Distinguishing symptom patterns in adults newly diagnosed with cancer: a latent class analysis

J Pain Symptom Manage. 2022 Apr 20;S0885-3924(22)00692-3. doi: 10.1016/j.jpainsymman.2022.04.172. Online ahead of print.


Context: Socio-demographic differences, including place of residence, socio-economic status, ethnicity, and gender, have been associated with various inequities in cancer care outcomes.

Objectives: The aims were to distinguish subgroups of patients with different symptom patterns at the time of the initial oncology visit and determine which clinical and socio-demographic variables are associated the different symptom patterns.

Method: Responses to the Edmonton Symptom Assessment Scale- revised and clinical and socio-demographic variables were obtained via the Ontario Cancer Registry and linked health data files. Latent class analyses were conducted to identify and compare the subgroups.

Results: The cohort (n= 216 110) with a mean age of 64.5 years consisted of 54.1% women. The analyses identified six latent classes (proportions ranging from 0.09 to 0.31) with distinct symptom patterns, including: 1) many severe symptoms, 2) many less severe symptoms, 3) predominantly mild symptoms, 4) severe psychosocial symptoms, 5) severe somatic symptoms, 6) few symptoms. The subgroups were associated not only with clinical differences (diagnoses and functional status), but also with various socio-demographic (age, sex) and community characteristics (neighborhood income, proportion of foreign born, rurality).

Conclusion: The results indicated that there were substantial differences in symptom patterns at the time of the initial oncology visit, which were associated with both clinical diagnoses and socio-demographic differences. These results point to the importance of taking the social situation of patients into account, and not just diagnosis, to better understand differences in symptom patterns of people living with cancer.

Keywords: Cancer; health equity; latent class analysis; oncology; patient reported outcome measure (PROM); symptom clusters.