Abstract
Evidence-based practice (EBP) in healthcare is imperative for optimum patient care. As part of the EBP process, clinicians must collect, appraise, and interpret clinical research to help guide their clinical practice. To help support EBP in clinical practice, patient-reported outcome (PRO) measures may be collected. Clinicians use PRO measures to assess the outcomes related to different health-related dimensions (e.g., pain, function, quality of life) spanning across several different intervention types (e.g., rehabilitation, surgical intervention). Not only do clinicians collect and interpret these PROs in a single setting, but also assess these outcomes over numerous visits (e.g., preoperative, postoperative). As PROs are important for EBP, focus now exists on the collection of these outcomes for purposes of reimbursement across healthcare facilities nationwide. More specifically, collecting outcomes associated with surgical interventions such as total hip arthroplasty (THA) are imperative as the number of procedures expected to rise substantially over the next decade. However, many PROs related to THA have minimally been tested using recommended contemporary psychometric analysis techniques. Therefore, the purpose of this dissertation was to assess the psychometric properties of two outcomes used in clinical practice to assess hip-related interventions such as THA: 1.) Hip Disability and Osteoarthritis Outcome Score Joint Replacement (HOOS-JR) and 2.) Hip Disability and Osteoarthritis Outcome Score (HOOS).
The HOOS-JR is a one-factor 6-item instrument designed to assess outcomes associated pain and function following a THA. The HOOS-JR was originally derived as a short-form from the original 40-item HOOS; however, minimal research has been identified pertaining to the assessment of the scale structure in a group of patients undergoing THA. Therefore, a confirmatory factor analysis (CFA) was conducted on the HOOS-JR to determine if the scale met the recommended goodness-of-fit indices in a large population of patients who underwent a THA. The HOOS-JR underwent further refinement due to the lack of model fit, which yielded a one-factor 5-item HOOS-JR. The alternate 5-item HOOS-JR was then subjected to invariance testing between age groups and sex (i.e., multi-group), and longitudinally across five time points (i.e., preoperatively, 6-months post-, 1-year post-, 2-years post-, and 3-years postoperatively). In addition, latent growth curve (LGC) modeling was performed to determine intraindividual (i.e., differences in mean scores over time) and interindividual (i.e., differences in mean scores between groups) differences in the responses to the outcome over time. In addition, LGC modeling was assessed to determine if the rate of change in these scores were linear. The 5-item HOOS-JR slightly exceeded the recommended cut-off values for testing; upon further assessment, item five demonstrated item-level bias within the model. In addition, no differences in mean scores were found between groups, and the change in mean scores were not consistent with a linear trajectory. These findings suggest a more curvilinear model was identified with most of the improvement in mean scores to occur between preoperative and 6-months postoperative. In its current form, the 6-item HOOS-JR is not recommended for use in clinical practice. However, the 5-item HOOS-JR maybe suitable though caution is warranted when attempting to make comparisons in mean scores across groups or over time. Continued refinement is recommended before implementation of the modified 5-item HOOS-JR in clinical practice and research.
The 40-item HOOS was created as a region- and disease-specific outcome that assesses five dimensions of hip-related health, including pain, symptoms, functional limitations, function associated with higher level activities, and hip related quality of life. Previous research conducted revealed the HOOS scale structure did not meet the recommended goodness-of-fit indices in a sample of mostly healthy individuals. In addition, analyses of the scale structure had not been confirmed in a large sample of patients who underwent a THA. Therefore, a CFA was conducted on the five-factor, 40-item HOOS. As model fit indices were not met, the scale was split into two samples (n1 and n2). An exploratory factor analysis (EFA) was performed on sample n1 to identify if a more parsimonious structure (i.e., alternate HOOS-9) could be found. Once an alternate model was found, CFA procedures were conducted on the sample n2. Recommended goodness-of-fit indices were met for the alternate model; therefore, the two samples were combined back together. The scale was then subjected to invariance testing between age groups and sex (i.e., multi-group) and longitudinally over five time points (i.e., preoperative and 6-months, 1-year, 2-years, and 3-years postoperatively). In addition, LGC modeling was conducted to determine if there were differences (i.e., intraindividual and interindividual) in mean scores between time points or between groups. Also, LGC modeling was used to determine if improvement in scores were linear. Upon our results, the alternate three-factor HOOS-9 met recommended goodness-of-fit indices for both groups, therefore comparisons in mean scores could be assessed across age groups and sex (i.e., males and females). Recommended model fit indices were also met at each time point and across time, suggesting the mean scores of the alternate HOOS-9 can be compared across time points. Lastly, the LGC model demonstrated the change in scores were not linear, however, most improvement in mean scores occurred between preoperatively to 6-months postoperatively. Additionally, there were significant changes in the mean scores across patients between preoperative to 3-years postoperatively, however, no significant differences were noted between groups. Therefore, in its current form, the 40-item HOOS should not be used in clinical practice and research; however, the alternate HOOS-9 may be a more viable option.