Statistical analysis
A descriptive analysis was performed. Mean/standard deviation & median/ interquartile range or frequency & percentage were reported for continuous and categorical variables respectively. Comparisons of demographic and clinical characteristics by lung metastases status were conducted using Pearson Chi-square and Wilcoxon rank-sum tests appropriately.
The incidence of lung metastases among hypo-pharyngeal cancer patients was defined as the number of hypo-pharyngeal cancer patients diagnosed with lung metastases divided by the total number of patients with hypo-pharyngeal cancer. The 95%CI of incidences was reported using the Wilson method. We also calculated incidences along with 95%CIs by age group, race, sex, histological grade, AJCC stage, tumor size by groups, metastases, and primary site.
A multivariable logistic regression model was conducted to evaluate the association between lung metastases and age, race, sex, histological grade, tumor size and primary site where both age and tumor size were treated as continuous covariates. Presence of bone, brain, or liver metastases at the time of cancer diagnosis is available in the SEER database and was used to characterize the extent of systemic disease among patients in this study.
In addition we performed a multivariable Cox Proportional Hazards regression model to estimate survival time between patients with or without lung metastases during a five year follow-up period while controlling for the covariates listed above. Subjects who died of other non- hypo-pharyngeal cancers and those who lost follow-up were censored.
Almost 35% of patients were missing tumor sizes; as a result multiple imputation with 100 imputed datasets was performed. The results of these models with or without multiple imputation were consistent and complete case model results were reported.
All statistical analyses were performed using STATA statistical software, Version 16.0 (College Station, TX) and R version 3.5.0. Statistical significance was defined as two-sided alpha level at 5%.
This study was approved by the institutional review board at Zhongshan Hospital of Fudan University; written and informed consent was waived.