Diabetic emergencies presenting during the COVID‐19 pandemic: A retrospective case analysis

Abstract Introduction COVID‐19 has triggered a global pandemic and is an emerging situation. Diabetes has been associated with significant mortality in SARS and MERS‐COV infections. Patients with diabetes are at risk of COVID‐19 triggering diabetic emergencies due to known and unknown mechanisms. There is little evidence overviewing the clinical course of COVID‐19 patients who either present or have diabetic emergencies during their disease course. Methods We conducted a retrospective case analysis of all patients admitted to our hospital during the COVID‐19 pandemic. The inclusion criteria were all patients receiving treatment for COVID‐19 and either presenting with a diabetic emergency on admission or developing an emergency during their admission. Data collected for the study were all routinely collected data as part of the admission. We compared these data to nine patients with no COVID‐19. Results Thirty patients received treatment for a diabetic emergency, of which 21 also received treatment for COVID‐19. Significant differences were found between pH and bicarbonate on admission between RT‐PCR‐positive and both RT‐PCR‐negative and non‐COVID‐19 patients. Other results approaching significance include ALP and eGFR. Discussion Patients suffering from COVID‐19 and diabetes concurrently can suffer from profound metabolic disturbance, with a significant difference in inpatient mortality. However further, prospective detailed investigation into biochemical processes is needed to fully elucidate underlying mechanisms that affect these patients' outcomes.


| ME THODS
We conducted a retrospective case analysis of patients with diabetes admitted to the medical wards of North Middlesex Hospital, London, during the COVID-19 pandemic between 15 March and 15 April 2020. The inclusion criteria were: (1) positive swab testing for COVID-19, (2) admission due to a diabetic emergency. We recorded details regarding the length of hospital stay, features of the metabolic emergency, time until resolution, treatment pathways utilized, clinical course of the patient, clinical outcome at 7 days post admission. Data for the study were extracted from routine hospital records. Notes were manually reviewed for results as well as for follow-up information. The study was part of a departmental audit, evaluating treatment pathways for patients with diabetic emergencies during the pandemic, and has been approved by the Ethics Department of our Hospital.

| Diagnosis of COVID-19
Primarily patients with clinical features matching the UK government case definition for COVID-19, 10

| Control group
We also had a group of patients who underwent treatment for a diabetic emergency without being investigated or treated for COVID-19. These patients were used as comparators to investigate significant metabolic differences. Qualitative characteristics were presented as mean values ±standard deviation while quantitative characteristics were presented as percentages (%). Differences between groups were evaluated using independent student t-test. Statistical significance was set at the level of p-value <0.05. We have split the patients into RT-PCR swab positive and RT-PCR swab negative but treated with a high clinical suspicion. Kaplan-Meier survival graph was calculated based on length of stay in days for the three patient groups with death as the event.

| RE SULTS
Thirty patients are included in this retrospective analysis, with their demographics demonstrated in Table 1 Seven patients were RT-PCR negative but treated for COVID-19 due to high degree of clinical suspicion. The average age was 47.33 ± 20.46 years, with five males and two females. 43% of these patients had type 1 diabetes. One patient had DKA, two had HHS, two had a mixed picture with one patient having hyperglycaemic ketosis.
67% of patients had been successfully discharged home within 7 days.
For the patients who were not treated for COVID-19 (n = 9), all patients had DKA despite the fact that 33% had type 2 diabetes.
There were four males and five females. 67% (n=6) of these patients were insulin treated, and 100% were discharged home within 7 days of admission. Table 2 demonstrates the admission gas results, blood tests and glycosylated haemoglobin (HbA1c) as well as selected COVID-19 markers. Broadly routine admission blood tests were similar when averaged and compared between the three groups. There was a non-significant rise in white cell count going from the RT-PCR patients to non-COVID-19 patients. No significant lymphopaenia was observed between any of the three groups. Kaplan-Meier survival curves were drawn for the groups, comparing all-cause mortality. This is shown in Figure 1. It demonstrates that the cumulative survival time drops if being treated for COVID-19, and more so if treated for COVID yet RT-PCR negative.

| DISCUSS ION
This is the first single-centre report on COVID-19-related diabetic emergencies. Some of the data in this study were part of the three-centre study which was published in Lancet Diabetes and Endocrinology recently. 13 We have observed that these pa- correlates with the observation that acute kidney injury (AKI) is more often observed in this cohort. 15 There was no difference in liver function tests unlike previous published trials. 16  disease severity. This was based on a retrospective study from Wuhan, the region of China first affected by COVID-19. 17 Our patients show elevated levels of these markers, with marginally reduced lymphocyte count, suggesting that there is severe metabolic derangement-however, whether this is due to COVID-19, the diabetic emergency or a combination is yet to be determined.

TA B L E 1
Baseline Demographics of all patients that were management for both a diabetic emergency and COVID-19, including destination at 7 days

TA B L E 2 Admission Blood test results including some COVID-19 markers
Blood test result mean  .828 .891 The values in bold were found to be statistically significant.

| Limitations
Our study was small (n = 30) which makes statistical analyses difficult and can lead to false significance. However, the findings of our retrospective analysis are corroborated by both Bornstein et al. 5

and
the Diabetes UK guidelines 6 whose recommendations are in line with our findings.
One other key factor of our analysis revolves around the  20 There are many barriers to the rapid roll out of serological tests including assessment of sensitivity and specificity of the tests, possibility of false-positive tests due to cross-reactivity with other viral pathogens and ensuring that mass production of the test is economically viable. 21 Another key limitation of this report is that this is a retrospective analysis of routinely collected data, meaning that there are some gaps in investigations. These primarily included lactate dehydrogenase, ferritin and aspartate transaminase in the non-COVID-19 patients and some of the patients treated during the COVID-19 pandemic.

| Conclusion
We present a single-centre observational analysis of COVID-19 patients who suffered from a diabetic emergency, in comparison to non-COVID-19 patients. Some significant differences were found; however, further investigations are needed into the relationship of COVID-19 and their impact on diabetic control and the risk of diabetic emergencies. A significant difference in mortality was observed. Prospectively collected data with a larger patient base, including data on COVID-19 markers, could help shed further light on the management of inpatient diabetics with COVID-19.

ACK N OWLED G EM ENTS
Not applicable.

CO N FLI C T O F I NTE R E S T
Nothing to declare by any author. Writing-original draft (lead); Writing-review & editing (lead).

E TH I C S A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
We followed the NHS REC decision flowchart and in line with this guidance ethical approval was not sought. The NHS REC flowchart can be found at http://www.hra-decis ionto ols.org.uk/ethic s/. We felt that this vital information should be shared with the wider medical community and so a manuscript was prepared and submitted to BMC.
Consent was not sought at the time of patient admission as they were not being enrolled into any study. This was part of a clinical audit registered with the trust for review of paper notes. All patient healthcare data were collected anonymously, retrospectively and no patient identifiable data were collected or described in the manuscript. This was in line with the NHS Health Research Authority guidance. 22

DATA AVA I L A B I L I T Y S TAT E M E N T
Available from authors on reasonable request.