The prevention and early detection of iron deficiency

Transcription

1 BLOOD DONORS AND BLOOD COLLECTION The iron status of South African blood donors: balancing donor safety and blood demand Karin van den Berg, 1 Ronel Swanevelder, 2 Charlotte Ingram, 3 Denise Lawrie, 4 Deborah Kim Glencross, 4 Caroline Hilton, 5 and Martin Nieuwoudt 6,7 BACKGROUND: Several studies in developed countries have demonstrated high levels of iron deficiency (ID) among blood donors. There is a paucity of data for developing countries where blood shortages remain a major concern. STUDY DESIGN AND METHODS: A total of 4412 donors were enrolled in the study. Specimens were collected for full blood count, iron, transferrin saturation, and ferritin assessment. Donor demographics were recorded. ID was indicated by a ferritin level of less than 20 ng/ml for men and less than 12 ng/ml for women. Anemia was defined as hemoglobin levels less than 12.5 g/dl. Regression models for predictors of ID were developed. RESULTS: A total of 17.5% of all donors had ID, with 16.3% prevalence in women and 18.6% in men. Low hemoglobin had the highest association with ID (adjusted odds ratio [AOR], ; 95% confidence interval [CI], ); male donors had twice the odds of ID compared to female donors (AOR, 2.501; 95% CI, ), while increasing age was associated with lower odds (AOD, 0.965; 95% CI, ). Among male donors, an interdonation interval of less than 3 months (AOR, 2.679; 95% CI, ) was associated with ID. Compared to other females combined, colored female donors (AOR, 2.335; 95% CI, ) had higher odds and black female donors (AOR, 0.559; 95% CI, ) lower odds of ID. CONCLUSION: ID is common among South African donors; low hemoglobin, gender, ethnicity, and past donation history is independently associated with ID. Recommendations aimed at protecting donor health may increase blood shortages in South Africa. The prevention and early detection of iron deficiency (ID) among blood donors remains the focus of donor wellness interventions in many developed countries. 1 While it has been shown that regular donors are safer donors, especially in high-hivprevalence settings such as South Africa, there is a growing awareness of the potential deleterious effects of highfrequency blood donation. 2 6 In developed countries, some of the proposed interventions to minimize the risk of ID include increased interdonation intervals, iron replacement ABBREVIATIONS: AOR = adjusted odds ratio; CHr = reticulocyte hemoglobin content; CI = confidence interval; ID = iron deficiency; IDA = iron-deficiency anemia; NHLS = National Health Laboratory Services; POC = point-of-care; SANBS = South African National Blood Service; TS = transferrin saturation; WPBTS = Western Province Blood Transfusion Service. From the 1 Medical Department, South African National Blood Service, Port Elizabeth, South Africa; 2 Business Intelligence Department, South African National Blood Service, Roodepoort, South Africa; 3 South African Bone Marrow Registry (SABMR), Cape Town, South Africa; 4 Department of Molecular Medicine and Haematology, National Health Laboratory Services and University of the Witwatersrand, Johannesburg, South Africa; 5 Medical Department, Western Province Blood Transfusion Service, Cape Town, South Africa; 6 South African Department of Science and Technology/National Research Foundation Centre of Excellence in Epidemiological Modelling and Analysis (SACEMA), Stellenbosch, South Africa; and the 7 Institute for Biomedical Engineering (IBE), Stellenbosch University, Stellenbosch, South Africa. Address reprint requests to: Karin van den Berg, South African National Blood Service, 2 Buckingham Road, Mount Croix, Port Elizabeth 6001, South Africa; karin.vandenberg@sanbs.org.za This study was funded by the South African National Blood Service and Western Province Blood Transfusion Service. Received for publication May 28, 2018; revision received August 21, 2018; and accepted August 21, doi: /trf AABB TRANSFUSION 2019;59; TRANSFUSION Volume 59, January 2019

2 SOUTH AFRICAN IRON STUDY for at-risk donors, and monitoring of ferritin levels. 1 The potential impact of these interventions on the blood supply of these countries may, at least in part, have been offset by the significant decrease in blood utilization due to implementation of restrictive transfusion policies in many of these countries. 7,8 In contrast, there is a paucity of data on the iron status of blood donors in sub-saharan Africa and other developing countries. While many of these countries are still reliant on family replacement donors, there is a growing movement toward voluntary nonremunerated donation focusing on the development of a regular donor pool. 9 South Africa is a notable exception where blood donation has always been from voluntary nonremunerated donors between the ages of 16 and 65 (and older with clinician approval). In addition, the two South African blood transfusion services, the South African National Blood Service (SANBS) and the Western Province Blood Transfusion Service (WPBTS) have been able to provide for the country s blood needs, albeit with occasional periods of blood shortage. South Africa has a highly regular blood donor pool, with male and female donors of all ages allowed to donate every 56 days, that is, up to six times per year. The current return rate for donors at the SANBS is approximately two times per year, which is higher than most sub-saharan countries and approaches that of high-development-index countries. 9 These factors, while contributing to a sustainable blood supply, may place South African donors at risk of iron deficiency. Few studies have assessed the iron profile of donors in the developing world. In one such study, Jeremiah and Koate 10 assessed the iron status of Nigerian male blood donors. They confirmed a high prevalence of low hemoglobin (Hb) (13.7%), isolated ID (20.6%) and iron-deficiency anemia (IDA) (12%) among Nigerian blood donors. Regular donors were more affected, and most of the affected donors were older than 35 years of age. To our knowledge, there has been no systematic investigation of the iron profile of South African blood donors. The South African donor pool also has relatively high proportions of young, female, and black donors, all of which have previously been noted to be risk factors for ID and Hb deferral. 6,11 For these reasons, we performed a multicenter cross-sectional study to determine the prevalence of and predictors for ID among whole blood donors in the various sectors of South Africa s populace, with a view to identify interventions to protect donors while ensuring a safe and sufficient blood supply. MATERIALS AND METHODS The South African blood donor iron study was conducted over a 3-month period between August and October 2014 as a multicenter cross-sectional study to determine the prevalence of andriskfactorsforidandidaamongwholeblooddonors. Both South African blood transfusion services participated in thestudy.existingroutineblooddonorconsentatthetimeof the study included consent for the use of donated blood in research aimed at improving blood safety. A waiver for separate individual consent was obtained from the SANBS Human Research Ethics Committee (Clearance Certificate 2013/19). Participant enrollment and eligibility Donor enrollment was staggered, with both fixed centers and mobile blood drives accepting participants. For inclusion in this study, participants had to meet the standard donor acceptance criteria, including having a point-of-care (POC) Hb assessment performed. The latter was performed by either a copper sulfate gravimetric method or a quantitative capillary POC Hb measurement. Both services had the same minimum Hb cutoff for female donors (12.5 g/dl), but WPBTS had a higher cutoff for male donors (13.5 g/dl) compared to SANBS (12.5 g/dl). The two services have otherwise similar eligibility criteria. Exclusion criteria were people donating for therapeutic reasons; autologous and designated donors; donors deferred for reasons other than low Hb levels as well as those found to test confirmed positive for HIV, hepatitis B and C, and syphilis. Donors younger than 18 were excluded, as South African law requires ministerial and parental consent when performing any form clinical research on children younger than 18 years of age. Specimen collection and testing Study-specific specimens included a 7-mL serum separator tube for the iron profile studies, and a 5-mL ethylenediaminetetraacetic acid tube for full blood count were collected. Only the study-specific specimens were collected from consenting donors who had been deferred for low Hb. A full blood count was performed using a hematology analyzer (Advia 2120, Siemens Healthcare Diagnostics Ltd.) The iron profile consisting of serum transferrin, transferrin saturation (TS), iron, and ferritin were performed using commercially available kits (Advia Chemistry Ferritin Assay, _EN Rev. B, ; Iron_2 Assay, _EN Rev. E, ; Transferrin Assay, _EN Rev. D, Ltd.). The stability of the iron analytes using this procedure was previously validated. 12 In response to the variation in reported minimum ferritin levels for the classification of iron deficiency, 5,6,13 we used South African clinical minimum levels for the classification of iron deficiency, which for men are below 20 ng/ml and for women below 12 ng/ml. 14 For the assessment of anemia, a minimum Hb level of 12.5 g/dl, as mandated in the Standards of Practice for Blood Transfusion in South Africa, was used for both men and women. 15 Donors with both low serum ferritin levels and Hb levels less than 12.5 g/dl were classified as having IDA. Donor demographics and donation history Donor demographic and donation information is routinely collected during each donation event and includes gender, Volume 59, January 2019 TRANSFUSION 233

3 VAN DEN BERG ET AL. ethnicity, age, type of donation, and donation site. History of previous donations and deferrals is determined from the donor s lifetime records. South Africa has four major ethnic groups: Asian, Colored, Black, and White. The South African Colored population is a mixed-population group dating back to slavery and the early settlers in South Africa and is made up of five source populations (African Khoisan, African Bantu, European, South Asian, and East Asian). 16 Statistical analysis Sample size At the time of this study, the SANBS active donor population comprised approximately 217,000 women and 220,000 men. The ID estimate in the South African donor populace used to calculate sample size was based on a prior Brazilian study by Terada et al. 4 In that study, ID among female donors ranged from 11.9% in first-time donors to 38.9% in repeat donors. For male donors, prevalence ranged from 0% on first donation to 5% in repeat donors. Sample sizes were calculated separately for women and men. The margin of error was set at 0.05% and power to 80%. To estimate 38.9% ID, a minimum of 1452 female participants was required. To detect 5% ID in male donors, 292 needed to be enrolled. To enable subanalysis and ensure representative sampling from all areas of the country, we choose to enroll all donors who presented during the study enrollment period at the various study sites, which yielded a significantly larger sample size. Data management A central data warehouse in the SANBS is populated by automated extract, transform, and load processes. During this process, test results, donor status, donation, and deferral history for each donation are calculated and linked. For this study, the above data were merged with the donor data from the WPBTS and then with the iron indices tested by the National Health Laboratory Services (NHLS). Data analysis All data were combined into a spreadsheet (Excel, Microsoft) file. Descriptive cross-tabulated statistics were calculated for all variables using statistical software (Stata SE version 15, StataCorp LP). Box and whisker plots were generated in Excel. Univariate statistical comparisons were done in either Stata or another statistical analysis program (Statistix version 9, Analytical Software). Statistical regression models for predictors of ID were executed in Stata. An overall model including both men and women was initially defined. Then, owing to differences observed in the descriptive statistics, separate models for each sex were developed. Dummy variables were created for categorical variables (such as age and ethnicity) in which individual categories had differing ID prevalence. A variable-inclusive approach was used to start, followed by stepwise variable exclusion processes combining collinear variable removal, confidence interval and p-value criteria, log likelihood maximization, and Akaike and Bayesian information criteria minimization. Final models were selected as being simultaneously parsimonious and maximally informative. RESULTS Of the 7594 donors who presented to donate during the study enrollment period, 4412 (58.1%) were included in this analysis (Fig. 1). Despite the strict inclusion and exclusion criteria, in-house comparison of the demographics of the study population to the overall South African donor population revealed that it was representative of the latter (data not shown). The descriptive results of the demographic and clinical characteristics for men and women are presented in Table 1. Female donors ( years) were on average younger than the male ( years) donors. The majority of the donors were White (60.8%); Black donors comprised 24.4% (1063) of the study population. Male and female donors displayed demographic differences in their donation habits; 242 (64.5%) of the new donors were female, while male donors, on average, made more donations in the previous 24 months ( vs donations), had almost double the lifetime collections of females ( vs donations), and had shorter interdonation intervals ( vs months). Average Hb differed by almost 2 g/dl between men ( g/dl) and women ( g/dl). In total, 176 (13.9%) female and 28 (1.2%) male donors were deferred for low Hb based on POC Hb assessment while 289 (13.9%) female donors and 49 (2.1%) male donors had confirmed low Hb levels based on venous sample assessment. Iron depletion (ferritin <12 μg/l in women and < 20 μg/l in men) was present in 337 (16.3%) of the female and 426 (18.6%) of the male donors. Overall, 17.5% (763) of South African blood donors displayed evidence of iron depletion. In addition, 7.7% (341) of all donors, 11.1% (232) of female donor and 4.7% (109) of male donors had absent ferritin stores. Analysis of the distributions of Hb, reticulocyte hemoglobin content (CHr), TS, and ferritin in relation to specific reference ranges are shown in Fig. 2. For both female and male donors, Hb, CHr, and TS were normally distributed. In comparison to the 11.6 g/dl for the female population suggested by Lawrie et al., 17 the SANBS/WPBTS cutoff of 12.5 g/dl for female donors was conservative; the 5th and 95th percentiles for study participants fell within the suggested normal reference ranges, with little variability among various donation frequencies. For male donors, the SANBS cutoff of 12.5 g/dl was lower than the reference range (13.4 g/dl), 17 with a number of donors with donation 234 TRANSFUSION Volume 59, January 2019

4 SOUTH AFRICAN IRON STUDY Fig. 1. Enrollment schema for the South African blood donor iron study. [Color figure can be viewed at wileyonlinelibrary.com] frequencies greater than three donations in 24 months demonstrating Hb values lower than the reference but still above the 12.5 g/dl cutoff mandated by the Standards of Practice for Blood Transfusion in South Africa. 15 Although normally distributed, CHr minimum and maximum values had a larger range in men than in women. Even for first-time donors, the Schapkaitz et al. 18 reference did not provide a good match to our results. CHr demonstrated low correlation with ferritin (data not shown). Similarly, there was no clear discernible trend for TS with increasing donation frequency in female donors, with a slight downward trend visible in male donors. Values for highly regular donors (>6 donations/24 months) were similar in women and men. Values for up to one-half of repeat donors, both male and female, were below the lower limit of normal as used by the NHLS. Distribution of ferritin values was skewed to the left in both men and women and differed markedly between the two groups. There was a major decline in ferritin levels in repeat male donors compared to first-time male donors, with a continued decline with increasing donation frequency. Although female donors also demonstrated a decline in ferritin levels with increasing donation frequency, the decline was not as marked. The prevalence of low Hb and low ferritin by interdonation interval differed between male and female donors (Table 2). Overall, 14% of female and 2.1% of male donors had low Hb. Low ferritin was more common among male donors (18.4%) than female donors (16.2%), while 7.4% of female donors had both low Hb and low ferritin compared to just 1.7% of male donors. Almost 20% of first-time female donors had Hb levels below 12.5 g/dl compared to only 1.6% of first-time male donors. The prevalence of low Hb was lower in repeat female donors compared to first-time female donors regardless of interdonation interval, with no significant difference between SANBS and WPBTS donors. In contrast, low ferritin levels were higher among repeat female and male donors, with significant differences between SANBS and WPBTS donors in both women and men. Iron depletion (normal Hb with low ferritin) was more common among male (16.7%) than female (8.8%) donors, with significantly higher prevalence in both SANBS male and female donors compared to WPBTS donors: 18.1% in SANBS male versus 6.0% in WPBTS male donors (p < 0.01); and 9.4% in SANBS female versus 4.3% in WPBTS female donors (p = 0.03). IDA was significantly more prevalent among SANBS female donors (7.9%) than those from WPBTS (3.2%) (p = 0.03). The difference in IDA prevalence between male donors from SANBS (1.8%) and WPBTS (0.9%) was not significant (p = 0.52), but the sample size was very small. Volume 59, January 2019 TRANSFUSION 235

5 VAN DEN BERG ET AL. TABLE 1. Demographic and clinical characteristics by sex of the 4412 participants enrolled in the South African iron study N (%) or Mean SD Characteristics Female Male Total Sex 2087 (47.3) 2323 (52.7) 4410 (100) Age distribution (59.2) 156 (40.8) 382 (8.7) (52.3) 531 (47.7) 1114 (25.2) (49.8) 487 (50.2) 971 (22.0) (44.6) 518 (55.4) 935 (21.2) > (37.5) 631 (62.5) 1010 (22.9) Age Ethnic distribution White 1200 (45.4) 1443 (54.6) 2643 (60.8) Colored 190 (52.1) 175 (47.9) 365 (8.4) Black 563 (53.0) 500 (47.0) 1063 (24.4) Asian 101 (36.3) 177 (63.7) 278 (6.4) Zone* (altitude in m) Eastern Cape ( ) 223 (51.9) 207 (48.1) 430 (9.8) Egoli ( ) 403 (48.3) 432 (51.7) 835 (18.9) Free State/North Cape ( ) 215 (46.6) 246 (53.4) 461 (10.5) KwaZulu Natal ( ) 255 (46.7) 291 (53.3) 546 (12.4) Mpumalanga ( ) 169 (47.5) 187 (52.5) 356 (8.1) Northern ( ) 370 (46.5) 425 (53.5) 795 (18.0) Vaal ( ) 267 (45.6) 318 (54.4) 585 (13.3) Western Cape ( ) 185 (46.0) 217 (54.0) 402 (9.1) Donor type New 242 (64.5) 133 (35.5) 375 (8.5) Repeat 1847 (45.8) 2190 (54.2) 4037 (91.5) Average collections in previous 24 months Average lifetime collections Average interdonation interval in months (from 2009) Average Hb Deferred for low Hb during study 176 (8.4) 28 (1.2) 204 (4.6) History of low Hb deferral 113 (64.2) 17 (60.7) 130 (63.7) Low Hb 289 (13.9) 49 (2.1) 338 (7.7) Average ferritin Low ferritin 337 (16.3) 426 (18.6) 763 (17.5) Absent ferritin (<10 μg/l) 232 (11.1) 109 (4.7) 341 (7.7) * Zone refers to a predefined geographic area for operational purposes, not limited to provincial boundaries. Hb = hemoglobin. An overall multivariate model including both sexes, with low ferritin as a binary response variable, and the following predictor variables low Hb, sex, age, ethnicity, number of collections in the preceding 24 months, interdonation interval, Hb deferral in the past 12 months, and lifetime collections produced a model that was significant for all predictors (Table 3). Altitude was initially included in the model but was subsequently dropped, as it had an insignificant effect. The model had a negative intercept, implying that iron deficiency preexists within the general population. The bivariate correlation coefficient (R 2 ) for Hb and ferritin levels was 0.306, suggesting low direct correlation. However, when incorporated in the multivariate model, low Hb (adjusted odds ratio [AOR], ; 95% confidence interval [CI], ) was the strongest predictor of low ferritin. In addition, male donors had significantly higher odds of ID compared to females (AOR, 2.501; 95% CI, ). Given the difference between the sexes, multivariate models were then generated separately for men and women, with age, ethnicity, and interdonation intervals included as categorical variables. Insignificant categories were dropped from the models. Male donors with low Hb had very high odds of low ferritin (OR: ; 95%CI: ) compared to those with normal Hb. Male donors aged 20 and younger, those with a greater number of donations in the preceding 24 months, a less than 3-month inter-donation interval and those with a history of having been deferred for low Hb in the previous 12 months had significant odds of having low ferritin levels. Being White was associated with significantly lower odds of low ferritin. For female donors, low Hb still had the highest association with low ferritin (odds ratio, ; 95% CI, ). Younger (<20 years old) women had significant odds of low ferritin. Female donors with a greater number of donations in the preceding 24 months, those with an interdonation interval of less than 3 months, and those with a history of deferral for low Hb had significant odds of low ferritin. Colored female donors had significantly higher odds of low ferritin, while Black female donors had significantly lower odds of low ferritin. 236 TRANSFUSION Volume 59, January 2019

6 SOUTH AFRICAN IRON STUDY Fig. 2. Box plots of selected iron biomarkers by increasing number of donations in preceding 24 months. [Color figure can be viewed at wileyonlinelibrary.com] Volume 59, January 2019 TRANSFUSION 237

7 VAN DEN BERG ET AL. TABLE 2. Hemoglobin and ferritin levels of the female and male participants by interdonation interval and by blood service SANBS First donation Duration between collections (average months, since 2009) Variable 0 3 >3 6 >6 All Donors SANBS Overall WPBTS Overall p value* (SANBS/WPBTS) Females (N) Normal Hb 195 (80.9%) 257 (87.7%) 461 (85.5%) 714 (87.1%) 1793 (86.0%) 1627 (85.9%) 164 (87.7%) 0.52 Low Hb (anemia) 46 (19.1%) 36 (12.3%) 78 (14.5%) 106 (12.9%) 289 (14.0%) 266 (14.1%) 23 (12.3%) 0.59 Low ferritin 30 (12.5%) 48 (16.6%) 127 (23.7%) 119 (14.7%) 338 (16.2%) 324 (17.3%) 14 (7.5%) <0.01 Low Hb + normal ferritin 23 (9.7%) 14 (4.8%) 29 (5.4%) 50 (6.2%) 133 (6.4%) 116 (6.2%) 17 (9.1%) 0.15 Normal Hb + low ferritin 7 (2.9%) 25 (8.6%) 79 (14.8%) 64 (7.9%) 183 (8.8%) 175 (9.4%) 8 (4.3%) 0.03 Low Hb + low ferritin 23 (9.7%) 22 (7.6%) 48 (9.0%) 55 (6.8%) 154 (7.4%) 148 (7.9%) 6 (3.2%) 0.03 Males (N) Normal Hb 126 (98.4%) 603 (98.1%) 673 (97.7%) 652 (98.3%) 2263 (97.6%) 2054 (98.0%) 209 (96.8%) 0.29 Low Hb (anemia) 2 (1.6%) 12 (1.9%) 16 (2.3%) 11 (1.7%) 48 (2.1%) 41 (2.0%) 7 (3.2%) 0.32 Low ferritin 2 (1.6%) 193 (31.8%) 155 (22.8%) 61 (9.3%) 426 (18.4%) 411 (19.9%) 15 (6.9%) < 0.01 Low Hb + normal ferritin (0.4%) 1 (0.2%) 8 (0.4%) 3 (0.2%) 5 (2.3%) < 0.01 Normal Hb + low ferritin (30.0%) 141 (20.8%) 51 (7.8%) 386 (16.7%) 373 (18.1%) 13 (6.0%) <0.01 Low Hb + low Ferritin 2 (1.6%) 11 (1.8%) 14 (2.1%) 10 (1.5%) 39 (1.7%) 37 (1.8%) 2 (0.9%) 0.52 * Fisher s exact test for two proportions comparing SANBS and WPBTS. Hb = hemoglobin; SANBS = South African National Blood Service; WPBTS = Western Province Blood Transfusion Service. DISCUSSION To our knowledge, this study represents the largest systematic assessment of the iron status of donors on the African continent and, with the inclusion of 1063 Black donors, also the largest assessment of ID among donors of African descent. Our study confirmed high rates of ID among South African blood donors of all ethnicities, with more than one in six donors displaying evidence of ID. Overall, male donors had a slightly higher prevalence of ID but lower prevalence of absent iron stores, had higher Hb levels, and were less likely to have been deferred for low Hb. However, after controlling for other variables, male donors had a 2.5 times higher odds of low ferritin than female donors. Other than sex, ethnicity, low Hb, age, and past donation history, including number of donations and interdonation interval, were significantly associated with low ferritin. The prevalence of ID in our study is consistent with what was found in other studies in a variety of settings. 2 4,10 Although the reported prevalence of ID in these studies varies significantly, they have all uniformly concluded high rates of ID especially among female, younger, and repeat donors. Jeremiah and Koate 10 demonstrated very high levels of ID (32.6%) among a small group of male blood donors in Nigeria. In the United States, the Retrovirus Epidemiology Donor Study-II (REDS-II) Donor Iron Status Evaluaiton (RISE) study 11 confirmed absent iron stores, defined as ferritin levels below 12 ng/dl in 15% of donors, with female donors more affected than males, but with no association with ethnicity. Similar findings have been noted in Europe and Australasia. 3,5,6 The variability in the reported ID prevalence may in part be due to the use of different ferritin levels to define ID and absent iron stores, but may also be due to different interdonation intervals and maximum number of donations per year allowed in various countries. 5,6,11,13 The use of different Hb cutoffs for acceptance of donors may further complicate comparisons across countries and blood services. Even within our study, the higher Hb cutoff used for male donors from WPBTS was associated with a significantly lower prevalence of ID compared to male donors from the SANBS. WPBTS started using quantitative finger prick Hb assessment exclusively in 2007 and implemented the higher Hb cutoff for male donors in 2011 (written communication, Greg Bellairs, June 2, 2017). A Cape Town study published in 2016 that assessed ID among healthy South Africans used a ferritin cutoff of 30 ng/dl. 19 This resulted in 39.8% of participants being classified as iron deficient, with even higher rates among women (56.6%) and Black Africans (50.7%). However, in a country with regulated iron-fortified staple foods 20 and general access to primary health care, it seems implausible that one-half of the South African population would be iron deficient. The challenge of what cutoff levels to use to define acceptable levels for blood donors was not limited to ferritin but included other indices such as Hb, CHr, and TS. Compared to local reference intervals, 17 the Hb distribution of donors, even repeat donors, fell well within the 5th and 95th percentiles; however, this was not true for TS or ferritin. Based on the NHLS lower limit of normal for TS, the levels noted in our study would suggest that our donor populace was iron deficient in all groups except first-time donors. This was not in agreement with what was seen for the ferritin levels, where only highly regular male donors had ferritin levels below the local minimum cutoff of 20. This variability in limits of normal would suggest that the 238 TRANSFUSION Volume 59, January 2019

8 SOUTH AFRICAN IRON STUDY TABLE 3. Logistic regression model findings for low ferritin A. For all donors Variable Odds ratio 95% CI p value Low Hb Male gender Age Ethnicity* Donations (24 months) Interdonation interval Hb deferrals (12 months) Lifetime collections Intercept (β 0 ) Likelihood ratio χ Log likelihood Prob > χ Pseudo R AIC BIC N (observations) 3877 B. For Men Variable Odds ratio 95% CI p value Low Hb Age Age White Donations (24 months) Interval <3 months Hb deferrals (12 months) Lifetime collections Intercept (β 0 ) Likelihood ratio χ Log likelihood Prob > χ Pseudo R AIC BIC N (observations) 2037 C. For women Variable Odds ratio 95% CI p value Low Hb Age Age Colored** Black Donations (24 months) Interval <3 months Hb-deferrals (12 months) Intercept (β 0 ) Likelihood ratio χ Log likelihood Prob > χ Pseudo R AIC BIC N (observations) 1840 * Ethnicity categories: white = 0 (reference), all others = = 1, all others = 0 (reference) = 1, all others = 0 (reference). White = 1, all others = 0 (reference). ** Colored = 1, all others = 0 (reference). Black = 1, all others = 0 (reference). AIC = Akaike information criteria; BIC = Bayesian information criteria; CI = confidence interval; Hb = hemoglobin. Volume 59, January 2019 TRANSFUSION 239

9 VAN DEN BERG ET AL. established prior reference intervals may require revision, considering that our study had a very large sample size, even of first-time donors, relative to most prior studies aimed at establishing local reference intervals. As noted by others, 21,22 we found only a weak direct correlation between low Hb and low ferritin. Despite this, low Hb remained the strongest predictor of ID in our multivariate model. Of concern is the number of donors who passed the POC Hb assessment but for whom a low Hb was later confirmed on venous sample measurement (13.9% of female donors and 2.1% of male donors). This raises the question of the reliability of current POC Hb measurement to accurately identify low Hb. A recent review of the variability in Hb cutoffs and Hb assessment noted inconsistent findings (both over- and underestimation of Hb) regarding the correlation between finger stick and venous Hb assessment. 23 Given the constraints associated with POC Hb assessment and the weak association between low Hb and low ferritin, development of a POC ferritin measurement would be the more ideal solution for predonation Hb assessment. We recognize that our study had some limitations. Chief among these may be the use of a ferritin cutoff of 20 ng/dl for defining ID in male donors, which may have resulted in an overestimation of ID in our male participants. Although higher than what was used in some studies, it is in line with current clinical practice in South Africa. While we used a convenience, consecutive enrollment schema, the demographic distribution of our study sample closely resembled that of the South African donor population. Not all donors deferred for low Hb consented to participate in the study and as such, overall, ID among blood donors may be more prevalent than reported here. However, the purpose of the study was in part to assess ID among donors accepted for donation, as continued bleeding of these donors may negatively affect their health. In addition, our large sample size limits potential enrollment bias. Finally, due to operational difficulties, WPBTS was able to enroll only repeat donors. However, overall, the 8.5% new donors in the study closely resembles the donor type distribution in the general donor population. Despite high rates of deferral for low Hb, 24 the findings of this study suggest that ID is a threat to the future sustainability of the South African blood supply. The country s blood supply is provided by a relatively small donor pool (<1% of the population) with a comparatively high donation frequency of approximately 2.06 per year, resulting in a national donation rate of 18 donations per 1000 population. 9 Attempts to increase the donor pool are hampered by the very high background HIV prevalence of approximately 18.9% in the target age groups, 25 a large rural population, and significant socioeconomic challenges within the country. These factors contribute to regular but well-managed blood shortages, which are counteracted by a highly regular donor pool. Recommendations from this study to protect donor health include increasing Hb cutoff for male donors, increasing interdonation intervals, improved POC Hb assessment, and increasing deferral periods for donors identified with low Hb. Implementation of these recommendations is almost certain to dramatically increase blood shortages in South Africa. Measures that may counteract the impact of these recommendations includes lowering Hb cutoff for Black female donors and offering iron replacement as well as proactive ferritin monitoring to at-risk donors. Careful consideration with further analysis and follow-up research is required to ensure policy development that adequately protects donors, while ensuring a sustainable blood supply for the patients of South Africa. ACKNOWLEDGMENTS The authors thank the staff and donors of the SANBS, WPBTS, and the NHLS, without whom the study would not have been possible. CONFLICT OF INTEREST The authors have disclosed no conflicts of interest. REFERENCES 1. Goldman M, Magnussen K, Gorlin J, et al. International forum regarding practices related to donor haemoglobin and iron. Vox Sang 2016;111: Cable RG, Glynn SA, Kiss JE, et al. Iron deficiency in blood donors: the REDS-II Donor Iron Status Evaluation (RISE) study. Transfusion 2012;52: Rigas AS, Srensen CJ, Pedersen OB, et al. Predictors of iron levels in 14,737 Danish blood donors: results from the Danish Blood Donor Study. Transfusion 2014;54(3pt2): Terada CT, Santos PC, Cançado RD, et al. Iron deficiency and frequency of HFE C282Y gene mutation in Brazilian blood donors. Transfusion Med 2009;19: Salvin HE, Pasricha SR, Marks DC, et al. Iron deficiency in blood donors: a national cross-sectional study. Transfusion 2014;54: Badami KG, Taylor K. Iron status and risk-profiling for deficiency in New Zealand blood donors. N Z Med J 2008;121: Roubinian NH, Escobar GJ, Liu V, et al. Trends in red blood cell transfusion and 30-day mortality among hospitalized patients. Transfusion 2014;54: Goodnough LT, Maggio P, Hadhazy E, et al. Restrictive blood transfusion practices are associated with improved patient outcomes. Transfusion 2014;54: World Health Organisation. Global status report on blood safety and availability Geneva: World Health Organisation; TRANSFUSION Volume 59, January 2019

10 SOUTH AFRICAN IRON STUDY 10. Jeremiah ZA, Koate BB. Anaemia, iron deficiency and iron deficiency anaemia among blood donors in Port Harcourt, Nigeria. Blood Transfusion 2010;8: Cable RG, Glynn SA, Kiss JE, et al. Iron deficiency in blood donors: analysis of enrollment data from the REDS-II Donor Iron Status Evaluation (RISE) study. Transfusion 2011;51: van den Berg K, Ingram C, Nagel S. The impact of time and centrifugation on the stability of iron studies. Kigali, Rwanda: African Society for Blood Transfusion; Baart AM, van Noord PA, Vergouwe Y, et al. High prevalence of subclinical iron deficiency in whole blood donors not deferred for low hemoglobin. Transfusion 2013;53: Siemens Healthcare Diagnostics Ltd. (editor). Camberley: Advia Chemistry Systems: Ferritin (FRT); Western Province Blood Transfusion, South African National Blood Service. Standards of practice for blood transfusion in South Africa. 6th ed. Cape Town: Western Province Blood Transfusion Service; South African National Blood Transfusion Service; p Daya M, van der Merwe L, Galal U, et al. A panel of ancestry informative markers for the complex five-way admixed South African coloured population. PLoS One 2013;8:e Lawrie D, Coetzee LM, Becker P, et al. Local reference ranges for full blood count and CD4 lymphocyte count testing. S Afr Med J 2009;99: Schapkaitz E, Buldeo S, Mahlangu JN. Diagnosis of iron deficiency anaemia in hospital patients: use of the reticulocyte haemoglobin content to differentiate iron deficiency anaemia from anaemia of chronic disease. S Afr Med J 2015; 106: Phatlhane DV, Zemlin AE, Matsha TE, et al. The iron status of a healthy South African adult population. Clin Chim Acta 2016; 460: South African National Department of Health. Regulations relating to the fortification of certain foodstuffs. Pretoria: Department of Health; Nadarajan V, Sthaneshwar P, Eow GI. Use of red blood cell indices for the identification of iron deficiency among blood donors. Transfusion Med 2008;18: Kiss JE, Steele WR, Wright DJ, et al. Laboratory variables for assessing iron deficiency in REDS-II Iron Status Evaluation (RISE) blood donors. Transfusion 2013;53: Chaudhary R, Dubey A, Sonker A. Techniques used for the screening of hemoglobin levels in blood donors: current insights and future directions. J Blood Med 2017;8: Manjezi M, Strydon K, Van den Berg K. Evaluation of the causes and rates of donor deferrals in the East London branch of SANBS. Mauritius: African Society of Blood Transfusion; Statistics South Africa. Mid-year population estimates Pretoria: Statistics South Africa; Volume 59, January 2019 TRANSFUSION 241