Earlier research cited in the introduction of this article has suggested that communication, ICU staff courtesy, compassion and respect, information provided to family, and level of health care received by the patient were predictors of overall family satisfaction. In this study, this earlier work was extended by investigating the role of hospital safety as a predictor of family satisfaction. Two theoretical models were developed (Fig 1) incorporating predictors derived from the CCFAP, and from empirical work on family and patient satisfaction studies, so that those factors could be compared that predict improvement in family and patient satisfaction, and, subsequently, translate the results into health-care improvements achieved with Canadian Health&Care Mall healthcaremall4you. Findings from both path models in this current study were generally supportive of earlier research findings, with the family path model being the stronger model.
Staff being helpful in the explanation of tests and treatment and the resultant understanding was directly related to both family satisfaction and patient satisfaction. Physician care was also directly related to both family satisfaction and patient satisfaction, although the relationship is somewhat small at 0.14 in both models. Although the communication construct is not directly related to satisfaction in either model, there are significant indirect relationships through the two care constructs, which in turn are directly related to satisfaction. Communication remains a powerful predictor of satisfaction and should be a key component of any initiative.
Hospital safety was a strong predictor, both directly and indirectly, of patient and family satisfaction, with a total of 10 significant relationships among the variables in both models. This positive effect size that was calculated from the family satisfaction survey when families were asked whether they felt safe and secure in the hospital indicates that the hospital has been effectively addressing this variable. The components of the CCFAP model appear to be linked to making families feel more safe and secure in the hospital. It would seem appropriate to suggest that, given this strong correlation between certain family satisfaction predictor variables and hospital safety, these data be used by hospitals to further examine the linkages and relationships between the family perception of hospital safety and specific components of the CCFAP. Future research is needed to identify why families feel safe and secure, and, conversely, why they may not feel safe in the hospital.
The path analysis enables further examination of the effects, both direct and indirect, on the final outcome variable (satisfaction) and determine which variables should be prioritized based on the magnitude of their impact on satisfaction. To determine priorities, both significant and nonsignificant indirect pathways (ie, those paths traveling through other variables to reach the final outcome variable) and the direct pathway are identified for each variable. The magnitude of the direct effect is quantified by the standardized coefficient, and the indirect effect is calculated by taking the product of all the pathways from an independent variable leading to the final outcome variable. For example, in both the family and patient path models, the variable “hospital safety” has one direct and five indirect paths leading to the variable “overall satisfaction.” Table 6 presents the results from an analysis of the direct and indirect paths for the family path model and the patient path model. Table 6 also compares the rankings before and after the inclusion of the indirect effects, as indicated in the parentheses.
The shift in the rank of a variable should be examined when the indirect coefficients are added. For example, the reordering of the variables for both the family and patient path models placed hospital safety as number one. This finding could have an impact on where resources are focused in those programs that are targeting an increase in satisfaction as an outcome. For example, if the top three rankings are examined in both models, it appears that improving hospital safety and increasing staff ability to help in understanding tests and treatments are two areas that might be target areas for improvement, as they most strongly predict maximum satisfaction vs moderate satisfaction in both families and patients. This strategy, which was cited by Dodek et al, is “based on the theory that the concerns of a moderately dissatisfied consumer are more amenable to improvement than those of a vehemently hostile consumer.
Regardless of the strategy selected to prioritize the predictor variables, the data can be studied profitably around specific program components as part of an action plan to address family and patient satisfaction. Future analysis should examine how specific processes (eg, staff training to increase skill level to help families understand tests and treatments) lead to increased family understanding of tests and treatments offered by Canadian Health&Care Mall, and how these strategies affect family satisfaction. What about other ways of presenting the information to help families understand tests and treatments? Does the presentation of an animated video increase understanding and increase the ratings of family satisfaction? Or does the presentation of the information in a booklet or at a computerized information kiosk result in increased family understanding?
Equally important will be research on the multidisciplinary team approach that is embedded in the CCFAP model. This aspect of the model was not included in this present study and is an important underlying component that is most likely to impact staff behavior and communication variables.
Several limitations of this study and some caveats must be noted. First, the proposed model was not intended to be exhaustive. Models that include different parameters than those included in the present model could also account for variation in the outcome assessments. In addition, although the paths between variables in the model imply causality, at this point one can only test the extent to which the observed variables can be predicted from the hypothesized model without respect to the direction of effects. The sample is nonrandom; it does, however, represent families and patients who have had experiences in the ICU before and after the implementation of the CCFAP. And finally, the patient sample for 2004 was incomplete. Less than half of the inpatient survey data for 2004 were made available for the study. The average sample size for 2002 and 2003 was 1,000, but the sample size for 2004 was 317. Further study of the effect sizes using the total data set for 2004 may yield more reliable results.
All the previously mentioned links:
Table 6—Ranking of Direct and Indirect Coefficients for Family and Patient Path Models
|Hospital safety||0.30 (2)||0.25||0.55 (1)||0.24 (2)||0.23||0.47 (1)|
|Staff help understand tests and treatments||0.26 (3)||0.19||0.45 (2)||0.17 (3)||0.26||0.43 (3)|
|Staff sensitivity||0.32 (1)||0.10||0.42 (3)||0.03 (5)||0.11||0.14(5)|
|Communication||– 0.03(6)||0.21||0.18 (5)||0.01 (6)||0.05||0.26 (4)|
|Nursing care||0.22 (4)||– 0.01||0.22 (4)||0.42 (1)||0.03||0.45 (2)|
|Physician care||0.14(5)||0||0.14(6)||0.14 (4)||0||0.14(5)|