📄 medicine

Age- and Gender-Dependent Symptom Architecture of Cancer-Related Fatigue: A Symptom Cluster Analysis Using the Edmonton Symptom Assessment System

This study reveals that the symptom architecture of cancer-related fatigue varies significantly by age and gender, transitioning from a tightly integrated network in younger patients to a drowsiness-predominant model in older adults, while females exhibit a multidimensional pattern balanced across sleep, depression, and drowsiness compared to males who show a drowsiness-driven profile.

Original authors: Mellar Davis, Alyssa Andrews

Published 2026-08-25

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Mellar Davis, Alyssa Andrews

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

For millions of people living with cancer, the most persistent and draining companion is not the tumor itself, but a profound sense of exhaustion known as cancer-related fatigue. This is not the ordinary tiredness that follows a long day of work or a sleepless night; it is a heavy, persistent weariness that does not improve with rest and interferes with the ability to function. Medical experts have long understood that this fatigue rarely travels alone. It tends to arrive in a package with other distressing symptoms, such as trouble sleeping, feelings of sadness or depression, and a constant sense of drowsiness. While doctors know these symptoms often appear together, the precise way they connect to one another has remained somewhat of a mystery. Specifically, it has been unclear whether the way these symptoms link together changes depending on a patient's age or gender, or if the same pattern holds true for everyone.

A team of researchers at the Levine Cancer Institute set out to map these connections with greater precision. They turned to a standard tool used in palliative care called the Edmonton Symptom Assessment System, which asks patients to rate the severity of nine different symptoms on a scale from zero to ten. By looking at data from hundreds of visits by over four hundred patients, the researchers could see not just how severe each symptom was, but how strongly the presence of one symptom predicted the presence of another. Their goal was to determine if the "architecture" of fatigue—the underlying structure of how these symptoms relate—shifts as people get older or differs between men and women. The findings reveal that the experience of fatigue is not a single, uniform condition, but rather a dynamic phenomenon that changes shape across different groups of people.

The study began by examining a large collection of records from an outpatient palliative care program. The researchers looked at 618 separate visits made by 422 unique patients, focusing on four specific symptoms: fatigue, sleep disturbance, depression, and drowsiness. They analyzed how these four factors moved together, asking whether a high rating in one area reliably predicted a high rating in the others. When they looked at the entire group of patients without separating them, a clear pattern emerged. Drowsiness was the strongest independent driver of fatigue, followed by sleep disturbance and then depression. However, this overall picture masked significant differences that only became visible when the researchers broke the data down by age and gender.

When the researchers separated the patients by age, a distinct transition appeared. In the youngest group, patients between 18 and 49 years old, the symptoms were tightly woven together. In these younger patients, fatigue, sleep problems, depression, and drowsiness were all strongly linked, forming a dense, integrated network where each symptom reinforced the others. As the researchers moved to the middle age group, 50 to 64 years old, the connections began to loosen, and drowsiness started to take the lead as the primary driver of fatigue. By the time they reached the oldest group, patients aged 65 and above, the network had simplified considerably. In these older patients, drowsiness remained the dominant force behind fatigue, but the link to depression had weakened to the point where it was no longer a statistically significant independent factor. Sleep disturbance remained relevant, but the tight, all-encompassing web seen in younger patients had given way to a model where physical sleepiness played the central role.

The differences between men and women were equally striking and revealed two structurally different patterns. For women, fatigue appeared to be a multidimensional experience. Their fatigue was driven by a balanced combination of sleep disturbance, depression, and drowsiness. None of these factors overwhelmed the others; instead, they contributed roughly equally, suggesting that for women, managing fatigue might require addressing all three areas simultaneously. In contrast, the pattern for men was far more focused. For male patients, fatigue was overwhelmingly driven by drowsiness. While sleep disturbance played a modest role, depression did not show a significant independent connection to fatigue in men. This suggests that for men, the feeling of exhaustion is more closely tied to a state of daytime sleepiness or sedation, whereas for women, it is a more complex interplay of physical and emotional factors.

The researchers also looked at whether the relationship between sleep and fatigue changed based on gender, but they found that this specific link remained stable across both men and women. The difference lay not in how sleep affected fatigue, but in how other symptoms, particularly drowsiness and depression, interacted with the picture. When the researchers combined age and gender, the most pronounced version of the "drowsiness-only" pattern appeared in men over the age of 50. Women, even in the older age groups, maintained a more distributed pattern where sleep and drowsiness both contributed, though the influence of depression did fade with age.

These findings suggest that a "one-size-fits-all" approach to treating cancer-related fatigue may not be the most effective strategy. Because the underlying structure of the symptoms changes depending on who is experiencing them, the way doctors assess and manage the problem should likely change as well. For a young woman struggling with fatigue, a treatment plan might need to address her sleep, her mood, and her energy levels all at once. For an older man, the focus might shift more aggressively toward managing medication side effects that cause drowsiness or treating specific sleep disorders, as his fatigue may be less about emotional distress and more about physical sedation. The study does not claim to have solved the problem of fatigue, nor does it prove exactly why these differences exist, but it provides a clear map showing that the landscape of cancer-related exhaustion is not uniform. It varies systematically, shaped by the demographics of the person experiencing it, offering a new path toward more precise and personalized care.

Technical Summary: Age- and Gender-Dependent Symptom Architecture of Cancer-Related Fatigue

Problem Statement

Cancer-related fatigue (CRF) is a highly prevalent and distressing symptom that rarely occurs in isolation, typically co-occurring with sleep disturbance, depression, drowsiness, and pain within symptom clusters. While the Edmonton Symptom Assessment System (ESAS) is widely used to capture these symptoms, the specific architecture of the relationships between fatigue and its correlates (sleep, depression, drowsiness) is not well characterized across demographic subgroups. Specifically, it remains unclear whether the underlying processes driving fatigue differ significantly by age and gender, which limits the ability to implement precision approaches to symptom management and stratified designs for intervention trials.

Methodology

The study employed a retrospective, cross-sectional analysis of ESAS data collected from an outpatient palliative care program.

Data Source: 618 visit-level observations from 422 unique patients (with a subset of 615 observations from 421 patients used for age-stratified analysis after excluding implausible ages).
Measures: Symptom intensity was assessed using the ESAS-revised (ESAS-r) on a 0–10 numerical rating scale. Fatigue (tiredness) served as the outcome variable, while sleep disturbance, depression, and drowsiness served as predictors.
Statistical Approach:
- Correlation: Bivariate associations were assessed using Spearman rank correlations.
- Regression: Multivariable linear regression models were constructed with standardized coefficients ( $\beta$ ) to determine independent predictors of fatigue.
- Stratification: Analyses were stratified by age (18–49, 50–64, $\ge$ 65 years) and gender (male, female).
- Interaction Testing: Interaction terms (e.g., Sleep $\times$ Male) were tested to evaluate effect modification.
- Sensitivity Analysis: A mixed-effects model with patient-level random intercepts was performed to account for repeated observations; results were consistent with primary visit-level models.
- Controls: Multicollinearity was assessed (VIF < 2.0), and complete-case analysis was utilized.

Key Results

Overall Symptom Architecture

In the overall cohort, drowsiness was the strongest independent correlate of fatigue ( $\beta = 0.410, p < 0.001$ ), followed by sleep disturbance ( $\beta = 0.175, p = 0.001$ ) and depression ( $\beta = 0.142, p < 0.001$ ). The model explained 31% of the variance ( $R^2 = 0.31$ ).

Age-Dependent Variations

The study identified a progressive restructuring of the symptom network as age increased:

Ages 18–49: Exhibited a tightly integrated, dense symptom network where sleep, depression, and drowsiness contributed with relatively balanced effect sizes. All symptoms were strongly intercorrelated.
Ages 50–64: Drowsiness emerged as the dominant predictor ( $\beta \approx 0.47$ ), while the coupling between sleep and drowsiness weakened. Depression remained moderately integrated.
Ages $\ge$ 65: The network simplified significantly. Drowsiness remained the strongest predictor ( $\beta \approx 0.37$ ), sleep disturbance re-emerged with a moderate effect, but depression lost its independent statistical significance ( $\beta = 0.06$ , non-significant). The mean off-diagonal correlation among symptoms declined from 0.49 in the youngest group to 0.33 in the oldest.

Gender-Dependent Variations

Distinct structural patterns were observed between sexes:

Females ( $R^2 = 0.329$ ): Demonstrated a multidimensional pattern with balanced contributions from sleep ( $\beta = 0.213$ ), depression ( $\beta = 0.212$ ), and drowsiness ( $\beta = 0.294$ ). All three were statistically significant independent predictors.
Males ( $R^2 = 0.452$ ): Exhibited a drowsiness-predominant pattern. Drowsiness was the primary driver ( $\beta = 0.554$ ), sleep disturbance contributed modestly ( $\beta = 0.226$ ), and depression was not a significant predictor ( $\beta = 0.038, p = 0.51$ ).
Interactions: The Sleep $\times$ Male interaction was non-significant, indicating the core sleep-fatigue relationship is stable across genders; differences lie in the relative contribution of drowsiness and depression.

Key Contributions

Demographic Stratification of Symptom Clusters: The study provides empirical evidence that the architecture of CRF is not unitary but varies systematically by age and gender. It challenges the assumption of a static symptom cluster across all populations.
Identification of Distinct Phenotypes:
- Younger Patients: Characterized by an "affective" network where fatigue is tightly coupled with mood and sleep.
- Older Patients: Characterized by a simplified model dominated by drowsiness, suggesting a shift toward physiologic or medication-related somnolence.
- Females: Characterized by a multidimensional model requiring attention to mood, sleep, and energy simultaneously.
- Males: Characterized by a drowsiness-driven model where depression plays a negligible role.
Methodological Validation: The use of mixed-effects sensitivity analysis confirmed that visit-level associations are robust despite repeated measures from the same patients.

Significance and Claims

The authors claim these findings support a precision approach to fatigue assessment and management in supportive oncology.

Clinical Implications: The study suggests that assessment priorities should be tailored. For females, evaluation should encompass the full cluster (sleep, mood, energy), potentially requiring combined interventions (e.g., CBT for insomnia, depression treatment). For males, the focus should shift toward aggressive screening for sedation, daytime sleepiness, and medication burden, with interventions like deprescribing or stimulant consideration. For older adults, the diminishing role of depression suggests a need to prioritize drowsiness and sleep fragmentation over mood management in fatigue protocols.
Research Implications: The findings argue for the use of moderation models and latent class analysis in symptom science to capture dynamic, context-dependent relationships. Furthermore, the authors suggest that future clinical trials for fatigue interventions should utilize age- and gender-based stratification to reduce signal dilution and improve the detection of treatment effects.

The paper concludes that while the NCCN Guidelines recommend screening for concurrent symptoms, the relative priority of these assessments should be informed by the patient's age and gender to optimize supportive care.