AMERICAN JOURNAL or
EPIDEMIOLOGY
Vol. 131, No. 4
Copyright © 1990 by The Johns Hoplani University School of Hygiene and Public Health All rights reserved
Printed mU S.A.
LINDA H. PEREIRA,1 JUAN A. EMBIL,1-*3 DAVID A. HAASE,4 AND KEVIN M. MANLEY1 Pereira, L H. (Dept of Community Health and Epidemiology, Dalhousie U., Halifax, Nova Scotia, Canada B3H 4H7), J. A. Embil, D. A. Haase, and K. M. Manley. Cytomegalovirus infection among women attending a sexually transmitted disease clinic: association with clinical symptoms and other sexually transmitted diseases. Am J Epidemiol 1990;131:683-92. The prevalence rates of cytomegalovirus, Nelsseria gonorrhoeas, Chlamydla trachomatis, Trichomonaa vaginalls, and herpes simplex virus infection were determined for 247 women attending a sexually transmitted disease clinic in Halifax, Nova Scotia between July 1983 and December 1985. Isolation rates were 8.5%, 32.8%, 27.1%, 7.3%, and 6.5% for the five infectious agents, respectively. With multiple logistic regression analysis, the presence of cervical cytomegalovirus infection was independently associated with age less than 23 years and with gonococcal infection. Factors predictive of C. trachomatis infection included age less than 23 years, gonococcal infection, oral contraceptive use, and purulent discharge. Number of lifetime sexual partners was statistically associated only with herpes simplex virus infection. N. gonorrhoeas, C. trachomatis, and T. vaginalls were all independently associated with purulent discharge. Cytomegalovirus, N. gonorrhoeae, and C. trachomatis were statistically more likely to be present concurrently with other organisms than to be present as a single infection. Women with another genital infection were 6.5 times more likely to have cytomegalovirus than were women with no other genital infection. Of the 21 women with cytomegalovirus, only two had no other sexually transmitted disease. These findings suggest that N. gonorrhoeae and other sexually transmitted diseases may play a role in either the sexual transmission of or the reactivation of cervical cytomegalovirus infection. chlamydia infections; cytomegaloviruses; gonorrhea; herpes genitalis; sexually transmitted diseases; trichomonas
Most people become infected with cyto- normally establishes latent infection, but megalovirus in the course of their lifetime, some women experience periods of reactiAfter primary exposure, cytomegalovirus" vation of the virus in the uterine cervix Received for publication November 4, 1988, and in final form September 15, 1989. 1 Department of Community Health and Epidemiology, Dalhousie University, Halifax, Nova Scotia, Canada. 1 Department of Microbiology, Dalhousie University, Halifax, Nova Scotia, Canada. 1 Department of Pediatrics, Dalhousie University, and the Izaak Walton KilLam Hospital for Children, Halifax, Nova Scotia, Canada. 683
4
Department of Medicine, Division of Infectious Diseases, Dalhousie University, and the Victoria General Hospital, Halifax, Nova Scotia, Canada, Reprint requests to Linda H. Pereira, Department of Community Health and Epidemiology, Clinical Research Center, Dalhousie University, 5849 University Avenue, Halifax, Nova Scotia, Canada B3H 4H7. Supported by grant 6603-1136-54 from Health and Welfare, Canada.
Downloaded from https://academic.oup.com/aje/article-abstract/131/4/683/79562 by Tulane University Medical Library user on 21 January 2019
CYTOMEGALOVIRUS INFECTION AMONG WOMEN ATTENDING A SEXUALLY TRANSMITTED DISEASE CLINIC: ASSOCIATION WITH CLINICAL SYMPTOMS AND OTHER SEXUALLY TRANSMITTED DISEASES
684
PEREIRA ET AL
for mucopurulent cervicitis, salpingitis, and subsequent high risk of ectopic pregnancy (16). Unlike cytomegalovirus, detection of N. gonorrhoeae, C. trachomatis, T. vaginalis, and herpes simplex virus is usually accompanied by identifiable symptoms, specifically discharge or ulcers. However, all of these infections can be subclinical. To identify factors predictive of each infection, examine possible interrelations, and evaluate the independent association of each agent with clinical symptoms of disease, we investigated the presence of the above-mentioned five infectious agents in 247 women attending a sexually transmitted disease clinic. MATERIALS AND METHODS
Study population Cervical specimens from all women attending the sexually transmitted disease clinic at the Victoria General Hospital in Halifax, Nova Scotia between July 1983 and December 1985 were cultured for cytomegalovirus and herpes simplex virus. Patients who had come for a follow-up visit after treatment were excluded from our sample. Endocervical swabs from 247 women were successfully cultured. Demographic information, observations from clinical examination, routine test results (for N. gonorrhoeae, C. trachomatis, T. vaginalis, and herpes simplex virus), and final diagnoses were collected from clinic charts. A questionnaire including information on contraceptive use and sexual history was administered by the attending physician for the purposes of this study. Approximately 40 percent of our sample of women came to the clinic because of known or suspected sexually transmitted disease in a sexual partner. A total of 25 percent of the sample had no symptoms of disease. Evaluation of vaginal and cervical secretions Clinic physicians described the quantity of discharge they observed from both the cervix and the vagina by selecting from
Downloaded from https://academic.oup.com/aje/article-abstract/131/4/683/79562 by Tulane University Medical Library user on 21 January 2019
despite the presence of antibody (1). There is no evidence that cytomegalovirus has an etiologic role in cervicitis, salpingitis, or pelvic inflammatory disease; however, infection with cytomegalovirus during pregnancy is the cause of most congenital viral infections (2). The more severe sequelae of neurologic defects have been associated with primary cytomegalovirus infection, but there is evidence that secondary infections (due to reactivation or reinfection) can also result in fetal damage and developmental disorders (3-5). The epidemiology of cytomegalovirus is similar to that of a sexually transmitted disease: Cytomegalovirus has been isolated more frequently from women attending sexually transmitted disease clinics than from the female patients of other clinic populations (6-8). Cytomegalovirus has been demonstrated to persist for several weeks in high concentrations in semen (9). Cytomegalovirus seropositivity has been found to be associated with number of lifetime sexual partners (10), and deoxyribonucleic acid restriction enzyme typing has demonstrated identical cytomegalovirus isolates from sexual partners (11). D. J. Lang has suggested, however, that "it would be preferable to identify this infection as sexually transmissible rather than sexually transmitted" (12, p. 476). There are few surveys of women which have included cervical cultures for cytomegalovirus along with routine tests for the more common sexually transmitted diseases Neisseria gonorrhoeae, Chlamydia trachomatis, Trichomonas vaginalis, and herpes simplex virus (7, 13). In one study (14), 80 percent of the cytomegalovirus culture-positive women attending a sexually transmitted disease clinic had other infections. In another, more recent study (10), chlamydia infection was a strong independent predictor of cervical cytomegalovirus shedding in seropositive women. A statistically significant relation between N. gonorrhoeae and C. trachomatis has been noted in several studies (15). These are the agents most often responsible
CYTOMEGALOVIRUS AND SEXUALLY TRANSMITTED DISEASES
clease restriction digestion analysis as described elsewhere (18, 19). Specimens were processed for N. gonorrhoeae and C. trachomatis in the microbiology laboratory of the Victoria General Hospital by using routine methods. Specimens taken from suspect lesions were also cultured at the hospital laboratory for herpes simplex virus. Wet mount preparations of vaginal secretions were examined at the sexually transmitted disease clinic for trichomonads and yeast species. The diagnosis of condyloma acuminata was based on clinical examination.
Statistical methods Associations between categorical variables were evaluated using the Pearson chi-square statistic. Multiple logistic regression analyses were performed to determine which combination of factors was most predictive of infection with specific agents. Linear regression analysis was used to identify the pathogens most closely associated with discharge quantity and apSpecimen collection and laboratory pearance. The prevalence ratio, which is methods the prevalence of infection in the exposed Specimens for identification of infectious divided by the prevalence of infection in agents were collected in a standardized the unexposed, was calculated to measure fashion. Swabs from the endocervix were the association between each pathogen sent to the hospital laboratory to be cul- and other sexually transmitted diseases. tured for N. gonorrhoeae and C. trachoma- SPSSx (20) and BMDP (21) statistical packages were used. tis. Cytomegalovirus and herpes simplex viRESULTS rus were cultured from an additional enPatient characteristics and isolation rates docervical swab which was sent to a research laboratory. Culture methods have The demographic and behavioral charpreviously been described (8). Specimens acteristics of the 247 female sexually transwere inoculated onto human foreskin fibro- mitted disease clinic patients are summablasts that were prepared according to the rized in table 1. The mean age was 23.3 (± technique of Embil and Faulkner (17) and ' 6.8) years, and the mean age at first interMcFarlane et al. (18). Cell cultures were course was 16.4 (±2.3) years. The median examined for cytopathic effect up to 5 number of sexual partners in the last 6 weeks before being reported as negative for months was 1.5 (range, 0-1,000), and the cytomegalovirus. Since herpes simplex vi- median number of lifetime sexual patterns rus grows more rapidly in human fibro- was 5 (range, 1-1,000). The distribution of blasts, it was not technically feasible to these two variables was skewed partly beisolate cytomegalovirus from a specimen cause of the presence in our sample of seven containing herpes simplex virus. Viral iso- prostitutes who had an extremely large lates were identified and typed by endonu- number of partners.
Downloaded from https://academic.oup.com/aje/article-abstract/131/4/683/79562 by Tulane University Medical Library user on 21 January 2019
three categories on the clinic chart: profuse, moderate, and scant. If the discharge was perceived to be profuse at either the cervix or in the vagina, the clinic chart was marked profuse. Otherwise, a general assessment of discharge quantity at both sites was described as either moderate or scant. As an independent variable in a multiple regression analysis, discharge quantity was coded as 1 for scant, 2 for moderate, and 3 for profuse, or, when appropriate, 1 for profuse and 0 for not profuse. The appearance of endocervical secretions on a white cotton-tipped swab was classified as clear, white, orpurulent (yellowgreen). If the cervical discharge was too scant to evaluate, the appearance of vaginal secretions on the speculum was classified instead. In multiple variable analysis, discharge appearance was either coded as 1 for clear, 2 for white, and 3 for purulent, or, when appropriate, as 1 for purulent and 0 for not purulent. When blood was present, data on discharge were coded as missing.
685
686
PEREIRA ET AL. TABLE 2
TABLE 1
Characteristics of female patients of a sexually transmuted disease chmc, Halifax, Nova Scotia,
CharacUrutic
Age (yean) 14-19 20-24 25-55 Mamed Contraceptive use None Oral contraceptives Intrautenne device Bamer method* Pregnant Age at first intercourse
EPIDEMIOLOGY
Vol. 131, No. 4
Copyright © 1990 by The Johns Hoplani University School of Hygiene and Public Health All rights reserved
Printed mU S.A.
LINDA H. PEREIRA,1 JUAN A. EMBIL,1-*3 DAVID A. HAASE,4 AND KEVIN M. MANLEY1 Pereira, L H. (Dept of Community Health and Epidemiology, Dalhousie U., Halifax, Nova Scotia, Canada B3H 4H7), J. A. Embil, D. A. Haase, and K. M. Manley. Cytomegalovirus infection among women attending a sexually transmitted disease clinic: association with clinical symptoms and other sexually transmitted diseases. Am J Epidemiol 1990;131:683-92. The prevalence rates of cytomegalovirus, Nelsseria gonorrhoeas, Chlamydla trachomatis, Trichomonaa vaginalls, and herpes simplex virus infection were determined for 247 women attending a sexually transmitted disease clinic in Halifax, Nova Scotia between July 1983 and December 1985. Isolation rates were 8.5%, 32.8%, 27.1%, 7.3%, and 6.5% for the five infectious agents, respectively. With multiple logistic regression analysis, the presence of cervical cytomegalovirus infection was independently associated with age less than 23 years and with gonococcal infection. Factors predictive of C. trachomatis infection included age less than 23 years, gonococcal infection, oral contraceptive use, and purulent discharge. Number of lifetime sexual partners was statistically associated only with herpes simplex virus infection. N. gonorrhoeas, C. trachomatis, and T. vaginalls were all independently associated with purulent discharge. Cytomegalovirus, N. gonorrhoeae, and C. trachomatis were statistically more likely to be present concurrently with other organisms than to be present as a single infection. Women with another genital infection were 6.5 times more likely to have cytomegalovirus than were women with no other genital infection. Of the 21 women with cytomegalovirus, only two had no other sexually transmitted disease. These findings suggest that N. gonorrhoeae and other sexually transmitted diseases may play a role in either the sexual transmission of or the reactivation of cervical cytomegalovirus infection. chlamydia infections; cytomegaloviruses; gonorrhea; herpes genitalis; sexually transmitted diseases; trichomonas
Most people become infected with cyto- normally establishes latent infection, but megalovirus in the course of their lifetime, some women experience periods of reactiAfter primary exposure, cytomegalovirus" vation of the virus in the uterine cervix Received for publication November 4, 1988, and in final form September 15, 1989. 1 Department of Community Health and Epidemiology, Dalhousie University, Halifax, Nova Scotia, Canada. 1 Department of Microbiology, Dalhousie University, Halifax, Nova Scotia, Canada. 1 Department of Pediatrics, Dalhousie University, and the Izaak Walton KilLam Hospital for Children, Halifax, Nova Scotia, Canada. 683
4
Department of Medicine, Division of Infectious Diseases, Dalhousie University, and the Victoria General Hospital, Halifax, Nova Scotia, Canada, Reprint requests to Linda H. Pereira, Department of Community Health and Epidemiology, Clinical Research Center, Dalhousie University, 5849 University Avenue, Halifax, Nova Scotia, Canada B3H 4H7. Supported by grant 6603-1136-54 from Health and Welfare, Canada.
Downloaded from https://academic.oup.com/aje/article-abstract/131/4/683/79562 by Tulane University Medical Library user on 21 January 2019
CYTOMEGALOVIRUS INFECTION AMONG WOMEN ATTENDING A SEXUALLY TRANSMITTED DISEASE CLINIC: ASSOCIATION WITH CLINICAL SYMPTOMS AND OTHER SEXUALLY TRANSMITTED DISEASES
684
PEREIRA ET AL
for mucopurulent cervicitis, salpingitis, and subsequent high risk of ectopic pregnancy (16). Unlike cytomegalovirus, detection of N. gonorrhoeae, C. trachomatis, T. vaginalis, and herpes simplex virus is usually accompanied by identifiable symptoms, specifically discharge or ulcers. However, all of these infections can be subclinical. To identify factors predictive of each infection, examine possible interrelations, and evaluate the independent association of each agent with clinical symptoms of disease, we investigated the presence of the above-mentioned five infectious agents in 247 women attending a sexually transmitted disease clinic. MATERIALS AND METHODS
Study population Cervical specimens from all women attending the sexually transmitted disease clinic at the Victoria General Hospital in Halifax, Nova Scotia between July 1983 and December 1985 were cultured for cytomegalovirus and herpes simplex virus. Patients who had come for a follow-up visit after treatment were excluded from our sample. Endocervical swabs from 247 women were successfully cultured. Demographic information, observations from clinical examination, routine test results (for N. gonorrhoeae, C. trachomatis, T. vaginalis, and herpes simplex virus), and final diagnoses were collected from clinic charts. A questionnaire including information on contraceptive use and sexual history was administered by the attending physician for the purposes of this study. Approximately 40 percent of our sample of women came to the clinic because of known or suspected sexually transmitted disease in a sexual partner. A total of 25 percent of the sample had no symptoms of disease. Evaluation of vaginal and cervical secretions Clinic physicians described the quantity of discharge they observed from both the cervix and the vagina by selecting from
Downloaded from https://academic.oup.com/aje/article-abstract/131/4/683/79562 by Tulane University Medical Library user on 21 January 2019
despite the presence of antibody (1). There is no evidence that cytomegalovirus has an etiologic role in cervicitis, salpingitis, or pelvic inflammatory disease; however, infection with cytomegalovirus during pregnancy is the cause of most congenital viral infections (2). The more severe sequelae of neurologic defects have been associated with primary cytomegalovirus infection, but there is evidence that secondary infections (due to reactivation or reinfection) can also result in fetal damage and developmental disorders (3-5). The epidemiology of cytomegalovirus is similar to that of a sexually transmitted disease: Cytomegalovirus has been isolated more frequently from women attending sexually transmitted disease clinics than from the female patients of other clinic populations (6-8). Cytomegalovirus has been demonstrated to persist for several weeks in high concentrations in semen (9). Cytomegalovirus seropositivity has been found to be associated with number of lifetime sexual partners (10), and deoxyribonucleic acid restriction enzyme typing has demonstrated identical cytomegalovirus isolates from sexual partners (11). D. J. Lang has suggested, however, that "it would be preferable to identify this infection as sexually transmissible rather than sexually transmitted" (12, p. 476). There are few surveys of women which have included cervical cultures for cytomegalovirus along with routine tests for the more common sexually transmitted diseases Neisseria gonorrhoeae, Chlamydia trachomatis, Trichomonas vaginalis, and herpes simplex virus (7, 13). In one study (14), 80 percent of the cytomegalovirus culture-positive women attending a sexually transmitted disease clinic had other infections. In another, more recent study (10), chlamydia infection was a strong independent predictor of cervical cytomegalovirus shedding in seropositive women. A statistically significant relation between N. gonorrhoeae and C. trachomatis has been noted in several studies (15). These are the agents most often responsible
CYTOMEGALOVIRUS AND SEXUALLY TRANSMITTED DISEASES
clease restriction digestion analysis as described elsewhere (18, 19). Specimens were processed for N. gonorrhoeae and C. trachomatis in the microbiology laboratory of the Victoria General Hospital by using routine methods. Specimens taken from suspect lesions were also cultured at the hospital laboratory for herpes simplex virus. Wet mount preparations of vaginal secretions were examined at the sexually transmitted disease clinic for trichomonads and yeast species. The diagnosis of condyloma acuminata was based on clinical examination.
Statistical methods Associations between categorical variables were evaluated using the Pearson chi-square statistic. Multiple logistic regression analyses were performed to determine which combination of factors was most predictive of infection with specific agents. Linear regression analysis was used to identify the pathogens most closely associated with discharge quantity and apSpecimen collection and laboratory pearance. The prevalence ratio, which is methods the prevalence of infection in the exposed Specimens for identification of infectious divided by the prevalence of infection in agents were collected in a standardized the unexposed, was calculated to measure fashion. Swabs from the endocervix were the association between each pathogen sent to the hospital laboratory to be cul- and other sexually transmitted diseases. tured for N. gonorrhoeae and C. trachoma- SPSSx (20) and BMDP (21) statistical packages were used. tis. Cytomegalovirus and herpes simplex viRESULTS rus were cultured from an additional enPatient characteristics and isolation rates docervical swab which was sent to a research laboratory. Culture methods have The demographic and behavioral charpreviously been described (8). Specimens acteristics of the 247 female sexually transwere inoculated onto human foreskin fibro- mitted disease clinic patients are summablasts that were prepared according to the rized in table 1. The mean age was 23.3 (± technique of Embil and Faulkner (17) and ' 6.8) years, and the mean age at first interMcFarlane et al. (18). Cell cultures were course was 16.4 (±2.3) years. The median examined for cytopathic effect up to 5 number of sexual partners in the last 6 weeks before being reported as negative for months was 1.5 (range, 0-1,000), and the cytomegalovirus. Since herpes simplex vi- median number of lifetime sexual patterns rus grows more rapidly in human fibro- was 5 (range, 1-1,000). The distribution of blasts, it was not technically feasible to these two variables was skewed partly beisolate cytomegalovirus from a specimen cause of the presence in our sample of seven containing herpes simplex virus. Viral iso- prostitutes who had an extremely large lates were identified and typed by endonu- number of partners.
Downloaded from https://academic.oup.com/aje/article-abstract/131/4/683/79562 by Tulane University Medical Library user on 21 January 2019
three categories on the clinic chart: profuse, moderate, and scant. If the discharge was perceived to be profuse at either the cervix or in the vagina, the clinic chart was marked profuse. Otherwise, a general assessment of discharge quantity at both sites was described as either moderate or scant. As an independent variable in a multiple regression analysis, discharge quantity was coded as 1 for scant, 2 for moderate, and 3 for profuse, or, when appropriate, 1 for profuse and 0 for not profuse. The appearance of endocervical secretions on a white cotton-tipped swab was classified as clear, white, orpurulent (yellowgreen). If the cervical discharge was too scant to evaluate, the appearance of vaginal secretions on the speculum was classified instead. In multiple variable analysis, discharge appearance was either coded as 1 for clear, 2 for white, and 3 for purulent, or, when appropriate, as 1 for purulent and 0 for not purulent. When blood was present, data on discharge were coded as missing.
685
686
PEREIRA ET AL. TABLE 2
TABLE 1
Characteristics of female patients of a sexually transmuted disease chmc, Halifax, Nova Scotia,
CharacUrutic
Age (yean) 14-19 20-24 25-55 Mamed Contraceptive use None Oral contraceptives Intrautenne device Bamer method* Pregnant Age at first intercourse
