Abstract

Background

An increased risk of diabetes mellitus (DM) after COVID-19 has been reported in the United States, Europe, and Asia. The burden of COVID-related DM has yet to be described in Africa, where the overall risk of DM has been increasing rapidly. Our objective was to compare the prevalence of pre-DM and DM in Nigerian individuals with a history of COVID-19 to individuals without known COVID-19 infection.

Methods

We undertook a retrospective cohort study with 256 individuals with a past medical history of COVID-19 with no history of pre-DM or DM and 256 individuals without a history of COVID-19 or pre-DM/DM. Participants were categorized as pre-DM (fasting capillary glucose 100–125Ìýmg/dL) or DM (fasting capillary glucose ≥ 126Ìýmg/dL). We employed univariate and multivariable logistic regression to identify key predictors and adjust for confounders related to hyperglycaemia risk factors. Additionally, we used multinomial logistic regression to analyze the relationship between COVID-19 history and diabetes status, distinguishing between normal, pre-diabetic, and diabetic glucose levels. All models were adjusted for age, gender, hypertension, physical activity, central adiposity, and family history of DM.

Results

Compared to the control group, those with a history of COVID-19 had a similar median age (38 vs. 40 years, p = 0.84), had a higher proportion of men (63% vs. 49%), and had a lower prevalence of central adiposity (waist: hip ratio ≥ 0.90 for males and WHR ≥ 0.85 for females) (48% vs. 56.3%, p = 0.06). Of the 256 with a history of COVID-19, 44 (17%) required in-patient care. The median (interquartile range) time interval between COVID-19 diagnosis and the glycaemic assessment was 19 (IQR: 14, 24) months. Pre-DM prevalence was 27% in the post-COVID-19 group and 4% in the control group, whereas the prevalence of DM was 7% in the post-COVID-19 group and 2% in the control group. After multivariable adjustment, the odds of pre-DM were 8.12 (95% confidence interval (CI): 3.98, 16.58; p < 0.001) higher, and the odds of DM were 3.97 (95% CI: 1.16, 13.63) higher in those with a history of COVID-19 compared to controls. In the adjusted multinomial logistic regression analysis, individuals with a history of COVID-19 exhibited significantly elevated risks for pre-diabetes (RRR = 7.55, 95% CI: 3.76–15.17) and diabetes (RRR = 3.44, 95% CI: 1.01–11.71) compared to those without COVID-19.

Conclusion

Previous COVID-19 was found to be a risk factor for prevalent pre-diabetes and diabetes mellitus in Nigeria. More intensive screening for DM in those with a history of COVID-19 should be considered.

Background

In addition to pneumonia, COVID-19 is associated with multi-systemic complications, such as coagulopathy, myocarditis, and acute kidney injury [1]. New onset diabetes mellitus (DM) after COVID-19 has also been reported worldwide, predominantly in Asia, Europe, and the US [2, 3] In a systematic review and metanalysis conducted in 2022, it was reported that males who had contracted COVID-19 had a two-fold risk of developing DM when compared to those without a history of COVID-19 [2]. Furthermore, in a related study, the incidence of diabetes post-COVID-19 was estimated to be 15.53 per 1000 person-years [2]. Moreover, the risk of developing type 2 diabetes following COVID-19 was significantly higher, with a relative risk (RR) of 1.70 (95% confidence interval (CI):1.32–2.19), when compared to individuals without a history of COVID-19 [2]. A Korean study also showed that the risk of diabetes in the post-acute phase following COVID-19 was increased when compared to those without COVID-19 (adjusted hazard ratio [HR], 1.30; 95% CI, 1.27 to 1.33) [4], while an Italian-based study found an increased risk (1.6%) of onset DM following COVID-19 [5]. Incident risk is highest in the first six months post-COVID-19 infection [6].

The etiological mechanisms contributing to post-COVID-19 DM have not been fully elucidated. Potential mechanisms include a direct impact on pancreatic beta cells, which may compromise insulin secretion [7, 8]. Other factors, such as steroid therapy in the management of COVID-19 infection, have also been implicated in the pathogenesis of post-COVID-19 diabetes [3, 7,8,9,10,11,12,13,14,15,16].

Although the impact of COVID-19 pandemic in Africa has been less pronounced compared to regions, Nigeria, like many other African countries, has faced significant infection rates, resulting in near-term and long-term morbidity and mortality. The data regarding the impact of COVID-19 on DM within the African context is notably limited. The ongoing COVID-19 pandemic may add to the rising burden of DM in Africa, which is already at an epidemic level [10]. Prior to the pandemic, the national pooled prevalence of DM in Nigeria was already a concern at 5.7% [17]. Remarkably, despite an overall decrease to 4.3% in the national pooled prevalence of DM post-COVID-19 [18], the prevalence of pre-diabetes in Nigeria remained high at 13.2% [19]. This study aims to retrospectively examines the comparative prevalence of diabetes and pre-diabetes in individuals with and without a history of COVID-19 in Nigeria, thereby investigating the evolving hypothesis that there is an epidemiological link between COVID-19 and abnormal glucose metabolism.

Methods

Study design and population

The study design comprised a retrospective cohort study of persons with a history of COVID-19 that were identified through [1] the database of the state’s epidemiology unit of the Ministry of Health, [2] the infectious disease and isolation centres in-patient database, and 3) identification of individuals who previously tested positive to COVID-19 conducted by the local disease surveillance notification officers (DSNOs), who are domiciled in the various communities across the two States in northeastern Nigeria (Adamawa and Gombe). The latter group was traced by these DSNOs, who tracked them to their various communities and enlisted them after obtaining informed consent. Conventionally, the DSNOs are trained staff who are responsible for tracking individuals suspected to have been infected in the event of any epidemic outbreak and also obtaining specimens for diagnosis.

The diagnosis of COVID-19 was previously established in this cohort using polymerase chain reaction (PCR) for hospitalized patients and urban dwellers. Rapid antigen test kits were used for non-hospitalized patients and rural dwellers, and were administered by DSNOs.

We excluded those below 18 or over 75 years, pregnant women, and those who self-reported having diabetes or pre-diabetes before the COVID-19 pandemic.

Individuals without a history of COVID-19 and who did not have a self-reported history of diabetes or pre-diabetes served as a control group. However, there was no record to ascertain their prior glycaemic status. We recruited this cohort during a free medical outreach organized in the community solely to recruit the control, during which capillary glucose testing and blood pressure measurements were offered free of charge. Those who self-reported having diabetes or pre-diabetes were excluded from the study. Additional control participants were recruited amongst individuals attending the healthcare facilities either on an outpatient basis to the general outpatient department or as a patient’s relative. Again, those who self-reported being diabetic or pre-diabetic were similarly excluded.

Ethics approval

The ethical approval for this study was granted by the institutional research board of the Modibbo Adama University Teaching Hospital Yola Adamawa State, Nigeria, with approval number MAUTHYOLA/HREC/22/237. All participants gave informed consent to participate in this study.

Data collection

Data were collected from December 2022 to April 2023. Trained DSNOs administered the questionnaire by tracking the COVID-19-positive cohort in their communities. A trained nurse enrolled healthcare workers who met the eligibility criteria in their various healthcare facilities. We collected data on sociodemographic information, behavioural risk factors for DM, personal history of hypertension, and family history of hypertension and DM. Glucose was tested using capillary blood (Accucheck) in participants after 8Ìýh of fasting, and their waist-hip ratio was taken. Blood pressure was measured by adhering to the WHO standardized procedure. Blood pressure was assessed with two separate measurements 30Ìýmin apart after resting for at least 5Ìýmin in accordance with WHO standardized procedures [20]. Waist circumference was measured using a flexible tape at the level of the midpoint between the ribs and the iliac crest from the front after exhalation [21]. Hip circumference was measured using a flexible tape at the point where the buttocks extend the most when viewed from the side [21].

We targeted a sample size of 256 in each group to detect a difference in diabetes prevalence with an alpha of 0.05 with a power of 80%, assuming a diabetes prevalence in the COVID-19 group of 20.8% [11].

The questionnaire is sectioned into socio-demographics and metabolic risk profiles, including physical activity status, the family history of DM or hypertension, smoking status, and glycaemic status. The Anthropometrics section compromised the waist-hip ratio and blood pressure. Besides the outcomes of the COVID-19 tests, the questionnaire had provisions for the type of care offered, such as in-patient or outpatient care, duration of admission, administration of steroids, and oxygen therapy.

Study outcome

The primary outcome was the prevalence of diabetes and pre-diabetes, defined using criteria established by the American Diabetes Association [2]. Diabetes was defined as a fasting glucose concentration ≥ 7.0Ìýmg/dL and pre-diabetes was defined as a fasting glucose concentration of 5.6–6.9 mmol/l. Capillary blood was used for glycaemic status assessment [22].

Covariates

Hypertension was defined as either a systolic blood pressure ≥ 140 or diastolic ≥ 90 or both or self-reported hypertension on treatment. Central obesity was defined using the adult waist-hip ratio (WHR), male WHR ≥ 0.90, and female WHR ≥ 0.85 [23]. Physical activity was measured using World Health Organization criteria for physical inactivity [24]. Physical inactivity was defined as the failure to meet WHO recommendations on physical activity for health, which are defined as engaging in at least 150Ìýmin of moderate-intensity activity per week or 75Ìýmin of vigorous-intensity activity per week, through any combination of walking and moderate or vigorous-intensity activities.

Statistical analysis

All statistical analyses were performed using Stata software (version 17.0) and R version 4.0.3. Data were presented as frequencies and percentages for categorical variables. The prevalence of diabetes and pre-diabetes with Clopper-Pearson 95% confidence interval was presented for individuals with and without a history of COVID-19. We used univariate and multivariable logistic regression models for the associated risk factors of hyperglycaemia (diabetes or pre-diabetes) to identify significant predictors and adjust for potential confounders. We assessed the relationship between those with and without a history of COVID-19 and diabetes (normal blood glucose, pre-diabetic and diabetic) using multinomial logistic regression. Normal blood glucose was used as the reference. To control for potential confounding variables, all models were adjusted for age (< 50 vs. ≥50 years), gender (male vs. female), hypertension (yes vs. no), family history of DM (yes vs. no), physical activity (active vs. sedentary), and waist-hip-ratio (normal vs. central obesity).

Results

FigureÌý1 describes the flowchart for participant recruitment. Initially, 294 potential post-COVID participants were identified; 31 with self-reported diabetes were excluded, and seven declined consent, leaving 256 participants that were included in this study. Of these, 44 were hospitalized, with 16 testing their baseline blood sugar levels at diagnosis; 4 of these developed new-onset diabetes, while the rest were euglycemic. The remaining 240 did not have baseline blood sugar checked. For the control cohort, 300 potential participants were identified, but 44 with self-reported diabetes were excluded.

Fig. 1

Flow chart of participant recruitment and outcomes

Participant characteristics for the two groups are given in TableÌý1. Among those with a history of COVID-19, a higher percentage were male compared to those without a history of COVID-19 (62.5% vs. 49.4%. p = 0.002). Significant differences in marital status between the two groups were observed. For instance, there were nearly twice as many single individuals in the COVID-19 cohort compared to the control group (26.5% vs. 14.5%). Also, while the majority were married in both cohorts, there were slightly more so in the control group (69.9% vs. 76.2%). Among those with a history of COVID-19, more were likely to have a tertiary education compared to the control group. The occupations varied significantly between the two groups (P < 0.001), with the COVID-19 cohort having more civil servants (27.3% vs. 8.2%) and healthcare workers (16% vs. 0%).

Table 1 Baseline demographic characteristics of study participants

We observed significant differences in the prevalence of diabetes in the COVID-19, and the control groups were 4.6% (0.046, 95% CI: 0.024–0.08) and 1.6% (0.016, 95% CI: 0.004–0.039), respectively (Fig.Ìý2). The prevalence of pre-diabetes in the COVID-19 and the control cohorts were 26.6% (0.266, 95% CI: 0.213–0.324) and 4.3% (0.043, 95% CI: 0.022–0.075), respectively. Results from univariate and multivariable logistic regression models for the associated risk factors of hyperglycaemia (diabetes or pre-diabetes) (Supplementary Table S1A-B) show a substantially increased likelihood of hyperglycaemia (fasting glucose 100Ìýmg/dL) among individuals with a past COVID-19 infection compared to those without such a history, in both univariate (OR = 7.45, 95% CI: 4.13–13.41, p < 0.0001) and multivariable models (aOR = 7.15, 95% CI: 3.79–13.50, p < 0.0001). Furthermore, hypertension (aOR = 2.07, 95% CI: 1.12–3.82, p = 0.021) and physical inactivity (aOR = 3.17, 95% CI: 1.81–5.57, p < 0.0001) were significant risk factors for diabetes after excluding those who were exposed to steroid. Other risk factors such as age, gender, family history of diabetes, and central obesity were not associated with prevalent diabetes in this cohort.

Fig. 2

Prevalence of hyperglyceamia between those with a history of COVID-19 COVID (+) and those without COVID (-)

In the multivariable multinomial logistic regression models (TableÌý2), the risk of pre-diabetes (vs. normoglycemia) was about eight-fold higher in those with a history of COVID-19 (RRR = 7.55, 95% CI: 3.76–15.17, p < 0.0001) compared to those without a history of COVID-19. Similarly, the risk of diabetes (vs. normoglycemia) was three-fold higher in those with a history of COVID-19 3.97 (RRR = 3.44, 95% CI: 1.01–11.71, p = 0.028) compared to those without a history of COVID-19. Age and gender were not a significant predictor of having pre-diabetes or diabetes. Hypertension was also not significantly associated with having pre-diabetes but was associated with having diabetes (RRR = 4.63, 95% CI: 1.45–14.84, p = 0.01). Individuals with low physical activity had higher odds of having pre-diabetes (RRR = 2.87, 95% CI: 1.57–5.24, p = 0.001) and diabetes (RRR = 4.86, 95% CI: 1.43–16.56, p = 0.011) compared to active individuals. Results from sensitivity analyses excluding the 16 participants who received steroids during admission for COVID-19 were similar (Supplementary Table S2).

Table 2 Multinomial logistic regression of the risk of diabetes and pre-diabetes in individuals with and without a past history of COVID-19 (inclusive of those treated with steroids)

Discussion

To the best of our knowledge, this is the first study to examine the prevalence of DM and pre-diabetes in those with and without a history of COVID-19 in Africa, an area that has one of the fastest-growing populations with DM. We showed that the prevalence of DM and pre-DM in persons with a known history of COVID-19 was higher than in those without a history of COVID-19. Our findings suggest that COVID-19 may be a risk factor for pre-diabetes and diabetes in this population.

In our study based in Northeastern Nigeria, we found the prevalence of diabetes mellitus to be higher among those with a history of COVID-19 compared to participants without a known history of COVID-19. The observed prevalence in those with a history of COVID-19 was higher than the subregional (northeastern Nigeria) prevalence of DM of 3.8% (95% CI: 2.7–4.7) [12]. Previous studies have also shown a higher risk of diabetes-related to COVID-19 in other areas of the world. In a recent Canadian study, the risk of diabetes in individuals with a history of COVID-19 was 0.5%, as opposed to 0.4% in those without COVID-19 [25]. The study further revealed that COVID-19 accounted for 3.41% of DM in the studied population. In their systematic review and metanalysis, Jiajun et al [26] revealed that a history of COVID-19 increased the risk of DM and hyperglycaemia by 1.7-fold compared to those without a history of COVID-19. In the Indian experience, new-onset diabetes among individuals with a history of COVID-19 was observed in 16.7% three months post COVID-19 [13]. Though this study implicated other factors such as age, adiposity, and family history of diabetes in post COVID-19 DM, the observed prevalence is higher than the national prevalence of DM in India of 9.3% [27] At variance with our study, a Chinese-based study reported a 10.3% prevalence of diabetes among COVID-19 patients, which showed almost no difference in the prevalence of 10.9% in the general Chinese population [14]. Additionally, Rathmann et al. [28] reported a higher incidence of diabetes in those with a history of COVID-19 than in those with other types of common acute viral upper respiratory infections (15.8 vs. 12.3 per 1000 person-years) in Germany [28].

The high prevalence of mild COVID-19 could account for the relatively lower prevalence of diabetes in our study. Studies have shown that the incidence of post COVID-19 DM increases with the severity of COVID-19 [16]. According to a systematic review, the global prevalence of pre-diabetes is 5.8% [29], while the pooled national prevalence of pre-diabetes in Nigeria was estimated at 13.2% [19]. Given that pre-diabetes is a major risk factor for the development of DM, identification of this population may have important public health consequences.

It is thought that COVID-19 triggers the pathogenic process of DM by first initiating insulin resistance, then pre-diabetes, and finally frank diabetes [7]. In addition, other pathophysiologic mechanisms are heightened inflammatory state, disruption of the angiotensin-converting enzyme ACEI/ACE2 balance, and subsequent dysfunction of the renin-angiotensin aldosterone-system RAAS [30]. Of note is the direct pancreatic beta cell destruction by the SARS-COV2 [30]. The extent to which this population is at risk of the development of diabetes mellitus deserves further study.

The relationship between COVID-19 and diabetes mellitus may be bidirectional [7, 14, 31]. Hence, it is possible that the cases of previously undiagnosed diabetes are being recognized during a clinical encounter related to COVID-19. This possibility was suggested by a US-based study which showed that persons with COVID-19 were 40% more likely to be diagnosed with diabetes mellitus compared to those without COVID-19 [32]. In our study, although participants with a history of COVID-19 reported no pre-existing diabetes before the study assessment, it is possible that hyperglycaemia was present at the time of COVID-19 diagnosis but was not clinically apparent.

For severe COVID-19, glucocorticoids are indicated for lung disease, which may, in turn, increase the risk of diabetes mellitus [13, 33, 34]. For example, in an Indian study, of the 31 individuals who had been treated with steroids while on admission for COVID-19, eleven (35.5%) went on to develop new-onset diabetes three months later [13]. Our study results are unlikely to be related to glucocorticoid use. In a sensitivity analysis, excluding those who reported glucocorticoid use as a treatment for COVID-19, there is a higher odds of prevalent diabetes and pre-diabetes among those with COVID-19 compared to the control group persisted [33].

Inpatient care, which connotes severe COVID-19 disease, has been associated with a higher risk of post-COVID-19 DM and, consequently, a higher burden of DM [35]. This study had only a 17% admission rate, which may possibly explain the relatively lower prevalence of post-COVID-19 DM observed. This finding of the high prevalence of pre-DM in individuals with history of COVID-19 speaks to the public health policymakers to enact a policy framework that would lead to regular monitoring of the glycaemic status of this cohort. In addition, the policy should provide them with health education on the need for lifestyle modification and behavioural changes to reduce the risk for DM in the future, thereby decreasing the burden of DM.

We acknowledge the following limitations in our study. First, most of our participants did not have a glucose determination at the time of or before their COVID-19 diagnosis. We, therefore, are unable to fully exclude persons with pre-existing diabetes. Second, we used capillary glucose for diabetes determination, which, while suitable for epidemiologic studies, may be slightly different than plasma glucose. An additional assessment with HbA1c could also have been useful to better categorize glycaemic status. Similarly, we did not carry out COVID-19. antibody testing for the control group. Being predominantly retrospective in design, it is not shielded from the inherent recall bias of this design. Lastly, the two cohorts differed on important characteristics which we attempted to balance with multivariable adjustment.

Recruiting participants through healthcare facilities could have introduced some selection bias. This is because some individuals who might have had asymptomatic or mild COVID-19 and never presented to the healthcare facilities for testing would not have been identified as having COVID-19. In addition, bias could have been introduced by virtue of socioeconomic status. People of low socioeconomic status are less likely to seek healthcare than those with higher socioeconomic status. In this regard, some participants of low socioeconomic status with characteristic symptoms of COVID-19 were unable to receive diagnostic testing due to their financial constraints. Furthermore, causal inference is limited in a cross-section study and longitudinal studies are needed to establish temporality and provide added evidence of a causal relationship between COVID-19 and DM. Finally, we did not screen for other viruses, such as hepatitis C viral infection, which is endemic and has been associated with DM.

Conclusion

In conclusion, our study suggests that a history of COVID-19 may be a risk factor for diabetes and pre-diabetes in Nigeria. If confirmed in longitudinal studies, more aggressive screening for DM may be warranted among those who have a history of COVID-19. Further studies are also needed to determine the risk of transition to DM in those with a history of COVID-19 with pre-diabetes and what interventions can be implemented to decrease this risk.