Statistical Analyses
Patient demographics and clinical characteristics were summarized by
weight category using means and standard deviations for continuous
variables and frequencies and percentages for categorical variables.
A mixed effects linear regression model was used to determine if there
was a significant change from baseline BMI over time among the cohort.
This model included a fixed effect for BMI and a random effect for the
patient. Covariates included in the model were patient baseline age at
time of transplant, antigen mismatch, sex, relapse, malignant,
ethnicity, TBI, and transplant source. Further, the relationship between
the change in BMI over time and age was explored by including an
interaction term between age and time.
Fisher’s exact test was used to determine if there were significant
differences in the proportion of patients with GVHD (chronic and acute)
and viral infections between the pre-transplant weight categories. The
Kaplan-Meier method was used to estimate the median time to neutrophil
and platelet engraftment and the overall survival rates at 5 years based
on pre-transplant weight category.
Cox proportional hazards models were
utilized to estimate the hazard ratio (HR) for overall survival, in
addition to times to platelet engraftment and neutrophil engraftment,
for patients in the overweight and obese pre-transplant weight
categories versus patients in the normal weight/underweight
pre-transplant weight category. Models were adjusted for age, transplant
source, and mismatch. For overall survival, if patients did not die,
they were censored at the time the data was pulled (i.e. December 31,
2019). For the analyses of time to neutrophil engraftment and time to
platelet engraftment, a cause-specific hazards regression model was used
to account for the competing risk of death for subjects who died before
they could engraft. Patients were excluded if they were missing
engraftment data due to death or transfer of care prior to engraftment.
All HRs were accompanied by their corresponding 95% confidence interval
(CI). SAS Version 9.4 (Cary, NC USA) was used for all statistical
analyses. A p-value < 0.05 was considered statistically
significant.