The QOPI methodology reflects a balance of feasibility and sustainability with meaningful, valid data. QOPI data are collected through retrospective chart reviews conducted twice per year. Data collection occurs during a defined time period, during which practices are given several weeks to pull and abstract charts.
Unit of Analysis
The default unit of analysis for QOPI is the practice setting. Practices with multiple offices may choose to collect data at multiple sites. Also, practices are offered the option of collecting physician-level data, although abstracting a sufficient number of charts per physician is not practical for many groups.
Data abstractors in the pilot range from administrative staff to physicians (who were instructed not to abstract their own charts). In most practices, data are abstracted by research nurses; however, efforts are taken to define the QOPI measures clearly enough that administrative staff can assist in abstraction. Training is offered for each abstracter before their participation in QOPI, and assistance is available from ASCO staff as needed.
To select charts for abstraction, practices are instructed to generate a list of patient visits for the preceding 6 months (back to the date of the previous QOPI data collection). Proceeding sequentially backwards along this list, charts for a total of 85 patients meeting basic criteria (e.g., an invasive cancer diagnosed within the last 5 years) are selected. Because QOPI includes measures specific to breast and colorectal cancers and lymphoma, minimum requirements for the number of charts abstracted are provided for those diagnoses. Also, because QOPI includes measures of end-of-life care, practices pull and abstract charts of patients who died within the previous 6 months.
To enter data, abstracters log in to a secure online data entry and reporting system developed and hosted by ASCO. Only anonymized data are submitted; practices maintain a key to link the QOPI unique identifier with patient identification information.
The QOPI data entry system leads the abstracter through the process of entering chart data. As the abstracter enters the answer to an initial question, additional items open depending on the response provided. For example, if the abstracter responds “yes” to the question, “Is this patient dead as a consequence of cancer?” an additional question will open automatically: “Is there a practitioner's notation documenting the patient's physical pain or lack thereof on his/her last visit to the office before death?”
Because only relevant questions are visible during abstraction, the abstracter must complete all fields before the chart can be submitted. The data entry system provides a user-friendly interface for the abstracters, and helps ensure data integrity by enforcing the intended skip patterns and eliminating the problem of missing data.