The demographic characteristics of our study group are shown in . Participants were predominantly of minority ethnicity and of low income. Almost 15% had abnormal Pap test results, although most of these were atypical or low grade.
Demographic characteristics of 1536 women completing stress assessments divided into by tertiles of stress score.
Median score for the PSS-10 perceived stress measure was 13 (range 0–38) and for the PCL-C was 24 (range 17–85), suggesting moderate stress. As shown in , stress as measured by PSS was associated with older age, lower income, less education, smoking, and drug use. PSS and PCL-C were correlated (P < 0.001 by test for nonzero correlation). Of the 1536 women with completed questionnaires, PCL-C scores were >50, indicating post-traumatic stress, for 149 (10%). Median CES-D score was 8 (range 0–57), suggesting minimal prevalence of depression. CES-D score was also correlated with PSS (P < 0.001).
In univariate analysis, SIL was associated with known cervical cancer risk factors, including HIV seropositivity (O.R. 12.62, 95% C.I. 3.96–40.27, P < 0.001, compared to seronegative women), ethnicity (O.R. 0.38, 95% C.I. 0.19–0.79, P = 0.004 for white and O.R. 0.46, 95% C.I. 0.23–0.95, P = 0.026, for other ethnicity compared to African-American), current employment (O.R. 0.28, 95% C.I. 0.15–0.51, P < 0.001 for compared to unemployed women), and current smoking (O.R. 3.24, 95% C.I. 1.71–6.15, P < 0.001, compared to never smokers).
The proportion of women with SIL was not significantly different across stress levels. For PSS, compared to women in the lowest tertile of reported stress, O.R. for SIL was 0.88 (95% C.I. 0.51–.53, P = 0.655) for women in the middle stress tertile and 0.96 (95% C.I. 0.55–1.68, P = 0.886) for women in the highest stress tertile. For PCL-C, compared to women in the lowest tertile of reported PTSD symptoms, O.R. for SIL was 0.78 (95% C.I. 0.44–1.40, P = .415) for women in the middle tertile and 1.17 (95% C.I. 0.69–2.00, P = .560) for women in the highest stress tertile.
Depressive symptoms also were not associated with SIL. SIL rates were similar for CES-D scores above 16 (O.R. 1.41, 95% C.I. 0.88–2.26, P = 0.152) and above 23 (O.R. 1.39, 95% C.I. 0.81–2.40, P = 0.244).
Because prior treatment might have masked significant associations, we repeated analyses to look for a possible correlation between SIL at any time during WIHS and either PSS or PCL-C at the index visit; again no association was found (not shown). Similarly, repeating subset analyses in only the HIV seropositive group did not reveal an association between stress and SIL (not shown). Because HIV-related immunosuppression may have dominated stress as a determinant of SIL, we repeated analyses, limiting assessment only to those women with HIV and CD4 lymphocyte counts >500 and to women without HIV; no association between SIL and PSS score, PCL-C, or CES-D was found. Finally, we found no associations between SIL and combinations of PCL-C and CES-D scores or potentially stressful events including trauma history and self-reported sexual abuse (not shown).
Results of multivariable analysis are shown in . Odds of SIL were elevated among women with HIV infection and were linked to degree of immunosuppression, race, and current smoking. PSS, PCL-C score >50, and CES-D score >15 were not significantly associated with SIL.
Multivariable analysis of associations of demographic and biological risk factors for squamous intraepithelial lesions on Pap testing.