Physical activity in association with mortality among Black women diagnosed with breast cancer in the Southern Community Cohort Study

Abstract

PURPOSE:

Physical activity (PA) is associated with many health benefits. While PA has been associated with reduced mortality after breast cancer diagnosis in many studies, few studies have examined the role of PA in breast cancer survival among underserved and minority populations, including Black women. We investigated PA in association with mortality among Black predominantly low-income breast cancer survivors in the Southern Community Cohort Study (SCCS).

METHODS:

Study participants were women diagnosed with incident breast cancer (n=949) in the SCCS, which is a prospective cohort study of predominantly low-income adults aged 40–79 years recruited from 12 Southeastern states between 2002–2009. Participants completed a detailed baseline questionnaire, with annual follow-up for mortality via registry linkages. Cox regression models were used to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for associations of pre-diagnosis PA (measured via a validated questionnaire) with all-cause and breast cancer-specific mortality.

RESULTS:

Breast cancer survivors had a mean age of 61.1 years and most (79.3%) had a household income of <$25,000. In adjusted models, higher levels of total PA (MET-hours/day) were inversely associated with all-cause mortality with HRs (95% CIs): 0.79 (0.59–1.06), 0.66 (0.49–0.90), and 0.60 (0.43–0.84), for Q2, Q3, and Q4 (reference: Q1), respectively, p_trend ≤0.01. A similar inverse association was found for breast cancer-specific mortality.

CONCLUSIONS:

Higher levels of pre-diagnosis PA were associated with improved survival among low-income Black breast cancer survivors. Resources to reduce barriers to PA participation and increase support for education and intervention efforts to promote PA among Black women are needed.

INTRODUCTION

Although breast cancer mortality has decreased for all racial/ethnic groups in the United States, Black women continue to have the lowest survival rates regardless of stage and subtype, and this disparity continues beyond 10 years post-diagnosis (1–6). Black women diagnosed with breast cancer are less likely than White women to be diagnosed with localized disease (7). Diagnosis at later stages can lead to increased short and long-term morbidity (for example, due to more aggressive cancer treatments) and increased risk for distant recurrence (8, 9). Lack of timely access to quality and non-biased health care and long-term survivorship care can further increase risk of poor prognosis after breast cancer diagnosis (9–11).

Compared to other race/ethnicities in the United States, Black women have a higher comorbidity burden, including prevalence of diabetes, hypertension, and obesity (12–14). The causative factors behind this higher comorbidity burden are complex, and involve social factors and structural inequities (15–18). Using data from a case-control study in Ohio, Thompson and colleagues reported that only 13.2% of Black breast cancer cases and 23.7% of Black breast cancer controls met American Cancer Society physical activity (PA) guidelines (13). Busen et al. reported that lower levels of total PA (recreational, household, and transportation) among Black breast cancer survivors was associated with obesity (19). Black women face many barriers to participating in PA that are related to the social determinants of health and social inequities, such as access to walkable and safe neighborhoods and greenways, economic burden, and need for social support (20–22).

PA has many health benefits among women, including improved body composition and muscle mass, reduced fatigue, reduced risk of metabolic and cardiovascular disease, and improved physical and mental health-related quality of life (23–28). Despite the substantial literature on PA and mortality after a diagnosis of breast cancer (29–31), very few studies have examined these associations among underserved, low-income, and minority populations, and in particular low-income Black women. We identified one previous study of PA and breast cancer survival among Black women, which used data from a pooling study of population-based case-control studies in California that included follow-up for mortality outcomes and focused on pre-diagnosis recreational PA (32). This study reported no association for pre-diagnosis recreational PA in hours per week with either all-cause or breast cancer-specific mortality among Black women.

The Southern Community Cohort Study (SCCS) is a large prospective cohort of predominantly low income Black and White adults from 12 Southeastern states (33). This cohort provides the opportunity to examine PA in a unique cohort of older and predominantly low-income Black breast cancer survivors (e.g., women diagnosed with incident breast cancer after enrollment in the cohort). We evaluated the association of pre-diagnosis PA and all-cause mortality and breast cancer-specific mortality among Black women diagnosed with incident breast cancer in the SCCS.

MATERIALS AND METHODS

Study Design and Study Population

The SCCS is a prospective cohort study of predominantly low-income Black and White adults aged 40–79 years at enrollment (n=84,797). The study methods and rationale have been previously published (33, 34). Participants were recruited from 12 Southeastern states between March 2002 and September 2009. At enrollment, a baseline questionnaire collected data on socio-demographics, medical history, health behaviors, reproductive history, employment history, weight and height, PA, and diet. The majority of participants (85%) were recruited through community health centers and completed computer-assisted in-person interviews (http://www.southerncommunitystudy.org). The remaining participants completed a mailed version of the baseline questionnaire.

Vital status, date and cause of death information are ascertained via annual linkages to the Social Security Administration and National Death Index (NDI). Participants are followed for new cancer diagnoses via annual linkages to statewide cancer registries. Statewide cancer registries provide data on type of cancer, age at diagnosis, initial primary cancer treatment (surgery, radiotherapy, and chemotherapy), tumor characteristics, and stage at diagnosis. The initial eligible population for the analyses herein included Black women diagnosed with primary invasive incident breast cancer from enrollment through 2018 (n=1,006 incident breast cancer cases). Six women had date of breast cancer diagnosis equal to date of death and these women were excluded from the analysis. In addition, women missing PA (n=51) were excluded, for a final sample of 949 breast cancer survivors. All participants provided written informed consent, and the SCCS was approved by the institutional review boards at Vanderbilt University Medical Center and Meharry Medical College.

Pre-diagnosis Physical Activity

A questionnaire specifically for use in the SCCS was developed to evaluate multiple types of PA and sedentary behavior (35). Data were collected on activities at home, work, and specifically for exercise. The full questionnaire is available online (http://www.southerncommunitystudy.org). Total PA was summarized in metabolic equivalent (MET)-hours/day (36). Total PA included light, moderate, and strenuous household and occupational activity, and moderate/vigorous sports, and this variable has been used in previous reports in the SCCS (37, 38). We also considered PA in hours/day to enable comparison to previous studies (32). We initially considered separately evaluating moderate/vigorous sport participation (e.g., leisure-time PA) for comparison to previous reports (23, 30, 39), but sample size was too small (72.5% reported no moderate or vigorous exercise participation). Total pre-diagnosis PA was categorized in quartiles with cutpoints shown in Table 1. We also examined walking in MET-hours per day.

Table 1.

Characteristics of Black breast cancer survivors^a, Southern Community Cohort Study

Age at diagnosis, mean, range	61.1	41.1–89.0
Age at SCCS cohort enrollment, mean, range	53.5	40.0–79.0
Time between enrollment and diagnosis, mean, range	7.51	0.2–16.8
Menopausal status at diagnosis, n %
Premenopausal	46	4.9
Postmenopausal	833	87.8
Missing	70	7.4
Education, n %
<High school	261	27.5
High school	313	33.0
Some college/technical training	244	25.7
≥College degree	131	13.8
Income, n %
<$15,000	547	57.6
$15,000-<$25,000	206	21.7
≥$25,000	187	19.7
Missing	9	1.0
Insurance Coverage, n %
None	317	33.4
Other type	253	26.7
Medicare	227	23.9
Medicaid	149	15.7
Missing	3	0.3
SEER Summary Stageb, n %
Local	494	52.1
Regional	355	37.4
Distant	52	5.5
Missing	48	5.1
ER/PRb, n %
ER+/PR+	450	47.4
ER+/PR− or ER−/PR+	127	13.4
ER−/PR−	247	26.0
Missing	125	13.2
Surgeryb, n %
None	30	3.2
Breast conserving surgery	97	10.2
Mastectomy	423	44.6
Missing	399	42.0
Radiotherapyb, n %
No	468	49.3
Yes	314	33.1
Missing	167	17.6
Chemotherapyb, n %
No	431	45.4
Yes	416	43.8
Missing	102	10.8
Hypertension
No	314	33.1
Yes	634	66.8
Missing	1	0.1
Diabetes
No	679	71.6
Yes	270	28.5
Pre-diagnosis BMI at baseline (kg/m2)
<24.99	115	12.1
25–29.99	234	24.7
30–34.99	250	26.3
35–39.99	172	18.1
≥40	161	17.0
Missing	17	1.8
Total PA (MET-hours/day)
<9.91	239	25.2
9.91–<17.2	229	24.1
17.2–27.29	243	25.6
≥27.29	238	25.1
Total PA (hours/day)
<3.14	240	25.3
3.14–<5.43	224	23.6
5.43<8.71	249	26.2
≥8.71	236	24.9

Abbreviations: Body mass index (BMI), estrogen receptor (ER), metabolic equivalents (MET), physical activity (PA), progesterone receptor (PR), Surveillance, Epidemiology, and End Results Program (SEER).

^aIncludes women diagnosed with incident breast cancer after enrollment in the cohort.

^bData source is from statewide cancer registries.

Clinical Characteristics and Additional Covariates

Potential confounding factors and breast cancer survivor characteristics considered in the analysis were selected a priori based on previous research (29, 32, 39) and included age at diagnosis, time between enrollment in the SCCS and diagnosis of breast cancer, Surveillance, Epidemiology, and End Results (SEER) Program summary stage (local, regional, distant, missing/unstaged), estrogen receptor (ER)/progesterone receptor (PR) status (classified as joint ER/PR status (ER+/PR+, ER+/PR-, ER-/PR+, ER-/PR-), chemotherapy (yes, no), radiotherapy (yes, no), breast cancer surgery (none, breast conserving surgery, mastectomy), menopausal status at diagnosis of breast cancer (premenopausal, postmenopausal), education (<high school, high school, some college/technical training, college degree or higher), household income (<$15,000, $15,000-<$25,000, ≥$25,000), hypertension, diabetes or high blood sugar, and pre-diagnosis body mass index (BMI). Weight and height were self-reported for most women (>90%) at the baseline questionnaire (cohort enrollment) and were used to calculate BMI. BMI was initially categorized using the World Health Organization/Centers for Disease Control and Prevention classifications: underweight (<18.5 kg/m²), normal weight (18.5–24.99 kg/m²), overweight (25–29.99 kg/m²), obese class I (BMI (30-<34.99 kg/m²), obese class II (BMI 35-<40.0 kg/m²), and obese class III (BMI ≥40.0 kg/m²) (40). However, the sample size for <18.5 kg/m² (n=6) was too small and combined with the normal weight category.

Data on cancer diagnosis, age at diagnosis, cancer treatment, ER/PR status, and SEER summary stage were collected from the statewide cancer registries while all other variables described above were part of the baseline SCCS questionnaire. Cancer registries generally systematically collect treatment information for the first course of treatment only. While completeness of surgery and radiotherapy may be accurately captured for many cases, as well as receipt of chemotherapy (yes, no) for initial diagnosis, hormonal therapy is incomplete (41, 42), therefore we do not present data on hormonal therapy for this study.

Outcome Ascertainment and Final Analytic Sample

Study outcomes included all-cause mortality and breast cancer-specific mortality. Mortality data were collected via annual linkages to the Social Security Administration vital status service for epidemiologic researchers and the NDI through 12/31/2020. Breast cancer deaths were classified using the underlying cause of death data from the NDI and the International Classification of Diseases, Tenth Revision code C509.

Statistical Analysis

Frequencies and percentages for categorical variables and means (ranges) for continuous variables were calculated for socio-demographic and clinical factors overall and by race. We also evaluated bivariate associations for clinical factors and key covariates by total PA. Statistical differences were assessed with the Chi-Square test for categorical variables and Kruskal-Wallis test for continuous variables. Survival time was calculated from diagnosis date to date of death or date of last follow-up via linkage to the NDI. Kaplan-Meier estimates of 5-year survival were calculated for overall survival and breast cancer-specific survival.

Adjusted hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were derived from Cox proportional hazards regression models (43). Entry time was defined as date of breast cancer diagnosis and exit time was defined as date at death or last follow-up. Multivariable models were first adjusted for age at diagnosis, education, time between cohort enrollment, BMI, and diabetes (as this was associated with PA in the descriptive analysis). As education and income were correlated, only education was included in the multivariable models. Multivariable models were additionally adjusted for clinical factors (stage, breast cancer surgery, radiotherapy and chemotherapy). Missing data (shown by covariate in Table 1) were included as a category for each covariate as the sample size was small (44). We conducted a sensitivity analysis excluding deaths in the first year after diagnosis from the analysis (45). We also conducted a sensitivity analysis excluding women diagnosed with breast cancer <1 year after the baseline interview, when PA was measured. The proportional hazards assumption was tested by examining interaction terms for PA and both overall survival time and breast cancer-specific survival time, and no evidence for violation of this assumption was found. P-values for trend were obtained by including the ordinal categorical variable for PA in the model as a continuous variable. All analyses were performed using SAS version 9.4. Tests of statistical significance were based on two-sided probability and p-values <0.05 were considered statistically significant.

RESULTS

The mean time between enrollment and breast cancer diagnosis was 4.9 years, ranging from zero to 13 years. The mean years of follow-up after breast cancer diagnosis was 6.9 years. During follow-up, there were 332 deaths (173 of these were due to breast cancer). Five-year overall survival and breast cancer-specific survival rates were 75.4% and 84.8%, respectively.

Table 1 displays characteristics of Black breast cancer survivors in the SCCS. The mean age was 61years, ranging from 41 to 89 years. Most women were postmenopausal at enrollment into the cohort (87.8%). Close to 14% had a college degree or higher, while 27.5% had less than a high school education. The most common type of health insurance status was not having coverage (33%). Over half of women were diagnosed with localized breast cancer (52%), followed by regional (37%), and distant (5.5%). Over 66% of women had a diagnosis of hypertension at enrollment into the cohort and 28.5% had diabetes.

We also evaluated sociodemographic and clinical factors associated with pre-diagnosis total PA (MET-h/day) (Table 2). Women with lower levels of daily total PA had lower education, lower income, and were more likely to report a diagnosis of diabetes.

Table 2.

Characteristics of Black breast cancer survivors by pre-diagnosis physical activity, Southern Community Cohort Study^a

	Total PA (MET-hours/day)
	<9.91		9.91–<17.2		17.2–27.29		≥27.29
	n	%	n	%	n	%	n	%
Education
<High school	84	35.2	68	29.7	68	28.0	41	17.2
High school	78	32.6	75	32.8	74	30.5	86	36.1
Some college	51	21.3	52	22.7	63	25.9	78	32.8
≥College degree	26	10.9	34	14.9	38	15.6	33	13.9b
Income
<$15,000	165	69.9	136	59.7	136	56.2	110	47.0
$15,000-<$25,000	43	18.2	43	18.9	58	24.0	62	26.5
≥$25,000	28	11.9	49	21.5	48	19.8	62	26.5b
Menopausal Status
Premenopausal	11	4.9	12	5.7	8	3.5	15	6.9
Postmenopausal	212	95.1	199	94.3	219	96.5	203	93.1
SEER Summary Stage
Local	116	51.1	125	56.6	125	54.8	128	56.9
Regional	96	42.3	88	39.8	85	37.3	86	38.2
Distant	15	6.6	8	3.6	18	7.9	11	4.9
ER/PR
ER+/PR+	105	51.2	107	52.5	118	58.1	120	56.6
ER+/PR− or ER−/PR+	29	14.2	37	18.1	38	18.7	23	10.9
ER−/PR−	71	34.6	60	29.4	47	23.2	69	32.6
Surgery
None	29	12.6	24	10.9	27	11.3	17	7.4
Lumpectomy	104	45.0	104	47.3	107	44.8	108	47.2
Mastectomy	98	42.4	92	41.8	105	43.9	104	45.4
Radiotherapy
No	119	60.4	109	56.5	128	64.7	112	57.7
Yes	78	39.6	84	43.5	70	35.4	82	42.3
Chemotherapy
No	113	52.8	110	52.9	114	52.5	94	45.2
Yes	101	47.2	98	47.1	103	47.5	114	54.8
Hypertension
No	74	31.0	85	37.1	71	29.3	84	35.3
Yes	165	69.0	144	62.9	171	70.7	154	64.7
Diabetes
No	149	62.3	166	72.5	174	71.6	190	79.8
Yes	90	37.7	63	27.5	69	28.4	48	20.2b
Smoking
Current	63	26.4	66	28.8	66	27.2	67	28.3
Former	59	24.7	54	23.6	46	18.9	48	20.3
Never	117	49.0	109	47.6	131	53.9	122	51.5
Pre-diagnosis BMI at baseline kg/m2)
<24.99c	21	8.9	29	13.0	36	15.2	29	12.3
25–29.99	58	24.6	58	25.9	61	25.7	57	24.3
30–34.99	73	30.9	52	23.2	65	27.4	60	25.5
35–39.99	43	18.2	48	21.4	36	15.2	45	19.2
≥40	41	17.4	37	16.5	39	16.5	44	18.7

Abbreviations: Body mass index (BMI), estrogen receptor (ER), metabolic equivalents (MET), physical activity (PA), progesterone receptor (PR), Surveillance, Epidemiology, and End Results Program (SEER).

^aTable excludes missing data by variable (see Table 1 for frequency and percentage missing for each variable as applicable). Table includes women diagnosed with incident breast cancer after enrollment in the cohort.

^bP-value from Chi-square test <0.05.

^cSix women in the lowest BMI category had BMI <18.5 kg/m².

Table 3 displays results for the associations of pre-diagnosis PA with all-cause mortality among Black breast cancer survivors. Higher levels of total PA both in MET-hours/day and hours/day were inversely associated with all-cause mortality with statistically significant trends across all models (p_trend <0.01). For example, HRs (95% CIs) for PA in MET-hours/day (reference: quartile 1) and all-cause mortality, adjusting for time between cohort enrollment and diagnosis, BMI, education, and diabetes were: 0.79 (0.59–1.05), 0.65 (0.48–0.88), 0.54 (0.39–0.74), for quartiles 2, 3, and 4 respectively). The inverse association remained after adjustment for clinical factors (stage, breast cancer surgery, chemotherapy, and radiotherapy). Results were similar when also looking at total PA in hours per day. We also examined walking in MET-hours/day and all-cause mortality. A statistically significant inverse association was observed in age-adjusted models, but the HRs were no longer statistically significant after adjustment, although the p_trend was 0.02 (Supplemental Table S1).

Table 3.

Hazard ratios for PA in association with all-cause mortality among Black breast cancer survivors^a in the Southern Community Cohort Study

		Age-adjustedb		Further-adjusted model 1c		Further-adjusted model 2d
	Event	HR	(95% CI)	HR	(95% CI)	HR	(95% CI)
Total PA (MET-hours/day)
<9.91	113	1.00	(reference)	1.00	(reference)	1.00	(reference)
9.91–<17.2	84	0.72	(0.54–0.96	0.79	(0.59–1.05)	0.79	(0.59–1.06)
17.2–27.29	76	0.62	(0.46–0.83	0.65	(0.48–0.88)	0.66	(0.49–0.90)
≥27.29	59	0.47	(0.34–0.65	0.54	(0.39–0.74	0.60	(0.43–0.84)
ptrend		<0.01		<0.01		<0.01
Total PA (hours/day)
<3.14	112	1.00	(reference)	1.00	(reference)	1.00	(reference)
3.14–<5.43	83	0.77	(0.58–1.02	0.82	(0.62–1.10)	0.84	(0.63–1.13)
5.43<8.71	75	0.61	(0.46–0.82	0.63	(0.47–0.85)	0.69	(0.51–0.94)
≥8.71	62	0.52	(0.38–0.71	0.57	(0.42–0.78)	0.64	(0.46–0.89)
ptrend		<0.01		<0.01		<0.01

Abbreviations: physical activity (PA), hazard ratios, metabolic equivalents (MET).

^aIncludes women diagnosed with incident breast cancer after enrollment in the cohort.

^bAdjusted for age at diagnosis.

^cAdditionally adjusted for education, time between cohort enrollment and breast cancer diagnosis, diabetes, and body mass index.

^dAdditionally adjusted for stage, breast cancer surgery, radiotherapy, and chemotherapy

Table 4 displays results for the associations of pre-diagnosis PA with breast cancer-specific mortality. Higher levels of total PA in MET-hours/day were inversely associated with breast cancer-specific mortality with statistically significant trends across all models. For example, HRs (95% CIs) for PA in MET-hours/day (reference: quartile 1) and breast cancer-specific mortality, adjusting for time between cohort enrollment and diagnosis, BMI, education, and diabetes were: 0.83 (0.55–1.25), 0.66 (0.43–1.01), 0.59 (0.38–0.91), for quartiles 2, 3, and 4 respectively (p_trend:<0.01). The inverse association remained after adjustment for clinical factors (stage, breast cancer surgery, chemotherapy, and radiotherapy). Patterns were similar when also looking at total PA in hours per day. We also examined walking in MET-hours/day and breast cancer-specific mortality. HRs in the highest two quartiles were suggestive of an inverse association, but not statistically significant in age-adjusted or fully adjusted models, the p-values for trend were 0.01 and 0.04, respectively (Supplemental Table S1).

Table 4.

Hazard ratios for PA in association with breast cancer-specific mortality among Black breast cancer survivors^a in the Southern Community Cohort Study

		Age-adjustedb		Further-adjusted model 1c		Further-adjusted model 2d
	Event	HR	(95% CI)	HR	(95% CI)	HR	(95% CI)
Total PA (MET-hours/day)
<9.91	55	1.00	(reference)	1.00	(reference)	1.00	(reference)
9.91–<17.2	43	0.74	(0.50–1.11)	0.83	(0.55–1.25)	0.86	(0.57–1.31)
17.2–27.29	40	0.65	(0.43–0.97)	0.66	(0.43–1.01)	0.65	(0.42–1.00)
≥27.29	35	0.52	(0.34–0.79)	0.59	(0.38–0.91)	0.62	(0.39–0.98)
ptrend		<0.01		<0.01		0.02
Total PA (hours/day)
<3.14	56	1.00	(reference)	1.00	(reference)	1.00	(reference)
3.14–<5.43	41	0.77	(0.51–1.15)	0.82	(0.54–1.23)	0.94	(0.62–1.44)
5.43<8.71	39	0.62	(0.41–0.93)	0.63	(0.41–0.96)	0.68	(0.44–1.05)
≥8.71	37	0.57	(0.37–0.86)	0.62	(0.41–0.96)	0.69	(0.44–1.08)
ptrend		<0.01		0.01		0.04

Abbreviations: physical activity (PA), hazard ratios, metabolic equivalents (MET).

^aIncludes women diagnosed with incident breast cancer after enrollment in the cohort.

^bAdjusted for age at diagnosis.

^cAdditionally adjusted for education, time between cohort enrollment and breast cancer diagnosis, diabetes, and body mass index.

^dAdditionally adjusted for stage, breast cancer surgery, radiotherapy, and chemotherapy

We also conducted two sensitivity analyses. We evaluated the associations of total PA and all-cause mortality and breast cancer-specific mortality excluding the first year of events after breast cancer diagnosis. The inverse association for increasing levels of PA for both outcomes was similar to overall findings (Supplemental Table S2). For example, HRs (95% CIs) for PA in MET-hours/day (reference: quartile 1) and all-cause mortality, adjusting for time between cohort enrollment and diagnosis, BMI, education, and diabetes were: 0.80 (0.58–1.09), 0.68 (0.49–0.94), 0.55 (0.39–0.79), for quartiles 2, 3, and 4 respectively (p_trend:<0.01). We evaluated the associations of total PA and all-cause mortality and breast cancer-specific mortality excluding women who were diagnosed with breast cancer in the cohort <1 year after the baseline interview (which measured PA). The inverse association for increasing levels of PA for both outcomes was similar to overall findings (Supplemental Table S3). For example, HRs (95% CIs) for PA in MET-hours/day (reference: quartile 1) and all-cause mortality, adjusting for time between cohort enrollment and diagnosis, BMI, education, and diabetes were: 0.77 (057–1.03), 0.65 (0.48–0.88), 0.51 (0.37–0.72), for quartiles 2, 3, and 4 respectively (p_trend:<0.01).

DISCUSSION

In this study of low-income Black breast cancer survivors from the Southeastern United States, we found higher levels of total PA prior to cancer diagnosis were inversely associated with reduced risk of all-cause mortality and breast cancer-specific mortality. For all-cause mortality, risk was reduced by about 34–40% in the highest categories of PA, while risk was reduced by about 35–38% in the highest categories of PA for breast cancer-specific mortality. Our cohort study is the first to report an inverse association between PA and all-cause mortality and breast-cancer-specific among low-income Black breast cancer survivors.

The association of PA and breast cancer survival has been well-studied (23–28), but with very limited data among Black women and women of lower socioeconomic status, who have lower survival after a cancer diagnosis (9). The findings of our study are in contrast with the one previous study of PA and mortality outcomes among Black women (32). This study pooled data from three population-based case control studies of women diagnosed with breast cancer in one state (California) during 1994–2002 and followed-up through 2010. Results among Black women for all-cause mortality and recreational PA 10 years before diagnosis in hours/week were as follows: HR=1.04, 95% CI: 0.83–1.29 and HR=0.85, 95 CI: 0.67–1.07 for tertile 2 and tertile 3, respectively (reference: tertile 1). Results for breast cancer-specific mortality were as follows: HR=1.00, 95% CI:0.75–1.32 and HR=1.08, 95% CI: 0.82–1.44) for tertile 2 and tertile 3, respectively (reference: tertile 1). This was a large and well-conducted study, but included cases from case-control studies with differing measures of PA. Our findings are similar to a recent study that showed benefits of pre-diagnosis PA on all-cause mortality in the Diet, Exercise, Lifestyle and Cancer Prognosis Study (46). This study was not focused on Black women or women of lower socioeconomic status. This study also had data on post-diagnosis PA. Future studies of Black breast cancer survivors that include data on pre-diagnosis and post-diagnosis PA are needed. Further, a critical need of future research is to address barriers to PA participation among Black women using a health equity approach (11).

Black women have increased barriers to participating in PA that are related to the social determinants of health and multilevel social inequities (20–22). Joseph and colleagues conducted an integrative literature review on barriers to physical activity among Black women (21). Key environmental barriers included lack of access to sidewalks and facilities and safety concerns. Inequity in the built environment has been highlighted as a major contributor to disparities in PA participation among minority populations and those living low-income neighborhoods (47, 48). Joseph et al. also highlighted economic factors including lack of access to exercise classes or gym membership, fatigue and family commitments (21, 49). Finally, Joseph et al. noted social support as a key barrier, which can be particularly important in populations that are both disadvantaged and older (50, 51).

Paxton and colleagues reviewed health behavior and lifestyle interventions specifically among Black breast cancer survivors in 2019 (52). The authors’ overall summary for PA interventions based on the current evidence was that while the literature on intervention studies targeting Black women was not substantial, most interventions demonstrated success in improving PA. They noted the key importance of utilizing partnerships with community-based organizations and support groups in recruiting and retaining Black women in future interventions. The IMPROVE study, a community-based study, is currently being conducted to evaluate a group-based exercise intervention compared to group-based social support among older Black and Non-Hispanic White breast cancer survivors (53). While intervention studies such as IMPROVE are a key part of improving PA, a multi-faceted health equity approach that considers comprehensively structural inequities, policies, and economic, physical, and social environments to address all barriers to PA among minority and disadvantaged populations is ultimately needed (11).

The SCCS provided a unique opportunity to evaluate the role of PA in an understudied population of low-income Black breast cancer survivors, who are at higher risk for poor survival compared to higher income women and White women. However, several limitations of our study need to be considered. One limitation of our analysis was the use of self-reported PA, which could result in misclassification and bias findings, most likely to the null. However, the PA questionnaire used in the SCCS has been validated (35) and used in many research studies (37, 38). Another limitation was the focus on pre-diagnosis PA. Studies have shown PA habits change after a breast cancer diagnosis and may decrease overtime in long-term survivors, however, part of this decrease could be due to aging as well (54, 55). Future studies with PA measured at multiple time points among Black breast cancer survivors before and after diagnosis are needed. In addition, sample sizes were too small for stratified analyses by potential effect modifiers. For example, previous studies have examined the association of PA and mortality by tumor characteristics including ER status and stage (32). In the present study, we could not adequately consider stratification by ER status or molecular subtypes. Future studies with a large sample size are needed to examine associations by tumor characteristics and molecular subtype.

Our study focused on total PA and too few women were participating in recreational PA to enable us to examine associations for recreational PA only. We were able to examine walking only, and an inverse association was observed in age-adjusted models, that was attenuated after adjustment, although the p-value for trend remained statistically significant. It may be particularly beneficial to examine the contributions of other types of PA, in particular among older women with high levels of comorbidities. We have shown previously that household and transportation-related PA are important sources of PA among Black breast cancer survivors (19). Further, light intensity activities (1.5-<3.0 METs) have been shown to improve function in older cancer survivors (26). Finally, this study did not include data on cardiovascular events, which could be informative in future studies of PA and breast cancer survival.

In summary, our study shows that pre-diagnosis PA is associated with improved survival in predominantly low-income Black cancer survivors. Our study is the first to show a protective influence of PA on survival among Black breast cancer survivors. Future research studies of PA and survival outcomes with multiple measure of PA over time before and after diagnosis among underserved and minority breast cancer survivor populations are needed. There is a key need to reduce barriers to regular PA participation among Black women using a multi-faceted approach that considers comprehensively the role of social determinants including structural inequities and living environments (11). Our study provides further support for the importance of PA in reducing mortality, highlighting the critical need for continued and increased resources for education and intervention efforts to promote and support PA participation among Black women in the United States.

ABBREVIATIONS

BMI

Body mass index

CI

Confidence interval

ER

Estrogen receptor status

HR

Hazard ratio

HER2

Human epidermal growth factor receptor 2

NDI

National Death Index

MET

Metabolic equivalents

PA

Physical activity

PR

Progesterone receptor status

SEER

Surveillance, Epidemiology, and End Results

SCCS

Southern Community Cohort Study

Data Availability Statement

Access to SCCS data is overseen by the SCCS Data and Biospecimen Use Committee, which meets on regular basis. Researchers may gain access to the SCCS data by submitting a request through the SCCS Online Request System with more details available at: https://www.southerncommunitystudy.org/research-opportunities.html