Phase I: development of the items pool

In the qualitative interview, a total of 12 nurses were interviewed, all female, aged 25–49 years, with an average age of 35.17 ± 6.97 years; their years of experience in hospice care ranged from 2 to 11 years; 3 (25.0%) had junior titles, 6 (50.0%) had intermediate titles, and 3 (25.0%) had senior titles, the results of which are shown in Table 1. The results of the semi-structured interviews are detailed in Supplementary Material 1. The initial version scale was grouped into two dimensions and ten items, respectively: overall social environment (3 items), and organizational policy and culture (7 items). Detailed items for the initial version of the scale are shown in Supplementary Material 2.

Table 1 Demographic information of participants in semi-structured interviewsPhase II: selection of environment indicators through consensus process

The consensus process strictly followed the Delphi method. A total of 2 rounds of expert consultation were completed in this study. In each round, 16 experts were involved in the selection and optimization of scale dimensions and items. Among them, 15 were female and 1 was male, working in the fields of geriatrics, geriatric nursing, oncology nursing, and nursing education. Details of the experts are shown in the Table 2.

Table 2 Demographic of delphi’s participants (N = 16)Round 1

The dimensions of the hospice care environment scale in the first round of expert consultation were 2. The coefficients of variation were 0.088 and 0.000, and the mean importance ratings were 3.875 and 4.000, with full score ratios of 87.5% and 100.0%, respectively, indicating a concentration of expert opinion on the importance of the two dimensions. The coefficients of variation of the 10 items of the scale ranged from 0.000 to 0.119, the mean importance scores ranged from 3.750 to 4.000, and the full score ratios ranged from 75.0 to 100.0%, and all indicators met the prespecified criteria, as shown in Supplementary Material 3. The first round of expert consultation form is available in Supplementary Material 4.

In the current round of expert consultation, experts suggested changing the dimensions to social and organizational environments. Three additional items were added: “Government administration established good hospice policy”, “Hospital/unit has good incentives for hospice work”, and “My department integrates multidisciplinary staffs (such as dietitians, social workers, volunteers, etc.) to provide hospice care services for patients”. In addition, some items were modified in expression. In this round of expert consultation, the I-CVI and S-CVI/Ave of the hospice care environment scale were both 1.000 and 1.000.

Round 2

In the second round of expert consultation, the coefficient of variation for each indicator ranged from 0.000 to 0.088, the mean importance score ranged from 3.875 to 4.000, and the full score ratio ranged from 87.5 to 100.0%, as shown in Supplementary Material 3. The second round of expert consultation form is available in Supplementary Material 4.

The experts pointed out that “multidisciplinary staff” should be listed in item 12. Therefore, we changed it to “The department integrates multidisciplinary staff (e.g., dietitians, social workers, volunteers, etc.) to provide hospice services to patients. There were no changes to the other items. In the current round of expert consultation, the I-CVI and S-CVI/Ave of the hospice care environment scale after adding 3 items were both 1.000.

After two rounds of expert consultation, the Hospice Care Environment Scale was developed, consisting of 2 dimensions and 13 items. A Likert 5-point scale was used (completely disagree = 1 point, disagree = 2 points, neutral = 3 points, agree = 4 points, and completely agree = 5 points).

Phase III: psychometric evaluation of the hospice care environment scaleParticipant characteristics

A total of 500 valid questionnaires were collected in this study. The valid questionnaires were randomly divided into two groups in a 1:1 ratio, Sample 1 (n1 = 250) for entry screening and Sample 2 (n2 = 250) for quality assessment of the formed scales. The descriptive statistics of the samples were used in Table 3, with the mean age of (32.36 ± 6.71) years for Sample 1, which ranged from 21 to 52 years, and (32.31 ± 6.91) years for Sample 2, which ranged from 20 to 54 years.

Table 3 Demographic and medical characteristics of participants (N = 500)Item analyses

The results of the critical ratio method showed that the scale items all had CR values > 3.0 and that the differences between the high and low subgroups were statistically significant for all entries (P < 0.0001). According to corrected item-total correlation analyses, since the correlation coefficients of the items were between 0.802 and 0.908, it was not considered necessary to remove items from the draft scale, and exploratory factor analysis was performed for the 13 items in the draft scale.

Exploratory factor analysis

In the factor analyses, the Kaiser-Meyer-Olkin value was calculated as 0.927, while Bartlett’s test of sphericity statistic was calculated as χ2 = 4505.403,P<0.0001.These results revealed that the data set was large enough for factor analyses, could be considered homogeneous, and was suitable for factor analyses. According to the scree plot (Fig. 1), there is an inflection point at the 3rd factor, therefore, 2 factors were set to be extracted, and the principal component analyses method was chosen to extract them. The cumulative variance explained by the 2 factors was 83.856%, and after rotation using the direct oblique intersection method, the loadings on the factors to which each question item belonged were at a high level (0.619–1.003), and there were no large cross-loadings (< 0.4), and the factor structure clearly (Table 4). Factor 1 contains items 1–4, named “social environment”, and factor 2 contains items 5–13, named “organizational environment”.

Fig. 1

Parallel analyses for factor extraction(n1 = 250)

Fig. 2

Standardized path chart of the hospice care environment scale

Table 4 Item characteristics and factor loadings of the hospice care environment scaleConfirmatory factor analyses

The fit indices of the tested scale model were determined as χ2/df = 2.689, RMSEA = 0.082, CFI = 0.982, GFI = 0.932, AGFI = 0.866, IFI = 0.983. The structure of the scale was confirmed with 2 factors and 13 items (Fig. 2). In Table 5, the fit index values of the model tested for 13 items and the acceptable limits of these values are presented.

Table 5 Fit index values of the CFA of the hospice care environment scaleInternal consistency reliability

The total Cronbach’s α coefficient of the hospice care environment scale was 0.970, among which the social environment dimension was 0.944 and the organizational environment dimension was 0.966, indicating that the internal consistency of the scale was good.