Quality of life assessment: Patient compliance with questionnaire completion

Quality of life assessment: Patient compliance with questionnaire completion

708 Abstracts the data quality is fundamental for the whole monitoring process and therefore has to be kept very up-todate and easy to access. In or...

154KB Sizes 0 Downloads 6 Views

708

Abstracts

the data quality is fundamental for the whole monitoring process and therefore has to be kept very up-todate and easy to access. In order to achieve this aim we propose to extend the medical database (MDB) of a clinical trial by adding a data file which contains important metainformation concerning the medical data forms which are not yet entered into the database. For example, for each medical data form a corresponding variable in the organizational data file is generated whose value indicates whether the medical data form is correctly stored in the database, whether it is completely missing, not stored because of data errors or missing values, or what other reasons may be responsible for its absence. The concept of extending the MDB by organizational data had been implemented 3 years ago in a multicenter trial on lung cancer and has been shown to improve and facilitate the monitoring of a mulUcanter study considerably. P105 DOES CENTRAL MONITORING OF THE TIMELINES OF FORMS RESULT IN A DECREASE IN THE QUALITY OF DATA Beverly Koekl and Joseph Pater National Cancer Institute of Canada Clinical Trials Group Kingston, Ontario, Canada The NCIC Clinical Trials Group (CTG) coordinates multicenter clinical trials in cancer therapy. In 1988 the CTG initiated a program for monitoring the timeliness of submission of case-report forms. Between the interval from July-Dec/88 and Jan-Jun/90 the proportion of onstudy and follow-up forms submitted to the central office on-time increased from 79% to 90%. There was concern however that within individual centers improvement in timeliness might have been accompanied by a decrease in data quality. Since July 1987 the CTG has maintained a database for recording queries resulting from central review of incoming data. The volume of queries generated for any given trial depends on the nature of the trial, the complexity of the case-report forms and the requirement of regulatory or pharmaceutical agencies. Each cancer therapy center participates in a unique subset of CTG tdals and the number of patients on-study varies by canter. We have estimated the quality of data for each center by calculating a query letter ratio (QLR) which adjusts for differences in the numbers of patients on-study using the indirect method of standardization often used in the comparison of mortality rates. Descriptions of the database and how the individual center QLR's are calculated will be presented. Changes in the proportion of forms submitted on-time and changes in QLRs were compared for 26 large centers between the intervals July-Dec/88 and Jan-June/90. Twenty-one (81%) of centers improved in timeliness. Overall, an individual center's improvement in timeliness did not appear to be accompanied by a relative decrease in the estimated quality of data. In fact there was a non-significant (p = 0.27) positive relationship between improvement in quality and improvement in timeliness (Spearman's r = 0.23). These findings put to rest our concerns that emphasizing timely data submission could result in a deterioration in data quality. P106 QUALITY OF LIFE ASSESSMENT: PATIENT COMPLIANCE WITH QUESTIONNAIRE COMPLETION Anna Sadura, Joseph Pater, Mark Levlne, and David Osoba National Cancer Institute of Canada Clinical Trials Group Queen's University Kingston, Ontario, Canada The NCIC CTG conducts national, mulUcentered clinical trials in cancer therapy. It has a mandate to consider quality of life (QL) as an endpoint in its randomized studies, and has done so in two recent phase III adjuvant trials. Patient compliance in completion of repeated QL assessments was measured. In trial A, 226/229 (98.7%) of questionnaires were completed and returned. (100% of baseline; 97.9% of subsequent). For trial B, compliance was 97.8% (546/558) (100% of baseline; 97.1% of subsequent). Very few questionnaires were incomplete though errors in timing of administration were noted. This excellent compliance is contrary to previously reported experience and is, we feel, attributable to comprehensive measures adopted to ensure cooperation of participants. In both cases, QL was a stated tdal objective and a detailed rationale was given in the study protocol. Baseline questionnaire completion was an eligibility criteda which was verified prior to patient entry on study. Pre-tdal educational workshops were held for data management and nursing staff, computerized prospective reminder programs were developed, re-

Abstracts

709

quired QL assessment schedule was reflected in forms design, and feedback and opportunity for input was provided to study participants at annual meetings. The NCIC CTG experience has demonstrated the feasibility of QL assessment as an end-point in clinical trials of adjuvant cancer therapy. P107 MANAGEMENT OF DATA FILES IN PREPARATION FOR STATISTICAL ANALYSIS Pauline Ralz, Margaret Frederick, and Sandra Forman Maryland Medical Research Institute Baltimore, Maryland Maryland Medical Research Institute has randomized and collected data for over 3,900 patients in a large, multicenter, clinical trial. Over 30 different forms were keyed and edited for analysis. Building a large crosssection of data for statistical processing presented a challenge. Among the requirements for setting up this database were: (1) forms were stored on a Data General INFOS database, and were to be analyzed as SAS®files; (2) data integrity was a primary concern, as multiple users were to be allowed access to these files; and (3) storing and retrieving this volume of data would require large computer resources, affecting I/O processing and system response time. The following procedures were implemented: 1. Important data were stored on a flat "Master" file. This file, combined with the INFOS database, was used to create several flat files. Many calculations and implicit values were stored in these files. 2. The flat files were converted to SAS ® files, available to all users. These SAS® files were smaller, easy to store, require less I/O processing, due to unique features built into the SAS ® language. This resulted in a new, uniform and manageable database; duplicate processing was avoided, CPU time and space were reduced, and data integrity ensured. P 108 TOWARDS REDUCING OBSERVER EFFECTS IN BLOOD PRESSURE (BP) MEASUREMENT Patrlcla E. Hogan Bowman Gray School of Medicine Winston-Salem, North Carolina BP readings are plagued by numerous sources of variability that threaten the validity and reliability of the measurements, especially observer differences in the measurement process. The Postmenopausal Estrogen/ Progestin Interventions Trial (PEPI, a clinical trial assessing the effect of various hormonal combinations on cardiovascular risk factors, including systolic BP) has implemented a standardized measurement protocol including use of a random zero sphygmomanometer. A BP training, certification, and monitoring program has been designed to reduce observer sources of variability. The monitoring component of the program consists of the following types of routine analyses: (1) reliability measures for duplicate readings; (2) longitudinal checks for consistency; and (3) examination of terminal digit preference by observer. The results and uses of these analyses will be discussed. Although many studies assess terminal digit preference, the best way to provide feedback to BP observers remains controversial The training and monitoring approaches taken by several other clinical trials with BP as a primary end-point will also be presented. P109 SELECTING A POINT ESTIMATOR OF BLOOD PRESSURE FROM CONSECUTIVE MEASUREMENTS Jennifer J. Gaaaman, R.L. Berg, R.P. Bylngton, L.A. Hebert, K.E. Lambdin, and A. Tanna National Institutes of Health Bethesda, Maryland The Modification of Diet in Renal Disease (MDRD) Study is a factorial design multicenter clinical trial testing the effect of blood pressure control and dietary protein and phosphorus intake on the progression of renal disease. Patients are randomly assigned to a blood pressure control regimen, either a usual care goal (mean arterial pressure <~107mmHg) or a lower special intervention goal (mean arterial pressure ~<92mmHg). Clinical Center staff follow a stepped care program to select treatment to achieve blood pressure goals. Patients arrive at the clinic, meet with the blood pressure measurer and rest for 5 minutes. Blood pressure is then estimated by taking three consecutive measures (using random zero sphygmomanometers) and