Sexual behavior and HPV infection in British women, by postal questionnaires and telephone interviews^{† ‡}

INTRODUCTION

Cervical cancer is the third most common cancer among women worldwide [Ferlay et al., 2010], and is more frequent in many socio-economically deprived populations [Almonte et al., 2008]. Human papillomaviruses (HPVs), most frequently HPV 16, are the primary cause of cervical carcinogenesis [International Agency for Research on Cancer, 1995]. Over 100 HPV types have now been described, including 13 “high-risk” types that are definitively associated with cervical cancer and several more that are probably carcinogenic [International Agency for Research on Cancer, 2007]. The overall prevalence of HPV in cervical cancers worldwide is over 99%, implying the highest attributable fraction ever identified for a specific cause of a major human cancer [Walboomers et al., 1999].

HPV is one of the most common sexually transmitted infections [Bosch et al., 2008]. Most women become infected during their lives, although the majority of the infections are transient and clear within 2 years [Plummer et al., 2007]. HPV genital acquisition is associated with sexual behavior patterns such as early age at first sexual intercourse, number of sexual partners, and sexual behavior of the partner; and certain co-factors (high parity, long-term use of hormonal contraceptives, and cigarette smoking) help in the progression from infection to cancer [Vaccarella et al., 2006; Almonte et al., 2008; Vaccarella et al., 2008].

This study describes sexual behavior patterns in relation to HPV prevalence and acquisition rates among British women who participated in the ARTISTIC trial (A Randomized Trial In Screening To Improve Cytology), which was conducted in Greater Manchester [Kitchener et al., 2006]. In order to compare responses and participation rates, the method of ascertaining sexual behavior was randomized between postal questionnaire, telephone interview, and both.

MATERIALS AND METHODS

RESULTS

Of the 1,880 women selected, 661 (34% in group 1, 37% in group 2, and 34% in group 3) agreed to participate (512 at first invitation), 182 (10%) refused, and 1,037 (55%) did not reply to the invitation or reminder. Among 1,037 non-respondents, two women died, 3 withdrew from the ARTISTIC trial, and 163 (16%) changed address during the invitation process (information from the trial coordination office or by returned mail). There were no significant differences between randomization groups (group 1: questionnaire; group 2: questionnaire excluding sexual history, short interview; group 3: questionnaire, longer interview) in response, refusal, and non-response rates. Nine women in group 2 and 14 in group 3 returned the questionnaire but were not interviewed. Table I shows overall participation rates in this study and numbers in each group by age, allocated screening protocol (HPV result revealed or concealed), cytology at entry, and HPV status at entry and at the second screening round in the ARTISTIC trial. Participation rates ranged from 29% to 39%, with a lower rate among women aged under 35 (P = 0.03) and those who had been randomized in the ARTISTIC trial to have their HPV results concealed (P = 0.01). Among women who responded there were no significant between-group differences for any of the screening results shown in Table I.

Similarly, there were no statistically significant between-group differences for any variables not related to sexual behavior collected in questionnaires (Supplementary Table I), therefore these data were pooled and analyze for all women (three groups). Age at second screening round was associated with both HPV acquisition and HPV prevalence, while marital status, ever been pregnant, menopausal status and ever used hormonal contraceptives were associated with HPV prevalence at second round (Table II). No associations remained significant after adjusting for age at second round (AORs: 0.6; 95% CI: 0.3–1.1) for being married or cohabiting 0.9 (95% CI: 0.4–1.8) for ever been pregnant, 1.4 (95% CI: 0.4–4.6) for being menopausal, and 3.9 (95% CI: 0.9–17.5) for ever use of hormonal contraceptives.

Table III shows responses to questions on sexual behavior from questionnaires for groups 1 and 3, from interviews for groups 2 and 3, and for all women (questionnaire data in groups 1 and 3, interview data in group 2). The only statistically significant difference was in the proportion of women reporting first intercourse at age 20 or later, which was higher in group 2 (short interview) than in group 3 (long interview, 29.2% vs. 19.1%; P = 0.02). The right-hand part of Table III shows percentage agreement (exact and within one category) on questionnaire and interview responses to these questions by women in group 3. There was good agreement (87.4–95.9% exact, 99.5–100% within one category) for all questions. A similar analysis for BMI and age at first smear also showed good agreement between responses in questionnaires and interviews (agreement for current BMI: 92.2% in group 2 and 90.4% in group 3 and for age at first smear: 85.3% in group 2 and 81.3% in group 3, Supplementary Table II).

Table IV shows age-specific rates for HPV prevalence at the next screening round and for HPV acquisition (HPV positive at round 2 in women who were HPV negative at entry) for all women with an HPV test at the second screening round. Table IV shows age-adjusted HPV prevalence and acquisition ORs for each sexual history question, and the multivariate ORs including age, number of earlier sexual partners and number of earlier partners. ORs for both prevalence and acquisition increased consistently with increasing number of lifetime sexual partners and lifetime regular partners. For HPV acquisition the effect of recent sexual partners (OR for 2+ recent partners: 4.4, 95% CI: 1.7–11.2) appeared to be stronger than that of earlier partners (OR for 2+ earlier partners: 2.2, 95% CI: 0.7–6.7).

Table V tabulates the number of sexual partners over lifetime and in the last 5 years according to age at first sexual intercourse and year of birth. There were marked changes in sexual behavior between women born before 1940 and those born in 1970–1979. The percentage with more than one lifetime sexual partner increased from 50% (10/20) to 94% (125/133) (P for trend <0.001) and the percentage reporting a sexual relationship before age 17 increased from 0% (0/20) to 44% (58/133; P for trend <0.001). Similar cohort trends were observed for number of sexual partners in the last 5 years.

Table VI shows the number of lifetime sexual partners by age, reported by women participating in a large National Survey of Sexual Attitudes and Lifestyles (NATSAL) conducted in 2000 [Johnson et al., 2001] and in women in the current study who were aged 17–45 in 2000. Lifetime number of sexual partners up to 2000 was estimated as the number of partners more than 5 years before the survey, as 89% of questionnaires/interviews were done between 2004 and 2006.) The distributions of the number of lifetime sexual partners in women ≥25 years old were similar in both studies (average number of sexual partners 7.1 in the current study and 6.9 in NATSAL 2000) but in women under age 25 a substantially larger proportion of respondents reported ≥5 sexual partners (63%) than in NATSAL. This presumably reflects, a tendency for women with several partners to attend for screening earlier.

DISCUSSION

Cervical cancer is the third most common cancer among women worldwide [Ferlay et al., 2010], with the highest attributable fraction ever identified for a specific cause of a major human cancer [Walboomers et al., 1999]. HPV 16 and 18 are the most frequent types associated with cervical cancer and two prophylactic vaccines against these high-risk types are available and are being used to vaccinate young women (mostly adolescents) in several countries. Most HPV infections are no longer detectable 2 years later [Plummer et al., 2007; Rodriguez et al., 2008], but it is unknown whether HPV in a woman who was HPV negative at a previous test is newly acquired or it is a recurring a latent infection. HPV testing may soon replace cytology as the primary cervical screening test, at least in older women, and the significance of transient and persistent infections is an important aspect of the natural history of HPV infection and cervical neoplasia in relation to screening policy and clinical management. Viral latency is also relevant to the effectiveness of HPV vaccination in older women, many of whom have cleared a previous HPV infection. This study describes the sexual behavior patterns in relation to HPV prevalence and HPV acquisition rates in a sample of British women attending regular cervical screening.

The multiple regression results suggest that most incident infections are newly acquired, as HPV acquisition was determined mainly by the number of new sexual partners in the last 5 years (OR for 2+: 4.4, 95% CI: 1.7–11.2). The smaller and non-significant effect of earlier partners (OR for 2+: 2.2, 95% CI: 0.7–6.7) may be attributable to residual confounding, as the partner of a woman who has had multiple partners is more likely to have had several partners himself. A weak association of lifetime number of sexual partners with HPV acquisition after adjusting for the effect of recent partners has been seen in a number of studies [Moscicki et al., 2001; Velicer et al., 2009]. Several studies have shown increased risk of new HPV infection in young women who started a new sexual relationship [Ho et al., 1998; Moscicki et al., 2001; Giuliano et al., 2002; Winer et al., 2006; Oh et al., 2008; Nielsen et al., 2009; Velicer et al., 2009]. A study in young women in the US [Moscicki et al., 2001] reported that 55% of uninfected women under age 22 years attending family planning clinics acquired an HPV infection within 3 years; the risk increasing with numbers of new partners. Data on HPV acquisition in older women are sparse, and the reason for the high HPV prevalence seen in older women in many developing countries [Herrero et al., 2000; Lazcano-Ponce et al., 2001; Molano et al., 2002; Ferreccio et al., 2004] is not known. Twenty-seven (10.0%) of the women aged over 45 in this study reported one or more new partners in the last 5 years and 11.1% (3/27) of these had a new HPV infection at the second round, compared with 3.7% (9/243) of those with no new partner.

The results of this study are limited because it was not powered to evaluate the association between sexual activity and HPV status; instead the sample size was fixed to a minimum of 200 women targeted by each method of data collection assuming a minimal response rate of 30%. However, these limited data clearly suggest that HPV is often recently acquired rather than persistent or recurrent even in older women.

Postal questionnaires are less expensive and perhaps less intrusive than personal interviews, but they may provide less reliable information on sensitive sexual behavior data. The comparison of responses on sexual behavior showed no substantial differences between British women returning postal questionnaires and those interviewed by telephone, or between questionnaire and interview responses by women who provided both. Among women aged 25 or older the distribution of sexual partners observed in this study was similar to the results of Natsal 2000 [Johnson et al., 2001], which was based on a national random sample of women interviewed at home (Table VI). Similarly, the patterns of sexual behavior reported by South Korean students returning a postal questionnaire [Shin et al., 2004] were similar to those obtained by a household interview in The National Health and Nutrition South Korea Survey [Oh et al., 2008]. A recent study conducted in the US suggests that reporting of potentially sensitive information such as very early sexual intercourse in women and late first intercourse in men is increased when telephone interviews are automated rather than conducted by an interviewer [Turner et al., 2009].

For both scientific and practical reasons it thus appears that personal interviews, at least in this field, have no advantages over more impersonal and less expensive alternatives. Postal questionnaires or automated telephone interviews are easier and less expensive to administer and code consistently than personal interviews. This is an important finding, as large samples will need to be surveyed to monitor future changes in sexual behavior among adolescents (age 12–13) who are now being vaccinated against HPV and will not begin cervical screening cervical screening (age 25) for more than a decade later.