Abstract
Purpose
In this mixed-methods study, we evaluated the factors that contribute to delayed breast cancer (BC) diagnosis and treatment at a Kenyan hospital.
Methods
Individuals with a diagnosis of BC, either as a referral or index patient, were recruited to participate in this study through convenience sampling. Data were collected on sociodemographics, health history, and cancer history, diagnosis, and treatment of patients at Kenyatta National Hospital (KNH). For the quantitative analyses, the relationship between sociodemographic and health history factors with stage at diagnosis, number of visits before diagnosis, time to diagnosis, and time to initial intervention, stratified by time to onset of symptoms, were examined using regression analyses. For the qualitative analysis, in-depth interviews of every fifth patient were completed to assess reasons for delayed diagnosis and treatment.
Results
The final analytic sample comprised of 378 female BC patients with an average age of 50. These females were generally of lower SES: 49.2% attained no or only primary-level education, 57.4% were unemployed, and the majority (74.6%) had a monthly household income of < 5000 Kenyan shillings (equivalent to ~ $41 USD). The median time from BC symptom onset to presentation at KNH was 13 (IQR = 3–36) weeks, from presentation to diagnosis was 17.5 (IQR = 7–36.5) weeks, and from diagnosis to receipt of the initial intervention was 6 (IQR = 3–13) weeks. Female BC patients who were never/unmarried, less educated, less affluent, users of hormonal contraception, and had ≥ 3 children were more likely to experience diagnosis and treatment delays. Qualitative data showed that financial constraints, lack of patient BC awareness, and healthcare practitioner misdiagnosis and/or strikes delayed patient diagnosis and treatment.
Conclusions
BC patients experience long healthcare system delays before diagnosis and treatment. Educating communities and providers about BC and expediting referrals may minimize such delays and subsequently BC mortality rates in Kenya.
Keywords: Delays, Diagnosis, Outcomes, Stages, Presentation
Introduction
Breast cancer (BC) is the most common cancer and leading cause of cancer death among women globally [1]. While breast cancer incidence rates in low- and middle-income countries (LMICs) are lower than in high-income countries (HICs), they are on the rise, particularly in Sub-Saharan Africa [2]. Women diagnosed with BC in Sub-Saharan Africa overwhelmingly present at later stages of disease [3, 4]. Later-stage BC diagnosis is associated with patient-related factors, including sociodemographics (age, marital status, ethnicity, education level, and income), familial and community factors (culture, social pressure, support networks), health beliefs (breast cancer awareness, belief in spiritual and traditional healers), and psychosocial factors (fear of BC treatment, denial) [5]. Provider-related factors (i.e., lack of BC awareness among healthcare professionals, lack of patient-centred care) and health system-related factors (i.e., limited availability of resources, geographic access issues, delayed referrals, lack of coordinated care) also play a role [6].
In 2020, approximately 6799 incident cases and 3,107 BC deaths were reported among women in Kenya. Age-standardized incidence rates in Kenya have steadily increased from 1990 (13.9 per 100,000) to 2019 (23.0 per 100,000), as observed elsewhere in Sub-Saharan Africa. The combination of delayed BC care due to presentation delay and diagnostic and treatment delay–with the latter frequently attributable to provider- or healthcare system-related factors–is likely a reason for Kenya’s rising BC mortality rates [2]. An audit into the breast clinic at Kenyatta National Hospital (KNH)–one of three national referral and teaching hospitals in Kenya hosting all BC treatment modalities, including surgery, chemotherapy, and radiotherapy–revealed an average 2.1-month delay in treatment, primarily due to diagnostic delays [6]. In the current study, we aimed to investigate the causes of presentation delay, diagnostic delay, and treatment delay among patients receiving care at KNH.
Methods
Study design, participants, and setting
From January to December 2018, this mixed-methods study was conducted at KNH’s breast cancer clinic, haemato-oncology clinic, cancer treatment centre, oncology wards, and surgical wards, where patients travel from across the country to receive specialized care. Every week, each breast clinic and haemato-oncologic clinic sees approximately sixty patients with breast pathology.
For the quantitative portion of the study, participants ≥ 18 years old with histopathologically confirmed BC were consecutively sampled, provided informed consent, and enrolled. Ineligible participants included those with a diagnosis of ductal carcinoma in situ (DCIS) and those who refused to consent. Data were collected on 386 males and females with BC. Due to small sample size, males were excluded from the current analysis, leaving an analytic sample of 378 females with BC. For the qualitative part of the study, an in-depth interview of every fifth patient was performed by trained medical students at KNH. Responses from these patients, who provided consent, were recorded, and transcribed using NVivo (QSR International, Inc.). Inductive content analysis was conducted until thematic saturation was reached.
Measures
For the quantitative analysis, independent variables of interest included sociodemographic factors (age in years, region of origin [Central, Coast/Upper Eastern, Lower Eastern, Nairobi, Nyanza, Rift Valley, Western], marital status [married, divorced/separated/widowed, never married], education level [lower (none, primary), higher secondary, tertiary], and income in Kenyan shillings [< 5000, 5000–10000, > 10000] and, health history factors (alcohol use [no, yes], age at menarche in years, use of hormonal contraceptives [no, yes], and parity [< 3, ≥ 3]. The sole confounder was smoking status (no, yes). Additional sociodemographic and health history characteristics, including employment status (employed, unemployed), history of diabetes (no, yes), age at first delivery in years, duration of hormonal contraceptive use (1, 1–2, 3–5, > 5 years), use of hormonal replacement therapy (no, yes), and history of breastfeeding (no, yes), were used to characterize the sample.
Outcomes of interest were number of visits before diagnosis, time to diagnosis in weeks (from presentation to health facility to diagnosis), stage at diagnosis (early stage [I, II], late stage [III, IV]), and time to initial intervention in weeks (from diagnosis to start of treatment). Associations between independent variables and these outcomes were hypothesized to vary by patients’ time of symptom onset to presentation at the health facility (≤ 1 month, > 1 month to 5 months, > 5 months). Other BC-related variables, for which information was collected and available for analysis, included family history of cancer (no, yes), mode of histopathological diagnosis (core biopsy, excisional biopsy, incisional biopsy, fine needle aspiration, core biopsy and fine needle aspiration, excisional biopsy, and fine needle aspiration, other), and time to other treatments/intervention modalities (surgery, chemotherapy, hormonal therapy, radiotherapy, immunotherapy) in weeks. A schematic diagram of BC management stages and flow within the KNH is shown in Fig. 1.
Fig. 1.
Time to different stages of breast cancer management at a Kenyan referral hospital. The illustrated diagram depicts the breast cancer management process at a referral hospital in Kenya. Next to each arrow is the median (interquartile range) time it takes for a patient to go through each stage, in weeks
Data analysis
Medians (accompanied by interquartile range [IQR]) and frequencies (accompanied by proportions) were used to summarize descriptive statistics for continuous and categorical variables, respectively. Independent linear regression models were run using robust standard errors to evaluate associations between independent variables and three continuous outcomes (number of visits before diagnosis, time to diagnosis, and time to initial intervention), which were log-transformed first because they had right-skewed distributions. Then, exponentiated betas were reported from these model results for ease of interpretation. To evaluate associations between independent variables of interest and stage at diagnosis (binary outcome), we used relative risk (RR) regression, by running the generalized linear model procedure with a Poisson distribution, log link, and robust error variances. Regression analyses for all four outcomes were stratified by time of symptom onset. Covariates included in each model were age, region of origin, marital status, education level, income, alcohol use, age at menarche, use of hormonal contraceptives, parity, and smoking status. For the model where time to initial intervention was the outcome, time to diagnosis was included as an additional covariate to adjust for. Altogether, we tested 36 hypotheses across nine predictors and four outcomes—therefore, to correct for multiple comparisons at an α of 0.05, we used the Benjamini-Hochberg (BH) procedure. In the supplementary analysis, we also performed linear regression analyses to determine which sociodemographic and health history characteristics were associated with time from initial intervention to receipt of additional modes of intervention, such as surgery, chemotherapy, hormonal therapy, and radiotherapy. Immunotherapy was not examined, as too few patients (5%) received this treatment modality. All analyses were performed using SPSS (Version 26.0, Chicago, IL).
Results
Participant characteristics
Of the 378 female BC patients in this study, who had a median age of 50 (range: 25–88), most were native to the Central region of Kenya (50.3%), were married (62.2%), and were primary school graduates (43.7%) (Table 1). Over half were unemployed (57.4%) and nearly three-quarters (74.9%) had household incomes of < 5000 Kenyan shillings (KES)/month. There were very few patients who reported smoking (2.9%) and having diabetes (7.7%), but alcohol use was more prevalent (18.3%). Patients reported a median age at menarche of 15 (IQR = 14–16) years. Over three-quarters (77.2%) used hormonal contraception, with the majority (50.3%) using them for more than five years. Use of hormonal replacement therapy was less common (10.6%). Also, most patients disclosed that they gave birth to fewer than three children (56.9%), if any. The median age at first delivery was 21 (IQR = 19–25) years and 93.4% of patients had a history of breastfeeding.
Table 1.
Characteristics of Female Breast Cancer Patients, N = 378
| Characteristics Sociodemographics |
Sample N (%) | Missing n(%) |
|---|---|---|
| Age (years), median (IQR) | 50.0 (43.0–57.0) | 0 (0.0) |
| Region of origin | 0 (0.0) | |
| Central | 190 (50.3) | |
| Coast | 5 (1.3) | |
| Upper eastern | 34 (9.0) | |
| Lower eastern | 51 (13.5) | |
| Nairobi | 24 (6.3) | |
| Nyanza | 28 (7.4) | |
| Rift valley | 26 (6.9) | |
| Western | 20 (5.3) | |
| Marital status | 1 (0.3) | |
| Married | 235 (62.2) | |
| Divorced/Separated/Widowed | 83 (22.0) | |
| Never married | 59 (15.6) | |
| Education level | 0 (0.0) | |
| Lower | 186 (49.2) | |
| None | 21 (5.6) | |
| Primary | 165 (43.7) | |
| Higher | 192 (50.8) | |
| Secondary | 138 (36.5) | |
| Tertiary | 54 (14.3) | |
| Employment status | 0 (0.0) | |
| Employed | 161 (42.6) | |
| Unemployed | 217 (57.4) | |
| Income per month (Kenyan shilling) | 0 (0.0) | |
| < 5000 | 283 (74.9) | |
| 5000–10000 | 41 (10.8) | |
| > 10,000 | 54 (14.3) | |
|
Health History | ||
| Smoking status | 0 (0.0) | |
| No | 367 (97.1) | |
| Yes | 11 (2.9) | |
| Alcohol use | 0 (0.0) | |
| No | 309 (81.7) | |
| Yes | 69 (18.3) | |
| Diabetes | 5 (1.3) | |
| No | 344 (91.0) | |
| Yes | 29 (7.7) | |
| Age at menarche (years), median (IQR) | 15.0 (14.0–16.0) | 20 (5.3) |
| Use of hormonal contraceptives | 1 (0.3) | |
| No | 85 (22.5) | |
| Yes | 292 (77.2) | |
| Duration of use (years) | ||
| < 1 | 22 (5.8) | |
| 1–2 | 39 (10.3) | |
| 3–5 | 41 (10.8) | |
| > 5 | 190 (50.3) | |
| Hormonal replacement therapy | 9 (2.4) | |
| No | 329 (87.0) | |
| Yes | 40 (10.6) | |
| Parity | 0 (0.0) | |
| < 3 | 215 (56.9) | |
| ≥ 3 | 163 (43.1) | |
| Age at first delivery (years), median (IQR) | 21.0 (19.0–25.0) | 20 (5.3) |
| History of breastfeeding | 0 (0.0) | |
| No | 25 (6.6) | |
| Yes | 353 (93.4) | |
|
Cancer History, Diagnosis, and Treatment | ||
| Family history of cancer | 0 (0.0) | |
| Yes | 129 (34.1) | |
| No | 249 (65.9) | |
| Time of onset of symptoms (weeks), median (IQR) | 13.0 (3.0–36.0) | 2 (0.5) |
| ≤ 1 month | 124 (32.8) | |
| > 1 month to 5 months | 106 (28.0) | |
| > 5 months | 146 (38.6) | |
| Number of visits before diagnosis, median (IQR) | 5.0 (3.0–6.0) | 2 (0.5) |
| Time to diagnosis (weeks), median (IQR) | 17.5 (7.0–36.5) | 0 (0.0) |
| Mode of histopathological diagnosis | 16 (4.2) | |
| Core biopsy | 250 (66.1) | |
| Excisional biopsy | 29 (7.7) | |
| Incisional biopsy | 5 (1.3) | |
| Fine needle aspiration | 14 (3.7) | |
| Core biopsy & fine needle aspiration | 42 (11.1) | |
| Excisional biopsy & fine needle aspiration | 7 (1.9) | |
| Other | 15 (4.0) | |
| Stage at diagnosis | 5 (1.3) | |
| Early stage | 156 (41.3) | |
| Stage I | 29 (7.7) | |
| Stage II | 127 (33.6) | |
| Late stage | 159 (42.1) | |
| Stage III | 100 (26.5) | |
| Stage IV | 59 (15.6) | |
| Unknown | 58 (15.3) | |
| Time to initial intervention (weeks), median (IQR) Time to other intervention modes (weeks), median (IQR) |
6.0 (3.0–13.0) | 8 (2.1) |
| Time to other intervention modes (weeks), median (IQR) | ||
| Surgery | 8.0 (3.0–26.0) | 150 (39.7) |
| Chemotherapy | 11.0 (6.0–20.0) | 66 (17.5) |
| Hormonal therapy | 42.0 (28.0–54.3) | 284 (75.1) |
| Radiotherapy | 42.0 (28.0–55.0) | 243 (64.3) |
| Immunotherapy | 40.0 (21.0–56.0) | 359 (95.0) |
IQR interquartile range
Approximately one-third of participants (34.1%) reported having a family history of cancer. The median time to onset of BC symptoms was 13 (IQR = 3–36) weeks, whereby 32.8% experienced symptoms for ≤ 1 month, 28.0% for > 1 month to 5 months, and 38.6% for > 5 months. The median number of visits before diagnosis was 5 (IQR = 3–6) and the median time to diagnosis was 17.5 (IQR = 7–36.5) weeks. The primary mode of histopathological diagnosis was core biopsy (66.1%), and most patients were diagnosed at stage II (33.6%). The median time from diagnosis to receipt of the initial intervention was 6 (IQR = 3–13) weeks. Following the initial intervention, the median time to receipt of other treatment modalities were as follows: surgery, 8 (IQR = 3–26) weeks (among 60.3% of patients who received surgery); chemotherapy, 11 (IQR = 6–20) weeks (among the 82.5% who received chemotherapy); hormonal therapy, 42 (IQR = 28–54.3) weeks (among the 24.9% who received hormonal therapy); radiotherapy, 42 (IQR = 28–55) weeks (among the 35.7% who received radiotherapy); and immunotherapy, 40 (IQR = 21–56) weeks (among the 5.0% who received immunotherapy).
Association between sociodemographic and health history factors with breast cancer management
Among patients who experienced symptoms for ≤ 1 month, we observed significant associations between several sociodemographic and health history factors with all four outcomes of interest, even after correction for multiple comparisons (Table 2). Compared to patients who originally came from the Central region of Kenya, the number of visits before cancer diagnosis was 40% greater among those who were native to the Western region (1.40, 95% CI: 1.17–1.67; P < 0.001), time to diagnosis was 95% longer among those who came from the Coast/Upper Eastern region (1.95, 95% CI 1.10–3.46; P = 0.022), and time to initial intervention was 170% longer for those who originated from the Rift valley (2.70 95% CI: 1.47–4.94; P = 0.001). Never married patients had about 107% longer time to diagnosis compared to married patients (2.07, 95% CI 1.18–3.63; P = 0.011). Compared to patients with secondary/tertiary education, the time to receipt of the initial intervention was 46% longer among those with no/primary-level education (1.46, 95% CI 1.08–1.98; P = 0.015). The likelihood of late- versus early-stage diagnosis was 22% lower among patients with families earning 5000–10000 KES/month relative to < 5000 KES/month (0.78, 95% CI 0.63–0.95; P = 0.013). With regards to associated health history factors, the number of visits before BC diagnosis was 23% lower among those who reported alcohol use (0.77, 95% CI 0.62–0.95; P = 0.016). On the other hand, patients who used hormonal contraceptives had 56% longer time to initial intervention (1.56, 95% CI 1.04–2.33; P = 0.032). Additionally, having ≥ 3 children was associated with 41% higher number of visits before cancer diagnosis (1.41, 95% CI 1.08–1.84; P = 0.011) and 84% longer the time to diagnosis (1.84, 95% CI 1.15–2.94; P = 0.011) when compared to having < 3 children.
Table 2.
Sociodemographic and health history characteristics associated with stage of cancer diagnosis, number of visits before diagnosis, time to diagnosis in weeks, and time to initial intervention in weeks, among patients who experienced symptoms for ≤ 1 month prior to presentation at health facility
| Characteristics | Late vs. Early Stage Cancer at Diagnosis†
N = 104 |
Number of Visits before Cancer Diagnosis††
N = 120 |
Time to Diagnosis††
N = 120 |
Time to Initial Intervention††b
N = 118 |
||||
|---|---|---|---|---|---|---|---|---|
| Sociodemographics | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a |
| Age (years) | 0.99 (0.98–1.00) | 0.006* | 0.99 (0.97–1.00) | 0.026* | 1.00 (0.98–1.03) | 0.796 | 1.00 (0.99–1.02) | 0.681 |
| Region of origin | ||||||||
| Central | ref | ref | ref | ref | ||||
| Coast/Upper Easterna | 1.06 (0.87–1.29) | 0.584 | 1.19 (0.84–1.67) | 0.326 | 1.95 (1.10–3.46) | 0.022* | 0.80 (0.48–1.33) | 0.385 |
| Lower Eastern | 1.03 (0.83–1.27) | 0.791 | 0.76 (0.57–1.02) | 0.068 | 0.87 (0.50–1.51) | 0.625 | 1.29 (0.81–2.05) | 0.278 |
| Nairobi | 1.21 (0.92–1.59) | 0.178 | 1.23 (0.71–2.11) | 0.459 | 1.53 (0.51–4.56) | 0.443 | 0.63 (0.35–1.16) | 0.137 |
| Nyanza | 0.88 (0.67–1.16) | 0.372 | 0.77 (0.56–1.06) | 0.111 | 0.73 (0.45–1.18) | 0.201 | 0.79 (0.49–1.28) | 0.345 |
| Rift valley | 1.14 (0.94–1.36) | 0.178 | 0.83 (0.64–1.08) | 0.165 | 0.76 (0.41–1.40) | 0.374 | 2.70 (1.47–4.94) | 0.001* |
| Western | 1.09 (0.86–1.37) | 0.483 | 1.40 (1.17–1.67) | < 0.001* | 0.62 (0.38–1.01) | 0.055 | 1.43 (0.78–2.60) | 0.247 |
| Marital status | ||||||||
| Married | ref | ref | ref | ref | ||||
| Divorced/Separated/Widowed | 1.08 (0.92–1.27) | 0.325 | 0.90 (0.74–1.09) | 0.283 | 1.30 (0.83–2.02) | 0.249 | 0.84 (0.58–1.22) | 0.356 |
| Never married | 0.94 (0.75–1.17) | 0.560 | 1.07 (0.79–1.45) | 0.669 | 2.07 (1.18–3.63) | 0.011* | 0.95 (0.62–1.45) | 0.822 |
| Education level | ||||||||
| Higher | ref | ref | ref | ref | ||||
| Lower | 0.93 (0.81–1.07) | 0.309 | 1.02 (0.85–1.23) | 0.811 | 1.03 (0.72–1.48) | 0.853 | 1.46 (1.08–1.98) | 0.015* |
| Income per month (Kenyan shilling) | ||||||||
| < 5000 | ref | ref | ref | ref | ||||
| 5000–10000 | 0.78 (0.63–0.95) | 0.013* | 0.84 (0.67–1.05) | 0.135 | 0.63 (0.38–1.04) | 0.072 | 0.90 (0.60–1.36) | 0.630 |
| > 10,000 | 0.89 (0.73–1.08) | 0.226 | 1.05 (0.81–1.37) | 0.695 | 0.89 (0.57–1.39) | 0.613 | 1.00 (0.62–1.59) | 0.989 |
| HEALTH HISTORY | ||||||||
| Alcohol use | ||||||||
| No | ref | ref | ref | ref | ||||
| Yes | 1.05 (0.92–1.21) | 0.463 | 0.77 (0.62–0.95) | 0.016* | 1.06 (0.64–1.76) | 0.807 | 0.78 (0.52–1.15) | 0.208 |
| Age at menarche (years) | 0.98 (0.94–1.01) | 0.207 | 0.98 (0.92–1.05) | 0.579 | 0.86 (0.75–0.99) | 0.034* | 1.01 (0.92–1.10) | 0.833 |
| Use of hormonal contraceptives | ||||||||
| No | ref | ref | ref | ref | ||||
| Yes | 1.05 (0.87–1.27) | 0.613 | 0.74 (0.54–1.00) | 0.047 | 0.97 (0.60–1.57) | 0.907 | 1.56 (1.04–2.33) | 0.032* |
| Parity | ||||||||
| < 3 | ref | ref | ref | ref | ||||
| ≥ 3 | 1.09 (0.90–1.31) | 0.372 | 1.41 (1.08–1.84) | 0.011* | 1.84 (1.15–2.94) | 0.011* | 0.96 (0.65–1.43) | 0.852 |
All regression analyses were performed in female patients who experienced breast cancer symptoms for ≤ 1 month prior to presentation at health facility. All models adjusted for smoking status
Relative Risk Regression was conducted using a Poisson distribution, log link, and robust error variances to obtain RRs that examined the association between sociodemographic and health history factors with stage at diagnosis. The outcome, stage at diagnosis, was modeled as a binary covariate, where early stage (Stage I or II) was the referent category
Linear regression was conducted using a normal distribution, log link, and robust error variances to determine the relationship between sociodemographic and health history factors with number of visits before cancer diagnosis, time to diagnosis, and time to initial intervention. All outcome covariates were continuous and log-transformed to ensure they had a normal distribution–therefore, exponentiated betas are reported in the table to allow for ease of interpretation
After correction for multiple comparisons using the Benjamini–Hochberg procedure, significant associations were denoted with an asterisk (*)
Model includes all covariates shown in the table, smoking status, and time to diagnosis
Among BC patients who experienced symptoms for < 1 to 5 months, of which there were fewer in number, we did not observe any statistically significant associations between health history factors and any of the four outcomes (Table 3). However, some significant associations were detected for the sociodemographic factors. After correction for multiple comparisons, we found that compared to those who were native to Central Kenya, the likelihood of late- versus early-stage cancer diagnosis was 36% higher for patients who came from the Lower Eastern region (1.36, 95% CI 1.12–1.64; P = 0.002), 25% higher among patients originating from Nyanza (1.25, 95% CI 1.04–1.49; P = 0.015), and 62% higher among patients native to the Rift Valley (1.62, 95% CI 1.35–1.96; P < 0.001). Also, the time to diagnosis was 64% shorter for patients who came from the Coast/Upper Eastern region (0.36, 95% CI 0.16–0.78; P = 0.010), and 43% shorter among patients who came from the Western region (0.57, 95% CI 0.38–0.85; P = 0.006) relative to those coming from the Central region. The likelihood of late- versus early-stage cancer diagnosis was 21% lower for never married patients (0.79, 95% CI 0.64–0.97; P = 0.025) and divorced/separated/widowed patients had an approximately 36% greater number of visits before BC diagnosis (1.36, 95% CI 1.01–1.83; P = 0.042), both in comparison to married patients. Finally, the likelihood of late- versus early-stage cancer diagnosis was 21% higher among patients whose family earned > 10,000 KES/month compared to < 5000 KES/month (1.21, 95% CI 1.02–1.44; P = 0.028).
Table 3.
Sociodemographic and health history characteristics associated with stage of cancer diagnosis, number of visits before diagnosis, time to diagnosis in weeks, and time to initial intervention in weeks, among patients who experienced symptoms for > 1 month to 5 months prior to presentation at health facility
| CHARACTERISTICS | Late vs. Early Stage Cancer at Diagnosis†
N = 83 |
Number of Visits before Cancer Diagnosis††
N = 101 |
Time to Diagnosis††
N = 101 |
Time to Initial Intervention††b
N = 99 |
||||
|---|---|---|---|---|---|---|---|---|
| Sociodemographics | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a |
| Age (years) | 0.99 (0.99–1.00) | 0.234 | 0.99 (0.98–1.01) | 0.392 | 0.99 (0.97–1.01) | 0.356 | 1.02 (0.99–1.04) | 0.166 |
| Region of origin | ||||||||
| Central | ref | ref | ref | ref | ||||
| Coast/Upper Eastern | 0.87 (0.68–1.10) | 0.243 | 0.90 (0.64–1.26) | 0.544 | 0.36 (0.16–0.78) | 0.010* | 1.28 (0.71–2.31) | 0.416 |
| Lower Eastern | 1.36 (1.12–1.64) | 0.002* | 0.81 (0.63–1.05) | 0.111 | 1.00 (0.55–1.79) | 0.987 | 1.53 (0.86–2.73) | 0.152 |
| Nairobi | 1.27 (0.96–1.67) | 0.091 | 1.13 (0.58–2.18) | 0.717 | 2.30 (0.94–5.64) | 0.068 | 1.22 (0.61–2.47) | 0.572 |
| Nyanza | 1.25 (1.04–1.49) | 0.015* | 0.79 (0.58–1.07) | 0.127 | 1.24 (0.61–2.51) | 0.551 | 0.96 (0.53–1.72) | 0.886 |
| Rift valley | 1.62 (1.35–1.96) | < 0.001* | 1.11 (0.78–1.58) | 0.553 | 1.16 (0.62–2.17) | 0.633 | 0.63 (0.35–1.16) | 0.137 |
| Western | 1.04 (0.80–1.36) | 0.769 | 1.07 (0.63–1.82) | 0.802 | 0.57 (0.38–0.85) | 0.006* | 1.48 (0.76–2.88) | 0.243 |
| Marital status | ||||||||
| Married | ref | ref | ref | ref | ||||
| Divorced/Separated/Widowed | 0.96 (0.80–1.16) | 0.688 | 1.36 (1.01–1.83) | 0.042* | 1.54 (0.96–2.47) | 0.074 | 1.38 (0.78–2.44) | 0.271 |
| Never married | 0.79 (0.64–0.97) | 0.025* | 1.04 (0.68–1.60) | 0.848 | 0.75 (0.37–1.51) | 0.422 | 1.67 (0.98–2.83) | 0.058 |
| Education level | ||||||||
| Higher | ref | ref | ref | ref | ||||
| Lower | 1.11 (0.96–1.29) | 0.153 | 1.12 (0.91–1.38) | 0.294 | 1.42 (0.96–2.12) | 0.081 | 0.79 (0.57–1.10) | 0.159 |
| Income per month (Kenyan shilling) | ||||||||
| < 5000 | ref | ref | ref | ref | ||||
| 5000–10000 | 1.24 (0.94–1.65) | 0.130 | 0.92 (0.63–1.35) | 0.681 | 1.24 (0.67–2.27) | 0.490 | 1.32 (0.67–2.59) | 0.420 |
| > 10,000 | 1.21 (1.02–1.44) | 0.028* | 1.06 (0.80–1.41) | 0.686 | 1.29 (0.81–2.04) | 0.282 | 0.85 (0.58–1.26) | 0.428 |
| HEALTH HISTORY | ||||||||
| Alcohol use | ||||||||
| No | ref | ref | ref | ref | ||||
| Yes | 1.19 (0.97–1.45) | 0.092 | 0.95 (0.64–1.39) | 0.784 | 0.82 (0.42–1.60) | 0.564 | 1.22 (0.76–1.94) | 0.407 |
| Age at menarche (years) | 0.99 (0.97–1.02) | 0.675 | 1.01 (0.97–1.05) | 0.710 | 1.05 (0.96–1.16) | 0.281 | 0.98 (0.91–1.05) | 0.491 |
| Use of hormonal contraceptives | ||||||||
| No | ref | ref | ref | ref | ||||
| Yes | 0.86 (0.72–1.04) | 0.116 | 1.27 (0.95–1.71) | 0.109 | 1.10 (0.69–1.76) | 0.687 | 1.59 (1.01–2.48) | 0.044 |
| Parity | ||||||||
| < 3 | ref | ref | ref | ref | ||||
| ≥ 3 | 0.94 (0.77–1.15) | 0.550 | 0.90 (0.68–1.19) | 0.458 | 0.76 (0.47–1.23) | 0.267 | 1.11 (0.69–1.78) | 0.660 |
All regression analyses were performed in female patients who experienced breast cancer symptoms for > 1 month to 5 months prior to presentation at health facility. All models adjusted for smoking status
Relative Risk Regression was conducted using a Poisson distribution, log link, and robust error variances to obtain RRs that examined the association between sociodemographic and health history factors with stage at diagnosis. The outcome, stage at diagnosis, was modeled as a binary covariate, where early stage (Stage I or II) was the referent category
Linear regression was conducted using a normal distribution, log link, and robust error variances to determine the relationship between sociodemographic and health history factors with number of visits before cancer diagnosis, time to diagnosis, and time to initial intervention. All outcome covariates were continuous and log-transformed to ensure they had a normal distribution–therefore, exponentiated betas are reported in the table to allow for ease of interpretation
After correction for multiple comparisons using the Benjamini–Hochberg procedure, significant associations were denoted with an asterisk (*)
Model includes all covariates shown in the table, smoking status, and time to diagnosis
After correction for multiple comparisons, among patients who experienced symptoms for > 5 months, several significant factors were found to be associated with BC management. As shown in Table 4, the only factors not associated with any of the outcomes in this group were age and use of hormonal contraceptives. For region of origin, compared to patients who came from Central Kenya, the likelihood of late- versus early-stage cancer diagnosis was 23% lower among those who came from Nairobi (0.77, 95% CI 0.61–0.97; P = 0.028) and 28% lower among those who came from Nyanza (0.72, 95% CI 0.57–0.92; P = 0.008). The time to diagnosis was 195% longer among those who came from the Coast/Upper Eastern region (2.95, 95% CI 1.38–6.33; P = 0.005) and 230% longer among those who came from the Western region (3.30, 95% CI 1.76–6.17; P < 0.001) when compared to those who were native to Central Kenya. Divorced/separated/widowed patients had 44% greater number of visits before cancer diagnosis in comparison to married patients (1.44, 95% CI 1.15–1.80; P = 0.001). The likelihood of late- versus early-stage cancer diagnosis was 14% lower among those with no/only primary education as opposed to secondary/tertiary education (0.86, 95% CI 0.76–0.97; P = 0.012). The time to initial intervention was also 50% lower for patients with a family income of > 10,000 KES/month compared to < 5000 KES/month (0.50, 95% CI 0.33–0.75; P = 0.005). With regards to associated health history factors, the likelihood of late- versus early-stage cancer diagnosis was 24% higher (1.24, 95% CI 1.08–1.43; P = 0.002) among those who reported alcohol use. For BC patients who had ≥ 3 children relative to those with < 3 children, the number of visits before cancer diagnosis was 17% higher (1.17, 95% CI 1.03–1.33; P = 0.019).
Table 4.
Sociodemographic and health history characteristics associated with stage of breast cancer diagnosis, number of visits before diagnosis, time to diagnosis in weeks, and time to initial intervention in weeks, among patients who experienced symptoms for > 5 months prior to presentation at health facility
| Characteristics | Late vs. Early Stage Cancer at Diagnosis†
N = 107 |
Number of Visits before Cancer Diagnosis††
N = 132 |
Time to Diagnosis††
N = 134 |
Time to Initial Intervention††b
N = 130 |
||||
|---|---|---|---|---|---|---|---|---|
| Sociodemographics | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a | exp(β) (95% CI) | P a |
| Age (years) | 1.00 (0.99–1.00) | 0.557 | 1.00 (0.99–1.01) | 0.889 | 1.01 (0.99–1.03) | 0.414 | 0.99 (0.98–1.01) | 0.513 |
| Region of origin | ||||||||
| Central | ref | ref | ref | ref | ||||
| Coast/Upper Eastern | 0.86 (0.70–1.07) | 0.183 | 1.09 (0.87–1.37) | 0.457 | 2.95 (1.38–6.33) | 0.005* | 0.72 (0.45–1.16) | 0.177 |
| Lower Eastern | 1.09 (0.94–1.28) | 0.258 | 1.18 (0.86–1.61) | 0.297 | 1.64 (0.91–2.96) | 0.102 | 0.65 (0.41–1.03) | 0.066 |
| Nairobi | 0.77 (0.61–0.97) | 0.028* | 1.04 (0.79–1.37) | 0.772 | 1.79 (0.78–4.11) | 0.167 | 0.96 (0.44–2.09) | 0.912 |
| Nyanza | 0.72 (0.57–0.92) | 0.008* | 0.89 (0.67–1.20) | 0.455 | 1.36 (0.62–2.99) | 0.445 | 1.12 (0.68–1.86) | 0.658 |
| Rift valley | 1.05 (0.89–1.25) | 0.572 | 0.89 (0.56–1.41) | 0.619 | 1.64 (0.59–4.57) | 0.344 | 0.95 (0.41–2.24) | 0.915 |
| Western | 0.82 (0.64–1.05) | 0.120 | 1.38 (0.96–2.00) | 0.084 | 3.30 (1.76–6.17) | < 0.001* | 1.18 (0.72–1.93) | 0.508 |
| Marital status | ||||||||
| Married | ref | ref | ref | ref | ||||
| Divorced/Separated/Widowed | 0.95 (0.82–1.10) | 0.496 | 1.44 (1.15–1.80) | 0.001* | 1.62 (0.96–2.76) | 0.073 | 0.99 (0.64–1.52) | 0.963 |
| Never married | 1.01 (0.87–1.17) | 0.925 | 0.95 (0.79–1.15) | 0.616 | 0.60 (0.35–1.02) | 0.060 | 1.10 (0.68–1.78) | 0.709 |
| Education level | ||||||||
| Higher | ref | ref | ref | ref | ||||
| Lower | 0.86 (0.76–0.97) | 0.012* | 0.95 (0.80–1.13) | 0.560 | 1.06 (0.66–1.69) | 0.819 | 1.04 (0.73–1.47) | 0.830 |
| Income per month (Kenyan shilling) | ||||||||
| < 5000 | ref | ref | ref | ref | ||||
| 5000–10000 | 1.03 (0.89–1.19) | 0.686 | 1.04 (0.80–1.35) | 0.763 | 1.18 (0.52–2.70) | 0.693 | 1.03 (0.63–1.68) | 0.919 |
| > 10,000 | 0.92 (0.77–1.11) | 0.381 | 0.87 (0.63–1.19) | 0.370 | 0.64 (0.36–1.14) | 0.130 | 0.50 (0.33–0.75) | 0.001* |
| HEALTH HISTORY | ||||||||
| Alcohol use | ||||||||
| No | ref | ref | ref | ref | ||||
| Yes | 1.24 (1.08–1.43) | 0.002* | 0.99 (0.77–1.27) | 0.939 | 0.71 (0.42–1.20) | 0.203 | 1.10 (0.67–1.81) | 0.699 |
| Age at menarche (years) | 0.98 (0.94–1.01) | 0.222 | 0.99 (0.94–1.04) | 0.660 | 0.91 (0.81–1.03) | 0.152 | 1.14 (1.05–1.25) | 0.003* |
| Use of hormonal contraceptives | ||||||||
| No | ref | ref | ref | ref | ||||
| Yes | 0.89 (0.78–1.02) | 0.099 | 0.98 (0.82–1.17) | 0.817 | 1.29 (0.82–2.04) | 0.265 | 1.21 (0.82–1.77) | 0.339 |
| Parity | ||||||||
| < 3 | ref | ref | ref | ref | ||||
| ≥ 3 | 1.17 (1.03–1.33) | 0.019* | 1.00 (0.83–1.21) | 0.982 | 0.75 (0.46–1.20) | 0.226 | 0.92 (0.65–1.32) | 0.661 |
All regression analyses were performed in female patients who experienced breast cancer symptoms for > 5 months prior to presentation at health facility. All models adjusted for smoking status
Relative Risk Regression was conducted using a Poisson distribution, log link, and robust error variances to obtain RRs that examined the association between sociodemographic and health history factors with stage at diagnosis. The outcome, stage at diagnosis, was modeled as a binary covariate, where early stage (Stage I or II) was the referent category
Linear regression was conducted using a normal distribution, log link, and robust error variances to determine the relationship between sociodemographic and health history factors with number of visits before cancer diagnosis, time to diagnosis, and time to initial intervention. All outcome covariates were continuous and log-transformed to ensure they had a normal distribution–therefore, exponentiated betas are reported in the table to allow for ease of interpretation
After correction for multiple comparisons using the Benjamini–Hochberg procedure, significant associations were denoted with an asterisk (*)
Model includes all covariates shown in the table, smoking status, and time to diagnosis
Qualitative feedback from BC patients
From our inductive analysis, we observed two major thematic areas with a possible central shared issue. The thematic areas were health system-related and patient-related factors as key contributors to delays in the continuum of BC care (Table 5). From participants’ perspective, the majority of healthcare system issues related to misdiagnosis and misinformation on the part of health care providers.
Table 5.
Distribution of codes on themes and quotes
| Theme | Codes | Quotes |
|---|---|---|
| Health system-related factors | Delayed decision making | “In one test, they said that there was nothing they could detect/observe. In the second test, they said that they could (see something). When I returned those results, they said it was unclear because one test was reporting that there was nothing and the other one was showing something. So, they sent me again to do more tests.” |
| Misdiagnosis and misinformation | “I used to do reviews and they told me there was nothing to worry about. But even when staying put, I felt the lump was still increasing in size, so I went back.” | |
| Inadequate resources | “No, it was the medicine that would run out here in KNH. So, it could pose a challenge.” | |
| Inadequate human resources | “So, I stayed like three months without starting treatment and getting a doctor.” | |
| Long turnaround time | “They told me it would take three weeks but now you see it took almost one month. I came back but they told me the results weren’t ready yet then they told me from there on it was a holiday period, so I come back on three months later.” | |
| Mistreatment by healthcare workers | “I got the result and took it back to them and they requested me to start clinic. In my mind, I was feeling—NO! I cannot go to these people because they were treating people like animals as they were taking samples. I went back home and rested.” | |
| Other medical priorities | “Afterwards Corona was hitting very hard. We were all told to go home and not come back because we could catch Corona at the clinic. I later came five months to the clinic to restart the clinic” | |
| Strike | “Because of the strike, I went to several hospitals, including Nakuru General. There was quite a while when I hung back at home.” | |
| Patient-related factors | Alternative medicine | “I took herbal medicines for one month later I noticed a lump was not reducing” |
| Denial by patient | “They told me I had cancer and I told them it was okay.” | |
| Other family priority matters | “I got pregnant and gave birth and was taking care of my child. The antenatal care personnel saw that this thing was not okay.” | |
| Fear of going to hospital | “It started a year ago and at first I was afraid to go to hospital to be told I have cancer.” | |
| Ignorance | “My problem started around 4 years ago, but it was inside when I laid on my back, when I pressed that’s when I could feel it. So, I assumed it was just something small but later I started seeing those big lumps.” | |
| Loss of hope | “And then I felt there was no need of me going ahead with the process because I thought how one experience could such? So, I gave up.” | |
| Inadequate resources | “I told them I didn’t have money and that month just passed. I got the money in December and got the x-ray done.” |
“It is because I went to hospital and was told that there wasn’t a problem and so I went home and continued with my life”.
“They said it was a fibrocystic disease of the breast”.
“I used to do reviews and they told me there was nothing to worry about. But even when staying put, I felt the lump was still increasing in size, so I went back”.
Delayed diagnosis also came from inability to make decision by being taken to variety of test and from one test to the other.
“In one test, they said that there was nothing they could detect/observe. In the second test, they said that they could (see something). When I returned those results, they said it was unclear because one test was reporting that there was nothing and the other one was showing something. So, they sent me again to do more tests”.
Discussion
The aim of this mixed-methods study was to identify factors that influence breast cancer presentation and delays in diagnosis and treatment among patients seeking care at a national referral and teaching hospital in Kenya. Among our most important findings is evidence that female BC patients seeking care at KNH experienced significant delays between symptom onset and presentation (median: 13 weeks or 3.3 months), between presentation and diagnosis (median: 17.5 weeks or 4.4 months), and between diagnosis and initial treatment (median: 6 weeks or 1.5 months). These delays were significantly more likely among BC patients who were never/unmarried, less educated, less affluent, users of hormonal contraception, and had ≥ 3 children.
Our findings are consistent with those discussed by Espina et al. (2017), a systematic review that found the average time from BC symptom recognition to diagnosis ranged from 4 to 15 months in North and Sub-Saharan Africa [2], owing to delays in both BC presentation and diagnosis. Studies in other LMICs showed comparable average time intervals from symptom recognition to diagnosis as North and Sub-Saharan Africa (i.e., 7.6 months in Brazil [7] and 5.5 months in Malaysia [8]), but HICs had significantly shorter time intervals (i.e., 1.1 months in France [9] and 1.6 months in the United States [10]). Separately, the average time to presentation and the average time to diagnosis, were also much longer in Africa than other HICs [2]. For example, time to presentation was only 0.3 months in the UK [11] and 0.5 months in Germany [12], while time to diagnosis ranged from 0.3 to 1.4 months in France, Germany, and the US [2]. Earlier studies have shown that presentation and diagnostic delay of > 3 months combined is strongly associated with late-stage BC presentation and poorer survival [13].
In Sub-Saharan Africa, of which Kenya is a part of, the average age of BC presentation ranges from 42 years in Côte d’Ivoire to 59 years in South Africa. [14–28]. There is a great deal of heterogeneity in the age at which patients present with BC in this area; we typically see the incidence begin in the early 20s and then taper off in the 50s [29], while in other countries the incidence of breast cancer predominantly averages around the 40s. Lower socioeconomic status is one of the most significant patient-related factors found to be associated with delayed BC presentation [2, 16]. Benbakhta et al. (2013) also found that residing more than 100 kms away from a health facility can affect delays as well as the number of visits before diagnosis (i.e., more than three visits is associated with delay) [16].
We observed through in-depth interviews that patient-related factors influencing delayed BC care included lack of awareness, use of alternative medicine, inability to afford healthcare costs, as well as health system-related factors such as misdiagnosis and misinformation by healthcare practitioners. By misinformation, we mean that some healthcare practitioners had provided incorrect reassurances to patients whose diagnoses had not been confirmed histologically. Other healthcare factors included insufficient resources, lengthy turnaround times, and the patient’s perception that they were mistreated by healthcare professionals. Extant literature confirms that issues around health illiteracy, a first visit to traditional healers, and the inability to afford the costs of care are reasons for delays in presentation among breast cancer patients in Sub-Saharan Africa [34–36].
Strengths and limitations
The collection of both quantitative and qualitative data for patients throughout all stages of BC management (symptom onset, presentation, diagnosis, and intervention) is a key strength of this study. In our quantitative analysis, we were able to examine how many patient-mediated factors (i.e., sociodemographics, health history) influenced delays in BC diagnosis and treatment, but not how health system-related factors also played a role. Only our qualitative analysis gave some insight into how delays were also influenced by health-system factors. As for other study limitations, the results should be interpreted in light of the fact that they were derived from a convenience sample of patients from only one hospital in Kenya. Patients were asked to remember the time to BC presentation, diagnosis, and treatment; however, this subjective form of measurement makes recall bias likely. Information on patients, including in-depth interviews, were obtained at the hospital, which is not a neutral location and may have influenced responses. However, we attempted to mitigate any biases that may have resulted from the interview setting by having medical students conduct the interviews (i.e., and not the medical practitioners themselves). In addition, as this facility is a referral hospital, enrolled patients were required to have received a BC diagnosis beforehand. As a result of the nature of referrals, most patients who sought treatment at KNH did so at the conclusion of their BC care journey. Hence, the timing of the care received could be exaggerated due to the number of facilities one needs to visit before receiving the final referral to seek care at KNH. Furthermore, the study period coincided with the COVID-19 pandemic, which may have contributed to the observed delays in diagnosis and treatment.
Conclusions
In this study, we discovered that BC patients at one medical facility in Kenya experience, on average, more than a 3-month delay between symptom onset and presentation, more than a 4-month delay between presentation and diagnosis, and a 1.5-month delay between diagnosis and initial treatment. Primarily socioeconomic status, health history factors, and geographic barriers contribute to these delays. Misdiagnoses, misinformation, and inadequate health system resources also contributed to BC treatment delays. This emphasizes the need for not only patient education, but also increased awareness among healthcare providers, which could contribute to a BC diagnosis at an earlier stage when the disease is more amenable to treatment and has better long-term outcomes.
Supplementary Material
Acknowledgements
We would like to acknowledge the Kenyatta National Hospital research unit for the research grant and the African Caribbean Cancer Consortium for the publication grant.
Funding
Kenyatta National Hospital, KNH/R&P/23 J/26/9, Daniel Kinyuru Ojuka
Footnotes
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s10549-023-07067-y.
Declarations
Conflict of interest There is no conflict of interest of authors of this article.
Data availability
Enquiries about data availability should be directed to the authors.
References
- 1.Moodley J, Cairncross L, Naiker T, Momberg M. Understanding pathways to breast cancer diagnosis among women in the Western Cape Province, South Africa: a qualitative study. BMJ Open [Internet]. 2016. Jan 4 [cited 2019 May 19];6(1). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4716174/ [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Espina C, McKenzie F, dos-Santos-Silva I (2017) Delayed presentation and diagnosis of breast cancer in African women: a systematic review. Ann Epidemiol. 27(10):659–671 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Jedy-Agba E, McCormack V, Adebamowo C, Dos-Santos-Silva I (2016) Stage at diagnosis of breast cancer in sub-Saharan Africa: a systematic review and meta-analysis. Lancet Glob Health 4(12):e923–e935 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Pace LE, Shulman LN (2016) Breast cancer in sub-saharan Africa: challenges and opportunities to reduce mortality. Oncologist 21(6):739–744 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Shin DW, Cho J, Kim SY, Guallar E, Hwang SS, Cho B et al. (2013) Delay to curative surgery greater than 12 weeks is associated with increased mortality in patients with colorectal and breast cancer but not lung or thyroid cancer. Ann Surg Oncol 20(8):2468–2476 [DOI] [PubMed] [Google Scholar]
- 6.Miima S, Muturi A, Ojuka D, Ndung’u A (2017) Evaluating compliance to Kenya national cancer guidelines on diagnosis and staging of breast cancer at Kenyatta national hospital. East Afr Med J 94(7):553–560 [Google Scholar]
- 7.Barros ÂF, Uemura G, de Macedo JLS (2013) Interval for access to treatment for breast cancer in the Federal District, Brazil. Rev Bras Ginecol E Obstet Rev Fed Bras Soc Ginecol E Obstet 35(10):458–463 [DOI] [PubMed] [Google Scholar]
- 8.Norsa’adah B, Rampal KG, Rahmah MA, Naing NN, Biswal BM (2011) Diagnosis delay of breast cancer and its associated factors in Malaysian women. BMC Cancer 11:141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Molinié F, Leux C, Delafosse P, Ayrault-Piault S, Arveux P, Woronoff AS et al. (2013) Waiting time disparities in breast cancer diagnosis and treatment: a population-based study in France. Breast Edinb Scotl 22(5):810–816 [DOI] [PubMed] [Google Scholar]
- 10.Richardson LC, Royalty J, Howe W, Helsel W, Kammerer W, Benard VB (2010) Timeliness of breast cancer diagnosis and initiation of treatment in the National Breast and Cervical Cancer Early Detection Program, 1996–2005. Am J Public Health 100(9):1769–1776 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Allgar VL, Neal RD (2005) Delays in the diagnosis of six cancers: analysis of data from the National Survey of NHS Patients: Cancer. Br J Cancer 92(11):1959–1970 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Arndt V, Stürmer T, Stegmaier C, Ziegler H, Dhom G, Brenner H (2002) Patient delay and stage of diagnosis among breast cancer patients in Germany—a population based study. Br J Cancer 86(7):1034–1040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Richards MA, Westcombe AM, Love SB, Littlejohns P, Ramirez AJ (1999) Influence of delay on survival in patients with breast cancer: a systematic review. Lancet Lond Engl 353(9159):1119–1126 [DOI] [PubMed] [Google Scholar]
- 14.Ahmed AA (2014) Clinicopathological profile of female Sudanese patients with locally advanced breast cancer. Breast Dis 34:131e4. [DOI] [PubMed] [Google Scholar]
- 15.Aloulou S, El Mahfoudi A, El Omrani A, Khouchani M (2015) Factors related to late diagnosis of breast cancer: experience of CHU Mohammed VI Marrakech. Pan Afr Med J 21:162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Benbakhta B, Tazi M, Benjaafar N, Khattabi A, Maaroufi A (2015) Determinants of patient and health system delays for women with breast cancer in Morocco, 2013. Rev Epidemiol Sante Publique 63:191e201. [DOI] [PubMed] [Google Scholar]
- 17.El-Shinawi M, Youssef A, Alsara M, Hablas A, Gaafar R, Seifeldin IA et al. (2013) Assessing the level of breast cancer awareness among recently diagnosed patients in Ain Shams University Hospital. Breast 22:1210e4. [DOI] [PubMed] [Google Scholar]
- 18.Ermiah E, Abdalla F, Buhmeida A, Larbesh E, Pyrhonen S, Collan Y (2012) Diagnosis delay in Libyan female breast cancer. BMC Res Notes 5:452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Landolsi A, Gahbiche S, Chaafii R, Chabchoub I, Ben Fatma L, Hochlef M et al. (2010) Reasons of diagnostic delay of breast cancer in Tunisian women (160 patients in the central region of Tunisia). Tunis Med 88:894e7. [PubMed] [Google Scholar]
- 20.Mousa SM, Seifeldin IA, Hablas A, Elbana ES, Soliman AS (2011) Patterns of seeking medical care among Egyptian breast cancer patients: relationship to late-stage presentation. Breast 20(6):555–561. 10.1016/j.breast.2011.07.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Clegg-Lamptey J, Dakubo J, Attobra YN (2009) Why do breast cancer patients report late or abscond during treatment in Ghana? A pilot study. Ghana Med J 43:127e31. [PMC free article] [PubMed] [Google Scholar]
- 22.Ezeome ER (2010) Delays in presentation and treatment of breast cancer in Enugu Nigeria. Niger J Clin Pract. 13:311e6. [PubMed] [Google Scholar]
- 23.Ibrahim NA, Oludara MA (2012) Socio-demographic factors and reasons associated with delay in breast cancer presentation: a study in Nigerian women. Breast 21:416e8. [DOI] [PubMed] [Google Scholar]
- 24.Marcus TS, Lunda S, Fernandez L (2013) Delayed breast cancer presentation: hospital data should inform proactive primary care. Afr J Prim Health Care Fam Med 5:1e7 [Google Scholar]
- 25.Otieno ES, Micheni JN, Kimende SK, Mutai KK (2010) Delayed presentation of breast cancer patients. East Afr Med J 87:147e50. [DOI] [PubMed] [Google Scholar]
- 26.Pace LE, Mpunga T, Hategekimana V, Dusengimana JM, Habineza H, Bigirimana JB et al. (2015) Delays in breast cancer presentation and diagnosis at two rural cancer referral centers in Rwanda. Oncologist 20:780e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Price AJ, Ndom P, Atenguena E, MambouNouemssi JP, Ryder RW (2012) Cancer care challenges in developing countries. Cancer 118:3627e35. [DOI] [PubMed] [Google Scholar]
- 28.Toure M, Nguessan E, Bambara AT, Kouassi YK, Dia JM, Adoubi I (2013) Factors linked to late diagnosis in breast cancer in Sub-Saharan Africa: case of Cote d’Ivoire. Gynecol Obstet Fertil 41:696e700. [DOI] [PubMed] [Google Scholar]
- 29.Akarolo-Anthony SN, Ogundiran OT, Adebamowo AC (2010) Emerging breast cancer epidemic: evidence from Africa. Breast Cancer Research 12(Suppl 4):S8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Barros AF, Uemura G, de Macedo JL (2013) Interval for access to treatment for breast cancer in the Federal District Brazil. Rev Bras Ginecol Obstet 35:458e63. [DOI] [PubMed] [Google Scholar]
- 31.Molinie F, Leux C, Delafosse P, Ayrault-Piault S, Arveux P, Woronoff AS et al. (2013) Waiting time disparities in breast cancer diagnosis and treatment: a population-based study in France. Breast 22:810e6. [DOI] [PubMed] [Google Scholar]
- 32.Richardson LC, Royalty J, Howe W, Helsel W, Kammerer W, Benard VB (2010) Timeliness of breast cancer diagnosis and initiation of treatment in the National Breast and Cervical Cancer Early Detection Program, 1996–2005. Am J Public Health. 100:1769e76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Richards MA, Westcombe AM, Love SB, Littlejohns P, Ramirez AJ (1999) Influence of delay on survival in patients with breast cancer: a systematic review. Lancet 353:1119e25. [DOI] [PubMed] [Google Scholar]
- 34.Brinton LA, Figueroa JD, Awuah B, Yarney J, Wiafe S, Wood SN et al. (2014) Breast cancer in Sub-Saharan Africa: opportunities for prevention. Breast Cancer Res Treat 144(3):467e78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Opoku SY, Benwell M, Yarney J (2012) Knowledge, attitudes, beliefs, behaviour and breast cancer screening practices in Ghana. West Africa Pan Afr Med J 11:28. [PMC free article] [PubMed] [Google Scholar]
- 36.Adisa AO, Gukas ID, Lawal OO, Adesunkanmi AR (2010) Breast cancer in Nigeria: is non-adherence to chemotherapy schedules a major factor in the reported poor treatment outcome? Breast J 16:206e7. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Enquiries about data availability should be directed to the authors.