Validity of self-assessment outcome questionnaires: patient-physician discrepancy in outcome interpretation

Abstract

Patient outcome following total hip arthroplasty (THA) was evaluated using a previously described patient assessment outcome index questionnaires. The questionnaire was distributed to 263 patients who underwent cementless THA. One hundred and three patients responded to the self-administered questionnaire and had updated clinical evaluation. We obtained a modified Harris Hip Score (HHS) based on patient assessments of their own pain and function and compared it with the clinical HHS obtained at the patients' last office visit. The mean follow up period was 4 years. Statistical analysis was performed between the two scores. The correlation between the scores from the self-administered questionnaire and the patients' last office visit revealed a fairly low correlation coefficient (r = 0.467, p < 0.001). Relative lack of correlation between the HHS's obtained from these two sources was especially noted for patients with a pain score of 30 points or less. These 26 patients were subsequently interviewed in detail about their pain to further explain these differences. The etiology of the perceived "hip pain" was found to be secondary to trochanteric bursitis in 13 patients, lumbar spondylosis in 7 patients, arthrosis of the contralateral hip in 5 patients, and from other causes in 8 patients. Pain that was hip related (anterior thigh or groin) was present in only 5 out of the 26 patients with a pain score of 30 or less. Another source of discrepancy between the total scores of the HHS was found to be on behalf of the physician in evaluating the presence of a limp. We also found that patients' expectations had changed from their preoperative expectations. Although outcome measures developed and administered by clinicians are subject to bias from several sources, results of this study suggest that self administered patient outcome measures also have their limitations. The validity of self-administered patient outcome questionnaires can be severely impacted by the patients' understanding of the questions asked, as even the most seemingly simple questions are subject to misinterpretation.