Longitudinal association of physical activity with quality of life domains among malaysian breast cancer survivors.

Introduction
Breast cancer (BC) is Malaysia's most commonly diagnosed cancer, accounting for 32.9% of newly diagnosed cases among women in 2020 [1]. Over the past three decades, BC's incidence and survival rates have increased due to advancements in treatments and early detection [2]. As a result, the number of BC survivors has also increased, highlighting the need to understand and address their quality of life (QOL) during the extended survivorship period.
Quality of life is a multidimensional construct encompassing various physical, emotional, social, and functional well-being [3] aspects. Impaired health-related QOL occurs when there is a gap between an individual's ideal standard and their actual functional level [4]. Previous studies conducted in Malaysia have identified several factors, including ethnicity, age, marital status, time since diagnosis, type of surgery, cancer stage, complementary medication use, and employment, as important determinants of QOL among BC survivors [5–9].
A substantial body of evidence from Western studies has established PA as a crucial modifiable factor influencing the QOL of BC survivors. Studies have reported that BC survivors who engage in moderate-to-vigorous PA experience improved QOL, while those who lead sedentary lifestyles have reduced functioning scores [10]. Additionally, high levels of PA have been associated with lower fatigue and pain scores and higher sexual functioning scores among BC survivors [11]. Combined aerobic and resistance training has also been shown to improve QOL in BC patients effectively [12].
Physical activity has been shown to improve QOL in BC survivors by enhancing physiological function and metabolic health, which contribute to better physical well-being and reduced treatment-related fatigue [13]. Aerobic and resistance exercise further improve physical fitness, and functional capacity, supporting the physical domain of QOL [14]. Collectively, these findings highlight the integral role of PA in promoting better physical and overall quality of life among BC survivors.
Recognizing the significant benefits of PA, organizations such as the National Cancer Society Malaysia recommend that cancer survivors engage in at least 2.5 h per week of moderate-intensity aerobic exercise or 1.25 h per week of vigorous-intensity aerobic activity, along with 2–3 resistance training sessions [15–17].
However, BC survivors often experience reduced PA levels following diagnosis, primarily due to treatment adverse effects like pain and upper body weakness [18–20]. Furthermore, PA levels tend to decrease over the follow-up period, indicating a need for interventions to promote and sustain PA among BC survivors [21]. Malaysian studies have also revealed that BC survivors have lower PA levels compared to the healthy population, and barriers to exercise vary among different ethnic groups [22, 23]. In the general population, PA participation is lower among females than males (66.6% vs. 72.2%), decreases with age, and varies by ethnicity, with Chinese adults having the highest prevalence of inactivity—while urban dwellers are less active than rural residents [24, 25].
Despite the growing body of evidence supporting the positive association between PA and QOL among BC survivors, several important uncertainties remain in the existing literature. First, the directionality of this relationship over time is not well established; it remains unclear whether increases in PA lead to subsequent improvements in QOL, or whether survivors with better baseline or evolving QOL are more likely to maintain or increase their PA levels. Second, longitudinal findings across specific QOL domains are inconsistent, with some studies reporting sustained benefits of PA on physical and role functioning, while others show null or time-limited effects on emotional, cognitive, or symptom-related domains [26]. Third, most longitudinal evidence originates from Western populations, and data from Asian and middle-income settings remain sparse, limiting the generalisability of findings to contexts with different cultural norms, healthcare systems, and survivorship support structures [10, 11, 27–30]. These gaps highlight the need for longitudinal studies that can better characterise the temporal and domain-specific associations between PA and QOL among BC survivors in Malaysia. Therefore, this study aimed to examine the longitudinal association between changes in PA and changes in different QOL domains among Malaysian BC survivors over a three-year follow-up period. We hypothesized that higher level of PA over time would be associated with improvements in QOL among participants.

Methods
This study is part of the larger Malaysian Breast Cancer Survivorship Cohort (MyBCC), a hospital-based prospective cohort study conducted at the Universiti Malaya Medical Centre (UMMC). MyBCC explores various aspects of survivorship among newly diagnosed, multiethnic BC survivors treated at UMMC, examining the relationship between socioeconomic factors, body composition, lifestyle factors, psychosocial characteristics, return to work, use of complementary and alternative medicine, overall survival, and QOL among multiethnic BC survivors newly diagnosed and treated at UMMC [31]. Building on an initial study that focused specifically on PA levels and related factors [32], this study expands the scope to examine the association of PA and QOL levels among the BC survivors.
Analyses included all available observations from patients with complete data for one or more assessment points (baseline, one year, and three years post-diagnosis) between 2013 and 2020. The sampling method used was universal sampling. All eligible subjects who consented were recruited. The QOL of the participants was assessed using the European Organization for Research and Treatment of Cancer Core Quality of Life Questionnaire (EORTC QLQ-C30, version 3) and the EORTC Quality of Life Questionnaire for Breast Cancer (QLQ-BR23) [33]. The questionnaires were available in English and Malay, and participants could choose their preferred language. The QLQ-C30 measured various aspects of QOL, including global health status, functional scales, and symptom scales. The EORTC QLQ-C30 questionnaire consists of 30 items, organised into five functional scales (physical, role, emotional, cognitive, and social) and nine symptom scales (fatigue, nausea and vomiting, pain, dyspnea, insomnia, appetite loss, constipation, diarrhoea, and financial difficulties), along with a Global health status scale [33]. Additionally, the QLQ-BR23 questionnaire comprises 23 items covering functional and symptom scales. The functional scale assesses body image, sexual functioning, sexual enjoyment, and future perspective. In contrast, the symptom scale evaluates systemic therapy side effects, breast symptoms, arm symptoms, and emotional distress related to hair loss [33].
Data gathered from the questionnaires underwent scoring procedures. Each subscale's raw score was computed and then transformed linearly into a standardised raw score ranging from 0 to 100, adhering to the EORTC scoring manual guidelines. Higher scores in functional scales (e.g., physical functioning, role functioning) reflect better or healthier functioning levels. Similarly, a high score in the global health status/summary domain score indicates better QOL. Conversely, higher symptom scale scores correspond to more severe symptoms and therefore reflect a lower QOL. The scoring methodology for QLQ-BR23 is identical to the functional and symptom scales in QLQ-C30, maintaining consistency in principles. The PA levels were measured using the Global Physical Activity Questionnaire (GPAQ) or its validated Malay version. The GPAQ assessed PA in three domains: work, transportation, and recreational activities. The participants' PA levels were classified as active (≥ 600 MET (metabolic equivalent of task)-min/week) or inactive (< 600 MET-min/week) according to the World Health Organization (WHO) recommendations [34]. Other independent variables collected were sociodemographic factors (age, ethnicity, marital status, income, education), clinicopathological factors (cancer stage, types of surgery), and treatment information. Body Mass Index (BMI) was categorized according to the World Health Organization (WHO) Expert Consultation guidelines for Asian populations. Individuals with a BMI of less than 18.5 kg/m2 were classified as underweight. Those with a BMI between 18.5 and 22.9 kg/m2 were considered to have normal weight. A BMI ranging from 23.0 to 27.4 kg/m2 was categorized as overweight, while individuals with a BMI of 27.5 kg/m2 or higher were classified as obese [35].
Cancer stage was categorized into two groups: Stage 1–2 (early-stage) and Stage 3–4 (advanced-stage), a classification commonly used in clinical and survivorship research to reflect meaningful differences in prognosis, treatment burden, and QOL outcomes [36].

Ethics approval and consent to participate
The study protocol received ethics approval from the Universiti Malaya Medical Centre (UMMC) Ethical Committee (MEC number 896.150). Written informed consent was obtained from all participants prior to initiating any study procedures. All aspects of the study were conducted in strict compliance with the Malaysian Good Clinical Practice Guidelines and the principles of the Declaration of Helsinki.

Statistical analysis method
Descriptive statistics summarised patient characteristics, and statistical tests (t-tests, chi-square or Fisher exact tests) compared variables between inactive and active groups. The scores for each QOL scale and the summary score were presented in mean and standard deviation (SD) for each time point. R package ‘PROscorer’ version 0.0.1 was used to generate the scores for QLQ-C30 [37]. The QLQ-BR23 scores were calculated according to the guidelines of the EORTC scoring manual [38]. Each QOL scale was analysed separately.
Generalised Estimating Equations (GEE) were used to examine three types of longitudinal associations: (1) changes in PA over time, (2) changes in QOL over time, and (3) the association between PA and specific QOL domains across three time points (baseline, one year, and three years). Time was treated as a categorical variable to account for potential non-linear changes throughout the follow-up period. PA was modelled as a time-varying variable, reflecting that each participant’s PA level could change across different timepoints.
Univariable and multivariable GEE models with a Gaussian distribution were applied to evaluate the relationship between PA and QOL outcomes. In the univariable models, the effects of PA and time on each QOL domain were assessed separately. Given the large number of QOL domains assessed, multivariable models were subsequently constructed only for domains that showed preliminary evidence of association with PA in univariable analyses. This approach allowed the analysis to focus on the most relevant outcomes while maintaining clarity in interpretation. Specifically, role functioning and constipation met this criterion and were further examined using multivariable GEE models. Independent variables with a p-value < 0.25 in the univariable analysis were considered for multivariable modelling [39]. A stepwise selection procedure, guided by the Corrected Quasi-likelihood under the Independence Model Criterion (QICC), was applied to identify the best-fitting and most stable final model. The correlation structure for each model was chosen based on the lowest Quasi-likelihood under the Independence Model Criterion (QIC). An exchangeable correlation structure was used for the GEE model evaluating PA over time, while an independence correlation structure was used for the models examining QOL over time and the association between PA and QOL. To explore specific differences between time points (baseline vs one year and one year vs three years), we conducted pairwise post-hoc comparisons using the sequential Bonferroni correction to control for multiple testing.
All GEE analyses were conducted using the geepack package in R statistical software version 4.4.1 [40]. Robust (sandwich) standard errors were used by default to account for within-subject correlation and to ensure valid inference even if the working correlation structure was misspecified. Statistical significance was defined as a two-sided p-value < 0.05. No multicollinearity was detected based on variance inflation factors. Interactions among covariates (age, sex, cancer type) were non-significant. Missing data were handled using all available observations, as GEE accommodates participants with incomplete follow-up.

Sample size calculation
A sample size of 42 was initially calculated to detect a small effect size (f = 0.20) for the association between PA and QOL, using repeated measures within-between subjects ANOVA [41], with power set at 0.80 and alpha at 0.05. This test was chosen because it closely matches the GEE available in GPower, both belonging to the general linear model family. The effect size aligns with Cohen’s conventional thresholds, where f = 0.10 is considered small, f = 0.25 medium, and f = 0.40 large [42]. Accounting for potential 20% attrition or ineligible data, the adjusted required sample was 51.
However, given the actual available sample of 133 participants who completed all three timepoints, this study is sufficiently powered to detect a medium effect size (f = 0.25), which reflects a clinically meaningful change in QOL associated with PA [43, 44]. In behavioral sciences and QOL research, medium effects are often considered practically significant due to the complex, multifactorial nature of outcomes like QOL [44].

Results

Patient characteristics
The study followed 133 women with an average age of 56.89 years (SD = 10.56) at baseline, one year, and three years. The majority of participants were Chinese (50.4%), married (70.7%), had at least a secondary education (73.7%), and were unemployed (60.9%) (Table 1). The majority were postmenopausal (68.4%) and diagnosed with early-stage BC (76.7%). Various treatment modalities were employed, including mastectomy (62.4%), chemotherapy (58.6%), radiotherapy (62.4%), and hormonal therapy (82.7%). Patients’ characteristics were shown to have no significant difference between the inactive (< 600 MET-min/week) and active (≥ 600 MET-min/week) groups at baseline (Table 1).
PA decreased progressively over time, with a decline in the percentage of active patients from baseline (48.1%) to one year (39.8%) and three years (35.3%) post-diagnosis (Table 2). The median total MET min/week was 960 (IQR = 2360) at baseline, which significantly declined to 600 (IQR = 1080) after one year and 360 (IQR = 960) after three years (p < 0.001) (Table 2).

QOL
Overall, BC survivors demonstrated a declining QOL. Global health status, physical functioning, cognitive functioning, and fatigue significantly declined over time (Table 3). Global health status decreased significantly from baseline to three years (p < 0.001). At baseline, the survivors scored 77.84 (SD = 12.89), then the score reduced to 71.99 (SD = 16.83) at one year and 71.09 (SD = 17.00) at three years. Post-hoc tests between groups also showed that the scores significantly differed between baseline and one year and baseline and three years.
The physical functioning scale was 95.09 (SD = 11.49) at baseline, 89.87 (SD = 12.91) at one year and 88.67 (SD = 13.38) at three years. The cognitive functioning scale decreased from 87.34 (SD = 15.64) to 83.21 (SD = 16.22) at one year and 80.58 (SD = 18.82) at three years. The physical and cognitive functioning scales significantly declined (p < 0.001 and p = 0.001, respectively). The scores between baseline and one year and baseline and three years were also significantly different, as shown by post-hoc tests. On the contrary, social functioning exhibited improvement (p = 0.047), though the subgroups did not show significance between groups in the post-hoc analysis.
Insomnia was the most severe symptom as patients reported the highest mean score of 18.55, while diarrhoea and nausea/vomiting were the least reported symptoms by the survivors. The fatigue scale significantly increased (p = 0.006), signifying those patients experienced more fatigue over time. The scores were also significantly different between baseline and one year and baseline and three years.
The results from the EORTC QLQ-BR23 questionnaire showed that systemic therapy side effects and arm symptoms increased significantly from baseline to three years (Table 4). The systemic side effects scores increased from 6.30 (SD = 8.65) at baseline to 9.78 (SD = 9.71) at one year and 10.06 (SD = 10.01) at three years (p < 0.001); the arm symptoms scores increased from 8.60 (SD = 17.65) at baseline to 13.78 (SD = 18.05) at one year and 14.62 (SD = 19.84) at three years (p = 0.010). The scores difference between baseline and one year, and baseline and three years were also significantly different.
While breast symptoms improved over time (p = 0.042), with mean scores decreasing from 9.71 (SD = 16.26) at baseline to 8.77 (SD = 11.80) at one year and 6.52 (SD = 9.02) at three years (p = 0.042), the post-hoc analysis showed no statistical significance between groups. Regarding functioning scales, body image and future perspective had higher scores, while sexual functioning had the lowest scores (Table 4).

Longitudinal association of PA and QOL
When investigating the impact of PA on QOL at various time points, active participation enhances role functioning (p = 0.026). This association translated to a 5.92 increase in role functioning (95% CI: 0.70, 11.10) when individuals were actively engaged (Table 5). Upon adjusting for other independent factors and three follow-up time points in the multivariable models, the positive impact of PA on role functioning persisted. The role function score increased by 6.22 (95% CI: 0.91, 11.50; p = 0.022) when subjects were active, even after adjusting for ethnicity and type of surgery.
For the constipation symptom score, being active at one year corresponded to a significant 6.62 reduction in the constipation score (95% CI: −13.20, −0.10; p = 0.047), indicating a substantial improvement in constipation symptoms (Table 6). The multivariable model also showed that being active at one year improved constipation symptoms, with a 6.62 reduction in constipation score (95% CI: −13.20, −0.06; p = 0.048). The significant improvement of constipation at one year showed that the effect is time specific. However, the overall main effect of PA across all time points was not significant (p = 0.112).
Multivariable models were constructed only for QOL domains that showed preliminary evidence of association with PA in the univariable analyses. Accordingly, role functioning and constipation were included in the multivariable analyses. The remaining QOL domains did not have association with PA at the univariable level and were therefore not examined further in multivariable analyses.

Discussion

Decreasing physical activity levels in breast cancer survivors
BC survivors exhibited low PA levels over time. Initially, 48.1% were active, but this declined to 39.8% after one year and 35.3% after three years. Total activity significantly reduced from 3503.0 (SD = 6838.3) MET-min/week at baseline to 792.5 (SD = 1364.0) MET-min/week three years post-diagnosis.
Compared to similar studies among Caucasian patients showing 21.4% to 39.5% activity post-diagnosis [21, 27, 45], our findings align with Asian populations (35% to 39%) [30, 46]. Declining PA are consistent with European and US studies [21, 47–50]. Unlike some studies showing a "V" shaped PA pattern [51, 52], our participants did not return to pre-diagnosis activity levels. In Hong Kong, a study showed a significant increase in post-diagnosis PA, attributed to a 'teachable moment' after diagnosis [46, 53].
In our study, only a small proportion increased their PA levels post-diagnosis (10% from baseline to one year; 14% from baseline to three years; 19% from one year to three years), lower than previous Malaysian studies (22.6%) [54]. This difference may reflect variations in recruitment settings, sample characteristics, or the time frame following treatment. Those who increased their PA were primarily Chinese and had lower family incomes, similar to the Hong Kong study [46]. Survivors with early-stage BC and lower incomes were more likely to increase PA and continue working post-diagnosis [47].

Longitudinal decline in quality of life among breast cancer survivors in Malaysia
This study investigated the longitudinal trajectory of QOL among BC survivors in Malaysia. While overall baseline QOL scores (77.84, SD = 12.89) indicated a good QOL state, exceeding the reference value of 61.8 [55], a significant decline was observed three years post-diagnosis (71.09, SD = 17.00). This finding aligns with similar studies in Malaysia, demonstrating a high initial physical domain score [6, 7, 9, 56]. Notably, our survivors exhibited more favourable QOL scores on global health status, functional scales, and symptom scales compared to other studies at the three-year follow-up mark [57–59]. This discrepancy may be attributed to our study's higher proportion of survivors with early-stage cancer.
However, longitudinal analysis revealed significant declines in global health status, physical and cognitive function, fatigue, systematic therapy side effects, and arm symptoms over time. While the initial psychological shock and stress following diagnosis and treatment are known to impact QOL, the continued decline deviates from the trajectory previously reported by Ng et al. [8] using the same cohort. Additionally, studies in Australia, Japan, and Shanghai have documented significant improvements in overall QOL at 12- and 36-month follow-ups, respectively [60–62].
Some cohort studies suggest that QOL deterioration over a more extended period may be influenced by factors such as fear of recurrence, subsequent primary cancers, ageing, or long-term hormonal therapy [59, 63]. Ageing populations were expected to experience decreased physical and cognitive functions over time, while systemic therapy side effects and arm symptoms could be exacerbated by hormonal therapy and surgeries, common treatments among Malaysian BC survivors. Conversely, fatigue, a common post-treatment symptom, could persist for up to ten years [64].
Social functioning was the only domain that improved statistically significantly in our study, consistent with findings from Ng et al. [8] in the same cohort. While we did not directly assess social support in the current analysis, a high level of perceived social support reported in our population in the previous study may partly explain this improvement. [8]. Furthermore, peer support services such as those offered by the Breast Cancer Welfare Association may also contribute to enhanced social functioning by providing accessible psychosocial support to patients during follow-up.
In the BC-specific questionnaire, our study participants scored highest in body image and future perspective but the lowest score in sexual functioning, consistent with similar findings in Malaysian and Iranian populations [5–8, 58]. Notably, half of our study population were Chinese, with the majority above 50 years old. Wong-Kim et al. [65] found that Chinese women prioritise survival after diagnosis and place less emphasis on body image. Furthermore, ageing tends to diminish the importance placed on body image. Considering that approximately 70% of the survivors in our study were post-menopausal, hormonal changes and menopausal symptoms could contribute to lower sexual functioning scores.

Longitudinal association of physical activity and quality of life
We hypothesised that increased PA over time would enhance the QOL. After adjusting for relevant sociodemographic and clinical factors in the multivariable analysis, our results suggest that PA was associated with improvements in the constipation and role-functioning domains. Notably, PA demonstrated a positive association with constipation improvement at one year. The significant interaction term of PA and time pointed towards this time-specific effect. This observation coincides with the highest constipation symptom scores reported at one year, suggesting PA's potential effectiveness in mitigating these symptoms by enhancing gastrointestinal function at this critical point. A local cross-sectional study by Zainordin et al. [66] further corroborates this finding.
We observed that higher PA levels were also associated with better role functioning, consistent with previous studies [10, 29, 67, 68]. The role functioning section of the questionnaire assessed survivors' occupational and social roles, encompassing limitations in work, daily activities, or leisure pursuits [69]. We propose that PA aids in regaining strength and mobility [70–72], enabling a return to pre-cancer roles. Additionally, PA's positive influence on mental well-being by releasing endorphins, serotonin, and endocannabinoids [73] might bolster self-confidence and facilitate role resumption. Although the magnitude of improvement (6–10 points) was small, it falls within the range considered clinically meaningful on the EORTC QLQ-C30 scale [74], suggesting modest but relevant benefits.
In contrast to our study, which identified positive associations of PA in two QOL domains, other cohort studies have demonstrated the positive effects of PA across multiple QOL domains over time. In the Shanghai Breast Cancer Survival Study, Chen et al. [75] reported a consistent positive association between PA and QOL at six months and three years post-diagnosis in various domains. Additionally, a cohort study with a five-year follow-up revealed a significant association between regular PA and improved general health status, physical functioning, role functioning, emotional functioning, and reduced symptoms of fatigue, pain, constipation, and loss of appetite [68]. The persistent positive effect of PA on QOL was noted over time in these studies. Our study's comparatively fewer positive outcomes may be attributed to differences in background characteristics and PA levels. Notably, our study has a smaller sample size (133 participants) compared to the studies by Chen et al. [75] and Lei et al. [68], which included over 1000 participants.

Limitations and strengths
This study captured the experiences of BC survivors at diagnosis, post-treatment adjustment at one year, and after lifestyle-habit consolidation at three years. While providing valuable insights, an even longer follow-up could explore conditions in extended survivorship. Recruitment from a single urban centre also limits generalizability to diverse populations, considering potentially differing lifestyles in rural settings. While the MyBCC instrument yielded valuable data, its time-consuming nature contributed to participant fatigue and dropouts, which is a limitation to consider. The COVID-19 pandemic’s impact on the PA of BC survivors should also be considered, as this study included patients who were followed up till 2020. Beginning on March 18, 2020, the nation was placed under lockdown, severely limiting citizen mobility and altering their daily behaviours (such as commuting to work and buying groceries). Therefore, their PA is expected to decrease significantly throughout this confinement period [76]. Some patients may also be lost to follow-up because their appointments were rescheduled and could not be reached by phone. In addition, this study lacked detailed treatment and molecular subtype information, as these data were not collected in the main dataset. As a result, we were unable to explore the potential impact of specific treatment regimens or tumour biology on PA or QOL outcomes. The demographic characteristics of the study population may limit the generalisability of our findings. A high proportion of participants were classified as obese, which—while consistent with national data indicating that over 53% of Malaysian adults are overweight or obese [77]—may not reflect populations with lower BMI distributions. Additionally, most participants were aged 50 years and above (mean 56.9, SD 10.6), limiting the applicability of the findings to younger BC survivors. These factors should be considered when interpreting the broader relevance of our results. Another limitation is PA in this study was assessed using the self-reported GPAQ, which may be prone to recall bias and over- or under-reporting. Another limitation relates to the study design. Although this prospective study demonstrated significant associations between PA and QOL over time, these findings should be interpreted cautiously, as causality and direction cannot be confirmed. The final sample size exceeded the initial power calculation due to continued recruitment, which may increase the likelihood of Type I error; thus, effect sizes and clinical relevance should be considered alongside p-values. Further studies are needed to clarify this association. Focusing multivariable analyses on QOL domains that were significant in univariable analyses may have also overlooked other potentially relevant associations. Consequently, the observed associations with role functioning and constipation may not represent the only domains influenced by PA but rather those that reached statistical significance in this study.
Despite these limitations, the study possesses several strengths. We demonstrate the feasibility of a prospective cohort study among long-term BC survivors in Southeast Asia, where formal survivorship programs are absent. This pioneering effort presents valuable longitudinal data on PA and QOL trajectories for this understudied population. Employing GEE strengthens our findings. GEE provides robust coefficient estimates, handles missing data efficiently, and controls for intercorrelations within repeated measurements [78]. This robust approach enhances the reliability of observed associations and furthers our understanding of potential causal relationships. Using widely validated and reported questionnaires [33, 34, 37] facilitates comparison with studies worldwide. This allows us to gauge similarities and differences in experiences among BC survivors across various nationalities and ethnicities. Additionally, the established nature of these tools ensures scoring accuracy through the availability of comprehensive statistical software and packages.

Conclusion
This study provides valuable information about the longitudinal changes of PA and QOL among BC survivors in Malaysia. PA levels significantly decreased over time, with a decrease in the proportion of active survivors from baseline to three years. Overall QOL declined across multiple domains, including global health status, physical functioning, cognitive functioning, and fatigue. Insomnia was the most prevalent symptom, while breast symptoms improved slightly over time. However, systemic therapy side effects and arm symptoms significantly increased from baseline to three years. Our results suggest a potential link between higher PA levels and better role functioning. Additionally, PA was associated with improved constipation symptoms in one year. This study highlights an urgent need for comprehensive strategies to sustain PA and promote QOL in BC survivors. These may include encouraging BC survivors to remain active through structured exercise counselling, tailored home-based programmes, and integration of PA support into routine follow-up care. Community-based initiatives such as peer-led activity groups and survivorship programmes that address both physical and psychosocial needs may also help survivors sustain social functioning, manage treatment-related symptoms, and maintain overall functional health. Future research should explore effective strategies to promote long-term PA maintenance and investigate mechanisms underlying the observed relationships between PA and QOL domains.