Cervical cancer prevention depends on screening cytology, triage by colposcopically directed biopsy, and eradication of cancer precursors. Although effective in decreasing cervical cancer rates,1–3 this strategy requires multiple visits. Noncompliance at any stage might result in missed opportunities for cancer prevention. Many studies have explored risk factors for noncompliance with cytology screening4,5 and colposcopy referral.6–9 Compliance with follow-up and treatment recommendations after colposcopy, although no less important, has been less intensively studied.10–12 We set out to determine whether correlates of compliance with follow-up after colposcopy can be predicted by applying multivariate analysis to data collected prospectively from a cohort of 451 women who had colposcopies.
Materials and Methods
Consecutive women who presented for their first colposcopies to the Resident Colposcopy Clinic at Rush-Presbyterian-St. Luke's Medical Center, Chicago, Illinois, between October 1, 1992, and June 30, 1997, were enrolled in a program approved by the local institutional review board. Medicaid enrollment was a requirement for clinic entry.
After recording standardized medical and gynecologic histories, fourth-year residents from the Department of Obstetrics and Gynecology did colposcopies under the direction of a single gynecologic oncologist (LSM). Directed biopsies were taken of the most advanced lesions. Findings were recorded on standard forms. Referral cytology results were categorized according to the Bethesda System for cervicovaginal cytologic diagnosis.13
Biopsies were categorized as benign, atypia including koilocytosis and changes consistent with infection with human papillomavirus (HPV), cervical intraepithelial neoplasia (CIN, graded 1–3), or cancer. Because biopsy results were not an outcome variable, review for purposes of study was not done. In analysis, women without colposcopic lesions were categorized with those who had benign findings.
Individualized follow-up recommendations were made based on cytologic, histologic, and colposcopic findings before counseling visits scheduled 2–3 weeks after colposcopy. In general, immediate outpatient loop excision was recommended for women with CIN 2 or 3, and offered to those with CIN 1 if they also had satisfactory colposcopies and negative endocervical curettages. Women with unsatisfactory colposcopies and smears suggestive of at least low-grade squamous intraepithelial lesions (SIL), those with dysplasia in endocervical curettings, and selected others were advised to have conization, usually scheduled in the outpatient surgical area within 1 month. Pregnant women were advised to have repeat colposcopies postpartum. For women with discrepancies between cytology and biopsy results, conization or repeat colposcopy to resolve discrepancies was advised. Women with negative colposcopies or nonneoplastic biopsy results were asked to return for surveillance cytologic screening in 6 months. After discussion of the small but finite malignant potential of their lesions, women with CIN 2–3 who refused therapy and all women with CIN 1 also were offered observation with serial cytology. To avoid selection bias in analysis of demographics and other variables, women who did not return for counseling were included and analyzed according to predetermined recommendations.
Charts were reviewed retrospectively 8–12 months after colposcopy to determine compliance. Data were entered prospectively in an Excel 4.0 spreadsheet (Microsoft, Inc., Redmond, WA) and tabulated. Univariate tests were carried out using Pearson χ2, Fisher exact test, Kruskal-Wallis test, and Mann-Whitney U test. Multivariate analysis was done using backward stepwise logistic regression. Likelihood ratio tests were used to determine whether to keep predictors. Statistical significance was P ≤ .05, and factors that achieved that level were included in multivariate models. Assessment of correlation was determined using Spearman rho.
The mean age of women who had colposcopy was 26.0 years (median 22, range 14–90 years). Of the women studied, 357 (79%) were black, 49 (11%) were white, 42 (9%) were Hispanic, and 3 (1%) were other or not recorded. Cigarette smoking was reported by 124 (27%) women. Smokers reported a mean of 7.6 pack-years (median 4, range 0.05–60 pack-years).
Gynecologic characteristics reported by women in the study are shown in Table 1. The reported mean age at first intercourse was 15.7 years (median 16, range 8–22 years). Most women reported multiple lifetime sexual partners (mean 5.2, median 3, range 0–100 partners). The mean age at first conception was 17.9 years (median 17, range 11–29 years). At the time of colposcopy, 104 women (23%) were pregnant.
Table 2 shows the referral cytology and colposcopic biopsy results. Women with normal cytology results were referred for visible cervical changes. Only 14% of cytology results were high grade, and only 14% of biopsies showed CIN 2–3 or cancer. Satisfactory view of the entire squamocolumnar junction was achieved in 355 women (79%) but not in 93 (21%) and was not recorded in 3 (1%).
Study distribution of recommendations for management after colposcopy is shown in Table 3. As anticipated under clinic guidelines, the majority of women directed toward serial cytology had negative findings or CIN 1, and most women who were advised to have treatment had CIN 2–3 or cancer.
One hundred eighty women (40%) returned for their first appointments after colposcopy or complied immediately with recommendations, whereas 63 (14%) complied only after missing initial appointments, and 205 (46%) never complied. Rates of ever complying with recommendations were 92 of 114 (81%) for loop excision, 23 of 27 (85%) for conization, 31 of 50 (62%) for repeat colposcopy, and 91 of 250 (36%) for repeat cytology. Compliance rates according to biopsy grade were 51 of 127 (40%) for women with negative colposcopic biopsies, 12 of 35 (34%) for those with atypical biopsies, 74 of 114 (65%) for those with CIN 1, 52 of 61 (85%) for those with CIN 2 or CIN 3, and 100% for the one woman with cancer.
Associations of categoric variables with ever complying with recommendations are shown in Table 4. In univariate analysis, significant predictors included biopsy grade and recommendation (P < .001 for both) and a history of genital herpes simplex virus (HSV) infection (P = .03). Referral cytology results approached significance (P = .07) and were entered in multivariate analysis. Continuous variables, including age, gravidity, parity, age at first intercourse, age at first conception, lifetime number of sexual partners, and pack-year smoking history were not associated with ever complying with follow-up recommendations. In logistic regression analysis, the only factors that remained associated with ever complying were history of HSV and nature of management recommendations (Table 5).
We also investigated predictors of compliance at first scheduled appointment. As with ever complying, the only factors that remained significantly associated with on-time compliance in logistic regression analysis were history of HSV (OR 0.10; 95% CI 0.01,0.87, P = .007) and type of management recommendation (compared with serial cytology, for loop excision OR 4.89; 95% CI 2.68, 8.93; for conization OR 4.16; 95% CI 1.54,11.2; for repeat colposcopy OR 2.39; 95% CI 1.24, 4.62, P < .001).
Recommendations were derived from integration of biopsy results, cytology results, colposcopic findings, and discussions of the risks and benefits of alternatives with individuals, so we explored the correlations between recommendation, biopsy, and cytology. The correlation between cytology and biopsy was 0.25 (n = 273), between cytology and recommendation 0.41 (n = 332), and between biopsy and recommendation 0.49 (n = 282). Correlations were significant (P < .01 for all) but at most moderate, so changes in one factor explain less than 25% of the variability of either of the others.
The poor compliance rate for women advised to return for repeat cytology in 6 months led us to investigate whether specific factors might predict compliance within that group. In logistic regression analysis, the only factors associated with compliance in that group were pregnancy at the time of evaluation (OR 0.62; 95% CI 0.39, 0.84) and history of HSV (OR 0.11; 95% CI 0.01, 0.84).
In the present study, the strongest predictor of compliance with recommended management after colposcopy was the nature of the follow-up advised. The 36% compliance rate among women advised to return in 6 months for surveillance Papanicolaou smears was significantly lower than rates for women advised to have immediate treatment with loop excision (81%) or cold-knife conization (85%), or those advised to have repeat colposcopies (62%).
Although unique in applying logistic regression analysis to colposcopy compliance, our results are consistent with reports by others. McCarthy and colleagues showed that compliance after mammography was lower when surveillance testing at 6 months rather than immediate intervention was recommended.14 Gold and colleagues, and Eger and Peipert, also working with urban, heavily minority populations, reported that management recommendations were significant correlates of compliance after colposcopy, although they found that biopsy grades were determinants also.10,11 In the former study, compliance with recommendations of repeat cytology at 6 months was 43%, similar to our 36% rate, and compliance with recommendations of early repeat colposcopy or excision was nearly as high as our reported 74% rate. However, neither of those studies controlled for potential confounding by multivariate logistic regression to estimate ORs for effects after adjustment for other predictors. The nature of management recommendation is a function of biopsy grade and other subject factors, so it was not surprising that recommendation remained significant in regression analysis whereas biopsy grade did not. As in our study, and in those of Gold et al and Eger and Peipert, Laedtke and Dignan found that compliance after colposcopy could not be predicted from demographic characteristics.12
Those results suggest that when noncompliance after colposcopy is a substantial problem, treatment of cervical cancer precursors should be favored over surveillance when both are valid options. Surveillance is appropriate mainly for women with CIN 1 because treatment cannot be justified without documented disease, and the risk of progression is too high to consider observation for most women with higher grades of CIN.15 The majority of women with CIN 1 have transient atypical changes associated with infection by HPV 16,17 Most of those lesions resolve spontaneously, so surveillance with serial cervical cytology would be optimal management, with ablation or excision reserved for those with disease progression or persistence. However, CIN 1 has a low but definite risk of progression to invasive cancer.15 At centers with high rates of noncompliance, triaging women with CIN 1 to observation might allow development of cervical cancers that would have been prevented by treatment.
We also tried to identify factors that distinguish compliant from noncompliant women who were recommended to have serial cytology. Despite the relatively large population directed to cytologic surveillance, we were unable to identify such factors, with the exception of history of HSV and pregnancy, the former present in a minority of women and the latter a contra-indication to treatment of CIN 1. Although only 44 women had CIN 1, in multivariate analysis, women with CIN 1 were not statistically significantly more likely than women with atypia or negative colposcopic findings to comply with follow-up. Our failure to identify discerning factors and our finding that women who were advised to have loop excisions were more than 6.8 times more likely to comply with recommendations suggested that excision might be appropriate for women with CIN 1 in clinics with similar populations.
Older women with CIN appear less likely to clear their disease without intervention, perhaps because older women are less likely to have recently acquired HPV infections at the time of detection.18 Our results showed that older women were not more likely to comply with follow-up recommendations, so clinicians caring for populations at high risk of noncompliance, but unwilling to treat all women with CIN 1, might consider limiting intervention to older women.
Several authors have outlined a strategy for managing abnormal cytology that incorporates colposcopy and loop excision in the same visit.19,20 That strategy was criticized because false-positive colposcopic examinations might lead to unnecessary excisions, which could be avoided by awaiting the results of colposcopic biopsy or observation for spontaneous regression. In one study, 40% of women with smears read as low-grade SIL had no CIN identified in loop-excision specimens despite colposcopic abnormalities.20 Among women at high risk for noncompliance, that objection might be outweighed by the risk of failing to treat disease. Noncompliance with a recommendation of loop excision was 19% in the present study. The majority of those women had CIN 2–3, so even that rate of noncompliance is of concern and suggests that in practices similar to the one studied a program that incorporates see-and-treat strategies for women with cytologic abnormalities and clear colposcopic abnormalities might result in better compliance and lower risk of progression to cancer than strategies that require women to return for treatment after results of colposcopic biopsies are available.
Our study also identified history of HSV as a risk factor for noncompliance after colposcopy. That might be an artifact of analysis, but because of the small number of women with that history, it also might be a proxy for behavioral characteristics that predispose to noncompliance, which deserves further study.
Our study had at least three limitations. Subjects were predominantly minority women in an urban setting, and although urban minority women face an increased risk of cervical cancer and a greater likelihood of noncompliance,21 results might not be generalizable to practices in which compliance is high regardless of recommendation. For most women, cytologic surveillance for CIN 1 and deferral of treatment pending results of colposcopic biopsies remain preferred management. It is not possible to use our data to differentiate between the effect of follow-up interval, follow-up intervention, and surgical versus nonsurgical intervention, all of which are integrated in recommendations and potentially affect a woman's view of her illness. We did not investigate the effect more intensive management might have on compliance, but such studies should be done.
1. Sigurdsson K. Quality assurance in cervical cancer screening: The Icelandic experience 1964–1993. Eur J Cancer 1995;1:728–34.
2. Adami HO, Ponten J, Sparen P, Bergstrom R, Gustafsson L, Friberg LG. Survival trend after invasive cervical cancer diagnosis in Sweden before and after cytologic screening. Cancer 1994;73:140–7.
3. Benedet JL, Anderson GH, Matisic JP. A comprehensive program for cervical cancer detection and management. Am J Obstet Gynecol 1992;166:1254–9.
4. Peters RK, Bear MB, Thomas D. Barriers to screening for cancer of the cervix. Prev Med 1989;18:133–46.
5. Beckmann CA, Beckmann CRB, Lipscomb GH, King RE, Steere E. Pap smear screening: Determinants of patient compliance. J Womens Health 1994;4:663–8.
6. Michielutte R, Diseker RA, Young LD, May WJ. Noncompliance in screening follow-up among family planning clinic patients with cervical dysplasia. Prev Med 1985;14:248–58.
7. Lacey L, Whitfield J, DeWhite W, Ansell D, Whitman S, Chen E, et al. Referral adherence in an inner city breast and cervical cancer screening program. Cancer 1993;72:950–5.
8. Funke BL, Nicholson ME. Factors affecting patient compliance among women with abnormal Pap smears Patient Educ. Couns 1993:20:5–15
9. Kaplan CP, Bastani R, Marcus A, Breslow L, Nasseri K, Chen L Low-income women with cervical abnormalities: Individual and system factors affecting follow-up. J Womens Health 1995;4:179–88.
10. Gold MA Dunton CJ Macones GA, Dixon L.H, Carlson JA Knowledge base as a predictor of follow up compliance after colposcopy. J Lower Genital Tract Dis 1997;1:132–6.
11. Eger RR. Peipert JF Risk factors for noncompliance in a colposcopy clinic J Reprod Med 1996;41:671–4
12. Laedtke TW, Dignan M Compliance with therapy for cervical dysplasia among women of low socioeconomic status. South Med J 1992;85:5–8
13. Broder S. The Bethesda System for reporting cervical/vaginal cytologic diagnosis Report of the 1991 Bethesda workshop. JAMA 1992;267:1892
14. McCarthy BD, Yood MU. Boohaker EA, Ward RE, Rebner M, Johnson CC. Inadequate follow-up of abnormal mammograms. Am J Prev Med 1996;12:282–8.
15. Ostor AG. Natural history of cervical intraepithelial neoplasia: A critical review. Int J Gynecol Pathol 1993;12:186–92
16. Kiviat N Natural history of cervical neoplasia: Overview and update. Am J Obstet Gynecol 1996;175:1099–104.
17. Schiffman MH. Recent progress in defining the epidemiology of human papillomavirus infection and cervical neoplasia. J Natl Cancer Inst 1992;84:394–8.
18. van Oortmarssen GJ, Habbema JDF Epidemiological evidence for age-dependent regression of pre-invasive cervical cancer. Br J Cancer 1991;64:559–65.
19. Bigrigg MA, Codling BW, Pearson P, Read MD, Swingler GR. Colposcopic diagnosis and treatment of cervical dysplasia at a single clinic visit. Experience of low-voltage diathermy loop in 1000 patients. Lancet 1990;336:229–31.
20. Ferris DG, Hainer BL, Pfenninger JL, Zuber TJ. 'See and treat' electrosurgical loop excision of the cervical transformation zone. J Fam Pract 1996;42:253–7.
21. Parker SP, Davis KJ, Wingo PA, Ries LA, Heath CW Jr. Cancer statistics by race and ethnicity. CA Cancer J Clin 1998;48:31–48.