Increasing Incidence of Primary Aldosteronism in Western Sweden During 3 Decades - Yet An Underdiagnosed Disorder.

CONTEXT: Primary aldosteronism (PA) is the most common cause of secondary hypertension. Yet, the incidence of PA in the general population has not been studied.
OBJECTIVE: To estimate the incidence of PA in the general population. DESIGN AND METHODS: Patients who had received a diagnostic code for PA between 1987 and 2016 were identified in the Swedish National Patient Registry. Assessment of clinical and biochemical data was used to validate the diagnosis. The annual incidence of PA was calculated by using the number of inhabitants in the Västra Götaland County as a reference.
RESULTS: Of 570 identified patients, 473 (83%) had confirmed PA. Eligible for the incidence analysis were 416 patients, 248 (60%) men and 168 (40%) women, diagnosed with PA between 1987 and 2016. The mean (± standard deviation) age at diagnosis was 56 ± 12 years. The median (interquartile range) annual incidence was 2 (1-2) cases per million between 1987 and 1996, 6 (4-9) cases per million between 1997 and 2006 and 17 (12-24) cases per million between 2007 and 2016. At the end of the study (December 31, 2016), 386 patients with confirmed PA were alive and living in the Västra Götaland County, giving a prevalence of 231 cases per million (0.022%).
CONCLUSIONS: Despite increasing incidence, the proportion of patients identified with PA is lower than expected. Given the serious consequences of untreated PA, the noticeably low prevalence at the end of the study stresses the need to increase the awareness of PA among health care providers.

Primary aldosteronism (PA) affects a significant proportion of the adult hypertensive population, ranging from 4% in patients with stage 1 hypertension, 12% in patients with stage 3 hypertension, and up to 20% in patients with treatment-resistant hypertension (1-5). A prevalence as high as 7% in prehypertensive patients (6), and 11% in normotensive patients (7), have also been reported. PA is associated with increased risk of several comorbidities, especially cardiovascular diseases, as well as increased mortality (8, 9). Importantly, the outcome improves substantially in patients with PA who receive disease-specific treatment with either adrenalectomy, or mineralocorticoid receptor antagonists, emphasizing the importance of disease detection (9, 10).

Unfortunately, PA remains an underdiagnosed disease. In a recent register-based study from northern Italy, only 2% of patients expected to have PA were correctly diagnosed with the disease (11). Similarly, according to a web-based cross-sectional survey, 19% and 36% of general practitioners in Germany and Italy, respectively, have never diagnosed PA among their hypertensive patients (12). Also, in a large cohort of US veterans with treatment-resistant hypertension, only 1.6% were screened for PA (13).

Most epidemiological studies on PA have focused on the prevalence in prespecified patient populations such as patients with hypertension (2). To the best of our knowledge, the incidence of PA in the general population has not yet been evaluated. The estimation of the current diagnostic rates of PA may be valuable for several reasons, including optimizing health care policies aiming at earlier and more widespread detection of the disease.

The objective of this study was to evaluate the incidence of PA in Västra Götaland Country in Sweden from 1987 to 2016. For this purpose, we have used the National Patient Register (NPR) that includes more than 99% of all diagnostic codes provided through hospital visits across Sweden. The reliability of data obtained from the NPR for diseases such as myocardial infarction and hip fracture is high (14), but unknown for PA. Thus, a secondary aim was to validate the diagnosis in patients with presumed PA, identified in the NPR, by review of medical records.

Methods

Design

This was a register study including all patients from Västra Götaland County, Sweden who were registered with a diagnostic code for PA in the NPR between January 1, 1987, and December 31, 2016. To validate the diagnosis, the medical records of the patients were reviewed.

Data sources

The NPR is a population-based registry that collects diagnostic codes from all inpatient hospital visits across Sweden since 1987 and all specialist outpatient visits since 2001. All diseases, as well as external causes of injury and surgical procedures, are registered in the NPR according to the International Classification of Diseases, 9th revision (ICD-9) between 1987 and 1996, and the 10th revision (ICD-10) since 1997. By using a personal identification number, unique for all residents in Sweden, it was ensured that each patient with PA was taken into account only once.

Identification of patients

Patients with PA were identified in the NPR by the following codes: E260 (ICD-10) and 255.B (ICD-9). Every occasion producing a diagnostic code for PA was recorded, providing data on personal identification number, sex, date of diagnosis, and the clinic at which the diagnosis was made. In addition, the presence of adrenal tumor (ICD-9: 227.0 and ICD-10: D350 and/or D441) in a patient with a diagnostic code for PA was also recorded.

Validation of the diagnosis

To validate the diagnosis of PA, the medical records of all patients identified in the NPR were reviewed by an experienced endocrinologist. By reviewing the clinical, biochemical, imaging, and histopathological data, the patients were assigned to one of the following groups: (1) confirmed PA, (2) PA probable but not confirmed, (3) not PA, or (4) medical records not available. Patients with one of the following were considered to have confirmed PA (15): (1) a diagnostic test confirming PA; (2) severe phenotype with high plasma aldosterone concentration (>555 pmol/L [>20 ng/dL]), low or suppressed direct plasma renin activity or plasma renin concentration, and spontaneous hypokalemia; or (3) patients treated medically or surgically for PA after endocrine evaluation, although raw data on the diagnostic procedures were unavailable.

Evaluation of coverage in the NPR

To study the proportion of patients with confirmed PA who were not registered with the diagnosis in the NPR, we used a cohort of patients with confirmed PA who had undergone adrenal venous sampling (AVS) at our institute since 2006 (16). These patients were identified through a search in the electronic database at the department of radiology by using the specific codes for AVS.

Statistical analysis

Descriptive data are presented as mean ± SD or median (interquartile range; range). For comparison between 2 groups, unpaired t test was used for normally distributed data and Mann-Whitney U test for nonnormally distributed data. For proportions, Pearson χ 2 or Fisher’s exact test was used. A P value < 0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 25.

Calculations of annual notified incidence and prevalence of PA

After validation of the diagnosis of PA for patients identified in the NPR, the annual notified incidence and prevalence for patients with confirmed diagnosis were calculated by using the number of residents in the Västra Götaland County as reference. Information on the number of residents was obtained from the Swedish Total Population Register (17). The Västra Götaland County is the second most populated county in Sweden, with 1.41 million inhabitants at the beginning of the study (January 1, 1987), and 1.67 million at the end of the study (December 31, 2016).

The annual incidence is presented as the number of patients diagnosed with PA per million inhabitants/year. The date of the first endocrine evaluation because of suspected, and later confirmed, PA was used as the date of diagnosis. Excluded from the incidence analysis were patients diagnosed before 1987, patients with PA not living in Västra Götaland County, and patients diagnosed before they moved to the county. The annual incidence was also calculated for 3 10-year periods (1987-1996, 1997-2006, and 2007-2016), where the first period encompassed patients with PA registered according to ICD-9 and the following 2 periods encompassed patients registered with an ICD-10 code.

The prevalence of PA at the end of the study was determined by calculating the proportion of inhabitants with PA living in the Västra Götaland County in December 2016. Thus, patients with PA who had died or moved from the county during the study period were excluded from the prevalence analysis.

Ethics approval

The study was approved by the local ethics committee of the University of Gothenburg, Göteborg, Sweden (reference number 029-17; approved February 22, 2017). The study was conducted according to the Declaration of Helsinki.

Results

In total, 570 patients received a diagnostic code for PA between 1987 and 2016 in Västra Götaland County (Fig. 1). For the validation of the diagnosis, medical records were available for 553 (97%) patients. Of all identified patients, 473 (83%) had confirmed PA and 17 (3%) had probable PA (Table 1). Sixty-three (11%) patients did not have PA. Of 17 (3%) patients whose medical records were unavailable, 13 had received the diagnostic code for PA during the ICD-9 period (between 1987 and 1996).

Figure 1.

Flow chart of patient identification.

Table 1.

Influence of different search algorithms on the accuracy of the diagnosis of PA

	Diagnostic code for PA in ICD-9 and/or ICD-10	Diagnostic code for PA in ICD-9	Diagnostic code for PA in ICD-10	Diagnostic code for PA in ICD-9 and/or ICD-10, registered on > 1 occasion	Diagnostic code for PA in ICD-10, registered on > 1 occasion	Diagnostic code for PA and an adrenal tumor in ICD-10
Total number	570	62	522a	411	393	178
Confirmed PA	473 (83%)	35 (56%)	452 (87%)	380 (92%)	368 (94%)	174 (98%)
Probable PA	17 (3%)	3 (5%)	14 (3%)	5 (1%)	5 (1%)	1 (<1%)
Not PA	63 (11%)	11 (18%)	52 (10%)	21 (5%)	18 (5%)	3 (2%)
Missing data	17 (3%)	13 (21%)	4 (<1%)	5 (1%)	2 (<1%)	0

Abbreviations: ICD-9, International Classification of Diseases, 9th revision; ICD-10, International Classification of Diseases, 10th revision; PA, primary aldosteronism.

14 patients had received a diagnostic code for PA according to both ICD-9 and ICD-10.

Of 473 patients with confirmed PA, 345 (73%) had a positive confirmatory test (saline infusion test, n = 325), and 53 (11%) had a severe phenotype, considered to be confirmative for PA without further testing. Of the remaining 75 (16%) patients, in whom the original biochemical evaluation was unavailable, 68 were considered to have confirmed PA based on administration of disease-specific treatment for presumed PA (i.e., adrenalectomy [n = 40] and medical treatment [n = 28]). The remaining 7 patients had undergone AVS because of PA, but clinical data were otherwise lacking.

Of 17 patients with probable PA, 15 had a high plasma aldosterone-to-renin ratio but were not investigated further because of multiple comorbidities (n = 9), unwillingness to proceed with confirmatory testing (n = 5), and concomitant pheochromocytoma (n = 1). For the remaining 2 patients, only information from radiological reports was available, stating that the patients had PA and an “aldosteronoma.”

The diagnostic code for PA was incorrectly given in 63 patients without the disease. In 31 of these, PA had been suspected but subsequently ruled out biochemically. Erroneous coding was found in 32 patients (Fig. 1).

Alternative search algorithms

Of 62 patients registered with a diagnostic code for PA between 1987 and 1996 (ICD-9), 35 (56%) had confirmed PA and 3 (5%) had probable PA (Table 1). Of 522 patients registered with a diagnostic code for PA between 1997 and 2016 (ICD-10), 452 (87%) had confirmed PA and 14 (3%) had probable PA.

Of 522 patients registered 1997 or later (ICD-10), 393 (75%) were registered with PA on more than 1 occasion (Table 1). Of these, 368 (94%) patients had confirmed PA. Of 178 patients registered 1997 or later, with a diagnostic code for adrenal tumor, 174 (98%) had confirmed PA.

Coverage in the NPR

Of 217 resident patients who had underwent AVS at our institute between 2006 and 2016 because of PA, 7 (3%) were not registered in the NPR with a diagnostic code for PA. In other words, 97% of this subgroup of patients with PA were identified in the NPR.

Incidence of confirmed PA in Västra Götaland County, 1987-2016

In total, 416 patients with confirmed PA, diagnosed between 1987 and 2016, 248 (60%) men and 168 (40%) women, were included in the incidence analysis (Fig. 1). The median (interquartile range) annual incidence during the whole study period was 6 (2-13) cases per million; 9 (1-17) cases per million for men and 5 (1-11) cases per million for women (P = 0.2; Fig. 2). The median (interquartile range) annual notified incidence was 2 (1-2) cases per million between 1987 and 1996, 6 (4-9) cases per million between 1997 and 2006 and 17 (12-24) cases per million between 2007 and 2016 (Fig. 2).

Figure 2.

The annual incidence of primary aldosteronism by calendar year (cases per 1 million inhabitants and year) in Västra Götaland Country between 1987 and 2016.

The mean age at diagnosis was 56 ± 12 years (range 21-86); 54 ± 13 in women and 57 ± 10 in men (P = 0.02; Table 2). At diagnosis, 233 (60%) patients were treated with multiple antihypertensive medication and hypokalemia was present in 298 (79%) patients. Of 392 patients, a confirmatory test had been performed in 316 (81%), most commonly the saline infusion test (n = 303). AVS had been performed in 282 (68%) patients, of whom 263 were successful (93% success rate). Four hundred (97%) patients had received disease-specific treatment, 176 (43%) with adrenalectomy, and 224 (54%) with MRA. The number of patients operated on with adrenalectomy because of PA increased from 7 patients between 1987 and 1996, to 48 between 1997 and 2006 and to 121 patients between 2007 and 2016 (Table 2, Fig. 3).

Table 2.

Characteristics of patients diagnosed with PA in Västra Götaland County between 1987 and 2016

	All	Men	Women	P	1987-1996	1997-2006	2007-2016	P a
	(n = 416)	(n = 248)	(n = 168)		(n = 24)	(n = 104)	(n = 288)
Age, mean ± SD	56 ± 12	57 ± 10	54 ± 13	0.02	53 ± 10	55 ± 10	57 ± 12	0.13
No of antihypertensive drugs at diagnosis, median (IQR)b	2 (1-3)	2 (2-3)	2 (1-2)	0.002	2 (1-2)	2 (1-3)	2 (1-3)	0.9
Potassium supplementation at diagnosis, n (%)b	298 (79)	177 (77)	121 (82)	0.2	8 (100)	71 (81)	219 (78)	0.6
Saline infusion test, n (%)	303 (73)	179 (72)	124 (74)	0.13	2 (8)	51 (49)	250 (87)	<0.001
Adrenal vein sampling, n (%)	282 (68)	177 (71)	105 (63)	0.06	4 (17)	48 (46)	230 (80)	<0.001
Unilateral disease, n (%)c	145 (51)	87 (49)	58 (55)	1.0	0	22 (45)	123 (53)	0.3
Bilateral disease, n (%)	118 (42)	75 (42)	43 (41)	0.8	1 (25)	17 (35)	100 (43)	0.3
Unsuccessful, n (%)	19 (7)	15 (8)	4 (4)	0.13	3 (75)	9 (19)	7 (3)	<0.001
Treatment, n (%)
Adrenalectomy	176 (43)	100 (41)	76 (46)	0.3	7 (35)	48 (46)	121 (42)	0.5
MRAd	224 (54)	139 (57)	85 (51)	0.3	10 (50)	51 (49)	163 (57)	0.2
Other treatment	12 (3)	7 (3)	5 (3)	0.9	3 (15)	5 (5)	4 (1)	-

Abbreviations: IQR, interquartile range; MRA, mineralocorticoid receptor antagonist; PA, primary aldosteronism.

Because of the small number of patients diagnosed between 1987 and 1996, only differences between 1997 and 2006 and 2007 and 2016 were compared.

Data on number of antihypertensive drugs at diagnosis, potassium supplementation, imaging, and treatment were not available for 78, 39, 27, and 4 patients, respectively.

A lateralization index greater than 4.0, or a lateralization index between 3 and 4 together with a contralateral index below 1.0, were considered to be compatible with unilateral disease.

Patients with a lateralization index < 3 on adrenal vein sampling, and patients not considered to be candidates for surgery, received treatment with MRA.

Figure 3.

The annual incidence of adrenalectomies for primary aldosteronism by calendar year (operations per 1 million inhabitants and year) in Västra Götaland Country between 1987 and 2016.

Prevalence

At the end of the study on December 31, 2016, 386 patients with confirmed PA were alive and living in Västra Götaland County, with a population of 1,671,783 inhabitants. Thus, the prevalence was 231 cases per million (0.022%).

Discussion

In this register-based study, we found an overall annual incidence of PA to be 6 cases per million. The mean number of notified incident cases increased in a linear fashion over time, from 2 cases per million/year (1987-1996), to 17 cases per million (2007-2016). At the end of the study, the prevalence of PA in Västra Götaland was 231 cases per million (0.022%). Dedicated work on PA in the primary health care (18), and quality improvement on AVS (16), corresponding to the dramatic increase of adrenalectomies around 2005 (Fig. 3), may have contributed to increased awareness of the disease in our region. Still, previous studies on Swedish hypertensive patients have shown a prevalence of PA between 2% and 6% (18-20). Also, in Swedish patients with type 2 diabetes and hypertension, the prevalence of PA was 1% (21). Based on these studies, and given that 27% of the Swedish adult population have hypertension (22), the estimated number of patients with PA in the Västra Götaland County should be between 3500 and 20 000 cases. Thus, despite an increasing detection rate, PA is obviously still a greatly underdiagnosed disorder.

PA is a serious condition associated with increased risk of cardio- and cerebrovascular diseases. In a recent meta-analysis, patients with PA were found to run a 2-fold increased risk for coronary artery disease and congestive heart failure, 3-fold increased risk for stroke, and 4-fold increased risk for atrial fibrillation, compared with patients with essential hypertension (8). Mortality is also increased in patients with PA (9). However, when adequately managed with disease-specific treatment, the mortality ratio seems to be comparable to that in patients with essential hypertension (9). It is therefore of utmost importance to identify patients with PA.

Diagnoses such as myocardial infarction (23), hip fracture (24), and pulmonary embolism (25) can be reliably obtained from the Swedish NPR by using diagnostic codes for identification. For other diseases, for example Cushing disease, the same approach is not suitable because at least one-third of the patients with a diagnostic code for the disease have diagnoses that are not related to hypercortisolism (26). An important question for the current study, and future epidemiological research, was therefore the reliability of a national health care registry such as Swedish NPR for PA. The accuracy of the diagnostic code for PA between 1987 and 1996 (ICD-9) was low (56%). The accuracy after 1997 (ICD-10) was considerably higher (87%), especially for patients who had received the diagnostic code on more than 1 occasion (94%). At the same time, however, by applying the diagnostic code for PA on more than 1 occasion in the ICD-10, 84 of 452 (19%) patients with confirmed PA would not have been captured. Similarly, by using a combination of diagnostic code for PA and adrenal adenoma, 98% of the patients had confirmed PA, but less than 40% would have been identified. Thus, for future epidemiological research on PA in Sweden, including patients who have received the diagnostic code for PA on more than 1 occasion seems to be a reasonable approach with an acceptable accuracy and sensitivity.

Another important issue for the current and future research is the validity of the results. By using another source to identify patients with PA, the coverage in the NPR could be evaluated. Fortunately, the vast majority (97%) of patients with confirmed PA who underwent AVS at our institute since 2006 were registered in the NPR, strongly suggesting a high external validity.

To our knowledge, this is the first study to evaluate the incidence of PA in the general population. The study provides clear evidence that PA is increasingly being detected. Yet, a large number of patients still remain undiagnosed. The increasing detection rate is most probably explained by increasing awareness of PA, mainly among primary care physicians, together with a simplified diagnostic workup in patients with suspected PA where “wash out” of antihypertensive medications affecting aldosterone and/or renin concentrations are no longer considered necessary. Another possible contributing factor is the widespread use of imaging in general, resulting in increased detection of patients with adrenal incidentalomas that are subsequently screened for PA.

A major strength of this study is that the diagnosis in the vast majority of patients who had received a diagnostic code for PA could be validated. In fact, medical records were unavailable for only 4 of 522 (<1% of patients identified in the NPR since 1997). Also, Västra Götaland County has a well‐defined population, making it possible to provide a reliable calculation of incidence rates of PA. However, the study has limitations, such as the retrospective design, and that the true prevalence and incidence of PA cannot be analyzed, only the notified incidence, which is much lower than the true occurrence of the disease (1). Also, because of the retrospective nature of the study, important subjects such as the number of patients with, and the differentiation between, low-renin essential hypertension and idiopathic hyperaldosteronism was not possible (27). Similarly, the majority of the patients in our cohort, especially during the most recent period, were diagnosed with PA based on nonsuppressible aldosterone during saline infusion test. This may underestimate the incidence of PA because not all patients with the disease have angiotensin II–independent autonomous hyperaldosteronism (28).

In conclusion, the detection rate of PA is increasing. Nevertheless, PA is still greatly underdiagnosed and only a small proportion of hypertensive patients are being diagnosed with PA. Given the serious consequences of untreated PA, the noticeable low prevalence at the end of the study clearly illustrates the need for new and optimized strategies aiming at increasing the awareness of PA among health care providers.