Multilevel Latent Class Modelling of Colorectal Cancer Survival Status at Three Years and Socioeconomic Background Whilst Incorporating Stage of Disease


  •  Wendy Harrison    
  •  Mark Gilthorpe    
  •  Amy Downing    
  •  Paul Baxter    

Abstract

Previous studies investigating survival from colorectal cancer have typically considered potential confounders to include stage of disease. Stage however may lie on the causal path and statistical adjustment with stage as a confounder can then introduce bias known as the reversal paradox. Classification of stage may also be imprecise and incomplete. Modelling using Latent Class Analysis (LCA) may minimise bias by including covariates on the causal path as `class predictors' and by accommodating uncertainty associated with confounder values explicitly via the latent class part of the model. We construct multilevel latent class models to allow for the multilevel structure of the data: patients nested within NHS Trusts. We use a dataset of patients in a large UK regional population diagnosed with colorectal cancer between 1998 and 2004. Death within three years is the outcome. The optimum number of latent classes at patient and Trust level is determined with reference to likelihood-based model-fit criteria. The three-patient five-Trust class multilevel LCA model was chosen. Patient classes were identified as good, reasonable or poor prognosis groups. The impact of stage differed across the patient classes. Socioeconomic background and older age were clearly associated with increased odds of death in all patient classes. Females had significantly decreased odds of death compared with males in the good prognosis class. The five Trust classes identified outlying Trusts, indicating that the standard multilevel model would not have been sufficient to model these data.


This work is licensed under a Creative Commons Attribution 4.0 License.
  • ISSN(Print): 1927-7032
  • ISSN(Online): 1927-7040
  • Started: 2012
  • Frequency: bimonthly

Journal Metrics

  • h-index (December 2021): 20
  • i10-index (December 2021): 51
  • h5-index (December 2021): N/A
  • h5-median(December 2021): N/A

( The data was calculated based on Google Scholar Citations. Click Here to Learn More. )

Contact