End-of-life care preferences of the general public and recommendations of healthcare providers: a nationwide survey in Japan

The aim of this study was to clarify differences between preferences of the general public and recommendations of healthcare providers with regard to treatment, EOL care, and life-sustaining treatment (LST) based on a hypothetical scenario involving a patient with advanced cancer.

This study comprised exploratory post-hoc analyses of “The Survey of Public Attitude Towards Medical Care at the End of life”, which was a population based, cross-sectional anonymous survey to investigate public attitudes toward medical care at the end of life conducted descriptively by the Ministry of Health, Labour and Welfare (MHLW), Japan in December 2017.

The source population of the original study was members of the general Japanese public aged 20 and over. The stratified two-stage random sampling method was used for selection. The survey was administered by mail in questionnaire packs. In addition to the questionnaire, the packs sent out by post also included explanatory letters from the MHLW, and stamped, addressed envelopes to return the questionnaire.

The participants of this study were members of the general public, physicians, nurses, and care staff. We used the general public data from the original study, and combined it with the data of healthcare providers in order to conduct exploratory post-hoc analyses, though the original study was conducted just for description not for analysis by the government. The questionnaire was sent by mail to 6000 members of the general public, 4500 physicians, 6000 nurses, and 2000 care staff, accompanied by a letter that concisely explained the survey. The details on how to select target participant and facilities, and how to distribute the questionnaire were described in our previous study [14]. The MHLW sent a reminder to all non-responders in January 2018. Completion and return of the questionnaire, in combination with the explanatory letter, was considered to indicate voluntary and informed consent to participate. The institutional review board of the University of Tsukuba approved the protocol of this study.

As there were no specific and validated instruments for evaluating the preferences of the general public and recommendations of healthcare providers regarding treatment, EOL care, and LST, we developed an original questionnaire based on data from previous studies [11, 15‐17] and discussion among the authors of this study. Subsequently, we submitted the draft questionnaire to the MHLW and they made the final decision about the questionnaire items.

We used a scenario about a hypothetical cancer patient to investigate preferences for cancer treatment, EOL care, and LST based on the previous nationwide survey conducted in 2011 [15]. We submitted the draft questionnaire to the MHLW and they made the final decision about the questionnaire items. The respondents were asked to imagine that they were diagnosed with terminal cancer and going to die within 1 year (see Table 1), and were asked to express their preferences (or recommendations in the case of physicians, nurses, and care staff) about cancer treatment, EOL care, and LST. The treatments investigated were chemotherapy or radiation therapy for cancer (chemotherapy or radiation), fluid infusion if unable to drink water (fluid infusion), total parenteral nutrition (TPN) if unable to intake sufficient nutrition orally (TPN), nasogastric (NG) tube feeding if unable to intake sufficient nutrition orally (NG tube feeding), percutaneous endoscopic gastrostomy (PEG) tube feeding if unable to intake sufficient nutrition orally (PEG tube feeding), mechanical ventilation if it became difficult to breathe (ventilation), and cardiopulmonary resuscitation (CPR) if the heart or breathing stopped (CPR). The general public was asked to select from the following options: want treatment or care, do not want treatment or care, or not sure. For physicians, nurses, and care staff, the options were as follows: recommend treatment or care, do not recommend treatment or care, or not sure.

We collected the following background characteristics of the general public based on previous studies and discussion among the authors of this study: age in 5-year intervals from 20 to over 85, gender, living status, education, family doctor, and death of a close person within 5 years [11, 15]. We also collected the following background characteristics of the healthcare providers based on previous studies [11, 16, 17] and discussion among the authors: years of practice, workplace, frequency of caring for dying patients, identifying the proxy decision maker as standard practice, and participation in a nationwide education program on EOL discussion (binary category) [Palliative Care Emphasis Program on Symptom Management and Assessment for Continuous Medical Education (PEACE) and/or Education For Implementing End-of-Life Discussion (E-FIELD) for healthcare providers] [18‐20]. PEACE is a 2-day interactive education program for physicians that integrates palliative care and psycho-oncology. It has been available since 2007. E-FIELD is a 1- or 2-day interactive education program for healthcare providers about respecting the patient’s wishes for EOL care and practicing based on the best interests of the patient, which has been held since 2016.

We initially conducted descriptive analyses of categorical variables. We defined “chemotherapy or radiation” as aggressive cancer treatment, “fluid infusion”, “NG”, “TPN”, and “PEG tube feeding” as EOL care, and “ventilation” and “CPR” as LST. We defined the answer “want/recommend treatment or care” as the preference for aggressive cancer treatment, EOL care and LST, and we defined “not sure” as missing data because it can be interpreted as both positive and negative opinions. We defined healthcare providers who answered “identify the proxy decision maker at some point (ex. when diagnosed as incurable disease or when death is approaching as the disease progresses)” as healthcare providers who identified the proxy decision maker.

Subsequently, we compared preferences about each type of cancer treatment, EOL care, and LST (seven in total) using the chi-square test or Fisher’s exact test. Lastly, we conducted multivariate logistic regression analysis of preferences for the seven treatments in each respondent group.

For logistic regression analysis of the general public, we used the following six independent variables based on previous reports and discussion among the authors of this study [15, 21, 22]: gender (reference: male), age <  64 years, education (reference: junior high or high school), living with family, death of a close person in the past 5 years, and presence of a family doctor.

The highest preferences of the general public for cancer treatment, EOL care, and LST were as follows": "Fluid infusion" (48.5%), "Chemotherapy or radiation" (27.5%), and “TPN” (13.8%) (Additional file 1). Similarly, the most common recommendation about cancer treatment, EOL care, and LST made by physicians, nurses, and care staff was "Fluid infusion" (physicians: 59.5%, nurses: 56.4%, care staff: 53.6%) (Additional file 1).

The proportion of the general public who wanted “Chemotherapy or radiation” and “Ventilation” was significantly higher than the level of recommendation by physicians, nurses, and care staff (physicians: p < 0.001, nurses: p < 0.001, care staff: p = 0.006) (Fig. 1).

In addition, the proportion of the general public who wanted “CPR” was significantly higher than the level of recommendation by physicians and nurses (p < 0.001), although it was significantly lower than that by care staff (p < 0.001) (Fig. 1).

Multivariate logistic regression analysis revealed that women among the general public had significantly lower preferences for “TPN”, “NG tube feeding”, and “CPR”. In addition, members of the general public over the age of 65 had significantly lower preferences for “TPN”, “NG tube feeding”, “PEG tube feeding”, “Ventilation”, and “CPR”. Other factors, such as education, living status, death of a close person in the past 5 years, and presence of a family doctor, did not have a significant relationship with preferences for cancer treatment, EOL care, and LST (Tables 3 and 4).

Physicians who had participated in a nationwide training program were significantly less likely to recommend “Fluid infusion”, “TPN”, “NG tube feeding”, and “CPR”, whereas those working in hospitals were significantly more likely to recommend “Fluid infusion”, “TPN”, “NG tube feeding”, “PEG tube feeding”, and “CPR”.

Nurses who cared for at least one dying patient per month were significantly less likely to recommend all seven types of cancer treatment, EOL care, and LST, although nurses working in hospitals were significantly more likely to recommend “Chemotherapy or radiation”, “Fluid infusion”, and “PEG tube feeding”.

The care staff who worked in hospitals were significantly more likely to recommend “Chemotherapy or radiation”, “TPN”, and “PEG tube feeding”, whereas those who cared for at least one dying patient per month were significantly less likely to recommend “CPR” (Tables 3 and 4).

Physicians were more likely to recommend NG tube feeding and PEG tube feeding than nurses, and were more likely to recommend chemotherapy or radiation therapy, TPN, and NG tube feeding than care staff.

Nurses were more likely to recommend TPN than physicians and care staff. Care staff had a higher rate of recommending CPR than physicians and nurses (Additional file 2).