Recipients’ satisfaction with the care that they receive is a common outcome. Currently, it is frequently being employed in health care to evaluate individual provider performance. Satisfaction is, however, a complex construct that is often measured globally. The result is that providers of care can be frustrated as they have little direction on how to improve scores when they are low, or alternatively, how to maintain scores when they are high. Behaviors on the other hand can be both assessed by recipients of care and directed by providers.
OBJECTIVE: The primary objective of this study was to establish the psychometric properties of an instrument designed to measure the Person with HIV or Acquired Immunodeficiency Syndrome (AIDS) perceptions of the caring behaviors of their health care providers.
DESIGN: Instrument development and validation using psychometric techniques guided the design of the study.
POPULATION, SAMPLE, SETTING, YEARS: The population is persons with HIV infection or Acquired Immunodeficiency Syndrome (AIDS) who receive their care from providers in the United States. Data were collected from persons with HIV/AIDS who received their care in ambulatory care setting throughout one western geographical region of a large integrated health care system from 1997 to 2001. The final non-probability sample (N=761) was divided into two groups. Exploratory analysis was done in Group 1 (n=608). Confirmatory analysis was done in Group 2 (n=153). The demographic characteristics between the groups were not significantly different. They were predominantly male (92%), Caucasian (65%), currently employed (64%), and perceived themselves as being healthy (45%).
CONCEPTS: This instrument intends to measure the caring behaviors of providers of care to persons with HIV/AIDS. Caring is seen as a psycho-social relational process based on interaction, meaningful interpretation, and action. The actions may be present (actual) or desired. These actions are manifest as behaviors that are goal directed, quality focused, safe, cost effective, competent, and ethical.
METHODS: In the proposed instrument, there were 30 items that measured actual caring behaviors and 18 items that measured desired caring behaviors. Item to inter-item correlations were used for item analysis. Construct validity of the instrument was established using exploratory factor analysis and structural equation modeling in Group 1 and confirmatory factor analysis and in Group 2. Linear regression was used to establish predictive validity. Reliability statistics in both groups included Cronbach’s alphas.
FINDINGS: Seven items had item to inter-item correlations below .30, indicating that those items probably measure other concepts, or above .70, indicating redundancy. After review, it was concluded that these items could be removed from the instrument.
Exploratory factor analysis of the actual caring behaviors resulted in 4 factors that explained 59% of the variance. The Goodness of Fix indices for the structural equation model were as follows: discrepancy/df=2.487, Comparative Fix Index=.97, and RMSEA=.088. The confirmatory factor analysis in Group 2 also resulted in the same 4 factors that explained 70% of the variance
Exploratory factor analysis of the desired caring behaviors resulted in 2 factors that explained 81% of the variance. Confirmatory factor analysis also resulted in 2 factors that explained 68% of the variance.
Predictive validity of each of the scales was calculated to a global measure of satisfaction. In a model using the subjects from Group 1 (Adjusted R2=.183), only Actual Caring Behaviors were predictive of satisfaction of with care of HIV team (b=.47, t=4.182, p=0.000). Desired Caring Behaviors were not predictive (b=.058, t=.519, p=.605). While the adjusted R2 was higher (.324) in Group 2, once more only the Actual Caring Behaviors Scale, not the Desired Caring Behaviors Scale, was a predictor of the satisfaction with the HIV team.
Cronbach’s alpha was greater than .90 for the total instrument. Each of the subscales (Actual and Desired Caring Behaviors) also had alphas greater than .90. The subscales for the Actual Caring Behaviors Scale had alphas >.70 and the Desired Caring Behaviors Scale had alphas >.80.
CONCLUSIONS: The resulting instrument has two scales with a total of 41 items. Twenty-three of the items measure Actual Caring Behaviors and 18 items measure Desired Caring Behaviors. The Actual Caring Behaviors Scale has four subscales: Personalized Care, Competent Care, Care of Significant Others, and Negative Caring Behaviors. The Desired caring Behaviors has two subscales: Professional and Intimate Caring Behaviors.
IMPLICATIONS: While this instrument has acceptable psychometric properties, it could further benefit from analysis using Item Response Theory techniques. Given the current status of the Caring Behaviors instrument, providers of care to persons with HIV/AIDS can use the instrument with confidence.
Back to Research Instruments and Practices
Back to The Advancing Nursing Practice Excellence: State of the Science