Study population
This cross-sectional study is based on the CONSTANCES cohort of volunteers aged 18–69 years in 22 selected health screening centers from the principal regions of France (link to the website). Participants are randomly selected from adults covered by the National Health Insurance Fund (i.e., salaried workers, currently working or retired, and their family members), stratified by age, sex, region, and socioeconomic position. CONSTANCES collects data on personal, environmental, behavioral, occupational, and social factors from questionnaires self-administered at inclusion and annually thereafter, mailed to and returned by participants (or collected in the health centers). This general-purpose epidemiological cohort is designed to study a wide range of health problems in various disciplines in the general population. Its long-term objective is to follow 200,000 members (men and women) of the French population, aged 18 to 69 years; inclusion in this cohort began in 2012 [13]. After enrolment, participants are followed up by an annual self-administered questionnaire sent to their homes (paper or web-based), and they are invited for a new health examination every 5 years. The data considered here were collected at inclusion and come from the questionnaires about lifestyle, women's health, and occupational exposures. The data were collected from 2012 to 2019. Body mass index (BMI) was obtained from participants' weight and height measurements, collected at the initial medical examination.
This analysis covers the women aged younger than 25 years recruited between January 2012 and April 2019. It excluded the following women: those who a priori did not require contraception: those who reported an history of hysterectomy or bilateral oophorectomy; those who reported never having had sex; and those who reported an history of cervical cancer.
Outcome measurements
The variable to be explained was premature screening (having had a Pap smear before the age of 25 years). Women younger than 25 years who had answered "Yes" to the question "Have you ever had one or more Pap tests (smears taken from the cervix)?" were considered to have undergone premature screening.
Contraception and sex life
For contraception, the principal explanatory variable, we distinguished the different types of contraceptives according to their degree of medicalization (does or does not require prescription and follow-up by a healthcare professional) and then in more detail, into 5 contraceptive choices:
- Medical-contraceptives: Combined estrogen-progestin or progestin-only contraceptives, regardless of their form (oral, patch, ring, injection), which require at least one medical consultation annually for their prescription. Because this contraceptive was the most frequently used, it was the reference category. Contraception by IUD: a device must be placed by a physician (or midwife) and changed every 3 to 10 years. Women who use it are advised to have annual clinical examinations. Contraception by implants, which must be placed by a physician (or midwife) and changed every 3 years, but for which annual consultations are not required (only a check-up visit 3 months after its placement is recommended).
- Non-medicalized contraception: These are the contraceptives that require no medical visit (condom, withdrawal, spermicides, and rhythm or Knaus-Ogino methods).
- Absence of contraception: This category included the women who reported sexual relations but not contraceptive use, although they did not want to become pregnant.
Women answered the following questions about their sex lives and reproduction: sex of partners (male/female/both/do not wish to answer [DNWA]), number of lifetime partners (number/DNWA), new partner in the past 12 months (yes/no/DNWA), pain during intercourse (never/sometimes /often/always), sexual satisfaction (currently your sex life seems: not at all satisfactory/not very satisfactory/satisfactory/very satisfactory /DNWA/not applicable). The responses to this question were summarized as satisfactory and unsatisfactory, with women who answered "very satisfactory or satisfactory" classified as satisfied. The women considered to have pain during intercourse (dyspareunia) were those who answered "often" or "always".
Social and demographic characteristics
The other indicators considered were age, parity, civil status, and geographic origin, defined according to place of birth. Educational level was defined by the highest diploma completed: less than the baccalaureate or school-leaving exam ("bac"), passed the "bac", some post-secondary education, other diplomas.
Health status
A specific question allowed respondents to classify their health status as good, medium, or poor.
The categories for smoking were: current smoker, ex-smoker, non-smoker; for alcohol consumption: irregular consumption (less than 4 times a month), regular consumption (one to several times a week), and not currently; for marijuana use the question was “In your life, have you ever use marijuana?”. The possible answers were yes/no/DNWA. The weight and height of each participant were measured at the medical examination at the health center and enabled calculation of her BMI. This variable was introduced in categories according to the WHO classification (< 18.5 kg/m2 underweight, 18.5-24.9 kg/m2 normal weight, 25.0-30 kg/m2 overweight, 30.0-39.9 kg/m2 obese, > 40 kg/m2 morbidly obese).
Statistical analyses
Quantitative variables (age, BMI) were described by their means and standard deviations, and the qualitative variables as percentages. The quantitative variables were then discretized into categories. To assess the association between premature CCS and the categorical variables, we performed Chi2 or Fisher's exact tests of independence.
To understand the role of each variable, we first studied the associations between the explanatory variables and premature CCS in a univariate analysis. Variables were retained when they were associated with premature CCS with a P value < 0.05. They were then included in 3 separate thematic logistic regressions (contraception and sex life, social and demographic characteristics, and health status). These models were simplified by backward elimination. A final model including the associated variables for each thematic model (at P< 0.05) also underwent the backward elimination procedure. The associations between premature CCS and the variables of interest were expressed by adjusted odds ratios, and their 95% confidence intervals.
Missing data were handled by using multiple imputations with fully conditional specification (SAS 2013) and assuming missingness at random (MAR). To make the MAR assumption more plausible, every previously described variable was used for the imputation model [14, 15], including the outcome. Excluding the outcome from the imputation model could have hidden some associations, and including it did not change the standard deviations [16]. Ten complete datasets were created. This method, known as MID (multiple imputations, then deletion), uses information about the dependent variable in the imputation model (as well as the standard imputation method), but cases with imputed outcomes are deleted before analyses [17]. Overall, variable had 10% or more missing data: sex life satisfactory (13.06%), number of life time partners (44.2%), socio-professional categories (14.64%), practitioner who performed the Pap test (52.04%). The rate of missing data was 4.6% for the Pap test status.
The analyses were performed with SAS software, version 9.3.
Ethics
The national council on statistical information (CNIL) approved the CONSTANCES study (CNIL authorization n°910486). An additional related application to the CNIL was approved on January 25, 2016 (CNIL authorization n°1881675).