Predictors of 30-Day Readmissions for Adrenal Insufficiency

A Retrospective National Database Study

Asim Kichloo; Zain El-amir; Hafeez Shaka; Farah Wani; Sofia Junaid Syed


Clin Endocrinol. 2021;95(2):269-276. 

In This Article


Understanding hospitalizations and readmissions related to AI is important in guiding clinical management. Previous research has discussed hospitalizations and comorbid conditions, but our study's focus on readmission characteristics compared with index admissions, readmission rates and predictors of readmission may help guide future clinical practice.

Reasons for Readmission

This study found that the all-cause readmission rate was 17.3%. The most common reason for readmission was AI, and additional causes of readmission included sepsis from an unspecified organism, unspecified pneumonia and unspecified acute renal failure. Adrenal crisis has been reported to make up a quarter of hospitalizations for patients with AI.[4,9] Patients with AI are reported to have 1.5 higher rates of infectious episodes than controls in one cohort study.[10] Additionally, the cohort study found that hospitalization rates for infection were greater in AI patients than without AI.[10] Patients with AI are reported to have a five times higher mortality rate from infectious disease than those without AI from an age-adjusted background population.[11] The higher infection rate and mortality may be due to treatment with glucocorticoids, which is known to impact the function of the adaptive cellular immune system.[10] The known higher incidence of infection may explain why second to AI as a primary reason of readmission, sepsis and pneumonia, two infectious processes, were commonly reported causes of readmission in patients with AI.

Comorbidities and Charlson Comorbidity Index in Readmissions

The Charlson Comorbidity Index (CCI) was developed to predict the one-year mortality based on comorbidity data gathered from hospital chart review on over 600 patients that was then weighted based on the potential to impact patient mortality.[12] Patients who were readmitted were more likely to have CCIs greater than 2. Readmissions also had higher proportions of comorbid congestive heart failure, chronic kidney disease and protein-energy malnutrition but a lower proportion of hypertension compared with index admissions.

Comorbidities have been identified as risk factors for adrenal crisis in patients with AI, which may contribute to hospitalization rates.[13] Patients with both primary and central AI have higher risk of comorbid conditions compared with the general population.[5] One retrospective study studying the annual incidence of AI from 1996 to 2008 in Taiwan reported that only 151 patients in the 13-year period had AI as the sole diagnosis upon discharge.[14] The most commonly reported comorbidity was pneumonia followed by urinary tract infection, diabetes mellitus, electrolyte imbalance and COPD.[14] Higher rates of hypertension, renal insufficiency, COPD and cancer and higher CCIs were present in patients with AI in a previous retrospective study.[5] The higher prevalence of comorbid conditions in patients with AI has been attributed to long-term excessive glucocorticoid exposure.[4] Previous reports have stated that comorbidities are the most influential factor in determining the incidence of future adrenal crisis, which accounts for a quarter of hospital admissions in patients with AI.[4,9] This relationship may explain the higher incidence of comorbid conditions in readmissions patients, and the higher comorbidity rates may increase risk of adrenal crisis and result in readmission.

Outcomes in Readmission and Predictors of Readmission

All-cause readmissions had higher odds of inpatient mortality, increased mean LOS, higher THC and higher COH. This may be secondary to the aforementioned higher CCIs and comorbidity rates in patients with AI. Higher rates of comorbid conditions mean that additional consultants may be necessary, which may result in higher THC and COH, as well as increased LOS. Regarding mortality, AI is known to be life-threatening, with previous studies reporting reduced life expectancy in patients with autoimmune primary AI.[15] Patients with hypopituitarism, which may lead to central AI, have excess mortality, as well.[16] Hospitalization with a primary diagnosis of AI, which was the most common readmission diagnosis, may reflect the deadly nature of the condition. Additionally, infectious processes were common in readmissions, and patients with AI have higher reported mortalities from infectious disease processes than matched counterparts, which may explain the higher odds of inpatient mortality in readmissions.[11]

The predictors of readmission identified in this study were as follows: CCI of 3 or more, being discharged against medical advice, protein-energy malnutrition and obesity. Higher CCIs may mean that more complications of comorbid conditions are possible, which may increase the risk of readmission. Some comorbid conditions that have been reported to be more prevalent in patients with AI, including diabetes and COPD, may result in serious complications and perhaps contribute to readmission risk.[14] Leaving against medical advice, or leaving before the treating physicians recommend discharge, was also a predictor of readmission. Patients discharged against medical advice have been found to be more likely to be homeless and to have multiple comorbidities, including HIV, addiction and psychiatric illness in a retrospective matched cohort study.[17] The higher propensity to have numerous comorbid conditions with the discovered higher risk of readmission in patients with CCI of 3 or more may explain association between leaving AMA and readmission.

Obesity was found to be a predictor of readmissions in patients with AI. Previous research has shown that obesity is an independent risk factor for readmission within 30 days of discharge, such as in patients with elective spine surgery.[18] However, it should be recognized that obesity and AI have a known association. Obesity and AI have been linked because long-term excesses of glucocorticoids in patients with adrenal insufficiency increase the risk of developing obesity.[5] Obesity and AI have also been associated through a POMC mutation.[19,20] POMC is a pro-peptide that is expressed in the immune system, skin, pituitary gland and hypothalamus.[19] POMC mutation(s) can result in early-onset obesity and adrenal insufficiency with or without red hair pigmentation and pale skin.[19,20] The rates of readmission in patients with obesity and AI may reflect an underlying genetic defect, which can be diagnosed in early childhood in individuals; moreover, since POMC-derived peptides have numerous biological functions, there is the possibility of other derangements and manifestations later in life.[19,20] An underlying genetic tie between obesity and AI may explain readmission. On the other hand, protein-energy malnutrition was also found to be a predictor of readmission. Previous studies have shown higher rates of malnourishment in patients with AI.[21] Despite overweight or obese BMIs, elderly individuals with AI have been reported to be more prone to develop malnutrition.[21] Malnutrition has been a predictor of readmission and death in early and late periods after hospital discharge.[22] Previous studies have reported that the acute condition resulting in index admission can weaken a patient's overall health, and malnutrition compounds the problem and results in a higher risk of further exacerbations or complications of what may have been previously stable comorbidities.[22] The relationship between AI and malnutrition and the known relationship between malnutrition and readmissions may provide an explanation to the findings of our study.

Strengths and Limitations

This study has several strengths that should be noted. The population is one of the strengths of this study, as the population used is drawn from what is thought to be a large, hospital-based, multiethnic registry. Additionally, because this study examines several outcome-oriented facets and demographics of AI admissions and readmissions, it offers a comprehensive and thorough overview of the readmissions in the US healthcare system and the seriousness of AI readmissions. However, as with any study, there are limitations that should be appreciated. Data from the NRD are subject to biases that are associated with retrospective studies, and the database reports information on hospitalizations rather than individual patients; therefore, patients admitted numerous times or with numerous readmissions would be included more than once in the data set. The NRD data do not include information about the severity of AI at the time of admission or readmission. Moreover, the NRD uses ICD-10 codes to report hospitalizations and thus is subject to the possibility of coding errors.

Despite the limitations, the large sample size paired with the scientific inquiries and analysis techniques make for a study that sheds lights on a relatively under-investigated subject while aiming to encourage further discourse and future large multicentre controlled prospective studies on AI readmissions.