(M-122) Overcome the Challenges of NONMEM PopPK Dataset for Late Stage Daily Dose Study

Jing Su, NA – Sr. Director. Statistical Programming, Merck & Co., Inc.; Mansur Kazi, NA – Statistical Programmer, Cytel Inc.; Richard Moreton, NA – Associate Principal Programmer, Merck & Co., Inc.

Abstract: Create PopPK dataset for late stage daily dose study is much more difficult that early stage single or multiple dose clinical trials, because of large number of participants and long term study period requires to collect medication history through paper-based medication diaries. The dosing data of late stage daily dose study is big and most of dose history are collected at home through diary, only the last dose prior to PK samples is collected at clinic. Because of long dose history and dose entries are folded with start and end datetime, that is why ADDL (additional dose) and II (dose interval) come to play an important role to reveal the entire dosing history given NONMEM software requires certain data structure. There is no single formula to derive ADDL, it depends on the study dosing interval and how EX data is collected. Another challenge and important aspect of creating NONMEM PopPK data for daily dose study is to make sure the individual standalone dose record is available and present before PK record. This paper will show you how to correctly derive ADDL and control the quality for daily dose PopPK dataset, as PopPK dataset quality is critical, the inappropriately built dataset can cause proc mixed model not able to converge and the inference draw from the model might not suitable.

Introduction: For phase 3 clinical trials with daily dosing regimen, some company collect dose history solely on study medication eCRF form, while majority of company collect dose history on both medication and PK eCRF form. This paper will focus on the design that collect dose history on both medication and PK form.

Use example to show how to make standalone dose records and derive ADDL:
Step 1: Provide example EX and PC raw dataset;
Step 2: Expand interval dosing records in EX domain into individual daily records, and merge with PC by date of daily dose from EX domain and PCRFTDTC from PC domain;
Step 3: Collapse dose records to create final PopPK dataset;
Step 4: Expand final PopPK dataset by ADDL, derive time since last dose based on dose history on final PopPK dataset and compare with time since last dose derived from collected PCRFTDTC;
--- Provide SAS code and example records.

Conclusion: Having the standalone dosing records preceding PK sample, and correctly derived ADDL and ii are critical to PopPK data quality. The way of data entry is different between studies, but you can rely on the general rules and SAS code provided above to control NONMEM PopPK dataset quality, regardless of your role as dataset developer or QC'er.

Citations: Shuqi Zhao, 2021. “Imputation for Missing Dosing Time in NONMEM PopPK Datasets”, Proceedings of PharmaSUG 2021.