Abstract
Nucleic acid testing (NAT) of blood donors provides opportunities for identifying West Nile virus (WNV)-infected persons before symptoms develop and for characterizing subsequent illness. From June 2003 through 2008, the American Red Cross performed follow-up interviews with and additional laboratory testing for 1436 donors whose donations had initial test results that were reactive for WNV RNA; 821 of the donors were subsequently confirmed to have WNV infection, and the remainder were unconfirmed or determined to have false-positive results. Symptoms attributed to WNV infection were determined by comparing symptom frequency among 576 donors identified with early WNV infection (immunoglobulin M antibody negative) and those with unconfirmed infection. We estimate that 26% of WNV-infected persons become symptomatic, defined by the presence of at least 3 of 8 indicator symptoms. Nearly one-half of symptomatic persons sought medical care; only 5% received a diagnosis of WNV infection. Female subjects and persons with higher viral loads detected in the index donation were more likely than other subjects to develop symptoms.
The constellation of symptoms referred to as West Nile fever (WNF) is by far the most commonly recognized clinical manifestation of West Nile virus (WNV) infection [1–3]. Although studies indicate that <1% of persons infected develop neuroinvasive disease [4–6], a serological survey conducted after the 1999 New York City outbreak indicated that 21% develop febrile illness after WNV infection [6]. However, this estimate was based on the identification of only 6 (32%) of 19 seropositive persons reporting recent febrile illness, compared with 70 (11%) of 648 seronegative participants.
The subsequent implementation of WNV blood donation screening by nucleic acid test (NAT) in the United States and Canada began in June 2003 and provides a unique opportunity to identify many WNV-infected persons very soon after infection and usually before symptom onset, thus eliminating the sample size limitations of serological surveys and the biases of clinical case detection [4, 7–12]. Three follow-up studies involving NAT-positive blood donors have studied the frequency, symptoms, and risk factors for WNF [11–13]. Brown et al [12] estimated that 30% of infected persons became symptomatic, but the study lacked a control group; Orton et al [11] estimated that 41% became symptomatic (61% minus 20% among controls); Custer et al [13] found that 34% of infected persons and 20% of controls had multiple symptoms that were compatible with WNF. These figures varied in part because of different definitions of WNF and different time periods studied surrounding the index donation. One study found that lower age and higher viral load independently correlated with symptomatic infection [12]; however, another study failed to identify a relationship between age and symptomatic infection [13].
The study reported here extends the work of Orton et al [11] at the American Red Cross (ARC) to further define the frequency of WNF-related symptoms, to define demographic and virologic factors associated with symptom development, and to examine health care seeking behavior of WNV-infected persons. This study represents, to our knowledge, the largest collection of WNV infections among otherwise healthy adults across different demographic groups and geographic areas that has ever been studied.
Methods
General approach. The study population was drawn from WNV RNA-reactive blood donors identified during routine blood donation screening from June 2003 through 2008, including those who were previously reported by Orton et al [11]. RNA-reactive donors were asked to return for a follow-up blood sample and interview. Based on subsequent laboratory testing, these donors were classified as WNV-confirmed (true positive) or unconfirmed (ie, donors with samples that had false-reactive test results during routine blood donation screening). Interviews using standardized questionnaires were conducted before completion of laboratory testing and prior to receipt of confirmatory results by donors, thus allowing a relatively unbiased assessment of symptoms, because donors did not know of their WNV confirmatory status at the time of interview. Symptoms were then compared among donors with confirmed WNV infection and donors without confirmed infection. Risk factors for symptom development among the donors with confirmed WNV infection, including viral load of the index donation, were analyzed.
Donor identification and laboratory testing. Since 2003, ARC blood donors have undergone WNV NAT screening using transcription-mediated amplification (TMA; Gen-Probe and Novartis) in minipools (MPs) of 16 donations [7, 11]. Reactive pools were resolved by individually testing each donation sample comprising the pool. Routine individual donation (ID) NAT was implemented in place of MP NAT in areas where reactive donations, defined as likely to confirm, exceeded an established trigger [7]. The triggers used to convert from MP to ID NAT in response to ongoing WNV activity progressively became more sensitive, such that currently 1 reactive donation having a high signal in a reactive MP is used to convert a defined geographic location to ID NAT [14–17]. All TMA-reactive index samples identified by ID or MP NAT were also tested for WNV RNA by research-based qualitative and quantitative polymerase chain reaction (PCR) assays (National Genetics Institute). Frozen plasma components from all TMA-reactive donations were retrieved, aliquots were prepared, and samples were tested for RNA (by PCR and TMA in replicates of up to 10) and for WNV-specific antibodies. WNV antibody testing was performed with a research immunoglobulin (Ig) M assay (Abbott Laboratories) in 2003 and a combination of US Food and Drug Administration-cleared tests for IgM and IgGstarting in 2004 (Focus Technologies) according to methods described elsewhere [7]. Follow-up samples collected from consenting TMA-reactive donors were also tested for RNA (by TMA and PCR) and WNV antibodies.
Donors whose index donation samples were reactive on initial screening for WNV RNA by TMA were confirmed to have WNV infection (confirmed or true positive) through replicate RNA testing by TMA and PCR of the index donation or through observed seroconversion in follow-up testing [7, 11]. False-positive donors lacked RNA reactivity upon replicate testing of samples from the index donation using 2 different methods (TMA and PCR) and did not have IgM antibodies detected in the index donation or did not seroconvert when followed. The sensitivity of the confirmatory testing algorithm, based on index donation results by TMA, PCR, or IgM testing and verified by subsequent seroconversion (IgM or IgM with IgG), was 99% [18]. Follow-up samples were collected and questionnaires were completed within 120 days after the index donation.
Epidemiological investigation. TMA-reactive donors were notified of their initial test results by letter and were contacted at the same time to schedule a follow-up visit. The follow-up visit consisted of collection of a blood sample and a face-to-face interview conducted by trained donor counselors or donor center physicians with use of standardized questionnaires. The questionnaires gathered information regarding demographic characteristics and the presence of 14 symptoms consistent with WNV infection on the day of and/or 2 weeks after the index donation. The interviewers, using a standardized 25-question survey, also queried donors about medical care seeking, including visiting a doctor as a result of WNV-related symptoms, hospitalization because of those symptoms, and whether a WNV infection was specifically diagnosed.
Donors did not know their WNV confirmatory status at the time of interview. The final study population considered for analysis consisted of donors with completed questionnaires who either did not have confirmed infection or had confirmed WNV infection and tested IgM-antibody negative at the index donation. Because donors with confirmed WNV infection who lacked IgM antibody at the index donation donated during the earliest phase of infection and were most likely to have donated before symptoms would have developed [19], including only these donors in the study minimized potential bias attributed to symptomatic individuals who failed to donate. Preliminary analysis did, in fact, demonstrate lower post-donation symptom frequency among IgM-positive (15% reported ≥1 symptom), compared with IgM-negative (53% reported ≥1 symptom), donors with confirmed WNV RNA in their samples.
Data analysis was performed with SAS software (SAS Institute) [20]. Comparison of categorical variables was assessed using the x2 test. Viral load was compared with use of analysis of variance. Multivariate analysis was performed by logistic regression [21].
The ARC Institutional Review Board reviewed and approved all aspects of the study, including donor notification and consent forms for follow-up sampling and administration of the questionnaire.
Results
Study population. From June 2003 through 2008, 1997 blood donors had samples that were initially reactive for WNV RNA by TMA; subsequent laboratory testing confirmed 1095 (55%) to have WNV infection, and 902 (45%) did not have infection confirmed or were found to have false-positive test results (Figure 1). A total of 966 (88%) of the 1095 donors had infection confirmed through replicate RNA testing by TMA and PCR at the time of the index donation, 116 (11%) had infection confirmed through IgM testing at the time of the index donation, and 13 (1%) had infection confirmed through IgM testing of a follow-up donation up to 63 days after the index donation (Table 1). The 1095 donors with confirmed WNV infection represented approximately one-half of the >2000 WNV-infected donors reported by all blood centers in the United States during the study period.
Completed surveys were received from 1436 (72%) of the 1997 donors. Donors confirmed as WNV RNA positive were more likely to have completed the survey (821 donors; 75%) than were donors who were not confirmed as being WNV RNA positive (615 donors; 68%; odds ratio [OR], 1.4; 95% confidence interval [CI], 1.2–1.7) (Figure 1). Seventy-eight percent of the interviews occurred within 90 days after the date of the index blood donation. There was no systematic difference in the distribution of intervals from index donation to interview between donors who were confirmed as being WNV RNA positive and those who were not confirmed as being positive (Figure 2). The frequencies of reported symptoms did not vary by duration of the follow-up interval. The final study population consisted of 576 donors who were confirmed to be positive for WNV infection and who were IgM-negative at the index donation (subjects) and 615 donors without confirmed infection (controls) (Figure 1). The remaining 245 of 821 donors with confirmed WNV infection had index donation samples that were positive for IgM. Subjects were more likely to be male (332 [58%] of 576) than were controls (292 [47%] of 615; P < .01).
Symptom frequency and association with WNV infection. The frequency of each symptom attributable to WNV infection was calculated by subtracting the frequency with which each symptom was reported among the 615 control donors from the frequency with which the symptom was reported among the 576 subject donors. Eight reported symptoms (new rash, generalized weakness, headache, severe muscle pain, joint pain, fever, chills and painful eyes) had an attributable frequency of >10% and had an odds of being reported by subjects that was at least 5 times the odds of being reported by controls (Table 2). Of the 576 subject donors, 270 (47%) reported none of the 8 indicator symptoms, with 85 (15%), 54 (9%), 75 (13%), 39 (7%), 23 (4%), 18 (3%), 7 (1%), and 5 (1%) reporting 1 to 8 symptoms, respectively. The symptoms were not independent from each other, judged by x2 analyses of each 2-symptom pair or through multivariate logistic regression analyses among subject donors, although some symptoms were more closely associated with each other (such as fever and chills or generalized weakness and severe muscle pain) than were others.
Symptomatic donors (cases) were defined as having ≥3 of the 8 symptoms; 167 (29%) of the 576 subjects and 20 (3%) of the 615 controls met this definition, of which the most commonly observed triad among subjects was headache, generalized weakness, and fever (55; 9.5%). Among the 167 cases, only 94 (56%) reported fever (Table 3). We estimate that 26% (29% minus 3%) of persons infected with WNV develop symptoms that meet the case definition of ≥3 of the 8 indicator symptoms due to the infection. However, it is noteworthy that 53% (306) of the subjects developed ≥1 of the 8 symptoms, compared with 11% (69) of the controls (OR, 9.0), which suggests that up to 42% of infections may result in symptoms.
Demographic characteristics and viral load. Among the 576 subjects, females (85 [35%] of 244) were more likely than were males (82 [25%] of 332) to have met the case definition for symptomatic infection (P < .01; OR, 1.6; 95% CI, 1.1–2.3). There was no consistent relationship between donor age and the development of symptoms (P < .05 among both males and females), and the females appeared to be more likely than males to develop symptoms regardless of age (Table 4).
Subjects meeting the case definition had higher viral loads detected in the index donation sample, compared with those not meeting the case definition (P < .01; Table 5). Viral loads were similar among males and females but differed among age groups (P=.02); donors 20–29 years of age had the lowest viral load (median viral load, 795 copies/mL; maximum viral load, 220,000 copies/mL). There was no linear trend of viral load with age for all other age groups (median viral load, 2035–7750 copies/mL with a maximum of 650,000 copies/mL). Logistic regression analysis of the relationship between viral load detected in the index donation, sex, and age and meeting the case definition showed that the odds of meeting the case definition were again significantly higher in women (OR, 1.7; 95% CI, 1.2–2.4) and increased ∼1.5 times for every 10-fold increase in viral load (OR, 1.5; 95% CI, 1.2–1.8); age was not statistically significant (P=.43).
Medical care seeking among donors with confirmed WNV infection. Of the 167 subject donors who met the case definition during the 2 weeks after blood donation, 73 (44%) reported having visited a doctor because of their WNV-related symptoms, and 5 (3%) were hospitalized. Among the 615 controls, 38 (6%) reported having visited a doctor because of their symptoms, and 6 (1%) were hospitalized; none received a diagnosis of WNV infection. These results suggest that 38% (44% minus 6%) of donors who met the case definition had a doctor visit attributable to WNV infection, and similarly, 2% (3% minus 1%) were hospitalized as a result of their symptoms. Among the 73 patients who reported seeking medical care, only 4 (5%) of the 5 hospitalized patients received a diagnosis of WNV infection, and none received a diagnosis of neuroinvasive disease. Because interviews were conducted before final confirmatory notification (that is, donors did not know their final WNV test results at the time of the interview), all donors who sought medical care should not have been influenced by a definite diagnosis of WNV infection. Therefore, comparing subject donors who met the case definition with control donors, as was done in this study, provides an estimate of the proportion of WNV-infected individuals who seek medical care.
Discussion
This study suggests that 26 percent of persons infected with WNV develop symptoms that meet our case definition and are attributable to the infection. Our case definition, which is based on the presence of at least 3 of 8 indicator symptoms (new rash, severe muscle pain, painful eyes, joint pain, generalized weakness, headache, fever, and chills), was highly correlated with WNV infection, having been met by 29% of the subjects and only 3% of the controls. Nevertheless, the estimate that 26% of WNV-infected persons become symptomatic may be low, because the presence of even 1 of the 8 indicator symptoms was highly predictive of WNV infection: ≥1 symptom was present in 53% of subjects versus 11% of controls (attributable percentage, 42%).
Our results can be compared with results from several other blood donor follow-up studies. Brown et al [12], who used a case definition of the presence of both fever and headache, estimated that 30% of infected persons become symptomatic; however, no WNV-negative control group was present for adjustment. In a study by Custer et al [13] that used a definition of symptomatic infection of ≥3 symptoms, 34% of WNV-infected donors and 20% of false-positive donors were symptomatic, which suggests that only 14% had symptoms that were attributable to WNV. It is unknown why the Custer et al [13] study had a relatively small difference observed between the percentage of subjects and the percentage of controls, which yielded a substantially lower rate of adjusted symptom reporting. This may have been related to a less rigorous laboratory-based definition of subjects and controls [13]. However, even with only 14% of individuals having symptoms attributable to WNV infection, Custer et al [13] reported a remarkably similar distribution of individual symptoms, with headache being the most common.
One notable finding of our study was the absence of reported fever among a substantial proportion of the symptomatic persons (44%). This finding was corroborated by a study involving 534 persons with WNF identified by surveillance in California, in which 31% of subjects did not report fever [22]. These findings suggest that many symptomatic WNV infections are without recognized fever.
Our data indicate that many WNV-related illnesses are clinically significant but are nonspecific and remain undiagnosed. Of the 26% of persons infected with WNV who developed symptoms meeting our case definition that were attributable to infection, we observed that 38% had visited a doctor for their symptoms, and 2% were hospitalized as a result of the infection. However, only 5% of those who sought medical care received a diagnosis of WNV infection. These results are higher than those reported in the study by Custer et al [13], which found that an adjusted 4% of individuals sought medical care (ie, 12% of all confirmed-positive donors, regardless of the number of symptoms, compared with 8% of false-positive donors). Similarly, only 2 (6%) of the 35 confirmed-positive donors who sought medical care received a diagnosis of WNV infection [13]. Only 12 (9%) of 135 viremic donors, or 29% of those meeting the study case definition for WNF, sought medical care in Colorado [12]. This low rate of recognition of WNF is consistent with surveillance data that shows that the 15,800 WNF cases reported through 2007 in the United States were a small fraction of the >300,000 infections that are estimated to have occurred [23].
Similar to the findings of Brown et al [12], we found that a higher plasma viral load at the time of index donation predicted the development of symptoms. However, there were inconsistencies with other studies regarding other possible factors associated with reported symptoms. Although Brown demonstrated decreasing proportions of infected persons developing symptoms as donor age increased, particularly among men [12], no such relationship was observed in our study or in the study of Custer et al [13]. In our study, women were more likely to have reported symptoms (OR, 1.6), similar to the report by Custer et al [13] (OR, 1.4). Nevertheless, Brown et al [12] found no relationship between sex and symptom development.We observed similar plasma viremia levels between men and women, which suggests that plasma viremia level is not responsible for increased reporting of symptoms among women.
Several limitations to our study exist. Approximately one-quarter to one-third of NAT-reactive donors did not consent to interview. If donors who developed symptoms were more likely to participate, this would bias the study towards an increased frequency of symptoms. The initial notification of preliminary WNV screening test results could have increased nonspecific symptom reporting, although there should not be a differential effect between subjects and controls. Blood donors are healthier than the general population and thus may not be representative of the entire population. However, underlying illnesses or conditions that may influence the development of WNF have not been identified in the general population. In addition, plasma viremia levels were measured only on the day of index donation and thus may not reflect peak levels or the extent of viral replication and dissemination throughout the acute infection period. Nevertheless, the time of donation in relationship to each donor's plasma viremia curve should be randomly distributed among persons within each age and sex subgroup, and thus, comparisons among these groups should reflect true population differences in viremia levels.
In summary, our data provide several insights into the clinical characteristics of WNV infection. Our data demonstrate that demographic risk factors for neuroinvasive disease strikingly contrast with those for WNF. Although surveillance data indicate that men are at greater risk than are women for neuroinvasive disease [24], our data suggest that women may be at higher risk for developing WNF. The substantial increase in the risk of neuroinvasive disease with advancing age [24] was not observed for WNF in our study. Although 26% of our WNF-confirmed study population developed symptoms that met our case definition and were attributable to infection, our analysis suggested that as many as 42% had ≥1 indicator symptom that was attributed to the infection. The fact that nearly one-half of the patients who met our symptomatic case definition sought medical care yet few received a diagnosis of WNV infection suggests that WNF has significant but largely unrecognized clinical impact [25].
Acknowledgments
We thank Dr. Tiepu Liu, Fatemeh Musavi, and Edward P. Notari IV, for technical advice and support in data analysis.
References
Figures and Tables
Study flow chart. IgM, immunoglobulin M; WNV, West Nile virus.
Study flow chart. IgM, immunoglobulin M; WNV, West Nile virus.
Distribution of intervals between index donation and interview for 821 donors confirmed as West Nile Virus RNA positive versus 615 donors without confirmed infection, American Red Cross, June 2003 through 2008.
Distribution of intervals between index donation and interview for 821 donors confirmed as West Nile Virus RNA positive versus 615 donors without confirmed infection, American Red Cross, June 2003 through 2008.
Breakdown of Confirmatory Criteria for 1095 West Nile Virus RNA—Infected donors from the American Red Cross, June 2003 through 2008
Breakdown of Confirmatory Criteria for 1095 West Nile Virus RNA—Infected donors from the American Red Cross, June 2003 through 2008
Frequency of Symptoms on the Day of Donation or during the 14 Days after Donation among 576 West Nile Virus—Infected Subjects and 615 Controls Identified by the American Red Cross, June 2003 through 2008
Frequency of Symptoms on the Day of Donation or during the 14 Days after Donation among 576 West Nile Virus—Infected Subjects and 615 Controls Identified by the American Red Cross, June 2003 through 2008
Frequency of Symptoms on the Day of Donation or during the 14 Days after Donation among 167 West Nile Virus—Infected Subjects Meeting the Clinical Case Definition (≥3 Indicator Symptoms) Identified by the American Red Cross, June 2003 through 2008
Frequency of Symptoms on the Day of Donation or during the 14 Days after Donation among 167 West Nile Virus—Infected Subjects Meeting the Clinical Case Definition (≥3 Indicator Symptoms) Identified by the American Red Cross, June 2003 through 2008
Analysis of Relationship of Donor Age to Case Definition Stratified by Sex among 576 West Nile Virus—Infected Subjects identified by the American Red Cross, June 2003 through 2008
Analysis of Relationship of Donor Age to Case Definition Stratified by Sex among 576 West Nile Virus—Infected Subjects identified by the American Red Cross, June 2003 through 2008
Viral Load among 576 West Nile Virus—Infected Subjects Identified by the American Red Cross, June 2003 through 2008
Viral Load among 576 West Nile Virus—Infected Subjects Identified by the American Red Cross, June 2003 through 2008
