the results because the proportion of patients using different antihypertensive classes was similar between groups.
Hypokalemia is a frequent finding of PA. In one study, the clinical and laboratory profiles of 309 individuals with either essential hypertension or PA were compared. A total of 260 patients had essential hypertension and 49 patients had PA. Patients with PA had substantially lower serum potassium levels vs those with essential hypertension (P< .001), but normal potassium levels were seen in 61% of PA patients (3.4–5.2 mmol/L).19 We report similar results, although the absolute differences were small and probably clinically irrelevant. Decreased potassium levels may be less indicative of hyperaldosteronism since more patients take antihypertensives that increase potassium levels such as angiotensin-converting enzyme inhibitors and angiotensin receptor blockers (ARBs). Of our 183 patients who obtained an ARR, 90.7% were taking an ACE inhibitor or ARB.
The ESCPG recommends that patients with confirmed PA undergo computed tomography, in part for subtype classification and to exclude potential adrenal carcinoma.5 The focus of our study was on identification of patients who could have PA. We did not address confirmation or subtype classification. Given the frequency of nonfunctional unilateral adrenal macroadenoma in our patient age group,20 computed tomography is probably not warranted as an early diagnostic modality.
Patients with hypertension were identified using diagnostic ICD-9 code, which may have been incorrectly inputted by the provider. Patients who were using antihypertensive medications provided by pharmacies outside the VA may not be identified. Some providers did not respond to our request for permission to obtain plasma aldosterone and renin values. In addition, we were unable to reach all patients who qualified for a laboratory blood draw. Medication adherence was determined by assessing patients’ medication refill history. This does not indicate that patients were taking their antihypertensive medications on a daily basis. Certain conditions such as renal impairment, pseudohypoaldosteronism type 2, renovascular hypertension, hypokalemia and hyperkalemia, and malignant hypertension can lead to false-positive and -negative ARR results. We did not screen for most of these conditions. Finally, not all patients who should have been tested obtained aldosterone and plasma renin activity. This limits the generalizability of our results.
The ARR is most sensitive when used in patients from samples collected in the morning.5,17 Patients were instructed to go to the VA laboratory in the morning but we had 15 patients who came in the afternoon. Three of the 15 patients who came in the afternoon had an elevated ARR. The Endocrine Society’s guidelines state that in many cases, the ARR still can be confidently interpreted even during suboptimal conditions or if the patient continues to use medications while obtaining the aldosterone and renin levels.
Despite these limitations, we believe that our study provides valuable information about the utility of using the computer to screen for resistant hypertension and identifying individuals who require further evaluation. Approximately 20% of patients with resistant hypertension will have primary aldosteronism and may respond to aldosterone antagonists, resulting in better BP control and a decrease in the morbidity and mortality associated with hypertension.
Acknowledgments and disclosures: Support was provided in part by the Research Service, VA Northern California Health Care System. Presented in preliminary form at the 92nd Annual Meeting, The Endocrine Society, San Diego, CA, June 2010. The authors have no conflict of interest to declare.