The present study involved the adaptation and validation of a measurement instrument in a non-randomized sample of relatives of patients admitted to a 32-bed medical-surgical ICU pertaining to a private teaching hospital in Santiago (Chile). The Unit had a regular 6 -h visiting period, with the added possibility of visiting the patient outside that time interval. The nurse-to-patient ratio was 1:2−3. Between October 2017 and December 2018, the relatives of patients with an ICU stay of ≥ 48 h and ≥ 18 years of age, and who had visited the patient at least once in the ICU, were invited to participate. The study considered the relative in charge of the patient and the visiting persons (direct relatives or not of the patient) authorized by him or her. Up to three relatives per patient were included. Relatives of patients with limitation of therapeutic effort or receiving end-of-life care were not considered.

Data collection was carried out using a sociodemographic characteristics registry form for relatives and the version in Spanish of the FS ICU-24.20 The FS ICU-24 was developed by Wall et al.9 and consists of 24 items with responses scored using a 5-point Likert scale and divided into two dimensions: satisfaction with care (14 items) and satisfaction with decision making (10 items). The FS ICU-24 uses a scoring system in which the Likert scores are transformed into a numerical scale (0–100), with higher mean scores indicating greater satisfaction. The original version of the FS ICU-24 reported high internal consistency.9 On the other hand, we recorded sociodemographic variables of the relatives, as well as basic clinical variables of the patient. In all the stages, the questionnaires were administered during the visiting times by research assistants previously trained in the use of the instrument and in managing the impact of ICU stay upon the relatives.

The adaptation and validation of the FS ICU-24 comprised the following stages: 1) content validity; 2) construct validity; and 3) reliability analysis.21,22

The version in Spanish of the FS ICU-24 was examined by a group of 10 experts composed of two physicians (specialized in intensive care and family medicine), 7 nurses specialized in intensive care, and a nurse with a Master specializing in family issues. The experts evaluated coherence between the items of the FS ICU-24 and the evaluated construct. Likewise, content validity was assessed using the Lynn index (> 0.8) and the coefficient of content validity, with a value of > 0.7 being considered adequate.23 The suggestions and modifications in this stage gave rise to the second version of the FS ICU-24.

In the language adaptation stage, the second version of the FS ICU-24 was administered to 10 relatives of patients who recently moved outside the ICU. Based on a semistructured interview, clarity, drafting and understanding of the items were evaluated. The suggestions made by these relatives in turn gave rise to the third version of the FS ICU-24, which was applied in the pilot stage to 10 relatives of patients admitted to the ICU.

The study sample consisted of 240 relatives of critical patients, established using a criterion of 10 subjects per item.24 A descriptive statistical analysis was made of the sociodemographic characteristics and items of the relatives and the patients. Data asymmetry and normality were assessed using the Shapiro-Wilk test. The items in turn were subjected to multivariate normality testing using the Mardia test to determine the extraction method to be used in the exploratory factor analysis. Since multivariate normality was not met, we used the principle axes extraction method.25

The choice of type of rotation (Oblimin or oblique) was determined establishing a cut-off point of 0.5 for the correlation between the factors and the factor solution.21 To identify the optimum number of factors, use was made of the Kaiser criterion,26 the scree plot27 and Horn’s parallel analysis.28 To incorporate an item in a certain factor, we considered factor loading ≥ 0.3.21 Lastly, we evaluated the discrimination of the items by analyzing the correlations between each item and its factor and the homogeneity of the factor solution on analyzing the differences between the factor loadings of one same item in the different factors.

The reliability of the version in Spanish of the FS ICU-24, in the same way as its dimensions, was evaluated using Cronbach’s alpha, with a value > 0.7 being considered adequate.

The statistical and psychometric analyses were made using R software.29 Statistical significance in hypothesis testing was considered for p < 0.05.

The study was approved by the Scientific Ethics Committee of the Faculty of Medicine of Pontificia Universidad Católica de Chile (Ref.: 170708001). All the relatives signed an informed consent document before participating in the study. Authorization was also obtained from the creator of the FS ICU-24 (Dr. Daren Heyland) for use of the instrument.

In the phase of evaluation by experts, all the items of the version in Spanish of the FS ICU-24 were classified as adequate. Nevertheless, suggestions were made regarding the semantics of some items and the structure of the questionnaire. With regard to the Lynn index, all the items scored well (>0.8), while in the evaluation of the coefficient of content validity, three items (7, 10 and 12) obtained a value < 0.7. These items were analyzed by the research team, with incorporation of the suggestions of the experts. In this stage, none of the items were eliminated. Subsequently, in the language adaptation stage, the relatives (n = 10) rated the items as being clear and easy to understand. In the pilot stage, the instrument administration time was found to be ≤ 10 min, with no additional comments being reported by the relatives.

The descriptive analysis of the items is shown in Table 2. The mean for the items was 86.4 (SD = 4.1), with a minimum of 76.9 (item 13) and a maximum of 92.2 (item 9). All the items exhibited a non-normal distribution, as well as negative asymmetry, reflected by multivariate normal behavior (p < 0.001) with the Mardia test in asymmetry and kurtosis. In turn, the KMO statistic for the matrix of correlations was 0.93, with rejection of the null hypothesis in the sphericity test (χ2 = 2518; p < 0.001), showing the matrix to be optimum for factor analysis. Oblique rotation was used, since the correlation between the factors and the factor solution was 0.5, presenting lesser complexity (oblique rotation = 1.3 versus orthogonal rotation = 1.7).

For the analysis of principal axes, the Kaiser criterion showed a solution of 2 factors, while the scree plot (Fig. 1) and Horn’s parallel analysis evidenced a solution of 3 factors. We tested the solutions with 2 and 3 factors, and chose the solution of 3 factors since it presents greater statistical evidence and is more consistent with the literature and the conceptual analysis of the solutions. The analysis of principal axes showed that the three factors globally explained 51% of the total variance of the items of the instrument under study (eigenvalue factor 1 = 9.78; variance explained = 41%; eigenvalue factor 2 = 1.49; variance explained = 6%; eigenvalue factor 3 = 1.2; variance explained = 4%).

Figure 1.

Scree plot of the number of factors in the Chilean-Spanish version of the FS ICU-24.

(0.12MB).

Regarding the factor loadings (Table 2), factor 1 was composed of items 1–9, 12 and 14, which mostly incorporated elements associated to patient care, and was referred to as the «satisfaction with care of patient and family» (SCPF) dimension. In the case of factor 2, we included items 10, 11, 15–20, which were related to the process of communication with the ICU staff, referred to as the «satisfaction with communication» (SC) dimension. Lastly, factor 3 was composed of items 21–23, related to decision making, and was referred to as the «satisfaction with decision making» (SDM) dimension. Item 13 (waiting room environment) did not fit adequately in any of the dimensions, and item 24 (adequate time in decision making) fit the SC dimension with the minimum acceptable value (0.3) and the SDM dimension with a fit of 0.05 points less; nevertheless, we decided to maintain both items, incorporating item 13 to the SCPF dimension and item 24 to the SDM dimension (Table 3).

With regard to the internal consistency of the FS ICU-24, adequate values were obtained for both the global scale (α = 0.93) and for the dimensions SCPF (α = 0.82), SC (α = 0.91) and SDM (α = 0.71).

The present study has some limitations. The use of a version in Spanish of the FS ICU-24, in contrast to a translation and back-translation process, could have influenced the cross-cultural adaptation process. On the other hand, since the study was carried out in a private teaching hospital, its applicability to non-teaching and public healthcare centers could be limited. The exclusion of relatives of patients with limitation of therapeutic effort restricts applicability of the findings in this population — limiting knowledge of the experience of these relatives regarding relevant processes in the ICU, such as end of life care. On the other hand, the sample of relatives was characterized by a high educational level, which may have influenced the questionnaire adaptation process — particularly in the language adjustment stage. Lastly, the interdisciplinary nature and the number of experts involved in the content validity stage stand out as strengths of the study.