Introduction
Cervical cancer affects women across the globe, with a disproportionately higher burden of morbidity and mortality in low- and middle-income countries (LMICs) [
1]. In 2018, more than 80% of the 311,000 cervical cancer deaths worldwide occurred in LMICs, predominantly in sub-Saharan Africa (SSA) [
1]. Cervical cancer is classified as a human immunodeficiency virus (HIV)-related malignancy, further contributing to the increasing cancer burden in SSA countries with a high prevalence of HIV [
2]. Additionally, late-stage cervical cancer at diagnosis is significantly associated with increased cervical cancer mortality, and approximately 68% of cervical cancers are diagnosed at a late stage in SSA [
3]. Thus, it is paramount to detect cervical cancer at an earlier, more treatable stage in order to significantly reduce cervical cancer deaths [
4,
5].
Cervical cancer screening aims to prevent invasive cancer [
6‐
10]. Other clinical and sociodemographic factors have been associated with late-stage cervical cancer at diagnosis, particularly in LMICS, including abnormal vaginal bleeding [
6,
7,
11,
12], age at diagnosis [
9,
13‐
15], marital status [
6,
15‐
18], and living in a rural area [
7,
15,
16,
18,
19]. In addition, the practice of traditional healers has been shown to be associated with an increase in late-stage cervical cancer at diagnosis [
20] and as a barrier to cervical cancer care in low-resource settings [
21].
In Botswana, an upper-middle-income country in SSA, cervical cancer is the leading cause of female cancer deaths [
22,
23]. However, there remains a dearth of information regarding the demographics and clinical factors that contribute to late-stage cervical cancer at diagnosis among women from SSA. Botswana also has a high prevalence of HIV, with 25.1% of females between 15–49 years of age living with HIV in 2019 [
24]. In recent decades, with the growing burden of HIV-related cancers [
25], the Botswana Ministry of Health and Wellness (MOHW) has prioritized reducing the cervical cancer burden by adapting American Society of Clinical Oncology (ASCO) resource stratified screening strategies for its citizens, with the majority of cervical cancers being detected through loop electrosurgical excision procedure or visual inspection with acetic acid [
26‐
29]. Despite these efforts by the Botswana MOHW, approximately 50% of cervical cancers are diagnosed at a late stage [
6,
30].
This study aims to identify potential clinical and sociodemographic factors associated with a late-stage diagnosis of cervical cancer in Botswana in order to help inform future interventions at the community and individual levels aimed at decreasing cervical cancer morbidity and mortality.
Characteristics by HIV status
Table
4 shows patient characteristics by HIV status. Of the 984 cervical cancer cases, the HIV status of 967 (98.3%) women was known. Women whose HIV status was unknown (n = 17) were excluded from the HIV stratified analyses. WLWH comprised 69.7% (n = 674) of the study population and were significantly younger than women without HIV (45.8 years vs. 60.5 years, p < 0.001). WLWH were also more likely to live in urban areas (24.0% vs. 14.7%, p = 0.001), were more likely to have never been married (74.0% vs. 47.8%, p < 0.001), and were more likely to have been screened for cervical cancer (61.8% vs. 48.4%, p < 0.001) than women without HIV. In addition, WLWH were more likely to have SCC pathology than women without HIV (88.9% vs. 83.7%, p = 0.020). Among the WLWH, 78.6% (n = 429) had a CD4 cell count > 250 cells/mm
3 at diagnosis, and 96.2% (n = 640) reported being on ART.
Table 4
Clinical and demographic characteristics of the study population by HIV status
| 293 | 30.3 | 674 | 69.7 | |
Age categories | | | | | |
< 30 | 2 | 0.7 | 11 | 1.6 | < 0.001* |
≥ 30–40 | 23 | 7.9 | 165 | 24.5 | |
≥ 40–50 | 44 | 15.1 | 324 | 48.1 | |
≥ 50–60 | 60 | 20.5 | 119 | 17.7 | |
≥ 60–70 | 98 | 33.6 | 47 | 7.0 | |
≥ 70 | 65 | 22.3 | 8 | 1.2 | |
Residence | | | | | |
Rural | 250 | 85.3 | 511 | 76.0 | < 0.001* |
Urban | 43 | 14.7 | 161 | 24.0 | |
Marital status | | | | | |
Never married/single | 140 | 47.8 | 498 | 74.0 | < 0.001* |
Married | 91 | 31.1 | 132 | 19.6 | |
Divorced | 3 | 1.0 | 9 | 1.3 | |
Widowed | 59 | 20.1 | 34 | 5.1 | |
Previous cervical cancer screening | | | | | |
Never screened | 143 | 51.6 | 248 | 38.2 | < 0.001* |
Screened | 134 | 48.4 | 401 | 61.8 | |
Visit with a traditional healer | | | | | |
No | 250 | 87.7 | 599 | 90.9 | 0.140 |
Yes | 35 | 12.3 | 60 | 9.1 | |
FIGO Stage | | | | | |
I | 48 | 17.0 | 115 | 17.8 | 0.83 |
II | 106 | 37.6 | 225 | 34.9 | |
III | 100 | 35.5 | 245 | 38.0 | |
IV | 28 | 9.9 | 60 | 9.3 | |
Pathology | | | | | |
SCC | 239 | 83.9 | 583 | 88.9 | 0.022* |
Adenocarcinoma | 31 | 10.9 | 33 | 5.0 | |
Invasive ductal | 1 | 0.4 | 3 | 0.5 | |
Other | 9 | 3.2 | 30 | 4.6 | |
Vascular invasion | 0 | 0 | 1 | 0.2 | |
Unknown | 5 | 1.8 | 6 | 0.9 | |
Abnormal vaginal bleeding | | | | | |
Not reported | 68 | 23.2 | 192 | 28.5 | 0.089 |
Reported | 225 | 76.8 | 482 | 71.5 | |
CD4 | | | | | |
< 250 cells/mm3 | – | – | 117 | 21.4 | – |
≥ 250 cells/mm3 | – | – | 429 | 78.6 | |
ART | | | | | |
No | – | – | 25 | 3.8 | – |
Yes | – | – | 640 | 96.2 | |
Table
5 shows the results of the multivariable logistic regression models by HIV status. Among the WLWH, prior cervical cancer screening showed decreased odds with late-stage disease at diagnosis (OR: 0.61, 95% CI 0.44–0.86), and increased odds with previous abnormal bleeding symptoms (OR: 2.10, 95% CI 1.46–3.01). Among women without HIV, factors associated with higher odds of late-stage disease at diagnosis included increasing age (OR: 1.02, 95% CI 1.00–1.14; p = 0.041) and abnormal vaginal bleeding (OR: 3.06, 95% CI 1.52–5.71). Results of complete case analyses by HIV status were similar to the imputed results.
Table 5
Imputed multivariable subgroup analyses of women living with HIV and women without HIV
Age | 1.01 | 0.99–1.02 | 0.480 | 1.02 | 1.00–1.04 | 0.041* |
Urban vs. rural residence | 0.79 | 0.54–1.16 | 0.234 | 0.88 | 0.43–1.81 | 0.730 |
Never married/single vs. married/widowed/divorced | 1.23 | 0.85–1.79 | 0.273 | 1.59 | 0.94–2.70 | 0.087 |
Cervical cancer screening vs. never screened | 0.61 | 0.44–0.86 | 0.004* | 0.80 | 0.48–1.35 | 0.413 |
Visit with a traditional healer (Ye/No) | 0.95 | 0.55–1.65 | 0.864 | 1.52 | 0.71–3.26 | 0.276 |
Abnormal vaginal bleeding (Yes/No) | 2.10 | 1.46–3.01 | < 0.001* | 3.06 | 1.52–5.71 | 0.001* |
CD4 count < 250 vs. ≥ 250 | 1.15 | 0.73–1.79 | 0.546 | | | |
Anti-retroviral treatment (Yes/No) | 1.11 | 0.46–2.65 | 0.821 | | | |
Discussion
This large study of women with cervical cancer in Botswana aimed to identify potential clinical and sociodemographic factors associated with a late-stage diagnosis of cervical cancer in Botswana. Our study showed that prior cervical cancer screening was associated with decreased odds of having late-stage cervical cancer at diagnosis, whereas experiencing abnormal vaginal bleeding and having never been married were associated with an increased odds of having late-stage cervical cancer at diagnosis. Having HIV was not associated with having late-stage cervical cancer at diagnosis. Furthermore, results suggested that women living in rural areas who visited a traditional healer were more likely to be diagnosed with late-stage cervical cancer.
Screening has been shown to lead to an earlier diagnosis of cervical cancer in high-income countries with established screening programs, and screening has also been shown to be effective in low resource settings [
7,
9,
38]. With the growing burden of HIV-related cancers in recent decades [
25], the Botswana MOHW has prioritized reducing the cervical cancer burden through implementing and supporting a national cervical cancer screening program as part of the HIV care continuum [
23,
27,
28,
39,
40]. A prior retrospective study by Nassali et al. (2018) reviewed 149 cervical cancer patients admitted to PMH from August 2016 to February 2017, which may include some overlap with our study sample. In that study, Nassali et al. defined late-stage cervical cancer as FIGO stage IB2-IVB, and found an increased odds of presenting with late-stage tumors in patients not previously screened for cervical cancer. Additionally, two qualitative studies in Botswana [
41,
42] have shown that lack of knowledge regarding the benefits of screening for cervical cancer can delay diagnosis. Our study provides further evidence supporting the finding that screening decreases the odds of presenting with late-stage cervical cancer at diagnosis when implemented in a low resource setting.
Early-stage asymptomatic cervical cancer can be detected through screening, but in the absence of screening, patients with cervical cancer can present with clinical symptoms including abnormal vaginal bleeding and post-coital bleeding [
34]. Reports of symptomatic bleeding and its association with late-stage disease at diagnosis in low resource settings have been inconsistent in the literature [
6,
7,
11,
12]. Studies have shown an increased risk [
6], no association [
7], and a decreased risk [
11,
12]. In Nepal, a decreased risk for having late-stage cervical cancer at diagnosis was noted if the symptoms of bleeding were reported first to the woman’s husband, who might encourage his wife to seek medical care. In Morocco, the decreased association between having late-stage cervical cancer at diagnosis and having symptomatic bleeding was hypothesized to be due to understanding the severity of bleeding as a cervical cancer symptom and seeking medical care without delay. Two studies in Botswana [
20,
43] showed that the perception of symptom severity was related to having advanced stage cervical cancer at diagnosis and to having a delay in health seeking behavior. In our study, reporting gynecological bleeding symptoms, including previous abnormal vaginal bleeding and/or post-coital bleeding, was associated with a two-fold increase in the odds of presenting with late-stage cervical cancer. These results support increasing awareness regarding abnormal vaginal bleeding and post-coital bleeding as indications of cervical cancer and emphasizing the need to seek medical care as soon as possible. In addition, these findings support screening asymptomatic women to be able to diagnose cervical cancer at an early stage.
In our study, there was no significant difference between WLWH and women without HIV with regard to their likelihood of presenting with advanced stage cervical cancer. Some studies have recognized HIV as a risk factor for late-stage cervical cancer at diagnosis [
16,
19], yet other studies [
6,
44] have reported no association with HIV and late-stage cervical cancer at diagnosis. The role of HIV in the diagnosis of late-stage cervical cancer remains unclear and should be investigated further. When comparing WLWH and women without HIV in our cohort, the WLWH were younger, were more likely to have undergone cervical cancer screening, had more often lived in urban areas, and were more likely to be married or to have been married than women without HIV. Also, in the subgroup analyses, associations with late-stage cervical cancer at diagnosis differed between the two subgroups of WLWH versus women without HIV. Increasing age was significantly associated with late-stage cervical cancer at diagnosis in women without HIV, but not in WLWH. WLWH with a history of cervical cancer screening had lower odds of presenting with late-stage cervical cancer at diagnosis; however cervical cancer screening was not significantly associated with late-stage cervical cancer at diagnosis in women without HIV. In Botswana, cervical cancer screening programs have been implemented as part of the HIV care continuum for women, making it is plausible that women without HIV do not access screening services to the same extent as WLWH. Therefore, increasing screening services among women without HIV could reduce the prevalence of late-stage cervical cancer at diagnosis for these women.
Our study saw a decrease in late-stage cervical cancer cases among women who had been married. Similarly, a study from Nepal [
11] noted that married women with symptomatic bleeding were less likely to present with a late-stage cervical cancer because husbands may encourage their wives to seek medical care. A study by Ibrahimi and Pinheiro (2017) in the United States reported that being married was an independent predictor of a more favorable prognosis of cervical cancer [
45]. While some studies have identified being unmarried as a risk factor for having late-stage cervical cancer at diagnosis [
6,
17,
18], other studies have found no such association [
15,
16]. Reasons for this association are unclear. Future studies investigating differences in financial, emotional, and sociocultural marital structures and the impact on prompt cancer diagnosis are warranted. These inconsistencies could be attributed to differences among the sociocultural marital structures and support systems across countries, which need to be explored further.
Areas of residence may also impact health care. For example, living in a rural area versus living in an urban area has been investigated as a potential contributing factor for women with late-stage cervical cancer at diagnosis [
7,
15,
16,
18,
19]. A study in Sudan reported that an increased risk for having late-stage cervical cancer at diagnosis was associated with living in a rural setting versus living in an urban setting [
15], but this result was not seen in our cohort, nor was any such association reported in other studies in Ghana [
7], Florida [
18], Ethiopia [
19], or SSA [
15]. The use of traditional healers has been shown to be associated with late-stage cervical cancer at diagnosis in Botswana [
20] and as a barrier to cervical cancer care in Uganda [
21]. In Botswana, over 95% of traditional healers live in rural areas [
46], and thus women living in a rural area may be more likely to consult with traditional healers as their first choice for healthcare. In our study, when examining only women living in a rural area, those who visited a traditional healer had a higher odds of presenting with late-stage cervical cancer. Points of intervention in rural areas could include educating traditional healers to recognize symptoms of cervical cancer in order to facilitate a referral for diagnosis and treatment.
This large study investigating late-stage cervical cancer at diagnosis in Botswana includes detailed demographic and clinical information on patients in Botswana collected over a five-year period, but it does have several potential limitations and challenges. It is important to note that, due to the cross-sectional study design, no decisive conclusions can be made about the temporality or causality among the study variables and late stage diagnosis. Patients were enrolled at PMH, a public tertiary hospital with oncology services, and at GPH, a private tertiary hospital with the only chemo-radiation oncology center in Botswana; thus, all patients who need radiotherapy should be sent to GPH. We were unable to account for patients who were not diagnosed with or treated for cervical cancer outside of these two facilities. In addition, the study collected data at the time of diagnosis and is therefore subject to recall bias, social desirability bias, and potential unmeasured confounding and missing data. Unfortunately, due to the retrospective nature of the study, we lacked important information and were unable to account for confounders including education level, knowledge and awareness of cervical cancer, and cervical cancer screening. To account for any bias due to missing data, we conducted MICE for the primary analysis [
35‐
37]. Results using MICE were similar to the complete case analysis. Although our findings represent a large proportion of cervical cancer cases in Botswana, they do not represent all cervical cancer patients in Botswana and are not generalizable to the entire country.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.