Background/Purpose: The erythrocyte sedimentation rate (ESR) is a laboratory test commonly used in clinical practice as an assessment of systemic inflammation. The glycohemoglobin (A1c) measures long-term serum glucose levels to screen for diabetes and evaluate glycemic control. Both of these tests use red blood cells to determine their values. The result of the ESR test is nonspecific and varies with factors including patient age, gender, body size, pregnancy, anemia, albumin, and more. This project sought to determine the correlation of A1c to ESR values while accounting for other relevant clinical factors.

Methods: Historical biomedical data from 2011 to 2014 were collected and analyzed. Patients with ESR values resulted within a 30-day period of glycohemoglobin values were included, with a total sample size of 22,984. Additional information was collected including age, gender, body mass index, smoking status, platelets, WBCs, C-reactive protein and hemoglobin. The dataset was then analyzed to determine the correlation of ESR values with glycohemoglobin values. Additional analyses evaluated the impact of other variables on the ESR value and construct a model based on this using forward selection methods.

Results: The correlation between ESR and A1c levels was 0.268, with a p-value of 2.2e-16 indicating a strong positive correlation. Other variables that were significantly associated with the ESR level included hemoglobin, gender, creatinine, platelets, white blood cells (WBCs), and age (see table 1).

Table 1: Effect of Individual Variables on ESR Values.

Fitting the full model using significant predicting variables yielded the following: Model: Expected (ESR) = 37.6 – 3.5(hemoglobin) + 2.15(glycohemoglobin) + 2.17(creatinine) + 0.02(platelets) + 0.24(WBCs) + 0.24(Age) – 2(Male) This model had an overall F-statistic of 1418 with a p-value of 2.2e-16, indicating a highly significant model. The R squared value for this model was 0.3102, indicating that 31% of the variability of ESR was explained by the model.

Conclusion: The traditional model used in interpreting erythrocyte sedimentation rates (ESR) is age divided by two plus five for males, though this is understood to be an incomplete model. It is established that many factors effect ESR values, and this project sought to quantify the effect of these variables on actual ESR values using a retrospectively collected large biomedical dataset. The variables that had a significant effect on ESR levels are shown in table 1 and in the model. The most significant variable was hemoglobin, which had an inverse association with ESR. Other variables had a direct correlation to ESR. Elevated ESR values often result in referrals to rheumatology, and if there was a more accurate way to reliably interpret what an ‘average’ ESR level would be given a set of specific clinical factors it may help physicians and rheumatologists to frame and understand the significance of a given patients’ ESR value.

« Back to 2016 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/correlation-of-erythrocyte-sedimentation-rate-with-glycohemoglobin-values-and-other-patient-factors/