The calculated versus the measured glycosylated haemoglobin (HbA1c) levels in patients with type 2 diabetes mellitus

Abstract Background Diabetes mellitus (DM) is a chronic metabolic disorder that is increasing globally. It is associated with chronic complications that are more common among patients with poor glycaemic control. Glycosylated haemoglobin (HbA1c) is the gold standard for monitoring glycaemic control. Measurements of HbA1c are relatively expensive and not available in some remote areas of developing countries. Methods We conducted a cross‐sectional study to evaluate the agreement between the calculated and measured HbA1c levels. The equation to compute the calculated HbA1c also incorporated the fasting blood glucose (FBG) level and was as follows: HbA1c = 2.6 + 0.03 × FBG (mg/dl). Result We enrolled 290 patients with type 2 DM in this study. Of these, 204 (70.3%) were females and the mean (SD) age was 54.9 (12.8) years. The mean (SD) diabetes duration was 6.8 (5.5) years. There were 211 (72.8%) patients using oral hypoglycaemic agents, 62 (21.4%) were using insulin and 17 (5.9%) were using both insulin and oral hypoglycaemic agents. There was a borderline difference between the mean (SD) calculated and measured HbA1c levels (p = 0.054). There was a significant correlation between the calculated and measured HbA1c (r = 0.595, p < 0.001). However, there was no agreement between the calculated and measured HbA1c. The bias ±SD (limits of agreement) for calculated versus measured HbA1c was −1.008 ± 2.02% (−5.05, 2.032). Conclusion Despite the presence of a significant correlation between the calculated and measured HbA1c, the calculated level has shown an unacceptable agreement with the measured HbA1c.

estimated to jump to 47 million by 2045. 1 These data underscore the huge burden DM will present for the future in developing countries if no efforts are taken to control it. Individuals with DM are at a high risk of serious complications (in particular those with type 2 DM), such as heart disease, renal disease, stroke, atherosclerosis, peripheral neuropathy and blindness. 1 Monitoring glycaemic control among diabetic patients is of paramount importance to modify the prognosis and the outcome. 1,2 The glycated haemoglobin A 1c (HbA 1c ) and fasting blood glucose (FBG) levels are important tools for the diagnosis and monitoring of glycaemic control with a cut-off of 6.5% and 126 mg/dl, respectively. 1,3 The role for a mathematical formula to calculate HbA 1c has gained interest since it is a simple and cost-effective method. 4 Basically, glycated haemoglobin is a form of haemoglobin used primarily to estimate the average plasma glucose concentrations over prolonged periods of three to four months. 5 Hence, it is promoted for monitoring glycaemic control in patients with DM as well as a method for diagnosing DM. 6,7 In developing countries, where there are several resource-poor settings and HbA 1c test availability is a concern, calculations for HbA 1c may offer hope and help. 4 Moreover, the estimated HbA 1c can be a reasonable means to reduce the financial burden, particularly in developing countries. 4,5,8 In fact, the health expenditure in Africa is relatively low, with the mean health expenditure per person with diabetes in Africa being less than 400 U.S. dollars, compared to 6,800 U.S. dollars in North America and the Caribbean nations. 1 The prevalence of DM in Sudan was 22.1% in 2019, and it is estimated to reach 24.2% by 2045. 1 This is in accordance with a recently published study from Sudan. 9,10 Sudan is not exempt when it comes to having limited healthcare resources and facilities.
Most of the populations are not covered by a health insurance system, and there is a high cost for health services as well as limited availability for HbA 1c measurements in most rural areas. Moreover, some recent studies from Sudan have demonstrated a high prevalence of DM and uncontrolled DM. 9,10 Hence, we conducted this study to evaluate the agreement between the calculated and measured HbA 1c among patients with type 2 DM.

| Study design and ethic approval
This was cross-sectional study conducted between April and October 2019. Patients were recruited from the outpatient diabetic/ endocrine clinic at Gadarif Teaching Hospital. We recruited all patients with type 2 DM. This included both males and females, aged 18 years or older and those with haemoglobin levels between 12 and 16 g/dl because HbA 1c results can be affected by several factors, including anaemia. The exclusion criteria were patients with type 1 DM, haemoglobinopathies, functional thyroid disorders, patients with hypertension and on diuretics, renal disorders, anaemia (haemoglobin ˂12 g/dl), bedridden patients, patients with an advanced malignancy and pregnant women.
The data were collected using a standardized questionnaire that included demographic data (age and sex), comorbidities, weight, height, haemoglobin level, FBG, measured HbA 1c and calculated HbA 1c levels based on the adopted equation.
Universal safety precautions were adopted for collecting the blood samples. Sterile disposable needles and vacutainers were used to obtain the samples. Correct procedures were adopted at all steps and included the venepuncture site and the pressure used to transfer blood into the vacutainer to prevent haemolysis. After sign-
There was a borderline difference between the mean (SD) calculated and measured HbA 1c (p = 0.054) ( Table 2). There was a significant correlation between the calculated and measured HbA 1c (r = 0.595, p < 0.001). Alternatively, there was no agreement between the calculated and measured HbA 1c . The bias ±SD (limits of agreement) for the calculated versus measured HbA 1c was HbA 1c -1.008 ± 2.02% (₋5.05, 2.032) (Table 3, Figure 1).

| DISCUSS ION
The current study showed that, although there was a significant positive correlation between the calculated and measured HbA 1c , there was an unacceptable agreement between the measured and calculated HbA 1c . Thus, the non-agreement obtained in our results pointed to an unreliability of the calculated HbA 1c using the equation to assess glycaemic control. Likewise, one study has shown a statistically significant negative bias among patients with or without DM when the Bland-Altman plot was made. 5 Moreover, they reported a non-significant difference between the measured and calculated HbA 1c levels. 5 They further explained the significant negative bias in patients with DM to hyperglycaemia in uncontrolled DM. 5 Previously, Nayal et al. observed a significant difference between the erythrocyte HbA 1c levels and the calculated HbA 1c levels. 6 In Sudan, it has recently been shown that HbA 1c has a poor reliability, insufficient sensitivity or specificity for diagnosing gestational DM. 3 Previous studies have recommended the use of the calculated HbA 1c based on self-measured glucose and past HbA 1c values for assessing glycaemic control in patients with DM, in particular among those with good glycaemic control. 6,7 HbA 1c is subject to variations, irrespective of the glycaemic control. Some of the putative factors for these variations include age, sex hormones, visceral fat distributions, physiologic and genetic factors and socioeconomic status. 15,16 The significant correlation between the calculated and measured HbA 1c which was obtained in our study may be explained by the observation that higher HbA 1c levels were seen in patients with persistently elevated blood glucose levels, particularly in patients with uncontrolled DM. 5,17 Additionally, those with good glycaemic control had HbA 1c levels close to, or within, the reference range, which might provide identical values using the same mathematical formula (HbA 1c = 2.6 + 0.03 × FBG [mg/dl]). 5 Interestingly, another study adopting the same formula obtained a significant difference between the measured and calculated HbA 1c among patients with type 2 DM and a control group to assess glycaemic control. 4 ThFBSey found that HbA 1c values derived and predicted by the formula were in accordance with measured values using the high-performance liquid chromatography (HPLC) BIORAD method. 4 Additionally, some studies have used other equations to estimate HbA 1c that were helpful in estimating glycaemic control based on a significant correlational difference. 17,18 Unfortunately, the limits of agreement were not assessed to consider it clinically acceptable. Likewise, one study has shown a significant difference between the levels of measured and calculated HbA 1c based on the calculated fasting blood glucose (FBG) (HbA 1c = 2.6 + 0.03 × blood glucose [mg/dl]). 19 Interestingly, they found the calculated HbA 1c was not identical to the measured HbA 1c in erythrocytes. Therefore, they recommended its use in patients with well-controlled DM only. 19 Likewise, another study restricted its use among patients with good glycaemic control. 7 Thus, the mathematical formula cannot be used interchangeably with measured HbA 1c levels. 19 Using the calculated HbA 1c was justified by concern regarding checking TA B L E 1 General characteristic of the enrolled patients

TA B L E 3 Correlation, bias and limits
of agreement between the calculated and measured haemoglobin A 1c levels HbA 1c every three months may be premature to evaluate the effect of basal insulin if the fasting blood glucose has not achieved the goal for two to three months. 18 Moreover, HbA 1c reflects the last 120-day average, which lags behind the current improvement in glycaemic control. 18 In addition, the mathematical model used to calculate HbA 1c by utilizing measured fasting plasma glucose levels provides the ability to monitor intermittent HbA 1c levels between scheduled check-up visits in patients on anti-diabetic therapy. 4 It can be used like the estimated average glucose for chronic glycaemia and acute glycaemia, as well as offer patients a better understanding of current glycaemic control based on daily glucose measurements. 20 Our findings of a significant difference between the calculated and measured HbA 1c levels might be explained by the significant correlation between the FBG and HbA 1c , which has been documented in recent studies. [21][22][23][24] Furthermore, no sex differences were observed in a linear relation between glucose and HbA 1c among Greek male and female patients with DM. 25 Interestingly, FBG was strongly correlated in a group of Japanese patients with uncontrolled type 2 DM (HbA 1c > 8.0%). 26 This was in accordance with the observation that higher blood sugar levels predicted higher levels of HbA 1c among patients with a pre-diabetic range in Indonesia. 24 A significant correlation between HbA 1c and FBG levels was reported among patients with and without DM. 27 In Zambia, it has been observed that there was correlation between HbA 1c and FBG. 28 Hence, they recommended FBG as a suitable and alternative tool to assess glycaemic control in the absence of HbA 1c . 28 The availability of facilities to conduct regular checks for HbA 1c , cost issues and poor health insurance was among the factors in favour of using the calculated HbA 1c . 5

| Limitation of the study
Other factors that could have effect on HbA 1c such as ferritin level and inflammatory markers were not assessed. Other factors which were reported to influence the control of DM, for example vitamin D level were not investigated. 30

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no competing interests.

AUTH O R CO NTR I B UTI O N S
Imad R Musa and Ishag Adam involved in conceptualization. Saeed

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the current study are available.
From the corresponding author on reasonable request.