This was a descriptive and analytical cross-sectional study with data collection conducted over a period of one week from October 1 to October 8, 2018. The study population included women aged 18 years or older on the day of the survey and residing in the Municipality of Abomey-Calavi. In the study, consenting women who were mentally competent, 18 years of age or older at the time of the survey, and residing continuously in the Municipality of Abomey-Calavi for the last six months prior to the survey were included. On the other hand, women who belonged to a breast cancer prevention service, women in whom secondary screening was noted, or non-consenting women were not included.

The minimum sample size was estimated with the Schwartz formula at 1729 participants:

n = K ∗ Z 2 P ( 1 − P ) / i 2

with n, the sample size; K, the cluster effect; Z, the reduced interval corresponding to the 5% risk; Z = 1.96; p, prevalence of breast cancer in Tunisia 2009 (p = 10%) and i: the desired precision (i = 2%). We used a four-stage WHO random cluster sampling.

The dependent variable of the study was breast cancer early screening. It was a composite variable that included the following conditions: having performed breast self-examination in the past 12 months or having been seen for breast cancer early screening in a qualified health care facility in the past 12 months or having had a mammogram in the past 12 months. The independent variables were sociodemographic characteristics, family history, and variables related to knowledge about early breast cancer screening. The socio-economic level was evaluated using the quartile method on the basis of a set of weighted items. The abusive consumption of alcohol was defined as a daily consumption of more than four glasses of 20 g ethanol equivalent per unit. The practice of physical activity was insufficient for moderate physical activity of at least 30 minutes or intense activity of at least 15 minutes during their activities or less than 5 days per week.

Regarding data collection, a questionnaire was developed and then pretested and validated prior to the multiplication of the collection forms. The questionnaire was administered during an individual face-to-face interview. The data collected was checked at the end of the day to ensure the quality of the information collected. The data collected were entered using Epidata 3.1. Fr. They were then cleaned and analyzed with R Studio 3.5.1. The qualitative variables were expressed as proportions and the quantitative variables as mean ± standard deviation after verification of their normality by the Shapiro-wilk test. The comparison of proportions was done with the chi-square test or Fisher’s exact test, depending on whether the theoretical number of participants was less than 5 or not. Univariate and multivariate analyses were performed to identify factors associated with early breast cancer screening. For all comparisons, a difference at the p-value threshold < 5% was considered statistically significant. The initial model for the multivariate analysis included variables associated in univariate, variables not associated in univariate but at the p-value threshold of less than 20%, and variables known in the literature to be associated with breast cancer early screening (forced variables) but that did not meet either of the previous conditions. All these variables were entered into an initial top-down binary stepwise logistic regression model. Interactions were checked. The Akaike Information Criterion (AIC) was used to select the model that best fit the data. A significance level at the p-value < 5% was applied to retain the associated factors after adjustment for the other variables in the final multivariate analysis model.

This study was based on declarative information about breast cancer in women. However, free and informed consent in written form with the respondents’ signature was systematically requested by the investigators before the questionnaire was administered. The data collected during the survey were strictly confidential and treated anonymously. In addition, this study received the agreement of the Abomey-Calavi communal authority under n˚01024/C-AC/DC/SG/DRH/SAC before the data collection.

The mean age of the women surveyed was 32.0 ± 11.5 years with a range of 18 and 71 years. Women between the ages of 25 and 34 represented 33.97% of the sample. Table 1 presents the socio-demographic and economic characteristics of the respondents.

Table 2 presents the gyneco-obstetrical history of the women surveyed.

The average age of onset of menarche was 14.8 ± 1.7 years with extreme values of 9 and 18 years for all the total of 1740 women included in this study. In addition, 268 (15.40%) reported having had their first sexual intercourse before the age of 15 and 1258 (72.30%) reported having become pregnant at least once during their lifetime. Alcohol consumption was reported by 145 women (8.33%) and the majority did not practice enough physical activity (82.24%) (Table 3).

Regarding their eating habits, 90.86% did not eat enough fruit and vegetables, 75.75% ate fast food, 41.15% said they liked to eat too much sugar, 27.13% too much fat and more than one in four liked to eat too much salt (Figure 1).

The majority of respondents knew about breast cancer through several information channels (Figure 2).

a: Primipare at least.

The proportion of women who said they had heard about breast cancer once was 75.17%, meaning 1308 of all women surveyed. According to the majority of them (68.50%), breast cancer is mainly manifested by the presence of a nodule (Figure 3).

Of the 1308 patients who had heard of breast cancer once, 726 (55.50%) said they had heard of breast cancer screening at least once (Figure 4).

The breast cancer is detectable early according to 512 meaning 70.52% of them (Figure 5).

The proportion of women who had ever been screened for breast cancer in the overall sample was 12.93%, meaning 225 women out of the 1740 surveyed. In addition, 203 (11.67%) reported that this screening was on their own initiative.

About the main means of early breast cancer screening used:

Our study reported that breast self-examination was the most commonly used method of early cancer screening in the last 12 months (85.78%) (Figure 6).

This technique was adopted by 193 women as the main means of breast cancer screening. Of these, 50 (25.91%) reported having visited a health facility for confirmatory examination. Table 4 presents information on breast self-examination.

Mammography is also a means of screening. According to the declarations of the 05 respondents who had undergone mammography in the last 12 months and of whom 03 suspected cases were confirmed and already treated.

After adjustment, factors associated with early breast cancer screening were age ≤ 50, education level (increasingly higher), socioeconomic level (average/high), residence setting (downtown), and marital status (married/coupled). Indeed, women aged 50 years or older were 2.40 times (AOR CI: 1.10 - 6.35; p = 0.0015) more susceptible to breast cancer early screening. Similarly, this susceptibility was 1.67 (AOR CI: 1.04 - 2.68), 3.46 (AOR CI: 2.29 - 5.30) and 9.90 (AOR CI: 5.78 - 17.15) times among women with primary, secondary and tertiary education, respectively, compared to those with no education (p = 0.0001). Also, women in couples or married women were 1.76 times (AOR CI: 1.27 - 2.46; p = 0.0001) more likely to be screened compared to those not in couples.

Furthermore, susceptibility to early breast cancer screening increased significantly with socioeconomic level (p = 0.0015). Thus, susceptibility ranged from 1.53 times (AOR CI: 1.02 - 2.33) for the middle level to 2.31 times (AOR CI: 1.46 - 3.69) for the high level, compared with the low level. Residence was also significantly associated with early screening of breast cancer in women. Women living downtown were 1.58 times (AOR CI: 1.16 - 2.16) more likely to be screened early for breast cancer compared with those living out of town. Figure 7 presents the result of the final multivariate analysis model.

The prevalence of early breast cancer screening among women was 12.93%, meaning 225 women of them. This prevalence is low compared to 16.7% found by Ouédraogo in France in 2013 [8] and the 23.3% revealed by the World Health Organization (WHO) in 2014 in Benin [9] . The difference observed in our study can be explained by the fact that according to the results of our research, 24.83% meaning 432 women have never heard of breast cancer; 13.53% of those who have once heard of it, have no knowledge of its manifestations and 29.48% of those who have heard of its screening, did not know that it can be screened early. Far from these previous figures, a study conducted by the “Association Lalla Salma de Lutte contre le Cancer” in 2006 in Morocco found a prevalence of breast cancer screening that was 40%, higher than ours in the general population similarly [10] . Awareness of early breast cancer screening in the general population is therefore necessary to try to significantly reduce breast cancer morbidity and mortality.

In our study 75.17% of the women collected had heard about breast cancer at least once and the sources of information were mainly friends 50.23%, radio 41.21% and television 38.91%. This same observation was made respectively by Suh et al. In 2012 in Buea, Cameroon, and by Gueye et al. in Senegal in 2009, where nearly three quarters (74.17%) of the participants had already heard of breast self-palpation [11] , and information on self-care comes mainly from television (52.9%) [12] . In addition, Mahdaoui in his study in Morocco among patients aged 30 to 69 years at the level of 15 health centers and two CRSR of the medical prefectures of Rabat and Skhirat-Témara in 2012 by making known that women declare to have information on breast cancer in 91% of the cases [13] . In contrast, Suh et al., in 2012 and the 13^th U.S. referenced study targeting African women who migrated to the U.S. on breast cancer in 2017 conducted by the “Alliance des Ligues Francophones Africaines et Méditerranéennes (ALIAM)”, emphasized that knowledge about self-care or access to screening was low [10] [11] .

In the said study, 68.50%, and 21.71% of women said respectively that breast cancer is manifested in the early stage by the presence of nodule and pain in the breast. Mahdaoui in 2012 in Morocco found the same result and mentioned that control of the symptom known by women for breast cancer is the breast mass (97%) [13] . This finding also appears to be similar to Vahabi’s finding in 2011 among immigrant Iranian women in the city of Toronto aged 25 years and older, where 72% of women associated pain with the onset of breast cancer and were unaware of the clinical breast examination or when and how to screen for the disease [14] . On the other hand, Zannou et al. in 2015 in Cotonou, Benin, in their study of patients seen in consultation or hospitalized in the internal medicine-medical oncology and visceral surgery departments for breast cancer, noted that patients who came to the consultation late, did not have any knowledge of the disease and mentioned reasons of absence of pain (55.6%) at the onset of breast symptoms; fear of the diagnosis (12.7%) and the postpartum period which reassured them (4.8%) [4] . Peltzer et al. in 2014 in their study to assess awareness of the links between breast cancer and certain factors (heredity, diet, overweight, exercise, alcohol consumption, smoking) in 24 low- and average-income countries on 03 continents (Asia, Africa, and the Americas) also found that 35% of women were unaware that any of these risk factors could influence cancer occurrence [15] .

Among the 1308 patients who had heard of breast cancer once, 726 meaning 55.50% reported having heard of breast cancer screening and 512 meaning 70.52% knew that breast cancer can be screened early; and the main means of screening mentioned were breast self-palpation (67.58%) and mammography (37.89%). A similar finding to ours was made by Suh et al. in 2012, where 95% of women believed that breast cancer was preventable, but only 37% knew that breast self-examination could be a screening method [11] . In addition, Mahdaoui in 2012 showed in his research work that women know and practice breast self-examination in 95% of cases [13] . In contrast, Vahabi in Toronto in 2015 in his study showed that lack of knowledge is a barrier to mammography; he found that the majority of women were unaware that the risk of breast cancer increases with age and didn’t know the difference between mammography and other cancer prevention or screening methods [14] . Zannou et al. in Benin were of the same opinion, mentioning that 58 of the 63 patients (92.1%) did not perform breast self-examination and only 5 patients (7.9%) did so [4] . These results highlight the gap between theoretical knowledge and practice [16] . A real breast cancer awareness and screening campaign would be ideal to improve women’s knowledge and practices.

The age of the respondents was significantly associated with early breast cancer screening (p = 0.0015). Indeed, women of 50 years of age or less had 2.4 times the chance of being screened early for breast cancer, compared to those over 50 years. Pujol et al. in 2008 in France made the same observation, as for the majority of cancers, the risk of being affected increases with age. They added that less than 10% of breast cancers occur before the age of 40 and the incidence then increases steadily until the age of 65. This, combined with the fact that the density of the mammary gland is less important at this age, justifies the choice of the age range of 50 to 74 years chosen for screening by mammography [3] . In addition, Mawadzoue in 2011 in Mali in a case-control study showed a significant association between late age at first menarche and an increased risk of developing breast cancer in premenopausal women versus a decreased risk in postmenopausal women (not significant) [17] .

The educational level of the women collected was statistically associated with early breast cancer screening (p < 0.0001). Our study showed that as the level of education increased, so did the percentage of early breast cancer screening in women increase also. Gueye et al., in 2009 in Senegal, made a remark almost identical to ours where knowledge and practices on breast self-examination are strongly influenced by the level of education and the level of financial income (p = 0.02) [12] . The level of education would therefore be a factor favorable to the adherence and periodic practice of early breast cancer screening among women. Accessibility to information on early breast cancer screening should be targeted at all segments of the population, especially those with lower levels of education and financial income, through communication in various local languages.

The marital status of the women investigated was significantly associated with early breast cancer screening (p < 0.0001). Married or partnered women were 1.76 times more likely to be screened for early breast cancer than unmarried women. Indeed, married women or women in couples are more likely to perform breast self-examination since, being in a household, they are more likely to have their breasts palpated by themselves or by their spouse, for example during sexual intercourse, or to perform it on gynecological advice in case of pregnancy or other gynecological consultations. It is important to intensify communication on early screening methods and in particular on self-care.

Women’s place of residence had a significant impact on early breast cancer screening (p = 0.0001). Indeed, women living in downtown were 1.58 times more likely to undergo early breast cancer screening, compared to those living out of town. Zannou et al. reported a finding almost similar to ours, in which patients’ difficult geographic accessibility to referral care sites would explain the use of alternative therapies within reach [4] . The more a woman lives in an urban area, the more likely she is to undergo early breast cancer screening, since she is in constant contact with the media, health care facilities and others. The implementation of a policy of accessibility to information on breast cancer screening and prevention in rural areas would be welcome.

The frequency of early breast cancer screening increased significantly with women’s socioeconomic level (p < 0.0015). This result seems to be consistent with that of ALIAM in 2017 where financial reasons occupy an important place in the process of early breast cancer screening in women; because the cost of consultations, complementary examinations and treatments in a context of lack of social coverage, remains prohibitive for many families; associated with breast cancer screening [10] . This result is reinforced by Elkaou in Morocco in 2014, in his study on the determinants associated with screening and diagnosis evoked during the national program of early breast cancer screening: case of the region of greater Casablanca, indicated respectively the fear of being positive, poverty, lack of financial means, social stigmatization, delay of care as reasons [18] .

To solve this problem, it would be preferable to encourage breast self-examina- tion, to carry out mass screening campaigns, to communicate on the possibility of curing this disease if it is screened early and to think about the implementation of a policy of free treatment.

The objectives of the study were achieved through a cross-sectional, descriptive and analytical observational study using a probability sampling technique (four- stage cluster survey according to WHO) that allowed for the inclusion of a reasonable sample of women (1740) in order to obtain reliable results.

Women’s knowledge, attitudes and practices regarding early breast cancer screening were described. The frequency of early breast cancer screening among women was determined (12.93% i.e. 225 women/1740). Factors associated with early breast cancer screening in women were identified. They were: age (≤50 years), educational level (increasingly higher), marital status (married/coupled), residence area (downtown), and socioeconomic level (average/high).

The data collection tool used was well adapted to the target group, using everyday language to facilitate understanding and limit information bias. The questionnaire interview has its limitations, as do all data collection techniques, because when it comes to women’s knowledge, attitudes, and practices regarding early breast cancer screening, a slight discrepancy could be observed between the actual response and the woman’s behavior. However, steps have been taken to address this and to ensure the validity of the results.