Article Text

Original research
Modification of social determinants of health by critical illness and consequences of that modification for recovery: an international qualitative study
  1. Joanne McPeake1,2,
  2. Leanne Boehm3,
  3. Elizabeth Hibbert4,
  4. Katrina Hauschildt5,
  5. Rita Bakhru6,
  6. Anthony Bastin7,
  7. Brad Butcher8,
  8. Tammy Eaton9,10,
  9. Wendy Harris11,
  10. Aluko Hope12,
  11. James Jackson13,
  12. Annie Johnson14,
  13. Janet Kloos15,
  14. Karen Korzick16,
  15. Judith McCartney2,
  16. Joel Meyer17,
  17. Ashley Montgomery-Yates18,
  18. Tara Quasim2,19,
  19. Andrew Slack20,
  20. Dorothy Wade21,
  21. Mary Still22,
  22. Giora Netzer23,
  23. Ramona O Hopkins24,25,
  24. Mark E Mikkelsen26,
  25. Theodore Iwashyna27,
  26. Kimberley Haines28,
  27. Carla Sevin29
  1. 1Critical Care, The Healthcare Improvement Studies Institute, University of Cambridge, Cambridge, UK
  2. 2Critical Care, NHS Greater Glasgow and Clyde, Glasgow, UK
  3. 3School of Nursing, Vanderbilt University, Nashville, Tennessee, USA
  4. 4Department of Physiotherapy, Western Health Foundation, Sunshine, Victoria, Australia
  5. 5Center for Clinical Management Research, VA Ann Arbor Healthcare System, Ann Arbor, Michigan, USA
  6. 6Department of Internal Medicine, Wake Forest University School of Medicine, Winston-Salem, North Carolina, USA
  7. 7Department of Peri-operative Medicine, Barts Health NHS Trust, London, UK
  8. 8Department of Critical Care Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
  9. 9VA HSR&D Center for the Study of Healthcare Innovation, Implementation, & Policy, University of Michigan, Ann Arbor, Michigan, US
  10. 10Department of Systems, Populations and Leadership, University of Michigan School of Nursing, Ann Arbor, Michigan, US
  11. 11Intensive Care Unit, University College London, London, UK
  12. 12Department of Medicine, Division of Pulmonary, Allergy and Critical Care Medicine, Oregon Health & Science University (OHSU), Portland, Oregon, USA
  13. 13Critical Illness, Brain Dysfunction, and Survivorship (CIBS) Center, Vanderbilt University, Nashville, Tennessee, USA
  14. 14Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, Minnesota, USA
  15. 15Department of Acute and Critical Care Nursing, University Hospitals Cleveland Medical Center, Cleveland, Ohio, USA
  16. 16Department of Pulmonary and Critical Care Medicine, Geisinger Medical Center, Danville, Pennsylvania, USA
  17. 17Department of Critical Care, Guy's and St Thomas' NHS Foundation Trust, London, UK
  18. 18University of Kentucky, Lexington, Kentucky, USA
  19. 19School of Medicine, Dentistry and Nursing, University of Glasgow College of Medical Veterinary and Life Sciences, Glasgow, UK
  20. 20Department of Critical Care, Guy's and St Thomas' Hospitals NHS Trust, London, UK
  21. 21Critical Care, University College London Hospitals NHS Foundation Trust, London, UK
  22. 22Critical Care, Emory University Hospital, Atlanta, Georgia, USA
  23. 23Division of Pulmonary and Critical Care, University of Maryland School of Medicine, Baltimore, Maryland, USA
  24. 24Center for Humanizing Critical Care and Pulmonary/Critical Care Medicine, Intermountain Medical Center, Murray, Utah, USA
  25. 25Psychology and Neuroscience, Brigham Young University, Provo, Utah, USA
  26. 26Division of Pulmonary, Allergy, and Critical Care Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania, USA
  27. 27Department of Internal Medicine, Division of Pulmonary & Critical Care, University of Michigan, Ann Arbor, Michigan, USA
  28. 28Department of Physiotherapy, Sunshine Hospital, Melbourne, Victoria, Australia
  29. 29Department of Medicine, Division of Allergy, Pulmonary, and Critical Care Medicine, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  1. Correspondence to Dr Joanne McPeake; joanne.mcpeake{at}


Objectives Social determinants of health (SDoH) contribute to health outcomes. We identified SDoH that were modified by critical illness, and the effect of such modifications on recovery from critical illness.

Design In-depth semistructured interviews following hospital discharge. Interview transcripts were mapped against a pre-existing social policy framework: money and work; skills and education; housing, transport and neighbourhoods; and family, friends and social connections.

Setting 14 hospital sites in the USA, UK and Australia.

Participants Patients and caregivers, who had been admitted to critical care from three continents.

Results 86 interviews were analysed (66 patients and 20 caregivers). SDoH, both financial and non-financial in nature, could be negatively influenced by exposure to critical illness, with a direct impact on health-related outcomes at an individual level. Financial modifications included changes to employment status due to critical illness-related disability, alongside changes to income and insurance status. Negative health impacts included the inability to access essential healthcare and an increase in mental health problems.

Conclusions Critical illness appears to modify SDoH for survivors and their family members, potentially impacting recovery and health. Our findings suggest that increased attention to issues such as one’s social network, economic security and access to healthcare is required following discharge from critical care.

  • adult intensive & critical care
  • public health
  • qualitative research

Data availability statement

Data are available upon reasonable request. The dataset used is available on reasonable request from the corresponding author.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

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.


  • This international multicentre study used in-depth semistructured interviews to understand how the social determinants of health were modified by critical illness.

  • Using replicable, rigorous qualitative methods, this reserach suggests a complex interplay between the social determinants of health and recovery from critical illness.

  • Although the sample size is considerable for a qualitative study, we recognise that we have used a convenience sample, with a small portion of all critical care survivors in these countries.


The social determinants of health (SDoH) are the non-medical factors which influence health outcomes and include both upstream policy, environmental, and context factors and their manifestations in terms of individual material and social hardship.1 There has been a growing realisation of their central role in shaping the capacity of individuals to not only access high-quality care, but also to benefit from such care.2 In some discussions, particularly around the role of SDoH impacting outcomes from acute illness and hospitalisation, such factors are conceptualised as causally antecedent to medical care and illness.

Illness can itself result in material and social hardship, opening the possibility of a bidirectional relationship.3 Financial toxicity resulting from oncological care has been documented and is also found after surgery and respiratory failure.4 5 Adverse employment outcomes are common after acute myocardial infarction, stroke and critical illness.6–8 While changes in individual economic and social situations may be exacerbated by gaps in the US social safety net, many have been documented throughout the industrialised world.9

Within the critical care context, more detailed evidence is emerging about the potential interaction between SDoH and recovery. For example, a recent multicentre study from the USA revealed that social isolation was a risk factor for poor outcomes (mortality) among older adults surviving critical illness.10 Similarly, in the UK, recent data have shown that almost one-third of participants requested a social and welfare consultation during an intensive care unit (ICU) recovery clinic, in order to alleviate ongoing welfare and social issues.11 However, a full and systematic understanding of the challenges survivors face, alongside how critical illness may worsen SDoH, is lacking.

The objective of this study, therefore, was to identify SDoH that were modified at the individual level by the experience of critical illness, and the effect of such modifications on patients’ and families’ recovery. We conducted an international qualitative study of the experience of recovery from critical illness with patients and caregivers from 14 different hospitals across Australia, the USA and the UK. We deliberately recruited patients from three different countries to ensure that any international differences in social contexts and recovery could be delineated. The social context in each country and indeed the healthcare systems are different. The USA, for example, has an insurance-based healthcare system, whereas the UK has a national healthcare system and Australia a mixture of both. By including these diverse approaches to healthcare delivery, we could understand the international context better.


Design and setting

We used a qualitative, descriptive design. Sites participating in the Society of Critical Care Medicine’s (SCCM) THRIVE Programme12 offered ICU recovery programmes in the form of post-ICU clinics and peer support programmes were invited. Patients who had not received any specific intervention were also interviewed (table 1).

Table 1

Participant demographics

The THRIVE Programme was a programme of work run via the SCCM for 5 years (2015–2019). It actively recruited hospitals internationally to work with the society to support innovation in the field of ICU recovery . It ran two collaboratives: peer support and post-ICU clinics. We invited sites to take part in this programme of work from these two collaboratives. All sites in the initial recruitment waves (2015, 2016) were from Australia, the USA and the UK, with the majority of sites from the USA.

We chose to undertake interviews internationally, in three developed nations, where the social context for support and health needs could be fully understood. The purpose of this was to understand if different developed healthcare systems supported care in diverse ways. The sample size was decided upon through analysing previous research in the field and through iterative discussions with the research team. This approach was taken across the entire sample and not a site level. The number of patients included in the final sample was based on the number of THRIVE sites in each country. Not all THRIVE sites were able to be involved in the research process due to staffing limitations and access to research support at these institutions.

All patients attending ICU recovery programmes were invited by professionals at each site if they met: (1) inclusion criteria—English-speaking patients older than 18 years admitted to the ICU, or caregiver of a patient who survived critical illness; and not (2) exclusion criteria—ongoing severe neurological and/or cognitive impairment or continued inpatient care in hospital or rehabilitation. Informed consent was obtained before each interview.

Data collection

Data were collected through in-depth semistructured scripted interviews via telephone. Separate questions were used for caregivers and patients; these were adapted for those who did not receive a post-critical illness intervention (online supplemental file 1). Questions were generated from previous literature and through iterative discussion within the research group (JMcP, LB, KJH, CS). All researchers, alongside patient representatives, discussed the interview script to ensure consistency. Some interviewers were known to the participants through their role in direct clinical care. Data were audio-recorded and transcribed verbatim.

Data analyses

We sought to understand how SDoH were potentially modified at the individual level by the experience of critical illness, and the effect of such modifications on recovery. To do so, we systematically mapped our analysis against a set of predefined concepts related to SDoH, adapted from a public policy framework,13 including money and work; education and skills; housing, transportation and neighbourhoods; and family, friends and social connections. In this analysis, we defined a ‘rural’ setting as a location out of a main town or city.

The Framework Analysis technique was used to analyse data across these concepts,14 through seven stages: (1) transcription; (2) familiarisation with the interview; (3) coding; (4) developing a working analytical framework; (5) applying the analytical framework; (6) charting data into the framework matrix; (7) interpreting the data.11 This analytical framework is widely used in critical care research and allows structured and systematic analysis of qualitative data.15 16 Three researchers (JMcP, LB, JMcC) independently undertook preliminary sweeps of the data. Key quotes to support the findings were then independently extracted by JMcP and JMcC. Member checking with a small number of interview participants was undertaken pre-analysis and post-analysis to enhance rigour of the reported results. Member checking, also known as participant or respondent validation, is a technique for exploring the credibility of results. Data or results are returned to participants to check for accuracy and resonance with their experiences.17 Participants who had agreed to ongoing contact with the research team were involved in the member checking process. The Consolidated Reporting of Qualitative Research checklist was used for this study.18

Patient and public involvement

Patients and caregivers who had previously been admitted to intensive care helped create the interview schedules used for this study. These representatives helped develop question content and structure via iterative discussion with the research team. These individuals had been part of ICU follow-up services previously; they were known to staff involved in the research and had given permission to be contacted about involvement such as this. They also supported the planning of the study conduct and reviewed all study documentation.


Across 14 sites, interviews were completed with 66 patients and 20 caregivers from Australia, the USA and the UK (table 1). Interviews took place between July 2018 and February 2019. We analysed the data across four main domains of SDoH: money and work; education and skills; housing, transportation and neighbourhoods; and family, friends and social connections. We have presented each SDoH as an individual category; however, how this particular category of SDoH influenced outcomes and indeed other SDoH is also explored. Supporting quotes, illustrating these concepts, can be found in table 2.

Table 2

Modification of the SDoH at the individual level and potential effects of such modifications on recovery from critical illness

Money and work

Loss of both money and work following critical illness was discussed frequently across the interviews. For example, a patient from Australia described how their new critical illness disability had changed their employment status:

I come home and now you’ve lost your job and you can’t work and then the realisation that you are on oxygen for the rest of your life.

Job loss and change had a direct negative impact on income, access to care and subsequent recovery from critical illness. For example, one patient experienced loss of not only insurance but also savings during critical illness and the recovery period following forced unemployment:

For the first surgery I had two insurances… second surgery, my school, they were nice enough to keep me on there, but then I lost it because I wasn’t working…. We had a little savings, but that’s all gone, because we have had to use it for medical bills and driving to the hospital an hour and a half.

There were health-related consequences of the changes. For example, participants described how they could no longer afford the treatments necessary for recovery:

My insurance company is messing with me right now… I’m out of my medicine. I’m out of one of them. I called to see how much it was and I can’t afford that. They want $55 for seven days. I can’t afford that.

Participants discussed how these changes to money and work were a direct mechanism for mental health problems. One participant from the USA spoke about how changes to her health insurance had increased feelings of anxiety:

There was an issue with the insurance….the first time said they weren’t going to cover any of our hospital bills. You can imagine…‘We’re going to lose our house, oh my God…’ The anxiety of it all.

A participant from the UK discussed changes to her employment situation following critical illness and subsequent emotional disruption:

I’ve changed jobs, I was a teacher before, but now I’ve gone into office work and I’m still trying to adjust. My GP [general practitioner] wrote a letter saying I wasn’t fit to go back to teaching because of the kind of asthma I’ve got…At the time I was devastated, I was really, really devastated.

Caregivers suffered similar loss of money and work, although via different mechanisms. One participant (patient) spoke about how their partner had lost their job due to new/increased caregiving responsibilities following hospital discharge:

I was working when I went in, and so was she and she had to take time off from work, and they had let her go from work, after she wasn’t making her units, and it made it rough on us.

Skills and education

The increased disability caused by critical illness resulted in an inability to return to previous roles and activities. One participant, who had a highly skilled role, described the impact of cognitive disability on function:

My brain’s just not making the connection…I’m retired but I was a judge and for a long time I wouldn’t go back to court…. I can tell my brain is not making the connection.

Similarly, a participant who had been in a skilled academic role could no longer return to their previous appointment, because of critical illness-caused cognitive decline:

My identity was a writer and professor had been built around being smart and so those issues manifested themselves as word finding, executive functioning…I couldn’t go back to work though. I was having cognitive difficulties…my request to have support for my particular cognitive deficits has been denied.

Such loss of skills required survivors to re-evaluate future career and employment prospects. One younger participant spoke of having to change educational pathways:

It’s changed my career path. I’m only 20…I just turned 21. And so I’m still really young. I’m in college. I changed my major because of it.

Housing, transportation and neighbourhoods

New critical illness-related disability significantly disrupted housing, transportation and residency in those survivors interviewed. Patients needed to move house due to physical inability:

It took, again, a lot of coordination and a lot of….Everyone pitched in…but it did provide a lot of stress, just in managing food and managing cleaning, and all the little things that you have to do from day to day. One of the reasons we began looking to move…having such a large house was a big struggle for her. Even going up and down stairs.

A participant from Australia spoke about the challenges to find funding for the housing adaptations which were required following critical illness and the financial and emotional stress this caused:

When he came home whether the house had to be reconfigured for a wheelchair or anything like that…the financial circumstances that’s another challenge… he was a breadwinner and now he’s not and who do we rely, what do we do?

The burden of transportation rose for survivors, who now needed to attend more medical appointments. Yet, many critical illness survivors could no longer drive, so this ongoing health burden caused disruption for the entire family unit:

It wasn’t necessarily one doctor’s appointment that stood out, but it was the fact that there were so many of them, and I think it was so significant because we had to drive an hour each way to get there.

These changes influenced recovery for many, especially for those living in rural areas, as accessing appropriate and reliable healthcare became harder:

I mean around here you’re not going to find any medical help that’s going to be decent. We’re just so far removed from everything….quite a disadvantage of being so far away from everything.

Changes to how participants could access transport also had a direct impact on recovery. For example, one participant spoke about how their inability to drive led to feelings of isolation:

I couldn’t drive for quite a long time, so I felt fairly dependent, I felt pretty trapped in.

Family, friends and social connections

Critical illness changed survivors’ social networks and relationships within families. For example, participants spoke about the negative impact of physical and emotional changes on wider social networks:

And so our community sort of vanished… I still have these huge scars and I’m not who I was before. I think for most people, they couldn’t deal with it.

Critical illness also led to fractures in family networks. In some cases, families struggled to manage the enormity of the situation, which led to challenges during recovery. For example, one survivor from the UK highlighted the impact which it had on the family unit:

I was doing something in the kitchen, and I couldn’t do it and I ended up smashing stuff all over the kitchen and my brother came in and I started shouting at him, saying you know this is my life?

These changes had consequences during recovery. Participants described isolation, challenges re-engaging with activities of daily living and mental health problems related to these changes. A participant from the USA described how these changes had impacted their mental health and behaviour:

No one talked to me about how I might be when I get home, like emotionally…I react to things… I feel bad ’cause its like hell for my family. I have these… I can’t control them…just absolute fits of anger and rage… and just crying.


This international, multicentre study suggests a complex interplay between SDoH and recovery from critical illness. It is already well established that upstream policy and contextual factors, as well as individual hardship, are associated with worsened onset of critical illness and outcomes; the early months of the COVID-19 epidemic particularly highlighted this.19 These qualitative results demonstrate how critical illness precipitated adverse changes in the recovery environment. These changes and other behaviour changes resulted in lost connections with family and social support, loss of the instrumental, social, and psychological benefits of work, and having numerous other practical difficulties. Together these changes impeded successful access to and benefit from even traditional health services. Our findings provide evidence that the relationship between social determinants and recovery plays out across multiple domains of SDoH and recovery, as well as bidirectionally. Further, they demonstrate that the ‘social determinants of recovery’ are not fixed nor the same as patients’ pre-illness statuses, suggesting a need to assess these mechanisms, and their impact, across time.

These findings can be situated in a broader body of work that suggests their generalisability. Inadequacies of the US health and social safety net are well described, but it is notable that the data for this study also included examples from Australia and the UK. Loss of employment and/or financial hardship after acute myocardial infarction, acute respiratory distress syndrome or traumatic injury are not uncommon.6 20 21 Studies have likewise found many suffer from significant financial costs and related material hardship due to chronic and ongoing illness, such as cancer22 and heart disease.23

Some, but not all, of the difficulties here seem to be a failure of the insurance functions of existing organisational arrangements to buffer patients and their families against purely financial shocks of critical illness. This interpretation is reinforced by findings that specifically financial stress is central after acute respiratory failure,5 as well as work on surprise billing24 and risk for high bills after surgery,25 or decades-old findings about patients’ self-management of diabetes in the presence of cost concerns.26 Additionally, substantial evidence supports our findings that critical illness leads to other financial hardships through job loss, unpaid family caregiving and new non-medical expenses due to new disability.4 7 27 Moreover, recent data from the UK suggest that patients who have ongoing disability due to critical illness are more likely to require government-funded welfare support in the years following discharge, in comparison with contemporary hospital controls28; concerns are emerging about patients with COVID-19.29 30

Efforts to address these resulting financial hardships through direct payments to patients and caregivers have shown promise; the Medicaid Cash and Counseling Programme found improved patient and caregiver well-being alongside improved health outcomes.31 32 Similarly, the US Department of Veteran Affairs’ Program of Comprehensive Assistance to Family Caregivers suggests financial support provides crucial assistance in varied ways, depending on specific needs.33 In the UK, efforts to include social welfare consultation as part of critical illness recovery programmes have also shown promise.11 These issues may become more prominent as moves to telehealth and remote monitoring require patients to bring more of their own technology to fully access services, potentially exacerbating inequities.34 35

Other challenges do not appear to be purely financial and would not be remediated by even theoretically complete insurance against total healthcare costs. Particularly prominent are the impacts of critical illness on social isolation. Social isolation is not benign; the influence of social relationships and social isolation on mortality is comparable with smoking, obesity and alcohol.36 Initiatives across the UK and the USA have successfully introduced innovation to support social isolation. For example, in Chicago, one health system added a social connection question to a pre-existing health screening tool, alongside care pathways such a friendly caller initiative to promote community socialisation during the COVID-19 pandemic.37 Peer support programmes which link individuals who have had similar healthcare experiences may also be advantageous, with evidence suggesting peer support could be a mechanism for reducing social isolation and improving global mental health.16

These data in the context of emerging literature have implications for clinicians and health systems seeking to promote full recovery of critically ill and other patients by addressing SDoH.38 39 First, the data suggest that assessments of risk might include not just current hardship, but the extent to which the patients’ reserves (broadly construed) are sufficient to prevent future hardship. It is unclear to what extent currently suggested patient-level risk assessments of SDoH40 41 are capable of predicting changes of the type these patients described. Second, given the myriad ways in which SDoH impacted each other following critical illness, understanding the impact of post-illness interventions (or lack thereof) must be also scrutinised across multiple domains, understanding that positive and negative consequences may be different for each patient. Certainly, targeted interventions to assure access to follow-up care, such as transportation support (eg, Uber or Lyft vouchers) or telehealth support (eg, device provision and training) can address some impacts. However, the multifaceted and interactive results of our study also suggest that broader programmes of direct assistance after critical illness, which address financial and non-financial resources, should be explored and potentially advocated for.


Strengths of this study include its international, multicentre approach to understanding the interplay between SDoH and recovery from critical illness. However, although the sample size is considerable for a qualitative study, we recognise that we have used a convenience sample, with a small portion of all critical care survivors in these countries. The sample size was not uniform across the three countries involved, which may have also influenced the reported results. We also do not have detailed in-hospital information for each patient.

We acknowledge that the primary aim of these interviews was not to delineate the socioeconomic problems which participants faced during recovery from critical illness. As such, important concepts may have been missed in this analysis. For example, we did not find cross-country differences; this is an important construct which could have been missed. Finally, we have used a broad definition of the term ‘rural’ in this analysis to ensure that it is applicable internationally. However, there are international variances in rural and urban interfaces, as such we may be under-reporting or over-reporting this as an issue.


In conclusion, this international, multicentre study has explored how critical illness changes social circumstances, impacting recovery and health. Our findings suggest that increased attention to issues such as one’s social network, economic security, education and skills, access to healthcare and living environment is required following critical care discharge. Targeting interventions toward these domains, including specific emphasis on social support and education, could potentially improve outcomes.

Data availability statement

Data are available upon reasonable request. The dataset used is available on reasonable request from the corresponding author.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants. The study design and protocol were approved by the Western Health Low Risk Human Research Ethics Panel (Australia, ref: HREC/17/WH/170); the Vanderbilt University Institutional Review Board (US coordinating site, ref: 171299); and the Southwest (Cornwall and Plymouth, ref: 18/SW/0137) Research Ethics Committee (UK). All participants provided informed consent.


The authors would like to acknowledge the funders for their support.


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.


  • Twitter @jomcpeake22, @lizzyhib, @katiehauschildt, @dwadepsych

  • Contributors Conception and design—JMcP, CS, KH, TI and MEM. Data extraction and primary analysis—JMcP, KH, CS, TI and LB. Analysis and interpretation—LB, JMcP and TI. Guarantor- JMcP. Reporting, drafting and revising the manuscript for important intellectual content—JMcP, CS, KH, TI, MEM, LB, EH, KH, RB, AB, BB, TE, WH, AH, JJ, AJ, JK, KK, JMcC, JM, AM-Y, TQ, AS, DW, MS, GN and ROH.

  • Funding KH, JMcP, LB, CS and TQ received funding from SCCM to undertake this work (no award number). LB is funded by NIH/NHLBI (K12 HL137943) so as is TI (K12 HL138039). JMcP is funded by a THIS.Institute (University of Cambridge) Fellowship (PD-2019-02-16/307748-01).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • 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.