Glucose control in diabetes during home confinement for the first pandemic wave of COVID-19: a meta-analysis of observational studies

Aim To assess the effect on glycaemic control of confinement due to lockdown measures, during COVID-19 pandemic, in people with type 1 (T1DM) and type 2 (T2DM) diabetes. Methods Meta-analysis of observational studies reporting measures of glucose control and variability before and during and/or after periods of confinement caused by COVID-19 in 2020 and/or 2021. Results We included 27 studies on T1DM. No significant change in Hba1c was observed after lockdown (WMD − 1.474 [− 3.26; 0.31] mmol/mol, I2 = 93.9). TIR significantly increased during and after lockdown (WMD: 2.73 1.47; 4.23 %, I2 = 81% and 3.73 [1.13; 5.33] %, I2 = 85%, respectively).We retrieved nine studies on T2DM patients. No significant variation in HbA1c was detected (WMD − 1.257 − 3.91; 1.39 mmol/mol, I2 = 98.3%). HbA1c had a more favourable trend in studies performed in Asia than in Europe (p = 0.022 between groups). Conclusion Lockdown showed no significant detrimental effect on HbA1c in either T1DM or T2DM. Conversely, home confinement led to a reduction in mean glucose and glucose variability in T1DM, although with a high heterogeneity of results. Supplementary Information The online version contains supplementary material available at 10.1007/s00592-021-01754-2.


Introduction
COVID-19 pandemic forced most Countries to adopt confinement measures to prevent the spreading of the disease. Those measures, although different across countries, all led to some extent to a reduction in physical activity, with the shutdown of gyms in many countries; moreover, the commitment to stay at home modified daily routine, increasing the time available for cooking and eating. All those changes in daily routine may have altered glycaemic control in people with diabetes mellitus [1].
The outbreak of COVID-19 during the spring of 2020 and the consequent lockdown in many countries also reduced the access to diabetes specialist care [2], metabolic monitoring through laboratory determinations [3], and visits and examinations for screening of diabetic complications. The reduced availability of medical care, associated with insufficient patient self-management [4], was a possible determinant of the observed increase in incidence and severity of diabetic complications, such as foot ulcers [5]. In order to maintain basic care, telemonitoring was implemented in many countries [6][7][8]. Telematic interactions were also used to help patients in developing coping strategies for managing home confinement [9][10][11] and maintaining physical activity [12,13]. Physical activity is crucial in diabetes mellitus management, especially during a pandemic, as it enhances immune response to viral infections [14]; therefore, many efforts have been performed to help patients in finding strategies to maintain it during lockdown [15].
The success of strategies implemented for the care of diabetes during prolonged lockdown for COVID-19 epidemic waves was assessed in several observational studies, providing discordant results [16][17][18]. This meta-analysis is aimed at collecting all evidence on the effect on glycaemic control of confinement due to lockdown measures, and the consequent adaptation of care, during the first wave of COVID-19 pandemic, in patients with type 1 and type 2 diabetes.
Managed by Antonio Secchi.

Materials and methods
Review Protocol has been submitted for registration to the PROSPERO website CRD42021234360https:// www. crd. york. ac. uk/ PROSP ERO/. Searches were performed in Pub-Med and Embase ("COVID-19" and "diabetes mellitus") up to March 10, 2021 (see the complete search strings in Supplementary Table 1). Further studies were searched among references from papers.
Observational studies written in English language and performed on humans, enrolling patients with type 1 or type 2 diabetes, and reporting measures of glucose control and variability before and during and/or after periods of confinement caused by COVID-19 in 2021 and/or 2021 were included.
The principal endpoints were variations from baseline in glycated haemoglobin (HbA1c) and Time in Range [19] (TIR, time during which glycaemia is maintained between 70 and 180 mg/dl) during and after lockdown. Additional outcomes were Time Above Range (TAR, time during which glycaemia is above 180 mg/dl), Time Below Range (TBR, time during which glycaemia is below 70 mg/dl), mean glucose, glucose variability (coefficient of variation), frequency of glucose monitoring, variations in eating habits and physical activity, perceived stress.
The following data were extracted: number of included patients, duration of diabetes, mean age, proportion of male patients, patients using flash glucose monitoring (FGM), continuous glucose monitoring (CGM), or self-monitoring of blood glucose (SMBG); proportion of patients in continuous subcutaneous insulin infusion (CSII) multiple daily insulin injections (MDI), basal insulin only, sodium glucose transporter 2 inhibitors (SGLT2i), dipeptiydil-4 inhibitors (DPP-4i), pioglitazone, metformin, sulphonylureas (SU); study duration, country of origin, duration of Lockdown, use of teleconsulting, values of HbA1c before and after lockdown, TIR,TAR, TBR), Coefficient of variability (CV), use of telemedicine, any variation in physical activity, diet, stress. The quality of the studies was assessed at study level, using the scale developed by Carmen-Moga and colleagues (Table 2S).
Titles and abstracts were screened independently by two authors. If one or more inclusion criteria were present, the whole article was read, in order to assess if all the inclusion criteria were present. Study selection, data retrieval and study quality assessment were performed independently by two investigators (C.D.P. and G.A.S.) and conflicts resolved by a third investigator (M.M.).
Begg's and Mandzumkar test were used to detect publication bias, with reference to all principal endpoints; funnel plots were used when more than 10 studies were available for the specific outcome. Weighed mean differences (WMD) during and after lockdown vs. pre-lockdown), with 95% confidence intervals, were calculated using random effect models. Rosenthal's conservative estimate of 0.7 [20] was adopted for pre-post correlation. Fixed effect models were used for sensitivity analysis. I 2 statistics was used for the assessment of heterogeneity. Subgroup analyses were performed, based on country, age group (children and adolescent < 18 years, adult > 18 years), type of monitoring (FGM, CGM, SMBG), insulin treatment (multiple injections, continuous subcutaneous infusion, hybrid closed-loop systems), structured telemonitoring (yes/no).
All the analyses were performed on Comprehensive Meta-Analysis Software, V3 edition, Biostat Inc. 14 North Dean Street, Englewood, NJ 07,631, USA.

Results
Out of 1634 results, 122 studies were selected on the basis of the titles and abstracts. Of those, 79 did not report data on glycaemic control during or after lockdown measures; 6 included both T1DM and T2DM patients without providing subgroup analysis [17]; two studies reported subgroup analyses with no overall analysis [21]. Thirty-six studies reported glycaemic control before and during or after the pandemic restrictions and were therefore included in the meta-analysis (Fig. 1S). Of those, 9 were performed in type 2 diabetes, whereas 27 were performed in type 1 diabetes. Characteristics of the included studies are reported in Table 1.

Time below range
Seventeen studies on type 1 diabetes estimated time below range (TBR) before and during lockdown, whereas nine studies reported TBR before and after lockdown: TBR did

Mean glucose
In the 14 studies with available data, mean glucose during lockdown was significantly lower than before lockdown

Patients' reported behaviours
Thirteen studies enrolling patients with T1DM reported data on patients' behaviours. The heterogeneity of instruments used for the assessment of patients' behaviour prevented a formal meta-analysis (Table 2). A reduction in physical activity was reported by 8-70% of patients, whereas those reporting an increase in food intake were 17-46%; moderateto high stress was found in 20-52% of patients.

Type 2 diabetes
Nine studies reporting HbA1c before and after lockdown were available in T2DM patients, including 9591 subjects with a median age of 60.5 years; five studies were performed in Asia (India, South Korea, Japan, Saudi Arabia), whereas four were performed in Europe (Turkey, Italy, Greece). No significant variation in HbA1c was detected (WMD − 1.257 [− 3.91; 1.39] mmol/mol; Fig. 1, panel B), with high heterogeneity (I 2 = 98,3%). No significant publication bias was detected (Kendall's tau = − 1, p = 0.88). A subgroups analysis revealed a significant difference between studies with mean baseline HbA1c below or above 64 mmol/mol (p = 0.045 between groups), with those with higher baseline HbA1c showing a greater reduction (Fig. 17S). A further subgroup analysis showed that HbA1c had a more favourable trend in studies performed in Asia than in Europe (p = 0.022 between groups) (Fig. 18S). No difference was found between age groups (p = 0.22 between studies with a mean age higher or lower than 60 years) (Fig. 19S).
Two studies on people with T2DM [22,23] both performed in India reported a modest reduction (20-24% of participants) in physical activity with no significant variation in food intake. On the other hand, two studies performed in Turkey and Japan [24,25] reported a frequent (54-70% of participants) reduction in physical activity and an increase in food intake (20-55% of participants). All the three studies reporting data on stress found a moderate increase in perceived stress and anxiety ( Table 2).

Discussion
Most studies on the glycaemic effects of lockdown were performed in T1DM, usually in patients using either FGM or CGM. The assessment of variations in interstitial glucose can be performed in a shorter time than that required for exploring modifications in HbA1c. For this reason, we already have a substantial body of evidence on the effects of lockdown during the first pandemic wave in T1DM, but not in T2DM.
In T1DM, an improvement in glycaemic control during lockdown was observed, together with a reduction in glucose variability. These results were obtained despite an observed reduction in physical activity and dietary compliance during lockdown, determined by the increased time spent at home; in addition, access to care was impaired during lockdown [26], and surveys on perceived glucose control revealed a high perceived difficulty in dealing with COVID-19 restrictions in people with T1DM [27].
The interpretation of results on glucose control in T1DM is problematic because of their high heterogeneity. The exploration of moderators of lockdown effect, using either subgroup analyses of trials or meta-regression, provides some further insight. Lockdown seems to have produced a greater beneficial effect on females than in males. This is consistent with a Chinese study in which males with T1DM had poorer glycaemic control than females during COVID-19 lockdown [28]; previous findings showed that females with DM, when compared to males, elicited more frequently behaviours aimed at disease prevention, health promotion and symptom recognition [29], which could have been of help in coping with confinement. An increase in glucose monitoring and an improvement in insulin titration during remote working or remote schooling may explain these improvement, as suggested by a study showing an improvement in glucose control only in patients working at home [21]; unfortunately, the information on glucose monitoring and on the proportion of patients on home schooling or working was insufficient to add these variables as moderators. Notably, in studies enrolling a majority of patients with T1DM with interstitial glucose monitoring systems, HbA1c was significantly reduced, suggesting that FGM or CGM could have been a relevant support during confinement. The difference in effects of lockdown between Asian and European studies could have been determined by the different proportion of patients on FGM/CGM (substantially higher in European studies), or to differences in lockdown measures (usually stricter in European countries).
No significant beneficial or detrimental effect of lockdown on glucose control was found in Type 2 diabetes, although the limited number and high heterogeneity of available studies suggests caution in drawing definitive conclusions. In patients with T2DM, studies performed in Asia showed a significant reduction in HbA1c, which was not observed in European studies. This geographic difference could have been determined by the differences in confinement measures (much stricter in Europe than in some Asian countries such as South Korea [30]), or by cultural differences possibly affecting the effect of lockdown on diet and physical activity; in studies performed in India, for example, lockdown appeared to have only a minor effect on physical activity [22,23]. Some limitations of the present meta-analysis should be recognized. Centers performing the studies were often thirdlevel clinics, which are not representative of all diabetes care facilities, because of a possible wider use of telemedicine, continuous glucose monitoring and more advanced treatments. Most of the studies performed in T1DM patients, furthermore, only enrolled patients which had performed at least 70% scans, thus excluding less compliant patients, who may be at higher risk of glucose deterioration. In addition, in our metanalysis, the mean age of the included patients with T2DM was low; accordingly, a survey has shown that patients contacted for telemedicine by a diabetes clinic during the pandemic were younger, with shorter disease duration and a lower prevalence of complications than the average pre-lockdown patients [31].
In conclusion, lockdown showed no significant detrimental effect on HbA1c in either T1DM or T2DM. Conversely, home confinement during the first pandemic wave led to a reduction in mean glucose and glucose variability in T1DM, although further studies are needed to better understand the high heterogeneity of results.
Author contribution GAS and EM were involved in each of the following points: 1. Design. 2. Data collection. 3. Analysis. 4. Writing manuscript. CDP, ID and MM were involved in each of the following points: 1. Data Collection 2. Manuscript revision.
Funding Open access funding provided by Università degli Studi di Firenze within the CRUI-CARE Agreement. This research was performed as a part of the institutional activity of the unit, with no specific funding. All expenses, including salaries of the investigators, were covered by public research funds assigned to the unit. The manuscript was drafted and revised by the authors in accordance with ICJME standards for authorship. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Declarations
Conflict of interest GAS and ID have received speaking fees from Astra Zeneca, Novonordisk, Eli Lilly outside the submitted work; CDP has no conflict of interest; MM has received speaking fees from Astra Zeneca, Bristol Myers Squibb, Boehringer-Ingelheim, Eli-Lilly, Merck, Novo Nordisk, Sanofi and Novartis and research grants from Bristol Myers Squibb outside the submitted work; EM has received consultancy fees from Merck and Novartis speaking fees from Astra Zeneca, Bristol Myers Squibb, Boehringer-Ingelheim, Eli-Lilly, Merck, Novo Nordisk, Sanofi and Novartis and research grants from Merck, Novartis and Takeda outside the submitted work. All the authors approved the final version of this manuscript. Dr. Giovanni Antonio Silverii is the person who takes full responsibility for the work as a whole, including the study design, access to data and the decision to submit and publish the manuscript.
Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors.
Informed consent For this type of study formal consent is not required.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.