The impact of diabetology consultations on length of stay in hospitalized patients with diabetes

Abstract Introduction Both hyperglycaemia and hypoglycaemia in hospitalized patients have been shown to be associated with a longer length of stay, higher readmission rates, and higher rates of morbidity and mortality. With 25%‐30% of all hospitalized patients carrying a diagnosis of diabetes, it is important to optimize glycaemic control. Current guidelines for care of inpatients with diabetes now suggest consulting a specialized diabetes team for all patients when possible. Aim This study was a retrospective cohort study to evaluate the impact of an inpatient diabetology consult within 48 hours of admission on patients’ length of stay. Methods All patients admitted to the general medicine service between 2013 and 2018 with a diagnosis of diabetes in their medical record were included, which consisted of 11 477 inpatient stays. We looked at the effect of an inpatient diabetology consultation within the first 48 hours on length of stay, complications and 30‐day readmission rates. Results We found that patients whose care included a diabetology consult within 48 hours of admission had a statistically significant shorter length of stay by 1.56 days compared to the remainder of the group. There was no difference in complications or 30‐day readmission rates between the groups. Conclusion Among general medicine patients with a diagnosis of diabetes, timely diabetology consultations reduced patients’ length of stay and have the potential to improve their care and lessen the economic impact.

An increasing number of studies show that consultations with specialized diabetes teams among hospitalized patients with diabetes resulted in a lower LOS, although these studies have all had fairly small patient populations. [5][6][7][8] The effects of specialized diabetes teams on readmission rates, however, are less clear. One study in Spain found no difference in readmission rates with or without inclusion of a specialized diabetes team in the patients' care. 8 On the other hand, a retrospective study comparing hospitalized patients cared for by a specialized diabetes team had significantly lower readmission rates when compared to patients cared for by a primary service team only. 5 Further, another retrospective study showed a reduction in composite morbidity among hospitalized patients with diabetes if a consultation by a specialized diabetes team was provided. 9 Given the limited evidence and the importance of diabetes care, a call to action was published in 2013 that outlined the need for further studies to evaluate morbidity, mortality, glycaemic control, and other outcomes in this patient population. 3 However, due to the increasing evidence of the effectiveness of utilizing dedicated diabetes teams, current guidelines from the American Diabetes Association (ADA) recommend consulting a specialized diabetes team for hospitalized patients with diabetes when possible. 13 Our aim was to look at a much larger patient population and to evaluate the impact of a diabetology consultation on LOS in patients with diabetes who were admitted to the general medicine service.

| ME THODS
Our study was a retrospective analysis of an electronic medical record (EMR) from a large hospital in the northeastern United States, which is an academic teaching institution, including an Internal Medicine residency programme and an Endocrinology, Diabetes and Metabolism fellowship programme. We collected hospital records for all stays between 2013 and 2018 for inpatients who were at least 18 years of age with either a diagnosis of diabetes in the patient's medical record or a diagnosis of diabetes from one of the hospital network's primary care professional sites 12 months prior to the inpatient event on the general internal medicine service. Patients admitted to the general medicine service are primarily cared for by a teaching team consisting of an attending physician, a resident, an intern and medical students, although a smaller subset of patients may be admitted and cared for by an advanced practice provider. Inpatient diabetology consultations (IDCs) are able to be called at any time, and patients are typically seen by an Endocrinology, Diabetes and Metabolism fellow and a supervising attending physician. Inpatient stays for pregnant women were excluded from the sample (n = 21 007).
The sample was further limited to those patients with a stay of more than 48 hours (n = 2933) because our key independent variable was IDCs within the first 48 hours. We also excluded those with at least 1 day in the intensive care unit (n = 1383) given that

| RE SULTS
Only 3% of the hospital stays we examined had an IDC within 48 hours of hospital admission (Table 1). Those inpatient stays that received an IDC tended to be for patients that were younger or had higher average POC BG scores. Further, only 7.42% of the hospital stays received an IDC at any time during their hospital admission (n = 825). Table 2 displays the results of the Poisson regression analysing LOS. When compared to the mean LOS for all inpatient stays reviewed in this study, an IDC within 48 hours of admission led to a 1.56 day shorter average LOS that was statistically significant (P < .001). LOS was reduced by nearly half a day for females compared to males (P = .043) and decreased as individuals aged (P < .001). As the level of SOI and ROM increased, LOS is significantly longer. Average POC BG values, BMI and having a consult prior to hospital admission did not significantly change LOS for patients with diabetes in the general internal medicine unit. Table 3 shows that IDCs had no statistically significant effect on the rate of inpatient complications (P = .167). Female patients had a 1.0% lower chance of complications compared to males (P = .017). Patients with level 4 SOI at admission were 4.2% more likely to experience complications during the hospital stay compared to those admitted with a SOI at level 1 (P = .002); SOI levels 2 and 3 were not statistically significant. There was a significant increase in the likelihood of complications for each ROM level increase. Patients admitted with an ROM equal to 4 experienced a 3.4% higher rate of complications compared to patients with a ROM equal to level 1 (P = .001). The likelihood of complications increased as a patient's average POC BG value increased (P = .028).
BMI, age and diabetology consultation prior to the hospital admission did not significantly impact a patient's likelihood of experiencing complications.
The probability of a readmission decreased for those with IDCs within 48 hours of admission, but not statistically significantly (Table 4). The probability of any readmission increased for those with an ROM at level 2 by 5.8% and by nearly 7.7% compared to those at level 1 (P < .001 and P < .001, respectively). The probability of readmission was 1.6% higher for women when compared to men (P = .025). SOI level, average POCT value, BMI and diabetology consults prior to admission did not significantly change the probability of readmission.
Patients who received an IDC at any time during their hospitalization compared to those who never received a consult had a significantly longer LOS by 3.16 days (P < .0001). The rate of complications was also significantly increased by 2.2% in those who received an IDC consult at any time (P < .0001). The probability of readmission for those with IDCs at any time during their hospitalization was increased by 0.29%, although this was not significant (P = .731).

| D ISCUSS I ON
This study showed that among hospitalized patients with diabetes admitted to the internal medicine service, a consultation with dia-

Mean (Standard Deviation) Mean (Standard Deviation)
Length

ACK N OWLED G EM ENTS
We would like to acknowledge Rebecca Aksdal for assistance in editing of the manuscript.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest. acquisition, writing and editing of manuscript. JJS: study conception/ design, data analysis, editing of manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Our data are not available for sharing given it contains confidential patient information, but is stored on a secure server should access be needed.