The health district of Koumpentoum is located Koumpentoum department, Tambacounda region. Its population was (164,409) inhabitants in (2020) on a surface area of (7652) Km², i.e. (21.48) inhabitants/km² [14]. The district had (23) health delivery points including a health centre which polarizes (22) health posts. In terms of human resources, the district had (03) physicians, (20) nurses, (10) nurses assistants and (24) midwives. Screening is carried out at all health posts and PCLs treatment at the health centre.

This was a cross-sectional, descriptive and analytical study during the whole month of September (2020).

Study population consisted of sexually active women aged between (18) and (69) years found in households in Koumpentoum district.

The corresponding author designed the questionnaire which included the following items: socio-demographic data, knowledge, attitudes and practices regarding PCLs screening. A pre-test was conducted with (30) women found in the health centre. The reliability and validity of the questionnaire were validated by the author and his collaborators, taking into account the suggestions of the members of the district management team who were responsible for conducting the pre-test surveys. The pre-test data were not included in the analysis.

All women aged between (18) and (69) years present in the households and who agreed to participate in the study were included.

Women who refused to participate in the study or who were absent at the time of the survey were not included.

The target of the study was sexually active women aged between (18) and (69) years living in the district. The sample size was calculated using Schwartz’s formula [15]:

N = Z 2 ∗ p ∗ ( 1 − p ) ∗ d / i 2

with:

- p = (90%) as the desired coverage of PCLs screening;

- q = (1) − p = (10%), corresponding to the complementary probability;

- i = (0.05) was the margin of error for precision of (95%);

- d = (1.5) was the effect of the sampling design;

N = (1.96)² * (0.90) * (0.10) * (1.5)/(0.05)² = (207).

We added (10%) of the non-respondents to the sample, which gives (228). The sampling was stratified in two stages. The first stage corresponded to villages or neighborhoods with a simple random selection of one village or neighborhood from each area of responsibility, while the second stage corresponded to households. For the selection of concessions, the interviewer stood at sociological centre of the village or neighborhood to determine a direction at random. With a pen thrown in the air, the tip of which indicated the direction of departure, he started with the first concession indicated. An exhaustive census of sexually active women aged between (18) and (69) years found in each compound was carried out. All sexually active women found in the houses who agreed to participate in the study were interviewed, which bringings the final sample size to (294).

The dependent variable was the practice of PCLs screening. The independent variables were socio-demographics, women’s knowledge of cervical cancer, attitudes and practices regarding PCLs screening.

To collect the data, we used a PCLs screening intake form administered as a closed-ended structured questionnaire directly to sexually active women aged between (18) and (69) years.

Epi-Info version (7) was used to create the data entry template and to obtain the database. Epi Info version (7.2.5.0), SPSS version (20) and R version (4.05) softwares was used to analyze the data. Quantitative variables were described in a descriptive analysis using extremes, means and standard deviations, while qualitative variables were described using frequencies with confidence intervals (95% CI). Pearson’s chi2 statistical tests and Fisher’s test under the conditions of applicability allowed us to verify the existence of a statistically significant relationship in the bivariate analysis. In the multivariate analysis, we used multiple logistic regression and included all independent variables with a p < (0.25) found in the bivariate analysis and those whose literature review had revealed a link with the independent variable. The top-down stepwise method allowed us to retain the variables associated with screening practice at a threshold of p < (5%). In the second step, these variables were removed one by one with a comparison of the nested models by the Aikake information criterion (AIC) [16]. The process was continued until no improvement was found by the maximum likelihood test. The Hosmer-Lemeshow test [17] verified the adequacy of the final model. The strength of the associations between screening practice and the independent variables was assessed in bivariate with the OR and in multivariate with the adjusted OR, surrounded by their (95%) CIs.

In order to obtain approval, the study protocol was submitted to the health authorities. With an information letter read in the local language, the survey was explained to all participants. Free consent was collected and documented on a form with the woman’s signature of consent to the survey. All data were collected anonymously, stored and used for research purposes only.

We interviewed (294) women aged between (18) and (69) years. The average age was (35.79) ± (9.8) years. Almost three quarters (74.49%) of the women were under the age of (40) years. Among them, (89.79%) were living in a couple, (33.7%) had attended French school and only (24.49%) of them were engaged in an income-generating activity (IGA) (Table 1).

While information about cervical cancer was found in (83%) of the cases, only

(43.87%) knew the risk factors. Knowledge of the availability of screening was (16.67%); knowledge of the place of screening (16.33%) and knowledge of the possibility to cure (62.59%). If knowledge of at least one symptom suggestive of the disease was found in (15.3%) of cases, recourse to care in case of signs suggestive of the disease was (95.58%). The health channel through the providers and community health workers was the main source of information (48.36%), followed by the radio, particularly Niani FM (community radio) which broadcasts throughout the district (37.7%) (Table 2).

The practice of screening was (16%), clearly dominated by the VIAA/VIL test in (89.36) of cases. More than half of the women (53.2%) were screened during the free campaigns and only (42.52%) of them had received screening advice from a health provider. The main reasons why they didn’t do screening were lack of knowledge in (41.3%) and fear of knowing the result in (23.07%) of cases. The frequency of PCLs was (4.26%) (Table 2).

Factors statistically and significantly associated with screening uptake were: information about cervical cancer (OR = 5.42 [1.27 - 23.16]; p = 0.003), knowledge of screening availability (OR = 9.26 [4.57 - 18.76]; p < 0.001), knowledge of at least one symptom suggestive of cervical cancer (6.57 [3.23 - 13.39]; p < 0.001) and prompting by a provider (OR = 4.48 [2.25 - 8.93]; p < 0.001) (Table 3).

In multiple logistic regression, only knowledge of the availability of screening (ORa = 7.73 [3.01 - 18.51], p < 0.001) and incentive from a health provider (ORa = 3.03 [1.25 - 7.31], p = 0.014) were statistically and significantly associated with screening uptake (Table 4).

Our study found a screening practice for PCLs of (16%) in the district. This frequency of PLCs screening in the district was higher than those observed by Faye et al. in Kaffrine (6.5%) [12] in Senegal, by Cunningham et al. in Tanzania (6%) [18], by Adrianampy H.A. et al. at the CHU of Fianarantsoa (2.33%) [19] and by Thiam in Kédougou (2%) [13]. However, much higher frequencies of screening were observed in Senegal by Guéye et al. in Joal with (28.32%) [11] and by Faye et al. in Thiès with (35.1%) [20]. These differences could be explained by the importance of organizing free screening days and by the routinization of screening in these districts [11].

The frequency of PLCs was (4.26%), similar to that found in Cameroon by Ngowa J.D.K. et al. with (6.4%) (n = 421) [21], Tebeu P.M. et al. with (5.1%) (n = 2485) [22] and Ngekoum B et al. with (7.4%) (n = 946) [23] and in Madagascar by Adrianampy H.A. et al. with 8.5% (n = 112) [19]. However, higher frequencies were observed in other studies carried out in Africa, notably by Bakhoum A. with (10.2%) (n = 899) [24], Makoussa D. et al. with (15.36%) (n = 1437) [25] and Keita M. et al. with (38.1%) in Bamako (n = 3302) [26]. The level of training and experience of providers as well as the technical quality of screening could influence the results. Periodic retraining of providers, regular supervision of screening sites and quality control of services are essential.

Factors statistically and significantly associated with participation in screening were: cervical cancer information (OR = 5.42 [1.27 - 23.16]; p = 0.003), knowledge of at least one symptom suggestive of cervical cancer (OR = 6.57 [3.23 - 13.39]; p < 0.001), knowledge of the availability of screening (ORa = 7.73 [3.01 - 18.51]; p < 0.001), and prompting by a health worker (ORa = 3.03 [1.25 - 7.31], p = 0.014). Age was not statistically associated with screening, unlike Cunningham et al. [18] in Tanzania and Compaoré et al. in Burkina Faso [27]. Indeed, WHO [28] and national programs [29] recommend screening every (3) years for women aged between (30) and (69) years. Women who were informed about cervical cancer were 5.42 times more likely (OR = 5.42 [1.27 - 23.16]; p < 0.001) to be screened than those who were not. Cunningham et al. in Tanzania [18], Faye et al. in Senegal [20], Guéye et al. in Joal [11] and Faye et al. [12] in Kaffrine had made the same observation. Similarly, knowledge of the availability of the screening was predictive of screening (ORa = 7.73 [3.01 - 18.51]; p < 0.001). There was also a statistically significant association between knowledge of at least one symptom of cervical cancer and screening uptake (OR = 6.57 [3.23 - 13.39]; p < 0.001), as in the Joal series by Guéye et al. [11] and Faye et al. [12] [20]. In our study, lack of awareness of the disease was the main obstacle to screening. Informing screening targets about the disease is an essential component of the cervical cancer elimination strategy [30] [31]. Provider prompting was predictive of participation in screening (ORa = 3.03 [1.25 - 7.31]; p = 0.014). Compaoré et al. in Burkina Faso [27] and Guéye et al. in Joal et al. [11] found the same relationship in their samples. Raising awareness of PLCs screening through the health channel in particular could significantly improve screening practice. Indeed, health facilities should systematize the offer of screening to sexually active women during the different types of consultations that constitute awareness-raising opportunities. This awareness raising should also be reinforced by increased use of community health workers. We did not find any link between the practice of screening and the practice of an IGA, nor between women’s education, unlike Faye et al. [12] and Compaoré et al. [27]. In the district, screening is charged at (1000) FCFA in primary care and this could be an obstacle to screening practice due to the low living standard of the population [14]. A high level of education, among women could facilitate awareness of the disease and systematic screening.

The main limitation of our study was the conduct of the survey by the district providers, which could lead to desirability bias. However, the interviewers were deployed outside their area of responsibility and our results are consistent with those found by other authors.