Assessing Change in Patient-reported Quality of Life After Elective Surgery: an Observational Comparison Study

Quality of Life, Surgical Procedure, Unspecified

outcome assessment, postoperative period

This study will compare two commonly-used methods for assessing patient-reported quality of life. The first is to assess quality of life before surgery and again after surgery using the same validated scale (ie Veterans Rand 12). The second is simply to ask patients whether or not they think their post-operative quality of life is better, worse, or the same. The investigators hypothesize that the second method may be inaccurate due to cognitive bias.

All preoperative and postoperative quality of life data comes from the SATISFY-SOS (NCT02032030) surveys, which are administered to patients at the preoperative assessment visit and then approximately 30 days after surgery. To maximize the follow-up survey response rate, patients are emailed the survey (once), mailed hard copy surveys (two times), and phoned (up to five times). The twelve items comprising the Veterans RAND 12 (VR-12) are items 24 through 35 on the survey, while the self-reported global quality of life question is item 1. The VR-12 is made up of two components: a physical component score (PCS), and a mental component score (MCS). Both scores are continuous on a scale from 0 to 100 (where higher is greater quality of life), and they are calibrated so that a score of 50 represents the US population mean (Kazis 2006, Selim 2009). The 30-day follow-up questionnaires also ask patients to self-report their change in quality of life. The question asks, How would you rate your quality of life now? with answer choices including Better than before your procedure, The same as before your procedure, and Worse than before your procedure. Surgical specialty for the procedure is obtained from the electronic medical record. Using queries in MetaVision (iMDsoft, Needham, MA), the informaticist will provide the requested survey and medical record data to the investigators. He performs rigorous data validation on each queried variable. SATISFY-SOS databases are hosted on a firewall-secured network server managed by the Department of Anesthesiology. The server is securely housed behind two locked doors within the departmental office suite and maintained and managed by the departmental IT team. Only the project Informaticist, Data Manager, and Director(s) have full access to these databases, which are also password-protected and encrypted for additional protection. Hardcopies of the baseline surveys are collected daily from the CPAP clinic and securely stored behind two locked doors within the Department of Anesthesiology. Baseline completed paper surveys are scanned into a digital image format (compressed TIFF). The digital image files are indexed and stored on a research file server that is attached to a private network with no public access. Survey email, mail and call lists are generated at Washington University in a similar manner to mailing lists for billing services. For each patient and date of service, a unique ID is generated and never duplicated. This unique ID is a nonsensical only meaningful to the research team. Baseline surveys are processed by Solutions Data Systems. The digital image files are transmitted to Solutions Data Systems via secure file transfer protocol. When data entry has been confirmed, Solutions Data Systems deletes the digital image file from their servers. Press Ganey, a vendor specializing in patient survey distribution and collection, disseminates, collects, and processes 30-day and 1-year surveys. Paper surveys processed through automated scanning are all manually checked, and a manager listens to 10% of telephone surveys. All telephone surveys are recorded and available for future quality checks for performance improvement. Press Ganey stores the survey hardcopies for 90 days while the study team conducts spot-check quality assessments of the scanned data. The company then shreds the paper copies. Similarly, Press Ganey will hold copies of the electronic files and electronic recordings for 90 days, after which the electronic files are removed permanently from their system (and then only maintained by Washington University). During this 90-day period, the study team conducts additional quality assessments of the converted data. Sample size considerations for this study were based on the primary outcome. The first component of the primary endpoint is comparing the median quality of life scores among the three self-reported change groups (better/same/worse). Using a minimum important difference of five points (Norman 2003), two tails, alpha of 0.05, and 80% power, the required sample size is 77 patients per group, or 231 total patients among the three groups. The second component of the primary outcome is the agreement between the two quality of life measures, as reflected by the kappa statistic. Kappa does not have sample size requirements beyond lack of sparse cells. The third component of the primary outcome is comparing the percentage agreement across the three different self-reported change groups. Since no studies have performed this type of comparison previously, the investigators pre-specified a 10% change as the minimum important change. Estimating 80% agreement, and using two tails, alpha=0.05, and 80% power, the required total sample size is 311 per group, or 933 total patients. Therefore, this study has adequate power for all of these endpoints. The following statistical analyses will be performed, using alpha=0.05 and 95 percent confidence intervals, where appropriate. All analyses will be performed twice, once for VR-12 physical quality of life, and again for the VR-12 mental quality of life. * Compare change in VR-12 QOL scores for those answering better/same/worse (Kruskal-Wallis). If significant, will use Wilcoxon Rank-Sum tests to compare each of the three groups, using a Bonferroni correction of alpha=0.017. * Calculate overall agreement between the self-reported and validated quality of life measures (using weighted kappa, which penalizes disagreements in proportion to their seriousness). * Calculate the overall percent of patients whose self-reported and validated quality of life scores matched (descriptive), including stratification by self-reported global change better/same/worse (compared using chi-square). * Calculate the quality of life where an equal proportion of patients reported better and same quality of life (MCID for improvement), or same and worse quality of life (MCID for deterioration). This is an anchor-based approach (Wright 2012) * Describe the change in VR-12 score and percent of patients reporting better, same, and worse quality of life for each of the following surgical specialties: neurosurgery, orthopedic, plastic, ophthalmologic, general, cardiac, gynecologic, otolaryngology, gastrointestinal/hepatobiliary, urologic, and other. Only those answering both the self-reported quality of life question and at least ten out of twelve VR-12 questions at baseline and 30 days will be included. The investigators will describe the characteristics of those with and without missing data. Multiple imputation will be used to fill in missing measurements for those missing two or fewer items on the VR-12 questionnaire.

United States