Background
Breast cancer accounts for 23% of all cases of female cancer in Kenya [
1]. The majority of patients present with late stage and locally advanced disease, resulting in high mortality rates [
2]. A necessary first step to improve breast cancer outcomes is improving breast healthcare and breast cancer knowledge in local communities. Another necessary step to mitigate late presentation of the disease is understanding the local barriers to early detection, accurate diagnosis and appropriate treatment [
3].
The limited data for breast cancer incidence that are available from Kenya are predominantly urban, [
1] and hospital-based [
4]. It is likely that breast cancer is also a major health problem in the underserved rural areas of Kenya, but currently very little data exist on breast cancer incidence, breast cancer knowledge and health seeking practice of women from these populations. Women in low- and middle-income countries (LMICs) like Kenya may face multiple socio-economic, religious, cultural, health care provider and health systems barriers to accessing optimal breast cancer care [
5].
In many rural communities of Kenya, men are important decision-makers in the family, however, there are no studies evaluating the influence of men in a woman’s breast healthcare seeking behaviour. Furthermore, there are no studies from Kenya directly assessing the knowledge of breast cancer among men.
We conducted a mixed methods study within a rural coastal community of Kenya to inform effective design of future interventions to improve access to breast cancer services. The aim of this study was to assess knowledge, perceptions and practices related to breast cancer among women, male heads of households, opinion leaders and healthcare providers in the coastal community of Kaloleni sub-county, Kilifi County, Kenya. Our secondary objective was to explore the role of male heads of households in influencing a woman’s breast health seeking behavior.
Methods
Study area and period
This study was performed in Kaloleni, a sub-county of Kilifi County, Kenya, between Sept 1st 2015 and Sept 30th 2016.
Study design and population
The convergence model of a triangulation design was used in this mixed methods study (6).
In this one-phase approach, quantitative and qualitative data were collected and analyzed separately and assigned equal importance, and the results of each component were then compared to achieve a broader understanding of the phenomenon from a diversity of perspectives [
6]. The quantitative component comprised female member and male head of household surveys. The qualitative part consisted of focus group discussions (FGDs), involving opinion leaders, and key informant interviews (KIIs) of health care providers. According to the 2009 Kenya National Population and Housing Census, Kaloleni had a population of about 290,000 living in about 44,000 households, with women of reproductive age comprising 25% of the population [
7]. Approximately 70% of the population lives below the poverty line and 81% of residents rely on subsistence agriculture for their livelihood [
8]. The Aga Khan University - East Africa (AKU-EA) and the University of California, San Francisco (UCSF), in partnership with the Ministry of Health of Kaloleni sub-county, Kilifi County, Kenya, run an Integrated Primary Health Care (IPHC) programme in Kaloleni sub-county. The IPHC programme has developed and maintains a regularly updated population registry of two Community Units (CUs), which provides a sampling frame of 2700 households and 16,005 residents and up-to-date denominator data for AKU - East Africa population-based studies.
The two study site CUs are served by one dispensary (the Kinarani Dispensary) and 100 Community Health Workers (CHWs) who are volunteers selected by the community to implement the community health strategy at the primary level. The Kinarani Dispensary and other nearby dispensaries provide basic primary care and these are linked to the sub-county (district) hospital at Mariakani Township. The Mariakani sub-county hospital is staffed by nurses, clinical officers and 8 medical officers who provide care across all health disciplines. All complex cases (including mastectomies) are referred to the tertiary level Coast Provincial General Hospital in the neighboring Mombasa County.
Sample size calculation and sampling techniques
We randomly selected women aged 15–49 years from the IPHC database. Male heads of households were also selected at random from the same households where women were interviewed. This was a simple random selection without replacement. We did not set an age limit when selecting the males. For the quantitative part of the study, the sample size was based on the formula (Z1-α/2)2*(p(1-p))/d2, where (Z1-α/2) = 1.96, p = the proportion with knowledge of screening methods, which is 40% among women in this region (unpublished pilot data), and d = 0.05, which was the knowledge variable with the lowest frequency in this population . We added 15% to the calculated sample size, to compensate for anticipated non-response, bringing the total sample size to 425 women. For our secondary objective, a total of 237 male heads of households were sampled randomly from 50% of the households from which female respondents were sampled (and this figure also included an additional 10% to compensate for anticipated non-response). For the qualitative part of the study, specific groups were targeted for the FGDs and KIIs based on demographics, societal rank, or occupation. The responsibility for selection of the participants for the FGDs and the KIIs was vested with the local Ministry of Health officials.
Prior to the fieldwork, the study team developed survey tools for males, females, the focus group discussions and the key informant interviews (Additional files
1,
2,
3,
4). The male survey tool had additional questions specifically addressing their attitude towards the health seeking behavior of the women in the household. The male and female survey tools were reviewed, translated into Kiswahili and the local Giriama language, and independently back-translated into English. There were a total of 55 questions in each of these tools, which included the following domains: socio-demographic characteristics; knowledge of breast cancer risk factors, symptoms and signs; knowledge of screening and diagnosis methods; and the perceived barriers to accessing breast health care. The FGD tool was adapted from the Racial and Ethnic Approaches to Community Health (REACH) 2010 Program of the Division of Preventive Medicine, Department of Medicine, University of Alabama at Birmingham. Some questions were revised to be appropriate in the local context. The KII tool was adapted from the Central Texas Affiliate of Susan G. Komen® (Qualitative Data: Ensuring Community Input). Ten field enumerators (CHWs), were trained on breast healthcare and the study protocol and tools by the study team. We piloted the study tools in a neighboring village. Based on the feedback provided by the CHWs, the questionnaires were further refined and adopted.
Focus group discussions
The study team conducted FGDs in Kiswahili and the local dialect at the Kinarani Dispensary. Facilitators were divided into 2 groups, each responsible for conducting three FGDs. A total of 6 groups participated; religious leaders, women elders, traditional healers, women’s group, an administration group comprising chiefs and assistant chiefs, and a group made up of the youth). Inclusion criteria for each FGD were distinct from each other and based on occupation/profession, thus maintaining participant homogeneity. No participant was enrolled in more than one FGD.
All FGDs were recorded. The field coordinator (FA) moderated the first session to set the pace and demonstrate the use of non-technical language, making it easy for the participants to understand terminologies. Each FGD session lasted about an hour and consisted of 17 questions to gain insight into participant understanding of cancer in general and breast cancer specifically, knowledge of screening and treatment options, and experiences in accessing health care services.
The study investigators (ZM, SS, and EN) and field coordinator (FA) conducted Key Informant Interviews in English and Kiswahili, as appropriate. The one-on-one KIIs targeted medical and clinical officers, nurses and technologists within Kaloleni. A total of 25 questions were administered, interrogating the knowledge and perceptions of the health care providers on breast cancer symptoms and signs, screening activities, challenges faced in provision of breast health care services, and perceived barriers that may result in delays in diagnosis. The KIIs also lasted about an hour and were audio recorded.
Data quality control
The interviewers (CHWS) administered the survey tools two weeks after their onsite training. Pairs of interviewers conducted the female and male interviews separately, but during the same household visit. The field coordinator (FA) undertook random onsite quality checks. Trained data clerks double-entered the data, which were verified in Epi Info v7.
Data analysis
Quantitative survey data analysis was performed in Stata ver13. The distributions of the knowledge, perceptions and practice results were tabulated, overall by gender (Additional files
5,
6). Coding and analysis of the qualitative data was modeled after the Rapid Assessment Process [
9], specifically (a) a multidisciplinary group was engaged to apply codes; and (b) an iterative process within the team was used to develop the coding system. Five researchers (MM, ZT, SS, TW and JK) each independently coded the first five transcripts, to establish inter-rater reliability, and then the rest of the transcripts were distributed across these five researchers for coding. New codes were added throughout the analysis. Themes were identified when coding saturation and thematic saturation was achieved (9).
Discussion
Our mixed methods study assessed the breast cancer knowledge, perceptions and practices of women, male heads of households, community opinion leaders, and health care workers in a rural coastal community in Kenya and describes gender-based and structural barriers that prevent women from accessing breast cancer care. To our knowledge, this is the first study from Kenya that also provides insight into the male heads of households’ knowledge of breast cancer and their role in influencing their wives’ access to breast health care.
Nearly half of the women in this study had no formal education, and nearly all of the others had only attended primary school. Most had heard of breast cancer, and knew it was a serious disease, but only a quarter knew any of its signs or symptoms, and only 6% had ever had a clinical breast examination performed by a clinician. Over 90% of the women said they would go to a health facility within a week if they felt a breast lump, but most said that the decisions about whether and where to go would be made by their husbands or other family members.
The male heads of households were better educated, with 80% having had formal education. Over 90% said they knew that breast cancer was a serious disease, and they would take their wife to a health facility within a week if she had a breast lump. They also said they would support her, and not leave her, if she had breast cancer.
The Focus Group Discussions and Key Informant Interviews confirmed that the population, its opinion leaders, and even some of its health care workers had a poor knowledge of the signs and symptoms of breast cancer. In contrast to the quantitative survey results, however, the opinion leaders and health care workers believed strongly that stigma, social rejection, and fear of losing their husbands were important reasons for women to hide or delay seeking help for breast lumps, leading to later breast cancer diagnoses and poorer outcomes. They also felt that using traditional healers was much more common than the quantitative surveys would suggest, reflecting both the population’s ignorance of breast cancer signs and the individual’s hope that the symptoms were due to a less important, less stigmatizing problem, again leading to delayed diagnosis and treatment. The FGDs and KIIs also identified lack of trust in the health care system, inadequate access to services, and lack of standard breast cancer management guidelines throughout the healthcare system as reasons for late diagnosis, suboptimal care and poor outcomes.
There was evidence from the qualitative study that late diagnoses are common. Many of the survey respondents and members of the various focus groups defined breast cancer as a “wound” of the breast. This provides indirect evidence that, similar to most LMICs [
10,
11], including Kenya [
2], breast cancer patients in Kaloleni probably present at an advanced stage of disease when the tumour has ulcerated through the overlying skin. Moderate level of evidence suggests a link between delay in patient presentation with various factors that include, lack of breast cancer awareness, alternative treatment use, and rural residency among others [
12].
Cultural, traditional and religious beliefs are some of the factors contributing to women not seeking breast cancer services from the health facilities. Key barriers to accessing breast health care were the lack of awareness of the risk factors, signs and symptoms of BC; the screening, diagnosis and treatment options for the disease; and access to accurate information. It is interesting to note that health workers also admitted they had inadequate knowledge about BC. They expressed a desire to be provided more educational and training opportunities about BC screening, diagnosis and treatment. This is not unique to Kaloleni; published data from a Nigerian teaching hospital reported that irrespective of age and professional qualification, more than 40% of the115 nurses who participated in their study regarded pain as a sign of early BC [
13]. Similarly, a recent review article from India [
14], reported that irrespective of their socio-economic and educational background, among the 7066 women aged 15–70 years surveyed, there was no improvement in the knowledge of risk factors for BC over an 8 year period; respondents demonstrated varying levels of awareness of risk factors such as family history (13–58%), reproductive history (1–88%) and obesity (11–51%) [
14].
Efforts to improve breast health care requires the recognition of the important role health workers play in public education [
15]. Community opinion leaders and health care providers can play a key role in educating and referring women for screening. Any proposed BC educational and training activity should therefore engage and target community leaders and various cadres of health professionals to ensure cascading of accurate information about BC at the community level and the public at large.
In our study, the data sources and the data collection methods for the quantitative and qualitative components differed, which may help explain some of the differences in the findings of these two components. The quantitative component consisted of interview-administered questionnaire surveys of female members and male heads of households, while the qualitative component consisted of group discussions with community leaders and relatively unstructured discussions with individual health care workers. Social desirability bias could have been a factor in the questionnaire surveys, especially because the questionnaires were administered by community health workers who were members of the community and were well known to those being interviewed. There were also differences seen in the results along gender lines. As noted above, the perception among health care workers, some heads of households, and FGDs of women was that there is often a lack of support for women with breast cancer from male heads of households. However, our survey findings and the FGD groups of men indicated that men would be financially and socially supportive of women in their household who were diagnosed with breast cancer. These discordant findings may be partly due to men answering questions the way they thought they “should” answer them, but it also may indicate that the real barrier may not be the men’s attitudes but the women’s perception of the men’s attitudes, and their (unfounded?) fear of rejection if their male partner found out about their cancer. This has important implications in terms of developing content for awareness programs. It is also important that some village chiefs and heads of households described a desire to be more involved in the breast health care of their wives and female cohabitants.
Perhaps the quantitative data and the openness expressed by some heads of households in the qualitative study reflect an opportunity for engagement of heads of households in breast health, either through community outreach or the CHWs. The role of husbands as a barrier to early detection practices has previously been reported; young Kamba women with breast cancer risked being abused and ostracized by their husbands [
16], and the subservient role of women in many societies leads to denial of symptoms and delay in diagnosis [
17]. There needs to be a continued commitment by civil society and governments to address issues related to the lack of female autonomy, lack of social support or social capital, and marginalization [
18]. In addition, enhanced engagement of men in health decision making and use of study models that are sensitive to gender and household dynamics may support the male role in health decisions that pertain to women’s health, including breast cancer [
19].
According to the quantitative results there was little interest among community members in using traditional healers to manage breast lumps; however the qualitative interviews with women, men, and health care workers described a much more significant role for these healers in responding to breast symptoms and in provision of health care in general. The role of traditional healers in breast cancer detection and treatment should be explored further, to understand whether they are, in fact, associated with delayed diagnosis and treatment in rural African communities. This has been previously researched in other geographical areas, and this belief has been identified among Chinese and South Indian women [
20]. A complex interplay of cultural, spiritual, and individual experiences and perceptions results in traditional and spiritual healers often being the first port-of-call for women with breast symptoms [
21]. Extending educational activities to and encouraging participation of traditional and spiritual healers in the implementation of local breast cancer control programs can be a bridge to build trust and mutual understanding among the various health sectors, potentially encouraging traditional healers to refer patients with breast cancer symptoms more rapidly to government health care facilities.
Structural barriers such as long distances to health facilities, lack of affordable services, lack of equipment to offer screening or treatment, lack of education about cancer, and minimal sensitization of both women and men to the symptoms of breast cancer and the options for diagnosis and treatment were cited as barriers to uptake of these services. Inadequate diagnostic facilities, requiring patients to travel long distances, and the costs of transportation, as well as lack of governmental support for screening, diagnosis, and treatment costs, which were all mentioned by our study participants, have previously been reported as major deterrents in getting timely advice and treatment for breast cancer [
22].
Poverty constitutes the underlying common denominator and most important barrier contributing to lack of awareness and delayed patient presentation. Evidence from multiple studies show that poverty is manifested by lower income, lower education level, rural residency, and lack of access to healthcare systems [
12]. In Kilifi County within which Kaloleni is located, the poverty rate is 70.8%, significantly higher than the national average of 45.9% [
8]. In 2005, the World Health Organization (WHO) member states endorsed a resolution to advocate for the adoption of universal health coverage and proposed health financing systems that can alleviate barriers to access to health care [
23]. However, there are multiple factors to consider, and financial solutions alone will not ensure access to health services [
18].
The use of the Community Health Workers (CHWs) was repeatedly mentioned in our focus group discussions as a good avenue to extend educational activities which are more acceptable to the community. Community education and continuous sensitization to the signs and symptoms of breast cancer, the fact that early detection increases the chances for cure, and the options for screening, diagnosis and treatment will be key to empowering people with accurate knowledge and dispelling some of the myths about breast cancer, which will encourage women to seek breast health services in a timely manner. Organized outreach breast cancer screening activities involving the CHWs would also enable these volunteers to reach remote populations within their units better. Lessons could be learnt from The Deep South Network for Cancer Control (DSN), a community–academic partnership operating in Alabama and Mississippi [
24], which formulated a community action plan to address cancer disparities through focusing on participation of local communities in education, research, and training. Some of the objectives were to increase cancer screening through raising awareness at the individual, provider, and system levels and training community partners to become effective advocates. The guiding principle and success of this plan depended upon trust, respect, and an appreciation of partners’ strengths and differences [
25].
One strength of this study was the combined quantitative and qualitative approach, which allowed additional insights to be revealed when the data from the two approaches were compared. Other strengths were the questioning of the male heads of households and the coordinated questioning of women, men, community leaders and health care professionals.
There were also several limitations to our study. As with most survey-based studies, our quantitative data could have been limited by selection bias and social desirability bias, especially since the surveys were administered by CHWs (community members) who served as the interviewers. The interpretation of the findings of our qualitative research may also have been limited by research bias, although we tried to mitigate this by involving a multidisciplinary team of researchers in this analysis. Finally, this study was confined to a small geographic region of rural Kenya, which may limit the generalizablility of the findings to similar low resource settings.