Demographics for the ED population are presented in Table 1. The mean age of the entire sample was 51 years, with the average age increasing with HbA1c. There was a slightly higher proportion of females to males in the sample, with 55% of the overall sample being female. Men made up a larger proportion of those with elevated HbA1c.
Table 1 – demographics by diabetes status.
|
No Diabetes
|
Pre-diabetes
|
Diabetes
|
All (n)
|
62,473 (52%)
|
36,751 (31%)
|
20,859 (17%)
|
Age*
|
42.3 (19.3)
|
59.1 (19.0)
|
64.2 (16.0)
|
HbA1c*
|
|
|
|
HbA1c (%)
|
5.3 (0.3)
|
6.0 (0.2)
|
8.0 (1.6)
|
HbA1c (mmol/mol)
|
34
|
42
|
64
|
Sex^
|
|
|
|
Female
|
36,321 (58.1%)
|
19,256 (52.4%)
|
10,151 (48.7%)
|
Male
|
26,152 (41.9%)
|
17,495 (47.6%)
|
10,708 (51.3%)
|
BMI (Kg/m^2)*
|
27.5 (7.2)
|
29.1 (8.0)
|
31.2 (8.8)
|
Aboriginality^
|
|
|
|
Aboriginal or Torres Strait Islander
|
3559 (5.7%)
|
1275 (3.5%)
|
863 (4.1%)
|
Neither Aboriginal nor Torres Strait Islander
|
58,914 (94.3%)
|
35,476 (96.5%)
|
19,996 (95.9%)
|
*mean with standard deviation in brackets ^ proportion of total. Diabetes/pre-diabetes defined using ADA criteria of >6.5% and 5.7-6.4% as above.
In the time period since the week beginning March 16th, the number of tests performed in these EDs has declined, from an average just below 642 per week to 556 per week. This appears to correspond to a decline in the rate of ED attendances, with roughly half the expected attendances per week in late March and April of 2020 compared to 2019. There have also been differences in the people attending ED, with fewer people over the age of 65, and a decline in the proportion who are female. This is shown in Table 2.
Table 2 – comparison of before/after COVID values for testing for the entire time period.
|
Pre-COVID
|
During COVID
|
Tests per week
|
642
|
556
|
Median age
|
50
|
47
|
% male
|
45.0
|
46.4
|
% over 65 years old
|
29.5
|
26.9
|
% under 40 years old
|
37.0
|
40.9
|
The primary findings are presented in Figure 1. During the time period from when NSW began legally enforcing isolation measures due to the epidemic of COVID-19, there was a significant decrease in the rate of people attending ED with elevated HbA1c consistent with diabetes. As can be seen in Figure 1, in most weeks this was below the 95% confidence interval from previous years. The rate has decreased from a mean varying between a low of 15% and a high of 25% to an average of just over 11% consistent with diabetes during this pandemic period.
Examining this rate by gender and age, there are some important findings. The reduction in the rate of tests consistent with diabetes appears to be largely driven by a decrease in the median age of people presenting. Once age-stratified, there appears to be some difference in diabetes percentages by age group. This is shown in Figure 2. There is also indication of a significant interaction with gender, with females in this sample showing a downward trend in high Hba1c results, however this trend was not apparent in the male population until the second wave in NSW. This is shown in Figure 3.
Comparing the pre-COVID-19 rates of diabetes with the pandemic rate using t-test gives a p-value <0.001, with the average rate of diabetes-consistent tests at 17.3% prior to COVID-19 and only 13.1% now.
These findings were not matched in the GP data. The mean rate of diabetes patient attendance in adults in the GP dataset prior to March in 2020 was 11.3%, which rose slightly to 11.8% during the pandemic period as seen in Figure 4. This then continued to rise throughout the pandemic, with the rate stabilizing at 12% towards the end of 2021. However, the average number of adults active in the system fell slightly from 633,228 to 609,936, a drop of 3.7%.
Moreover, there did not appear to be a greater rate of diabetes in hospitalized adults, with similar trends both before and during the pandemic.
During the pandemic, the proportion of services offered through VC rose sharply from an average of 9.1% prior to March 2020 to 76.8% after that time (p<0.0001), despite the average monthly number of consultations remaining steady during this time-period. This can be seen quite clearly in figure 5, which shows the stark divide between the proportion of services used before and during the pandemic. There was something of an initial lag during March/April 2020 as VC was being set up. Virtual Care had been temporarily halted at these clinics before March 2020 due to the setting up of a new service that was due to begin mid-2020 – the pandemic brought forward this new service substantially.
There was also a noticeable change in the proportion of services used when there were outbreaks compared to none. The first lockdown in NSW occurred at the point of the red line, with a second wave of restrictions coming through due to an outbreak in October 2020. These were relaxed in January 2020, at which point VC fell dramatically, and then reinstated during the 2021 lockdown in July 2021. Looking at the proportion of services provided through VC, where there were interventions in place to control outbreaks (April, May, June, November, December in 2020 and January, July, August, September, October in 2021) the average use of virtual care was 90.3% compared to 65.8% in non-outbreak months (p=0.0014). All services continued to be provided during the pandemic period, although the denominator increased modestly as a new clinic was started in May 2020.
Hospital services appeared to be unaffected, with similar numbers pre and during pandemic for admissions due to hypoglycaemia, retinopathy, and diabetic chronic kidney disease (p>0.05) in these hospitals.
Finally, there was an interesting trend whereby during periods where NSW experienced lockdowns, there were fewer patients who attended the ED with tests consistent with diabetes. Comparing the average rate of diabetes using a t-test between the times where the state was locked down and those it was not produces a p-value below 0.001.