Seroprevalence and Demographic Determinants of Human T-Lymphotropic Virus Type 1 and 2 Infections Among First-Time Blood Donors—United States, 2000–2009

Abstract

Background. Human T-lymphotropic virus type 1 (HTLV-1) and HTLV-2 are prevalent at low levels among US blood donors, but recent data on their prevalence is lacking.

Methods. Data on all first-time blood donors in a large network of US blood centers were examined during 2000–2009. HTLV-1 and HTLV-2 antibodies were measured by enzyme immunoassay (EIA) with confirmation by immunofluorescence or recombinant immunoblot. Prevalence rates were calculated, and odds ratios were assessed using multivariable logistic regression.

Results. Among 2 047 740 first-time donors, 104 were seropositive for HTLV-1 (prevalence, 5.1 cases/per 100 000; 95% confidence interval [CI], 4.1–6.1), and 300 were seropositive for HTLV-2 (prevalence, 14.7 cases/per 100 000; 95% CI, 13.0–16.3). The prevalence was lower than reported in the 1990s but stable from 2000 to 2009. HTLV-1 seropositivity was associated with female sex, older age, and black and Asian race/ethnicity. HTLV-2 seropositivity was associated with female sex, older age, nonwhite race/ethnicity, lower educational level, and residence in the western and southwestern United States.

Conclusions. The HTLV-1 and HTLV-2 prevalences among US blood donors has declined since the early 1990s. A higher prevalence of HTLV-2 in the west and southwest may be attributed to endemic foci among Amerindians.

(See the editorial commentary by Cook and Taylor on pages 486–7.)

Human T-lymphotropic virus type 1 (HTLV-1) has been etiologically linked to adult T-cell leukemia/lymphoma, HTLV-associated myelopathy/tropical spastic paraparesis, uveitis, and various neurologic disorders [1]. HTLV-1 infection is endemic in sub-Saharan Africa, the Caribbean and parts of South America, and southern Japan and whereas HTLV-2 infection is endemic among Pygmies in Africa and Amerindians in North, Central, and South America [2–4]. In the United States, injection drug users, American Indians, and African Americans have all been found to have an elevated HTLV-2 seroprevalence [4, 5]. Endemic transmission routes for HTLV-1 and HTLV-2 include vertical transmission from mother to child through breast-feeding, and sexual transmission (especially from index partners with a high proviral load) [6–9]. HTLV infection can be amplified to epidemic levels by contaminated blood transfusions, commercial sex work, and needle sharing among injection drug users [5, 10–12]. Transmission of HTLV-1 and HTLV-2 by blood transfusion is highly efficient, with a seroconversion rate of 27%–63% after exposed to anti–HTLV-1– and anti–HTLV-2–positive cellular blood components, but has largely been prevented by testing donated blood [10, 13, 14].

Because of their large numbers, uniform sampling frame, and availability of testing data with screening enzyme immunoassays, blood donors may serve as a convenient sentinel population when attempting to estimate the prevalence of viral infections. One caveat is that the population prevalence will be underestimated because of self-selection by donors and exclusion by blood banks of unhealthy or risky candidates. HTLV-1 seroprevalence among blood donors was 0.66% in men and 1.02% in women in Japan and ranged from 0.1% to 1.0%, depending on region, in Brazil [2, 15]. Europe has a much lower prevalence, with HTLV-1 and HTLV-2 seroprevalence of 0 cases per 100 000 first-time blood donors in Norway, Denmark, and Finland; 4.8 cases per 100 000 first-time blood donors in France; and up to 53.3 cases per 100 000 first-time blood donors in Romania [16, 17]. A previous study by Murphy et al observed a peak in HTLV-2 age- and sex-specific prevalence among 40–49-year-old men and women, followed by a decrease in older age groups [18]. The authors interpreted these data as indicative of a birth cohort effect in the United States that was probably due to an epidemic of HTLV-2 infection in the 1960s and 1970s related to injection drug use and secondary sexual transmission.

The HTLV-1 and HTLV-2 seroprevalence in the United States has not been measured since multicenter studies by the American Red Cross in 2001 and the Retrovirus Epidemiology Donor Study in 1996 [18–20]. We therefore measured the prevalence and predictors of HTLV-1 and HTLV-2 infection in a similar US blood donor population during 2000–2009.

METHODS

Study Design and Population

We assessed HTLV-1 and HTLV-2 infection and gathered demographic data routinely collected from individuals donating blood at a large network of Blood Systems blood centers, located mainly in the western, southern, and northern United States. Donor history questionnaires are routinely completed at the time of donation from blood donors, and responses are entered into a data warehouse and include information on blood center location and donor age, sex, self-reported race/ethnicity, and the level of educational attainment. Only first-time, nonautologous donors who gave at least 1 allogeneic whole blood donation or blood product between 1 January 2000 and 31 December 2009 were included. First-time blood donors were defined as having no previous donations recorded in the Blood Systems Data Warehouse. Autologous donors were excluded because they donated blood for reinfusion during their own surgery and are not comparable to other donors. Although the Blood Systems Data Warehouse was first established in 1999, it includes a summary of earlier donation data on included donors.

Data were grouped into geographic regions based on blood centers location. The southwest region included the Arizona center; the El Paso, Lubbock, and McAllen, Texas, centers; and the Albuquerque, New Mexico center. The west region included the Reno and Las Vegas, Nevada, centers; and the San Francisco, Central Coast, and Ventura, California, centers. The southeast region included the Meridian and Tupelo, Mississippi, centers; the Lafayette, Louisiana, center, and the Fort Smith, Arkansas center. The north-central region included the Billings, Montana, center; the Rapid City, South Dakota, center; the Bismarck, North Dakota, center; and the Cheyenne, Wyoming center.

Serologic Testing

Blood samples from each blood donor were screened for HTLV-1 and HTLV-2 seropositivity, using the Abbott Prism Anti-HTLV-1/2 enzyme immunoassay (EIA; Abbott Laboratories, Abbott Park, IL). This assay uses antigen-coated microparticles and a chemiluminescent readout. Reactive samples underwent further testing using an alternative test, the Abbott Anti-HTLV I/II EIA (Abbott Laboratories). This assay uses an antigen-coated bead and antibody sandwich technique with detection using a more conventional colorimetric readout. Further supplementary/confirmatory testing was performed on samples that were repeatedly reactive on the screening assays to confirm the presence of HTLV, using an immunofluorescence assay (IFA) developed and performed by the California Department of Health Service, from January 2000 to June 2008, and the Inno-LIA recombinant immunoblot HTLV-1/2 score assay (Innogenetics, Ghent, Belgium), from July 2008 to December 2009 [21]. A blood donation was confirmed to be positive if it was repeatedly reactive on EIA (ie, on at least 2 of 3 occasions) and if it was reactive on supplemental/confirmatory testing.

Statistical Analysis

To determine the prevalence of HTLV-1 and HTLV-2, the number of confirmed HTLV-1– and HTLV-2–positive first-time donors was first divided by the number of first-time blood donors in each year or subgroup (eg, age group and race/ethnicity group). Crude and subgroup-specific prevalences with 95% confidence intervals (CIs) were calculated. Subgroups included sex, age (10-year brackets), race/ethnicity, year of donation, blood center region, and educational levels. χ² tests were performed to determine significant differences in HTLV-1 and HTLV-2 prevalence between subgroups. Age- and sex-specific seroprevalences during 2000–2009 were graphed together with previously published data on first-time US blood donors from 1991 through 1995 [18]. Although the sampling frame for the previous study was different, there was some overlap in the blood centers.

SAS 9.3 and Arc GIS 9.3 were used to construct an HTLV prevalence map. To avoid random variation, counties with donor denominators of <5000 were excluded and colored white on the map. The prevalence of combined HTLV-1 seropositivity, HTLV-2 seropositivity, and untypable HTLV seropositivity was calculated for each county and categorized into quintiles.

Last, we performed a multivariable logistic regression to determine the independent association of various demographic and geographic factors with HTLV-1 and HTLV-2. A separate model for each HTLV type was fit as a full model before using backward stepwise regression to determine significant covariates. The likelihood ratio test was used to assess possible cross-products. All variables with statistically significant univariate associations with HTLV-1 or HTLV-2 were included in the final models (P < .05). All statistical analyses were performed using STATA/IC 11.2 for Windows (StataCorp, College Station, TX) and SAS 9.1.3 Service Pack 4 (SAS Institute, Cary, NC).

RESULTS

Study Population

A total of 2 047 740 first-time blood donors from 2000 through 2009 were included in the final sample population (Table 1). The average age was 32 years, with 75% of these donors <43 years of age. There were similar proportions of male blood donors and female blood donors. Over the 10-year interval, the proportion of minority first-time donors increased from 21.1% of the donor population in 2000 to 37.5% in 2009. Of note, the proportion of Hispanic first-time donors more than doubled, from 9.0% of the donor population in 2000 to 23.4% in 2009, consistent with a 43% increase in the Hispanic population between April 2000 and April 2010 in the US census and blood bank recruitment of underrepresented minorities. Geographically, 68.1% of donations were in the western and southwestern regions of the United States. There was a higher proportion of blood donors with a bachelor's degree, some college, or a technical school background than blood donors with a high school diploma.

Table 1.

Demographic Characteristics and Subgroup Human T-Lymphotropic Virus (HTLV) Seroprevalence Among First-Time Blood Donors, 2000–2009

Characteristics	Donors, No.	Confirmed Positive for HTLV-1 or HTLV-2a		Confirmed Positive for HTLV-1		Confirmed Positive for HTLV-2
		Donors, No.	Donors, %b (95% CI)	Donors, No.	Donors, %b (95% CI)	Donors, No.	Donors, %b (95% CI)
Overall	2 047 740	448	21.9 (19.9–23.9)	104	5.08 (4.10–6.05)	300	14.7 (13.0–16.3)
Sex
Male	1 025 794	144	14.0 [11.8, 16.3]	41	4.00 [2.77, 5.22]	92	8.97 [7.14, 10.8]
Female	1 021 946	304	29.8 [26.4, 33.1]	63	6.16 [4.64, 7.69]	208	20.4 [17.6, 23.1]
Age, y
≤19	638 076	43	6.74 [4.72, 8.75]	9	1.41 [.49, 2.33]	25	3.92 [2.38, 5.75]
20–29	441 581	22	4.98 [2.90, 7.06]	5	1.13 [.14, 2.12]	14	3.17 [1.51, 4.83]
30–39	346 893	51	14.7 [10.7, 18.7]	18	5.19 [2.79, 7.59]	26	7.50 [4.61, 10.4]
40–49	302 973	144	47.5 [39.8, 55.3]	25	8.25 [5.02, 11.5]	106	35.0 [28.3, 41.7]
50–59	199 944	130	65.0 [53.9, 76.2]	34	17.0 [11.3, 22.7]	87	43.5 [34.4, 52.7]
60–69	89 402	40	44.7 [30.9, 58.6]	7	7.83 [2.03, 13.6]	31	34.7 [22.5, 46.9]
≥70	28 871	18	62.4 [33.6, 91.1]	6	20.8 [4.15, 37.4]	11	38.1 [15.6, 60.6]
Race
White	995 884	99	9.94 [7.98, 11.9]	17	1.71 [.90, 2.52]	69	6.93 [5.29, 8.56]
Black	72 997	85	116 [91.7, 141]	24	32.9 [19.7, 46.0]	56	76.7 [56.6, 96.8]
Hispanic	259 177	54	20.8 [15.3, 26.4]	6	2.32 [.46, 41.7]	42	16.2 [11.3, 21.1]
Asian	53 078	18	33.9 [18.3, 49.6]	15	28.3 [14.0, 42.6]	2	3.77 [.00, 8.99]
Other	68 682	43	62.6 [43.9, 81.3]	6	8.74 [1.75, 15.7]	34	49.5 [32.9, 66.1]
Missing data	597 922	149	24.9 [20.9, 28.9]	36	6.02 [4.05, 7.99]	97	16.2 [13.0, 19.5]
Year of donation
2000	226 935	52	22.9 [16.7, 29.1]	12	5.29 [2.30, 8.28]	37	16.3 [11.1, 21.6]
2001	233 706	35	15.0 [10.0, 19.9]	9	3.85 [1.34, 6.37]	25	10.7 [6.50, 14.9]
2002	191 846	40	20.9 [14.4, 27.3]	9	4.69 [1.63, 7.76]	25	13.0 [7.92, 18.2]
2003	186 985	45	24.1 [17.0, 31.1]	7	3.74 [.97, 6.52]	32	17.1 [11.2, 23.0]
2004	195 336	45	23.0 [16.3, 29.8]	8	4.10 [1.26, 6.93]	34	17.4 [11.6, 23.3]
2005	196 318	54	27.5 [20.2, 34.8]	12	6.11 [2.65, 9.57]	38	19.3 [13.2, 25.5]
2006	192 420	37	19.2 [13.0, 25.4]	10	5.20 [1.98, 8.42]	26	13.5 [8.32, 18.7]
2007	202 652	56	27.6 [20.4, 34.9]	15	7.40 [3.66, 11.2]	35	17.3 [11.6, 23.0]
2008	219 124	46	21.0 [14.9, 27.1]	13	5.93 [2.71, 9.16]	26	11.9 [7.30, 16.4]
2009	202 418	38	18.8 [12.8, 24.7]	9	4.45 [1.54, 7.35]	22	10.9 [6.33, 15.4]
Blood center region
Southwest	751 290	143	19.0 [15.9, 22.2]	23	3.06 [1.81, 4.31]	104	13.8 [11.2, 16.5]
West	643 734	231	35.9 [31.3, 40.5]	49	7.61 [5.48, 9.74]	160	24.9 [21.0, 28.7]
Southeast	420 449	61	14.5 [10.9, 18.2]	29	6.90 [4.39, 9.41]	26	6.18 [3.81, 8.56]
North-central	232 256	13	5.60 [2.55, 8.64]	3	1.29 [.17, 3.75]	10	4.31 [1.64, 6.97]
Missing data	11	…	…	…	…	…	…
Education level
High school diploma	542 410	123	22.7 [18.7, 26.7]	27	4.98 [3.10, 6.86]	82	15.1 [11.9, 18.4]
Some college/technical  school	525 979	111	21.1 [17.2, 25.0]	20	3.80 [2.14, 5.47]	82	15.6 [12.2, 19.0]
Bachelor's degree	226 796	25	11.0 [6.70, 15.3]	10	4.41 [1.68, 7.14]	12	5.29 [2.30, 8.28]
Master's degree/doctorate	99 840	17	17.0 [8.90, 25.1]	7	7.01 [1.82, 12.2]	9	9.01 [3.13, 14.9]
Missing	652 715	172	26.4 [22.4, 30.3]	40	6.13 [4.23, 8.03]	115	17.6 [14.4, 20.8]

^a Data include untypable HTLV. Therefore, the number of infected subjects is greater than sum of the number of subjects infected with HTLV-1 and the number infected with HTLV-2.

^b Data are percentage of HTLV-positive first-time donors per 100 000 first-time donors.

Prevalence of HTLV-1 and HTLV-2 Infections

There were 2190 subjects with positive HTLV antibody screening tests, of whom 448 (20.5%) were confirmed to be positive for HTLV-1 or HTLV-2 infection (including confirmed but untypable samples), yielding an overall HTLV-1 or HTLV-2 prevalence of 21.9 infections per 100 000 first-time US blood donors (95% CI, 19.9–23.9), with 104 donors confirmed to be positive for HTLV-I infection (prevalence, 5.1 per 100 000; 95% CI, 4.1–6.1), 300 donors confirmed to be positive for HTLV-2 infection (prevalence, 14.7 per 100 000; 95% CI, 13.0–16.3), and 44 donors confirmed to be positive for untypable HTLV (prevalence, 2.1 per 100 000) during the 10-year period. No secular trends in prevalence were observed during the 10-year interval (Table 1).

Age- and sex-specific HTLV seroprevalences showed interesting patterns when compared to previously published blood donor data from 1991 through 1995. For HTLV-1, seroprevalence was higher among females and those in older age groups (Table 1 and Figure 1A). Compared with data from 1991 through 1995, both male and female prevalences retained an increase with age, although the slopes of the curves were more shallow. For HTLV-2, the seroprevalence among individuals aged ≥40 years was greater among female blood donors, compared with their male counterparts (Figure 1B). Compared with 1996 data, the increased prevalence among females occurred in the 40–49-year-old group instead of the 30–39-year-old group, and there was no decline among women aged ≥50 years. Similarly, the male prevalence showed less of a middle-aged peak than it did in 1996, although the prevalence was still increased among men aged >50 years.

Figure 1.

Age- and sex-specific seroprevalence of human T-lymphotropic virus type 1 (HTLV-1) and HTLV-2 among first-time blood donors in the current study and previously published data [18]. Age-specific seroprevalences per 100 000 donors for men (♦) and women (▪) in the 1991–1995 study and for men (▴) and women (×) in 2000–2009. The scale of the y-axis in panel A differs from that in panel B.

Ultimately, women had nearly twice the prevalence of HTLV-2 infection as men (Figure 1B). The 2 HTLV types showed different patterns according to race/ethnicity, with the HTLV-I seroprevalence highest among blacks and Asians and the HTLV-2 seroprevalence highest among blacks, Hispanics, and individuals who reported “other” as their race/ethnicity. HTLV-1 seroprevalence differed little by educational attainment, but HTLV-2 seroprevalence was inversely associated with education level. The unadjusted prevalence of confirmed HTLV-1 infection was highest in the western and southeastern regions, whereas the prevalence of HTLV-2 infection was highest in the western and southwestern regions.

Figure 2 illustrates the geographic distribution of the HTLV-1 and HTLV-2 prevalences among blood donors. Scattered counties in Mississippi, Alabama, North Dakota, and Texas had the highest HTLV-1 prevalence. However, a high prevalence of HTLV-2 infection had a more regional distribution, with counties in Nevada, Arizona, New Mexico, and Texas showing increased values.

Figure 2.

Human T-lymphotropic virus type 1 (HTLV-1; A) and HTLV-2 (B) prevalence, by US county of donor residence, 2000–2009. Counties with <5000 donors or no data are shown in white. The prevalence ranges differ between the 2 maps.

Multivariable Logistic Regression

HTLV-1 infection was independently associated with female sex, increasing age, and black and Asian race/ethnicity (Table 2). There was a steady increase in the odds of seropositivity with age, although CIs were quite wide for estimates for the oldest age groups. Donors who were black or Asian were >20 times as likely to be HTLV-1 seropositive, compared with donors who were white. There was a borderline association of HTLV-1 seropositivity with residence in the western United States.

Table 2.

Multivariable Logistic Model of Factors Associated With an Increased Odds of Human T-Lymphotropic Virus Type 1 Infection

Factor	OR (95% CI)
Sex
Male	1.00
Female	1.56 (1.05–2.32)
Age, y
≤19	1.00	…
20–29	0.82	[.27, 2.44]
30–39	3.89	[1.74, 8.66]
40–49	7.10	[3.31, 15.3]
50–59	16.6	[7.94, 34.8]
60–69	9.44	[3.48, 25.6]
≥70	27.7	[9.69, 79.0]
Race/ethnicity
White	1.00	…
Black	25.3	[13.1, 48.7]
Hispanic	2.59	[.98, 6.86]
Asian	21.4	[10.3, 44.5]
Other	7.46	[2.89, 19.2]
Missing data	3.99	[2.24, 7.12]
Region
North-central	1.00	…
Southwest	1.90	[.56, 6.41]
West	3.12	[.96, 10.2]
Southeast	2.95	[.88, 9.98]

Abbreviations: CI, confidence interval; OR, odds ratio.

HTLV-2 infection was independently associated with female sex, older age, black race/ethnicity, Hispanic race/ethnicity, other race/ethnicity, lower educational attainment, and residence in the western and southwest United States (Table 3). Women were twice as likely to be HTLV-2 seropositive. The odds of HTLV-2 infection increased substantially in the 40–49-year-old group and remained relatively flat in older age groups. Black donors were, again, >20 times as likely to be infected; smaller associations were seen for donors with a self-reported race/ethnicity of “other” or Hispanic. There was a marked inverse association of HTLV-2 infection with age. Donors residing in the western United States had a 4-fold increased HTLV-2 prevalence, and donors in the southwestern United States had a >2-fold increased HTLV-2 seroprevalence, compared with those living in the north-central United States.

Table 3.

Multivariable Logistic Model of Factors Associated With an Increased Odds of Human T-Lymphotropic Virus Type 2 Infection

Effect	OR (95% CI)
Sex
Male	1.00
Female	2.13 (1.67–2.73)
Age, y
≤19	1.00	…
20–29	1.11	[.57, 2.15]
30–39	2.81	[1.61, 4.93]
40–49	14.2	[9.08, 22.3]
50–59	19.2	[12.1, 30.5]
60–69	16.6	[9.68, 28.6]
≥70	18.8	[9.13, 38.7]
Race/ethnicity
White	1.00	…
Black	21.5	[14.8, 31.1]
Hispanic	2.93	[1.94, 4.44]
Asian	0.67	[.16, 2.73]
Other	7.63	[5.00, 11.6]
Missing data	1.36	[.83, 2.23]
Education level
High school diploma	1.00	…
Some college/technical school	0.73	[.53, 1.00]
Bachelor's degree	0.22	[.12, .40]
Master's degree/doctorate	0.31	[.15, .61]
Missing data	1.24	[.77, 2.00]
Region
North-central	1.00	…
Southwest	2.47	[1.28, 4.78]
West	4.10	[2.16, 7.82]
Southeast	0.68	[.32, 1.44]

Abbreviations: CI, confidence interval; OR, odds ratio.

DISCUSSION

We report a downward secular trend in the combined prevalence of HTLV-1 and HTLV-2 infections, from 31.4 cases per 100 000 donors during 1991–1995 to the current prevalence of 21.8 cases per 100 000 donors [18]. However, during the 10-year period of the study, from 2000 through 2009, there was little evidence for a continuing downward secular trend. Demographic associations were similar to those of earlier reports: females and individuals reporting a race/ethnicity of “Black, Hispanic or Asian” had an elevated prevalence of HTLV-1 and HTLV-2 infections, and educational attainment was inversely associated with both HTLV-1 and HTLV-2 seropositivity. Finally, this study provides additional evidence for a geographic clustering of HTLV-2, with West Coast and southwest blood donors more likely to be seropositive for HTLV-2 [18].

In the study by Murphy et al from 1991 to 1995, the high prevalence of HTLV-2 among individuals then aged 40–49 years suggested a birth cohort effect, consistent with a 30-year-long epidemic of HTLV-2 infection among blood donors in western United States, primarily due to needle sharing among injection drug users and secondarily due to sexual transmission from heterosexual men to women [22, 23]. A birth cohort effect is used to describe variations in the prevalence or incidence among individuals who are defined by some shared temporal experience or common life experience, such as year of birth or year of exposure. The epidemic of injection drug use in the 1960s and 1970s was first reported by unpublished data from the US National Household Survey on Drug Abuse, suggesting that the same birth cohort experienced a high prevalence of lifetime injection drug use [24]. With our current data, obtained during 2000–2009, we now have further evidence of a birth cohort effect, as seen in a shift in the age group with the highest prevalence of HTLV-2 infection, from 40–49 years during 1991–1995 to 50–59 years during 2000–2009 (Figure 1B).

Geographically, blood donors in the west and southwest had a significantly higher prevalence of HTLV-2 infection than those in the southeast and north-central United States. Previous studies have attributed this phenomenon to the epidemic of HTLV infection among injection drug users along the West Coast in the 1960s and 1970s [18]. Another explanation for the higher prevalence of HTLV-2 infection in the west and southwest could be the presence of Amerindians with endemic HTLV-2 infection This hypothesis is supported by the association of HTLV-2 with “other” and Hispanic race/ethnicity because both categories are likely to harbor more individuals of Amerindian ancestry. Apparent clustering of HTLV-2 in the southwestern region, where many Amerindians reside, also supports this hypothesis of Amerindian origin.

The preponderance of HTLV-2 infection over HTLV-1 infection has been observed in most but not all prior studies involving the US blood donor population [25, 26]. At more than twice the prevalence of HTLV-1 infection, the prevalence of HTLV-2 infection was substantially higher among women than men and increased steadily with age (Figure 1B), likely because of sexual transmission from injection drug users or other HTLV-2 risk groups. The female excess in HTLV-2 may also be attributed to the likely possibility that more female partners of male injection drug users are able to pass donor screenings at blood centers when undergoing blood donation [27].

In contrast to the prevalence of HTLV-2 infection, the prevalence of HTLV-1 infection is lower across all age groups in the current study than in the 1990s, and there is little evidence for birth cohort effect (Figure 1A). We believe that for unknown reasons HTLV-1 did not substantially enter the injection drug use population. Instead, these data support the concept of low-level endemicity of HTLV-1 in the United States (ie, a low HTLV-1 infection prevalence that is maintained through vertical and sexual transmission).

We observed several associations between HTLV infection prevalence and nonwhite race. Asian race was associated with a significantly higher prevalence of HTLV-1 infection (Table 1), which may be due to immigration to the United States from HTLV-1 hyperendemic areas, such as Japan [28]. The higher prevalence of HTLV-1 infections among black blood donors may also be attributable to immigration from HTLV-1–endemic Caribbean or sub-Saharan African countries. In contrast, the high prevalence of HTLV-2 infection among black donors is more likely due to a higher prevalence of injection drug use during military service in Vietnam or in the United States [29]. Liu et al reported an epidemic of HTLV-2 restriction fragment length polymorphism subtype a0 infection independently among individuals aged ≥30 years and among black populations across the United States [30]. Conversely, HTLV-2 RFLP subtypes (b4 and b5) were predominantly found in HTLV-2–seropositive American Indians in Oklahoma [30]. Among whites, HTLV-1 and HTLV-2 seroprevalence remains much lower than in other racial and ethnic groups (Figure 1A).

An inverse relationship was observed between infections with HTLV-1 and HTLV-2 and level of educational attainment. Lower socioeconomic status is likely a surrogate for risk behaviors, such as injection drug use or risky sexual behavior, and lack of access to preventive measures. Lower social class as measured through rural residence or educational attainment has been associated with a higher risk of HTLV-1 seropositivity among Jamaican men and women and with greater HTLV-1 and HTLV-2 seropositivity among Brazilian blood donors [31, 32].

The identification of demographic and geographic risk factors for HTLV-1 and HTLV-2 infections in a large national sample of blood donors allows cautious extrapolation of our findings on the HTLV-1 and HTLV-2 infection prevalences to the US general population. Blood donors are subject to selection for good health and low risk behaviors and therefore would be expected to have a 2–5-fold lower prevalence of HTLV infection. Nonetheless, in the absence of population-based serological surveys, our findings on demographic associations and a decreasing secular trend are likely to be applicable to the broader US population. In addition, these data are valuable for international comparisons because studies blood donors are often reported.

Strengths of the study include its large sample size, broad geographic scope, and availability of computerized data from and laboratory testing in a Food and Drug Administration–regulated blood bank environment. An additional strength of the study is the use of 2-stage EIA screening and immunofluorescence assay for repeatedly reactive confirmatory testing; studies relying only on EIA screening tend to overestimate the HTLV infection prevalence [33, 34]. However, some limitations must be taken into account. First, because we relied on administrative data, we did not have information on risk behaviors. Second, as noted above, blood donors are subject to selection by the blood bank and by self-selection for good health and altruism, introducing bias and likely underestimation of the HTLV seroprevalence in the general population. Finally, comparisons with the previously published study are limited by differences in the blood centers included in the 2 analyses.

In conclusion, the prevalence of HTLV-1 and HTLV-2 infection among US blood donors has declined since the 1990s but appears to have stabilized at lower levels over the past decade. With 3.2 million first-time donors annually, we estimate that US blood banks still detect almost 700 HTLV infections per year. This is both a benefit to public health and a potential burden on the blood centers because of the need for accurate counseling and medical referral. Further studies of the origin of HTLV-2 infection in the United States may wish to concentrate on the potential link between Amerindians in the southwestern United States. Finally, despite recent market withdrawals of HTLV test kits, blood centers will continue to require Food and Drug Administration–licensed and internationally licensed screening and confirmatory serologic assays for detection of HTLV-1 and HTLV-2.