Design and setting
A descriptive cross-sectional design was used in this study. The study employed a mobile phone app-based survey containing a three-part questionnaire (Wenjuanxing, www.wjx.cn). The data collection period was from 8th October to 22nd October 2020 (i.e., within a period of two weeks). The ethical approval was obtained from JNA and the First affiliated hospital of Nanjing Medical University (approval number: 2016-SRFA-076). The research was presented according to the STROBE checklist for cross-sectional research.
JNA distributed this anonymous, self-rated questionnaire to all nurses who attended the JNA training program in 2019 and other nursing members in Jiangsu via a WeChat applet and issued a notice to invite them to participate in the study. All study participants were provided with informed consent electronically prior to participating in the study. The informed consent page presented two options (Yes/No). Only subjects who chose “Yes” were guided to the questionnaire pages, and participants could quit the process at any time.
All responses from the nurses who had attended the JNA training program 2019 were included in the study. In order to represent the situation of nurses proportionally in the whole province, a regional-stratified sampling method was applied for responses from nurses who did not attend the training program. Registered nurses located in 13 prefecture-level cities in Jiangsu province were divided into 13 regions based on their geographic locations while the online questionnaire was distributed to all regions. All valid answers of the questionnaire from one region were included in the study when there were less than 1,000 responses in that region, otherwise 1,000 valid answers were randomly selected in case there were more than 1,000 responses in that region.
The online questionnaire consists of three parts: basic demographic and professional data, Chinese version of The Palliative Care Quiz for Nursing (PCQN-C), and Chinese version of the Frommelt Attitude Toward Care of the Dying scale (FATCOD-B-C).
Demographic and professional data
The first part of the questionnaire included participants demographics, nursing profession and their experience in palliative care. Demographic characteristic of nurses included gender, age, marital status, and personal beliefs. Nursing professional background included hospital classification, working years, level of nursing job, current working department, whether is an oncology nursing specialist, whether is a palliative care nursing specialist. Palliative care experience, and whether attended the JNA training program 2019.
The Palliative Care Quiz for Nursing (PCQN) instrument
The PCQN was originally designed by Ross et al (18), and was translated into Chinese and validated by Zou (19). The test-retest reliability was 0.782 and an internal consistency reliability of 0.758. The PCQN consisted of 20 questions with three possible responses including true, false and do not know. The PCQN contained three categories: (1) philosophy and principles of palliative care; (2) pain and symptom management; and (3) psychosocial and spiritual care (20). The total score ranged from 0-20, the higher the score the better the knowledge of palliative care.
The FATCOD-B-C instrument
The FATCOD-B was first developed by Formmelt to test nurses’ attitudes toward caring of the dying (21). It was modified and transferred into Chinese version (the FATCOD-B-C) by Wang to assess nurses’ and nursing students’ attitude towards caring for end-of-life patients in Mainland China (22). The FATCOD-B-C scale consisted of 29 items and it also classified into six subsets: Subset 1. attitude toward the interests of the dying person; Subset 2. attitude toward caring for the dying person; Subset 3. attitude toward the necessity of family support; Subset 4, attitude toward communication with the dying person; Subset 5. attitude toward fear of caring of dying person; Subset 6. attitude toward caring for the dying person’s families. It reported the construct validity of FATCOD-B-C was acceptable with Cronbach’s alpha coefficient 0.796 and the six subscales range form 0.610-0.863. The score from the 29 questions were added together to calculate a total FATCOD-B-C score with a possible range from 29-145, with the higher scores indicating more positive attitudes.
Before data collection, a pilot study was conducted with 20 nurses from one local hospital to assess the clarity of the questionnaire and to evaluate the feasibility and clarity of the study. The questionnaire took on average 8 to 12 minutes to complete, and all the nurses in the pilot study found the questionnaire was clear and easy to understand. The data generated from the pilot study was not included in the data analysis.
To minimize confounding and selection biases, a propensity score matched analysis was performed between the nurses who attained the JNA training program 2019 and whose who did not. We estimated propensity scores using logistic regression modelling of baseline demographics and other covariates (hospital classification, working experience, level of nursing job, working department, oncology nursing specialist, palliative care nursing specialist, experience of caring for dying patients, and experience of discussing death with patients or their family members) which could have impact on the knowledge and attitude. We conducted propensity score-matching using 1:1 nearest neighbor matching without replacement with a caliper width of 0.1. Covariate distributions between trained nurses and untrained nurses were balanced after conditioning on the propensity score (Supplementary table 3).
Data were analyzed using IBM SPSS Version 23.0. Descriptive analysis was used to describe the general data. For count data, frequencies and percentages were used. Comparisons between trained nurses versus untrained nurses were conducted in both the overall cohort as well as the propensity score matched subset by using independent samples t-test for continuous variables and the results were presented as mean difference with 95% confidence interval of the difference. P-value < 0.05 indicated that a difference was statistically significant.