Skip to main content

Using Markov multistate models to examine the progression of symptom severity among an ambulatory population of cancer patients: are certain symptoms better managed than others?

Jia J, Barbera L, Sutradhar R. J Pain Symptom Manage. 2016; 51(2):232-9. Epub 2015 Oct 15.


Context — Patient-reported assessments of symptom severity can assist providers in monitoring and managing symptoms for cancer patients, which is important for offering patients optimal cancer care. Understanding which symptoms deteriorate at a faster rate over time can help identify areas for improving symptom management.

Objectives — This paper aimed to longitudinally examine the transitions in symptom severity over time and determine which symptoms deteriorate most rapidly.

Methods — This was an Ontario-wide cohort study from 2007 to 2011 of adult outpatients diagnosed with cancer. During every symptom assessment at the cancer center, patients reported their level of severity for each of nine symptoms. A Markov multistate model under an intermittent observation scheme was implemented to examine the progression of symptom severity over time among cancer patients.

Results — This study included 55,883 patients with over 280,000 symptom assessments. The median time between assessments was 29 days, and the majority of patients at least three assessments. The symptoms deteriorating most rapidly over time were fatigue and well-being, whereas the symptom deteriorating least rapidly over time was nausea.

Conclusion — The availability of numerous medications for treating nausea, compared to fatigue and well-being, may be a reasonable explanation for our findings. Alternate management for these symptoms, such as exercise for reducing fatigue, should be investigated to improve patients' quality of life. The use of multistate modeling methods is also unique in the study of symptom progression and provides a more in depth understanding of the likelihood of symptom deterioration and improvement over time.

View full text

Keywords: Cancer Research and statistical methods

×