Abstract

Background

Long-term deleterious effects of repeated blood donations may be masked by the donors’ healthy lifestyle. To investigate possible effects of blood donation and iron loss through blood donation on cancer incidence while minimizing “healthy donor effects,” we made dose–response comparisons within a cohort of Swedish and Danish blood donors.

Methods

We used a nested case–control study design, in which case patients were defined as all donors who were diagnosed with a malignancy between their first recorded blood donation and study termination (n = 10866). Control subjects (n = 107140) were individually matched on sex, age, and county of residence. Using conditional logistic regression, we estimated relative risks of cancer according to number of blood donations made or estimated iron loss 3–12 years before a case patient was diagnosed with cancer. All statistical tests were two-sided.

Results

No clear association was observed between number of donations and risk of cancer overall. However, between the lowest (≤median, <0.75 g) and highest (>90th percentile, >2.7 g) categories of estimated iron loss, there was a trend ( Ptrend < .001) of decreasing risk for cancers of the liver, lung, colon, stomach, and esophagus, which are thought to be promoted by iron overload (combined odds ratio [OR] = 0.70, 95% confidence interval [CI] = 0.58 to 0.84), but only among men and only with a latency of 3–7 years. The risk of non-Hodgkin lymphoma was higher among frequent plasma donors (>25 vs 0 donations, OR = 2.14, 95% CI = 1.22 to 3.74).

Conclusions

Repeated blood donation was not associated with increased or decreased risk of cancer overall. The lack of consistency across latency periods casts doubt on an apparent association between reduced cancer risk and iron loss in men. The positive association between frequent plasma donation and risk of non-Hodgkin lymphoma deserves further exploration.

CONTEXT AND CAVEATS
Prior knowledge

Low rates of cancer incidence and mortality among blood donors suggest that frequent blood donation does not have long-term health risks.

Study design

Nested case–control study within a cohort of blood donors in Denmark and Sweden.

Contributions

No association between number of blood or plasma donations and overall risk of cancer was observed. A dose-dependent decreasing risk of cancers of the liver, lung, colon, stomach, and esophagus—which are suggested to be promoted by iron overload—was observed among men who donated blood 3–7 years earlier. An increased risk of non-Hodgkin lymphoma was observed among frequent plasma donors.

Implications

Risk of overall cancer is not associated with repeated blood donation.

Limitations

Factors that do or may influence cancer risk, such as smoking, alcohol consumption, diet, physical activity, anthropometric measurement, and occupational exposure to carcinogens were not available in the analysis.

Although low rates of cancer incidence and all-cause mortality among blood donors ( 1–4 ) suggest that blood donation is not harmful, possible deleterious effects of repeated blood letting may be masked by the healthy lifestyle and good health of the donor population. Apart from immediate complications, such as syncope and local bleeding ( 5 ), short- and long-term health effects of repeated blood donations per se have not been thoroughly investigated.

It has been suggested that depletion of iron, which catalyzes the production of a range of reactive oxygen species and free radicals, may have beneficial health effects ( 2 , 4 , 6–10 ). Low iron stores have been associated with reduced risks of cardiovascular disease ( 11–15 ); however, most recent studies, including a clinical trial in which subjects with peripheral arterial disease were randomly assigned to reduction of stored iron through blood letting ( 16 ), did not result in the purported protection ( 17–20 ).

High levels of iron may be associated with cancers of the lung, liver, colon, stomach, and esophagus ( 8–10 , 21 , 22 ). Several reports have investigated the association between levels of stored iron, as indicated by various laboratory markers, and cancer risk, but results have varied considerably ( 21–24 ). It has been suggested that increased cell proliferation, which is itself a possible risk factor for cancer ( 25 , 26 ) and is stimulated by repeated removal of blood cells, may increase the risk of hematopoietic malignancies in blood donors ( 4 ). Some have speculated that the immunologic effects of blood donation could also increase cancer risk ( 27–31 ).

To quantify cancer risk in relation to donation intensity and iron loss through blood donation, we conducted a case–control study nested within a large cohort of Swedish and Danish blood donors.

Subjects and Methods

Data Sources

Starting in the mid-1960s, an increasing number of Swedish blood banks and transfusion medicine clinics began computerized registration of blood donations and transfusions ( 32 ). Similar registration had started in Denmark in the 1960s ( 33 ), but data storage on a lasting electronic medium was delayed until 1981. Near nationwide coverage was attained in Sweden by 1992 and in Denmark by 1997.

We assembled all available computerized blood donation and transfusion registers from Swedish and Danish blood banks and transfusion medicine clinics into the Scandinavian Donations and Transfusions (SCANDAT) database, which has been described in greater detail previously ( 34 ). Hemoglobin concentration was recorded for each donation, and information is also available regarding the type of donation (whole blood, plasma, platelets, or other). Using the individually unique national registration numbers that are assigned to all residents of both Sweden and Denmark, we performed record linkages with the complete and nationwide registers of total population, death, and migration, thus ensuring the correctness of the national registration numbers and correct censoring on emigration and death. Record linkages with nationwide and essentially complete Swedish and Danish cancer registers ( 35 , 36 ) provided diagnosis dates and anatomic sites of all malignancies. After record linkage, the database was permanently de-identified by replacing the national registration numbers with anonymous but unique identification numbers.

Study Design

From the SCANDAT database we identified all donors for whom at least one successful whole blood or plasma donation was recorded between January 1, 1968, and December 31, 2002, and who had no history of cancer at the time of the first recorded donation. We conducted a matched case–control study nested within this cohort of 1110212 blood donors who were followed from the date of first donation until the first cancer diagnosis, emigration, death, or December 31, 2002, whichever occurred first. Case patients were all blood donors who were diagnosed with a first primary cancer during follow-up. For each identified case patient, we used incidence density sampling to randomly select 10 control subjects from the cohort, who were alive and cancer free at the time of diagnosis of the case patient (ie, the index date). The control subjects were matched to the case patients by sex, age (±180 days), and county of residence. For meaningful comparisons, both case patients and control subjects were required to have donated blood at least once during years 3–12, counting backward from the index date.

To obtain measures of donation frequency in comparable periods, we restricted the exposure assessment to fixed exposure windows, which were predefined in terms of time before the index date (3–7, 8–12, or 3–12 years). To be eligible for the analyses that went back beyond 7 years, the case patients and control subjects had to have resided during the entire length of the time window in counties that were fully covered by the electronic registrations that were assembled in the SCANDAT database. Hence, subjects living in counties in which the registration was not in full operation from the first day of the time window or who, due to relocations, had lived in noncovered counties during parts of the time window, were not eligible. The dates of such relocations were obtained from national registers. Hence, both the case patients and control subjects had to have at least 7 years of register coverage to be eligible.

For each case patient and his or her matched control subjects, we counted the number of whole blood and plasma donations performed in the respective time windows. We estimated iron loss using the approximate volume of whole blood loss for each donation (480 mL for whole blood, 50 mL for plasma) and the hemoglobin concentration measured immediately before each donation.

Because of early symptoms, donors with an incipient malignancy are likely to donate blood less often than healthy donors, which could lead to reverse causation. We therefore carried out the following analysis to determine the point from which the case patients' donation activity first started to decline. We first divided the 5 years immediately before cancer diagnosis into 6-month segments. For each segment, we calculated the relative donation intensity of the case patients as the ratio of the observed frequency to the expected frequency (the average frequency of a randomly selected control subject from the matched set). Confidence intervals (CIs) were calculated for the relative donation intensities under the assumption that the number of donations followed a Poisson distribution. Based on these calculations, which revealed a decreasing donation intensity among case patients starting as early as 2 years before the index date ( Figure 1 ), we decided to exclude from the analyses the 2 years immediately before cancer diagnosis.

Figure 1

Donation intensity among case patients relative to control subjects presented by time to diagnosis ( triangles ). 95% confidence intervals ( error bars ) are shown.

Figure 1

Donation intensity among case patients relative to control subjects presented by time to diagnosis ( triangles ). 95% confidence intervals ( error bars ) are shown.

Statistical Analysis

We analyzed associations between cancer incidence and the number of donations (any type) in the respective exposure windows using conditional logistic regression models with the exposure variable treated as a categorical variable (to allow for nonlinear relationships) with cut points chosen at approximately the median, 75th percentile, and 90th percentile. Thus, the four categories were 1–8, 9–16, 17–25, and more than 25 donations for the 10-year window from 3–12 years before diagnosis and 0–4, 5–8, 9–12, and more than 12 donations for the shorter 5-year windows (ie, 3–8 and 8–12 years before diagnosis of the case patient). We also examined the association with the number of donations as a continuous variable to test for a linear trend. We then considered whole blood and plasma donations separately, again both as categorical and continuous variables. Both the number of whole blood and the number of plasma donations were categorized as above. Because the type of donation was recorded with less precision in the Danish data, the analyses considering whole blood and plasma donations separately were restricted to Swedish data only.

After we examined associations with overall cancer risk, we performed further analyses for three broad groups of malignancies: solid tumors (defined according to the International Classification of Diseases, Seventh Revision [ ICD-7 ] as codes 140–199), non-Hodgkin lymphomas (including chronic lymphocytic leukemia, ICD-7 codes 200, 202, and 204.1), and other hematologic or lymphatic malignancies ( ICD-7 codes 201 and 203–207 [excluding 204.1]). In a post hoc analysis, we stratified the analyses of risk of non-Hodgkin lymphoma by whether the donors performed their first donation before or after HIV screening was initiated (ie, <1986 vs ≥1986).

We also investigated the relationship between iron loss and risk of selected solid tumors (lung, liver, colon, stomach, and esophageal cancers) for which high iron levels have previously been implicated ( 8–10 , 21 , 22 ). Iron loss was categorized according to the median, 75th percentile, and 90th percentile, as less than 1.5, 1.5–3.4, 3.5–5.4, and greater than 5.4 g for the longer exposure window and as less than 0.75, 0.75–1.7, 1.8–2.7, and greater than 2.7 g for the shorter windows.

To assess the potential for confounding by lifestyle factors, we used the Swedish and Danish inpatient registers to investigate, among the control subjects only, associations between donation frequency and incidence of hospitalization for the following diseases that are associated with tobacco and alcohol use: alcoholism, alcoholic hepatitis, and chronic obstructive pulmonary disease. The control donors were followed from the index date until the first occurrence of any of these diseases, death, emigration, or December 31, 2002. Also, among all female donors who had been pregnant after 1983, whose smoking habits were recorded in the Swedish medical birth register, we calculated the number of donations during the most recent 10 years before delivery among smokers and nonsmokers.

Odds ratios (ORs) were used as estimates of relative risks. All data processing and statistical analyses were performed using SAS version 8 or higher (SAS Institute, Cary, NC). All statistical tests were two-sided. P values less than .05 were considered to be statistically significant. The conduct of this study was approved by all regional ethics committees in Sweden, the Danish Scientific Ethics Committee System, and the Danish Data Protection Agency.

Results

We identified a total of 30729 case patients with cancer among donors with a known area of residence. Of these, 13810 case patients had less than 7 years of donation register coverage and were consequently ineligible for the analyses. Another 6053 case patients had not donated blood in the window 3–12 years before diagnosis. For the remaining 10866 case patients, we selected a total of 107140 individually matched control subjects who had all donated blood in the window from 3–12 years before cancer diagnosis of their respective case patient and who also had at least 7 years of register coverage. A total of 4720 (43.4%) case patients were women ( Table 1 ), and the median age at diagnosis of the case patients was 57 years (interquartile range = 48–64 years). Almost three-quarters (73%) of case patients were Swedish. The number of previous donations was similar for the case patients and control subjects (the median number of donations in the windows from 3–7 and 8–12 years before diagnosis was four and five for both case patients and control subjects, respectively). Of the 10866 case patients, 5319 (49%) had at least 12 years of blood donation register coverage.

Table 1

Characteristics of the study subjects *

Characteristic Cancer case patients Matched control subjects 
Total number of individuals 10866 107140 
Site of cancer in case patient, No. (%)   
    Solid cancers 10028 (92.3)  98871 (92.3) † 
    Non-Hodgkin lymphomas 482 (4.4)  4757 (4.4) † 
    Other hematologic/lymphatic malignancies 356 (3.3)  3512 (3.3) † 
Female sex, No. (%) 4720 (43.4) 46492 (43.4) 
Swedish, No. (%) 7949 (73.2) 78553 (73.3) 
Age in y, median (IQR) 57 (48–64) 57 (48–64) 
No. of recorded donations   
    <5 2071 (19.1) 19579 (18.3) 
    6–15 3942 (36.3) 38232 (35.7) 
    16–50 3667 (33.7) 37895 (35.4) 
    >50 1186 (10.9) 11434 (10.7) 
No. of donations, median (IQR)   
    In window 3–7 y 4 (0–9) 4 (0–9) 
    In window 8–12 y 5 (1–9) 5 (1–10) 
Iron loss in g, median (IQR)   
    In window 3–7 y 0.8 (0.0–1.9) 0.8 (0.0–2.0) 
    In window 8–12 y 1.0 (0.2–2.1) 1.0 (0.2–2.1) 
Characteristic Cancer case patients Matched control subjects 
Total number of individuals 10866 107140 
Site of cancer in case patient, No. (%)   
    Solid cancers 10028 (92.3)  98871 (92.3) † 
    Non-Hodgkin lymphomas 482 (4.4)  4757 (4.4) † 
    Other hematologic/lymphatic malignancies 356 (3.3)  3512 (3.3) † 
Female sex, No. (%) 4720 (43.4) 46492 (43.4) 
Swedish, No. (%) 7949 (73.2) 78553 (73.3) 
Age in y, median (IQR) 57 (48–64) 57 (48–64) 
No. of recorded donations   
    <5 2071 (19.1) 19579 (18.3) 
    6–15 3942 (36.3) 38232 (35.7) 
    16–50 3667 (33.7) 37895 (35.4) 
    >50 1186 (10.9) 11434 (10.7) 
No. of donations, median (IQR)   
    In window 3–7 y 4 (0–9) 4 (0–9) 
    In window 8–12 y 5 (1–9) 5 (1–10) 
Iron loss in g, median (IQR)   
    In window 3–7 y 0.8 (0.0–1.9) 0.8 (0.0–2.0) 
    In window 8–12 y 1.0 (0.2–2.1) 1.0 (0.2–2.1) 
*

IQR = interquartile range. Case patients and control subjects (10 per case patient, alive and free of cancer on the date of the respective case patient's cancer diagnosis) were matched on sex, age (±180 days), and county of residence. For some case patients, it was not possible to find 10 control subjects who fulfilled the eligibility and matching criteria, so the ratio of case patients to control subjects is slightly more than 0.1. Percentages may not add to 100 due to rounding.

Cancer site in the corresponding, individually matched case patient.

No clear dose–response relationship was observed between the number of donations in any of the exposure windows and cancer risk overall ( Table 2 ). However, men with at least five donations in the time window of 3–7 years were at a slightly lower risk than men with fewer donations ( Table 2 ). This risk reduction was statistically significant among men who made 9–12 donations ( Table 2 ).

Table 2

RRs, expressed as ORs, for all cancers combined in relation to number of whole blood and plasma donations, by latency and stratified by sex *

Exposure window OR (95% CI) according to No. of donations Ptrend 
3–12 y before diagnosis 1–8 9–16 17–25 >25  
    Both sexes 1.00 (referent) 0.98 (0.91 to 1.05) 0.97 (0.89 to 1.05) 0.99 (0.90 to 1.10) .84 
    Women 1.00 (referent) 0.96 (0.86 to 1.07) 0.99 (0.87 to 1.13) 0.98 (0.83 to 1.17) .78 
    Men 1.00 (referent) 0.99 (0.90 to 1.09) 0.96 (0.86 to 1.07) 1.00 (0.88 to 1.13) .68 
3–7 y before diagnosis 0–4 5–8 9–12 >12  
    Both sexes 1.00 (referent) 0.98 (0.93 to 1.03) 0.90 (0.85 to 0.96) 0.99 (0.93 to 1.06) .46 
    Women 1.00 (referent) 1.02 (0.94 to 1.10) 0.95 (0.87 to 1.05) 1.01 (0.91 to 1.13) .93 
    Men 1.00 (referent) 0.94 (0.87 to 1.01) 0.86 (0.80 to 0.94) 0.97 (0.90 to 1.05) .39 
8–12 y before diagnosis 0–4 5–8 9–12 >12  
    Both sexes 1.00 (referent) 0.99 (0.92 to 1.07) 0.98 (0.90 to 1.07) 1.00 (0.91 to 1.09) .72 
    Women 1.00 (referent) 0.99 (0.89 to 1.11) 1.02 (0.90 to 1.17) 0.97 (0.82 to 1.14) .76 
    Men 1.00 (referent) 1.00 (0.90 to 1.10) 0.96 (0.86 to 1.07) 1.01 (0.90 to 1.13) .82 
Exposure window OR (95% CI) according to No. of donations Ptrend 
3–12 y before diagnosis 1–8 9–16 17–25 >25  
    Both sexes 1.00 (referent) 0.98 (0.91 to 1.05) 0.97 (0.89 to 1.05) 0.99 (0.90 to 1.10) .84 
    Women 1.00 (referent) 0.96 (0.86 to 1.07) 0.99 (0.87 to 1.13) 0.98 (0.83 to 1.17) .78 
    Men 1.00 (referent) 0.99 (0.90 to 1.09) 0.96 (0.86 to 1.07) 1.00 (0.88 to 1.13) .68 
3–7 y before diagnosis 0–4 5–8 9–12 >12  
    Both sexes 1.00 (referent) 0.98 (0.93 to 1.03) 0.90 (0.85 to 0.96) 0.99 (0.93 to 1.06) .46 
    Women 1.00 (referent) 1.02 (0.94 to 1.10) 0.95 (0.87 to 1.05) 1.01 (0.91 to 1.13) .93 
    Men 1.00 (referent) 0.94 (0.87 to 1.01) 0.86 (0.80 to 0.94) 0.97 (0.90 to 1.05) .39 
8–12 y before diagnosis 0–4 5–8 9–12 >12  
    Both sexes 1.00 (referent) 0.99 (0.92 to 1.07) 0.98 (0.90 to 1.07) 1.00 (0.91 to 1.09) .72 
    Women 1.00 (referent) 0.99 (0.89 to 1.11) 1.02 (0.90 to 1.17) 0.97 (0.82 to 1.14) .76 
    Men 1.00 (referent) 1.00 (0.90 to 1.10) 0.96 (0.86 to 1.07) 1.01 (0.90 to 1.13) .82 
*

RR = relative risk; OR = odds ratio; CI = confidence interval. The analyses considering the two exposure windows of 3–12 and 8–12 years were restricted to the 5319 case patients and 52369 control subjects who had at least 12 years of donation register coverage.

The analyses of different groups of malignancies included a total of 10028 donors with solid cancers, 482 with non-Hodgkin lymphomas, and 356 with other hematologic or lymphatic malignancies ( Table 1 ). The association between number of donations and risk of solid malignancies followed much the same pattern as for cancer overall, with a small but statistically significant reduction in risk for donors with 9–12 donations in the 3–7 years before diagnosis (OR = 0.91, 95% CI = 0.85 to 0.97) ( Table 3 ). For non-Hodgkin lymphoma, the data signaled a slightly different pattern, with a suggested but not statistically significant ( Ptrend = .07) increased risk with increasing number of donations in the time window of 3–12 years ( Table 3 ).

Table 3

RRs, expressed as ORs, for solid cancers, non-Hodgkin lymphoma, and other hematologic/lymphatic malignancies in relation to number of donations, by latency *

Exposure window OR (95% CI) according to No. of donations Ptrend 
3–12 y before diagnosis 1–8 9–16 17–25 >25  
    Solid cancer 1.00 (referent) 0.98 (0.91 to 1.06) 0.97 (0.89 to 1.06) 0.99 (0.89 to 1.10) .73 
    Non-Hodgkin lymphoma 1.00 (referent) 0.92 (0.65 to 1.31) 0.89 (0.60 to 1.32) 1.27 (0.84 to 1.94) .07 
    Other 1.00 (referent) 0.87 (0.57 to 1.31) 1.10 (0.68 to 1.76) 0.69 (0.35 to 1.36) .16 
3–7 y before diagnosis 0–4 5–8 9–12 >12  
    Solid cancer 1.00 (referent) 0.98 (0.93 to 1.04) 0.91 (0.85 to 0.97) 0.98 (0.91 to 1.05) .37 
    Non-Hodgkin lymphoma 1.00 (referent) 0.97 (0.75 to 1.25) 0.81 (0.60 to 1.10) 1.24 (0.95 to 1.62) .16 
    Other 1.00 (referent) 0.93 (0.70 to 1.25) 0.80 (0.56 to 1.15) 0.99 (0.70 to 1.42) .29 
8–12 y before diagnosis 0–4 5–8 9–12 >12  
    Solid cancer 1.00 (referent) 1.01 (0.94 to 1.10) 0.99 (0.90 to 1.08) 1.01 (0.91 to 1.11) .71 
    Non-Hodgkin lymphoma 1.00 (referent) 0.75 (0.51 to 1.09) 0.87 (0.59 to 1.28) 1.11 (0.74 to 1.68) .16 
    Other 1.00 (referent) 0.80 (0.52 to 1.25) 1.00 (0.62 to 1.62) 0.56 (0.30 to 1.03) .09 
Exposure window OR (95% CI) according to No. of donations Ptrend 
3–12 y before diagnosis 1–8 9–16 17–25 >25  
    Solid cancer 1.00 (referent) 0.98 (0.91 to 1.06) 0.97 (0.89 to 1.06) 0.99 (0.89 to 1.10) .73 
    Non-Hodgkin lymphoma 1.00 (referent) 0.92 (0.65 to 1.31) 0.89 (0.60 to 1.32) 1.27 (0.84 to 1.94) .07 
    Other 1.00 (referent) 0.87 (0.57 to 1.31) 1.10 (0.68 to 1.76) 0.69 (0.35 to 1.36) .16 
3–7 y before diagnosis 0–4 5–8 9–12 >12  
    Solid cancer 1.00 (referent) 0.98 (0.93 to 1.04) 0.91 (0.85 to 0.97) 0.98 (0.91 to 1.05) .37 
    Non-Hodgkin lymphoma 1.00 (referent) 0.97 (0.75 to 1.25) 0.81 (0.60 to 1.10) 1.24 (0.95 to 1.62) .16 
    Other 1.00 (referent) 0.93 (0.70 to 1.25) 0.80 (0.56 to 1.15) 0.99 (0.70 to 1.42) .29 
8–12 y before diagnosis 0–4 5–8 9–12 >12  
    Solid cancer 1.00 (referent) 1.01 (0.94 to 1.10) 0.99 (0.90 to 1.08) 1.01 (0.91 to 1.11) .71 
    Non-Hodgkin lymphoma 1.00 (referent) 0.75 (0.51 to 1.09) 0.87 (0.59 to 1.28) 1.11 (0.74 to 1.68) .16 
    Other 1.00 (referent) 0.80 (0.52 to 1.25) 1.00 (0.62 to 1.62) 0.56 (0.30 to 1.03) .09 
*

RR = relative risk; OR = odds ratio; CI = confidence interval. The analyses considering the two exposure windows of 3–12 and 8–12 years were restricted to the 5319 case patients and 52369 control subjects for whom we had at least 12 years of donation register coverage.

The data from the Swedish component were further broken down into donations of whole blood or plasma ( Table 4 ). Apart from a small but statistically significant reduction in risk of solid cancers with an increasing number of whole blood donations made 3–7 years before the index date ( Ptrend = .02), the analyses revealed no other associations between whole blood donations and the risk of solid cancers ( Table 4 ). However, in the 3–12 year window, the highest category of plasma donations was associated with an increased risk of non-Hodgkin lymphoma (>25 vs 0 plasma donations, OR = 2.14, 95% CI = 1.22 to 3.74) ( Table 4 ). Although the Ptrend value for dose–risk was .05, the trend was erratic in all exposure windows. No consistent associations were observed between number of whole blood or plasma donations and other hematologic or lymphatic malignancies.

Table 4

RRs, expressed as ORs, for solid cancers, non-Hodgkin lymphoma, and other hematologic/lymphatic malignancies among Swedish blood donors by latency in relation to number of whole blood and plasma donations *

Exposure window OR (95% CI) according to No. of blood donations Ptrend OR (95% CI) according to No. of plasma donations Ptrend 
3 12 y before diagnosis  1–8 9–16 17–25  > 25  0–8 9–25  > 25  
    Solid cancer 1.00 (referent) 1.02 (0.93 to 1.11) 0.99 (0.89 to 1.10) 0.93 (0.81 to 1.08) .49 1.00 (referent) 0.99 (0.83 to 1.20) 1.05(0.90 to 1.22) .89 
    Non-Hodgkin lymphoma 1.00 (referent) 0.91 (0.60 to 1.38) 1.13 (0.73 to 1.75) 1.11 (0.63 to 1.95) .69 1.00 (referent) 0.47 (0.15 to 1.52) 2.14(1.22 to 3.74) .05 
    Other 1.00 (referent) 0.90 (0.54 to 1.50) 1.43 (0.82 to 2.50) 0.84 (0.31 to 2.22) .85 1.00 (referent) 0.72 (0.22 to 2.37) 0.47(0.14 to 1.57) .18 
3 7 y before diagnosis  0–4 5–8 9–12  > 12  0–4 5–12  > 12  
    Solid cancer 1.00 (referent) 0.96 (0.90 to 1.03) 0.93 (0.86 to 1.00) 0.93 (0.85 to 1.01) .02 1.00 (referent) 0.84 (0.70 to 1.01) 1.04(0.92 to 1.17) .99 
    Non-Hodgkin lymphoma 1.00 (referent) 0.85 (0.62 to 1.17) 0.99 (0.71 to 1.38) 1.04 (0.75 to 1.44) .80 1.00 (referent) 2.19 (1.16 to 4.16) 1.42(0.86 to 2.32) .08 
    Other 1.00 (referent) 0.85 (0.58 to 1.24) 0.90 (0.58 to 1.38) 1.19 (0.76 to 1.85) .90 1.00 (referent) 1.03 (0.48 to 2.18) 0.49(0.22 to 1.10) .24 
8 12 y before diagnosis  0–4 5–8 9–12  > 12  0–4 5–12  > 12  
    Solid cancer 1.00 (referent) 1.02 (0.93 to 1.11) 0.98 (0.88 to 1.08) 1.01 (0.90 to 1.14) .85 1.00 (referent) 1.05 (0.83 to 1.32) 0.99(0.84 to 1.17) .72 
    Non-Hodgkin lymphoma 1.00 (referent) 0.97 (0.64 to 1.47) 0.86 (0.54 to 1.37) 1.14 (0.69 to 1.88) .72 1.00 (referent) 0.27 (0.04 to 2.03) 1.80(0.97 to 3.35) .12 
    Other 1.00 (referent) 1.00 (0.59 to 1.69) 1.19 (0.66 to 2.14) 0.86 (0.41 to 1.80) .95 1.00 (referent) 0.97 (0.22 to 4.32) 0.18(0.02 to 1.32) .15 
Exposure window OR (95% CI) according to No. of blood donations Ptrend OR (95% CI) according to No. of plasma donations Ptrend 
3 12 y before diagnosis  1–8 9–16 17–25  > 25  0–8 9–25  > 25  
    Solid cancer 1.00 (referent) 1.02 (0.93 to 1.11) 0.99 (0.89 to 1.10) 0.93 (0.81 to 1.08) .49 1.00 (referent) 0.99 (0.83 to 1.20) 1.05(0.90 to 1.22) .89 
    Non-Hodgkin lymphoma 1.00 (referent) 0.91 (0.60 to 1.38) 1.13 (0.73 to 1.75) 1.11 (0.63 to 1.95) .69 1.00 (referent) 0.47 (0.15 to 1.52) 2.14(1.22 to 3.74) .05 
    Other 1.00 (referent) 0.90 (0.54 to 1.50) 1.43 (0.82 to 2.50) 0.84 (0.31 to 2.22) .85 1.00 (referent) 0.72 (0.22 to 2.37) 0.47(0.14 to 1.57) .18 
3 7 y before diagnosis  0–4 5–8 9–12  > 12  0–4 5–12  > 12  
    Solid cancer 1.00 (referent) 0.96 (0.90 to 1.03) 0.93 (0.86 to 1.00) 0.93 (0.85 to 1.01) .02 1.00 (referent) 0.84 (0.70 to 1.01) 1.04(0.92 to 1.17) .99 
    Non-Hodgkin lymphoma 1.00 (referent) 0.85 (0.62 to 1.17) 0.99 (0.71 to 1.38) 1.04 (0.75 to 1.44) .80 1.00 (referent) 2.19 (1.16 to 4.16) 1.42(0.86 to 2.32) .08 
    Other 1.00 (referent) 0.85 (0.58 to 1.24) 0.90 (0.58 to 1.38) 1.19 (0.76 to 1.85) .90 1.00 (referent) 1.03 (0.48 to 2.18) 0.49(0.22 to 1.10) .24 
8 12 y before diagnosis  0–4 5–8 9–12  > 12  0–4 5–12  > 12  
    Solid cancer 1.00 (referent) 1.02 (0.93 to 1.11) 0.98 (0.88 to 1.08) 1.01 (0.90 to 1.14) .85 1.00 (referent) 1.05 (0.83 to 1.32) 0.99(0.84 to 1.17) .72 
    Non-Hodgkin lymphoma 1.00 (referent) 0.97 (0.64 to 1.47) 0.86 (0.54 to 1.37) 1.14 (0.69 to 1.88) .72 1.00 (referent) 0.27 (0.04 to 2.03) 1.80(0.97 to 3.35) .12 
    Other 1.00 (referent) 1.00 (0.59 to 1.69) 1.19 (0.66 to 2.14) 0.86 (0.41 to 1.80) .95 1.00 (referent) 0.97 (0.22 to 4.32) 0.18(0.02 to 1.32) .15 
*

RR = relative risk; OR = odds ratio; CI = confidence interval. The analyses considered only the Swedish portion of the data; the analyses considering the two exposure windows of 3–12 and 8–12 years were restricted to the 3941 case patients and 38883 control subjects for whom we had at least 12 years of donation register coverage.

To further explore the association between number of plasma donations and risk of non-Hodgkin lymphoma, we stratified the analyses by whether the donors made their first donation before or after HIV screening was initiated. We found that the positive association described above was confined to those who started their donation career in the earlier period (>25 plasma donations in the 3–12 year exposure window: <1986 vs ≥1986, OR = 2.37, 95% CI = 1.20 to 4.67, vs OR = 1.29, 95% CI = 0.39 to 4.30).

We next performed analyses of iron loss in relation to risk of cancers of the liver, lung, colon, stomach, and esophagus ( Table 5 ). In the exposure window 3–7 years before index date, we noted an inverse association between iron loss and risk with a highly statistically significant trend ( Ptrend < .001). This inverse association was seemingly driven by male donors (iron loss: >2.7 vs <0.75 g, OR = 0.70, 95% CI = 0.58 to 0.84) ( Table 5 ). Similarly, in the exposure window 8–12 years before index date, risks were lower, especially among male donors, but none of the point estimates or tests for trend were statistically significant. In the combined exposure window (3–12 years), the point estimates for men and women in the highest donation category were statistically significant, as was the linear trend in male donors ( Ptrend = .03). In supplementary analyses, we recalculated the iron loss analyses after dividing the case patients and control subjects into three groups (cancers of the gastrointestinal tract, lung cancer, and liver cancer). These analyses revealed that the association between iron loss and reduced cancer risk was driven largely by a reduced risk of lung cancer (Supplementary Table 1, available online).

Table 5

RRs, expressed as ORs, for cancers of the lung, liver, esophagus, stomach, and colon in relation to iron loss, by latency and stratified by sex *

Exposure window OR (95% CI) according to iron loss through blood donation Ptrend 
3 12 y before diagnosis   < 1.5 g 1.5–3.4 g 3.5–5.4 g  > 5.4 g  
    Both sexes 1.00 (referent) 0.99 (0.84 to 1.16) 0.92 (0.75 to 1.12) 0.77 (0.59 to 0.99) .06 
    Women 1.00 (referent) 1.20 (0.92 to 1.56) 1.02 (0.71 to 1.45) 0.64 (0.33 to 1.26) .99 
    Men 1.00 (referent) 0.88 (0.71 to 1.08) 0.86 (0.68 to 1.10) 0.76 (0.57 to 1.01) .03 
3 7 y before diagnosis   < 0.75 g 0.75–1.7 g 1.8–2.7 g  > 2.7 g  
    Both sexes 1.00 (referent) 0.99 (0.87 to 1.12) 0.90 (0.79 to 1.04) 0.75 (0.64 to 0.88) <.001 
    Women 1.00 (referent) 1.13 (0.92 to 1.39) 1.06 (0.84 to 1.35) 0.94 (0.66 to 1.32) .87 
    Men 1.00 (referent) 0.92 (0.78 to 1.07) 0.83 (0.70 to 0.99) 0.70 (0.58 to 0.84) <.001 
8 12 y before diagnosis   < 0.75 g 0.75–1.7 g 1.8–2.7 g  > 2.7 g  
    Both sexes 1.00 (referent) 0.92 (0.77 to 1.10) 1.01 (0.84 to 1.22) 0.84 (0.67 to 1.06) .73 
    Women 1.00 (referent) 1.09 (0.83 to 1.44) 1.20 (0.87 to 1.66) 0.73 (0.41 to 1.28) .53 
    Men 1.00 (referent) 0.82 (0.66 to 1.03) 0.92 (0.73 to 1.16) 0.83 (0.64 to 1.07) .46 
Exposure window OR (95% CI) according to iron loss through blood donation Ptrend 
3 12 y before diagnosis   < 1.5 g 1.5–3.4 g 3.5–5.4 g  > 5.4 g  
    Both sexes 1.00 (referent) 0.99 (0.84 to 1.16) 0.92 (0.75 to 1.12) 0.77 (0.59 to 0.99) .06 
    Women 1.00 (referent) 1.20 (0.92 to 1.56) 1.02 (0.71 to 1.45) 0.64 (0.33 to 1.26) .99 
    Men 1.00 (referent) 0.88 (0.71 to 1.08) 0.86 (0.68 to 1.10) 0.76 (0.57 to 1.01) .03 
3 7 y before diagnosis   < 0.75 g 0.75–1.7 g 1.8–2.7 g  > 2.7 g  
    Both sexes 1.00 (referent) 0.99 (0.87 to 1.12) 0.90 (0.79 to 1.04) 0.75 (0.64 to 0.88) <.001 
    Women 1.00 (referent) 1.13 (0.92 to 1.39) 1.06 (0.84 to 1.35) 0.94 (0.66 to 1.32) .87 
    Men 1.00 (referent) 0.92 (0.78 to 1.07) 0.83 (0.70 to 0.99) 0.70 (0.58 to 0.84) <.001 
8 12 y before diagnosis   < 0.75 g 0.75–1.7 g 1.8–2.7 g  > 2.7 g  
    Both sexes 1.00 (referent) 0.92 (0.77 to 1.10) 1.01 (0.84 to 1.22) 0.84 (0.67 to 1.06) .73 
    Women 1.00 (referent) 1.09 (0.83 to 1.44) 1.20 (0.87 to 1.66) 0.73 (0.41 to 1.28) .53 
    Men 1.00 (referent) 0.82 (0.66 to 1.03) 0.92 (0.73 to 1.16) 0.83 (0.64 to 1.07) .46 
*

RR = relative risk; OR = odds ratio; CI = confidence interval. The analyses considering the two exposure windows of 3–12 and 8–12 years were restricted to the 945 case patients and 9312 control subjects for whom we had at least 12 years of donation register coverage.

As an indirect measure of the alcohol consumption and tobacco use among blood donors with different donation activity, we calculated the incidence of alcohol- and tobacco-related diagnoses among control subjects only using the respective nations' inpatient registries (Supplementary Table 2). There were only minor variations in the incidence of alcoholism and alcoholic hepatitis with donation activity. Although the incidence of chronic obstructive pulmonary disease was similar in the two lowest donation categories, the incidence in the two highest categories was nearly 50% lower (Supplementary Table 2). Among the 23471 female donors who had been pregnant in 1983 or later, those who reported being smokers had on average made 5.0 donations in the previous 10 years, compared with 6.6 among nonsmokers.

Discussion

In this case–control study nested within a large cohort of Scandinavian blood donors, we found little support for any important associations between repeated whole blood donations and cancer risk. There was some suggestion that the risk of non-Hodgkin lymphoma might be increased among frequent plasma donors who began donating before 1986, but the dose–risk trend was erratic and the data not very convincing.

Our a priori hypothesis, which was based on previously published observations ( 8–10 , 21 , 22 ), was that iron depletion may be associated with reduced risks of pulmonary, hepatic, esophageal, gastric, and colon cancer. Seemingly in support of this hypothesis, we noted that the risk of these cancers combined was 30% lower among men who fell into the 90th percentile of iron loss in the 3–7 year exposure window than men who had the smallest iron loss in the same period. The dose–risk trend among men in the 3–7 year window was monotonic and highly statistically significant. The absence of a similar risk reduction in women in the same period may be biologically plausible because women regularly dispose of some iron through menstrual blood, and, thus, the loss through blood donation may be less important. As an estimate of the relevance of the reduced cancer risk with iron loss, we estimated the risk difference between the lowest and highest exposure groups using incidence data from the overall cohort of blood donors. We found, among male donors with an accumulated iron loss of less than 0.75 g, that the incidence of the cancers associated with high iron levels was 41.9 cases per 100 000 person-years. Thus, a relative risk estimate of 0.70, comparing the highest exposure category with the lowest, would amount to 12.6 fewer cases of the studied cancers per 100 000 person-years. The data also suggested a lower risk of cancer among male donors who had higher than the median iron loss in the period 8–12 years before the index date, but these estimates were not statistically significant and did not follow a linear trend. We postulated that any deleterious effect of iron overload would be mediated mainly through mechanisms requiring substantial induction time. Therefore, it is surprising that the association with iron depletion was stronger in the 3–7 year window than the 8–12 year window, suggesting a suspiciously short induction and latency time. Although iron has been implicated as a late-stage carcinogen ( 37 ), it is also conceivable that subtle symptoms of cancers or their precursor lesions that are associated with iron levels may develop earlier than most other malignancies and that the 2-year latency—notwithstanding our initial data exploration—might have been insufficient for excluding the possibility that developing cancers may have affected the donation activity in the latter part of the 3–7 year exposure window. It is further possible that the hemoglobin concentration, and therefore also the iron loss, among case patients was affected by the developing malignancy beyond 2 years before the diagnosis. In ad hoc analyses, we found that the association of iron loss with reduced cancer risk was driven largely by a reduced risk of lung cancers. Therefore, the possibility of residual confounding by lifestyle factors such as smoking must also be considered. In fact, a supplementary analysis revealed a small gradient of smoking- and lifestyle-related morbidity in relation to donation intensity, suggesting that some residual confounding by a “healthy donor effect” may exist despite the restriction of our analyses to donors. Furthermore, among female donors who had been pregnant after 1983 and whose smoking habits were recorded in the Swedish medical birth register, the average number of previous donations was higher among smokers than among nonsmokers. The magnitude of confounding by lifestyle, however, is likely to be small and would hardly explain the difference by sex.

To our knowledge, only three studies of cancer occurrence among blood donors have been published previously ( 1 , 4 , 30 ). They all compared donors with nondonors, and their results mainly reflected the healthy donor effect. None appropriately addressed dose–response relationships between number of donations and cancer risk within a donor cohort. Hence, our study is not readily comparable with previous studies.

A noteworthy strength of this study is the well-defined cohort, which includes the majority of blood donors in recent years in Sweden and Denmark (essentially all donors after the register coverage became complete). Virtually complete high-quality population and health registers ensured essentially complete follow-up, and the large sample size allowed us to precisely characterize the overall cancer risk profile of both male and female blood donors. However, the statistical power was clearly insufficient to detect small effects in site-specific analyses. To limit the scope for both type I and type II errors, we restricted our analyses to a limited number of outcome categories that were defined a priori.

Although the blood donation registers were introduced gradually, typically one county at a time, the availability of data on internal migration allowed us to overcome exposure misclassification due to left truncation by considering only individuals who were fully covered by the donation registers throughout the full exposure window. The restriction to individuals who had donated blood at least once in the most recent 12 years reduced the number of study participants considerably, but in this way the reference group included only active donors. Accordingly, the effect of recently introduced screening tests and eligibility criteria, which may otherwise have contaminated the lowest exposure category with former donors who were no longer eligible for donation, was minimized.

Although the median follow-up time exceeded 10 years in the cohort within which this case–control study was nested ( 1 ), an insufficient number of donors were covered by the SCANDAT database for a meaningful assessment of relative risks associated with exposure in a third 5-year window (13–17 years before the index date). Therefore, it must be pointed out that our results do not rule out the possibility of associations with a longer latency than 12 years.

An important limitation of this study is the lack of information about possibly important confounding factors, such as smoking, alcohol consumption, diet, physical activity, anthropometric measures, and occupational exposures. Although the restriction of our study to donors only with internal comparisons across strata that were defined by donation activity should have limited the scope for important confounding by lifestyle factors, the analyses among the control subjects suggested an association between donation frequency and alcohol- and tobacco-related diseases. However, we cannot directly quantify the impact of the suspected confounding.

Given international differences in blood collection practices, involving maximum permitted blood donations annually and iron substitution use, the results from this study may not be entirely generalizable to other settings. Donors in the most heavily exposed group in this study made more than two donations per year and often approached the maximum of three or four for women and men, respectively. It is unknown to what extent these blood donors experienced mitotic stress or depletion of stored iron, but considering that the iron loss in the highest exposure category, which corresponds to an average daily iron loss of at least 1.4 mg, exceeds the normal average daily iron uptake, some level of iron depletion is likely to have occurred ( 38–41 ). Our database did not contain information about whether or not iron supplementation pills were dispensed, let alone whether they were taken. Although iron supplementation might have partly masked a biologic effect, it seems unlikely to have totally canceled or reversed it ( 42 ).

Although whole blood donation has been found to cause a transient, yet noticeable, immunomodulation ( 27–29 , 31 ), the risk elevation we observed for non-Hodgkin lymphoma among frequent plasma donors cannot be readily explained. From the cohort within which the study was nested, we calculated an incidence of 9.4 cases of non-Hodgkin lymphoma per 100 000 person-years among donors with fewer than nine plasma donations. Given a relative risk of 2.14, the incidence rate theoretically translates into an additional 10.7 cases per 100 000 person-years of follow-up. Because this association seemed to be confined to those who began donating before 1986, a time when donor selection underwent major changes due to the risk of HIV transmission, and there was generally a stronger economic incentive for plasma donations than whole blood donations, the finding could conceivably be confounded by some lifestyle-related risk factors. However, if the induction and latency time for non-Hodgkin lymphoma is long, the follow-up period among plasma donors who started donating after 1986 may not yet have been sufficient. The possible link between blood donation and risk of non-Hodgkin lymphoma was one of few a priori hypotheses, but because chance cannot confidently be ruled out in view of multiple testing and the small number of case patients, more studies of the association may be warranted. Our finding might constitute one of few clues to the etiology of this rapidly increasing malignancy.

In conclusion, we found that repeated blood donation was not associated with risk of cancer overall. The overriding hypothesis of a positive association between iron stores and risk of a number of cancers that have previously been associated with iron loss ( 8–10 , 21 , 22 ) gets at least some support from our data; however, residual confounding from unmeasured health-promoting factors that are related to donation activity remains a possibility. A positive association between frequent plasma donation and risk of non-Hodgkin lymphoma must be interpreted cautiously, and further exploration might generate new insights about factors involved in the etiology of this malignancy.

Funding

National Cancer Institute and the National Heart Lung and Blood Institute of the US National Institutes of Health (contract #NO1-CP21175); Swedish Cancer Society (contract 070724).

References

1.
Edgren
G
Tran
TN
Hjalgrim
H
, et al.  . 
Improving health profile of blood donors as a consequence of transfusion safety efforts
Transfusion
 , 
2007
, vol. 
47
 
11
(pg. 
2017
-
2024
)
2.
Casale
G
Bignamini
M
de Nicola
P
Does blood donation prolong life expectancy?
Vox Sang
 , 
1983
, vol. 
45
 
5
(pg. 
398
-
399
)
3.
Casale
G
Bignamini
M
Study of survival in 332 blood donors and 399 non-donors [Italian]
Riv Emoter Immunoematol
 , 
1983
, vol. 
30
 
3
(pg. 
304
-
311
)
4.
Merk
K
Mattsson
B
Mattsson
A
Holm
G
Gullbring
B
Bjorkholm
M
The incidence of cancer among blood donors
Int J Epidemiol
 , 
1990
, vol. 
19
 
3
(pg. 
505
-
509
)
5.
Newman
BH
Donor reactions and injuries from whole blood donation
Transfus Med Rev.
 , 
1997
, vol. 
11
 
1
(pg. 
64
-
75
)
6.
Sullivan
JL
Iron and the sex difference in heart disease risk
Lancet
 , 
1981
, vol. 
1
 
8233
(pg. 
1293
-
1294
)
7.
Sullivan
JL
Blood donation may be good for the donor. Iron, heart disease, and donor recruitment
Vox Sang
 , 
1991
, vol. 
61
 
3
(pg. 
161
-
164
)
8.
Tiniakos
G
Williams
R
Cirrhotic process, liver cell carcinoma and extrahepatic malignant tumors in idiopathic haemochromatosis. Study of 71 patients treated with venesection therapy
Appl Pathol
 , 
1988
, vol. 
6
 
2
(pg. 
128
-
138
)
9.
Selby
JV
Friedman
GD
Epidemiologic evidence of an association between body iron stores and risk of cancer
Int J Cancer
 , 
1988
, vol. 
41
 
5
(pg. 
677
-
682
)
10.
Knekt
P
Reunanen
A
Takkunen
H
Aromaa
A
Heliovaara
M
Hakulinen
T
Body iron stores and risk of cancer
Int J Cancer
 , 
1994
, vol. 
56
 
3
(pg. 
379
-
382
)
11.
Haidari
M
Javadi
E
Sanati
A
Hajilooi
M
Ghanbili
J
Association of increased ferritin with premature coronary stenosis in men
Clin Chem.
 , 
2001
, vol. 
47
 
9
(pg. 
1666
-
1672
)
12.
Kiechl
S
Willeit
J
Egger
G
Poewe
W
Oberhollenzer
F
Body iron stores and the risk of carotid atherosclerosis: prospective results from the Bruneck study
Circulation
 , 
1997
, vol. 
96
 
10
(pg. 
3300
-
3307
)
13.
Salonen
JT
Nyyssonen
K
Korpela
H
Tuomilehto
J
Seppanen
R
Salonen
R
High stored iron levels are associated with excess risk of myocardial infarction in eastern Finnish men
Circulation
 , 
1992
, vol. 
86
 
3
(pg. 
803
-
811
)
14.
de Valk
B
Marx
JJ
Iron, atherosclerosis, and ischemic heart disease
Arch Intern Med
 , 
1999
, vol. 
159
 
14
(pg. 
1542
-
1548
)
15.
Tuomainen
TP
Salonen
R
Nyyssonen
K
Salonen
JT
Cohort study of relation between donating blood and risk of myocardial infarction in 2682 men in eastern Finland
BMJ
 , 
1997
, vol. 
314
 
7083
(pg. 
793
-
794
)
16.
Zacharski
LR
Chow
BK
Howes
PS
, et al.  . 
Reduction of iron stores and cardiovascular outcomes in patients with peripheral arterial disease: a randomized controlled trial
JAMA
 , 
2007
, vol. 
297
 
6
(pg. 
603
-
610
)
17.
Sempos
CT
Looker
AC
Gillum
RF
Makuc
DM
Body iron stores and the risk of coronary heart disease
N Engl J Med
 , 
1994
, vol. 
330
 
16
(pg. 
1119
-
1124
)
18.
Danesh
J
Appleby
P
Coronary heart disease and iron status: meta-analyses of prospective studies
Circulation
 , 
1999
, vol. 
99
 
7
(pg. 
852
-
854
)
19.
Ascherio
A
Rimm
EB
Giovannucci
E
Willett
WC
Stampfer
MJ
Blood donations and risk of coronary heart disease in men
Circulation
 , 
2001
, vol. 
103
 
1
(pg. 
52
-
57
)
20.
Knuiman
MW
Divitini
ML
Olynyk
JK
Cullen
DJ
Bartholomew
HC
Serum ferritin and cardiovascular disease: a 17-year follow-up study in Busselton, Western Australia
Am J Epidemiol
 , 
2003
, vol. 
158
 
2
(pg. 
144
-
149
)
21.
Nelson
RL
Davis
FG
Sutter
E
Sobin
LH
Kikendall
JW
Bowen
P
Body iron stores and risk of colonic neoplasia
J Natl Cancer Inst
 , 
1994
, vol. 
86
 
6
(pg. 
455
-
460
)
22.
Nelson
RL
Iron and colorectal cancer risk: human studies
Nutr Rev.
 , 
2001
, vol. 
59
 
5
(pg. 
140
-
148
)
23.
Stevens
RG
Jones
DY
Micozzi
MS
Taylor
PR
Body iron stores and the risk of cancer
N Engl J Med
 , 
1988
, vol. 
319
 
16
(pg. 
1047
-
1052
)
24.
Hercberg
S
Estaquio
C
Czernichow
S
, et al.  . 
Iron status and risk of cancers in the SU.VI.MAX cohort
J Nutr
 , 
2005
, vol. 
135
 
11
(pg. 
2664
-
2668
)
25.
Albanes
D
Winick
M
Are cell number and cell proliferation risk factors for cancer?
J Natl Cancer Inst
 , 
1988
, vol. 
80
 
10
(pg. 
772
-
774
)
26.
Cohen
SM
Ellwein
LB
Cell proliferation in carcinogenesis
Science
 , 
1990
, vol. 
249
 
4972
(pg. 
1007
-
1011
)
27.
Ieromnimon
V
Kruger
J
Schmidt
R
Sehrbundt
M
Effect of blood donations on the profile of lymphocytic cells
Vox Sang
 , 
1981
, vol. 
41
 
3
(pg. 
165
-
171
)
28.
Lasek
W
Jakobisiak
M
Plodziszewska
M
Gorecki
D
The influence of blood donation on antibody-dependent cellular cytotoxicity (ADCC) in voluntary blood bank donors
Arch Immunol Ther Exp (Warsz)
 , 
1988
, vol. 
36
 
1
(pg. 
37
-
43
)
29.
Lasek
W
Jakobisiak
M
Grochowska
M
Plodziszewska
M
Szczytnicki
W
Two patterns of NK activity changes following blood donation: decrease in the beginners and restoration in regular blood bank donors
Arch Immunol Ther Exp (Warsz)
 , 
1992
, vol. 
40
 
3–4
(pg. 
191
-
194
)
30.
Lasek
W
Jakobisiak
M
Stoklosa
T
Decreased natural killer cell activity in whole-blood donors does not seem to result in increased cancer incidence
Transfusion
 , 
1994
, vol. 
34
 
4
(pg. 
359
-
360
)
31.
Karger
R
Weber
C
Schmidt
J
Kretschmer
V
Characterization of immune system alterations following preoperative autologous blood donation for elective hip replacement surgery
Transfus Med
 , 
2007
, vol. 
17
 
1
(pg. 
45
-
53
)
32.
Högman
CF
Ramgren
O
Computer system for blood transfusion service
Transfusion
 , 
1970
, vol. 
10
 
3
(pg. 
121
-
132
)
33.
Freiesleben
E
Jensen
KG
Tamborg
O
Donorregistrering og elektronisk databehandling
Ugeskr Laeger
 , 
1965
, vol. 
127
 
37
(pg. 
1171
-
1176
)
34.
Edgren
G
Hjalgrim
H
Tran
TN
, et al.  . 
A population-based binational register for monitoring long-term outcome and possible disease concordance among blood donors and recipients
Vox Sang
 , 
2006
, vol. 
91
 
4
(pg. 
316
-
323
)
35.
Storm
HH
Michelsen
EV
Clemmensen
IH
Pihl
J
The Danish Cancer Registry—history, content, quality and use
Dan Med Bull.
 , 
1997
, vol. 
44
 
5
(pg. 
535
-
539
)
36.
Ekström
AM
Signorello
LB
Hansson
LE
Bergström
R
Lindgren
A
Nyren
O
Evaluating gastric cancer misclassification: a potential explanation for the rise in cardia cancer incidence
J Natl Cancer Inst
 , 
1999
, vol. 
91
 
9
(pg. 
786
-
790
)
37.
Weinberg
ED
The role of iron in cancer
Eur J Cancer Prev
 , 
1996
, vol. 
5
 
1
(pg. 
19
-
36
)
38.
Simon
TL
Garry
PJ
Hooper
EM
Iron stores in blood donors
JAMA
 , 
1981
, vol. 
245
 
20
(pg. 
2038
-
2043
)
39.
Milman
N
Sondergaard
M
Iron stores in male blood donors evaluated by serum ferritin
Transfusion
 , 
1984
, vol. 
24
 
6
(pg. 
464
-
468
)
40.
Milman
N
Kirchhoff
M
The influence of blood donation on iron stores assessed by serum ferritin and hemoglobin in a population survey of 1359 Danish women
Ann Hematol
 , 
1991
, vol. 
63
 
1
(pg. 
27
-
32
)
41.
Milman
N
Kirchhoff
M
Influence of blood donation on iron stores assessed by serum ferritin and haemoglobin in a population survey of 1433 Danish males
Eur J Haematol
 , 
1991
, vol. 
47
 
2
(pg. 
134
-
139
)
42.
Garry
PJ
Koehler
KM
Simon
TL
Iron stores and iron absorption: effects of repeated blood donations
Am J Clin Nutr
 , 
1995
, vol. 
62
 
3
(pg. 
611
-
620
)
We are greatly indebted to all blood banks in Denmark and Sweden for collecting and contributing data for this study. The sponsors had no role in the study design, the data collection and analysis, the interpretation of the results, the preparation of the manuscript, or the decision to submit the manuscript for publication.