Further Psychometric Testing and Validation of the Errors of Care Omission Survey (EoCOS)

Although a large proportion of healthcare care is delivered in primary care settings, research on patient safety in primary care has lagged behind that of inpatient care. Patient safety issues occur as frequently in primary care as in inpatient settings, but are not well studied. Moreover, the extant studies mainly focus on errors of commissiondoing something wrong such as administering wrong medicationas opposed to errors of omissionfailure of action such as missed care. Although omission errors outnumber commission errors and threaten patient safety, there is a critical gap in the literature on omission errors. This lack of evidence restricts the evaluation of clinical practice and challenges administrators', policy makers' and clinicians' abilities to make changes to promote patient safety. There are no reliable and valid tools to measure errors of care omission in primary care. We are developing the Errors of Care Omission Survey (EoCOS) through literature review, investigation of existing tools, interviews with primary care providers (PCPs), content validation, and pretesting to measure PCP perceptions of errors of care omission. Four domains underlying care omission are proposed: a) self-management support; b) follow-up; c) emotional health; and d) care integration. Each EoCOS subscale measures one domain through items asking PCPs to rate their perceptions on a 4-point scale. The purpose of the proposed study is to further evaluate the psychometric properties of the EoCOS and refine and validate the tool. Specific aims include 1) Determine the factorial structure of EoCOS and finalize the subscales measuring care omission domains; 2) Investigate performance of the items in EoCOS using Item Response Theory (IRT) models; and 3) Examine whether the items on each EoCOS subscale measure the intended construct by fitting data from different PCPs utilizing Confirmatory Factor Analysis (CFA). A cross-sectional survey design with a sample of PCPs (physicians and nurse practitioners) recruited in New York State will be used (n=1,328). Each PCP will receive a mailed survey along with cover letter, consent form, and a prepaid return envelope for returning the surveys to the research team. Post card reminders and a second mailing will be conducted to increase the response rate. SPSS, MPlus and IRTPro software will be used for data analysis. We will randomly split the sample into two: a derivation and a validation sample. Using data from the derivation sample, we will establish the factorial structure of the EoCOS using Exploratory Factor Analysis and assess the characteristics of each individual item using Item Response Theory models. Using the validation sample, we will test whether the emerged factor structure fits the data from this sample. These analytical procedures will assess the discriminant, convergent, and construct validity of the EoCOS and will produce sample invariant parameters. EoCOS may be used with various groups of PCPs for conducting meaningful comparisons of care omission errors across practices and states increasing the utility of the tool. This evidence will help create patient safety systems in primary care to make care omission errors visible so they can be corrected before harming patients.