Studying the PAGODA Algorithm for Chemotherapy Dose Changes to Prevent Unplanned Treatment Delays

Incidence of unplanned chemotherapy delays

Will employ a generalized linear mixed effects model with logit link function, with a random patient effect to account for clustering of cycles within patients. Will test if additional random effects are needed (e.g., provider or clinic level). Statistical significance will be assessed at the 5% level. Delays will be assessed over cycles 2 through 7, and a delay will be defined as an interval of \> 18 days since day 1 of the previous cycle. An unplanned delay will be defined as any delay that was not prospectively planned by day 3 of the previous cycle.

Time toxicity

Will use health care contact days as a patient-level measure of time toxicity, as previously described by Gupta et al. Health care contact days will be assessed from electronic health record (EHR) data by recording the number of unique days on which the participant has any in-person contact with the health care system (e.g. for an office visit, chemotherapy treatment, laboratory study, emergency department visit, or other health care service). The mean number of health care contact days will be summarized and compared between the intervention arms using an independent two-sample t-test or Wilcoxon rank sum test. To adjust for multiple comparisons, statistical significance will be evaluated at the 1.25% alpha level.

Moderate-to-severe neutropenia

We will assess the patient-level occurrence of ANC less than 1000/mm3 through 30 days after day 1 of the last cycle of protocol treatment. The proportion of patients experiencing ANC less than 1000/mm3 will be summarized and compared between the intervention arms using a Chi-square test for independence or Fisher's Exact test.

Chemotherapy relative dose intensity

Relative dose intensity (RDI) will be individually calculated for 5-FU bolus, oxaliplatin, and infusional 5-FU at the patient-level as the ratio of the delivered dose intensity (dose delivered divided by days elapsed between cycle 1 day 1 and cycle 6 day 14) and the standard dose intensity (standard dose divided by standard treatment interval). The RDI will be summarized and compared between the intervention arms using an independent two-sample t-test or Wilcoxon rank sum test. To adjust for multiple comparisons, statistical significance will be evaluated at the 1.25% alpha level. RDI will be calculated for the entire sample and within treatment intent. For subgroup analyses, statistical significance will be assessed at the 5% level and any positive findings will be considered hypothesis-generating.