Article Text

Quality of dying and death in intensive care units: family satisfaction
  1. Fur-Hsing Wen1,
  2. Ming Chu Chiang2,
  3. Chung-Chi Huang3,4,
  4. Tsung-Hui Hu5,
  5. Wen-Chi Chou6,7,
  6. Li-Pang Chuang3 and
  7. Siew Tzuh Tang8,9
  1. 1 Department of International Business, Soochow University - Downtown Campus, Taipei, Taiwan
  2. 2 Department of Nursing, Chang Gung Memorial Hospital Kaohsiung Branch, Kaohsiung, Taiwan
  3. 3 Department of Internal Medicine, Chang Gung Memorial Hospital Linkou Main Branch, Taoyuan, Taiwan
  4. 4 Department of Respiratory Therapy, Chang Gung University, Taoyuan, Taiwan
  5. 5 Department of Internal Medicine, Chang Gung Memorial Hospital Kaohsiung Branch, Kaohsiung, Taiwan
  6. 6 Department of Hematology-Oncology, Chang Gung Memorial Hospital Linkou Main Branch, Taoyuan, Taiwan
  7. 7 School of Medicine, Chang Gung University, Taoyuan, Taiwan
  8. 8 School of Nursing, Chang Gung University College of Medicine, Taoyuan, Taiwan
  9. 9 Division of Hematology-Oncology, Chang Gung Memorial Hospital Linkou Main Branch, Taoyuan, Taiwan
  1. Correspondence to Dr Siew Tzuh Tang, School of Nursing, Chang Gung University College of Medicine, Taoyuan 333, Taiwan; sttang{at}mail.cgu.edu.tw

Abstract

Objective This cohort study identified patterns/classes of surrogates’ assessment of their relative’s quality of dying and death (QODD) and to evaluate their associations with family satisfaction with intensive care unit (ICU) care.

Methods We identified QODD classes through latent class analysis of the frequency component of the QODD questionnaire and examined their differences in summary questions on the QODD and scores of the Family Satisfaction in the ICU questionnaire among 309 bereaved surrogates of ICU decedents.

Results Four distinct classes (prevalence) were identified: high (41.3%), moderate (20.1%), poor-to-uncertain (21.7%) and worst (16.9%) QODD classes. Characteristics differentiate these QODD classes including physical symptom control, emotional preparedness for death, and amount of life-sustaining treatments (LSTs) received. Patients in the high QODD class had optimal physical symptom control, moderate-to-sufficient emotional preparedness for death and few LSTs received. Patients in the moderate QODD class had adequate physical symptom control, moderate-to-sufficient emotional preparedness for death and the least LSTs received. Patients in the poor-to-uncertain QODD class had inadequate physical symptom control, insufficient-uncertain emotional preparedness for death and some LSTs received. Patients in the worst QODD class had poorest physical symptom control, insufficient-to-moderate emotional preparedness for death and substantial LSTs received. Bereaved surrogates in the worst QODD class scored significantly lower in evaluations of the patient’s overall QODD, and satisfaction with ICU care and decision-making process than those in the other classes.

Conclusions The identified distinct QODD classes offer potential actionable direction for improving quality of end-of-life ICU care.

  • End of life care
  • Family management
  • Terminal care
  • Bereavement
  • Psychological care

Data availability statement

Data may be obtained from a third party and are not publicly available. The sharing of anonymised data from this study is restricted due to ethical and legal constrictions. Data contains sensitive personal health information, which is protected under The Personal Data Protection Act, thus making all data requests subject to Institutional Review Board (IRB) approval. Per Chang Gung Memorial Hospital (CGMH) IRB, the data that support the findings of this study are restricted for transmission to those in the primary investigative team. Data sharing with investigators outside the team requires IRB approval. All requests for anonymised data will be reviewed by the research team and then submitted to the CGMH IRB for approval.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Improving end-of-life intensive care unit (ICU) care is an international priority for critical care to improve quality of death and dying and counteract escalating and unsustainable ICU expenditures.

  • Family members play a significant role in evaluating the quality of end-of-life care in ICUs but their heterogeneous evaluations of their relative’s quality of dying and death (QODD) are not fully understood.

WHAT THIS STUDY ADDS

  • High, moderate, poor-to-uncertain and worst QODD classes were identified, whereas prominent characteristics differentiate these classes including physical symptom control, emotional preparedness for death and amount of life-sustaining treatments received.

  • The four QODD classes significantly differed in surrogates’ evaluations of their relative’s overall QODD and ICU care and in family satisfaction with ICU care and decision making.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • The four distinct QODD classes offer actionable direction for improving end-of-life ICU care quality.

  • Providing adequate physical symptom management, facilitating sufficient emotional preparedness for death and initiating timely end-of-life-care discussions to redress family surrogates’ optimistic expectations of the effectiveness of life-sustaining treatments can improve end-of-life ICU care quality.

Introduction

Intensive care has grown substantially over the three past decades worldwide1–3 to one of the most costly and resource-intensive acute hospital services.1 Heavy utilisation of intensive care unit (ICU) care over the disease trajectory contributes to high costs of health care1 and raises concerns about optimal use of ICU resources,4 especially for end-of-life (EOL) care, which has increased in the last decades.5 Indeed, landmark studies highlight the improving but unsatisfactory quality of EOL ICU care.6–8 Improving EOL ICU care thus becomes a priority9 to promote value-concordant EOL care,10 improve quality of death and dying,9 reduce family surrogates’ bereavement grief,11 and counteract escalating and unsustainable ICU expenditures.4 10

To understand and improve EOL ICU care, well-developed measurements are prerequisite.12 The Quality of Dying and Death questionnaire (QODD) is widely used and validated to evaluate QODD among a variety of populations.13 However, the QODD has important shortcomings.12 First, studies that used the QODD commonly failed to capture dying patients’ values and preferences12 because most assessments are made retrospectively by bereaved family members.13 Yet, this approach not only overcomes the difficulties of obtaining evaluations of QODD from dying patients14 but also centres family members’ perspectives for improving EOL care. Family members are an integral part of ICU care and commonly play a critical surrogate decision-making role for their relative,11 therefore their evaluations of EOL-care quality in ICUs are legitimate. Second, neither the factor structure for the proposed concept of QODD nor the underlying domains of the QODD have been satisfactorily settled.12–17 However, the QODD items reflect aspects identified by patients and family members as most important for high-quality EOL care,18 19 thereby representing important values to measure EOL-care quality in ICUs and no items on the QODD are neglectable. Third, though each QODD item is composed of frequency and quality-rating components, the quality-rating component was often used alone to ascertain QODD.7 14–17 20–22 When the frequency component was reported,7 23 24 multiple items make it difficult to comprehensively understand family members’ heterogeneous evaluations of their relative’s QODD. Therefore, the purposes of this study were to identify the different patterns of bereaved family surrogates’ evaluation of their relative’s QODD through latent class analysis (LCA) of the frequency component of the QODD and to initially assess validity of these innovatively identified latent classes in evaluating QODD.

Materials and methods

Study design, setting and sampling strategy

Study data were from a prospective cohort study focused on quality of EOL care in ICUs25 and its impact on bereavement adjustment.26 Participants were consecutively recruited from January 2018 to January 2020 and followed through June 2021 from two academically affiliated hospitals’ level III medical ICUs in Taiwan. Study ICUs were staffed by intensivists, had a 1:2 nurse-to-patient ratio per shift, and implemented an open visiting policy. In Taiwan, the government and the National Health Insurance programme promote palliative care, a do-not-resuscitate order for dying patients, and formal family meetings. Bedside physician-family prognostic and EOL-care discussions were commonly conducted.

Eligible ICU patients were identified as those at high risk of death, that is, had an Acute Physiology and Chronic Health Evaluation II score >20. Patients who died within 3 days of ICU admission were excluded, as their short ICU stay may not allow for implementation of high-quality EOL care.17 20 24 26 Adult family members who self-identified as having legal authority to act as surrogate for their relative’s medical decisions were eligible to participate; one surrogate was recruited per patient. Each surrogate signed informed consent for their participation and for review of the patient’s medical record.

Data collection

Family surrogates’ evaluation of their relative’s QODD and their satisfaction with EOL ICU care was assessed by phone interviews with experienced, trained research assistants approximately 1 month post patient death as adopted by previous studies.16 20 21 23

Measures

The QODD in ICUs was measured by the ICU-QODD. The 23-item ICU-QODD is an interviewer-administered questionnaire that assesses on six domains: Symptoms and Personal Care, Preparation for Death, Family Concerns, Treatment Received, Whole Person Concerns and Moment of Death.24 This measure excludes items that may not be applicable for evaluating QODD in ICUs (eg, being able to feed oneself, spending time with pets) or for cultural considerations (eg, having means to end life if needed). Only the frequency component of each QODD item was used in this study and was evaluated from 0 (none) to 5 (always) or existence (yes or no) with options of ‘don't know’. Two single-item summary questions explore overall quality of the patient’s dying and EOL-care in ICUs on a scale from 0 (terrible experience/worst care) to 10 (almost perfect experience/best care).

Family satisfaction with EOL care in ICUs was measured with the Family Satisfaction in the ICU questionnaire (FS-ICU),27 28 the most widely validated instrument for assessing FS-ICU care.29 The FS-ICU Care subscale (14 items) assesses (1) satisfaction with information access, (2) care quality, continuity and accessibility, and (3) ICU and waiting room atmosphere. The FS-ICU Decision-Making subscale (10 items) measures family satisfaction with the information and support received during the decision-making process. Item responses were rescaled from the original 1–5 Likert scale to a scale from 0 (least satisfied) to 100 (most satisfied). Mean scores of each subscale were calculated.

Statistical analysis

Each item on the ICU-QODD was first examined with descriptive statistics. We retained items with a ‘don't know’ response and classified the six-category responses of the frequency component into dichotomous variables: none/a little bit/some/a good bit of the time vs most/all of the time23 24 for identifying patterns of bereaved surrogates’ evaluation of their relative’s QODD using LCA by Mplus (V.8.6). In this approach, evaluations of QODD were treated as a pattern or set of QODD items (‘latent class’) rather than individual items.30 LCA divides bereaved surrogates into mutually exclusive probabilistic classes based on shared characteristics that discriminate among members of each class. LCA simultaneously estimates class probabilities (the number and relative size of classes) and conditional probabilities across classes. Conditional probability reflects the probability that a surrogate in a given class experienced that QODD item none/a little bit/some/a good bit of the time, most/all of the time or unknown.30

Best model solutions with an increasing number of classes were selected by model fit indices, including several information criteria (IC): highest log-likelihood, smallest Akaike (AIC), consistent AIC, Bayesian (BIC) and sample-size adjusted BIC.30 The BIC is commonly recognised as the best of the IC but when the IC values between consecutive numbers of hidden classes in plots of IC values versus class number flat, this situation suggests that the increase in class number is not meaningful.30 31 These criteria, parsimony, and the clinical meaningfulness of the latent class identification were factored in deciding the optimal number of classes.

To validate our innovatively identified QODD latent classes, we examined differences across the classes in the scores of the two single-item summary questions on the QODD and the FS-ICU Care and FS-ICU Decision-Making subscales by one-way analysis of variance with Dunnett’s tests for post hoc analyses.

Results

Participant characteristics

Among the 353 consecutive eligible patients who died in the ICUs, 309 family members (87.5%) participated in the first month bereavement survey and constituted the study participants (figure 1). Participants and non-participants of bereavement surveys did not differ in patient or family demographics (online supplemental table 1–2). Characteristics of patient and family participants are shown in table 1. Family evaluations of QODD, quality of ICU care, satisfaction with ICU care and satisfaction with decision-making process in ICUs were low to moderate.

Supplemental material

Figure 1

Participant flow chart. ICU, intensive care unit.

Table 1

Characteristics of participants (N=309)

Bereaved surrogates’ QODD-item responses

Most bereaved surrogates reported that their relative experienced the following most/all of the time: spending time with family/friends (63.1%), feeling at peace with dying (59.9%), and having pain under control (59.2%) (table 2). Approximately half of bereaved surrogates reported that their relative had no concerns of imposing strains on family (50.2%), could breathe comfortably (49.5%) and keep dignity and self-respect (49.2%) most/all of the time. A minority of surrogates reported that patients had control of events (12.9%) and did not fear dying (20.1%) most/all of the time.

Table 2

Frequency of experiencing each indicator reported by bereaved family members of deceased ICU patients (N=309)

Nearly all patients had healthcare costs covered (99.7%) (table 2), were unconscious at death (99.7%), and were intubated with mechanical ventilation support (99.0%). Surrogates reported that patients commonly had families present at the moment of death (86.1%), were touched/hugged by family (83.2%), had a religious ceremony take place (69.3%) and funeral arrangement in order (52.8%), and said goodbye to family (53.7%). A minority of patients received cardiopulmonary resuscitation (CPR) (11.3%) or dialysis (31.7%) in ICUs and had religious visits (19.7%). An unknown response was endorsed by >20% of bereaved surrogates for three items: whether their relative cleared up bad feelings (67.0%), feared dying (47.9%), and worried of imposing strains on family (36.6%).

Identification of QODD classes by LCA

Four items without variances (99.0%–100.0%) across participants (table 2) were omitted for LCA. Model fit indexes for the one- through nine-class LCA solutions of QODD classes and plots of ICs values are in online supplemental appendix 1 and 2, respectively. The four-class solution was selected as optimal due to best combination of ICs (giving higher priority to the plots of ICs values), parsimony and clinical meaningfulness. These four distinct classes (prevalence) were named high (41.3%), moderate (20.1%), poor-to-uncertain (21.7%) and worst (16.9%) QODD.

Supplemental material

The high QODD class is characterised by optimal physical symptom control, moderate-to-sufficient emotional preparedness for death with family uncertainty about patient fear of dying and slight control of events, ample religious support, strong family support, few life-sustaining treatments (LSTs) received and no funeral arrangement in order (table 3). Patients in the moderate QODD class had adequate physical symptom control, moderate-to-sufficient emotional preparedness for death with family uncertainty about their clearing up bad feelings and fearlessness of death, some control of events, ample religious support, moderate-to-strong family support, funeral arrangement in order and the least LSTs received (table 3). Patients in the poor-to-uncertain QODD class had inadequate physical symptom control, insufficient/uncertain emotional preparedness for death, slight control of events, adequate religious support, moderate-to-strong family support, some LSTs received and funeral arrangement in order (table 3). Patients in the worst QODD class had poorest physical symptom control; insufficient-to-moderate emotional preparedness for death but minimal concerns of imposing strains on family, and uncertainty about clearing up bad feelings; little control of events; limited religious support; poor-to-moderate family support; substantial LSTs received and funeral arrangement in order (table 3).

Table 3

Emission probabilities and sizes of the four classes of quality of dying and death in ICUs

Associations of QODD classes with bereaved surrogates’ overall evaluation of QODD and ICU care, and satisfaction with EOL care in ICUs

The four QODD classes significantly differed in bereaved surrogates’ overall evaluations of their relative’s QODD and ICU care as measured by the single-item summary questions on the QODD (table 4). Bereaved surrogates in the high and moderate QODD classes evaluated their relative’s overall QODD higher than those in the poor-to-uncertain and worst QODD classes. Bereaved surrogates in the worst and the moderate QODD classes evaluated overall quality of EOL ICU care as poorest and highest, respectively.

Table 4

Bereaved family members’ evaluation of overall quality of end-of-life care and satisfaction with ICU care across the four classes of quality of dying and death

Bereaved surrogates’ satisfaction with ICU care and decision-making process significantly differed by QODD class (table 4). Bereaved surrogates in the worst QODD class scored lowest in the FS-ICU Care and FS-ICU Decision-Making subscales. Furthermore, bereaved surrogates in the high and moderate QODD classes scored significantly higher in the FS-ICU Care than those in the poor-to-uncertain QODD class, whereas only surrogates in the high QODD class scored significantly higher in the FS-ICU Decision-Making subscale than those in the poor-to-uncertain QODD class.

Discussion

We observed that Taiwanese patients dying in the ICUs commonly had healthcare costs covered, families present at death, a religious ceremony take place and funeral arrangement in order, were touched/hugged by families, and did not receive CPR or dialysis (table 2). However, bereaved surrogates reported their relative’s poor QODD in physical symptom control and emotional preparedness for death (ie, feeling at peace with dying, feeling unafraid of dying, clearing up bad feelings, keeping dignity and self-respect, saying goodbye to family and not worrying about imposing strains on family). Proportion of patients who died with adequate physical symptom control (49.5%–59.2%), sufficient emotional preparedness for death (20.1%–59.9%) and sense of control of events most or all of the time (12.9%) is similar to or higher than other studies using the QODD (3%–47%, 30%–47% and 8%, respectively from the USA24 and 21.4%–39.3%, 7.9%–30.3% and 6.8%, respectively from the Netherlands).7 However, Levy et al reported substantially higher rates of adequate physical symptom control (58%–88%) and sufficient emotional preparedness for death (63%–79%) (indicated by maintaining dignity and keeping peace with death only).23 Taiwanese ICUs may have room to improve in providing physical symptom management and facilitating emotional preparedness for death.

We used LCA to identify four distinct QODD classes from the frequency component of the QODD. The four QODD classes significantly differed in surrogates’ evaluations of their relative’s overall QODD and ICU care and in FS-ICU care and decision making. Bereaved surrogates in the worst QODD class scored lowest in all aspects (table 4). Prominent underlying characteristics differentiate these four QODD classes, including physical symptom control, emotional preparedness for death and amount of LSTs received (online supplemental table 3). These unique features, unlike religious and family support or funeral arrangement, are all modifiable by high-quality EOL ICU care. Surrogates in the high/moderate QODD classes indicated optimal/adequate physical symptom control, moderate-to-sufficient emotional preparedness for death, and least or few LSTs received (table 3), markedly distinguishing these two classes from the poor-to-uncertain and worst QODD classes (online supplemental table 3). These underlying modifiable factors are associated with surrogates’ evaluation of overall QODD, EOL care and satisfaction with ICU care and may represent important targets for future interventions (table 4).

Our identified high/moderate versus poor-to-uncertain/worst QODD classes and their differential associations with bereaved surrogates’ evaluation of QODD and satisfaction with ICU care further support the importance of appropriate symptom management to achieve a good death for critically ill patients.32 The Institute of Medicine32 characterises good deaths as free from avoidable symptom distress. Minimising pain and other distressing symptoms at EOL is a widespread goal for dying patients and their family members.18 19 Adequate assessment and management of distressing symptoms like pain and dyspnoea at EOL remain the cornerstone of achieving comfort and are a top priority for providing dignified death in ICUs.33–35 Indeed, implementation of assessment-driven and standardised symptom management protocols such as the ABCDEF bundle36 improves ICU patient-centred outcomes and clinical practice.36

Emotional preparedness for forthcoming death19 37 also differentiates high/moderate versus poor-to-uncertain/worst QODD classes. Emotional preparedness for death implies emotionally accepting one’s dying role,37 feeling at peace with dying, maintaining dignity and self-respect within the constraints of present circumstances, and preparing family members for life without oneself, including closing and reconciling relationship bonds with family members, saying goodbyes to family and not imposing unendurable strains/burden on them.19 37 EOL ICU care that promotes patient dignity, self-respect, fearlessness and peacefulness may lead bereaved surrogates to evaluate emotional preparedness for death as moderate to sufficient and improve evaluation scores for QODD and satisfaction with care.

Amount of LSTs received further differentiates the high/moderate QODD classes from the poor-to-uncertain/worst QODD classes with few/least and some/substantial received LSTs, respectively (table 3 and online supplemental table 3). This characteristic may have driven bereaved surrogates in the moderate QODD class to evaluate overall quality of ICU care for their relative higher than those in the high or poor-to-uncertain QODD classes (table 4). Amount of LSTs received coupled with optimal physical symptom management and strong family support (online supplemental table 3) also distinguished bereaved surrogates in the high QODD class (but not bereaved surrogates in the moderate QODD class) from the poor-to-uncertain QODD class (table 4), which may have driven higher satisfaction with the decision-making process in ICUs. Though evidence shows surrogates commonly disagree with physicians on treatment appropriateness with overaggressive treatment endorsed by surrogates,38 critically ill patients are willing to decline aggressive LSTs39 to avoid states worse than death.40 Thus, in the absence of substantial LSTs, especially CPR, bereaved surrogates may retrospectively evaluate better QODD and ICU care, and greater satisfaction with EOL-care decision making due to accordance with patients’ wishes,10 32 39 40 avoidance of suffering and peacefulness of death.

Our study has several limitations. Participants were sampled from two hospitals in Taiwan, which may limit generalisability of findings to national or international target populations with substantially different cultural, societal and healthcare contexts. We acknowledge the high proportion of study participants with a cancer diagnosis which may not be representative of ICU decedents in Taiwan or globally. Preparedness for the loved one’s forthcoming death and appraisal of QODD may be different between family surrogates of ICU patients dying with cancer and with other diseases which will be examined in our future studies. Our findings warrant further validation with a broader sample. Second, our results do not apply to family members of patients who died within 3 days of ICU admission, or to those who did not participate in post-death surveys. Third, only the frequency component of the QODD was used in this study, and our validation of the four distinct QODD classes requires external validation. Whether items aggregated in each dimension identified in this study could represent ‘factor structure’ of the QODD in other samples, whether each item contributed independently or synergistically to its proposed dimension, and whether the four distinct QODD classes are associated with family surrogates’ bereavement outcomes warrant further validation. We have not controlled for covariates and cannot exclude the possibility of unmeasured residuals commonly found in observational studies impacting associations between QODD class and bereaved surrogates’ evaluations of QODD and quality of ICU care.

Conclusion and clinical implications

Our identification of four distinct QODD classes and their unique, non-equivalent associations with bereaved surrogates’ evaluations of the patient’s QODD and ICU care offer potential actionable direction for improving quality of EOL ICU care. Providing appropriate physical symptom management, facilitating sufficient emotional preparedness for death and initiating timely EOL-care discussions to redress family surrogates’ optimistic misunderstanding/expectations of the effectiveness of LSTs may optimise surrogates’ evaluation of their relative’s QODD and quality of EOL care and decision making in ICUs.

Data availability statement

Data may be obtained from a third party and are not publicly available. The sharing of anonymised data from this study is restricted due to ethical and legal constrictions. Data contains sensitive personal health information, which is protected under The Personal Data Protection Act, thus making all data requests subject to Institutional Review Board (IRB) approval. Per Chang Gung Memorial Hospital (CGMH) IRB, the data that support the findings of this study are restricted for transmission to those in the primary investigative team. Data sharing with investigators outside the team requires IRB approval. All requests for anonymised data will be reviewed by the research team and then submitted to the CGMH IRB for approval.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by Chang Gung Medical Foundation, Institutional Review Board; approval number: No: 201700343B0. Participants gave informed consent to participate in the study before taking part.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • F-HW and MCC contributed equally.

  • Contributors F-HW and MCC had equal contribution. F-HW, MCC, T-HH, C-CH, W-CC, L-PC and STT contributed substantially to the study conception and design. T-HH, MCC, C-CH, W-CC, L-PC contributed by providing study patients. T-HH, MCC, C-CH, W-CC, L-PC and STT contributed to collection and/or assembly of data. F-HW, MCC, T-HH, C-CH, W-CC, L-PC and STT contributed to data analysis and interpretation. F-HW, MCC, T-HH, C-CH, W-CC, L-PC and STT contributed to the writing and final approval of the manuscript. The corresponding author (STT) takes responsibility for the content of the manuscript, has full access to all of the data in the study, and is responsible for the integrity of the data, the accuracy of the data analysis, including and especially any adverse effects.

  • Funding National Health Research Institutes (NHRI-EX110-10704PI) with partial support from Ministry of Science and Technology (MOST 108-2314-B-182-061-MY3) and Chang Gung Memorial Hospital (BMRP888).

  • Disclaimer No funding sources had any role in designing and conducting the study; collecting, managing, analyzing, and interpreting the data; or preparing, reviewing, or approving the article.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.