A cross-sectional study was performed using an epidemiological database from an established phase II clinical trial that aimed to evaluate emerging technologies for diagnosing cervical neoplasia (15–17). IRB consent was obtained from subjects for their participation in the index trial, and prior (July 31, 2011 to present) approval for research in this secondary analysis was granted from the Baylor College of Medicine IRB (H-29389). Trial participants included women with abnormal cervical cytology presenting for colposcopy at one of three participating hospitals in the US and Canada between 1999 and 2005 (diagnostic group) as well as community volunteers with no prior history of dysplasia or cervical treatments (screening group). Cytology and histology tests were performed on all patients, as well as low and high-risk HPV types via Hybrid Capture II and PCR. Tissue biopsy was also performed and interpreted by skilled pathologists. All 1,858 subjects in the database were initially included. Cases with unknown dysplasia status, those who reported never having sexual intercourse and those with missing data regarding date of last Pap test or history of birth control use were excluded.

Sociodemographic variables in these analyses included age, race, nativity, marital status, household income, education level and employment status (18–20). Behavioral risk factors included age at first sex, number of sex partners (during past 12 months and lifetime), number of vaginal deliveries, history of oral contraceptive use, current binge drinking and current smoking (21–24). Number of vaginal deliveries was used instead of parity due to limitations of ascertainment bias and for having better accuracy. Current binge drinking was defined as consuming five or more alcoholic beverages in a single occasion more than one time in the past month.

Upon entering the study, subjects filled out a questionnaire regarding lifetime history of birth control use. Listed methods included oral contraceptives, intrauterine device, diaphragm, cream, foam/jelly/sponge, condom and surgical sterilization (25, 26). Women who used other methods were asked to specify the method and the number of years used. In this study, women with a history of surgical sterilization were defined as those who reported history of bilateral tubal ligation. Although women with no history of surgical sterilization were those who did not report receiving a bilateral tubal ligation, they did not also miss any data regarding birth control use.

Cervical cytology screening adherence was self-reported by participants on a questionnaire inquiring about date of last Pap test. Time since last cervical cytology screening was calculated as the difference between the date of the last test and the date of study enrollment. Participants were considered to be non-adherent with screening guidelines if the time duration since their cervical cytology-screening test was more than or equal to 15 months. Based on the compilation of different guidelines from the American College of Obstetrics and Gynecology (ACOG), American Cancer Society (ACS), and US Preventative Services Task Force, it appeared that most practitioners at the time of data collection had performed annual Pap testing except “low-risk“ patients with 3 consecutive negative tests (5, 25, 27, 28). Current guidelines suggest that low-risk women older than 30 years of age only need Pap screening every three years (27, 29), however, many studies have shown that the majority of practitioners in the US continue to perform annual cervical cytology screening anyway (30–32). Thus, in this study, the definition of adherence was based on the common practice of annual Pap testing.

Cervical dysplasia status was based on cytology and histology data. In cases where the diagnoses were discordant, the worst diagnosis recorded was used. For descriptive statistics, cervical dysplasia was categorized as normal (negative intraepithelial lesion, atypia, HPV-associated changes), low-grade (CIN-1, CIN-2), and high-grade (CIN-3, carcinoma in situ, invasive cancer) (33). In logistic regression analyses regarding factors associated with dysplasia, cervical dysplasia was dichotomized as normal versus other types (low- or high-grade).

Statistical analysis was performed using Stata 12.0 software package (StataCorp, LP, College Station, TX). Descriptive statistics and chi-squared test were used to evaluate sociodemo-graphic and behavioral risk factor differences between sterilized and non-sterilized women in both the screening and diagnostic groups. Analyses regarding surgical sterilization status and cervical cytology screening adherence were performed among screening group participants because the date of their last Pap test was independent of their enrollment in the trial. Analyses regarding surgical sterilization status and cervical dysplasia were performed among participants in the diagnostic group, in which the prevalence of cervical dysplasia was sufficiently high to evaluate the association. These analyses were further limited to US patients given the discrepancies between the US and Canadian guidelines of clinical management of cervical dysplasia (25, 34).

For both analyses, univariable logistic regression was first used to assess the crude association between surgical sterilization status and the outcome (Pap test non-adherence or cervical dysplasia). Next, multivariable logistic regression models were applied to evaluate the association between surgical sterilization status and the respective outcome after adjusting for potential confounders. Confounders were defined as variables significantly associated with both sterilization status and the respective outcome (p≤0.10) and that changed the odds ratio by 10% or more. A p-value ≤0.5 was considered statistically significant.

In Table 3, variables that made surgically sterilized subjects in the screening group more likely to be non-adherent to Pap test guidelines were further explored. Overall, non-adherent women were 39% more likely to be surgically sterilized than those who were adherent (p≤0.05). Women who were non-adherent were also significantly more likely to be divorced, separated or widowed (OR=1.62), Hispanic (OR=1.57) and to have had four or more vaginal deliveries (OR=2.00). They were also less likely to have a household income above $40,000 per year (OR=0.48) or education level above high school (OR=0.62). It was found that education was an effect measure modifier of the association between surgical sterilization and screening non-adherence and therefore stratified by education level in subsequent analyses. In univariable analyses, surgical sterilization was significantly associated with non-adherence among women with a high school education or less (OR=1.60) but not among those with a higher level of education. The association between surgical sterilization and non-adherence remained significant after adjusting for age, marital status, and number of vaginal deliveries (AOR=1.47).

In Table 4, factors that made surgically sterilized subjects in the diagnostic group more likely to have dysplasia were evaluated. Overall, there was no association between cervical dysplasia and surgical sterilization status; however, stratified analyses indicated that among women older than 40 years, surgical sterilization was associated with an 80% increased chance of dysplasia. This association was not statistically significant, likely due to the smaller sample size of women in these strata compared to the overall study size. The magnitude of the association did not change after adjusting for variables significantly associated with cervical dysplasia and surgical sterilization (race, marital status, education, and age of sexual debut).

This study explored whether surgically sterilized women with certain sociodemo-graphic risk factors were more likely to be non-adherent with cervical cytology and dysplasia screening. The data suggests that women with surgical sterilization are more likely to be non adherent with screening guidelines, even after controlling for age, marital status, education and number of vaginal deliveries, which were found to be significantly associated with screening non adherence. Among women over 40 years of age, the data suggest that surgical sterilization may be associated with increased prevalence of cervical dysplasia.

Although cervical cancer is presently attributed to be a disease of the developing world, evidence also suggests that increased cervical cancer morbidity and mortality in the US exists for those with decreased access to care or residing in low-resource settings. This is notably true in areas of the US with large immigrant and underserved populations who also utilize surgical sterilization as a prevalent means of contraception (9, 18, 19).

There were both inherent strengths and limitations to the study. A major strength of the study is its large and diverse sample size. Despite applying a convenience sampling, the large screening group was more likely indicative of the general US population than those populations examined in prior studies (10, 12–14). Data on cervical dysplasia can be considered reliable and accurate as it was based on a consensus diagnosis of cytological and histological evaluation, thus minimizing any potential misclassification resulting from using cytology alone. Moreover, this study identified education level less than or equal to a high school degree as an effect measure modifier in tubal ligation and adherence with screening. This has independent value for practitioners and will serve as an immediate potential avenue for awareness and outreach in clinical settings.

Despite these strengths, the study was limited by the fact that the diagnostic group was a hospital-based sample while the screening group was a convenience sample. While inferences can be drawn from these data, a true cause and effect relationship between non-adherence with Pap tests and increased cervical dysplasia in a single population cannot be shown. In addition, as the study's database only included history of number of vaginal deliveries and had no information about the number of Cesarean deliveries, this limited the ability to evaluate the association of true parity and the outcomes (although it can be well approximated given the overall low rate or Primary Cesarean in the cohort). Since complete medical chart of each patient was not available, date of last Pap test and surgical sterilization status were based on self-report and thus subject to recall bias. Prior studies have suggested that there may be a temporary protective effect of surgical sterilization on cervical cancer risk due to pre-operative screening routines (11, 12). Unfortunately, historical dates of surgical sterilization were not available and therefore it was not possible to draw any temporal relationships between sterilization and date of last Pap test or date of dysplastic lesions. Finally, the definition of non-compliance was not based on today's ACOG guidelines for less frequent screening as the study chose to evaluate the data based on guidelines current at the time the study was conducted. (27, 28). However, the link between surgical sterilization and adherence with Pap testing was evaluated every 3 years and findings were similarly striking for decreased adherence between both screening regimens.