The analysis that was conducted over a 16-month period has indicated that many factors already known to influence the likelihood of serious consequences of a COVID-19 also influenced mortality rate in the 16-month follow-up period of our study. However when all factors were considered together, only age, male sex and social disadvantage were associated with an increased risk of death, as was a lower LDL-cholesterol level, likely as marker of use of high dose statin treatment in people with a history of cardiovascular disease. This suggests that it is the constitutional characteristics of individuals that are the most important independent determinants of mortality outcome following a COVID-19 infection in individuals with T2DM.
Data obtained over a 16-month period show that 7.7% of people with T2DM who had a GP-confirmed COVID-19 infection died compared to a figure of 6.0% for age and gender-matched individuals without diabetes. Thus, the mortality rate was 1.7% higher in people with T2DM than in age and sex-matched controls. Although there was a very small difference in mean age between individuals with T2DM and individuals without T2DM, the large sample size means we obtain a significant p value. Also, not all individuals had three matches; some only had two and this contributed to a small but statistically significant difference.
We report that for people with T2DM, prescriptions of metformin, an SGLT2i or GLP-1 agonist were each associated with a lower mortality rate for individuals infected with COVID-19. A number of studies have indicated that COVID-19 infection is associated with changes in vascular endothelial function including increased production of pro-inflammatory cytokines and of micro-particles . Metformin, SGLT2is and GLP-1 agonists are all known to reduce cardiovascular event rate in individuals with T2DM  and the mechanisms by which they do this may be linked to their potential protective effect in COVID-19-infected individuals with T2DM. Prescription of insulin was associated with a higher mortality rate. A potential explanation is that insulin is usually given to people with T2DM who have had T2DM for longer, have historically poorer blood glucose control and potentially a greater burden of multiple morbidity/diabetes tissue complications.
In the pathogenesis of SARS-CoV-2 (COVID-19) infection, it is important to reflect on the role played by serine protease expression in human cells, a leading cause of the entrance and replication of SARS-CoV-2 and in parallel the different cellular expression of ACE2 in humans , as has been described in T2DM and in altered glycemia .
It has been suggested that the population at higher risk of worse prognosis are people with hypertension , and in people with diabetes [19, 20]. Indeed, these patients are at higher risk for intensive care unit (ICU) admission and death. It is likely that the risk factors are closely linked.
The analysis described an association of lower BMI and lower HbA1c with a higher mortality rate in T2DM. This has been reported previously . The observation was found to correspond to a U-shaped relation between BMI/HbA1c and mortality. In other words, people with a both a high and very low BMI were found to have a higher mortality rate. This is likely due to confounding factors as previously described .
The association seen here between South Asian ethnicity and a lower mortality rate may be related to there being a higher number of younger South Asian people with T2DM vs Caucasians, given the younger age of diagnosis of many South Asians with T2DM. Nevertheless the age of death of South Asian individuals with T2DM was significantly lower than for other ethnic groups. The absence of an ethnicity effect but presence of social disadvantage in the final model, as described in Table 4, may be a consequence of the link between socio-demographic situation and greater likelihood of adverse outcome following a COVID-19 infection, such that this outweighs any specific ethnicity effects, because of the close linkage between socio-economic status and ethnicity.
When the T2DM and matched controls are analysed together, the effects of higher age, prior diagnosis of T2DM, male gender and greater social disadvantage were independent determinants for risk of dying following a confirmed COVID-19 infection. The significant relation between socio-economic disadvantage and higher mortality rate is important even if the underlying factors are not easily modified. The association of lower LDL-cholesterol with a higher mortality rate may be associated with a more aggressive lipid-lowering therapeutic strategy for people with a history of cardiovascular events (themselves being at greater risk of adverse consequences of a COVID-19 infection) [22, 23] resulting in a lower LDL-cholesterol. We would speculate that people who know that they are at greater cardiovascular risk may be more likely to take lipid-lowering medication as prescribed and to be prescribed a more potent statin. Clearly this finding needs to be examined in other post-COVID-19 infection population cohorts. The absence of any independent effect of prior diagnosis of hypertension and of ethnicity in the final combined analysis suggests that there is no specific effect of these factors in individuals with T2DM, above and beyond that seen in the population as a whole.
These findings can be compared with an early COVID-19 pandemic study from Mexico  which concluded that early-onset T2DM conferred an increased risk of hospitalization and obesity conferred an increased risk for ICU admission and intubation. The predictive score for COVID-19-related death included age 65 years or older, diabetes, early-onset diabetes, obesity, age less than 40 years, chronic kidney disease (CKD), and hypertension.
A number of previous studies have described a relation between diagnosis of COPD [23, 25] and SMI  with more adverse outcome following a COVID-19 infection. Although the final multiple regression model in our study did not find an independent influence of these factors, they remain important in terms of evaluating risk of any patient becoming seriously unwell following COVID-19 infection.
It should be pointed out that the advent of treatment of COVID-19 hospitalised patients with dexamethasone has significantly reduced mortality rates across the world. In hospitalised patients, use of dexamethasone resulted in significantly lower 28-day mortality in those who were receiving either invasive mechanical ventilation or oxygen alone at randomization . However, close monitoring of blood glucose is recommended in all patients treated with dexamethasone .
Additional agents are also now available for COVID-19-affected patients who become severely unwell  including tocilizumab, a recombinant humanised monoclonal antibody that inhibits binding of interleukin-6 (IL-6) to both membrane and soluble IL-6 receptors. These continue to reduce mortality rate in severely unwell patients infected with COVID-19.
Potential pathogenic links between the SARS-CoV-2 virus and diabetes include the influence of glucose homeostasis and potentially altered immune status on the progression of the viral infection once established [30,31,32]. COVID-19 infection aggravates inflammation and alters immune system responses, leading to difficulties in blood glucose control. COVID-19 infection also increases the risk of thromboembolism and is more likely to induce cardiopulmonary failure in patients with diabetes than in patients without diabetes . All of these mechanisms are now believed to contribute to the poor prognosis of some patients with pre-existing diabetes and a COVID-19 infection .
The reason why studies may show a difference in outcome for statin users vs non statin users may relate to differences in the level of risk in the groups studied. For example, the Coronado study  described a higher mortality rate in statin users whereas other studies have described a relative protective effect of statins in COVID-19-affected individuals with T2DM . There is no question that statin use is overall hugely beneficial as reported in a recent very large study on US veterans .
During the COVID-19 pandemic, the achievement of tight blood glucose control and targeted management of cardiovascular risk factors are important for patients with diabetes mellitus [37, 38]. Medications used for both T2DM and cardiovascular disease need to be adjusted accordingly for people at high risk of COVID-19 infection in order to optimize their risk profile .
We recognise that a limitation of our study is that not all COVID-19 tests were fed back to general practice surgeries and coded. Furthermore we used the cut point of death up to 28 days after a COVID-19 positive test as the definition of a COVID-19-related death.
There is thus an underestimate of the total number of COVID-19 positive test results. However there is no reason to suspect that this would affect individuals with diabetes vs individuals without diabetes differentially.
The risk of severe outcomes has varied over time as the vaccination programme in the UK took hold in during December 2020–January 2021, so our findings may not reflect current risks given impact of vaccination, natural immunity and new variants of concern. This is the subject of ongoing work.
The data only covers the Greater Manchester conurbation and we have relied on general practice record coded diagnoses. Nevertheless the cohort whose outcomes we have analysed covers all general practices in the culturally and ethnically diverse Greater Manchester conurbation and is therefore representative of the 2.82 million people who live there in relation to the consequences of a COVID-19 infection.