THE HUMAN PAPILLOMAVIRUS (HPV) has been recognized as the causative agent of cervical cancer since the 1970s, when zur Hausen and Meisels and their colleagues detected HPV DNA in cervical lesions. 1,2 The link between HPV and cervical cancer has remained strong, as over 93% of all cervical cancers have been found to contain high-risk types of HPV. 3–5 In addition to its oncogenic potential, HPV is the most common sexually transmitted pathogen; the annual incidence of HPV infection in the United States is estimated to be as high as 5.5 million, and an estimated 20 million people are currently infected. 6
Testing for high-risk types of HPV was recently approved as a clinical management strategy to determine which women with an atypical squamous cell of undetermined significance (ASCUS) cytology report are most likely to have cervical intraepithelial neoplasia (CIN). 7 Testing for high-risk types of HPV as a primary screen for cervical cancer is equal, if not superior, to the Papanicolaou (Pap) smear in identifying older women at risk for CIN. 8,9 Testing for high-risk types of HPV shows promise for identifying disease regression in women with biopsy-confirmed CIN 1 7 or for follow-up after ablative or excisional treatment for high-grade or glandular disease. 11,12 The methods used to sample the cervicovaginal area for the presence of HPV have included clinician-directed cervical sampling and self-administered techniques. The self-administered techniques can be more comfortable and less invasive than the traditional speculum-assisted examination while providing equivalent HPV detection. Specifically, Dacron swabs, 13–15 a cytobrush, 16 and tampons 17–20 have been shown to be equivalent to clinician-directed sampling.
Among self-collection methods, tampons—with which women are more familiar and comfortable than novel self-collection methods—are particularly attractive for those populations where tampons are used. 13,19 Familiar, convenient, and comfortable self-sampling devices may be more likely to motivate unscreened or underscreened women to be screened 21–23 and are more likely to improve women's compliance in research protocols requiring frequent sampling. However, the benefits of using tampons as a self-sampling device should be weighed against the tampon's ability to detect HPV. A tampon exposed to cervicovaginal cells for 10 seconds identified fewer women with high-risk HPV types than clinician-directed swabs. 13 No studies to date have examined the use of longer tampon insertion times to detect high-risk HPV in the cervicovaginal vault.
This study was designed to compare the detection of high-risk HPV by tampons with longer exposure times in the cervicovaginal vault versus self-swab collection techniques. Women's acceptance of sampling with a tampon for longer periods was also assessed.
Between March 1998 and July 1999, women at Dartmouth teaching clinics were asked to participate, without compensation, in this prospective longitudinal study. All women had been referred to colposcopy for either abnormal Pap smear findings or normal Pap smear findings with abnormal cervical examination findings, were at least 18 years of age, and were not pregnant. Ten women were colposcopically normal; 93 women had abnormalities and were biopsied to determine disease severity. No woman was treated until after the sample collection period was completed. The final disease status was determined by biopsy.
This study required four tampon self-samplings. Women were instructed with verbal and written materials on how to self-sample while at the office. The first sampling occurred at the colposcopy visit, at which time the woman self-sampled with two consecutive swabs and a tampon for 10 to 15 seconds. The clinician sampled the cervix with two swabs before the colposcopy. The order of these five samplings was randomized to avoid sampling-order bias. The second through fourth tampon self-samplings were scheduled to occur in the participants’ homes 1 week, 2 weeks, and 3 weeks after the initial office visit.
At each of the home-based samplings, two consecutive swabs were used concurrently with the tampon in the same random order assigned at study entry. Concurrent swabs were used as a control for possible fluctuations in positivity that might occur with weekly testing. 24 The tampon remained in the vagina for progressively longer times: 1 hour at week 1, 4 hours at week 2, and overnight at week 3 (women were instructed to insert the tampon before going to sleep and to remove the tampon immediately upon arising in the morning).
The Committee for the Protection of Human Subjects at Dartmouth Medical School approved the study, and consent was obtained from all participants. Demographic and behavioral information was collected after consent was obtained.
Each home self-sampling kit contained two Dacron swabs, a tampon, self-sealing tubes for swab and tampon storage and shipping, and a mailing kit to return all samples after each sampling period. Tubes were provided with 1 ml and 20 ml of PreservCyt (Cytyc Corporation, Boxborough, MA) for the swabs and tampon, respectively. The 1-ml and 20-ml quantities of PreservCyt allowed the swab and the tampon to become completely saturated, with a small reservoir of PreservCyt pooling in the bottom of the respective tubes. Written instructions on how to assemble the mailing kit were included, and the mailing kit was labeled “medical specimen” to meet the shipping requirements of the package delivery services. Participants received reminder phone calls the day before each home self-sampling episode. If women were experiencing menses on the scheduled day of sampling, they were instructed to postpone sampling by 1 week. A calendar was provided for each participant to record the actual days of each sampling, along with any days of menstrual bleeding. Thus, 4 sets of 2 consecutive swabs and tampon self-collection samples were tested (i.e., 8 swabs and the 10-second tampon, 1-hour tampon, 4-hour tampon, and overnight tampon).
Samples were tested for 18 high-risk HPV types (16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 55, 56, 58, 59, 68, MM4, MM7, and MM9) and 9 low-risk HPV types (6, 11, 40, 42, 53, 54, 57, 66, and MM8) by polymerase chain reaction (PCR) amplification using the MY09-MY11-HMBB01 L1 consensus primer system 25 and a reverse line blot detection strip, which individually identifies each of these HPV types (Roche Molecular Systems, Alameda, CA). The method has been described in detail elsewhere. 13–15
Participants received an acceptability questionnaire at the time of the clinic visit and were asked to complete and return the questionnaire following completion of all home samplings. The questionnaire, which was designed for the current study, consisted of approximately 30 closed-ended (i.e., dichotomous or Likert-scale format) and open-ended questions. Questions assessed history of tampon use, feasibility of tampon use, preferences for cervical cancer screening, and experiences with using tampons for self-sampling.
Sampling techniques were compared by means of a two-tailed McNemar test at a 0.05 level of significance with Statistica 26 software looking at alpha- and beta-type discordant pairs. The data were stratified by histology (normal, CIN 1, CIN 2/3). The sampling techniques were analyzed with chi-square statistics for perfect, partial, and no agreement proportions among high-risk HPV types at each sampling. Logistic regression was used to show the odds ratios of detecting high-risk HPV and CIN 2/3 for the tampons in comparison with their contemporaneous swabs, with a type I error of 0.05. The testing characteristics of each tampon method included determination of 95% confidence intervals in addition to chi-square testing. The acceptability responses were compared by Kruskal–Wallis one-way analyses of variance and post hoc tests.
We enrolled 103 rural, mostly white women in the study who had presented for colposcopy because of ASCUS (53%), low-grade squamous intraepithelial lesions (LSIL) (23%), high-grade squamous intraepithelial lesions (HSIL) (12%), atypical glandular cells of undetermined significance (AGUS) (2%), and normal cytology with visual cervical abnormality (10%). Fifty were aged 35 years or younger, and 53 were older than age 35 years (range, 18–68). Sixty-nine women had given birth to one or more children. Biopsy revealed CIN 1 in 50 women (48.5%) and CIN 2/3 in 6 (5.8%). Other descriptors of this population have been published previously. 13
Contemporaneous comparisons of sampling techniques removed any temporal biases that could have occurred because of detection variations among techniques, viral clearing, or new acquisition of HPV infections. Two consecutive swabs were chosen as the comparator because they are equivalent to clinician-directed ecto-endocervical swabs, after adjustment for all other techniques, for detecting high-risk HPV (aOR = 1.00; 95% CI: 0.56–1.77); a single self-sampled swab was inferior to the clinician-directed swabs (aOR = 0.63; 95% CI: 0.40–0.91). 13
Participants received 412 tampons and 824 swabs at study enrollment, accounting for the 4 weekly tampons and the 8 self-swabs for each; 103 tampons and 206 swabs were processed from the initial clinic visit. From the 309 home kits, 257 tampons and 510 swabs were mailed back to our laboratory, for an overall return rate of 83%; 14.6% of the original cohort of women completing the office visit did not complete the remainder of the samplings. Of those 88 women who continued in the study, 2.3% did not complete the third sampling, and an additional 4.7% did not complete the last sampling. Overall, 82 women (79.6%) returned all four sets of samples.
With all swabs and tampons, high-risk HPV types were detected 266 times; type 16 was detected in 63 samples (23.7%), type 39 in 35 (13.2%), type 18 in 27 (10.2%), and type 52 in 25 (9.4%). The remaining high-risk types, in descending order of detection rate, were 51, MM9, 58, 33, 45, 59, 26, 55, 56, and MM4.
There were 42 women (40.8%) who tested positive for any high-risk HPV in at least 1 of the 5 sampling methods (concurrent consecutive swabs, 10-second tampon, 1-hour tampon, 4-hour tampon, and overnight tampon). Of those 42, 26 tested positive with the 10-second tampon, 24 with the 4-hour tampon, 20 with the overnight tampon, and 16 with the 1-hour tampon. A paired comparison of the two consecutive swabs and each tampon with use of McNemar paired proportional testing (shown in Table 1) indicated that the 10-second tampon was significantly less likely to detect high-risk HPV types than the swabs (chi-square = 7.11;P = 0.0077). The 1-hour and 4-hour tampons for all-aged women together identified high-risk HPV in a number of women that was similar to the number identified with their consecutive swabs, as demonstrated by the nonsignificant McNemar test findings. The McNemar test values, substratified by age (≤35 and >35 years), showed no change in the results (data not shown).
Discriminating results were found when high-risk HPV infections were subdivided by the woman's associated histology (Table 2). Among those with normal histology, the 10-second tampon identified fewer women with high-risk HPV than the contemporaneous consecutive swabs (chi-square = 4.17;P = 0.0412). The tampons exposed to cervicovaginal cells for 1 hour, 4 hours, and overnight in histologically normal women identified a number of women with high-risk HPV that was similar to the number identified with the swabs, as shown by the nonsignificant McNemar test findings.
Of the women whose histology finding was CIN 1 or CIN 2/3 (CIN 1+), tampons from all four time exposures identified numbers with high-risk HPV that were equivalent to the numbers identified with the consecutive swabs.
When agreement was defined as the positivity of the tampon and the concurrent self-swabs for any high-risk HPV types, with either perfect or partial agreement with regard to type, the overnight tampon had the highest agreement at 80%, followed by the 10-second, 4-hour, and 1-hour tampons, at 72.2%, 68.9%, and 58.3%, respectively (calculated by adding the perfect and partial rates in Table 3). None of the four agreement rates was statistically significant. The four agreement rates for women with CIN likewise did not differ among sampling times.
Perfect agreement meant that there was concordance of all high-risk HPV types between the tampon and its concurrent swabs. For all women, the perfect agreement rates for the tampon–swab pairs were equivalent among all tampon exposures and ranged from 33.3% to 64.0%. The perfect agreement rates for women with CIN 1+ similarly showed equivalence among all tampon–swab pairs. The perfect agreement rate for women with CIN 2/3 was 100% for all tampon exposures except the 1-hour exposure, for which it was 0%. Overall, there was a significantly greater proportion of perfect high-risk HPV type agreement among women with CIN than among women with normal histology (76.8% versus 23.2%; chi-square = 11.07;P = 0.0009).
Partial agreement occurred in 24 cases when either the tampon or the swabs detected 1 or more high-risk HPV types than the other. For all women and women with CIN, partial agreement ranged from 11% to 29%. Overall, there were equal numbers of women whose tampons detected at least one more high-risk HPV type than the swabs and whose swabs detected at least one more high-risk HPV type than the tampons (43.5% versus 56.5%;P = NS). The tampons and swabs each detected more high-risk HPV an equal number of times at all four time exposures. Women with CIN tended to have more partial agreements than histologically normal women (70.8% versus 29.2%; chi-square = 2.83;P = 0.0924).
Among all cases in which samples had no agreement, 33 involved detection of high-risk HPV by one technique but not the other. One case, that was verified, involved the detection of different high-risk HPV types with each technique. Among the 33 no-agreement tampon–swab pairs, 81.3% involved detection of one or more high-risk HPV types by the swabs alone, whereas 18.8% involved detection of one or more high-risk HPV types by the tampons alone (chi-square = 8.71;P = 0.0032). The swabs exclusively detected the high-risk HPV at the 10-second and overnight times, and at 1 and 4 hours detected the high-risk HPV twice as often as the tampons did. Of the tampon–swab pairs that did not agree, the proportion from histologically normal women was equivalent to the proportion from women with CIN (51.5% versus 48.5%;P = NS).
Table 4 supports the equivalency of all tampon exposures with the concurrent swabs. With adjustment for age and comparison with each set of contemporaneous swabs, the odds of detecting high-risk HPV were 0.58, 0.68, 0.71, and 1.00 for the 10-second, overnight, 1-hour, and 4-hour tampons, respectively, with confidence intervals overlapping by 1.0. All tampons were equivalent to the consecutive self-swabs for CIN 2/3 detection.
The test characteristics of the four tampon groups are presented in Table 5. For the detection of CIN 2/3, the 10-second and 4-hour tampons had excellent sensitivity (83.3% and 80.0%) and specificity (89.4% and 85.0%), with high negative predictive values (97.7% and 97.1%). The sensitivity and negative predictive value of the four tampon groups for detecting CIN 1+ disease was lower than for detecting CIN 2/3. There was no difference in the test characteristics of different tampon times for each histology grouping, CIN 1+ and CIN 2/3, by chi-square analyses.
The feasibility of using home sampling was documented. Sixty-seven women completed the survey, for an overall response rate of 65% (n = 28, or 56%, of the women aged 35 years or younger; n = 39, or 74%, of the women older than age 35 years). The survey asked participants whether they experienced any difficulty putting the home tampon samples in the tubes of PreservCyt and whether they experienced any difficulty putting these tubes in mailing kits. None of the women who responded to the survey reported any problems with either of these tasks.
Upon receipt of the samples mailed from home, the study manager documented any problems with the tubes. The most frequently stated reported for 16 of the 257 mailed tampons (6.2%), was leakage of the surplus PreservCyt from the container into the self-sealing biohazard bag during return-mailing. The tampons retained the PreservCyt they had originally absorbed. Fourteen of the 257 tampons (5.4%) were returned without any surplus PreservCyt but with the tampons still moist. Neither the leakage nor the absence of surplus PreservCyt affected our ability to use the tampons. All samples formed sizable cellular pellets with processing and tested positive for the β-globin control, ensuring that there was a cellular yield from each sample. For 4 of the 257 tampons (1.9%), there was more liquid in the tube than originally placed. The unknown diluent did not affect our ability to detect cellular DNA.
The acceptability of using tampons for varying times was ascertained at the completion of the study. For each self-sampled tampon, we asked participants to rate how “bothersome” the tampon was for the duration of sampling and during removal, on a 5-point Likert scale (1 = not bothersome; 5 = extremely bothersome). The acceptability of the four possible durations for tampon sampling was compared with Kruskal–Wallis one-way ANOVAs and post hoc tests using an alpha of 0.05. Women were increasingly more bothered by the increasing duration of time the tampon remained in the vagina (P = 0.0041), as seen in Table 6. The 10-second tampon and the 1-hour tampons engendered significantly less bother than the overnight tampon, as revealed by post hoc tests (mean [SD]: 1.29 [0.85] and 1.34 [0.82] versus 1.84 [1.22];P = 0.0152 and P = 0.0151, respectively).
Parity and age affected the woman's level of acceptance. All women, regardless of age, who had never given birth ranked all four tampon durations equivalently, ranging from 1.11 (0.47) to 1.71 (1.26). Regardless of age, women who had given birth accepted the 10-second and 1-hour tampon times better than the overnight time (H = 12.87;P = 0.0049). Women aged 35 years or younger who had given birth ranked the overnight duration worse than the 10-second duration in post hoc testing (2.44 [1.33] versus 1.39 [0.96];P = 0.0289). Women older than 35 years of age who had given birth ranked the 1-hour tampon most comfortable, at 1.13 (0.46), which differed in post hoc testing from the overnight duration rating of 1.83 (1.19) (P = 0.0120). All women ranked the removal of the tampon after 10 seconds, 1 hour, 4 hours, and overnight equivalently minimally bothersome.
Forty-two (63%) of participants expressed no concerns at all about using tampons for self-sampling. For women who did, the overwhelming concern was vaginal dryness. Only 13% (n = 9) reported that they worried about toxic shock syndrome while using the tampons.
In comparison, 21% of participants (n = 14) expressed concerns about using swabs for self-sampling. The most frequently voiced concern was whether the swab sampling was done correctly (e.g., “I couldn't tell if I was touching the cervix,” “I wasn't sure if I inserted it far enough”). One other concern was the possibility of swab breakage during sampling, which occurred for three women.
Women use tampons to detect Chlamydia, gonorrhea, and Trichomonas, in addition to HPV. 13–18,21–23 In cultures where tampons are used, they are familiar and acceptable to women. 13 The present work shows that the 10-second tampon duration is less effective than two consecutive swabs only in a cohort of histologically normal women, such as one to be observed for HPV acquisition. The restriction to histologically normal women had not been associated with the inferiority of the 10-second tampon insertion in previous work because only clinician-directed sampling was used as the comparator. 13 The current study suggests that in women with CIN, all tampon exposures, from 10 seconds to overnight, are equivalent to their respective swabs for detecting high-risk HPV.
All high-risk HPV agreement rates (perfect, partial, and none) reinforced the equivalence of the four tampon exposure times for detecting high-risk HPV. Perfect agreement rates occurred much more frequently for women with CIN than for histologically normal women, a finding suggesting that repeated HPV testing for perfect agreement as a testing strategy may improve cervical cancer screening.
The perfect type-specific agreement rates of the 10-second, 1-hour, 4-hour, and overnight tampon insertions with the respective swabs averaged 50%, ranging from 35% to 64%, a rate consistent with perfect type-specific agreement rates found in other studies. 15
The rates of partial and no agreement were higher than expected, ranging from 16% to 42%. Reasons for this discordance include laboratory and collection technique issues. The laboratory issues include the processing of the tampon or swab and the PCR detection of HPV DNA. Our processing step is limited by the lack of standardization of the number of cells aliquotted for the PCR step. Both the swabs and tampons could have collected differing amounts of cells at each of the four sampling times.
The tampon is proven to collect a sizable cellular pellet that the swab cannot (unpublished data), which could increase the possible variability in cell concentration aliquotted from each tampon sample for the PCR. Any PCR detection technique is limited by competition kinetics that occur when many types of HPV are present in abundance. Competition can cause reaction inhibition, resulting in nondetection of other low- or high-risk HPV types present. Twelve of our discordant pairs were positive for many types of HPV, all occurring with very strong signal intensity, possibly outcompeting other low- or high-risk HPV types for detection on both the tampons and swabs. PCR competition may have caused these samples to lack perfect type agreement. Standardization of the cell concentration in the processing step may provide a constant DNA concentration and prevent the lack of detection of some HPV types because of competition.
Four of our discordant pairs had one weakly positive PCR signal for a type of HPV not present in the other sampling technique. Our PCR system used the MY09/MY11 primer set. Subsequent to this study, a PGMY primer set was introduced as a more analytically sensitive system for type-specific HPV identification. 27 Specifically, the PGMY primer set uses multiple primer sequences to detect high-risk HPV types 26, 52, 55, and 59 more often than the MY09-MY11 primers do. These four HPV types accounted for 15.4% of our HPV results, indicating that these types could have been present in a higher proportion and with stronger signal intensity if the more sensitive analytic method had been available to use.
Contamination of PCR products must always be considered when there is a sample weakly positive for an HPV type occurring in none of the other swabs and/or tampons. Two of our samples had a single weak signal over all four sampling times. Contamination is actively prevented in our laboratory by keeping the processing and the reaction steps in separate rooms, by changing body suits after leaving the processing room and before entering the PCR room, by changing gloves both after the processing of each sample and after each PCR run, and by the use of multiple positive and negative controls. It is highly unlikely that the two singly positive weak-signal samples were due to laboratory contamination.
Discordance could also be due to the specific collection techniques. The collection techniques used in this study could have inherently different abilities to detect cells infected with high-risk HPV. Our study was designed to determine this. Forty-two (72.4%) of the discordant samples could not be attributed to laboratory causes and thus were considered to be due to collection technique differences. The discordant pairs caused only by collection technique differences were attributed equally to the tampons and swabs, indicating no intrinsic difference in the ability of tampons to detect any high-risk HPV in comparison with the swabs. This would indicate that the use of tampons, regardless of time exposed to cervicovaginal cells or the woman's associated histology, is a viable option for testing for high-risk HPV.
However, even if the PGMY primer system eliminated all the discordant pairs due to laboratory issues found in this study, this more sensitive PCR system could have created new discordant pairs by detecting high-risk HPV in previously negative samples. Hence, it is not possible to separate the effects of the collection technique from the analytic laboratory method used. We recommend that future studies use the PGMY primer set to minimize any discordance from laboratory causes.
The test characteristics (rates) of the 10-second, 4-hour, and overnight tampons for detecting high-risk HPV to predict CIN 2/3 were as high as those reported for clinician-directed samples. 7 The lower sensitivity rates for predicting any CIN reflected the combination of low- and high-risk HPV types present in CIN 1; the lower rates have been previously reported. 28 The increasing time exposures did decrease the acceptability ratings of tampons used for longer intervals, especially for women who had given birth. Potentially, the responsibilities associated with motherhood led to less tolerance of a test that could require a long sampling duration.
The technical feasibility of sending sample kits home with women to be used at a later date and mailed to the processing laboratory is excellent. The loss of the surplus transport medium did not diminish the cellular fixation that occurred with the tampon absorption of the original PreservCyt. The cellular matter tested positive for human β-globin, the control for human DNA presence, in all cases. We recommend a double container system for future mailing kits to reduce the loss and/or evaporation of transport medium.
The initial lost-to-follow-up rate was 14.6% from the first office visit to the first home sampling, an excellent retention rate for longitudinal studies. Subsequently, the loss of participation was 2% to 4% between the three home samplings, also an outstanding compliance rate, given that the subjects were reminded no more than once per sample to complete the sampling and received no compensation for participation in any part of the study. No other screening test done repeatedly, such as fecal occult blood testing for colorectal cancer screening, has equivalent compliance rates.
In summary, a self-sampling technique for high-risk HPV detection is clinically important for the success of primary screening for cervical cancer, for triage of abnormal Pap test findings, and for follow-up testing after high-grade squamous or atypical glandular lesions have been excised. Self-sampling could be an efficient method to reach many women rapidly. For instance, to monitor the prevalence of HPV in public health efforts, 29,30 a self-sampling kit could be mass-mailed or mass-distributed at a health gathering. The choice of using swabs or tampons for high-risk HPV detection is dependent on the woman, her culture, and her clinician. Both methods allow women the comfort and convenience of self-sampling without the barriers associated with a clinician-directed speculum examination.
1. zur Hausen H. Condylomata acuminata and human genital cancer. Cancer Res 1976; 36: 794.
2. Meisels A, Fortin R. Condylomatous lesions of the cervix and vagina. I. Cytologic patterns. Acta Cytol 1976; 20: 505–509.
3. IARC. IARC Monograph on the Evaluation of Carcinogenic Risks to Humans. Vol 64: Human Papillomaviruses. Lyon, France: International Agency for Research on Cancer, 1995.
4. Bosch FX, Manos MM, Muñoz N, et al. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective. J Natl Cancer Inst 1995; 87: 796–802.
5. Walboomers J, Jacobs M, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol 1999; 189: 12–19.
6. Cates W, Am Social Health Association Panel. Estimates of the incidence and prevalence of sexually transmitted diseases in the United States. Sex Transm Dis 1999; 26: 52–57.
7. Wright TC Jr, Cox JT, Massad LS, Twiggs LB, Wilkinson EJ. 2001 Consensus Guidelines for the management of women with cervical cytogical abnormalities [Review]. JAMA 2002; 287: 2120–2129.
8. Schiffman MH, Herrero R, Hildesheim A, et al. HPV DNA testing in cervical cancer screening results from women in a high-risk province of Costa Rica. JAMA 2000; 283: 87–93.
9. Wright TC, Denny L, Kuhn L, Pollack A, Lorincz A. HPV DNA testing of self-collected vaginal samples compared with cytologic screening to detect cervical cancer. JAMA 2000; 283: 81–86.
11. Jain S, Tseng CJ, Horng SG, Soong YK, Pao CC. Negative predictive value of human papillomavirus test following conization of the cervix uteri. Gyn Oncol 2000; 82: 177–180.
12. Ronnett BM, Manos MM, Ransley JE, et al. Atypical glandular cells of undetermined significance (AGUS): cytopathologic features, histopathologic results, and human papillomavirus DNA detection. Human Pathol 1999; 30: 816–825.
13. Harper DM, Noll WW, Belloni DR, Cole BF. Randomized clinical trial of PCR-determined human papillomavirus detection methods: self-sampling versus clinician-directed–biologic concordance and women's preferences. Am J Obstet Gynecol 2002; 186: 365–373.
14. Moscicki AB. Comparison between methods for human papillomavirus DNA testing: a model for self-testing in young women. J Infect Dis 1993; 167: 723–725.
15. Gravitt PE, Lacey JV, Brinton LA, et al. Evaluation of self-collected cervicovaginal cell samples for human papillomavirus testing by polymerase chain reaction. Cancer Epidemiol Biomarkers Prev 2001; 10: 95–100.
16. Sellors JW, Lorincz AT, Mahony JB, et al. Comparison of self-collected vaginal, vulvar and urine samples with physician-collected cervical samples for human papillomavirus testing to detect high-grade squamous intraepithelial lesions. CMAJ 2000; 163: 513–518.
17. Hillemanns P, Kimmig R, Hüttemann U, Dannecker C, Thaler CJ. Screening for cervical neoplasia by self-assessment for human papillomavirus DNA. Lancet 1999; 354: 1970.
18. Harper DM, Hildesheim A, Cobb JL, Greenberg M, Vaught J, Lorincz AT. Collection devices for human papillomavirus. J Fam Pract 1999; 48: 531–535.
19. Fairley CK, Chen S, Tabrizi SN, Quinn MA, McNeil JJ, Garland SM. Tampons: a novel patient-administered method for the assessment of genital human papillomavirus infection. J Infect Dis 1992; 165: 1103–1106.
20. Coutlee F, Hankins C, Lapointe N. Comparison between vaginal tampon and cervicovaginal lavage specimen collection for detection of human papillomavirus DNA by the polymerase chain reaction. The Canadian Women's HIV Study Group. J Med Virol 1997; 51: 42–47.
21. Tabrizi SN, Fairley CK, Chen S, et al. Evaluation of patient-administered tampon specimens for Chlamydia trachomatis
and Neisseria gonorrhoeae
. Sex Transm Dis 2000; 27: 133–137.
22. Paterson B, Tabrizi SN, Garland SM, Fairley CK, Bowden FJ. The tampon test for trichomoniasis: a comparison between conventional methods and a polymerase chain reaction for Trichomonas vaginalis in women. Genitourin Med 1998; 74: 136–139.
23. Bowden FJ, Paterson BA, Mein J, et al. Estimating the prevalence of Trichomonas vaginalis, Chlamydia trachomatis, Neisseria gonorrhoeae
, and human papillomavirus infection in indigenous women in northern Australia. Sex Transm Infect 1999; 26: 431–434.
24. Wheeler CM, Greer CE, Becker TM, Hunt WC, Anderson SM, Manos MM. Short-term fluctuations in the detection of cervical human papillomavirus DNA. Obstet Gynecol 1996; 88: 261–268.
25. Bauer HM, Greer CE, Manos MM. Determination of genital human papillomavirus infection using consensus PCR. In: Herrington CS, McGee JOD, eds. Diagnostic Molecular Pathology: A Practical Approach. Oxford: Oxford University Press, 1992: 132–152.
26. StatSoft, Inc. Statistica for Windows [computer program manual]. Tulsa, Oklahoma: StatSoft, 2001.
27. Gravitt PE, Peyton CL, Alessi TQ, et al. Improved amplification of genital human papillomaviruses. J Clin Microbiol 2000; 38: 357–361.
28. Ratnam S, Franco EL, Ferenczy A. Human papillomavirus testing for primary screening of cervical cancer precursors. Cancer Epidemiol Biomarkers Prev 2000; 9: 945–951.
29. Rep. Thomas A. Coburn (R-OK), HR 3248, 106th Congress, 1st Session, 2001.
30. Castellsague X, Menendex C, Loscertales MP, et al. Human papillomavirus genotypes in rural Mozambique. Lancet 2001; 358: 1429–1430.