PT - JOURNAL ARTICLE AU - Baker, R AU - Whitfield, M TI - Measuring patient satisfaction: a test of construct validity. AID - 10.1136/qshc.1.2.104 DP - 1992 Jun 01 TA - Quality and Safety in Health Care PG - 104--109 VI - 1 IP - 2 4099 - http://qualitysafety.bmj.com/content/1/2/104.short 4100 - http://qualitysafety.bmj.com/content/1/2/104.full SO - Qual Saf Health Care1992 Jun 01; 1 AB - OBJECTIVE--To establish the validity of two patient satisfaction questionnaires (surgery satisfaction questionnaire (SSQ) and consultation satisfaction questionnaire (CSQ)) developed for use in general practice. DESIGN--Prospective study of performance of SSQ and CSQ in patients selected for their predicted levels of satisfaction. SETTING--Avon Family Health Services Authority (FHSA); general practices in Bristol (practice A) and in Cheltenham (practice B). PATIENTS--400 patients selected by Avon FHSA who had changed practices but not their home address and whose original practice had not changed its services (group 1); 869 randomly selected patients (221 from practice A, 648 from practice B) (group 2). MAIN MEASURES--Median difference in satisfaction scores for each questionnaire between groups 1 and 2 and between subgroups of group 2 patients according to assessed level in continuity of care (< 50%, > or = 50%) in the past 12 consultations. RESULTS--272 (68.0%) patients in group 1 completed the SSQ and CSQ. 711 (81.2%) patients in group 2 (178/221 (80.5%) in practice A, 533/648(82.3%) in practice B) completed the SSQ and 374(88/106(83.0%), 286/335(85.4%)) completed the CSQ. Both questionnaires classified patients in groups 1 and 2 according to the construct of satisfaction; thus the difference in median scores for every component of satisfaction in each questionnaire was significant and occurred in the direction predicted by the construct. Each questionnaire also discriminated between patients grouped according to their assessed level of continuity of care. CONCLUSION--SSQ and CSQ are valid measures of satisfaction for these types of patients. IMPLICATIONS--Valid measures of patient satisfaction can be developed; untested instruments should no longer be used.