Modeling the “causal cascade” may clarify elusive health-related qualify of life findings in oncology trials

Patient-reported outcomes are a cornerstone of modern oncology trials, particularly in advanced cancers such as non-small cell lung cancer (NSCLC), where maintaining quality of life can be as important as extending survival. Yet clinical trials frequently show inconsistent or nonsignificant effects of treatment on health-related quality of life (HRQoL) scales, even when patients report meaningful symptom relief.

A study published in January 2026 in JCO Clinical Cancer Informatics suggests the disconnect may stem less from what is being measured and more from how it is analyzed.

Donald E. Stull, Ph.D., senior director of statistics and psychometrics at IQVIA, examined whether symptoms, functioning and HRQoL follow what he calls a “causal cascade” in patients with NSCLC and metastatic breast cancer. Rather than assuming that individual symptoms directly affect quality of life, Stull argues that the effects are typically indirect and unfold in sequence.

“The treatments are not designed to improve HRQoL,” Stull told Managed Healthcare Executive in an interview. “That is a downstream consequence of reducing tumor size, thus reducing pain or dyspnea, thus reducing fatigue, which allows the patient to increase their physical and social functioning.”

Stull says his interest in the issue dates back more than a decade, when he reviewed clinical study reports that included HRQoL as exploratory end points. “Sometimes they would find a weak relationship between treatment and HRQoL; oftentimes, they did not find any significant relationship,” he says. That recurring pattern led him to explore whether quality of life effects in oncology is more often indirect than direct.

In NSCLC and other cancers, treatment may first shrink tumors or ease disease-related symptoms. Those symptom changes can reduce fatigue, which then improves functioning, ultimately influencing overall quality of life. If analysts look only for a direct treatment-to-HRQoL relationship, Stull says, they may overlook how symptom changes drive downstream effects.

To test this idea, Stull analyzed data from two multicenter randomized trials available through Project Data Sphere: one involving patients with NSCLC and one involving patients with metastatic breast cancer. Both trials used the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30), a widely used ,patient-reported outcome measure that captures symptoms, functional domains and global health status/quality of life.

Using structural equation modeling, Stull evaluated symptom items and functional domains at two time points, estimating direct and indirect relationships simultaneously. The results showed a consistent and logical ordering of effects across both cancers.

Most symptoms did not have significant direct effects on functional domains or the global quality of life scale. Instead, their influence flowed primarily through fatigue. Pain and other symptoms were associated with increased fatigue, and fatigue, in turn, had direct effects on physical and social functioning and on HRQoL.

For NSCLC in particular, where symptoms such as pain and dyspnea can be especially burdensome, the cascade model aligned closely with clinical experience. Stull says that when he shares these findings, they tend to resonate with practicing clinicians because the framework mirrors how they approach patient care.

“This is the way they think about treating patients,” Stull says. “A patient presents with symptoms; the clinician makes an assessment about the cause of those symptoms, prescribes a treatment, and waits to see if the treatment makes the symptoms go away or reduces them.”

Clinicians are not directly treating HRQoL, he explains. Rather, they are targeting the underlying symptoms and then observing whether patients’ lives improve as a result. Quality of life is a downstream outcome of symptom control.

Clinical trial biostatisticians, however, often approach the question differently, Stull says. Their goal is typically to show a difference in quality of life between treatment arms, adjusting for other variables. But, Stull argues, those “other things” — such as pain and fatigue — are not merely factors to control for. They are part of the causal cascade and should be modeled explicitly rather than treated as background covariates.

The analysis also helps explain why HRQoL findings in NSCLC trials and other oncology studies can appear neutral. A treatment may reduce tumor size and thus tumor-related pain, improving quality of life through one pathway, Stull notes, while also causing nausea or diarrhea that worsens quality of life through another. “Taken as a whole, these inverse relationships may cancel each other out,” he says, giving the impression that treatment has no effect on HRQoL, “when in fact it does, but in opposite ways depending on the path of relationships one follows.”

A striking aspect of the study was the consistency of the cascade across both NSCLC and metastatic breast cancer and across two time points. That reproducibility suggests the pattern may reflect a broader process captured by the QLQ-C30 rather than a tumor-specific phenomenon.

The study has limitations. The models were based on single arms of two trials and did not directly test treatment effects within the cascade. In addition, the study did not examine causal relationships among the functional domains themselves or incorporate longer-term lagged effects. Still, Stull argues that explicitly anticipating complex symptom pathways may better capture patients lived experience during treatment.

“These are much more complicated relationships than simply showing mean differences between treatment arms,” he says.

By modeling how pain and fatigue mediate downstream effects, researchers may gain a clearer picture of how therapies influence functioning and quality of life.

“Fortunately, in oncology trials, HRQoL is now considered an important dimension of the patient’s experience of their disease and treatment,” Stull says.