Baseline Symptom Phenotypes Separated Recovery Paths in Work Injury Rehabilitation
A short, standardized rehabilitation program did not produce a single uniform recovery pattern in injured workers. It produced at least three pain trajectories and two health-related quality-of-life trajectories, and the baseline phenotype assigned at intake tracked closely with who stalled and who improved.
In this prospective cohort of adults with non-catastrophic, work-related musculoskeletal injuries, investigators followed patients through an 8-week community-based rehabilitation pathway in Ontario, Canada. Pain was measured with the Numeric Pain Rating Scale, and health-related quality of life was tracked with the EQ-5D-5L. At intake, participants had already been grouped into four self-report phenotypes from earlier work: Localized/Resilient, Mood-Dominant, Neurosensory-Dominant, and Complex/Central. Those classes were not academic labels alone. They represented distinct clusters of symptom burden and distress that appear to map onto very different biologic and behavioral recovery constraints.
The cohort included 146 enrolled workers, with 134 contributing to pain trajectory analyses and 138 to quality-of-life trajectory analyses. Growth mixture modeling identified three pain patterns: Rapid Recovery in 34.1% of participants, Incomplete Recovery in 42.2%, and No Recovery in 23.7%. For quality of life, 86.2% followed a Good QoL trajectory, while 13.8% showed No Improvement. The shape of those curves matters clinically because it shows that early rehabilitation is not simply a matter of time passing. Some workers improve quickly, a larger share plateau with persistent pain, and a smaller but meaningful subgroup fails to improve at all.
The most favorable baseline class, Localized/Resilient, served as the reference point. Compared with that group, the odds of landing in the pain No Recovery trajectory were higher in every other phenotype: the Mood-Dominant class had an odds ratio of 3.97, the Neurosensory-Dominant class 10.99, and the Complex/Central class 22.67. The pattern became even more striking for quality of life. Odds of EQ-5D-5L No Recovery were 9.00 times higher in the Mood-Dominant class, 7.50 times higher in the Neurosensory-Dominant class, and 150.00 times higher in the Complex/Central class, although the estimate for that smallest subgroup was imprecise.
Work Participation Lagged Behind Symptom Change in the Highest-Risk Group
The rehabilitation signal was not limited to pain scores. Workers in the Complex/Central class also had higher odds of being not fully returned to work, with an odds ratio of 4.18, and higher odds of being coded as requiring ongoing care, with an odds ratio of 5.74, both relative to the Localized/Resilient class. Those administrative outcomes are clinically important because they reflect how symptom burden spills into actual work participation and service needs. In occupational rehabilitation, pain reduction alone is not the endpoint that matters. Functional restoration and durable return-to-work status determine whether the patient is moving toward independence or toward prolonged case management.
The physiologic logic behind these class differences is plausible. Workers with more diffuse sensory symptoms and stronger mood-related burden may be carrying a broader pain processing disturbance, not just a local tissue injury. Central sensitization, altered pain modulation, and psychological distress can each amplify symptom persistence, reduce tolerance for progressive activity, and blunt the response to standard rehabilitation components. In contrast, the Localized/Resilient phenotype likely reflects a narrower injury profile with less systemic interference, making a short, active program more likely to succeed.
One result stood apart from the rest. Self-rated global recovery did not differ significantly between classes. That finding suggests that a single global question may miss the separation captured by repeated pain scores, quality-of-life measures, and administrative indicators. A worker may report some sense of recovery while still having measurable pain persistence or work restrictions, and that mismatch is exactly why multidimensional follow-up is more informative than one broad recovery label.
Intake Phenotyping May Help Identify Who Needs More Than Standard Care
The study’s main clinical contribution is its demonstration that brief, fully self-reported intake measures can sort injured workers into meaningfully different short-term recovery pathways before treatment progress is even underway. This is not a prediction exercise built on complicated imaging or invasive testing. The phenotypes were derived from questionnaire data already available in routine care, which makes the approach practical for occupational rehabilitation settings where decisions must be made quickly and with limited resources.
The data also reinforce the value of trajectory modeling. Static follow-up at a single time point can obscure the fact that two people with similar baseline pain may diverge sharply over 8 weeks, one regaining function and another remaining stuck. By identifying Rapid Recovery, Incomplete Recovery, and No Recovery pain trajectories, the analysis captures clinical course rather than just endpoint status. That is a more honest reflection of rehabilitation medicine, where rate of change often carries as much meaning as the final score.
A practical triage framework emerges from these findings. Workers in the Localized/Resilient class may be suitable for standard protocol-driven care with routine reassessment. Workers in the Mood-Dominant or Neurosensory-Dominant classes may warrant closer monitoring for stalled recovery, more deliberate attention to distress, and a lower threshold for adjusting the rehabilitation plan. The Complex/Central class appears to represent the highest-risk subgroup, with the most pronounced odds of poor pain recovery, poor quality-of-life recovery, delayed return to work, and ongoing care designation.
The investigation still leaves important constraints. The Complex/Central subgroup was small, which makes the very large odds ratio estimates less precise. Follow-up was short, limited to the 8-week program window, so the study does not answer whether the phenotype differences persist, narrow, or widen over longer occupational recovery. Even so, the short-term separation was strong enough to suggest that phenotype-informed rehabilitation could improve risk stratification at intake, especially in systems built around early, active musculoskeletal care.
The next step is not to ask whether injured workers vary in recovery. They plainly do. The harder question is how quickly clinics can convert these simple intake phenotypes into treatment decisions that reduce persistent pain, prevent work disability, and identify the patients who need escalation before recovery stalls.