Study setting

The study took place in Bamako, the capital of Mali, in Western Africa and was part of a larger project on breast cancer diagnosis in Mali [10]. Breast cancer diagnoses are routinely performed at the only pathology department in the country, at the University Hospital Point G, where this study took place. Fine needle aspiration is the standard diagnostic procedure; some patients received a diagnosis of histopathology after biopsy or surgical interventions. Immune histopathology was not performed. Chemotherapy, surgery and radiotherapy are available in hospitals and private clinics in Bamako. At the beginning of the project, an association of breast cancer patients was contacted to discuss the project and to obtain advice on important research topics. For this study, breast cancer patients were involved in the pilot testing of the questionnaire to improve the comfortability of the interviews for patients and to include important items from the patient’s point of view.

Data collection

We used a convenient sample as our study has an explorative character. All the female patients who received a breast cancer diagnosis at the department of pathology of the University Hospital Point G in Bamako, Mali, between January 1st and April 30th 2016 were included in this study (N = 64). Depending on the type of diagnosis, patients were interviewed at the pathology department the day of the fine needle aspiration or biopsy, or when they collected the results after an excisional biopsy. All patients were interviewed personally with a standardised questionnaire by two female medical students who were familiar with the study and interview techniques.

Patients were informed about the study and asked to confirm their willingness to participate. Ethical clearance was obtained by permission from the local ethics committee of the Medical Faculty in Bamako, Mali, with permission given for either written or oral consent. The implementation of written consent in the pilot phase confused and alienated the majority of patients, who were mainly illiterate and unfamiliar with a signing praxis, so we decided to obtain consent orally. Patients were included if they clearly stated their willingness to participate in the study, including participation in the interview, use of medical records and possible phone contact for follow-up.

The questionnaire was designed to analyse the patient pathway from first symptom recognition to beginning of treatment. The questionnaire was developed in accordance with questionnaires used for similar studies in Mexico [19] and Morocco [15] and in relation to the model of pathway to treatment [20]. The questionnaire can be found as supporting information in the original French (S1 Questionnaire) and translated English version (S2 Questionnaire). Dates of symptom recognition, first healthcare visit, diagnosis and start of treatment were collected. A calendar technique [21] was used to help patients to recall dates of symptom recognition and first healthcare visit due to breast related symptoms. The obtained information was confirmed by medical records, when available. The questionnaire also included items to describe the events in more detail (e.g., type of symptom recognised, type and number of medical doctors consulted, type of healthcare services) and factors that could influence the time intervals between the measured time points (e.g., visited a traditional healer, no referral). The collected data was linked with sociodemographic and clinical data routinely collected at the pathology service. The questionnaire was pretested with the interviewers and four patients in personal interviews, lasting about one hour. The questionnaire was in French but, if needed, interviews were conducted in the local language, Bambara. In the pilot phase, the interviewers were trained to translate the meaning of the questions into Bambara. Minor changes helped to improve understanding and the fluency of interviews. The personal interviews lasted about 20 minutes and took place in a protected room at the pathology department.

All interviewed patients or their families were contacted by telephone for follow-up between 18 and 23 months after diagnosis to receive information about the beginning of treatment and eventual date of death. Medical records were additionally searched at the main hospitals, where breast cancer surgery and chemotherapy are available. No information on treatment and survival was gathered for 9 patients.

Data analysis

Of the total 66 patients registered at the pathology department of University Hospital Point G, 64 consented to participate. Simple Cox regression analyses were performed to analyse the influence of sociodemographic factors (age, occupation, marital status, residence, health insurance), patient related factors (knowledge of breast self-examination, visit of a traditional healer, first symptom interpretation), healthcare system related factors (facility and type of medical doctor at the first healthcare visit, first diagnosis, facility and doctor of diagnosis) and pathological diagnosis (tumour size, lymph node involvement) on the lengths of three different time intervals and on survival. The three time intervals were the time to first healthcare visit (time between recognition of the first symptom and first healthcare visit), time to diagnosis (time from first healthcare visit to diagnosis) and time to treatment (time from diagnosis to start of treatment). Crude hazard ratios (HR), 95% confidence intervals (CI) and p values are reported and interpreted in an explorative sense.

For the variables with the largest effect on overall survival in the simple Cox regressions (tumour size T3/T4, traditional healer, and health insurance status), Kaplan-Meier survival curves and log-rank test were additionally performed to illustrate the findings.

Principal findings

We presented time lengths to first healthcare visit, diagnosis and treatment start among Malian breast cancer patients and compared them with results from similar studies in sub-Saharan Africa published within the last 10 years (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Comparisons of time to first health care visit, diagnosis and treatment reported in studies from Africa.

https://doi.org/10.1371/journal.pone.0207928.t003

A problem in comparing the results is the reporting of study results: some studies only reported mean or median times and some only the proportions of women with a certain time delay. Independent of the reporting style, large time differences between countries and within countries for all time intervals, but especially for the interval between first symptom recognition and first healthcare visit, were reported. These differences probably reflect differences in the study population and study setting, but might also be explained by the different methodologies used to collect time points, such as the use of medical records, differences in interview technique, reporting biases and definitions of time points. To make study results more comparable in the future, the Aarhus statement to improve the design and reporting of studies on early cancer diagnosis should be considered [20].

This study shows that the median time length to first healthcare visit is almost five times as long as the consecutive time to diagnosis and this data supports the results from comparable studies focusing on early breast cancer diagnosis [7, 8], which found that awareness and knowledge about breast cancer and breast self-examination trigger recognition of initial breast cancer symptoms and timely healthcare seeking. The positive association of advanced diseases (T3/T4 tumours) with time to first healthcare visit [28] and shorter survival times [5] was further supported in our study. These results generally led to the conclusion that population-wide awareness and education programs are needed to improve the recognition of symptoms and promote prompt healthcare seeking to improve early diagnosis and survival [2]. Use of traditional medicine was another factor associated with prolonged time to first healthcare visit in our and comparable studies from sub-Saharan Africa [7, 8]. We also found shorter overall survival times for women visiting traditional healers before the first healthcare visit. Use of traditional medicine might have a negative impact on survival when it leads to advanced stages of the disease and when it is used as an alternative to standard cancer therapy. Cooperation with traditional healers and their education about symptoms have been suggested to improve early breast cancer diagnoses [30]. However, qualitative studies found that patients use traditional medicine during the whole patient journey, from first symptom recognition to treatment with a complementary or alternative treatment [10], and that reasons for using traditional medicine are manifold and also related to a lack of access and availability or low quality of healthcare services or it can be due to financial barriers, as also reported in other health contexts in Mali [31, 32]. Having no health insurance increased the time to first healthcare visit and reduced the survival time in our study. Health insurance was often not considered in comparable studies. In Mali as in most countries in sub-Saharan Africa, only a minority of the population has any health insurance coverage. We do not know if the better general socioeconomic status of health insured patients might have an impact on survival or if this is related to reduced patients’ out-of-pocket costs and the possibility of adhering to the treatment plan or seeking better treatment. Nevertheless, as costs for breast cancer treatment exceed the average income of families in sub-Saharan Africa [33], universal health coverage that includes costs for breast cancer care might have a great impact on survival, if adequate treatment options are available.

Half of the patients had a diagnostic service performed at the pathology department within 1 month, but 25% took 6 months and longer. Older women seem to receive a diagnosis faster after a first healthcare visit, probably because medical doctors associate cancer more with older women and, among younger women, breast related symptoms might often fall in times of breast feeding and be difficult to differentiate from breast cancer symptoms [34]. Working women, which includes all women having another status than being a housewife, took longer to receive a diagnosis and to begin treatment, which was probably due to time constraints and the need to support the family financially. The longer time to treatment start might also reflect more extensive and time-consuming diagnostic investigations and could also explain the increased time to beginning treatment for women with health insurance. Women who first visited a community healthcare center or a generalist took longer to receive a diagnosis, probably due to limited inward diagnostic facilities and the need to refer patients for further investigations. Similar results were reported in a study in Egypt [25] and can also reflect shortcoming in diagnosis and referral at this level of the healthcare system, for example, women who were told they had an infection or nothing at their first healthcare visit took longer to receive a diagnosis. Education programs for medical doctors or nurses working at the entry level of the healthcare system to enable them to recognize and communicate breast cancer symptoms and to provide referral might be useful to improve early diagnosis [2], but since other studies showed that low social support, lack of money, fear of diagnosis and mistrust in modern medicine are also barriers for early diagnosis [10, 11], such programs can only be part of larger health system strengthening and education programs.

Married women took longer to obtain a diagnosis and begin treatment in our study, which might reflect their dependency on their husband to make important decisions about healthcare and the sometimes low or negative support from husbands as reported elsewhere [10, 32].

An unexpected finding was that women living outside Bamako received a diagnosis faster than women living in Bamako, since the distance to the pathology service is longer, which was shown to be a barrier for early stage diagnosis of breast cancer in patients [35]. Whether this result is random or describes systematic differences for women from Bamako and outside the area needs to be investigated in future studies. Seeing that patients from Bamako began treatment earlier, was an expected finding since treatment facilities are close by and transport and accommodation of the accompanying family is cheaper and easier to organize. Shorter times to treatment were also found for women with larger tumour size, probably because of the greater urgency for treatment perceived by the medical doctors and especially the patient and her family.

While we found several factors associated with the lengths of the different time intervals and survival, we could also show that at least 16%, but probably up to 28% (assuming that the lost to follow-up cases did not start any treatment), of all patients did not start any treatment. This result highlights the bottleneck of early detection efforts. We further found that at least 37%, but probably up to 47%, of all patients died within 18 months after receiving a diagnosis. A study from Nigeria that followed-up breast cancer patients between 1 week and 6 years reported that 90% of all patients (from between 39% and 87%, including all patients lost to follow-up) who died, did so within the first year after diagnosis [36]. This high mortality rates are probably related to the advanced disease at time of diagnosis and lack of access, availability and affordability of treatment options in the countries.

Limitations

This study followed-up all patients for at least 18 months that had received a diagnosis of breast cancer at the only pathology department in Mali during the first four months of 2016. A considerable effort was made to obtain precise time points for first symptom recognition, first healthcare visit and beginning of treatment and death (use of calendar techniques, verification by medical records provided by the patients and by searching medical records in the two main hospitals for cancer treatment in Bamako). Still, we lost ten patients during the follow-up time, who very likely did not receive any treatment and have probably died. A weakness of the study is the relatively small number of patients that were included in the study, therefore multivariable analyses on factors leading to delays were not conducted. As with other patient studies, this study does not include breast cancer patients who did not arrive at the hospital for diagnostic services and therefore important barriers for first healthcare seeking and diagnosis might be missing. It is also unknown how many women get lost after a first healthcare visit or never seek healthcare and these proportions might also vary largely between studies. These limitations effect the generalizability of our results.

Implications

The results show that a continuum of breast cancer care is currently not given for the majority of patients in Mali, but has to be considered to fully benefit from any early detection programs and to improve survival of breast cancer patients. Focus should be on financial barriers at the patient level but also on barriers in providing adequate diagnostic and treatment services, including the needed medical work force. Only if the healthcare system can warrant a continuum of care for the majority of diagnosed breast cancer patients, would breast cancer awareness campaigns with down-staging strategies be useful. International efforts in providing standard diagnostic and treatment options, palliative care at low cost and training for medical doctors should be prioritised to improve breast cancer survival in Mali and other sub-Saharan countries.

Future research

Prospective studies that accompany breast cancer patients from diagnosis to treatment outcome, analysing barriers to start of treatment, treatment adherence and clinical and patient reported outcomes are needed to describe in detail the situation of diagnosed breast cancer patients in sub-Saharan Africa and to highlight where public health interventions are needed most. The impact of the disease on the socioeconomic situation of the patient and her family needs to be studied and should be more the focus when discussing strategies to improve breast cancer early detection, treatment and survival in sub-Saharan Africa. Studies focusing on the macro level, for example, international relations, trade and laws, could be helpful to highlight barriers in providing care for breast cancer patients in sub-Saharan countries.