HERPES SIMPLEX VIRUS TYPE 2 (HSV-2) infection is relatively common but is often unrecognized. In a national study ending in 1994, 21.9% of Americans 12 years of age or older—representing 45 million people—were seropositive for HSV-2.1 In that study, only 9.2% of respondents with HSV-2 infection reported “ever having had genital herpes.” In addition to its numeric importance, HSV-2 infection occasionally can cause devastating systemic disease, primarily in neonates and in immunosuppressed persons. Another public health concern about HSV-2 infection is its role as a facilitator of human immunodeficiency virus (HIV) spread, increasing the risk of both transmission and acquisition of HIV.2,3
Presence of HSV-2 antibody generally indicates genital infection. The majority of persons seropositive for HSV-2 do experience symptoms, often nonspecific ones, such as genital itching or irritation, which frequently can be linked directly to herpetic lesions on subsequent examination and testing for virus.4–9 Previous studies have shown that the large majority of genital herpes outbreaks appear in an “atypical” form, with nonspecific ulceration, erosions, fissures, excoriations, erythematous patches, or other nonspecific lesions. HSV-2 outbreaks may not be recognized as a manifestation of a viral infection, but instead may be ignored or attributed to some other genital or perianal irritation caused by yeast infection, urinary tract infection, folliculitis, latex allergy, trauma, or other genital condition.4–9
Genital symptoms are commonly seen in primary practice, but many remain without a confirmed diagnosis. Little is known about the proportion that might be caused by HSV-2 infection. Similarly, it is unknown how often HSV-2 outbreaks are mistakenly diagnosed as urinary tract infection, candidal yeast infection, or other conditions that appear to respond to empirical therapy.9 In this context, improving diagnosis of HSV-2 infection in primary care practice could reduce inappropriate use of antibiotics and antifungal medications for misdiagnosed infections, as well as enable patients to receive effective treatment and take appropriate precautions to prevent spreading the disease.
A recent study of approximately 5400 primary care patients offers an opportunity to close this information gap.10 Previously reported analyses of this data set have described the high prevalence of HSV-2 antibody in the population sample (22.0% in men; 28.3% in women). We are able to use the data set to describe the relationship between HSV-2 infection and genital symptoms because study participants responded to a questionnaire asking whether they had experienced specific genital symptoms or conditions within the 12 months preceding the study.
We undertook the present analyses of the data set to explore which genital symptoms, if any, were associated with HSV-2 infection and to assess the strength of the association for each one. We also sought to determine whether associations between HSV-2 infection and genital symptoms persist after adjustment for potential confounders, such as age, education, lifetime number of sex partners, and history of other sexually transmitted diseases (STDs).
Materials and Methods
Study Population and Study Design
The study design, methods, and study population have already been described in detail.10 Briefly, the study took place between September and December 2002 at 36 randomly selected primary care physician (PCP) offices in suburban areas surrounding 6 US cities (Atlanta, Baltimore, Boston, Chicago, Dallas, and Denver). The cities were selected on the basis of population, income (state level), and geographic location. Suburban area selection was based on frequency of home ownership above the median and population density below the 90th percentile, household income above the median, and housing values above the median for that particular metropolitan area. Within each area, PCP offices were randomly selected from a commercial database of more than 2200 PCP offices within the suburban areas surrounding the 6 cities. The PCP sample was selected according to probability proportional-to-size random sampling, with the volume of total prescriptions written for any medication used as a surrogate for the practice size.
At each PCP office, a field interviewer (FI) recruited approximately 150 individuals aged 18 to 59. Equal distributions of men and women were sought across 4 age categories (18–29, 30–39, 40–49, and 50–59). Eligible patients had to be able to read and comprehend English. Patients known to be pregnant were not eligible. During a normal PCP office visit for general medical care, office staff presented cards to individuals on check-in to inform them that the physician’s office was participating in a research study; no mention of HSV-2 or genital herpes was made. The card directed patients to the FI to learn more about the study. Office staff also recorded the number of eligible patients, by age and by sex, for 2 consecutive weeks; these data were used for sample weighting.
The study was approved by the institutional review board of RTI International, which conducted the initial study with support from GlaxoSmithKline, Inc. RTI International has a multiple-project federalwide assurance (OHRP number 00003331, with an expiration date of September 16, 2005). The data set used in the present analysis contained no information traceable to individual patients.
Serological Testing and Responses to the Questionnaire
Serum samples were tested for antibodies to HSV-2 using the HerpeSelect HSV 2 enzyme-linked immunosorbent assay (ELISA) IgG kit (Focus Technologies, Cypress, CA). This test has been licensed by the U.S. Food and Drug Administration.
Patients enrolled in the study were asked to complete a 15-minute questionnaire, using audio computer-assisted self-interview (ACASI) technology. ACASI technology has been shown to improve the accuracy of reporting of sensitive information.11 The questionnaire included demographic questions and a series of questions designed to elicit sexual behavior and history of STDs. To assess the lifetime number of sex partners, patients were asked: “How many sex partners have you had in your lifetime?” To assess the STD history, patients were asked (in separate questions) whether they had “ever been told by a health care provider” that they had “chlamydia, gonorrhea (clap, GC),” “pelvic inflammatory disease (PID),” “syphilis or a genital ulcer,” “trichomonas (trich),” “genital warts, human papilloma virus (HPV), cervical dysplasia, or condyloma,” or “genital herpes.” Patients also were asked about their history of genital symptoms, signs and diagnoses during the 12 months preceding the office visit (see Table 1). The questionnaire asked 10 such questions; 2 of the 10 (about vaginal yeast infections and about shaving or douching) were asked only of women. For simplicity of expression, we refer to all these symptoms, signs, and diagnoses as “symptoms.”
Patient samples were weighted to represent the targeted study population (residents who visited PCP offices in the selected suburban areas), with adjustments for nonresponse, as described previously.10 To account for the survey sampling weighting, we used SUDAAN statistical software for sample survey data analysis (Research Triangle Institute, Research Triangle Park, NC).12
The investigators specified the main analyses at the outset. Only 4.3% of study participants reported ever being told they had genital herpes and were excluded from in this analysis. The results of primary interest were specified during the planning process to be the adjusted odds ratios for the relationship between each symptom and HSV-2 serostatus, adjusted for possible demographic and behavioral confounders, and patient-reported previous history of nonherpes STDs. Analyses were conducted separately for men and for women, since a given genital symptom might have a different physical or anatomical meaning for the 2 sexes.
Multiple logistic regression (conducted in SUDAAN) was used to adjust for potential confounding by demographic and behavioral variables. We fitted a separate model for each genital symptom cited in the questionnaire, with the genital symptom as the dependent variable. Independent variables included demographic and behavioral variables that had been shown in previous analyses to predict HSV-2 serostatus.10 In addition, because some of the STDs cited in the questionnaire may produce symptoms, we included, a priori, the self-reported history of a previous STD.
We began with regression models that fully coded (as separate variables) the categories of each independent variable, as in the questionnaire. To allow comparison of odds ratios across genital symptoms, we did not remove independent variable from the models found to be nonsignificant. However, small cell sizes precluded accurate calculations of P values for some of the less prevalent genital symptoms (symptoms 3–8 for men, 3 and 6 for women). To achieve adequate cell sizes for these models, categories within each demographic and behavioral variable were combined. For example, the 4 age categories (18–29, 30–39, 40–49, and 50–59) were combined into 2 (18–39 and 40–59). In addition, the separate self-reported STD diagnoses were collapsed into an “any STD” variable, and insurance status was deleted from the model. With one exception, the odds ratios in the resulting simplified models were within 4% relative to those in the full models; symptom 8 in men fell by nearly 9%. All the simplified models were built with each one having the same independent variables and categories as the other.
To estimate the importance of HSV-2 infection as a cause of genital symptoms, we calculated population-attributable risk (PAR). As described by Vittinghoff and Padian,13 this statistic “combines information on the prevalence of an exposure with a measure of the associated increment in risk, providing an estimate of the proportion of incident or prevalent disease that might be avoided by eliminating the exposure.” We used the formula PAR = (prevalence of HSV-2 among those with the symptom)(1 − 1/adjusted odds ratio). PAR was calculated only for those genital symptoms found to be significantly associated with HSV-2 infection.
Frequency of Symptoms in Men and Women
Self-reported genital symptoms were relatively common in general, but their prevalence varied greatly (Table 1). Symptom 1 (“itching, pain, burning, or tingling”) and symptom 2 (“redness, irritation, or a rash”) described general irritative symptoms in the genital area, and both were reported by more than 10% of both men and women. In addition, complaints of “tiny pimples or ingrown hair” (symptom 7) were reported by more than 10% of both men and women. In contrast, symptom 6 (“swelling in the groin”) was uncommon (reported by fewer than 2% of men and women).
Every genital symptom was more common in women than in men, sometimes by a large margin. Symptom 4 (“urethritis”) and symptom 5 (“discharge”) were twice as prevalent in women, and symptom 8 (history “urinary tract infection”) was almost 8 times as prevalent. Symptom 9 (“vaginal yeast infection”) and symptom 10 (“genital irritation from shaving or douching”) were reported only by women; both were common (reported by 29% and 17.3%, respectively).
Genital Symptoms and HSV-2: Descriptive Analysis
The prevalence of HSV-2 among men and women who report or do not report genital symptoms is shown in Table 2. HSV-2 was most prevalent in men who report symptom 8 (“urinary tract infection”) at 40.8% and women who report symptom 3 (“sores, blisters, ulcers, crusts, or small cuts/slits”) at 38.0%. The difference in HSV-2 prevalence among those who report and do not report genital symptoms is greater than 10 percentage points in 3 cases: (1) men reporting symptom 3, (2) women reporting symptom 3, and (3) men reporting symptom 8.
Genital Symptoms and HSV-2: Multivariate Analyses
Among men, the likelihood of reporting symptom 3 (“sores, blisters, ulcers, crusts, or small cuts/slits”) was 79% greater in those with HSV-2 infection than in those without the infection [adjusted odds ratio (OR), 1.79; 95% CI, 1.24–2.58; Fig. 1]. Of all men reporting symptom 3, 32.1% were HSV-2 positive (Table 2). Symptom 1 (“itching, pain, burning, or tingling”) and symptom 2 (“redness, irritation, or rash”) showed a trend toward an association, but neither trend reached statistical significance. The largest adjusted odds ratio, at 2.34, was for symptom 8 (history of “urinary tract infection”), but the relatively small prevalence of this symptom among men meant that confidence intervals were wide (adj OR, 2.34; 95% CI, 0.89–6.10). Overall, the likelihood of reporting any genital symptom among men was significantly increased in men with HSV-2, by 34% (adj OR, 1.34; 95% CI, 1.05–1.63).
Among women, the likelihood of reporting symptom 2 was 50% greater in those with HSV-2 infection than in those without the infection (adj OR, 1.50; 95% CI, 1.06–2.11). Of all women reporting symptom 2, 33.6% were HSV-2 positive (Table 2). Symptom 3 showed a trend toward an association, but the trend did not reach statistical significance. Symptom 4 (“urethritis/dysuria”) showed a statistically significant inverse association with HSV-2 infection, with an approximately 33% lower likelihood of reporting this symptom in those with HSV-2 infection, compared with those without the infection (adj OR, 0.67; 95% CI, 0.51–0.87). Of all women reporting symptom 4, 20.2% were HSV-2 positive (Table 2). No significant association between HSV-2 infection and having any genital symptom overall was found among women (adj OR, 0.96; 95% CI, 0.78–1.18).
Among both men and women, adjusted odds ratios for genital symptoms 5, 6, and 7 clustered at or slightly below the null value. In these cases, 95% CIs widely overlapped the null value of 1.0.
Logistic Regression Models for Symptom 3, for Men and Women (Table 2)
As an example of the logistic regression models produced in the analysis, the two models for symptom 3 (“sores, blisters, ulcers, crusts or small cuts/slits”) are shown in Table 3, for men and for women. As described in the preceding section, the adjusted odds ratio for HSV-2 infection was significant for men (OR, 1.79; 95% CI, 1.24–2.58), and though there was a similar trend for women, it was not statistically significant. In men, membership in the older age group (age 40–59) was associated with decreased reporting of symptom 3 (OR, 0.49; 95% CI, 0.38–0.64). Also in men, a history of any STD more than doubled the likelihood of reporting symptom 3 (OR, 2.28; 95% CI, 1.05–4.98). Other variables were only mildly associated with symptom 3, and 95% confidence intervals overlapped the null value of 1.0. In both men and women, increasing lifetime numbers of sex partners was associated with increased likelihood of reporting symptom 3, but this association was not statistically significant.
Calculation of PAR
As described above, the first term of the equation for PAR of HSV-2 infection for a given symptom is the prevalence of HSV-2 infection among those who reported the symptom. The second term is equal to (1 − 1/adjusted odds ratio).
Weighted prevalence of HSV-2 infection among men who reported symptom 3 (“sores, blisters, ulcers, crusts, or small cuts/slits”) was 32.1% versus 22% overall. Given the adjusted odds ratio of 1.79 (as in Fig. 1A), we estimated the PAR at 14.1%, or approximately 1 in 7.
Similarly, weighted prevalence of HSV-2 infection among women who reported symptom 2 (“redness, irritation, or rash”) was 33.6% versus 28% overall. Given that the adjusted odds ratio was 1.50 (as in Fig. 1B), we estimated the PAR at 11.2%, or approximately 1 in 9.
These results underline the importance and challenge of diagnosing HSV-2 infection as a potential cause of a variety of genital symptoms. Some of the genital symptoms assessed by the questionnaire were highly prevalent among the sample population (especially among women), and all of them were potential perceived manifestations of genital herpes. The analysis was designed to identify which genital symptoms were most closely associated with HSV-2 infection in a general primary care population and to lay the groundwork for future studies of strategies to improve the diagnosis of genital herpes. Only patients with no known history of genital herpes were included in the analyses.
The groups of symptoms with elevated odds ratios for both men and women were symptom group 3 (“sores, blisters, ulcers, crusts, or small cuts/slits”) and symptom group 2 (“redness, irritation, or a rash”). As assessed in the questionnaire, these symptom groups encompassed a variety of possible manifestations of genital herpes, including the vesicular and ulcerative-type symptoms most closely associated with “classic” genital herpes, as well as less “classic” but more common manifestations, such as cuts and slits and general irritation or rash. Because of the way in which the questionnaire grouped symptoms, it is not possible to determine the type of discomfort to which patients were referring in their answers.
The magnitude of the associations of HSV-2 infection with symptoms 2 and 3 was modest. The associations reached statistical significance for symptom 3 in men, with a 79% increase (adj OR, 1.79), and for symptom 2 in women, with a 50% increase (adj OR, 1.50). However, because HSV-2 was highly prevalent in the patient population, even these modest HSV-2-related increases in the risk for a given symptom could translate into increases in the incidence of the symptom population-wide.
Our estimates of the PAR indicate that HSV-2 may account for roughly 14.1% (1 in every 7 reports) of symptom 3 in men and roughly 11.2% (1 in every 9 reports) of symptom 2 in women. Estimates of PAR are variable and must be interpreted cautiously.13 However, as an approximate measure, the PARs for these common symptoms indicate that unrecognized HSV-2 infection causes morbidity in significant numbers of patients.
In men, the symptom that was most elevated in those with HSV-2 infection was symptom 8 (history of “urinary tract infection”), which was more than twice as likely to be reported among those with HSV-2 infection as among those without (adj OR, 2.34). However, only 2.4% of men reported this symptom, and confidence intervals were wide, overlapping the null value. A history of urinary tract infection was highly prevalent among women (at 18.9%), but it was not associated with HSV-2 infection.
The inverse association in women between HSV-2 infection and symptom 4 (“been diagnosed or treated for urethritis [“pain, tingling, itching of the urethra”] or had pain, burning, itching, tingling or discharge on urination”) was unexplained. With P at 0.0046, the association seems unlikely to have arisen by chance, but it seems to have no plausible biologic basis. We can only speculate that it resulted from some residual confounding or from patients’ misunderstanding of the lengthy question.
Other self-reported genital symptoms, such as itching, discharge, swelling, pimples, and history of vaginal yeast infection, did not correlate with HSV-2 infection. It is possible that HSV-2-related symptoms may have been mild or outnumbered by genital symptoms from a variety of other causes, so that the study power was insufficient to detect an association with HSV-2 infection.
The study has limitations that should be mentioned. First, patients’ reports of symptoms were retrospective, so they may have been nonspecific or vaguely remembered since the time the symptoms were experienced. Second, the questionnaire did not assess whether the symptoms were recurrent; thus, symptoms that occurred only once could not be distinguished from recurrent ones, which might have been more specific for HSV-2 outbreaks. However, another study that evaluated similar symptoms by recurrence frequency also failed to identify strong associations with HSV-2 antibody status.14 In addition, the severity of the symptoms was not assessed. Most patients had been visiting their physicians because of general medical complaints or to receive routine health checks, and the questionnaire did not ask about the symptoms’ clinical relevance for the patients. Finally, caution should be used in interpreting the statistical significance of individual results; among the 18 adjusted odds ratios (including the 10 symptoms reported by women and the 8 reported by men), about 1 result would have been expected to be “significant” at P = 0.05 by chance alone.15,16
In future studies, these limitations could be overcome in several ways. First, questions about symptoms should specify, as precisely as possible, the type of pain or irritation experienced. Second, the severity and recurrent nature of the symptoms should be carefully assessed. Third—and perhaps most important—future research may best focus on patients who are visiting their clinicians specifically for genital symptoms. In these cases, symptoms are more likely to be clinically relevant. Furthermore, when symptoms are recent or are present during the visit, clinicians can send samples for viral culture or for testing by polymerase chain reaction. Ultimately, detection of virus in the genital tract is the best way to document the connection between genital symptoms and HSV-2.
In the meantime, the associations between HSV-2 infection and several genital symptoms should encourage clinicians to consider HSV-2 as a possible cause of these symptoms among primary care patients. Even if the presence of HSV-2 in the genital tract cannot be documented in a given patient, type-specific serological testing (which is now commercially available) and culture or PCR for HSV-2 may help establish whether HSV-2 could or could not be causing the patient’s symptoms.