https://orcid.org/0000-0003-4392-269X
Abstract
Mortality studies have established that cardiovascular disease is the leading cause of death in patients with adrenal insufficiency and the risk is greater than that observed in individually matched controls.
Here we have performed a detailed analysis of cardiovascular morbidity and mortality, taking account of the role of comorbidities.
We performed a retrospective cohort study using the Clinical Practice Research Datalink (CPRD), a UK general practitioner database. The participant population comprised 6821 patients with adrenal insufficiency (primary, 2052; secondary, 3948) compared with 67 564 individually matched controls, with and without adjustment for comorbidities (diabetes, hypertension, dyslipidemia, previous cardiovascular disease, and smoking). The main outcome measures were composite cardiovascular events recorded in the CPRD and cardiovascular mortality in participants with linked national mortality data.
Hazard ratios (95% CI) for composite cardiovascular events in patients with adrenal insufficiency of any cause were 1.28 (1.20-1.36, unadjusted) and 1.07 (1.01-1.14, adjusted). Increased cerebrovascular events in patients with secondary adrenal insufficiency accounted for most of the increased hazard (1.53 [1.34-1.74, adjusted]) and were associated with cranial irradiation therapy. Cardiovascular mortality data were available for 3547 patients and 34 944 controls. The adjusted hazard ratio for ischemic heart disease mortality was 1.86 (1.25-2.78) for primary adrenal insufficiency and 1.39 (1.02-1.89) for secondary.
Comorbidities largely accounted for the increased cardiovascular events but in secondary adrenal insufficiency, cerebrovascular events were independently increased and associated with irradiation treatment. However, the risk of cardiovascular mortality remained increased even following adjustment for comorbidities in both primary and secondary adrenal insufficiency.
Adrenal insufficiency is a rare disorder in which the adrenal glands produce insufficient glucocorticoids. In primary adrenal insufficiency (Addison disease) the defect resides in the adrenal glands themselves. Secondary adrenal insufficiency is caused by pituitary or hypothalamic disease and is usually accompanied by other pituitary hormone deficiency. The mainstay of treatment is glucocorticoid replacement therapy. The treatment dose is titrated individually, and inappropriately high glucocorticoid doses have been associated with increased mortality (1-4), mainly from cardiovascular disease (5-15).
Cardiovascular disease, particularly ischemic heart disease, appears to be the leading cause of death in primary adrenal insufficiency (9, 11, 16), although not always disproportionately relative to the general population (16). However, previous studies were conducted over a period during which care for cardiovascular disease and for adrenal insufficiency were improving, especially in relation to glucocorticoid replacement doses. A recent study in autoimmune Addison disease found ischemic heart disease incidence still increased in women, although there was no increase in cerebrovascular disease (17).
Increased cardiovascular mortality has been reported in patients with pituitary disorders (5-8, 12, 13, 15), from both ischemic heart disease and cerebrovascular disease, but data on cardiovascular morbidity are limited (18, 19). Much of the focus of large hypothalamic-pituitary studies has been on the role of growth hormone deficiency (12, 13, 18, 19) and cardiovascular risk may have complex origins. One study showed increased cardiovascular risk factors, specifically hyperlipidemia, diabetes, and hypertension, irrespective of whether the adrenal insufficiency was primary or secondary (20). However, matched population-based data on cardiovascular outcomes are very limited and it is uncertain if there are differences between primary and secondary disease.
We therefore performed a population-based matched cohort study of cardiovascular events and mortality in primary and secondary adrenal insufficiency, taking account of cardiovascular comorbidities. Ischemic heart disease and cerebrovascular disease were examined separately, and the roles of some cardiovascular risk factors were assessed.
Methods
Study Design and Dataset
This retrospective cohort study used longitudinal patient data from a UK general practitioner database, the Clinical Practice Research Datalink (CPRD) (21). The dataset analyzed in this study (CPRD GOLD database), which was extracted in April 2018, contains information for 15 354 125 deceased and living patients from 734 practices. Data are encoded as medical codes (Read codes) for diagnoses and product codes for prescribed medications.
CPRD has provided a linkage scheme to expand health data to secondary care—58% of practices in CPRD have participated in the linkage scheme (21). From 1997, information for people registered in participating practices can be linked with the underlying cause of death provided by the Office for National Statistics (ONS), and inpatient diagnoses provided by Hospital Episode Statistics (HES). Diagnostic codes in HES data have been formatted using the International Classification of Diseases, 10th revision (ICD-10) and for ONS data using the International Classification of Diseases, 9th revision (ICD-9) up to 2001 and ICD-10 from 2001. The study protocol was approved by the Independent Scientific Advisory Committee for MHRA Database Research (protocol number 18_179; Appendix) (22).
Study Patients, Controls, and Study Period
We extracted records for study patients aged <100 years from the CPRD dataset, recorded from 1987 to 2017, using medical codes for adrenal insufficiency and product codes for glucocorticoid prescription. Medical codes were organized into 3 groups: (1) primary; (2) secondary; and (3) unknown type (Supplementary Appendix A) (22). Patients of “unknown type” were classified as unspecified adrenal insufficiency unless they were later assigned codes for primary or secondary adrenal insufficiency.
We created a list of product codes for glucocorticoids (Supplementary Appendix B) (22). These were combined with the medical codes and then used to extract study patients from the CPRD dataset. All extracted patients were required to have both medical codes and product codes. For inclusion, the product codes had to have been recorded within 3 months after the first record of the medical codes to ensure that glucocorticoid use was related to treating adrenal insufficiency. We excluded patients having, at any time, medical codes for acromegaly, Cushing disease, Cushing syndrome, congenital adrenal hyperplasia, and malignant neoplasm of the adrenal or pituitary glands (Supplementary Appendix C) (22), as these patients might have had increased risk of cardiovascular disease or mortality, independent of adrenal insufficiency per se. We also excluded those having a valid follow-up period of <1 month. The number of qualifying study patients was 6825 (Supplementary Figure 1A) (22).
Controls were selected from participants in the same CPRD dataset. We extracted up to 10 controls for each patient by matching sex, 5-year strata of year of birth, general practitioner practice, and 5-year strata of year at start of follow-up. We could not find any matched controls for 4 patients, who were excluded. The number of matched patients and controls was 6821 and 67 564. Start of follow-up was defined by the most recent date at which: (1) participants were registered to the practice; (2) adrenal insufficiency was first recorded (study patients only); or (3) CPRD considered that the practice provided research-standard information. Follow-up finished at death, exit from the practice, exit from CPRD, or the end of 2017, whichever occurred first. (Supplementary Figure 1B) (22).
Primary and Secondary Outcomes
Primary outcomes were the first occurrence of fatal or nonfatal cardiovascular events recorded in CPRD, including: (1) composite cardiovascular disease; (2) ischemic heart disease; and (3) cerebrovascular disease. Composite cardiovascular disease included ischemic heart disease, congestive cardiac failure aortic dissecting/ aneurysm, atrial fibrillation, cerebrovascular disease, transient ischemic attack, and peripheral arterial disease. Medical codes related to these conditions were listed and also used to define previous cardiovascular disease (Supplementary Appendix D) (22). Our interest is primarily with atherosclerotic vascular disease; therefore, deep venous thrombosis and pulmonary embolism were not included in the primary outcome.
Secondary outcomes, evaluated using ONS and HES data linked with CPRD, were cardiovascular mortality and hospitalization due to cardiovascular disease. “Death from cardiovascular disease” was assigned. ONS ICD-10 or ICD-9 codes specified the main cause of death as (1) disease of the circulatory system; (2) ischemic heart disease; and (3) cerebrovascular disease or transient cerebral ischemic attack (Supplementary Appendix E) (22). Hospitalization due to cardiovascular disease was defined by the primary diagnosis of the first hospitalization from (1) circulatory system diseases; (2) ischemic heart disease; and (3) cerebrovascular disease or transient cerebral ischemic attack using ICD10 codes (Appendix E) (22).
Statistical Analysis
Primary and secondary analyses
We used univariable and multivariable Cox models to analyze primary and secondary outcomes provided by CPRD in the full cohort and in the linked subset, respectively. Models were adjusted for the potential confounders: (1) previous cardiovascular disease; (2) diabetes; (3) hypertension; (4) use of lipid-lowering agents at baseline; and (5) smoking at any time. Incidence rates for cardiovascular events and mortality were calculated per thousand person-years.
Subanalyses
Multivariable Cox regression models and interaction tests explored effects on cardiovascular events and mortality of adrenal insufficiency by sex and age at the start of follow-up for each category. Cardiovascular events and mortality in primary and secondary adrenal insufficiency were directly compared using univariable and multivariable Cox regression analyses. In patients with secondary adrenal insufficiency, the association between radiotherapy and cerebrovascular disease was examined using logistic regression analysis.
Sensitivity analyses (validating cardiovascular disease outcomes in CPRD)
Cardiovascular outcomes commonly diagnosed in the hospital setting might have been misdiagnosed in the general practitioner setting. We, therefore, examined the agreement between HES and CPRD datasets, with HES records as reference, specifically for ischemic heart disease or cerebrovascular disease (including transient cerebral ischemic attack) as the primary diagnosis, underlying disease, or disease developed during the admission. False positive and false negative rates in the CPRD records were calculated.
We previously validated the diagnosis of adrenal insufficiency recorded in CPRD using the linked HES data (23).
Results
A total of 6821 patients with adrenal insufficiency (2052 primary; 3948 secondary) matched with 67 564 controls (20 366 matched for primary and 39 134 for secondary) were included in cardiovascular event analysis. Proportions of males and smoking, ages at start, and follow-up periods were comparable between patients and controls. Patients had higher proportions of baseline cardiovascular disease, diabetes, hypertension, dyslipidemia, and use of statins. Age at diagnosis was on average ~1 year before start of follow-up (Table 1). Participants with linkage to HES and ONS data had comparable baseline characteristics to the full study population (Supplementary Table 1) (22).
Baseline Characteristics of Adrenal Insufficiency Patients and Controls
| All adrenal insufficiency | Primary adrenal insufficiency | Secondary adrenal insufficiency | ||||
|---|---|---|---|---|---|---|
| Patients n = 6821 | Controls n = 67 564 | Patients n = 2052 | Controls n = 20 366 | Patients n = 3948 | Controls n = 39 134 | |
| Male; n (%) | 3173 (46.5) | 31 283 (46.3) | 860 (41.9) | 8483 (41.7) | 1971 (49.9) | 19 474 (49.8) |
| Age at start of follow-up; years, median (IQR) | 53 (38-68) | 53 (37-68) | 51 (36.5-67) | 51 (36-67) | 54 (38-68) | 53 (38-67) |
| Age at diagnosis; years, median (IQR) | 52 (36-67) | ·· | 50 (35-60) | ·· | 53 (37-67) | ·· |
| Follow-up period; years, median (IQR) | 4.3 (1.7-8.8) | 4.0 (1.6-9.0) | 4.6 (1.7-9.6) | 4.3 (1.7-10.0) | 4.4 (1.8-8.9) | 4.0 (1.6-8.9) |
| Previous cardiovascular disease; n (%)a | 1190 (17.5) | 7586 (11.2) | 318 (15.5) | 2055 (10.1) | 680 (17.2) | 4438 (11.3) |
| Diabetes; n (%)a | 712 (10.4) | 3217 (4.8) | 260 (12.7) | 841 (4.1) | 358 (9.1) | 1942 (5.0) |
| Type 1 diabetes; n (%)a | 222 (3.3) | 376 (0.6) | 153 (7.5) | 99 (0.5) | 49 (1.2) | 235 (0.6) |
| Hypertension; n (%)a | 1508 (22.1) | 9191 (13.6) | 349 (17.0) | 2625 (12.9) | 938 (23.8) | 5327 (13.6) |
| Dyslipidemia; n (%)a | 1397 (20.5) | 3407 (5.0) | 348 (17.0) | 960 (4.7) | 858 (21.7) | 1998 (5.1) |
| Statin use; n (%)a | 1339 (19.6) | 3305 (4.9) | 330 (16.1) | 928 (4.6) | 828 (21.0) | 1941 (5.0) |
| Smoking at any time; n (%) | 2928 (42.9) | 29 114 (43.1) | 832 (40.6) | 8660 (42.5) | 1721 (43.6) | 17 170 (43.9) |
| All adrenal insufficiency | Primary adrenal insufficiency | Secondary adrenal insufficiency | ||||
|---|---|---|---|---|---|---|
| Patients n = 6821 | Controls n = 67 564 | Patients n = 2052 | Controls n = 20 366 | Patients n = 3948 | Controls n = 39 134 | |
| Male; n (%) | 3173 (46.5) | 31 283 (46.3) | 860 (41.9) | 8483 (41.7) | 1971 (49.9) | 19 474 (49.8) |
| Age at start of follow-up; years, median (IQR) | 53 (38-68) | 53 (37-68) | 51 (36.5-67) | 51 (36-67) | 54 (38-68) | 53 (38-67) |
| Age at diagnosis; years, median (IQR) | 52 (36-67) | ·· | 50 (35-60) | ·· | 53 (37-67) | ·· |
| Follow-up period; years, median (IQR) | 4.3 (1.7-8.8) | 4.0 (1.6-9.0) | 4.6 (1.7-9.6) | 4.3 (1.7-10.0) | 4.4 (1.8-8.9) | 4.0 (1.6-8.9) |
| Previous cardiovascular disease; n (%)a | 1190 (17.5) | 7586 (11.2) | 318 (15.5) | 2055 (10.1) | 680 (17.2) | 4438 (11.3) |
| Diabetes; n (%)a | 712 (10.4) | 3217 (4.8) | 260 (12.7) | 841 (4.1) | 358 (9.1) | 1942 (5.0) |
| Type 1 diabetes; n (%)a | 222 (3.3) | 376 (0.6) | 153 (7.5) | 99 (0.5) | 49 (1.2) | 235 (0.6) |
| Hypertension; n (%)a | 1508 (22.1) | 9191 (13.6) | 349 (17.0) | 2625 (12.9) | 938 (23.8) | 5327 (13.6) |
| Dyslipidemia; n (%)a | 1397 (20.5) | 3407 (5.0) | 348 (17.0) | 960 (4.7) | 858 (21.7) | 1998 (5.1) |
| Statin use; n (%)a | 1339 (19.6) | 3305 (4.9) | 330 (16.1) | 928 (4.6) | 828 (21.0) | 1941 (5.0) |
| Smoking at any time; n (%) | 2928 (42.9) | 29 114 (43.1) | 832 (40.6) | 8660 (42.5) | 1721 (43.6) | 17 170 (43.9) |
Abbreviation: IQR, interquartile range.
a Data are from the start of follow-up.
Baseline Characteristics of Adrenal Insufficiency Patients and Controls
| All adrenal insufficiency | Primary adrenal insufficiency | Secondary adrenal insufficiency | ||||
|---|---|---|---|---|---|---|
| Patients n = 6821 | Controls n = 67 564 | Patients n = 2052 | Controls n = 20 366 | Patients n = 3948 | Controls n = 39 134 | |
| Male; n (%) | 3173 (46.5) | 31 283 (46.3) | 860 (41.9) | 8483 (41.7) | 1971 (49.9) | 19 474 (49.8) |
| Age at start of follow-up; years, median (IQR) | 53 (38-68) | 53 (37-68) | 51 (36.5-67) | 51 (36-67) | 54 (38-68) | 53 (38-67) |
| Age at diagnosis; years, median (IQR) | 52 (36-67) | ·· | 50 (35-60) | ·· | 53 (37-67) | ·· |
| Follow-up period; years, median (IQR) | 4.3 (1.7-8.8) | 4.0 (1.6-9.0) | 4.6 (1.7-9.6) | 4.3 (1.7-10.0) | 4.4 (1.8-8.9) | 4.0 (1.6-8.9) |
| Previous cardiovascular disease; n (%)a | 1190 (17.5) | 7586 (11.2) | 318 (15.5) | 2055 (10.1) | 680 (17.2) | 4438 (11.3) |
| Diabetes; n (%)a | 712 (10.4) | 3217 (4.8) | 260 (12.7) | 841 (4.1) | 358 (9.1) | 1942 (5.0) |
| Type 1 diabetes; n (%)a | 222 (3.3) | 376 (0.6) | 153 (7.5) | 99 (0.5) | 49 (1.2) | 235 (0.6) |
| Hypertension; n (%)a | 1508 (22.1) | 9191 (13.6) | 349 (17.0) | 2625 (12.9) | 938 (23.8) | 5327 (13.6) |
| Dyslipidemia; n (%)a | 1397 (20.5) | 3407 (5.0) | 348 (17.0) | 960 (4.7) | 858 (21.7) | 1998 (5.1) |
| Statin use; n (%)a | 1339 (19.6) | 3305 (4.9) | 330 (16.1) | 928 (4.6) | 828 (21.0) | 1941 (5.0) |
| Smoking at any time; n (%) | 2928 (42.9) | 29 114 (43.1) | 832 (40.6) | 8660 (42.5) | 1721 (43.6) | 17 170 (43.9) |
| All adrenal insufficiency | Primary adrenal insufficiency | Secondary adrenal insufficiency | ||||
|---|---|---|---|---|---|---|
| Patients n = 6821 | Controls n = 67 564 | Patients n = 2052 | Controls n = 20 366 | Patients n = 3948 | Controls n = 39 134 | |
| Male; n (%) | 3173 (46.5) | 31 283 (46.3) | 860 (41.9) | 8483 (41.7) | 1971 (49.9) | 19 474 (49.8) |
| Age at start of follow-up; years, median (IQR) | 53 (38-68) | 53 (37-68) | 51 (36.5-67) | 51 (36-67) | 54 (38-68) | 53 (38-67) |
| Age at diagnosis; years, median (IQR) | 52 (36-67) | ·· | 50 (35-60) | ·· | 53 (37-67) | ·· |
| Follow-up period; years, median (IQR) | 4.3 (1.7-8.8) | 4.0 (1.6-9.0) | 4.6 (1.7-9.6) | 4.3 (1.7-10.0) | 4.4 (1.8-8.9) | 4.0 (1.6-8.9) |
| Previous cardiovascular disease; n (%)a | 1190 (17.5) | 7586 (11.2) | 318 (15.5) | 2055 (10.1) | 680 (17.2) | 4438 (11.3) |
| Diabetes; n (%)a | 712 (10.4) | 3217 (4.8) | 260 (12.7) | 841 (4.1) | 358 (9.1) | 1942 (5.0) |
| Type 1 diabetes; n (%)a | 222 (3.3) | 376 (0.6) | 153 (7.5) | 99 (0.5) | 49 (1.2) | 235 (0.6) |
| Hypertension; n (%)a | 1508 (22.1) | 9191 (13.6) | 349 (17.0) | 2625 (12.9) | 938 (23.8) | 5327 (13.6) |
| Dyslipidemia; n (%)a | 1397 (20.5) | 3407 (5.0) | 348 (17.0) | 960 (4.7) | 858 (21.7) | 1998 (5.1) |
| Statin use; n (%)a | 1339 (19.6) | 3305 (4.9) | 330 (16.1) | 928 (4.6) | 828 (21.0) | 1941 (5.0) |
| Smoking at any time; n (%) | 2928 (42.9) | 29 114 (43.1) | 832 (40.6) | 8660 (42.5) | 1721 (43.6) | 17 170 (43.9) |
Abbreviation: IQR, interquartile range.
a Data are from the start of follow-up.
Cardiovascular Events
The incidence rates of composite cardiovascular events in the patients with adrenal insufficiency (primary or secondary) vs their matched controls were 31.4 (95% CI, 29.6-33.3) vs 24.4 (95% CI, 23.9-24.9; P < 0.0001) per 1000 person-years with follow-up periods of 36 492 and 366 711 person-years, respectively (unadjusted hazard ratio [HR], 1.28 [95% CI, 1.20-1.36]; Figure 1; Supplementary Table 2) (22). After adjustment for previous cardiovascular disease, diabetes, hypertension, dyslipidemia, and ever having smoked, the HR was lower to a marginally significant level (adjusted HR, 1.07 [95% CI, 1.01-1.14]). Likewise, adjusted HRs of composite cardiovascular events in primary and secondary disease were 1.08 (95% CI, 0.96-1.22) and 1.10 (95% CI, 1.01-1.19; Figure 1), respectively.
Cardiovascular events of patients with adrenal insufficiency relative to matched controls.
The incidence rates of ischemic heart disease in patients with adrenal insufficiency of any cause vs controls were 12.3 (95% CI, 11.2-13.4) vs 10.5 (95% CI, 10.2-10.8; P = 0.0010) per 1000 person-years (unadjusted HR, 1.16 [95% CI, 1.06-1.28]; adjusted HR, 0.95 [95% CI, 0.86-1.04]; Figure 1; Supplementary Table 2) (22). Examining primary and secondary disease separately, adjusted HRs of ischemic heart disease were 1.00 (95% CI, 0.83-1.19) and 0.91 (95% CI, 0.80-1.04; Figure 1), respectively.
The incidence rates of cerebrovascular disease in the patient group overall and controls were 10.4 (95% CI, 9.5-11.5) vs 7.2 (95% CI, 7.0-7.5; P < 0.0001) per 1000 person-years (unadjusted HR, 1.44 [95% CI, 1.30-1.60]; adjusted HR, 1.27 [95% CI, 1.15-1.42]; Figure 1; Supplementary Table 2) (22). In primary and secondary disease, adjusted HRs of cerebrovascular disease were 1.00 (95% CI, 0.80-1.25) and 1.53 (95% CI, 1.34-1.74; Figure 1), respectively.
Cardiovascular Mortality
A total of 3547 patients with adrenal insufficiency and 34 944 controls were included in the cardiovascular mortality analysis. Mortality rates from circulatory system diseases for all patients and controls were 9.9 (95% CI, 8.6-11.4) and 6.4 (95% CI, 6.1-6.8) per 1000 person-years, respectively (P < 0.0001; Supplementary Table 2) (22). The adjusted HR for all patients was 1.31 (95% CI, 1.12-1.54); for primary adrenal insufficiency, 1.58 (95% CI, 1.19-2.10); and for secondary adrenal insufficiency, 1.23 (95% CI, 0.99-1.52; Figure 2).
Cardiovascular mortality of patients with adrenal insufficiency relative to matched controls.
Mortality rates from ischemic heart disease for all patients and controls were 4.9 (95% CI, 4.0-6.1) and 2.7 (95% CI, 2.5-3.0) per 1000 person-years, respectively (P < 0.0001; Supplementary Table 2) (22). The adjusted HR for all patients was 1.49 (95% CI, 1.18-1.88); for primary, 1.86 (95% CI, 1.25-2.78); and for secondary adrenal insufficiency, 1.39 (95% CI, 1.02-1.89; Figure 2).
Mortality rates from cerebrovascular disease for all patients and controls were 2.5 (95% CI, 1.9-3.3) and 2.0 (95% CI, 1.8-2.3) per 1000 person-years (P = 0.10; Supplementary Table 2) (22), respectively. For primary and secondary disease separately, mortality rates were 1.8 (95% CI, 1.0-3.4) and 3.0 (95% CI, 2.1-4.2) per 1000 person-years, respectively (Supplementary Table 2) (22). The equivalent adjusted HRs were 1.03 (95% CI, 0.75-1.41), 0.85 (95% CI, 0.44-1.64), and 1.14 (95% CI, 0.78-1.67; Figure 2), respectively.
Hospitalization Due to Cardiovascular Disease
A total of 2674 (75%) patients and 15 962 (46%) controls were admitted during the follow-up period. The incidence rates for the first admission due to circulatory system diseases for patients and controls were 30.9 (95% CI, 28.3-33.6) vs 19.0 (95% CI, 18.4-19.7; P < 0.0001; Supplementary Table 2) (22) per 1000 person-years, respectively (unadjusted HR 1.62 [1.48-1.78]; Figure 3). The adjusted HRs for hospitalization from circulatory system diseases were 1.41 (95% CI, 1.28-1.55), 1.50 (95% CI, 1.25-1.80), and 1.31 (95% CI, 1.15-1.48) in the total, primary, and secondary adrenal insufficiency groups, respectively (Figure 3). The adjusted HRs for hospitalization due specifically to ischemic heart disease were not increased in any patient group (Figure 3). The adjusted HRs for hospitalization due to cerebrovascular disease were increased in the patient group due solely to an increase in those with secondary adrenal insufficiency (Figure 3).
Hospitalization due to cardiovascular disease of patients with adrenal insufficiency relative to matched controls.
Differences in Cardiovascular Events and Mortality by Age and Sex
For composite cardiovascular events, there was no statistically significant difference in the adjusted HRs stratified by sex or age at the start of follow-up in primary adrenal insufficiency. For secondary disease, younger individuals and women had higher adjusted HRs (Figure 4A). However, in both primary and secondary adrenal insufficiency, women or younger individuals had lower incidence rates of composite cardiovascular events compared with men or older individuals (Supplementary Table 3) (22).
A, Composite cardiovascular events in patients with primary and secondary adrenal insufficiency stratified by sex and age at the start of follow-up. B, Ischemic heart disease events in patients with primary and secondary adrenal insufficiency stratified by sex and age at the start of follow-up. C, Cerebrovascular disease events in patients with primary and secondary adrenal insufficiency stratified by sex and age at the start of follow-up.
For ischemic heart disease events, there were no significant differences in the adjusted HRs stratified by sex or age in either primary or secondary adrenal insufficiency (Figure 4B).
For cerebrovascular disease events, there were no significant differences between adjusted HRs stratified by sex or age in primary adrenal insufficiency. In those with secondary disease, younger patients had a significantly higher adjusted HR compared with older patients and women had a greater HR than men, but this was not statistically significant (Figure 4C).
For circulatory system diseases, mortality rates were high in women with primary and men with secondary adrenal insufficiency (Supplementary Table 4) (22), but the adjusted HRs of primary and secondary adrenal insufficiency were not different between sexes (Supplementary Figure 2) (22). In primary disease, the adjusted HR for circulatory system disease was significantly higher in younger ages whereas in secondary, the adjusted HR was not different between age groups (Supplementary Figure 2) (22).
For mortality from ischemic heart disease, the adjusted HRs were not different between sexes (Supplementary Figure 3) (22) in primary or secondary adrenal insufficiency. Young patients with primary adrenal insufficiency had a higher HR than older patients (Supplementary Figure 3) (22).
For mortality from cerebrovascular disease, the HRs stratified by sex and age were not analyzed because of low numbers of deaths (Supplementary Table 4) (22).
Primary Versus Secondary Adrenal Insufficiency
People with primary disease were less likely than those with secondary to experience a cerebrovascular event (adjusted HR, 0.62 (95% CI, 0.48-0.78); P < 0.0001; Supplementary Table 5) (22). Otherwise, there were no differences between primary and secondary adrenal insufficiency in cardiovascular outcomes.
Role of Radiotherapy in Cerebrovascular Disease in Patients With Secondary Adrenal Insufficiency
Of 3948 patients with secondary disease, 520 (13%) received radiotherapy. Of these patients, 11.4% developed cerebrovascular events compared with 6.5% of those who did not (unadjusted odds ratio 1.85 [95% CI, 1.36-2.50], P < 0.0001). After adjustment for age, sex, previous cardiovascular disease, diabetes, hypertension, dyslipidemia, and smoking, the odds ratio was 1.94 (95% CI, 1.42-2.64, P < 0.0001). There was no significant difference in mortality regarding radiotherapy although numbers were small. Admission to hospital for cerebrovascular disease was more common in patients who had radiotherapy than in those who did not (Supplementary Table 6) (22).
Sensitivity Analyses
To validate recording cardiovascular disease in CPRD, 18 636 participants could be extracted. For the diagnosis of ischemic heart disease, false positive and false negative rates were 4.1% and 25.5%, respectively (Supplementary Table 7) (22). For the diagnosis of cerebrovascular disease, false positive and false negative rates were 6.5% and 34.3%, respectively (Supplementary Table 8) (22). Importantly, the false negative and false positive rates in recording ischemic heart disease and cerebrovascular disease in the study patients and controls were comparable.
Discussion
The study design has strengths and limitations. Matching patients with controls is a major strength. As adrenal insufficiency was usually diagnosed and managed in a specialist center, the diagnoses of adrenal insufficiency in the primary care database might have included errors. We minimized this by having glucocorticoid replacement therapy as a condition of entry, ensuring that the treatment was timed to be in close relation to the diagnostic record of adrenal insufficiency. We also cross-checked the diagnosis with hospital admission information, and the sensitivity analysis yielded acceptable positive predictive values for the diagnosis of both primary and secondary adrenal insufficiency (23). Although disparities were observed with the different coding systems used for CPRD and the hospital episode statistics, recording cardiovascular disease in patients and matched controls was affected to a similar extent in both. In a minority (24%), adrenal insufficiency had been diagnosed before the start of follow-up and this may have influenced the analyses. Any effect is unlikely to have been substantial, as existing cardiovascular disease was present to a similar extent in these cases as in the majority in whom the timing of diagnosis and follow-up were coincident. The patients had a higher prevalence of cardiovascular risk factors and disease than controls at baseline. This could have resulted from the underlying disorder(s) (or treatment in the minority in whom diagnosis preceded study entry), or alternatively it could reflect informed presence bias whereby health information is more likely to be recorded in the electronic records of patients than in those of heathy controls. We have addressed such bias by adjustment for comorbidities in Cox regression models, accepting that this may overlook some biologically important influences. Finally, steroid dosage, especially overdosage, has been suggested as a possible cause of excess cardiovascular risk (17), but the extensive primary care database that we analyzed did not include sufficient glucocorticoid dosing information for reliable conclusions to be drawn, since doses were recorded in only 20.6% of patients.
We found that patterns of cardiovascular disease differed between primary and secondary adrenal insufficiency, according to type of cardiovascular disease, and events or mortality.
Cardiovascular events were higher in patients with adrenal insufficiency than in matched controls, but this could be explained largely by higher event rates from cerebrovascular disease in those with secondary adrenal insufficiency. Risk of mortality from cardiovascular disease, specifically ischemic heart disease, was increased in both primary and secondary adrenal insufficiency.
The disparity that we observed in risk of cardiovascular events between primary and secondary adrenal insufficiency suggests that factors beyond glucocorticoid replacement contribute to the excess of cardiovascular disease, which has been reported in other studies of hypothalamic-pituitary disorders (5-8, 12, 13, 15, 18, 19). Other contributing factors include deficiencies in other pituitary hormones, for example, growth hormone, where there is some evidence that replacement may be associated with reduced cardiovascular risk (18, 19). Similarly, excessive thyroid hormone replacement could affect cardiovascular risk. However, our data derive from routine care by general practitioners and it was not feasible to obtain information on biochemical testing for the diagnosis of other pituitary hormone deficiencies and their replacements that would have been recorded in the secondary care setting. In our study, the excess of cerebrovascular events without the excess of ischemic heart disease events does, however, suggest local factors such as hypothalamic or pituitary tumor mass or radiotherapy are more important than any systemic effect of pituitary hormone deficiency or its nonphysiologic replacement. In support of an irradiation-induced effect, we found a marked association between radiotherapy and cerebrovascular disease. This association has also been noted in other studies (7, 12, 15). We observed that cerebrovascular events were predominantly increased in younger patients, consistent with previous reports of increased cerebrovascular mortality (5) and all-cause mortality (7, 13, 15) in this group. Cerebrovascular events have been associated with cranial irradiation in young patients with various types of brain tumor (24) and radiation-induced vasculopathy has been proposed as the mechanism (25). We found only a nonsignificantly higher risk of cerebrovascular disease in women relative to men. Some (5, 8, 12) but not all previous studies (6, 15) have reported increased cerebrovascular mortality especially in women.
Although the incidence rates of composite cardiovascular disease, ischemic heart disease, or cerebrovascular disease events in patients with primary adrenal insufficiency were higher than controls, the hazard ratios after adjustment for established cardiovascular risk factors were no longer significant and were similar in both sexes. This finding differs from a recent study that reported an increased risk of ischemic heart disease in women with autoimmune Addison disease (17), but that study did not adjust for confounders and the patient group differed (17).
The increased risk of death from ischemic heart disease in patients with either primary or secondary adrenal insufficiency accords with previous reports in primary adrenal insufficiency (9) and in hypothalamic-pituitary disorders (5-8, 12, 13, 15). A plausible explanation for the increased risk of death when event rates did not appear to be increased could be a reduced ability to survive the stress of a cardiovascular event.
The mortality rates from cerebrovascular disease in primary or even in secondary adrenal insufficiency were not different from controls. This is in line with a previous study of autoimmune Addison disease (17) but contrasts with previous findings in hypothalamic-pituitary disorders where standardized cerebrovascular mortality ratios have been elevated (5-8, 12, 13, 15). These previous reports did not match patients with controls (instead using merely national database estimates) and were thus unable to adjust for potential confounders, regional differences in quality of care, or temporal changes in the quality of care.
In conclusion, in this study, composite cardiovascular events were increased in patients with adrenal insufficiency and this could be largely explained by cardiovascular comorbidities. Cerebrovascular events were, however, independently increased in secondary adrenal insufficiency and previous irradiation therapy was a major factor. Cardiovascular mortality, specifically ischemic heart disease, was increased in patients with adrenal insufficiency of any type. These findings support further optimization of glucocorticoid replacement in conjunction with cardioprotective interventions in patients with adrenal insufficiency.
Abbreviations
- CPRD
Clinical Practice Research Datalink
- HES
Hospital Episode Statistics
- HR
hazard ratio
- ICD-10
International Classification of Diseases, 10th revision
- ONS
Office for National Statistics
Acknowledgments
K.N. is supported by the Scholarship in Commemoration of HM King Bhumibol Adulyadej’s 90th Birthday Anniversary. A.M. is supported by the NIHR Applied Research Collaboration for NW London. The views expressed in this publication are those of the authors. The sponsors had no role in study design, analysis, or manuscript content.
Author Contributions: K.N. designed the work, analyzed and interpreted the data, designed the illustrations, drew the figures, and drafted the article. D.G.J. designed the work, interpreted the data, designed the illustrations, and critically revised the manuscript. I.F.G. analyzed and interpreted the data, designed the illustrations, and critically revised the manuscript. J.C. and N.O. interpreted the data and reviewed the manuscript. A.M. and J.K.Q. designed the work and reviewed the manuscript. S.R. designed the work, interpreted the data, and critically revised the manuscript. The corresponding author had full access to all data and had final responsibility for the decision to submit for publication.
Ethics Committee Approval: The study protocol was approved by the Independent Scientific Advisory Committee for MHRA Database Research (protocol number 18_179).
Additional Information
Disclosures: The authors declare that there are no conflicts of interest related to this manuscript.
Data Availability
The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
References
