JOURNAL OF CLINICAL MICROBIOLOGY, Mar. 1992, p. 716-718 0095-1137/92/030716-03$02.00/0 Copyright X) 1992, American Society for Microbiology
Vol. 30, No. 3
NOTES Latex Agglutination Testing Directly from Throat Swabs for Rapid Detection of Beta-Hemolytic Streptococci from Lancefield Serogroup C GREGORY F. HAYDEN,1* JAMES C. TURNER,2 DARIA KISELICA,3 MELISSA DUNN,' AND J. OWEN HENDLEY1 Departments of Pediatrics' and Student Health,3 University of Virginia Health Sciences Center, Charlottesville, Virginia 22908, and Thomson Student Health Center, University of South Carolina, Columbia, South Carolina 292082 Received 25 July 1991/Accepted 21 November 1991
A latex agglutination method for the rapid detection of beta-hemolytic streptococci from Lancefield serogroup C in throat swabs from 403 university students with symptomatic pharyngitis was evaluated. Compared with culture, the rapid test was poorly sensitive (34.4%) but very specific (98.4%) in detecting group C beta-hemolytic streptococci. The sensitivity of the rapid test improved with an increasing quantity of growth on culture.
Outbreaks of pharyngitis have been attributed to betahemolytic streptococci (BHS) from Lancefield serogroup C (1, 5), and an important role for these organisms in causing endemic pharyngitis among adolescents and young adults has been suggested (3, 4, 10, 11, 15, 16). In one recent study, group C organisms were isolated more than twice as often from the throats of ill university students with pharyngitis than from those of age-matched, healthy controls (18). The issue of causality has remained controversial, however, because many of the earlier studies had not been thorough or properly controlled and because the serologic correlates of group C infection have not been well defined. Latex agglutination methods for the rapid detection of group A BHS directly from throat-swab specimens are widely used in clinical practice (6, 8, 14). The negativecontrol latex reagent in one commercially available kit consists of latex particles coated with antibodies directed against the group C carbohydrate. This control reagent has been shown to be sensitive and specific in identifying group C BHS in veterinary specimens (7). The purpose of this study was to determine the accuracy of this reagent compared with that of culture in detecting group C BHS in throat swab specimens from young adults with symptomatic pharyngitis. Students presenting to the Student Health Centers at the University of Virginia and the University of South Carolina with a chief complaint of sore throat were studied from 1 October 1989 until 31 January 1990. Throat swab specimens were obtained simultaneously with two (University of Virginia) or three (University of South Carolina) rayon-tipped swabs. One swab was used to perform the latex agglutination testing (PathoDx Strep A Kit; Diagnostic Products Corp., Los Angeles, Calif.). The swab was immersed in an acidic solution for 1 min at room temperature. Any liberated *
antigen was then agglutinated by sensitized latex particles during a 4-min incubation on a clear glass plate on a mechanical rotator. Tests for group A and group C carbohydrate were performed in adjacent test circles. Agglutination was assessed by using indirect lighting against a black background. In accordance with the manufacturer's instructions, the test was considered positive when there was moderate or strong agglutination and negative when there was no agglutination and both circles remained milky white. The test was considered uninterpretable, and counted as negative, when there was a finely granular appearance (a slight roughening) in the test circles. The potencies of the test reagents were verified daily with positive control swabs impregnated with inactive group C BHS. The other swabs were used to inoculate two 5% sheep blood agar plates. At the University of Virginia, a single swab was used to inoculate both plates. The plate inoculated first was incubated at 35'C in a 5% C02-enriched atmosphere. The second plate was incubated at 35°C in an anaerobic environment (Anaerobic GasPak System; Baltimore Biologic Laboratory, Cockeysville, Md.). At the University of South Carolina, a separate swab was obtained to inoculate each plate. The growth of typical beta-hemolytic colonies at 24 and 48 h was quantitated as follows: 1+, 100 colonies. Overall test sensitivity and specificity were calculated by considering a culture positive when either the anaerobic or the C02-enriched culture contained group C BHS and negative when neither culture contained group C BHS. BHS were grouped by latex agglutination (PathoDx Strep Grouping Kit; Diagnostic Products Corp.). Isolates of group C BHS were frozen in sterile skim milk for later identification to the species level. The group C BHS include three species (Streptococcus equi, S. equisimilis, and S. zooepidemicus) which produce large colonies and a proportion of S. anginosus ("S. milleri") which produce small colonies and may also belong to other serogroups. The three large-colony species are common veterinary pathogens, but S. equisimilis
Corresponding author. 716
NOTES 1992 VOL. 1992 VOL. 30,30,
TABLE 1. Likelihood of a positive rapid test for group C BHS according to streptococcal species and quantity of growth on culture No. of positive rapid tests/no. testeda with the
following quantity of growth on culture
Species
S. equisimilis S. anginosus Both Not determined a
1+
2+
3+
4+
0/3 2/12 0/0 1/4
1/5 1/20
2/7 4/10
13/16 4/8
2/2 0/1
0/0 0/2
1/1 1/2
Categories for quantity of growth are as defined in the text.
is the most frequent human isolate. When reisolated, colonies with a positive Voges-Proskauer reaction were classified as S. anginosus (2). Colonies which were VogesProskauer negative were classified as S. equisimilis if they fermented 1% trehalose but not sorbitol. Group C BHS were detected among 93 (23.1%) of 403 students tested. Compared with combined culture, the rapid latex agglutination test was poorly sensitive (32 of 93, or 34.4%) but very specific (305 of 310, or 98.4%) in detecting group C BHS. Test sensitivity increased only slightly when the rapid-test result was compared with that of the C02enriched culture only (25 of 68, or 36.8%) or that of the anaerobic culture only (32 of 82, or 39.0%). The predictive values of positive and negative rapid tests were 86.5 and 83.3%, respectively. Test accuracies were similar at both study sites. The likelihood of a positive rapid test increased with an increasing quantity of growth on the culture plate (X2 for linear trend = 19.4; P < 0.001) (Table 1). More than 70% of cultures with 4+ growth of group C BHS were accompanied by a positive rapid test. A total of 84 (90%) patients were identified as infected with group C BHS, as follows: 31 (37%) with S. equisimilis only; 50 (60%) with S. anginosus only; and 3 (4%) with both S. equisimilis and S. anginosus. The rapid test was positive for 16 (52%) of the 31 patients with S. equisimilis, compared with only 11 (22%) of the 50 patients with S. anginosus group C (P < 0.01). Much of this variation in test sensitivity could be attributed to species differences in heaviness of growth. S. equisimilis strains were more commonly isolated in 4+ quantity than were S. anginosus strains (52 versus 16%; P < 0.01). The sensitivity of this latex agglutination method for the rapid detection of group C BHS directly from throat swabs was similar to that reported for a trivalent test utilizing colorized latex reagents (13). The sensitivity was much lower, however, than has been reported for rapid detection of group A BHS by similar methods (8, 14). This difference may relate, at least in part, to the numbers of group A or C organisms in the pharynx and on the throat swab. The positive correlation between test sensitivity and number of beta-hemolytic colonies isolated in throat culture has been described previously for group A BHS (8, 14). There are several possible explanations for the few positive rapid tests which were not confirmed by culture. Antigenic cross-reactions between group C BHS and S. pneumoniae and Klebsiella pneumoniae have been reported by using latex agglutination methods with blood culture and skin swab specimens (9, 12, 17). False-positive results of the trivalent color test for group C organisms have also been reported for individuals with blood group A (13). Alternatively, the rapid tests may have been true positives, whereas
717
the corresponding throat cultures may have been falsely negative. Swab-to-swab sampling error has been described for 5 to 10% of patients who have two cultures for group A BHS performed, and this error presumably can also occur in the detection of group C BHS. Furthermore, some group C organisms (S. dysgalactiae) may not be recognized upon culture because they appear alpha-hemolytic (3). Falsely negative cultures could also occur if the patients had used an antibacterial gargle or an unprescribed antibiotic. In such cases, nonviable group C BHS might be detected by the rapid test but not by culture. In summary, this latex agglutination test was very specific but poorly sensitive compared with culture for the rapid detection of group C BHS directly from throat swab specimens. The sensitivity was higher for cultures with high colony counts of group C organisms. Many uncertainties remain concerning the role of group C BHS in causing endemic pharyngitis, including the frequency, seasonality, and duration of asymptomatic carriage of group C BHS; the relationship between colony count and the likelihood of carriage versus infection; the role of S. equisimilis versus S. anginosus in causing pharyngitis; the serologic correlates of group C infection; and the microbiologic and clinical effects of antibiotic therapy. Additional research to clarify these issues should help to define the potential usefulness of the rapid latex test in clinical practice. On the basis of current information, however, rapid antigen detection tests cannot be recommended as a substitute for culture and identification to the species level of group C isolates. We appreciate the laboratory assistance of Sharon Dudley, Tami Geuder, Cindy Ross, Jill Tartaglino, Kathleen Ashe, William Herron, and Carol Brunson. The reagents for the latex agglutination testing were provided by Diagnostic Products Corp.
REFERENCES 1. Benjamin, J. T., and V. A. Perriello, Jr. 1976. Pharyngitis due to group C hemolytic streptococci in children. J. Pediatr. 89:254-
256. 2. Bucher, C., and A. vonGraevenitz. 1984. Differentiation in throat cultures of group C and G streptococci from Streptococcus millen with identical antigens. Eur. J. Clin. Microbiol. 3:44 45. 3. Cimolai, N., R. W. Elford, L. Bryan, C. Anand, and P. Berger. 1988. Do the beta-hemolytic non-group A streptococci cause pharyngitis? Rev. Infect. Dis. 10:587-601. 4. Corson, A. P., V. F. Garagusi, and J. H. Chretien. 1989. Group C beta-hemolytic streptococci causing pharyngitis and scarlet fever. South. Med. J. 82:1119-1121. 5. Duca, E., G. Teodorovici, C. Radu, A. Vita, P. TalasmanNiculescu, E. Bernescu, C. Feldi, and V. Rosca. 1969. A new nephritogenic streptococcus. J. Hyg. Camb. 67:691-698. 6. Facklam, R. R. 1987. Specificity study of kits for detection of group A streptococci directly from throat swabs. J. Clin. Microbiol. 25:504-508. 7. Inzana, T. J., and B. Iritani. 1989. Rapid detection of group C streptococci from animals by latex agglutination. J. Clin. Microbiol. 27:309-312. 8. Kaplan, E. L. 1988. The rapid identification of group A betahemolytic streptococci in the upper respiratory tract: current status. Pediatr. Clin. North Am. 35:535-542. 9. Lee, P. C., and B. L. Wetherall. 1987. Cross-reaction between Streptococcus pneumoniae and group C streptococcal latex reagent. J. Clin. Microbiol. 25:152-153. 10. Meier, F. A., R. M. Centor, L. Graham, Jr., and H. P. Dalton. 1990. Clinical and microbiologic evidence for endemic pharyngitis among adults due to group C streptococci. Arch. Intern. Med. 150:825-829. 11. Mogabgab, W. J. 1970. Beta-hemolytic streptococcal and con-
718
NOTES
current infections in adults and children with respiratory dis1958 to 1969. Ani. Rev. Respir. Dis. 102:23-34. 12. Petts, D. N. 1984. Early detection of streptococci in swabs by ease,
latex agglutination before culture. J. Clin. Microbiol. 19:432433. 13. Petts, D. N., A. Lane, P. Kennedy, S. G. Hadfield, and M. B. McIllmurray. 1988. Direct detection of groups A, C and G streptococci in clinical specimens by a trivalent colour test. Eur. J. Clin. Microbiol. Infect. Dis. 7:34-39. 14. Radetsky, M., J. A. Solomon, and J. T. Todd. 1987. Identification of streptococcal pharyngitis in the office laboratory: reassessment of new technology. Pediatr. Infect. Dis. J. 6:556563.
J. CLIN. MICROBIOL. 15. Ruch, C. R., and D. G. Beckwith. 1984. Pharyngitis-tonsillitis in a college population. J. Am. Coll. Health 32:222-226. 16. Schwartz, R. H., and S. T. Shulman. 1986. Group C and group G streptococci. In-office isolation from children and adolescents with pharyngitis. Clin. Pediatr. 25:496-502. 17. Shlaes, D. M., Z. Toossi, and A. Patel. 1984. Comparison of latex agglutination and immunofluorescence for direct Lancefield grouping of streptococci from blood cultures. J. Clin. Microbiol. 20:195-198. 18. Turner, J. C., G. F. Hayden, D. Kiselica, J. A. Lohr, C. F. Fishburne, and D. Murren. 1990. Association of group C betahemolytic streptococci with endemic pharyngitis among college students. JAMA 264:2644-2647.
Vol. 30, No. 3
NOTES Latex Agglutination Testing Directly from Throat Swabs for Rapid Detection of Beta-Hemolytic Streptococci from Lancefield Serogroup C GREGORY F. HAYDEN,1* JAMES C. TURNER,2 DARIA KISELICA,3 MELISSA DUNN,' AND J. OWEN HENDLEY1 Departments of Pediatrics' and Student Health,3 University of Virginia Health Sciences Center, Charlottesville, Virginia 22908, and Thomson Student Health Center, University of South Carolina, Columbia, South Carolina 292082 Received 25 July 1991/Accepted 21 November 1991
A latex agglutination method for the rapid detection of beta-hemolytic streptococci from Lancefield serogroup C in throat swabs from 403 university students with symptomatic pharyngitis was evaluated. Compared with culture, the rapid test was poorly sensitive (34.4%) but very specific (98.4%) in detecting group C beta-hemolytic streptococci. The sensitivity of the rapid test improved with an increasing quantity of growth on culture.
Outbreaks of pharyngitis have been attributed to betahemolytic streptococci (BHS) from Lancefield serogroup C (1, 5), and an important role for these organisms in causing endemic pharyngitis among adolescents and young adults has been suggested (3, 4, 10, 11, 15, 16). In one recent study, group C organisms were isolated more than twice as often from the throats of ill university students with pharyngitis than from those of age-matched, healthy controls (18). The issue of causality has remained controversial, however, because many of the earlier studies had not been thorough or properly controlled and because the serologic correlates of group C infection have not been well defined. Latex agglutination methods for the rapid detection of group A BHS directly from throat-swab specimens are widely used in clinical practice (6, 8, 14). The negativecontrol latex reagent in one commercially available kit consists of latex particles coated with antibodies directed against the group C carbohydrate. This control reagent has been shown to be sensitive and specific in identifying group C BHS in veterinary specimens (7). The purpose of this study was to determine the accuracy of this reagent compared with that of culture in detecting group C BHS in throat swab specimens from young adults with symptomatic pharyngitis. Students presenting to the Student Health Centers at the University of Virginia and the University of South Carolina with a chief complaint of sore throat were studied from 1 October 1989 until 31 January 1990. Throat swab specimens were obtained simultaneously with two (University of Virginia) or three (University of South Carolina) rayon-tipped swabs. One swab was used to perform the latex agglutination testing (PathoDx Strep A Kit; Diagnostic Products Corp., Los Angeles, Calif.). The swab was immersed in an acidic solution for 1 min at room temperature. Any liberated *
antigen was then agglutinated by sensitized latex particles during a 4-min incubation on a clear glass plate on a mechanical rotator. Tests for group A and group C carbohydrate were performed in adjacent test circles. Agglutination was assessed by using indirect lighting against a black background. In accordance with the manufacturer's instructions, the test was considered positive when there was moderate or strong agglutination and negative when there was no agglutination and both circles remained milky white. The test was considered uninterpretable, and counted as negative, when there was a finely granular appearance (a slight roughening) in the test circles. The potencies of the test reagents were verified daily with positive control swabs impregnated with inactive group C BHS. The other swabs were used to inoculate two 5% sheep blood agar plates. At the University of Virginia, a single swab was used to inoculate both plates. The plate inoculated first was incubated at 35'C in a 5% C02-enriched atmosphere. The second plate was incubated at 35°C in an anaerobic environment (Anaerobic GasPak System; Baltimore Biologic Laboratory, Cockeysville, Md.). At the University of South Carolina, a separate swab was obtained to inoculate each plate. The growth of typical beta-hemolytic colonies at 24 and 48 h was quantitated as follows: 1+, 100 colonies. Overall test sensitivity and specificity were calculated by considering a culture positive when either the anaerobic or the C02-enriched culture contained group C BHS and negative when neither culture contained group C BHS. BHS were grouped by latex agglutination (PathoDx Strep Grouping Kit; Diagnostic Products Corp.). Isolates of group C BHS were frozen in sterile skim milk for later identification to the species level. The group C BHS include three species (Streptococcus equi, S. equisimilis, and S. zooepidemicus) which produce large colonies and a proportion of S. anginosus ("S. milleri") which produce small colonies and may also belong to other serogroups. The three large-colony species are common veterinary pathogens, but S. equisimilis
Corresponding author. 716
NOTES 1992 VOL. 1992 VOL. 30,30,
TABLE 1. Likelihood of a positive rapid test for group C BHS according to streptococcal species and quantity of growth on culture No. of positive rapid tests/no. testeda with the
following quantity of growth on culture
Species
S. equisimilis S. anginosus Both Not determined a
1+
2+
3+
4+
0/3 2/12 0/0 1/4
1/5 1/20
2/7 4/10
13/16 4/8
2/2 0/1
0/0 0/2
1/1 1/2
Categories for quantity of growth are as defined in the text.
is the most frequent human isolate. When reisolated, colonies with a positive Voges-Proskauer reaction were classified as S. anginosus (2). Colonies which were VogesProskauer negative were classified as S. equisimilis if they fermented 1% trehalose but not sorbitol. Group C BHS were detected among 93 (23.1%) of 403 students tested. Compared with combined culture, the rapid latex agglutination test was poorly sensitive (32 of 93, or 34.4%) but very specific (305 of 310, or 98.4%) in detecting group C BHS. Test sensitivity increased only slightly when the rapid-test result was compared with that of the C02enriched culture only (25 of 68, or 36.8%) or that of the anaerobic culture only (32 of 82, or 39.0%). The predictive values of positive and negative rapid tests were 86.5 and 83.3%, respectively. Test accuracies were similar at both study sites. The likelihood of a positive rapid test increased with an increasing quantity of growth on the culture plate (X2 for linear trend = 19.4; P < 0.001) (Table 1). More than 70% of cultures with 4+ growth of group C BHS were accompanied by a positive rapid test. A total of 84 (90%) patients were identified as infected with group C BHS, as follows: 31 (37%) with S. equisimilis only; 50 (60%) with S. anginosus only; and 3 (4%) with both S. equisimilis and S. anginosus. The rapid test was positive for 16 (52%) of the 31 patients with S. equisimilis, compared with only 11 (22%) of the 50 patients with S. anginosus group C (P < 0.01). Much of this variation in test sensitivity could be attributed to species differences in heaviness of growth. S. equisimilis strains were more commonly isolated in 4+ quantity than were S. anginosus strains (52 versus 16%; P < 0.01). The sensitivity of this latex agglutination method for the rapid detection of group C BHS directly from throat swabs was similar to that reported for a trivalent test utilizing colorized latex reagents (13). The sensitivity was much lower, however, than has been reported for rapid detection of group A BHS by similar methods (8, 14). This difference may relate, at least in part, to the numbers of group A or C organisms in the pharynx and on the throat swab. The positive correlation between test sensitivity and number of beta-hemolytic colonies isolated in throat culture has been described previously for group A BHS (8, 14). There are several possible explanations for the few positive rapid tests which were not confirmed by culture. Antigenic cross-reactions between group C BHS and S. pneumoniae and Klebsiella pneumoniae have been reported by using latex agglutination methods with blood culture and skin swab specimens (9, 12, 17). False-positive results of the trivalent color test for group C organisms have also been reported for individuals with blood group A (13). Alternatively, the rapid tests may have been true positives, whereas
717
the corresponding throat cultures may have been falsely negative. Swab-to-swab sampling error has been described for 5 to 10% of patients who have two cultures for group A BHS performed, and this error presumably can also occur in the detection of group C BHS. Furthermore, some group C organisms (S. dysgalactiae) may not be recognized upon culture because they appear alpha-hemolytic (3). Falsely negative cultures could also occur if the patients had used an antibacterial gargle or an unprescribed antibiotic. In such cases, nonviable group C BHS might be detected by the rapid test but not by culture. In summary, this latex agglutination test was very specific but poorly sensitive compared with culture for the rapid detection of group C BHS directly from throat swab specimens. The sensitivity was higher for cultures with high colony counts of group C organisms. Many uncertainties remain concerning the role of group C BHS in causing endemic pharyngitis, including the frequency, seasonality, and duration of asymptomatic carriage of group C BHS; the relationship between colony count and the likelihood of carriage versus infection; the role of S. equisimilis versus S. anginosus in causing pharyngitis; the serologic correlates of group C infection; and the microbiologic and clinical effects of antibiotic therapy. Additional research to clarify these issues should help to define the potential usefulness of the rapid latex test in clinical practice. On the basis of current information, however, rapid antigen detection tests cannot be recommended as a substitute for culture and identification to the species level of group C isolates. We appreciate the laboratory assistance of Sharon Dudley, Tami Geuder, Cindy Ross, Jill Tartaglino, Kathleen Ashe, William Herron, and Carol Brunson. The reagents for the latex agglutination testing were provided by Diagnostic Products Corp.
REFERENCES 1. Benjamin, J. T., and V. A. Perriello, Jr. 1976. Pharyngitis due to group C hemolytic streptococci in children. J. Pediatr. 89:254-
256. 2. Bucher, C., and A. vonGraevenitz. 1984. Differentiation in throat cultures of group C and G streptococci from Streptococcus millen with identical antigens. Eur. J. Clin. Microbiol. 3:44 45. 3. Cimolai, N., R. W. Elford, L. Bryan, C. Anand, and P. Berger. 1988. Do the beta-hemolytic non-group A streptococci cause pharyngitis? Rev. Infect. Dis. 10:587-601. 4. Corson, A. P., V. F. Garagusi, and J. H. Chretien. 1989. Group C beta-hemolytic streptococci causing pharyngitis and scarlet fever. South. Med. J. 82:1119-1121. 5. Duca, E., G. Teodorovici, C. Radu, A. Vita, P. TalasmanNiculescu, E. Bernescu, C. Feldi, and V. Rosca. 1969. A new nephritogenic streptococcus. J. Hyg. Camb. 67:691-698. 6. Facklam, R. R. 1987. Specificity study of kits for detection of group A streptococci directly from throat swabs. J. Clin. Microbiol. 25:504-508. 7. Inzana, T. J., and B. Iritani. 1989. Rapid detection of group C streptococci from animals by latex agglutination. J. Clin. Microbiol. 27:309-312. 8. Kaplan, E. L. 1988. The rapid identification of group A betahemolytic streptococci in the upper respiratory tract: current status. Pediatr. Clin. North Am. 35:535-542. 9. Lee, P. C., and B. L. Wetherall. 1987. Cross-reaction between Streptococcus pneumoniae and group C streptococcal latex reagent. J. Clin. Microbiol. 25:152-153. 10. Meier, F. A., R. M. Centor, L. Graham, Jr., and H. P. Dalton. 1990. Clinical and microbiologic evidence for endemic pharyngitis among adults due to group C streptococci. Arch. Intern. Med. 150:825-829. 11. Mogabgab, W. J. 1970. Beta-hemolytic streptococcal and con-
718
NOTES
current infections in adults and children with respiratory dis1958 to 1969. Ani. Rev. Respir. Dis. 102:23-34. 12. Petts, D. N. 1984. Early detection of streptococci in swabs by ease,
latex agglutination before culture. J. Clin. Microbiol. 19:432433. 13. Petts, D. N., A. Lane, P. Kennedy, S. G. Hadfield, and M. B. McIllmurray. 1988. Direct detection of groups A, C and G streptococci in clinical specimens by a trivalent colour test. Eur. J. Clin. Microbiol. Infect. Dis. 7:34-39. 14. Radetsky, M., J. A. Solomon, and J. T. Todd. 1987. Identification of streptococcal pharyngitis in the office laboratory: reassessment of new technology. Pediatr. Infect. Dis. J. 6:556563.
J. CLIN. MICROBIOL. 15. Ruch, C. R., and D. G. Beckwith. 1984. Pharyngitis-tonsillitis in a college population. J. Am. Coll. Health 32:222-226. 16. Schwartz, R. H., and S. T. Shulman. 1986. Group C and group G streptococci. In-office isolation from children and adolescents with pharyngitis. Clin. Pediatr. 25:496-502. 17. Shlaes, D. M., Z. Toossi, and A. Patel. 1984. Comparison of latex agglutination and immunofluorescence for direct Lancefield grouping of streptococci from blood cultures. J. Clin. Microbiol. 20:195-198. 18. Turner, J. C., G. F. Hayden, D. Kiselica, J. A. Lohr, C. F. Fishburne, and D. Murren. 1990. Association of group C betahemolytic streptococci with endemic pharyngitis among college students. JAMA 264:2644-2647.
