This study aimed to translate and validate the POPAC scale into Dutch and test the Dutch version of the questionnaires for psychometric properties among Dutch nurses working in acute hospital settings (28). Data were collected with the online questionnaire program QualtricsXM (version 2018, Provo, UT USA).
The instrument
The POPAC scale consists of 15 items, as shown in Table 1. The items describe care procedures and processes in patients with dementia in hospitals (19). With the self-report of nurses in hospitals, the POPAC scale measures the extent to which nursing interventions are based on best practices in association with person-centred care. The items are divided into three subscales: cognitive assessments and care interventions (items 1–5), evidence and cognitive expertise (items 6–8), and individualizing care (items 9–15) (19). The scores can be evaluated per subscale, or the score of the total scale can be used. The subscale or total scale scores can be calculated by dividing the sum of the scores by the number of items, whereby higher scores imply higher levels of person-centredness (10,19,25).
Translation of the person-centred care of older people with cognitive impairment on the acute care scale
The instrument's principal author was involved in the translation, validation, and writing of the evaluation. Therefore, the instrument was translated according to the guidelines described by Sousa (29). Two independent translators from a certified translation agency translated the questionnaire into Dutch. Two researchers (AK and EF) independently assessed these two translations to determine the optimal translation of the question formulations and the answer options.
During the translation process, there was some discussion about using the term ‘cognitive functioning’ or ‘cognitive status’, whereby all translators agreed upon the choice for ‘cognitive functioning’ because this term is commonly used in nursing care in the Netherlands. There were no disagreements on a lingual or cultural basis. There was unanimous consensus for the final selection of all items.
This Dutch version was also translated back into English by two other independent translators from the same certified translation agency. These translations were again independently assessed by the same researchers to decide on the best translation. This time, there was consensus on all of the items. The author reviewed this final English version, and the conclusion was that the outcomes closely resembled the original version. There were no specific reasons to expect systematic errors during the translation due to linguistic or cultural differences (30). The final version is attached as Additional file 1.
Sample size
According to the scientific literature, the sample size depends on the number of factors and the factor load, where a minimum sample size of 100 is recommended, and a sample size of 150 is suggested for three-phase models (30, 31). The COSMIN (Consensus-based Standards for the selection of health status measurement instruments) checklist advocates seven times the number of items (33). Based on this knowledge, the optimal sample size was at least 150 (34, 35). It may be noted that in the post hoc analysis, the sample size was sufficient for almost all estimated parameters to be (highly) significant.
Setting, recruitment, and participants
Six hospitals in the northern part of the Netherlands participated in this study and were supplemented by Dutch nurses who were recruited via LinkedIn and Facebook. The data were collected in one university hospital, two non-university teaching hospitals, and three rural hospitals. The capacity of the hospitals varied from 241 to 1300 beds, with additional outpatients.
Nurses with at least three months of experience in the clinical setting, working in the direct care of people with dementia, and willing to participate were included in the study. All hospital departments were included, except for paediatrics and obstetrics. The data collection took place from July 2018 to March 2019.
The recruitment of participants in the hospitals was performed by contact persons working in the hospital based on a convenience sample (27). The authors also used LinkedIn and Facebook to recruit hospital nurses. A general request was made for nurses to participate via LinkedIn, in which nursing managers are active and then the call was repeated once. For Facebook, which is often used by Dutch nurses, a different approach was used for which the authors requested two groups on Facebook. One was in a private group for questionnaires of a professional nursing magazine, and the other was in an open group for nurses in general. On Facebook, a daily update of the response was provided. This Facebook group has many members; however, it is not known how many members are active.
Data analysis
For the data analysis, we used IBM SPSS statistics (for Macintosh, version 25, Armonk, NY: IBM Corp.). Only complete scales were used in the data analysis. To perform confirmatory factor analysis (CFA), JASP (Version 0.11.1) with Lavaan was used (34). Before starting the analysis, Item 5 was reverse coded due to the negative wording of this item. The decision to use only completed scales was made based on the response rate of 159 complete cases instead of 164 with the inclusion of incomplete scales. Because the sample was sufficiently large and the differences in outcomes were minimal, it was decided that only completed questionnaires would be included. This makes the data as accurate as possible.
The Shapiro-Wilk test was used to assess the normality of the distribution. Descriptive analyses were used to describe the sample. Item performance was assessed by calculating item means and standard deviations, the inter-item correlation matrix, and the corrected item-total correlation.
The CFA was performed by robust maximum likelihood estimation, after which four types of fit indices were used to evaluate the fit of the model to the data: the chi-square model fit, the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean-square residual (SRMR). The Hoelter index was utilized to check the smallest sample size at which the chi-square interpretation would not be significant. As a criterion for significance, a p-value <0.05 was used. The model fit was considered acceptable if the following criteria were met: p-value for the χ2 model fit compared to the baseline model smaller than 0.05, CFI and GFI values between 0.90 and 0.95 or above RMSEA and RMR values of 0.08 or below (35).
Cronbach's alpha on the total scale and its subscales were assessed to determine the internal consistency.
Ethics approval and consent to participate
The study was performed following the Helsinki declaration, and all of the participants provided written informed consent before filling out the questionnaire. Nurses had an option to choose whether the results would also be available for further research. The Medical Ethical Committee of the University Medical Center Groningen considered approval unnecessary (decision M17.221048) because the questionnaire was intended for staff. The questionnaire was completely anonymous; no one could be identified based on the results. The managers received an email with a general explanation and a link to the questionnaire to forward it to the nurses of their team. Managers were not informed about the number of participating nurses from their ward or about their responses. Based on the contact persons' information and the response per ward, there was no reason to believe that nurses felt obliged to participate in this survey. The voluntary nature of participation was emphasized in the explanations.