Abstracts 9 9 S

Although a one-stage design would have required more patients, a single stage design study would have been completed earlier given the rate of patient accrual. The pros and cons of the two-stage design will be considered in the context of this AMC trial.

P41 A CENTRALIZED MULTI-DISCIPLINARY APPROACH

TO DETERMINING SPECIFIC CAUSE OF DEATH

Susan Sunkle, Victoria Konig, Claire Langston and Mark Schluchter for

The Pediatric Pulmonary and Cardiovascular Complications of Vertically Transmitted HIV Infection

(P2C2) Study Group The Cleveland Clinic

Cleveland, Ohio

Specific cause of death (COD), "the single disease or injury that initiated the events resulting in death"*, is critical in the analysis off outcome data. In P2C2, a multi-centered, NHLBI-funded natural history study of heart and lung disease in children born to HIV-infected mothers, patients were followed prospectively at 10 sites. 110/805 enrolled infants and children died over a 7 year period and COD was self-reported by clinical centers. Because of the complex nature of disease in the P2C2 cohort, a Mortality Review Committee (MRC) was established mid-way through the project to conduct a centralized death review.

Death certificates (79/110172%1), autopsy reports (39/110135%]), relevant discharge summaries (87/110179%]), and a detailed patient summary from a physician or nurse familiar with the child near death (35/110132%]) were available. Serial study data (lab, physical exam, CXR, echocardiogram, biopsy, etc.) were compiled in table format, color-coded by organ system and distributed to MRC members two weeks prior to each meeting for review. The MRC met in five intensive two-day sessions over 1.5 years. This unique "paper autopsy" allowed discussion and debate by the pathology, cardiology, pulmonology and infectious disease specialists who made up the MRC. Standardized COD definitions were developed. Consensus could not be reached in 11/I 10(10%) of cases because relevant data near death were unavailable. MRC and self-reported CODs differed in 84/110(76%) of cases.

This intensive interactive review process is critical to obtain specific COD in a population with complex, multi-system disease. The need for centralized death review should be considered in the design phase of a study. "1989 US Standard Certificate of Death.

P42 QUALITY CONTROL METHODS FOR TELEPHONE

INTERVIEWING IN RANDOMIZED CONTROLLED TRIALS

Donna L. Perlmutter and Roberta Scherer for the Surgical Treatments Outcomes Project

for Dysfunctional Uterine Bleeding (STOP-DUB) Olga Lurye and Lucy Howard

for the Ischemic Optic Neuropathy Decompression Trial (IONDT)

University of Maryland Baltimore, Maryland

Background: Quality control (QC) and quality assurance (QA) of telephone interviewing for data collection in randomized controlled trials (RCTs) is desirable to ensure validity and reliability. Some QC and QA methods can be directly adapted from observational studies to RCTs, while others require modification or development.

IOOS Abstracts

Purpose: To develop QC and QA methods for use in RCTs to ensure objectivity and standardization of data collection by telephone.

Methods: We are conducting telephone interviews for two RCTs: STOP-DUB and IONDT Follow-up Study. We modified telephone QC and QA procedures originally developed for observational studies, conducted pilot studies, and further refined our methods. We first adapted QC and QA methods for STOP-DUB and subsequently adapted STOP-DUB methods to the IONDT Follow-up Study.

Results: Most methods appeared generalizable from observational studies to RCTs, and adaption from one RCT to another required only study specific changes. Techniques for developing forms, interviewer training, monitoring and feedback, and review of completed forms were adapted directly from observational study methods to STOP-DUB; development of the interviewer manual, explanation of the study structure, and certification procedures required moderate modifications; and development of masking techniques required major modifications.

Conclusions: Many of the standard techniques of interview QC and QA used in observational studies can be applied directly to RCTs. Additional QC and QA methods specific to RCTs are required to monitor and maintain interviewer objectivity and masking with regard to randomization.

P43 TRAINING AND CERTIFICATION OF PERSONNEL IN THE VITAMIN

INTERVENTION FOR STROKE PREVENTION STUDY (VISP)

Elizabeth Sides, Jan Smith, Hope Bryan, Sara Quandt, Carol Wasilauskas, Myra Carpenter and Virginia Howard

Wake Forest University School of Medicine, Winston-Salem, North Carolina and University of North Carolina at Chapel Hill, Chapel Hill, North Carolina

For a clinical center to be approved to begin the screening and randomization phase of the multicenter Vitamin Intervention for Stroke Prevention clinical trial, extensive training and certification of personnel are required. The complexity of the study requires that study coordinators and neurologists each be certified to perform certain tasks.

Using a variety of learning strategies appropriate for adult learners, the two and one-half day comprehensive centralized training course includes demonstration and practice of proper interviewing techniques and the computerized data management system; and instruction on blood collection and processing, all study questionnaires, assessment tools and data collection forms. For individual learning on site, a videotaped demonstration of the neurological examination is provided.

Certification for various aspects of the study includes: the study coordinator conducting two stroke assessment questionnaires via telephone with coordinating center staff, written tests on specific areas of the study procedures, and a pilot test with two w~lunteer stroke patients. The pilot test provides active practice on the data collection, the processing and shipping of blood samples, the dietary intake assessment, chart reviews and the computerized data entry and transti~r system. An audiotape of the dietary intake interview is made and reviewed for certification. Submission of the data forms as well as transmission of the data to the coordinating center complete the pilot test. When the center's research team has successfully completed these various components, then approval is given for their center to begin its participation in the study.

This presentation focuses on initial training and certification procedures and the policies and methods for managing recertification when there are personnel changes.