Association between physical activity and quality of life among breast cancer survivors in China: a cross-sectional mediation analysis

Abstract

Objectives

Physical activity is essential for breast cancer survivorship, yet its relationship with quality of life (QoL) remains inconsistent. This study investigated the psychological associations underlying this ‘activity paradox’ by examining the parallel mediating roles of unmet healthcare-seeking and health behaviour needs.

Design

A cross-sectional survey design.

Setting

A tertiary cancer centre in Hunan, China, between May 2025 and November 2025.

Participants

630 breast cancer survivors who had completed primary anti-tumour treatments within the past 1–5 years.

Main outcome measures

Physical activity was assessed using the International Physical Activity Questionnaire-Short Form. QoL was measured using the Functional Assessment of Cancer Therapy-Breast. Unmet needs were evaluated using the Cancer Survivor Profile-Breast Cancer. The parallel mediation model was tested using the PROCESS macro.

Results

The overall association between physical activity and QoL was non-significant (p=0.152), indicating a potential statistical suppression relationship. Physical activity was associated with lower unmet healthcare-seeking needs (B=−1.18, p=0.026). However, higher physical activity was associated with significantly higher unmet health behaviour needs (B=1.90, p<0.001). Through unmet health behaviour needs, physical activity showed a negative indirect association with lower overall QoL (indirect estimate: −1.57, 95% CI −2.96 to −0.25), which statistically counterbalanced the potential positive indirect association through unmet healthcare-seeking needs.

Conclusions

In this cross-sectional cohort, higher physical activity levels showed opposite-direction associations with unmet needs. Higher physical activity was associated with fewer unmet healthcare-seeking needs but with higher unmet needs for advanced lifestyle guidance. These findings suggest that clinical survivorship programmes may benefit from addressing the specific informational gaps reported by active survivors rather than relying solely on generic advice.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study used a relatively large sample of 630 breast cancer survivors from a tertiary cancer centre, providing robust statistical power to detect complex psychological associations.

• The research integrates two distinct theoretical perspectives—Self-Efficacy Theory and the Patient Activation perspective—to systematically examine the ‘activity paradox’ through parallel mediation analysis.

• All measurement tools, including the Cancer Survivor Profile-Breast Cancer and International Physical Activity Questionnaire-Short Form, were rigorously validated for the Chinese population, ensuring high internal consistency and cultural relevance.

• The cross-sectional nature of the data precludes the establishment of temporal precedence or causal relationships between physical activity and unmet needs.

• The use of convenience sampling at a single urban tertiary hospital and the reliance on self-reported physical activity measures may limit the generalisability of findings to rural populations and introduce recall bias.

Introduction

Breast cancer is the most common malignancy among women worldwide,¹ with incidence rates continuing to rise in China.^{2 3} As advances in early detection and therapies have been associated with improved survival rates,¹ the primary goal of clinical care has shifted from solely extending life to addressing the long-term quality of life (QoL) of survivors.⁴ The transition from active treatment to post-treatment rehabilitation represents a vulnerable period. During active treatment, patients receive frequent clinical monitoring and maintain contact with healthcare providers. This support decreases during the rehabilitation phase when survivors return home and manage their health independently. They often report persistent challenges, including physical symptoms like fatigue and lymphedema, alongside psychosocial distress such as fear of recurrence⁵ and financial burden. These unresolved issues constitute unmet supportive care needs.⁶ Research identifies these needs as factors of poor outcomes, where high levels of unmet needs correlate with higher anxiety and lower care satisfaction. Identifying the factors associated with these needs remains a priority for cancer care.⁷

Physical activity is recognised as a standard component of survivorship care, and international guidelines recommend regular exercise for breast cancer survivors.⁸ Physical activity has been associated with multiple benefits, including better physical function and fewer treatment-related side effects,⁹ alongside lower depressive symptoms. While conventional wisdom assumes that higher levels of physical activity correspond to better QoL, the relationship between physical effort and perceived well-being is not always consistent.^{10 11} Clinical observations often reveal an activity paradox where some survivors who engage in high levels of physical activity still report psychological burden and suboptimal QoL.¹² This phenomenon suggests that physical activity alone is not a guarantee for high QoL and that other psychosocial factors may influence this relationship. It is necessary to understand the factors associated with why physical activity is not consistently associated with a lower survivorship burden.¹³

Unmet supportive care needs may serve as a relevant factor. Two theoretical frameworks offer different perspectives on the relationship between physical activity and unmet needs. The Self-Efficacy Theory suggests a positive association in which regular physical activity may provide a mastery experience and may build self-efficacy. Survivors maintaining an exercise routine often report feeling more competent in managing their overall health. This generalised confidence is positively associated with the ability to navigate the healthcare system and communicate with medical professionals, which is associated with lower healthcare-seeking barriers and better QoL.^14–16 The Patient Activation perspective offers a contrasting view. Physical activity requires proactive effort and indicates that a patient is activated.¹⁷ These survivors may become active health managers with higher health literacy and outcome expectations. In this framework, they may desire advanced guidance regarding exercise safety and dietary plans. Standard follow-up care typically provides generic advice, which may reflect a service discrepancy.¹⁸ The active survivor becomes aware of this gap, and this awareness is associated with higher unmet needs, cognitive dissonance and distress, which inversely correlate with QoL.¹⁹ In this study, we focused on two domains of unmet needs from the Cancer Survivor Profile-Breast Cancer (CSPro-BC) scale: healthcare-seeking needs (deficits in navigating the healthcare system) and health behaviour needs (deficits in receiving professional lifestyle guidance).

We propose that these two pathways exist simultaneously. Physical activity may be linked to higher competence and lower healthcare-seeking barriers while also being accompanied by higher-level needs and a greater burden of health behaviour needs. If physical activity is associated with both indirect associations, the total statistical association may appear non-significant, consistent with a suppression relationship,²⁰ in which the negative indirect association statistically counterbalances the positive indirect association. Previous studies using variable-centred approaches like linear regression focus on overall associations and often overlook these opposing internal associations,²¹ leaving a gap in understanding the persistent burden reported by active survivors.

To date, few studies have empirically tested these competing indirect associations within a single framework. To address this gap, we analysed cross-sectional data from a sample of Chinese breast cancer survivors using a parallel mediation model to isolate specific indirect associations. We proposed three hypotheses. First, physical activity is negatively associated with unmet healthcare-seeking needs (supporting the competence indirect association). Second, physical activity is positively associated with unmet health behaviour needs (supporting the activation indirect association). Third, these two distinct unmet needs have opposing associations with QoL. Specifically, we hypothesise that higher unmet health behaviour needs statistically counterbalance the positive indirect association between physical activity and QoL. By elucidating these associations, this study provides an evidence base for precision survivorship care, emphasising the transition from generic advice to addressing the specific guidance needs of active survivors.

Methods

Study design and participants

A cross-sectional survey was conducted between May 2025 and November 2025 at Hunan Cancer Hospital. Participants were recruited from the breast surgery inpatient wards and outpatient clinics using convenience sampling. The inclusion criteria were: (1) pathologically confirmed diagnosis of breast cancer; (2) completion of primary anti-tumour treatments (ie, surgery, chemotherapy and/or radiotherapy, excluding long-term endocrine therapy) within the past 1–5 years; (3) age ≥18 years and (4) ability to read and understand Chinese. Patients with cognitive impairments, severe psychiatric disorders, concurrent other malignancies or those who experienced active disease recurrence during the study period were strictly excluded. The sample size was estimated using G*Power 3.1.9.7²² software based on the parameters for a linear multiple regression model (Fixed model, R² deviation from 0). To ensure the study had sufficient statistical power to detect a small-to-medium effect size (f²=0.05)²³ with 12 potential factors (including the independent variable, mediators and sociodemographic covariates), we set the significance level (α) at 0.05 and the statistical power (1−β) at 0.90. The calculation indicated a minimum requirement of 415 participants. Considering a potential invalid questionnaire rate of approximately 20%, we expanded the recruitment target. Ultimately, a total of 710 questionnaires were distributed. After excluding invalid responses, 630 valid questionnaires were included in the final analysis, which exceeded the required sample size and ensured robust statistical power.

Procedure

Data were collected by two trained oncology nurses and one postgraduate student. To ensure data quality, the investigators explained the purpose of the study and provided standardised instructions before the survey. Participants completed the self-report questionnaires independently in a quiet room, which took approximately 15–20 min. Questionnaires were checked immediately on retrieval, and those with missing data >20% or obvious pattern responding were excluded. Procedurally, we ensured participant anonymity, emphasised that there were no right or wrong answers and used established scales with clear instructions.

Patient and public involvement

Patients and the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Measures

Sociodemographic and clinical characteristics

We designed a structured questionnaire to collect participant data. Sociodemographic variables included age, marital status, education level, occupation, monthly household income and payment method for medical expenses. Clinical variables included pathological stage, treatment modality, time since last clinical follow-up and family history of cancer.

Unmet supportive care needs

We assessed unmet needs using the Chinese version of the CSPro-BC.²⁴ The original instrument contains five domains and 73 items.²⁵ The revised Chinese version underwent rigorous cross-cultural adaptation and validation. The finalised version comprises 71 items across five domains: symptom burden (27 items), function (18 items), health behaviour (4 items), financial strain (4 items) and healthcare-seeking (18 items). Participants rate items on a 5-point Likert-type scale. Previous research demonstrated favourable psychometric properties for this scale. Cronbach’s alpha coefficients for the subscales ranged from 0.65 to 0.92, indicating robust internal consistency.

Physical activity

Physical activity was measured using the International Physical Activity Questionnaire-Short Form (IPAQ-SF).²⁶ This instrument assesses the frequency and duration of vigorous-intensity activity, moderate-intensity activity and walking over the last 7 days. The Chinese version of the IPAQ-SF has demonstrated good test-retest reliability (Intraclass Correlation Coefficient=0.79) and acceptable validity in Chinese populations.²⁷ We converted data into Metabolic Equivalent of Task (MET)-minutes per week according to the official IPAQ scoring protocol. Participants were categorised into three levels. Low level included those not meeting criteria for moderate or high levels. Moderate level required at least 3 days of vigorous activity (≥20 min/day) or 5 days of moderate activity/walking (≥30 min/day) or achieving ≥600 MET-min/week. High level required at least 3 days of vigorous activity achieving ≥1500 MET-min/week or 7 days of any combination achieving ≥3000 MET-min/week. For the purpose of correlation and regression analyses, physical activity levels were coded as an ordinal variable (1=Low, 2=Moderate, 3=High).

Quality of life

To assess the multidimensional QoL in patients with breast cancer, the Functional Assessment of Cancer Therapy-Breast (FACT-B) scale was used.²⁸ The scale consists of 36 items across five subscales: physical well-being, functional well-being, emotional well-being, social/family well-being and breast cancer-specific concerns. Each item is scored on a five-point Likert-type scale (0=not at all to 4=very much). The FACT-B has been translated into Chinese. Previous studies reported that the internal consistency of the Chinese version for these five domains ranged from 0.59 to 0.85.²⁹

Statistical analysis

Data analyses were performed using IBM SPSS Statistics V.26.0 (IBM Corp). Prior to analysis, continuous variables were assessed for normality using skewness and kurtosis indices (all values were strictly within the ±2 range) to justify the application of parametric tests. Missing data were handled using mean imputation at the subscale level, as preliminary analysis confirmed the data were missing at random (MAR). Continuous variables were described using mean±SD and categorical variables using frequency (n) and percentage (%).

Control for common method bias: since all data were self-reported, we employed both procedural and statistical remedies to minimise common method bias.³⁰ Statistically, we conducted Harman’s single-factor test. The exploratory factor analysis showed that the first factor accounted for 42.64% of the total variance, which is below the recommended threshold of 50%. This indicates that common method bias was not a pervasive issue in this study.

Hypothesis testing: univariate analyses (t-tests and analysis of variance[ANOVA]) were first conducted to screen for potential sociodemographic and clinical confounders. Pearson correlation analysis was used to examine the bivariate relationships among key variables. Finally, the hypothesised parallel mediation model was tested using the PROCESS macro for SPSS (Model 4) developed by Hayes. Physical activity was entered as the independent variable (X), QoL as the dependent variable (Y) and the two unmet needs domains as parallel mediators (M1, M2). The model was adjusted for age, education level, current endocrine therapy status and surgical modality. We used the bootstrapping method with 5000 resamples to estimate the indirect pathways.³¹ An indirect estimate was considered statistically significant if the 95% bias-corrected CI did not include 0. Finally, a sensitivity analysis was conducted using complete-case analysis (listwise deletion, n=604) to strictly verify the robustness of the mediation model against the mean imputation method (detailed in online supplemental table S1).

Results

Participant flow and missing data

A total of 710 questionnaires were initially distributed, of which 685 were returned. On rigorous screening, 55 responses were completely excluded due to obvious pattern responding or missing data exceeding 20%. To assess potential attrition bias, we compared these 55 excluded participants with the final cohort and found no significant differences in age or basic clinical staging (p>0.05). The final analysis yielded 630 valid questionnaires (an effective response rate of 88.7%). Within this final sample, missing data were minimal (26 cases, 4.1%). Independent t-tests and χ² tests confirmed no significant demographic or clinical differences between participants with complete data and those with missing data (all p>0.05), satisfying the MAR assumption for subsequent imputation.

Baseline characteristics

The demographic and clinical profiles of the final cohort are detailed in table 1. The mean age of participants was 51.00 years (SD=9.91). The majority were married (95.6%) and had an education level of middle school or below (76.0%). Clinically, 77.9% had completed primary treatment within the past year. Univariate analyses (t-tests and ANOVA) were conducted to identify demographic factors associated with QoL. The results indicated that marital status was significantly associated with QoL (F=3.26, p=0.021). However, this specific nominal association is statistically fragile and should be interpreted with caution due to the highly imbalanced group distribution (Married n=602 vs Unmarried/Divorced/Widowed n=28). Other sociodemographic and clinical variables, including age, education, income and physical activity level, did not show statistically significant associations with QoL in these univariate comparisons (p>0.05). Covariates included in the final parallel mediation model were age, education level, surgical modality and current endocrine therapy status. Age and education were selected a priori based on their established relevance to health literacy and overall QoL in cancer survivorship.³² Additionally, surgical modality and endocrine therapy were adjusted for due to their well-documented associations with physical symptom burden and psychosocial well-being.^{33 34} Although marital status showed a preliminary univariate association with QoL, it was excluded from the multivariable model due to extreme group imbalance (Married 95.6% vs Others 4.4%) to prevent statistical instability and potential overfitting.

Table 1. Characteristics of participants (n=630).

Variables	Category	n (%)	FACT-B score (Mean±SD)	F/t	P value
Age (years)	–	630	51.00±9.91	−0.012	0.768
Ethnicity	Han Chinese	615 (97.6)	95.23±18.47	0.228	0.820
	Ethnic minorities	15 (2.4)	94.13±16.37
Marital status	Unmarried	2 (0.3)	78.50±9.19	3.259	0.021*
	Married	602 (95.6)	94.86±18.41
	Divorced	16 (2.5)	101.50±17.47
	Widowed	10 (1.6)	109.40±13.50
Education level	Middle school or below	479 (76.0)	95.07±18.23	0.137	0.872
	High school/technical	45 (7.1)	94.67±22.49
	College or above	106 (16.8)	96.03±17.49
Occupation	Unemployed/retired	200 (31.7)	93.36±18.90	1.442	0.229
	Farmer/worker	166 (26.3)	95.84±17.51
	Employed/professional	240 (38.1)	96.61±18.63
	Other/unknown	24 (3.8)	92.04±17.65
Monthly income (RMB)	<2000	88 (14.0)	97.02±18.05	0.847	0.468
	2000–5000	204 (32.4)	96.13±17.46
	5000–10 000	271 (43.0)	94.32±18.94
	>10 000	67 (10.6)	93.54±19.61
Time Since Last Clinical Follow-up	<12 months	491 (77.9)	94.71±18.33	0.910	0.403
	12–36 months	81 (12.9)	97.63±19.02
	>36 months	56 (8.9)	95.82±18.41
Medical insurance	Employee-basic	215 (34.1)	95.60±17.00	1.302	0.273
	Resident-basic	346 (54.9)	94.37±18.98
	Self-pay/other	68 (10.8)	98.19±19.80
Family history	No	503 (79.8)	95.49±18.22	0.760	0.445
	Yes	127 (20.2)	94.09±19.22
Cancer stage	Stage I	122 (19.4)	92.95±19.31	1.004	0.405
	Stage II	180 (28.6)	95.70±18.53
	Stage III	114 (18.1)	94.68±19.08
	Stage IV (stable)	42 (6.7)	99.07±15.44
	Unknown/unverified	172 (27.3)	95.68±17.85
Surgical modality	Mastectomy	441 (70.0)	94.83±18.45	−0.771	0.440
	Breast-conserving	189 (30.0)	96.07±18.37
Hormonal therapy	No	200 (31.7)	97.03±19.05	1.696	0.090
	Yes	430 (68.3)	94.36±18.07
Physical activity	Low	159 (25.2)	97.33±17.44	1.427	0.241
	Moderate	395 (62.7)	94.51±18.53
	High	76 (12.1)	94.37±19.65

Comparisons were performed using independent sample t-tests (t) or one-way ANOVA (F).

^*p<0.05.

RMB, Renminbi, the official currency of China.

Correlations among key variables

Table 2 presents the descriptive statistics and bivariate correlations. A key finding emerged regarding the relationship between physical activity and unmet needs. Physical activity showed a moderate, positive correlation with unmet health behaviour needs (r=0.415, p<0.001), suggesting that more active survivors experienced a greater deficit in health guidance. Conversely, physical activity showed a weak, negative correlation with unmet healthcare-seeking needs (r=−0.087, p<0.05). Notably, the direct bivariate correlation between physical activity and QoL was not statistically significant (r=−0.058, p>0.05), hinting at a complex underlying association.

Table 2. Descriptive statistics and bivariate correlations of study variables (n=630).

Variables	Mean±SD	1	2	3	4	5	6	7
1. Age	51.00±9.91	1
2. Physical activity	1.87±0.60	0.002	1
3. Symptom burden	76.65±18.34	0.034	0.042	1
4. Function needs	50.56±11.57	0.068	0.025	0.575***	1
5. Health behaviour needs	8.74±2.75	0.028	0.415***	0.467***	0.491***	1
6. Financial strain	12.08±4.73	0.05	0.057	0.516***	0.439***	0.463***	1
7. Healthcare-seeking needs	55.98±7.95	0.032	−0.087*	0.491***	0.463***	0.451***	0.412***	1
8. QoL	95.20±18.42	−0.012	−0.058	−0.07	−0.133***	−0.168***	−0.097*	−0.138***

The correlations shown are Pearson correlation coefficients.

*p<0.05, **p<0.01, ***p<0.001.

QoL, quality of life.

Parallel mediation analysis: unpacking the suppression relationship

We used the PROCESS macro (Model 4) to test the hypothesised parallel mediation model (figure 1). The model summary and specific estimates are detailed in table 3. The model explained 18.2% of the variance in unmet health behaviour needs (R²=0.182). The analysis revealed a complete suppression relationship, characterised by the following indirect associations:

Figure 1. The parallel mediation model. Note: This figure illustrates the relationship between physical activity (X) and quality of life (Y), mediated by unmet health behaviour needs (M1) and unmet healthcare-seeking needs (M2). The solid lines represent significant indirect associations, while the dashed lines represent non-significant indirect associations.

Table 3. Summary of direct, indirect and total estimates of the associations between physical activity and QoL (n=630).

Path	Estimate	SE	P value	95% CI (LL to UL)
Total association (c)	−1.76	1.23	0.152	(−4.18 to 0.65)
Direct association (c′)	−0.42	1.42	0.768	(−3.20 to 2.37)
Specific indirect associations
Via unmet health behaviour needs (M1)	−1.57	0.70	–	(−2.96 to −0.25)
Via unmet healthcare-seeking needs (M2)	0.23	0.17	–	(−0.05 to 0.62)
Total indirect association	−1.34	0.78	–	(−2.90 to 0.15)
Path coefficients
Path a1 (physical activity→M1)	1.90	0.17	<0.001	(1.57 to 2.23)
Path b1 (M1→QoL)	−0.83	0.35	0.017	(−1.50 to −0.15)
Path a2 (physical activity→M2)	−1.18	0.53	0.026	(−2.22 to −0.14)
Path b2 (M2→QoL)	−0.19	0.11	0.079	(−0.40 to 0.02)

Indirect associations were tested using 5000 bootstrap samples. ‘–’ indicates that p values are not available for indirect estimates; statistical significance is determined if the 95% bias-corrected CI does not contain 0. Model adjusted for age, education, surgical modality and endocrine therapy.

LL, lower limit; M1, unmet health behaviour needs; M2, unmet healthcare-seeking needs; QoL, quality of life; UL, upper limit.

Total association: the total association between physical activity and QoL was non-significant (estimate=−1.76, p=0.152), consistent with the correlation results. The negative ‘Activation’ indirect association: physical activity was strongly and positively associated with unmet health behaviour needs (B=1.90, p<0.001). In the model, these unmet needs were significantly and negatively associated with QoL (B=−0.83, p=0.017). The bootstrapping analysis confirmed a significant negative indirect association through unmet health behaviour needs (estimate=−1.57, 95% CI −2.96 to −0.25). This suggests that the informational burden associated with physical activity was inversely correlated with QoL.

The positive ‘Competence’ indirect association: physical activity was significantly and negatively associated with unmet healthcare-seeking needs (B=−1.18, p=0.026). However, the downstream association of these needs with QoL was not statistically significant in the full model (B=−0.19, p=0.079). Consequently, the indirect association through unmet healthcare-seeking needs was negligible (estimate=0.23, 95% CI −0.05 to 0.62)

In summary, the negative indirect association through health behaviour needs completely counterbalanced the potential beneficial associations of physical activity, resulting in a null total association. Furthermore, the sensitivity analysis using complete-case data (n=604) confirmed that this complete suppression relationship and all significant indirect associations remained robust (online supplemental table S1).

Discussion

This study investigated the complex relationship between physical activity and QoL in Chinese breast cancer survivors using a parallel mediation framework. The results challenge the conventional clinical assumption that ‘more physical activity is always accompanied by better QoL’ by identifying a statistical ‘suppression relationship’.²⁰ Specifically, we found that physical activity was significantly associated with higher unmet needs for health behaviour guidance, which appeared to counterbalance the potential benefits of exercise. This discovery provides a psychological perspective on the ‘activity paradox’.¹²

The activation paradox: high engagement, high needs

The most significant finding is the strong negative indirect association through unmet health behaviour needs. Our model explained 18.2% of the variance in this domain, suggesting a robust link between physical activity and the perceived need for professional guidance. Previous research often assumes that active survivors are self-sufficient and viewed as ‘model patients’ who require less support.^{32 35} Our findings are inconsistent with this view: higher physical activity was associated with significantly higher unmet needs for lifestyle guidance. In the model, these unmet needs were linked to lower QoL. This phenomenon aligns with the ‘Patient Activation’ framework in health psychology. Engagement in physical activity may reflect a shift in the patient’s mindset. Inactive patients are often passive recipients of care, whereas active survivors may become active seekers. These individuals generally possess higher health literacy and invest greater effort in their recovery, which correlates with higher expectations for their healthcare providers. Rather than generic advice like ‘rest well’, they desire specific guidance, such as protocols for exercise safety, detailed dietary plans to prevent recurrence³⁶ and strategies to manage lymphedema during weight training. However, the current healthcare system in China often struggles to meet these advanced needs.³⁷ Post-treatment follow-up typically focuses on medical surveillance, such as detecting recurrence,³⁸ rather than providing comprehensive rehabilitation coaching.¹⁸ This service discrepancy represents a critical ‘guidance gap’. When an active survivor invests effort but receives insufficient feedback, this disconnect may be associated with psychological distress and cognitive dissonance. The survivor behaves proactively but feels unsupported informationally. This unfulfilled desire for guidance is associated with anxiety and fear of recurrence. In this context, physical activity without matching professional support may be associated with greater stress, characterising a ‘vulnerable active’ phenotype. These survivors appear healthy on the outside but report uncertainty on the inside. Current literature indicates that QoL in breast cancer survivorship is a complex concept involving both physical symptom management and psychosocial distress.^{39 40} Within Asian populations, specific healthcare delivery models and cultural backgrounds are associated with how survivors report and manage these unmet needs.⁴¹ These regional care characteristics may relate to our specific finding that high physical activity correlates with an increased demand for professional guidance rather than improved overall well-being. This alignment with recent evidence suggests that survivorship care should be adjusted to fit regional clinical contexts.

The competence indirect association: a positive but insufficient buffer

On the positive side, our study found a negative association between physical activity and unmet healthcare-seeking needs. This finding aligns with Self-Efficacy Theory, suggesting that success in one behavioural domain may generalise to others, a phenomenon known as ‘co-action’ or ‘behavioural spillover’.^{42 43} Survivors who maintain an exercise routine may develop a sense of mastery and trust their ability to manage challenges. This generalised confidence is linked to easier navigation of the healthcare system, better communication with physicians and fewer perceived barriers when seeking medical help. However, this positive indirect association was not significantly associated with overall QoL in our statistical model, as the indirect association through healthcare-seeking needs was non-significant. This pattern suggests a hierarchy of needs during survivorship. Competence in hospital navigation functions as a baseline requirement, acting as a ‘hygiene factor’⁴⁴ or fundamental supportive care need.⁴⁵ While it may be associated with less frustration, it is not necessarily associated with better well-being. In contrast, the need for lifestyle optimisation represents a growth need central to a survivor’s daily identity. If this core need remains unfulfilled, it may dominate their psychological state. Consequently, the negative indirect association through the guidance gap statistically outweighed the positive indirect association through healthcare competence.

Unpacking the suppression relationship

The statistical identification of a suppression relationship is a central contribution of this study. Simple correlation analyses showed no significant link between physical activity and QoL. A superficial interpretation might suggest that exercise is unrelated to well-being in this population. However, the mediation analysis reveals a more complex dynamic in which physical activity is associated with two mediators that have opposing associations with QoL. It is positively associated with competence and simultaneously associated with new, higher-level informational needs. These opposing indirect associations statistically counterbalance each other, resulting in a non-significant total association.²⁰ This finding highlights the limitations of traditional variable-centred approaches, such as linear regression models, which focus on overall associations and often overlook these internal associations.²¹ Our results underscore the necessity of process-oriented research. We must look inside the ‘black box’ to understand how behavioural factors relate to psychological outcomes.⁴⁶

Clinical implications

These findings provide practical implications for survivorship care. The current approach often relies on general advice. It is necessary to develop a tiered support system. Routine follow-up mainly evaluates physical compliance through basic questions about exercise habits. This practice may miss specific patient needs. Survivorship care protocols should add structured informational needs assessments. A high level of physical activity may serve as a clinical marker. It may help identify survivors who require advanced lifestyle guidance.

For active survivors, clinical support should include specific empowerment coaching. However, a recent perspective notes that oncology needs better systems rather than expecting individual clinicians to do everything.⁴⁷ Asking busy frontline clinicians to serve as health coaches is difficult and may be associated with higher provider burnout. Addressing this guidance gap depends on systemic resources. Hospitals can integrate multidisciplinary teams, including rehabilitation nurses and exercise specialists, to develop advanced survivorship care plans.⁴⁸ These plans can target specific goals like returning to work or engaging in sports. Additionally, digital health technologies such as mobile apps provide alternative options to meet these rehabilitation needs when face-to-face resources are limited.⁴⁹

Psychological support is equally important. Active survivors might experience an effort-reward imbalance when their physical efforts lack professional validation. Addressing these psychological and informational gaps through multidisciplinary and digital methods may attenuate the negative indirect association through unmet health behaviour needs. This system-level approach may help reveal a positive indirect association between physical activity and QoL.

Limitations

We acknowledge several limitations. First, due to the cross-sectional design, causal inference cannot be drawn. While we hypothesised that physical activity is an antecedent to higher needs, it is also plausible that survivors with high unmet needs or anxiety engage in more physical activity as a coping strategy. Longitudinal studies are required to verify the temporal order. Second, participants were recruited via convenience sampling from a single tertiary cancer centre in an urban area. These survivors may possess higher health literacy and better access to medical resources compared with the general rural population, introducing selection bias. Consequently, the levels of unmet needs reported here might be underestimated compared with the broader population. Third, although we rigorously adjusted for key sociodemographic and clinical covariates (including age, education, surgical modality and endocrine therapy), residual confounding from other variables cannot be completely ruled out. Importantly, because this study used a pre-designed survey protocol, detailed data on body mass index, specific comorbidities and menstrual status were lacking, restricting our ability to account for these factors. Finally, physical activity and basic clinical characteristics were assessed via self-report, which limited measurement granularity and resulted in a notable proportion of ‘Unknown’ cancer stages.

Conclusion

In conclusion, this cross-sectional study identified a pattern of opposing statistical associations in breast cancer survivorship. Physical activity was associated with fewer unmet healthcare-seeking needs but also with higher unmet needs for advanced health guidance. The indirect association through health behaviour needs was significantly and inversely associated with QoL, whereas the indirect association through healthcare-seeking needs was not significant. This pattern resulted in a non-significant total association between physical activity and QoL. Physical activity may therefore have dual associations with survivor needs. Healthcare providers should consider these opposing associations. Future interventions could provide targeted informational support beyond general exercise recommendations. Longitudinal studies are needed to determine whether addressing these specific needs may be associated with better long-term well-being.

Data availability statement

Data are available upon reasonable request.