Increased infection risk in Addison’s disease and congenital adrenal hyperplasia: a primary care database cohort study

Context Mortality and infection-related hospital admissions are increased in patients with primary adrenal insufficiency (PAI). However, the risk of primary care-managed infections in patients with PAI is unknown. Objective To estimate infection risk in PAI due to Addison’s disease (AD) and congenital adrenal hyperplasia (CAH) in a primary care setting. Design Retrospective cohort study using UK data collected from 1995 to 2018. Main outcome measures Incidence of lower respiratory tract infections (LRTIs), urinary tract infections (UTIs), gastrointestinal infections (GIIs), and prescription counts of antimicrobials in adult PAI patients compared to unexposed controls. Results A diagnosis of PAI was established in 1580 AD patients (mean age 51.7 years) and 602 CAH patients (mean age 35.4 years). All AD patients and 42% of CAH patients were prescribed glucocorticoids, most frequently hydrocortisone in AD (82%) and prednisolone in CAH (50%). AD and CAH patients exposed to glucocorticoids, but not CAH patients without glucocorticoid treatment, had a significantly increased risk of LRTIs (adjusted incidence rate ratio AD 2.11 [95% confidence interval 1.64-2.69], CAH 3.23 [1.21-8.61]), UTIs (AD 1.51 [1.29-1.77], CAH 2.20 [1.43-3.34]), and GIIs (AD 3.80 [2.99-4.84], CAH 1.93 [1.06-3.52]). This was mirrored by increased prescription of antibiotics (AD 1.73 [1.69-1.77], CAH 1.77 [1.66-1.89]) and antifungals (AD 1.89 [1.74-2.05], CAH 1.91 [1.50-2.43]). Conclusions There is an increased risk of infections and antimicrobial use in PAI in the primary care setting at least partially linked to glucocorticoid treatment. Future studies will need to address whether more physiological glucocorticoid replacement modes could reduce this risk. Précis Using data from 1580 AD patients and 602 CAH patients collected in a UK primary care database from 1995 to 2018, we identified increased risk of infections and antimicrobial prescription counts.

For the second outcome, antimicrobial use, we used the codes for antibiotics and antifungals as 1 3 9 classified in the British National Formulary. We then calculated the total number of prescriptions 1 4 0 for every antimicrobial in each cohort.

4 1
For each of the study groups we analyzed age, sex, body mass index (BMI), smoking status, Descriptive statistics were used to summarize the baseline characteristics for the exposed and 1 5 0 unexposed groups of patients. Categorical variables were investigated using Chi-square test 1 5 1 and continuous variables were analyzed using t-test. Adjusted incidence rate ratios (aIRRs) for specific infections and antimicrobial prescriptions 1 5 3 were calculated after adjustment for age, sex, smoking status, BMI, Townsend Deprivation 1 5 4 Index, and Charlson Comorbidity Index, using multivariate Poisson regression analysis.  Committee in 2003. The present study was reviewed and approved (study reference: In total, 1580 patients fulfilled the AD criteria; these were matched with 3158 unexposed 1 6 4 individuals ( Table 1). The mean age of AD patients was 51.7 years and the majority were 1 6 5 women (57.8%). Compared to unexposed individuals, AD patients had a lower median BMI, 1 6 6 while the Townsend Deprivation Index did not differ significantly between the two groups. The Charlson Comorbidity Index showed that AD patients had an increased burden of comorbidities In total, 602 patients fulfilled the CAH criteria and were subdivided into 254 glucocorticoid- with 508 and 696 unexposed controls, respectively (Table 2). The majority of CAH patients were female (72.3%), with a lower mean age in glucocorticoid-1 7 6 treated patients at cohort entry (33.4 vs. 36.9 years). CAH patients had a higher median BMI 1 7 7 compared to controls, and this was evident for both glucocorticoid-treated sub-cohort and the  The most commonly prescribed type of glucocorticoid in the AD cohort was hydrocortisone 1 8 5 (1296 patients, 82%), followed by prednisolone (187 patients, 11.8%). Only a minority of 1 8 6 patients were prescribed cortisone acetate (91 patients, 5.8%, no longer available in the UK) 1 8 7 and dexamethasone (6 patients, 0.4%).

8 8
In the glucocorticoid-treated CAH cohort, prednisolone was most commonly prescribed (127 1 8 9 patients, 50.0%), followed by hydrocortisone (96 patients, 37.8%), with a small minority 1 9 0 receiving dexamethasone (15 patients, 5.9%) or cortisone acetate (11 patients, 4.3%). Only five 1 9 1 CAH patients (2%) were prescribed a combination of short-and long-acting glucocorticoids. The risk of LRTIs, UTIs and GIIs was significantly increased in the AD cohort compared to 1 9 4 unexposed patients, with the highest relative risk observed for GIIs (adjusted incidence rate  Figure 1). These results were confirmed in 1 9 7 the sub-analysis of patients who had at least two prescriptions of both glucocorticoids and 1 9 8 mineralocorticoids (94.5% of the total cohort) (Suppl.  Figure 1). However, when analyzing the population accordingly to 2 0 2 glucocorticoid use, only patients exposed to glucocorticoids had an increased risk of infections,  Figure 1). In contrast, infection risk in CAH patients not treated with glucocorticoids did not differ from that   and Figure 1).

1 7
Given the higher incidence of type 1 diabetes mellitus (T1DM) in our AD cohort (8% vs. 0.5% in 2 1 8 matched controls), and given the potentially higher risk of infections in T1DM patients, we  In this population-based study we found that the risk of three common infections (lower 2 2 4 respiratory tract, urinary tract, and gastrointestinal infections) was increased in the primary care 2 2 5 setting in patients with PAI, as compared to population-based matched controls. This was also 2 2 6 supported by our finding of increased prescription rates of antimicrobials in in patients with PAI.

7
Moreover, we found that CAH patients not receiving glucocorticoids did not have an increased infection risk observed in PAI. To our knowledge, our study is the first to analyze the risk of 2 3 0 infection in PAI according to different etiologies and also the first to evaluate these outcomes in 2 3 1 a primary care setting. focused on hospital-treated infections only, possibly overestimating the actual incidence of this 2 3 7 complication due to a lower threshold for admission in PAI patients. In addition, information on 2 3 8 the actual etiology of PAI was not available in this study, as PAI was diagnosed based on 2 3 9 concomitant glucocorticoid and mineralocorticoid prescriptions, which did not allow to 2 4 0 differentiate between AD, CAH, and other causes of PAI. patients frequently also suffer from other autoimmune comorbidities (18), and this was 2 4 5 confirmed in our study, with more prevalent autoimmune disease in our AD cohort, which can be 2 4 6 safely assumed to consist of a large majority of patients with AD of autoimmune origin. (19), and as we found similar increases in infection risk in the CAH cohort, potential etiology-2 4 9 related immune function is unlikely to explain the increased susceptibility to infections we 2 5 0 observed.

5 1
Supraphysiological glucocorticoid doses, as usually administered in the context of chronic 2 5 2 inflammatory disease, is well known to cause changes in the immune system, with consequently  occur with glucocorticoid replacement in PAI. A recent paper documented significantly 2 6 8 decreased natural killer cytotoxicity in patients with PAI (7), which was present in both patients with autoimmune adrenalitis and those with PAI following bilateral adrenalectomy, indicating that 2 7 0 the underlying etiology did not play a role in these changes in immune function. A recent 2 7 1 randomized control trial including patients with primary and secondary adrenal insufficiency 2 7 2 reported a reduction in respiratory tract infections with modified-release hydrocortisone (8). peripheral blood mononuclear cells in the PAI patients at baseline, which attenuated after the 2 7 7 switch to modified-release hydrocortisone therapy (24). The findings of our study, including both 2 7 8 patients with AD and CAH, suggest that exogenous glucocorticoid is at least a contributory 2 7 9 factor to the increased infection risk we observed, given that no significant increase in infection 2 8 0 risk was observed in the CAH patients not receiving glucocorticoid therapy. Both our AD and CAH populations had increased prescription rates for antibiotics and 2 8 2 antifungals. Interestingly, increased prescription rates were also noted in the CAH patients not   The highest increase in risk of infection in our AD cohort was seen in GIIs, while for the CAH 2 8 9 cohort on glucocorticoids the most significant increase in risk was seen in LRTIs and UTIs; 2 9 0 however, the differences between the three infection groups was not statistically significant. This 2 9 1 may be explained by the age difference between AD and CAH patients, with mean ages of 51.7 2 9 2 and 35.4 years, respectively. Indeed, LRTIs and UTIs are more frequently diagnosed in older 1.4%, UTIs 8.5%). Therefore, the higher aIRR of LRTIs and UTIs in CAH patients is probably 2 9 6 related to a difference in age-related background risk.

9 7
Our AD cohort had an age and sex distribution similar to the one reported in other papers (2,5), 2 9 8 and the types of prescribed glucocorticoid preparations at baseline in this cohort were not 2 9 9 different from the ones reported in a recent worldwide survey (28). Our CAH cohort was 3 0 0 younger that the AD cohort, consistent with the different etiology of these two diseases, and the 3 0 1 types of glucocorticoids prescribed was similar to those reported in the cross-sectional UK 3 0 2 CaHASE study (29), with the possible exception of lower numbers of dexamethasone users in 3 0 3 our study. Taking this into account, our results can be assumed to be representative of the UK 3 0 4 AD and CAH populations. Our study has several strengths. We used a large population-based sample of patients of both GPs and this could have resulted in some degree of misclassification of the exposed cohorts visiting GP might be lower in patients with PAI who receive regular education on the importance 3 1 6 of treating infections promptly to avoid adrenal crisis. This may be a factor resulting in a degree 3 1 7 of overestimation of the difference in the infection rates we found between these cohorts.

1 8
However, since patients with PAI are generally more medicalized, it is also possible that they  Furthermore, even though we tried to assess the impact of associated comorbidities by there is some evidence of an immune-modulatory effect of androgens (30), we could not take 3 2 7 this into account in our population as we had no data on DHEA replacement therapy in AD 3 2 8 patients, since this is a hospital-prescribed drug in the UK; similarly, we did not have data on 3 2 9 biochemical control of androgen excess in the CAH patients. patients. This may also provide a case for recommending a vaccination strategy in PAI, e.g. patients (32). Future studies will have to clarify whether achieving a more physiological delivery 3 4 0 of glucocorticoid replacement will decrease the risk of infections in PAI, with the potential to 3 4 1 result in reduced morbidity and mortality in these patients.  recommendations of the Advisory Committee on Immunization Practices (ACIP).