Breast Cancer Early Detection in Eswatini: Evaluation of a Training Curriculum and Patient Receipt of Recommended Follow-Up Care.

[Figure: see text].

BACKGROUND

Breast cancer (BC) is the most common malignancy diagnosed in women globally.[1] In sub-Saharan Africa (SSA), approximately 75% of Black patients have stage III or IV BC at diagnosis, more than double the comparable proportion in the United States.[2,3] No country in SSA provides population-level mammographic screening, and few health care providers (HCPs) are trained to recognize or manage breast abnormalities.[4,5] In both Rwanda and South Africa, patients with BC have reported that HCPs failed to recognize their early BC symptoms.[6,7]

CONTEXT

Key Objective

Can a breast cancer (BC) early-detection program in Eswatini improve HIV clinic nurses' breast evaluation skills and improve evaluation of their patients' breast symptoms?

Knowledge Generated

Following a 2-day training, nurses demonstrated significantly improved breast health knowledge and clinical breast examination skills. Those improvements were durable as well, with practically no deterioration at 90 days after training. However, when these nurses put their skills to use in the clinic, only 61% of patients with abnormal breast examinations completed their referral for further testing, reducing the program's potential for early BC detection.

Relevance

Successful efforts to decrease the stage of BC diagnosis in resource-limited health systems will require both effective detection strategies and accessible services for timely diagnosis.

In countries where mammographic screening is available, clinical breast examinations (CBEs) have not been shown to decrease BC mortality.[8] In countries where mammography is not widely available, investigators have explored using CBE as a stand-alone BC screening or early-detection technique.[9-16] Recent data from India suggest that CBE screening decreases BC stage at diagnosis and may decrease BC mortality in women > 50 years old.[17] However, there is consensus that before any screening program can be implemented, health systems must have the capacity to facilitate early diagnosis of symptomatic disease by building community awareness, enhancing providers' ability to evaluate abnormal findings, and ensuring robust referral systems to facilitate timely diagnosis.[18] In Rwanda, Pace et al developed a BC early-diagnosis program that trained nurses from primary health centers to address breast complaints and to perform CBE. Nurses learned about BC symptoms and received hands-on training in CBE skills. Over 2 years, 127 nurses evaluated 1,560 women who presented with breast complaints at primary health centers. Of these women, 18 were diagnosed with BC.[11]

In SSA, 4.5 million people living with HIV are > 50 years old; that number is expected to triple over the next 25 years.[19,20] As women living with HIV (WLHIV) age, their risk of developing BC increases, as does that of other women. Among patients with BC in both the United States and SSA, mortality is higher among WLHIV than among HIV-negative individuals.[21-25] Integration of HIV care with other noncommunicable diseases services is recommended, but has been limited.

In collaboration with the Ministry of Health of Eswatini (MOH), we piloted an adapted version of the program of Pace et al in five HIV and/or antiretroviral therapy (ART) clinics in Eswatini. We aimed to evaluate the pilot program's effectiveness in improving breast health knowledge and clinical skills of HIV and/or ART clinic nurses and HIV counselors. We also aimed to determine the volume of clinic patients who would be referred, because of breast abnormalities, by the trained nurses and the proportion of patients who would complete referrals.

METHODS

Context

The Kingdom of Eswatini (formerly Swaziland) is a small country in Southern Africa with a population of 1.3 million people. Eswatini has the world's highest prevalence of HIV in adults (27%), but > 90% of PLHIV use ART.[20] Eswatini's Essential Health Care Package, issued by the MOH, specifies that routine BC screening by CBE should be offered at every facility in the public health care system, but screening breast examinations are rarely performed.[26]

This study included high-volume HIV and/or ART clinics at four regional hospitals and one national referral hospital located in the four regions of the country. The national referral hospital has a breast specialty clinic where women with breast abnormalities can be referred for further evaluation. At the regional hospitals, women can also be referred to onsite surgical clinics.

Study Design and Participants

We conducted a prospective, two-phase pilot study of a training program for the early detection of BC in WLHIV and receiving care at a study clinic. First, we evaluated the impact of this training on breast health knowledge and clinical skills among nurses and HIV counselors working in the clinics. We subsequently evaluated the proportion of patients who completed follow-up care for an identified breast abnormality.

Nurses and HIV counselors were eligible for the study if they were ≥ 18 years old, were employed at a study clinic, attended the training session, were not planning to leave the clinic within 3 months of the planned training, and provided written, informed consent.

Patients were eligible for screening if they were female, ≥ 18 years old, diagnosed with HIV, and receiving care at a study clinic. They were eligible for the study if they had a CBE-detected breast abnormality, had at least one working telephone number, and provided verbal, informed consent.

Procedures

In May and June 2019, participating nurses and counselors underwent a 2-day BC early-detection training program conducted by the MOH and study staff, with a curriculum adapted from the aforementioned Rwandan program.[11] Content included normal breast physiology, benign pathology, and BC. Substantial time was devoted to CBE skills, including practice on silicone training mannequins.[27] Nurses were trained in a clinical algorithm that emphasized referring women with abnormal breast examination to a surgeon or the national breast clinic. The curriculum included standardized patient simulations to allow trainees to practice these skills. Following the training, study staff visited study clinics monthly to reinforce breast evaluation skills and monitor study activities.

Before the training, nurses and counselors completed a survey on demographics and prior breast health experience. Before and immediately after the training, all participants took a written examination to evaluate knowledge about breast health and BC. Before and after training, nurses also completed a case-based practical skills examination on evaluating a breast complaint, including performance of a CBE on the anatomical model and execution of the referral algorithm. Approximately 90 days after the training, nurses completed the knowledge and skills examinations for a third time.

Following training, nurses were instructed to screen all female patients ≥ 18 years for five specific breast symptoms: a breast mass, breast pain, breast skin or nipple changes, nipple discharge, and swelling in the armpit. Patients who reported breast symptoms underwent CBE. If an abnormality was found, referral to a surgeon or specialty breast clinic was provided. Nurses also asked referred patients for permission for a study coordinator to contact them. If the patient agreed, a study coordinator contacted them via telephone within 7 days and, if the patient provided verbal, informed consent, administered a questionnaire on demographics, BC risk factors, and knowledge of other women diagnosed with BC. Patient participants were phoned 30 days after enrollment to complete a questionnaire on whether they had completed the referral visit and the reasons for not completing the referral or the clinical outcome, as appropriate. Participants who had not completed referral at 30 days were phoned again at 60 days and completed the same questionnaire. Nurses logged counts of all women who were screened for breast symptoms and who were referred for clinical follow-up. Clinic charts were flagged postscreening to prevent rescreening at subsequent visits to the ART clinic. This phase of the pilot was planned for 1 year, but participant enrollment was ended 2 months early because of the COVID-19 pandemic.

Statistical Analysis

Among trainees, we used counts and percentages to describe demographics and experience with HIV and BC. To evaluate the impact of training curriculum on breast-related knowledge and clinical skills, pretraining, immediate post-training, and 90-day post-training examination scores were compared using Wilcoxon signed-rank tests. Among referred WLHIV, we determined the percentage completing referral and reported barriers to doing so. To explore factors associated with referral completion, demographic information and BC-related knowledge were compared between patients who did and did not undergo further evaluation using Z-score testing, chi-squared testing, and Wilcoxon-Mann-Whitney tests, as appropriate.

Ethical Considerations

The study was approved and monitored by the Eswatini Human and Health Research Review Board and the Columbia University Institutional Review Board. All participants provided written or verbal informed consent.

RESULTS

Trainees and Curriculum Evaluation

A total of 44 nurses and 36 counselors were trained in May-June 2019. The median age of the nurses was 35 years (interquartile range [IQR] 31-42), and 34 (78%) were female (Table 1). The median age of the counselors was 38 years (IQR 32-42), and 33 (92%) were female. Nurses reported a median of 10 years (IQR 7-16) of total work experience. Few had on-the-job experience or training related to BC. Just 10 nurses (23%) reported having cared for someone with BC, and five (11%) reported formal training related to BC. Four counselors (11%) reported having cared for a patient with BC.

TABLE 1

Characteristics of Nurses and Counselors Undergoing BC Early-Detection Training, Eswatini, 2019

All 44 nurses and 36 counselors completed the pretraining and immediate post-training knowledge examinations. Of 23 possible points, nurses scored a median of 18 (IQR 16-19) points on the pretraining examination (Fig 1A). On the immediate post-training examination, nurses scored a median of 20 (IQR 19-21) points, a significant improvement (P < .0001). Thirty-eight nurses (86%) completed the 90-day post-training examination, with a mean actual time from the training of 105 (range: 92-105) days. On the 90-day examination, nurses scored a median of 20 (IQR 19-21) on the written examination, also an improvement compared with pretraining examination scores (P < .0001). Among counselors, the median pretraining examination score was 16 (IQR 14-18) and the immediate post-training score was 18 (IQR 17-20), a significant improvement (P < .0001; Fig 1B).

FIG 1

Change in BC knowledge and practical skills examination scores with training, Eswatini, 2019. (A) BC knowledge examination scores among nurses. Immediate post-training and 90-day post-training scores both significantly improved, compared with pretesting scores (P < .0001 for both). (B) BC knowledge examination scores among counselors: pretraining and immediate post-training. Immediate post-training scores significantly improved, compared with pretesting scores (P < .0001). (C) BC practical skills examination scores among nurses: pretraining, immediate post-training, and 90-day post-training. Immediate post-training and 90-day post-training scores both significantly improved, compared with pretesting scores (P < .0001 for both). Wilcoxon signed-rank tests used for all comparisons. BC, breast cancer.

The nurses' scores on the practical skills examination also improved after training. Of 28 possible points, median scores were 10 points (IQR 7-11) before training, 24 points (IQR 22-25) immediately after training, and 23 points (IQR 22-24) at 90 days after training (Fig 1C). Both post-training scores were improvements over the pretraining scores (P < .0001).

Completion of Referrals

From June 2019 through April 2020, a total of 15,408 women ≥ 18 years old visited a participating ART clinic at least once. Trained nurses screened 9,502 WLHIV for breast symptoms. Of women screened, 150 (2%) reported symptoms and underwent a documented CBE, and 93 (1%) had an abnormal CBE, which prompted referral for further evaluation. Ninety (97%) of the referred women consented to participate in this study. Two consenting women could not be reached for any further follow-up; information on the remaining 88 women is reported here.

The median age of the referred patients was 39 years (IQR 33-46; Table 2). The largest group of patients came from the Hhohho region (n = 40; 45%), which contains the capital city of Mbabane.

TABLE 2

Demographic Characteristics of Referred Patients, Eswatini, 2019-2020

At the 30-day follow-up, 38 of 88 patients (43%) reported completing the referral; 38 patients (43%) had not done so, and 12 (14%) could not be contacted. At the 60-day follow-up, among the 50 patients not known to have completed a referral at 30 days, 16 (32%) had completed the referral and 31 (62%) had not. Patients who did and did not complete referral did not differ in age, region, employment status, or relationship status (Table 2). The reasons most frequently cited by the patients for not yet having completed follow-up were the costs of transportation and follow-up care, non–work-related time constraints, and not being convinced that further evaluation of their breast complaint was needed (Table 3). Only three patients mentioned COVID-19 precautions as a reason for not completing the referral; however, enrollment stopped in April 2020, early in the pandemic.

TABLE 3

Barriers Cited for Not Completing Referral (multiple answers allowed), Eswatini, 2019-2020

Nearly all referred patients (n = 80; 91%) had heard of BC (Table 4). Only 44 (50%) reported having known a woman with BC personally, and 22 (25%) reported knowing a woman who had been cured of BC. Although 54 women (61%) reported having ever performed a breast self-examination, only 31 (35%) reported having ever undergone CBE. Patients who did and did not complete referral by 60 days did not differ in BC knowledge or exposure (all P > .05). Women completing referral reported a median of 4 months of symptoms before their CBE, whereas women who did not complete referral reported a median of 6 months; the difference was not statistically significant.

TABLE 4

Patient Participants' Breast-Related Knowledge and Referral Completion, Eswatini, 2019-2020

Of the 54 women who completed referral, 42 (81%) self-reported not being diagnosed with any specific illness, six women (11%) reported that they were still undergoing evaluation with additional imaging or a biopsy to determine their diagnosis, one (2%) reported being diagnosed with a breast abscess, and three (6%) reported being newly diagnosed with BC.

DISCUSSION

We describe early findings from a BC early-detection pilot program conducted in five HIV and/or ART clinics in Eswatini. The program's training curriculum successfully improved nurse and counselor breast health knowledge, and nurses maintained that improvement for at least 90 days. Nurses also showed a large and durable improvement in their breast evaluation clinical skills. The program had substantial reach: over 10 months, these trained nurses screened more than 9,500 women for breast symptoms and referred 93 for breast abnormalities. However, referral completion was suboptimal. Of the 88 referred women who enrolled in the study, 54 (61.4%) self-reported having completed their referral within 60 days of enrollment. Participants who did not complete referral frequently cited the costs or time required as reasons. On the basis of self-report, three new BCs were diagnosed.

Although Eswatini's national guidelines call for use of CBE in routine screening, nurses lacked practical skills for assessing a breast complaint, performing a CBE, and escalating care when appropriate. Our program adapted a training curriculum that has previously shown success in Rwanda.[11] The training combined didactic presentations, hands-on examination practice with training mannequins, and practical case studies. Nurse trainees immediately put their skills to use in the clinic. This combination of multimodal training techniques and routine real-world practice likely contributed to the excellent retention seen 3 months after training. Our findings were consistent with the experience of Pace et al in Rwanda, where median knowledge examination scores improved from 73.9% to 91.3% and practical examination scores improved from 24.0% to 88.0%, also with near total retention at 3-6 months.[11]

The curriculum's success in two different health care systems suggests that it could similarly improve knowledge and skills in any setting where HCPs lack skills in basic breast health. The curriculum is inexpensive, short, and scalable. If implemented widely, training group size would be limited only by the number of mannequins on hand for CBE practice. For countries aiming to offer universally available CBE, we would recommend broad implementation of our curriculum among HCPs. The Rwanda program included mentorship for nurses in the months after training, whereas our nurses were visited monthly by a trained study coordinator, which may have contributed to skill retention. Refresher trainings and ongoing supervision therefore might also be beneficial.

The major challenge our program faced was patient retention following abnormal examination. Fewer than two in three women obtained further evaluation after their abnormal breast examination. If women who did and did not complete referral had similar BC prevalence, then two of the 34 women who did not complete referral had BC and went undiagnosed. Poor follow-up undermines the potential of early-detection programs to reduce mortality. A CBE screening program in Manila, Philippines, was discontinued after 1 year when only 35% of referred women completed follow-up, reducing that program's CBE sensitivity for BC from 53% to 26%.[12] Participants in our study cited associated costs—both financial and time—as the greatest barrier to completing follow-up. The time and money needed to obtain follow-up care could be reduced by offering those services at the hospitals where participants already receive HIV care. Breast ultrasound and fine-needle aspiration may be feasible outside tertiary referral hospitals. Experience from Rwanda suggests that general practitioners and nurses can be trained in breast ultrasound.[28]

A successful screening or early-diagnosis program also requires downstream diagnostic systems able to accommodate an influx of new patients, as even the best screening tests discover far more false positives than true positives. Six patients reported that their referral clinic was not able to see them when they presented. At the end of the study period, six women who had pursued follow-up care were awaiting further testing to determine their diagnosis. If diagnostic services cannot be offered in a timely fashion, follow-up rates are likely to decrease over time.

Our strategy of examining only women who reported breast symptoms may reduce the diagnostic burden from false-positive CBEs. The program of Pisani et al in the Philippines targeted all women age 35-64 years, and the program of Sankaranarayanan et al offered screening CBE to all women age 30-69 years in the Trivandrum district of India. In both, CBE had a positive predictive value (PPV) of approximately 1%.[12,13] Alternatively, the program of Pace et al in Rwanda targeted women who self-presented with a breast complaint and produced a PPV of 6%-7%.[11] In Kenya, Botswana, and Tanzania, at discrete BC screening events open to interested women, CBE yielded PPVs of 9%, 3%-4%, and 5%, respectively.[10,14,15]

BC incidence data would be needed to determine how much examining only symptomatic women affected CBE's sensitivity, but the strategy detected 32 new BC cases per 100,000 women examined. That proportion is similar to those reported by Pisani et al and Sankaranarayanan et al and double that currently documented by Eswatini's cancer registry (32 v 15 per 100,000 person-years).[1,12,13] Our study was not designed to evaluate this program's impact on BC stage at diagnosis or mortality. However, Rwanda's program appears to have increased the diagnosis of early-stage BC, and population-level CBE screening in India is reported to have reduced BC-related mortality among women > 50 years old.[17,29] Individual ministries in SSA will need to decide if the suggested benefits and costs of either approach to BC early detection are sufficient for broad implementation. However, our findings demonstrate that, before any widespread early-detection efforts in Eswatini, strategies are needed to ensure that patients can complete recommended evaluation after an abnormal test.

This pilot study has some inherent limitations. The frequency of BC cases detected may shift with multiple rounds of testing as prevalent cancers are diagnosed in the first round. The rate of completed referral might be subject to a Hawthorne Effect and diminish in real-world practice. Our reliance on patient self-report may have resulted in inaccurate counts of new BC diagnoses.

Much larger studies of CBE as a BC early-detection technique are needed to characterize its impact on BC stage at diagnosis and mortality in SSA. Strategies to improve follow-up—such as patient tracking, patient navigation, transport support, or decentralization of diagnostic services—are critical to the success of early-detection efforts and should be evaluated. However, the sum of work from lower middle-income countries suggests that CBE may be a feasible and effective tool for improving early detection of BC and, possibly, decreasing BC mortality in regions without mammography.