Incidence and Clinical Presentation of Pheochromocytoma and Sympathetic Paraganglioma: A Population-based Study

Abstract

Context

Pheochromocytoma and sympathetic paraganglioma (PPGL) are rare catecholamine-secreting tumors but recent studies suggest increasing incidence. Traditionally, PPGL are described to present with paroxysmal symptoms and hypertension, but existing data on clinical presentation of PPGL come from referral centers.

Objective

We aimed to describe time trends in clinical presentation and incidence of PPGL in a population-based study.

Methods

We conducted a nationwide retrospective cohort study of a previously validated cohort of 567 patients diagnosed with PPGL in Denmark 1977-2015. We collected clinical data from medical records of a geographic subcohort of 192 patients. We calculated age-standardized incidence rates (SIRs) and prevalence for the nationwide cohort and descriptive statistics on presentation for the subset with clinical data.

Results

SIRs increased from 1.4 (95% CI 0.2-2.5) per million person-years in 1977 to 6.6 (95% CI 4.4-8.7) per million person-years in 2015, corresponding to a 4.8-fold increase. The increase was mainly due to incidentally found tumors that were less than 4 cm and diagnosed in patients older than 50 years with no or limited paroxysmal symptoms of catecholamine excess. On December 31, 2015, prevalence of PPGL was 64.4 (CI 95% 57.7-71.2) per million inhabitants. Of 192 patients with clinical data, 171 (89.1%) had unilateral pheochromocytoma, while unilateral paraganglioma (n = 13, 6.8%) and multifocal PPGL (n = 8, 4.2%) were rare.

Conclusion

Incidence of PPGL has increased 4.8-fold from 1977 to 2015 due to a “new” group of older patients presenting with smaller incidentally found PPGL tumors and few or no paroxysmal symptoms.

In the “classical” descriptions of pheochromocytoma and sympathetic paraganglioma (PPGL), patients presented with severe hypertension and spectacular symptoms of catecholamine excess, including paroxysmal headache, sweating, and palpitations (1). Today, the diagnostic patterns have changed with the advent of improved biochemical testing and the discovery of PPGL-predisposing mutations leading to improved case detection in affected families. Moreover, the increase in the use of cross-sectional image modalities has led to a virtual tsunami of adrenal incidentalomas, of which less than 1% are diagnosed as pheochromocytomas (2). Since physicians may encounter patients with adrenal incidentalomas in all specialties, updated and population-based real-world data on the epidemiology and clinical presentation of PPGL are necessary to continually improve evidence-based clinical guidelines on how to recognize these rare patients.

Recently, a large Dutch study suggested that the incidence of patients diagnosed with PPGL has increased with approximately 50% from 1996 to 2015 (3). Studies from referral centers have also shown that today a large part of PPGL patients are diagnosed primarily as adrenal incidentalomas, and that these incidentally found PPGL patients present differently from the “classical” PPGL patients in terms of symptoms, tumor size, and biochemical activity (4).

However, the majority of previous studies on the epidemiology and clinical presentation of PPGL may have missed patients diagnosed and treated outside of highly specialized centers, and no studies have provided population-based data on clinical presentation. This potential selection bias raises the concern that current guidelines and textbooks base their descriptions of the clinical presentation of PPGL patients upon data, which might not apply to all PPGL patients seen in today’s everyday clinical care.

Therefore, we aimed to describe time trends in incidence, prevalence, and clinical presentation of all PPGL patients diagnosed in a large population-based study.

Materials and Methods

Study design

We have previously identified and validated a population-based cohort of patients diagnosed with PPGL in Denmark (5). Here we present nationwide data on the incidence and prevalence of PPGL from 1977 to 2015. We also present population-based data on the clinical presentation of PPGL that were collected from medical records for all PPGL patients residing in 2 of the 5 Danish regions at time of diagnosis. We followed the RECORD guideline (The REporting of studies Conducted using Observational Routinely collected health Data, Supplementary 1 (6)) (7). The study was approved by the Danish Data Protection Agency (reference 2014-41-3198). The Danish Health Authorities gave us permission to review medical records without patient consent (reference 3-3013-1021/1).

Setting

Since 1967, all persons living or working in Denmark have had a unique, permanent, and centrally registered social security number. All public and many private registers use this number for identification, which allows linkage of data from numerous different registries on an individual level. Since reporting to public health registries is mandatory for physicians and healthcare is tax paid and universally accessible, coverage of these registries is considered almost complete (8-11). This essentially turns the entire Danish population into a cohort study with near complete follow-up on diseases, treatments, and outcomes since 1977 (12). The Danish population was 5.08 million in 1977 and 5.66 million in 2015 (mean population 1977 to 2015 was 5.29 million).

PPGL patients

We have previously identified and validated a nationwide cohort of all PPGL patients diagnosed in Denmark 1977-2016 (5). In brief, we identified all persons in Denmark registered with a diagnosis code for PPGL in either the Danish National Patient Registry (established 1977), the Danish National Pathology Registry (established 1970 with nationwide coverage in 1997), or the Danish Registry of Causes of Death (established 1970) between January 1, 1977, and December 31, 2016. In 2 of the 5 Danish regions (the north and central regions, mean population 1.75 million), we located medical records to confirm the PPGL diagnosis and collect detailed clinical data. For sympathetic paragangliomas, we excluded patients without any clinical or biochemical signs of catecholamine excess (ie, nonfunctioning paragangliomas, n = 3). Based on this cohort of confirmed PPGL patients, we then developed an algorithm that could identify PPGL patients in the remaining 3 regions (mean population 3.55 million) using only health registry data. The algorithm had a positive predictive value of 93.1% and a completeness of 88.9% (5). The algorithm’s positive predictive value was 95.3% in an external sample (5).

Due to technical problems during the implementation of a new electronic healthcare system, some regions did not report complete data for 2016. Therefore, in this paper we focus on the 567 PPGL patients identified in Denmark from 1977 to 2015, of whom we had detailed clinical data on 192 patients (referred to as the clinical subcohort).

Besides the 567 PPGL patients diagnosed in Denmark from 1977 to 2015, we also by chance identified 7 patients who were diagnosed before 1977, and 2 patients who did not have permanent residence in Denmark at time of diagnosis (5). These patients were excluded from all analyses, except prevalence estimates, as described below.

Clinical data

We collected data for the clinical subcohort on date of diagnosis, symptoms, mode of discovery, tumor size, and other clinical data (n = 192). For the remaining Danish PPGL patients (n = 375), we obtained data on patient demographics from administrative health registries and first registered PPGL diagnosis code from health registries, which is an accurate estimate of date of diagnosis (5).

In the clinical subcohort, we determined how and why PPGL had been diagnosed based on medical records (eg, the reason for referral to biochemical testing or pathological examination). We categorized the reasons for evaluation into 7 overall modes of discovery: (1) paroxysmal symptoms (ie, the classical triad of headache, sweating, and palpitations, 1 or 2 of the 3 classical symptoms, or other paroxysmal symptoms), (2) hypertension (suspected secondary hypertension or pressor response during surgery or pregnancy), (3) adrenal incidentalomas (as defined by guidelines (13)), (4) cancer imaging (cancer staging of active extra-adrenal malignancy or follow-up of cancer in remission), (5) genetic (familial disposition or syndromic presentation), (6) autopsy, and (7) other (including B-symptoms, palpable mass, and other unclassified reasons). Details on criteria and number of patients are found elsewhere (Supplementary 2 (6)).

We considered paroxysmal symptoms and hypertension present if documented in medical records up to 6 months before date of diagnosis and absent if consistently described absent, consistently described as nonparoxysmal, or not mentioned at all. We recorded paroxysmal symptoms and hypertension data as missing if medical records could not be found. We used patient reported date of first symptom to calculate symptom duration. We did not include symptoms or hypertension documented in medical records after confirmation of the PPGL, as the diagnosis could affect the perception of signs and symptoms, rendering collected data less clinically useful.

We defined tumor size as the largest diameter described in radiology reports (of the largest tumor if more than 1 PPGL tumor was found), preferring measurements from computed tomography scans and magnetic resonance imaging over ultrasound and measurements from pathology reports. We categorized tumor size in 4-cm increments (ie, <4, 4-7.9, and ≥8), since 4 cm is often used as a cutoff when distinguishing between small and large adrenal tumors (13).

We defined PPGL tumors as norepinephrine (NE) producing, if norepinephrine or normetanephrine in plasma or urine was above upper normal range; epinephrine (E) producing, if epinephrine or metanephrines was above upper normal range; and NE and E producing, if both were above upper normal range. We preferred plasma metanephrines over other biochemical tests when calculating fold elevation and used mean fold elevation when patients had multiple tests before diagnosis.

Statistical analysis

We calculated incidence rates (IRs) as new patients diagnosed with PPGL in Denmark per million person years with 95% CI based on the Poisson distribution. Following the same approach as previous studies (3), we age-standardized incidence rates (SIRs) to the most recent European Standard Population from 2013 to allow comparing estimates in Denmark over time and with other populations.

Due to the life-long risk of recurrence, we considered all PPGL patients prevalent from date of diagnosis until death or emigration (dates obtained from the Civil Person Registry) (11). We calculated age-standardized prevalence per million persons as the proportion of the Danish population who had previously been diagnosed with PPGL and were still alive and residing in Denmark at December 31 each year. When calculating prevalence, we also included the above-mentioned patients who were diagnosed with PPGL before 1977 (n = 7) or diagnosed in another country before moving to Denmark (n = 2).

For descriptive statistics, we included patients with missing data in figures as a separate missing category but excluded them from tables and provided the reasons for missing data in footnotes. Data on the size of the Danish background population was obtained from Statistics Denmark using a Stata package, dstpop, developed for this paper (https://www.github.com/andreasebbehoj/dstpop). All analyses were conducted in Stata Statistical Software, Release 16.1 (StataCorp, College Station, Texas).

Role of the funding source

None of the funding sources had any role in designing the study, in collecting, analyzing, or interpreting the data, or in writing the manuscript.

Results

Incidence

Of the 567 PPGL patients diagnosed in all of Denmark from 1977 to 2015, 315 (55.6%) were women and median age at diagnosis was 56.2 years (range 8.1-94.0) (Table 1 and Supplementary 3A (6)).

Mean SIR of PPGL in Denmark 1977-2015 was 3.00 (95% CI 2.75-3.25) per million person-years, increasing 4.8-fold from 1.4 (95% CI 0.2-2.5) per million person-years in 1977 to 6.6 (95% CI 4.4-8.7) per million person-years in 2015 (Fig. 1).

During the study period, incidence of PPGL mainly increased in the 50- to 74-year-old and ≥75-year-old population (Fig. 2). Incidence increased similarly for men and women (Supplementary 4 (6)). There were no differences in incidence between regions where the diagnosis of PPGL had been confirmed in medical records and regions where PPGL patients had been identified using the algorithm described in methods (Supplementary 5 (6)).

Prevalence

Prevalence increased throughout the study period without reaching a plateau (Fig. 3). At the end of the study period (December 31, 2015), there were 353 patients with current or previously cured PPGL still alive and residing in Denmark, resulting in an age-standardized prevalence of 64.4 patients (95% CI 57.7-71.2) per million Danish residents (Table 2). Age-standardized prevalence was higher for women (73.4 [95% CI 63.4-83.4] per million women) than men (55.7 [95% CI 46.5-64.9] per million men). Age-specific prevalence was highest in the ≥75-year-old population (140.9 [95% CI 107.5-181.4] per million) and lowest in the <25-year-old population (5.9 [95% CI 2.8-10.9] per million) with some differences between men and women (Table 2).

Mode of discovery

Of the 192 PPGL patients diagnosed in north and central Denmark available for detailed review of medical records, mode of discovery was available for 191 patients. The most common mode of discovery was evaluation for an adrenal incidentaloma (n = 44, 23.0%), followed by evaluation for paroxysmal symptoms (n = 40, 20.9%), and hypertension (n = 34, 17.8%) (Table 1). Patient age at diagnosis varied with mode of discovery. For example, median age was 34.3 years (range 11.7-74.7) in patients diagnosed due to genetic causes and 63.0 years (range 42.2-81.9) in patients diagnosed due to cancer imaging (Supplementary 3B and 6 (6)).

In 1977-1986, the most common modes of discovery were hypertension (n = 13, 31.7%), autopsy (n = 10, 24.4%), and paroxysmal symptoms (n = 8, 19.5%) (Table 1). In 2007-2015, the most common modes of discovery changed to adrenal incidentalomas (n = 29, 36.7%), cancer-related imaging (n = 15, 19.0%), and paroxysmal symptoms (n = 12, 15.2%), while none was diagnosed at autopsy. The number of patients diagnosed due to work-up for familial disposition or syndromic presentation increased from 3 (7.3%) patients in 1977-1986 to 10 (12.7%) in 2007-2015 (Table 1). In absolute numbers, SIR for patients diagnosed due to paroxysmal symptoms or hypertension combined was constant throughout the study period, while patients diagnosed due to adrenal incidentalomas, cancer imaging, and genetic disposition accounted for the majority of the increase in SIR (Fig. 4A).

Symptoms at presentation

The majority of patients had documented paroxysmal symptoms suggestive of catecholamine excess before diagnosis. Of 189 PPGL patients with available medical records describing symptoms, 40 (21.2%) patients had the classical triad of headache, sweating, and palpitations, 64 (33.9%) had 1 or 2 of the classical symptoms, 24 (12.7%) had other paroxysmal symptoms, and 61 (32.3%) had no documented paroxysmal symptoms (Table 1). Patients with paroxysmal symptoms and available data on symptom duration (n = 125) reported that symptoms had begun for a median of 1.7 years (range 0.0-23.5) before diagnosis and 33 (26.4%) patients reported 5 years of symptoms or more (Table 1). Symptoms varied with mode of discovery, and approximately half of the patients diagnosed due to adrenal incidentalomas, cancer imaging, or genetic disposition had no paroxysmal symptoms (Supplementary 6 (6)). Out of 189 patients with available records, 76 (40.2%) had documented paroxysmal hypertension (Table 1).

Towards the end of study period, a decreasing proportion (but relatively stable number) of patients presented with the classical triad or 1 or 2 of the classical symptoms, while the proportion and number of patients without paroxysmal symptoms increased (Table 1 and Fig. 4B). Median symptom duration and hypertension did not show any obvious patterns of variation during the study period (Table 1).

Tumor size

Median tumor diameter was 4.5 cm (range 1.0-30.0) and varied during the study period, without an obvious pattern (Table 1). Tumors were smaller in patients diagnosed due to genetic disposition compared to most other patient groups (Supplementary 6 (6)). During the study period, the proportion of patients with PPGL tumors <4 cm increased; the proportion of 4- to 7.9-cm tumors varied; and proportion of tumors ≥8 cm decreased (Table 1). In absolute numbers, SIR of <4 cm and 4- to 7.9-cm tumors increased, while SIR of ≥8 cm tumors was stable (Fig. 4C).

Tumor location, genetics, preoperative diagnosis, and other clinical findings

The majority of patients presented with unilateral pheochromocytoma (n = 171, 89.1%), while unilateral paraganglioma (n = 13, 6.8%) and multifocal PPGL (n = 8, 4.2%) were rare (Table 1). Five patients (2.6%) were later diagnosed with new primary PPGL tumors (data not shown). Five (2.6%) of the 192 patients had evidence of metastatic PPGL at diagnosis and 9 (4.7%) patients later developed metastatic PPGL (data not shown).

The median increase in catecholamines was 6.1 fold (range 1.1-65.0) above upper normal laboratory range (Table 1). Tumors diagnosed in patients due to genetic disposition were less biochemically active than other patient groups (Supplementary 6 (6)).

Thirty-four (17.9%) patients had a genetically or clinically confirmed PPGL-disposing mutation or syndrome, 67 (35.3%) had at least 1 negative genetic test, and 89 (46.8%) had never been tested (Table 1). Eighty-two patients were tested for mutations in the RET-gene, while only 30 were tested for mutations in any of the succinate dehydrogenase subunit (SDH) genes. The proportion of patients undergoing genetic testing increased during the study period (Table 1) and varied with mode of discovery (Supplementary 6 (6)).

Hereditary PPGL syndromes and mutations included multiple endocrine neoplasia 2A (n = 14) and 2B (n = 2), neurofibromatosis type 1 (n = 7), von Hippel-Lindau (n = 2), mutations in SDHB (n = 5) and SDHC (n = 1), a likely pathogenic CDKN2B variant (n = 1), and familial PPGL of unknown genetic cause (n = 2).

Out of 167 patients diagnosed alive with available data, 28 (16.8%) patients had undergone an invasive procedure on the PPGL tumor before PPGL had been diagnosed and thus without alpha-blockade. Procedures included fine-needle aspiration or biopsy (13 patients), surgical resection of PPGL (9 patients), or both aspiration/biopsy and subsequent resection (6 patients). Reasons for these procedures are described in Supplementary 2 (6). Ten (6.0%) of 167 patients were never operated for various reasons (Supplementary 6 (6)). Of the 157 patients who underwent surgery, 142 (90.4%) patients were diagnosed with PPGL and started on alpha-blockade before surgery, while 15 (9.6%) were diagnosed after surgery. Perioperative mortality (within 30 days of surgery) was 1.4% (2 of 142 patients) in the first group and 6.7% (1 of 15 patients) in the latter (data not shown). The proportion of patients operated without a preoperative diagnosis decreased from 13.3% (4 out of 30) in 1977-1986 to 5.5% (4 out of 73) in 2007-2015 (data not shown).

Discussion

We present a nationwide study of the epidemiology of PPGL (567 patients) and large population-based study of the clinical presentation of PPGL (192 patients). This is the first study to present population-based data on the clinical presentation of PPGL and the second-largest epidemiological study using a validated PPGL cohort. First, we show that the increasing incidence of PPGL is due to incidentally found tumors. Secondly, we find that this new group of incidentally found patients has changed the clinical presentation of PPGL to older patients with smaller tumors and fewer “classic” symptoms. Finally, we find that the clinical presentation of PPGL differs in our population-based study compared to what has previously been reported from specialized referral centers.

Up until 2018, only 4 nationwide studies of the incidence of PPGL had been published (3). These studies have reported an increase in incidence rate from 1.9 to 2.1 per million person-years in the 1970s and 1980s to 5.7 per million person-years in 2011-2015 (3). Our data are in in concordance with these previous findings. In 2020, the first non-European nationwide study was published, with 1048 patients diagnosed 2003-2014 in Korea, corresponding to a relatively low age-standardized incidence rate of 1.8 per million person-years (14). This study used a rather strict and unvalidated algorithm to identify patients in a single health registry, thus the findings could be explained by a low completeness (ie, sensitivity) of the algorithm.

While the increasing incidence of PPGL has been suspected for many years, and was confirmed by Berends et al. in 2018 (3), the explanation for this increase has been more controversial. Given that tumor size has decreased over time, as observed by Berends et al. and in our study, it has been speculated that earlier diagnosis of PPGL due to incidental imaging findings might explain some of the increase in incidence (3). However, the fact that patients are getting increasingly older at time of diagnosis, is a strong argument against the notion, that earlier diagnosis of the same patients should have any important role in the increasing incidence. In a recent study from Mayo Clinic, a large, well-known US referral center for PPGL patients, Gruber et al. (4) reported that the proportion of incidentally found PPGL tumors had increased from 31.4% (44 of 140 tumors) in 1995-2004 to 59.5% (176 of 296 tumors) in 2005-2016 (4). This result is similar to what we found in our study in 2007-2015 (36.7% patients were diagnosed due to adrenal incidentalomas and 19.0% due to cancer staging or follow-up, Table 1). In conclusion, we find that the increasing incidence of PPGL is best explained by the incidental detection of a new group of “nonclassical” PPGL patients, who were likely missed in the era before the widespread availability of cross-sectional imaging.

Regarding clinical presentation, we observed several interesting differences between our population-based study and other more selected PPGL cohorts from referral centers. First, patients were older in our study (median age 56.2 years in 1977-2015 with median age 59.9 years in 2006-2015, Table 1) than in the single-center study by Gruber et al. (median age 52.0 years) (4) and older than in the nationwide Dutch study by Berends et al. (median age approx. 53 years in 2015) (3). Secondly, 9.6% (15 of 157 patients) of patients who underwent surgical resection of PPGL were not diagnosed with PPGL before surgery. This is a much higher proportion than other studies, which suggests that these patients may be missed in studies from referral centers, though it may also reflect a time effect, as we found a clear preponderance of nonpreoperatively diagnosed patients early in our study period. One (6.7%) out of these 15 patients died perioperatively. While it is difficult to conclude on a such a small sample, the perioperative mortality is very different from a 1979 review (15), reporting that 16 out of 19 PPGL patients operated for an unsuspected PPGL died during surgery, but more in line with a more recent review reporting an 8% (5 of 60 patients) perioperative mortality (16). Nevertheless, it is worth emphasizing that surgical resection of PPGL without alpha-blockade is potentially dangerous and invasive procedures involving adrenal tumors are discouraged unless PPGL has been biochemically ruled out (17). Thirdly, only 17.9% (34 of 190 patients, Table 1) in our study had been diagnosed with a PPGL-predisposing mutation or syndrome. While this could very well be explained by a low number of patients undergoing a complete genetic workup in the early part of our study period, it is still considerably lower than the 30% to 40% hereditary PPGL patients reported in multicenter studies (18-20), and will need to be investigated in the future. Fourthly, in our study only 6.8% had unilateral sympathetic paraganglioma compared with 18.3% (3), 21.6% (among adults) (21), and 25% (22) previously reported. Multifocal PPGL (including bilateral pheochromocytoma) and metastatic PPGL were also rare in our study. This again indicates that presentation may differ between referral centers and in the general population of PPGL patients, however the results could also be subject to diagnostic changes during the study period, which our patient cohort was too small to identify. Finally, approximately half of the patients diagnosed due to incidentalomas, cancer-related imaging, or genetic disposition had paroxysmal symptoms in our study (Supplementary 6 (6)). This is considerably less than the 63.6% to 83.5% symptomatic patients reported by Gruber et al. (4) for the same modes of discovery. However, different health care systems and study protocols might make comparisons difficult. Irrespective of the exact numbers, our results highlights that absence of paroxysmal symptoms does not rule out PPGL (13, 23).

Strengths and limitations

The main strengths of this study are its population-based design and large validated cohort of PPGL, which effectively nullifies referral bias, allows for precise epidemiological estimates, and ensures that data on clinical presentation are generalizable outside of specialized centers. The main limitation of the study is its reliance on routine care databases, where only patients diagnosed and registered with PPGL are identified, thus potentially underestimating the true incidence and prevalence of PPGL. Moreover, the retrospective data collection from health records caused missing data, and the documented symptoms and other clinical variables might be affected by changes over time in clinical awareness, guidelines and documentation practices (eg, electronic patient records). Finally, the Danish population is mainly Caucasian, which may limit generalizability to more diverse populations.

Conclusions

In conclusion, we found that the incidence of PPGL has increased almost fivefold from 1977 to 2015, and that the “new” PPGL patients in recent years primarily are diagnosed as adrenal incidentalomas, present at an older age, and have smaller tumors and fewer symptoms. We also found that clinical presentation of PPGL is different in a population-based cohort compared with studies from referral centers. The high proportion of asymptomatic PPGL patients in our study highlights that the absence of paroxysmal symptoms in no way excludes a catecholamine-secreting PPGL, and that adrenal incidentalomas are changing the epidemiology and clinical presentation of PPGL.

Abbreviations

Abbreviations
 
• E

  epinephrine

 
• NE

  norepinephrine

 
• SIR

  standardized incidence rate

 
• PPGL

  pheochromocytoma and sympathetic paraganglioma

 
• SDH

  succinate dehydrogenase subunit.

Acknowledgments

We thank the 82 physicians and other health care professionals from all of Denmark, who assisted in locating and collecting data from patient medical records.

Financial Support: This paper was funded by a pre-graduate research-year scholarship (A.E.) and a PhD scholarship (A.E.) from Department of Clinical Medicine, Aarhus University, with additional funding from the Department of Endocrinology and Internal Medicine, Aarhus University Hospital (E.S.), Fonden til Lægevidenskabens Fremme (C.T., grant number 15-407), Aase og Ejnar Danielsens Fond (P.L.P., grant number 10-001601), and Bagermester August Jensen og Hustrus legat (E.S.).

Author Contributions: Project idea and protocol was by E.S. and P.L.P. Data extraction from registries, majority of data collection, analyses, and first draft was done by A.E. with technical assistance from and under supervision of K.S., C.T., R.W.T., P.J., E.S., and P.L.P. S.F.J. performed data collection for validating the algorithm in the Capital Region of Denmark under supervision of Å.K.R. and U.F.R. M.G.R. assisted in collecting data in the Northern Danish Region. All authors contributed to the writing and approval of the final manuscript.

Additional Information

Disclosures: The authors have nothing to disclose. The authors declare no conflicts of interest.

Data Availability

Supplementary figures and tables are available on https://doi.org/10.6084/m9.figshare.13325405.v1. The Stata code for the analysis is available on https://github.com/andreasebbehoj/2020-PPGL-incidence-prevalence-paper, including history of all changes and the log file. In order to preserve patient confidentiality and uphold Danish law on data protection, restrictions apply to the availability and sharing of patient data. Upon request, the corresponding author will detail these restrictions and under which conditions access to some of the data may be provided.

References

Author notes