Trends in Incidence and Prevalence of Major Transfusion-Transmissible Viral Infections in US Blood Donors, 1991 to 1996 FREE

The US blood supply is the safest it has ever been due to a combination of donor education, donor screening, and new laboratory test procedures. Risks of transfusion-transmitted viral infections are extremely low^1- 3: estimated to be 1 in 677,000 units for human immunodeficiency virus (HIV);³ 1 in 641,000 for human T-lymphotropic virus (HTLV);¹ 1 in 103,000 for hepatitis C virus (HCV);¹ and 1 in 63,000 for hepatitis B virus (HBV).¹ Blood centers have implemented education programs and screening procedures aimed at reducing risk of transfusion-transmitted viral infections,⁴ and efforts are under way to improve behavioral screening of donors. Most infectious units are believed to be from donors who donate in the window period, the time between infection and detectability by screening tests. Thus, there has also been continuous effort to develop more sensitive and specific screening tests as exemplified by the introduction of HCV enzyme immunoassay (EIA) 2.0 in 1992, implementation of HIV-1 p24 antigen screening and HCV EIA 3.0 in 1996, and introduction of HCV nucleic acid amplification testing in April 1999.⁵

Monitoring time trends in infectious disease rates in the blood donor population provides a mechanism to assess the safety of the blood supply and the effectiveness of donor deferral criteria and other screening measures. Changes in blood donor infectious disease rates may also reflect changes in population risks. For example, acquired immunodeficiency syndrome (AIDS) incidence rates (IRs), while decreasing in homosexual men, increased in women and racial and ethnic minorities in the mid 1990s.⁶

Monitoring of changes in laboratory tests is crucial because apparent changes in prevalence or IRs may actually be secondary to introduction of a new test. Although the actual infection rate in donors may be constant, use of new screening or confirmatory assays may result in improved detection of infected individuals, an increased number of false-positive results, or both.

We evaluated changes in rates of transfusion-transmissible viral infections (TTVIs) between 1991 and 1996 using test data collected on all allogeneic (community, directed, and apheresis) donations given at 5 US blood centers taking part in the National Heart, Lung, and Blood Institute–sponsored Retrovirus Epidemiology Donor Study (REDS). This study represents an additional 3 years of data compared with a prior report on the incidence of viral markers,¹ allowing analysis of trends.

Data on donation type and date, demographic characteristics (age, sex, race/ethnicity, and education level), and serological test results have been collected by REDS on an ongoing basis since 1991 at 5 blood centers: the American Red Cross (ARC) Biomedical Services Greater Chesapeake and Potomac (Baltimore, Md, and Washington, DC), Southeastern Michigan (Detroit), and Southern California (Los Angeles) Regions; the Blood Centers of the Pacific in San Francisco; and the Oklahoma Blood Institute in Oklahoma City. Westat (Rockville, Md) serves as the coordinating center. REDS compiles information on approximately 8% of all US allogeneic collections. The REDS protocol was approved by the institutional review board at each center.

To assess changes in rates of TTVIs from 1991 to 1996, we analyzed data on 5.9 million allogeneic donations from 1.9 million donors between 1991 and 1996. Serological tests were done according to standard operating procedures at each center and included the following: tests for antibodies to HIV type 1 (types 1 and 2 after March 1992) and tests for HTLV antibody (Ab), hepatitis B surface antigen (HBsAg), hepatitis B core Ab, and HCV Ab.

Confirmatory tests included the HIV-1 Western blot (WB) assay (Cambridge Biotech Corp, Worcester, Mass) or the recombinant immunoblot assay (RIBA, Chiron Corporation, Emeryville, Calif) for HIV; the RIBA 2.0 (Chiron) for HCV; and the monoclonal neutralization assay (Abbott Laboratories, Abbott Park, Ill) for HBsAg. The confirmatory algorithms for HTLV varied by blood centers and changed over time. At ARC centers, both an in-house HTLV viral lysate WB and Cambridge p21e EIA were used as confirmatory tests from January 1991 to November 1993 with the presence of p24 on WB and a p21e EIA sample/cutoff ratio exceeding 6.0 needed for HTLV confirmation.⁷ Radioimmune precipitation assays were performed if the combined results of the first 2 assays were indeterminate. The 6.0 EIA sample/cutoff ratio (rather than a sample/cutoff ratio of 1.0, which is the usual interpretation of a positive EIA) was chosen to minimize false-positive reactivity to the recombinant p21e antigen. This algorithm was replaced in November 1993 by a viral lysate WB that contained an additional p21e recombinant protein antigen (Cambridge p21e spiked WB); the criteria for a positive interpretation of this test followed the manufacturer's instructions. At non-ARC centers, the Diagnostic Biotechnology (Pte) Ltd (Singapore) p21e, rgp-I/II spiked WB 2.3 (April 1994 to November 1995 or December 1996), and the Genelabs Diagnostics Pte Ltd (Singapore) HTLV blot 2.4 (November 1995 to December 1996) replaced the Abbott WB (January 1991 to March 1994). Radioimmune precipitation assays were also performed as needed for HTLV confirmation.

We attempted to sort out false-positive confirmatory results for several viruses. For HIV, the WB interpretive criteria of February 1993 were applied to the 6 years of data, and WBs lacking p31 were further investigated by polymerase chain reaction (PCR) and/or by follow-up serology.⁸ Review of these results led to identification of 5 HIV false-positive donations. For HTLV, we conducted peptide EIA tests (BioChem ImmunoSystèmes Inc, Montreal, Quebec) on HTLV WB-positive samples to discriminate HTLV-I from HTLV-II. If HTLV type was discordant by interpretation of relative strengths of WB bands vs peptide testing, cells from that donor (if available) were sent for HTLV DNA testing by PCR. Eighty-nine percent of 702 WB-positive allogeneic samples were evaluated using this protocol; we identified 57 HTLV peptide EIA- and PCR-negative donations (presumed false-positive), with 52 occurring at ARC centers and 5 at non-ARC centers. For HBsAg, 92 allogeneic donations were identified as possibly false-positive because 1 of their 3 EIA optical densities was 0.2 or less and their Ab to hepatitis B core antigen was negative.⁹ Kleinman et al⁹ determined that at least 70% of HBsAg-neutralized positive tests with these characteristics are negative on HBV DNA PCR and are therefore presumed to be false-positive.

All analyses were conducted using the number of true-positive tests (ie, false-positive tests were considered negative) except for HBsAg. For the latter, we calculated prevalence and IRs in 2 ways: (1) using all HBsAg-confirmed positive donations (this approach assumes that all 92 potential false-positive donations are true-positive and is comparable to analyses performed in other studies) and (2) by converting the 92 potential false-positive test results to negative (this approach assumes that all 92 potential false-positive donations are negative and better approximates the true value).

To assess frequency of infection in new donors, we calculated prevalence of viral markers per 100,000 first-time donations and 95% confidence interval (CI) for each year. Prevalence was defined as number of seropositive first-time donations divided by total number of first-time donations for each year. A donation was defined as first time if the donor specified this was the first time he or she had given blood at that center and there was no history of prior donations. To determine if a temporal trend existed, we used logistic regression analysis (PROC CATMOD)¹⁰ with prevalence as outcome variable and year of donation as independent variable; this trend test evaluated whether change in prevalence rates occurred from 1991 to 1996 by testing for the presence of a common prevalence rate.

We also assessed whether the risk of TTVI had changed between 1991 and 1996 among donors who had previously gone through the screening process (repeat donors). Incidence rates and exact 95% CIs were calculated for each viral marker for 5 overlapping 2-year intervals, which allowed us to compare IRs among similar groups of donors; ie, donors who had given at least 2 allogeneic donations in 2 years, the first donation being nonreactive for the viral marker of interest. In contrast to a method that would derive annual estimates, the number of incident cases obtained in each category was larger, while establishment of 5 categories still enabled us to assess trend. The IR for each 2-year period was derived by dividing the number of seroconverters in each 2-year period by the total person-years at risk in the same period. Person-years were calculated by summing follow-up time over all donors where follow-up time for a donor was the time between their first and last donation in any 2-year interval. For seroconverters, follow-up time was adjusted by assuming that seroconversion occurred halfway between the last negative and first positive donation. Donations were considered seronegative if nonreactive for the viral marker under study, without considering test results for other markers.

For HCV incidence calculation, we only considered data collected during screening with the second generation EIA test (from about March 1992 to June 1996). Because IRs based on the presence of HBsAg underestimate the true hepatitis B rates due to the transient nature of HBsAg, we derived HBV IRs from the number of HBsAg seroconversions, number of interdonation intervals, and total person-years, calculated as follows: estimated HBV IR=[No. of HBsAg seroconversions]/[(5% × total person-years)+(70% × 63 days/365 days × total No. of intervals)].¹¹ Of the HBV incident subjects, 70% were assumed to be transient HBsAg carriers with a 63-day antigenemia period, and 5% were assumed to be chronic HBsAg carriers.^11- 12

A Pearson χ² test of a common IR among the five 2-year intervals was used to test whether any change in IRs could be detected.¹³ Because this statistic requires independent categories, person-years and the number of incident cases in each 2-year interval were further subcategorized into the following: (1) person-years and number of incident cases in common with the preceding 2-year interval, (2) person-years and number of incident cases unique to the 2-year interval, and (3) person-years and number of incident cases in common with the subsequent 2-year interval.

During 1991 to 1996, 1,946,462 donors gave 5,912,681 allogeneic donations at the 5 REDS blood centers. First-time donations represented 19% of total donations with their number decreasing from 215,148 (22% of allogeneic donations) in 1991 to 183,321 (19% of allogeneic donations) in 1996. Most first-time donors were white (75%) and 35 years old or younger (66%). Thirty-one percent had a college degree or higher level of education. More first-time donations were given by men (52%). Over the 6-year period, some shifts in demographic characteristics of first-time donors occurred. There was a gradual increase in percentage of Hispanic and Asian first-time donors (χ²P=.001): 5.8% (n=12,094) were Hispanic donors in 1991 vs 12.6% (n=21,450) in 1996, and 3.9% (n=8065) were Asian donors in 1991 vs 5.8% (n=9924) in 1996. We observed a corresponding decrease in first-time non-Hispanic white donors, with 80.8% (n=169,215) of donors being white in 1991 vs 70.1% (n=119,568) in 1996.

Hepatitis infections were detected at an order of magnitude greater frequency than were retroviral infections among first-time donors (Figure 1). We observed a gradual decline in HCV prevalence in first-time donors, with estimates decreasing from 0.63% in 1992 to 0.40% in 1996 (P<.001, Figure 1). In contrast, the proportion of HBsAg-positive first-time donors (about 0.2%) remained relatively constant (P=.36). The prevalence estimates shown in Figure 1 were calculated treating presumed HBsAg false-positive results as negatives, but similar estimates were obtained without this adjustment.

The prevalence of HIV decreased after 1992 by a factor of 1.5 to 2.0, with estimates varying from 0.030% to 0.028% in 1991-1992 to 0.018% to 0.015% in 1993-1996 (P=.0001 comparing 1991-1992 with 1993-1996, P for trend = .006 for 1991-1996). The apparent increase in HTLV prevalence in 1994-1996 compared with earlier years (from about 0.035% to 0.046%), although not significant (P=.10, Figure 1), warranted further investigation. We calculated the proportion of HTLV donors who were HTLV-I and HTLV-II positive to assess whether the trends differed by virus type. Furthermore, because the revision in HTLV confirmatory testing that occurred at the ARC centers in November 1993 (see "Methods") could have resulted in detection of a larger number of false-positive results, we also compared the proportion of HTLV-positive donors at ARC centers with that at other REDS centers. As shown in Figure 1, HTLV-II is about 3 times more prevalent than HTLV-I in first-time blood donors (24-33 per 100,000 for HTLV-II vs 6-15 per 100,000 for HTLV-I). The average HTLV prevalence estimate obtained at ARC centers between 1994 and 1996 was significantly higher (P<.001, Figure 2) than the average of the remaining HTLV prevalence estimates (ie, prevalence at non-ARC centers between 1991 and 1996 and prevalence at ARC centers from 1991-1993).

A total of 903,228 repeat donors were included in the analysis; 52% were men and 84% were non-Hispanic white. Forty percent had a college degree or higher level of education, and 38% were younger than 35 years (as of December 31, 1996). The percentage of Hispanic and Asian repeat donors increased from 1991 to 1996 (3.6% to 6.0% for Hispanics; 2.0% to 2.7% for Asians).

The overall IRs per 100,000 person-years (and 95% CIs) for 1991 to 1996 were 2.92 (2.26-3.70) for HIV; 1.59 (1.12-2.19) for HTLV; 3.25 (2.36-4.36) for HCV; and 2.66 (2.04-3.41) for HBsAg (with presumed false-positive results considered negative), which translated into an estimated HBV IR of 10.43 (7.99-13.37). (Corresponding estimates with false-positive results considered positive [for comparability to previous analyses {see "Methods"}] were 4.54 [3.72-5.50] for HBsAg and 17.83 [14.60-21.56] for HBV). Incidence rates calculated for each overlapping 2-year interval are presented in Figure 3. About 400,000 donors contributed approximately 1.4 million donations to each 2-year IR calculation. Although IRs appeared to increase for HTLV (from 0.71 to 2.00) and decrease for HIV (from 2.61 to 1.50) and HBV (from 13.9 to 6.9), these trends were not statistically significant (P=.20 for HTLV, P=.14 for HIV, and P=.17 for HBV). Trend was also not statistically significant for HBV calculations conducted with presumed false-positive results considered positive, with HBV IR estimates varying from 21.6 to 13.9 (P=.74). Hence, there was no detectable pattern between 1991 and 1996, with IRs remaining stable for both retroviruses and hepatitis B and C. The small number of incident cases prevented us from evaluating trends in HTLV incidence by type or by center (ARC vs non-ARC).

All blood donors undergo predonation screening that has evolved through the years.¹⁴ Between 1983 and 1992, donors were asked not to donate if they had AIDS-related symptoms or HIV-related risk behaviors including injecting drugs, being a male who had sex with another male (MSM) since 1977, or having a sexual relationship with a prostitute, injection drug user (IDU), or MSM.¹⁴ These deferral procedures led to a rapid decline in risk of transfusion-transmissible HIV infection (from an estimated peak of 1.1% per transfused unit in 1982 in San Francisco¹⁵), with an estimated 90% of high-risk men self-deferring as a result.^15- 16 In 1987, it was estimated that recruitment procedures, education, and self-screening led to deferral of 49 of every 50 HIV-positive persons attempting to donate.¹⁷ In the late 1980s and more formally following recommendation by the Food and Drug Administration in 1992, predonation screening changed from voluntary self-deferral to a system where potential donors are directly asked about specific behavioral risks.⁴ Current screening procedures include reading materials informing potential donors of HIV and other TTVI risk factors; a self-administered questionnaire about risk behaviors, health, and recent international travel history; and direct questioning by trained staff about specific high-risk behaviors.

In the REDS blood centers, HIV prevalence decreased from 0.03% (1991-1992) to less than 0.02% (1993-1996) in first-time donors. In contrast, the prevalence of HIV infection in the US population increased between 1984 and 1992, with 0.3% of the US population (about 1 million people) estimated to be infected as of 1992.¹⁸ First-time donor HIV-1 prevalence at the beginning of this study was about one tenth that of the general population, with a further 50% reduction during the course of the study. This decline in first-time donors may reflect (1) a decline in HIV prevalence in the segment of the population recruited to donate, (2) the continuing increase in the proportion of the population that has been tested for HIV by clinics or their physician, or (3) improved effectiveness of education and screening procedures leading to increased deferral of persons reporting high-risk behaviors.

The most common TTVI in first-time donors is HCV. We saw a decrease in prevalence from 0.63% in 1992 to 0.40% in 1996, consistent with a prior report.¹⁹ The prevalence of HCV was estimated at 1.8% in the general population for 1988-1994,²⁰ which is about 3- to 5-fold higher than what we saw in first-time blood donors. This difference could be due to self-selection of blood donors or reflect the effectiveness of predonation screening. Nationwide, no new transfusion-associated HCV cases have been detected by the Centers for Disease Control and Prevention Sentinel Counties Viral Hepatitis Surveillance System since 1994.²¹ Incidence rates of HCV have decreased since 1989 in the general population, with most of the decline apparently resulting from a corresponding unexplained decrease in newly acquired IDU-associated HCV infection.²²

The decline in HCV prevalence in first-time blood donors is not due to changes in screening or confirmatory testing since all donations were screened by HCV EIA 2.0 and confirmed by HCV RIBA 2.0. New deferral criteria to reduce HIV transmission may have also led to increased deferral of HCV-positive persons, since HIV and HCV share major routes of transmission (except for MSM transmission, which has not been reported for HCV²¹).

Prevalence of HBsAg did not change between 1991 and 1996, with about 0.2% of first-time donors testing positive for HBsAg every year. Bastiaans et al²³ estimated that HBsAg prevalence in first-time donors at 3 ARC Blood Services Regions was 0.26% between September 1977 and December 1978.^23- 24 Hence, prevalence seems to have stayed constant in first-time blood donors since the late 1970s. Prevalence of HBV has also stayed relatively constant in the last 2 decades in the general population. Data from the second (1976-1980) and third (1988-1994) National Health and Nutrition Examination Surveys (NHANES II, NHANES III) indicated no significant changes in prevalence estimates between 1976 and 1994.²⁵ Prevalence of HBV was estimated at 3.6% in NHANES II vs 3.1% in NHANES III in whites, and 13.7% vs 11.9% in blacks. These rates are at least 15-fold higher than those for first-time donors.

Since HBV appears to be stable since the late 1970s in the general and blood donor populations, it is unlikely that changes in predonation screening in the 1980s and 1992 had a major impact on deferral of first-time donors infected with hepatitis B. Although current criteria are likely effective in screening out some persons at high risk for hepatitis B, it is difficult to develop screening criteria sensitive enough to effectively reduce the prevalence of hepatitis B in first-time blood donors and specific enough to avoid unnecessary deferral of healthy donors (no risk factor can be identified in 30% to 40% of HBV cases).²⁶ A widespread hepatitis B immunization program in effect in the United States since late 1992 may lead to a decrease in hepatitis B prevalence in the future²⁵ in first-time donors (by preventing early childhood infection and providing immunity for adolescents and adults before they engage in high-risk behaviors).

On introduction of HTLV screening in 1989, 0.014% to 0.021% of all donations were confirmed positive for HTLV-I or II.²⁷ Seroprevalence rates are higher in high-risk groups such as prostitutes (7%), sexually transmitted disease clinic patients (0.4%), and homosexual men (0.1%).²⁸ A substantial proportion of HTLV infections in the United States is represented by HTLV-II; it is relatively highly prevalent in IDUs (0.4%-17.6%).²⁹ In our study, HTLV prevalence in first-time donors ranged from 0.035% in 1991 to 0.046% in 1996, with significantly higher prevalence at ARC centers after the introduction of a new confirmatory test algorithm³⁰ in November 1993. Although a real increase in HTLV infection in donors at the 3 ARC centers in 1994-1996 cannot be excluded, we believe that a more likely explanation for this increase is the change in the ARC testing algorithm for the following reasons: false-positive results have been previously reported with this algorithm³⁰; the increase in confirmed positives was confined to ARC centers in 1994-1996; most false-positive results identified in this analysis occurred at the 3 ARC centers after November 1993 (47/57) and we suspect that our stringent criteria did not capture all false-positive results (about 11% of WB-positive allogeneic samples could not be evaluated and we interpreted the few [n=16] EIA-positive and PCR-negative results as HTLV-infected); there was no obvious geographical pattern to the increase (we observed an increase at the ARC center located on the West Coast but not at the non-ARC center located in the same geographical area); and the HTLV screening assays at ARC centers were no more sensitive than those used at non-ARC centers.

Repeat donors must not only deny risk behaviors at their predonation screening, but must have tested negative on their prior donation. Thus, TTVI rates in repeat donors would be expected to be lower than in first-time donors. It is also likely that most repeat donors will not suddenly engage in a new high-risk behavior. Our results from donors who gave at least 2 allogeneic donations in 2 years confirm that IRs as described previously^1- 2 for these viral infections are small. Incidence rates calculated for each overlapping 2-year interval did not statistically significantly differ from one another; thus, IRs and consequently transfusion-transmitted risks have remained stable.

This low level of TTVI risk in repeat donors could in part be explained by some failure in the screening process. An estimated 1.9% of acceptable donors surveyed in 1993 admitted to a risk factor that should have resulted in deferral if it had been reported at time of donation.³¹ Units given by these high-risk donors are more likely to be infectious and may not be detected by screening tests (window-period units), thereby raising TTVI risk.

The small number of seroincident donors shows the difficulty of studying rare events and may have prevented us from finding significant trends in IRs or evaluating IRs for HTLV by type or by ARC vs non-ARC centers. The low number of prevalent and incident cases also precluded development of valid models to assess the influence of demographics. To avoid bias that could be introduced by varying sensitivities of different generation HCV assays, we considered only the data collected following implementation of the second generation EIA. Hence, for HCV, the 1992-1993 and 1995-1996 IRs only reflect the rates in donors who gave at least 2 donations during a portion of these intervals.

The lower prevalence estimates in first-time donors vs the general population suggest that screening for behavioral risk factors associated with these major viral infections is effective. Using a combination of sensitive and less sensitive HIV-1 EIAs, Janssen et al³² estimated that 12.6% of HIV-1–infected first-time donors were recently infected. Thus, the recent downward trend in HIV and HCV prevalence in first-time donors may translate into a lower number of infectious window-period units, indicating that the safety of the US blood supply continues to improve. Although these data are reassuring, further improvement in behavioral screening is desirable. Although some strategies may be difficult to implement in light of chronic blood shortages, approaches to improve deferral procedures include the following: (1) increasing education about and awareness of risk factors and the importance of safe blood donations in the general population, (2) ensuring that donor recruiters and sponsors do not inappropriately pressure persons who may be ineligible for donation, (3) decreasing test-seeking behavior by encouraging testing at alternative test sites, (4) further identification or characterization of important risk behaviors, and (5) strategies to reach donors who are reluctant to admit that they engaged in high-risk behaviors. Proper application of deferral procedures could also be strengthened if donor perceptions of privacy and confidentiality were improved, if information was given in a user-friendly format that was clearly understood, and if a validated standardized questionnaire was used by all blood centers.³³ Finally, implementation of more sensitive tests (such as nucleic acid amplification testing for HIV and HCV) that detect infection earlier (ie, reduce the window period) will further decrease risks of transfusion-transmitted viral infections.