The glycosylated hemoglobin (HbA1c) assay is used worldwide as a point of care test to diagnose diabetes and to predict diabetic complications. ^{1– 3} It is formed by an enzymatic glycation reaction between glucose and the amino group of the terminal valine in the beta chain of adult hemoglobin. The amount of HbA1c in a particular individual depends mainly on the glucose levels and is also postulated to depend on red cell life span, ethnicity, oxidative stress, type of hemoglobin and the principles or methods used to determine its levels in blood. ^{4– 9}

Iron deficiency anaemia (IDA) is one of the most common causes of nutritional anaemia. Diabetics are at increased risk of iron deficiency due to chronic inflammatory states, diabetic kidney disease and as a complication of gastrointestinal bleeding induced by antiplatelets which are commonly prescribed in these group of patients for various indications. ¹⁰ Iron deficiency increases HbA1c levels independent of glucose levels by increasing red cell life span along with decreasing hemoglobin concentration causing an increase in the proportion of glycated hemoglobin. Additionally, iron deficiency induced oxidative stress may further enhance hemoglobin glycation. ^{4, 11, 12} As diabetes and iron deficiency can both affect HbA1c levels, we attempted to study the influence of iron deficiency on HbA1c levels in the absence of diabetes mellitus.

This cross-sectional observational study was conducted at the Kasturba Hospital Manipal, India. Subjects for the study were sourced from the medical records of KMC Manipal, after clearance from the institutional ethics council who were admitted as in patients of the Department of Medicine over the three years from August 2018 to August 2021. Data were collected and analyzed during the study period as stipulated by the IEC. Eighty-two nondiabetic patients with iron deficiency were identified as cases and 82 nondiabetic patients without iron deficiency were identified as controls. The population in both the groups were matched for age and sex.

Age and sex were matched for cases and controls. The median ferritin levels were 5.9 ng/dl (n=82) and 144 ng/dl (n=82) among controls. Transferrin saturation was distributed among cases, with a median of 3.4% for males (n=35) and 3.2% for females (n=45) ( Table 1).

Median hemoglobin levels were compared among each age group, with more severe anemia being seen as age increased (7.7 g/dL and 13.7 g/dL for cases and controls in the 20-44 year category; 6.2 g/dL and 13.65 g/dL for cases and controls in the 45-64 year category; 5.6 g/dL and 13.85 g/dL for cases and controls in the ≥65 year category).

HbA1C was also compared age wise, with results being statistically significant across each category as well (median A1C in the 20-44 year category being 5.4% for cases vs 5.2% for controls, n=36, p≤0.0001; for the 45-64 year category being 5.75% for cases vs 5.3% for controls, n= 30, p≤0.0001; and in the ≥65 year category having 6.3% for cases vs 5.5% for controls, n=16, p=0.0003) ( Figure 1 & Figure 2).

In our study patients with anemia and iron deficiency showed a statistically significant negative correlation between HbA1c and haemoglobin levels (Spearman’s correlation coefficient (-0.285) (p=0.009)).

Gender wise, there was a statistically negative correlation between HbA1C and hemoglobin in males (spearman’s correlation coefficient (-0.346) (p=0.039)) whereas in females there was a negative correlation but it was not statistically significant (Spearman’s correlation coefficient (-0.210) (p=0.161)). There was negative but not statistically significant correlation observed between HbA1c level and and serum ferritin level (n=82) (Spearman’s correlation coefficient (-0.050) (p=0.659)), HbA1C and serum iron (n=80) (Spearman’s correlation coefficient (-0.073) (p= 0.52)), HbA1C and transferrin saturation (n=80) (Spearman’s correlation coefficient (-0.105) (p=0.353)) and positive but not statistically significant correlation between HbA1c and TIBC (n=80) (Spearman’s correlation coefficient (0.132) (p=0.242)).

HbA1c alone, if taken as a marker of diabetes in a patient with iron deficiency and asymptomatic diabetes, will result in a false diagnosis of diabetes and complications related to antidiabetic therapy. Several studies have stressed the futility of diagnosing diabetes based on HbA1C alone as a marker of diabetes severity in an iron deficient population.

HbA1C shows a linear relationship with red blood cell life span (cohen et al), which is inversely proportional to the mean cell haemoglobin (MCH) in patients with anemia due to iron deficiency as found in a study conducted by Koga et al., where women in the pre-menopausal age group with low MCH had higher HbA1c levels independent of plasma glucose levels. ^{11, 13} This explains the negative correlation between haemoglobin and HbA1c found in our study and is consistent with studies conducted by Madhu et al., Silva et al and Rajagopal et al. ^{14– 16} These findings provide additional support to the notion that iron deficiency elevates HbA1c levels. Grossman et al. ¹⁷ found no correlation between hemoglobin and HbA1c levels suggesting that HbA1c levels in nondiabetic elderly individuals may be interpreted independently of hemoglobin or nutritional anaemia factors. However, their study categorized patients based on HbA1c and assessed haemoglobin levels while our study and others (Madhu et al., Silva et al and Rajagopal et al) adopted the opposite approach. This difference in study design might have contributed to confounding results.

A large retrospective study conducted by Rao et al. ¹⁸ not only showed a statistically significant difference in HbA1c levels in subjects with iron deficiency and normal individuals but also in both males (6% versus 5.6%) and females (5.7% versus 5.4%). These findings were similar to those of the present study.

Differences in HbA1c levels among subjects in the study group with and without iron deficiency were observed irrespective of the method employed to assess HbA1c levels in various studies. Silva et al. ¹⁵ have done a head-to-head comparison between ion exchange HPLC and immunoturbidimetry and found that in both the methods HbA1c was considerably higher in patients with iron deficiency than without iron deficiency ( Table 2).

In a study conducted by Christy et al., ¹⁹ subjects with diabetes and well-controlled sugar levels and iron deficiency exhibited significantly higher HbA1c levels than iron-sufficient diabetic individuals, corroborating our finding that iron deficiency anaemia independently influences HbA1c levels. In contrast, Solomon et al., ²⁰ found significantly lower HbA1c levels (6.18 ± 1.57) in the IDA group than in the control group (7.74 ± 1.81) (p < 0.05) within a diabetic population. However, it is worth noting that this study did not standardize blood sugar levels using fasting, random, postprandial blood glucose levels, or oral glucose tolerance tests. It must be emphasized that in the present study, random blood sugar levels were comparable between both the study groups with median RBS values of 106 mg/dL and 102 mg/dL among cases and controls, respectively. A difference was observed in the HbA1c levels. The highest HbA1c level was 8.6% noted for a 40-year-old gentleman with anal fissures with a haemoglobin of 3.8 g/dL. His RBS at admission was 89 mg/dL and FPG was 84 mg/dL. Ferritin was 4.5 ng/mL. 19.5% of cases had A1C in the diabetic range despite nondiabetic range plasma sugars. None of the patients had started receiving antidiabetic drugs.

Lyons et al., ²¹ in a study of non-diabetic individuals from the RODAM study group, found a significant association between iron deficiency and increased odds of having HbA1c levels in the prediabetes range (5.5-6.5%) compared to women without iron deficiency (OR: 1.43, p<0.01). This aligns with our findings of higher median HbA1c levels in women with iron deficiency anaemia (5.6%) than in those without (5.2%). However, the study did not identify a similar association between iron deficiency and HbA1c levels in men, contradicting our findings.

The results of the present study showed a higher median HbA1c in men with iron deficiency anaemia (5.7%) than in those without (5.2%). This discrepancy may be attributed to the smaller sample size of male patients with and without iron deficiency (n/n’=68/1177) in the study by Lyons et al.

In a large-scale study conducted by Ford et al., ²² where both diabetic and non-diabetic patients were categorized into four groups according to their haemoglobin and iron indices, and proposed that in a population, IDA has negligible effect on HbA1c levels or on diabetes prevalence which contradicts our observation. However, the same study also reported higher glycosylated hemoglobin values in individuals with iron deficiency than in those with normal iron status, suggesting that the results of this study might not be applicable to patients with severe IDA.

Correction of iron deficiency results in an increase in the young erythrocyte population with a decrease in the proportion of glycated haemoglobin levels, as observed in studies done in both diabetic and nondiabetic iron deficient individuals. ^{14 , 23} Aydin et al. ²⁴ studied diabetic iron deficient individuals and showed a significant statistical decline in the mean HbA1c from 7.09 ± 0.51(%) to 6.69 ± 0.53 (%) with the rise in haemoglobin levels (Spearman rho = −0.362; p < 0.001), and with similar baseline FBS levels before 118 (mg/dL) (108–132) mg/dL and after iron therapy 116 (mg/dL) (106–125). Similar observations were found in a study ²⁵ where iron therapy led to a statistically significant decline in HbA1c levels before (5.01 ± 0.39) and post-iron (4.69 ± 0.38) therapy in nondiabetic iron deficient pregnant women. The results of these independent investigations corroborate our findings regarding the relationship between IDA and HbA1c levels. In contrast, a study by Sinha et al. ²⁶ involving nondiabetic iron deficiency anemia observed increased HbA1c levels following iron therapy. However, the small sample size and potential confounding factors may have influenced the results.

Another study from southern India by Kalasker et al. ²⁷ reported that among their 40 non-diabetic test subjects with iron deficiency, the mean HbA1c was lower (5.91%) compared to their control group peers (6.54%). These findings contradict our study, which may be due to the higher mean hemoglobin level in their cases (8.74 ± 1.98 g/dL) compared to the cases in our study (6.6 g/dL, range 4.9–9 g/dL). The lower hemoglobin levels in our study likely indicate more severe anemia and greater red blood cell turnover, which may have contributed to the observed differences in HbA1c levels.

Iron deficiency induces oxidative stress, resulting in increased malonaldehyde, which accelerates glycation of haemoglobin. The same concept was reinforced by a study conducted by Ghazy et al., who showed a decrease in glycated haemoglobin and malonaldehyde accompanied by correction of iron deficiency in nondiabetic patients, which reinforces our finding of a positive correlation between iron deficiency and HbA1c. ^{9 , 28}

Our study’s strengths include a substantial sample size and careful matching of participants based on age and sex. A limitation of our study was the inability to examine the impact of treating iron deficiency anemia on HbA1c levels.

The findings of our study emphasize the importance of carefully interpreting HbA1c levels in individuals with iron deficiency, especially in areas with a high anemia prevalence, to avoid misdiagnosing diabetes or prediabetes. Moreover, the substantial prevalence of iron deficiency among patients with diabetes underscores the potential unreliability of HbA1c as the sole indicator of glycaemic control. Inaccurate assessment of glycaemic status through HbA1c in diabetics who also have iron deficiency anaemia that is co-existing may lead to unnecessary escalation of antidiabetic treatments, additional financial burden, and, most importantly, dangerous complications of hypoglycemia.

This study was approved by the review board of our institution -Kasturba Medical College and Kasturba Hospital ethics committee on 2 ^nd June 2021 (Reference number- IEC 200/2021). Written consent was taken from all participants Data were collected from the hospital’s medical records, and patient confidentiality was protected.