Self-Reported ABO Blood Type Compared With DNA-Derived Blood Group

To the Editor: Recent literature suggests ABO blood type may be associated with a variety of diseases, including pancreatic cancer,1,2 gastric cancer,3 and coronary heart disease,4 with studies often relying on self-report. Although ABO blood group is becoming important for epidemiologic studies, DNA may not always be available, or lack of funds may prohibit DNA genotyping or antigen tests to determine ABO group. Therefore, it is important to know the validity of self-reported ABO blood type. We assessed the validity of self-reported ABO blood type using DNA-derived ABO blood group as the gold standard. Evaluation of this has been minimal, with 1 hospital-based study in Japan assessing validity5 and 2 cancer studies using a very small sample subset to assess percent agreement between self-report and DNA-derived blood group.1,6 Subjects were controls participating in a population-based case-control cancer study conducted during 2011 in Ontario, Canada. Of the 1731 persons mailed a study package, 1,285 (74%) completed the questionnaire, and of those, 1,226 (95%) responded to, “What is your blood type?” Of these, 537 (44%) indicated “Don’t Know,” and 689 (56%) indicated ABO blood type. Subjects were also invited to provide a saliva sample through an Oragene DNA Self Collection Kit (DNA Genotek, Kanata, Ontario), or a blood sample drawn at a laboratory. Six hundred eight (47%) of 1285 persons who completed the questionnaire provided a blood/saliva sample from which DNA was extracted. ABO blood group was determined after genotyping (iPLEX assay; Sequenom, San Diego, CA) two single nucleotide polymorphisms in the ABO gene (rs8176719 and rs8176746). Allele combinations were used to determine the ABO status for each person, as previously done by others.7,8 Only participants who provided both self-reported blood type and DNA were included in this validity study (n = 337). The unweighted κ statistic (with 95% confidence interval [CI]) was used to assess chance-corrected agreement between self-reported and DNA-derived blood group. Using genotyped ABO as the gold standard, we calculated sensitivity, specificity, and percent agreement for each of the ABO blood groups individually. Almost perfect chance-corrected overall agreement was found between self-reported and DNA-derived ABO blood group (κ = 0.83 [95% CI = 0.77–0.88]) (Table).The specificity and sensitivity of self-reported ABO group were assessed against the gold standard of DNA-derived ABO group (Table). Sensitivity was highest for blood group O (0.93 [95% CI = 0.89–0.97]) and lowest in blood group AB (0.79 [0.61–0.97]). Specificity for all blood groups were above 0.90, ranging from blood group B (0.99 [0.97–1.00]) to blood group O (0.91 [0.92–0.989]). Percent agreement was greater than 90% for all blood groups. Of the 337 persons answering the blood-type question, 300 (89%) correctly reported their blood type. The distribution of DNA-derived ABO blood group was O (47%), A (37%), B (11%), and AB (6%), as expected.TABLE: Distribution of DNA-Derived ABO Blood Group by Self-Reported ABO Blood Group, and Sensitivity, Specificity, κ, and Percent Agreement (n = 337)Our study found almost perfect agreement between self-report and DNA-derived ABO blood group. High sensitivity and specificity further indicate that self-reported blood type is a valid measure of true ABO blood group. Similar to a Japanese study that assessed the validity of self-reported blood type among patients from one hospital,5 we found self-report to be a valid measure; however, our study has the advantage of being population-based. Two cancer studies used a very small subset of their study population and reported percent agreement to be 91%1 and 92%,6 similar to our finding of 89% agreement. In summary, self-report is an accurate measure of ABO blood group, suggesting self-report could be used as a proxy for DNA-derived ABO blood group in epidemiologic studies where DNA is not available. However, if a high proportion of subjects do not know their blood type, then this would be a limitation. ACKNOWLEDGMENT We gratefully acknowledge the dedication of Beth Lowcock as project coordinator. Zoë Bider-Canfield Prevention and Cancer Control Cancer Care Ontario Toronto, Ontario Canada [email protected] Michelle Cotterchio Prevention and Cancer Control Cancer Care Ontario Toronto, Ontario, Canada Dalla Lana School of Public Health University of Toronto Toronto, Ontario, Canada