Article Text

Download PDFPDF
Original research
Beyond borders: cystic fibrosis survival between Australia, Canada, France and New Zealand
  1. Adèle Coriati1,
  2. Xiayi Ma1,
  3. Jenna Sykes1,
  4. Sanja Stanojevic2,
  5. Rasa Ruseckaite3,
  6. Lydie Lemonnier4,
  7. Clémence Dehillotte4,
  8. Jan Tate5,
  9. Catherine Ann Byrnes6,7,
  10. Scott C Bell8,9,
  11. Pierre Regis Burgel10,11,12,
  12. Anne L Stephenson1,13,14
  1. 1Department of Respirology, St Michael's Hospital, Toronto, Ontario, Canada
  2. 2Community Health and Epidemiology, Dalhousie University, Halifax, Nova Scotia, Canada
  3. 3Public Health and Preventative Medicine, Monash University, Clayton, Victoria, Australia
  4. 4Association Vaincre la Mucoviscidose, Paris, Île-de-France, France
  5. 5Paediatric Department, Starship Children's Health, Auckland, New Zealand
  6. 6Paediatrics, Child and Youth Health, The University of Auckland School of Medicine, Auckland, New Zealand
  7. 7Paediatric Respiratory Department, Starship Children's Health, Auckland, New Zealand
  8. 8Department of Thoracic Medicine, The Prince Charles Hospital, Chermside, Queensland, Australia
  9. 9Children’s Health Research Centre, The University of Queensland Faculty of Medicine and Biomedical Sciences, Herston, Queensland, Australia
  10. 10Assistance Publique Hôpitaux de Paris, Department of Respiratory Medicine and French Cystic Fibrosis Reference Center, Hopital Cochin Pneumologie, Paris, Île-de-France, France
  11. 11Institut Cochin, Université de Paris, Paris, Île-de-France, France
  12. 12European Reference Network Respiratory Diseases, Frankfurt, Germany
  13. 13St Michael's Hospital Li Ka Shing Knowledge Institute, Toronto, Ontario, Canada
  14. 14Department of Medicine, University of Toronto, Toronto, Ontario, Canada
  1. Correspondence to Dr Anne L Stephenson, Dept. of Respirology, St Michael's Hospital, Toronto, Canada; Anne.Stephenson{at}


Background Life expectancy for people with cystic fibrosis (CF) varies considerably both within and between countries. The objective of this study was to compare survival among countries with single-payer healthcare systems while accounting for markers of disease severity.

Methods This cohort study used data from established national CF registries in Australia, Canada, France and New Zealand from 2015 to 2019. Median age of survival for each of the four countries was estimated using the Kaplan-Meier method. A Cox proportional hazards model was used to compare risk of death between Canada, France and Australia after adjusting for prognostic factors. Due to low number of deaths, New Zealand was not included in final adjusted models.

Results Between 2015 and 2019, a total of 14 842 people (3537 Australia, 4434 Canada, 6411 France and 460 New Zealand) were included. The median age of survival was highest in France 65.9 years (95% CI: 59.8 to 76.0) versus 53.3 years (95% CI: 48.9 to 59.8) for Australia, 55.4 years (95% CI: 51.3 to 59.2) for Canada and 54.8 years (95% CI: 40.7 to not available) for New Zealand. After adjusting for individual-level factors, the risk of death was significantly higher in Canada (HR 1.85, 95% CI: 1.48 to 2.32; p<0.001) and Australia (HR 2.08, 95% CI: 1.64 to 2.64; p<0.001) versus France.

Interpretation We observed significantly higher survival in France compared with countries with single-payer healthcare systems. The median age of survival in France exceeded 60 years of age despite having the highest proportion of underweight patients which may be due to differences in availability of transplant.

  • cystic fibrosis

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.

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.


  • International comparisons have the potential to yield important results, as distinct differences, as well as similarities, in cystic fibrosis care delivery exist in different countries.


  • After accounting for the variation in patient characteristics, cystic fibrosis survival was the highest in France compared with Canada, Australia and New Zealand.


  • Disparities in outcomes between countries could be explained by modifiable factors such as access to transplant.


Between country comparisons of morbidity and mortality can be useful to inform policies at a national level. Prior research has shown large differences in prognosis for those living with cystic fibrosis (CF) when comparing high-income countries with low/middle-income countries as well as comparisons between high-income nations.1–3 Although many countries with established CF registries produce annual registry reports, direct comparisons are often inappropriate or not possible due to variation in definitions or differing categorisation of key metrics used to monitor health. Even with outcomes that have an unequivocal definition, such as mortality, direct comparisons can be challenging due to variations in data processing, missing death data or a lack of clarity around those included or excluded from the survival analyses.4 Furthermore, when comparing survival metrics, many national registry reports lack median age of survival estimates making survival comparisons impossible.5

International comparisons of CF populations can ultimately change practice—a survival comparison between Toronto and Boston discovered a higher age of survival in Toronto and led to the recommendation of the high fat diet in the 1980s6. More recently, Stephenson et al confirmed a 10-year survival advantage for Canadians living with CF compared with similar people in the USA after accounting for prognostic factors.3 The same group found that transplantation and post-transplant survival explained 30% of the survival gap between the countries highlighting a potentially modifiable factor that could boost survival for individuals living with CF in the USA.7 Transplantation, however, does not fully explain the Canada–USA survival gap and another distinct difference between the countries relates to the model of healthcare delivery as Canadians have access to a universal, single-payer healthcare system whereas Americans use a multipayer healthcare scheme which could limit access to medical care depending on insurance coverage and financial ability. The Canada–USA survival study found that individuals in the USA with no insurance or on medicare/medicaid were more likely to die than people in the USA with private insurance. Furthermore, there was no statistically significant difference in risk of death between those with private insurance in the USA and Canadians.3 Comparing health metrics between countries with similar healthcare models could minimise this as a contributing factor.

The main objective of our study was to directly compare demographic and clinical characteristics as well as survival outcomes in a contemporary cohort of individuals living with CF in countries with single-payer healthcare systems namely, Australia, Canada, France and New Zealand (NZ). We hypothesise that there will be no statistically significant difference in survival between the countries.

Study design and methods


This population-based cohort study used four established, national longitudinal CF registries: Australian CF registry (ACFDR), Canadian CF registry (CCFR), French CF registry (FCFR) and NZ CF registry (NZCFR). Individuals captured in each of the registries provide informed consent to have their data collected and used for research purposes.

Study period

A contemporary period between 1 January 2015 and 31 December 2019 was used.

Data resources

The ACFDR is managed by Monash University and has been in existence since 1998. The registry captures data on more than 4500 individuals followed in 23 CF centres which is reported to be at least 90% of the CF population. The CCFR, managed by CF Canada, has been in existence since the early 1970s and is estimated to capture 98% of the CF population with a low rate of lost-to-follow-up (~5%). The registry currently contains annual records on more than 6700 unique individuals with CF who are followed in the 42 Canadian CF clinics. The FCFR is managed by the Vaincre la Mucoviscidose. Since its inception in 1992, the registry has data on more than 9450 individuals followed in 47 CF centres, capturing at least 95% of the French CF population. The NZCFR is managed by CF NZ and follows more than 540 individuals. Established in 2011, the NZCFR registry captures data from 19 NZ clinics. It is reported that about 98% of New Zealanders with CF are followed in the NZCFR. Registry data undergo routine validation checks to ensure that they are free of duplicates and errors.5 Only individuals with a diagnosis of CF were included in this analysis while individuals with cystic fibrosis transmembrane conductance regulator (CFTR)-related disease or cystic fibrosis screen positive, inconclusive diagnosis were excluded.

Demographic and clinical data are recorded annually. Demographic data included date of birth, sex, age at diagnosis, newborn screening status (NBS), CFTR variant, pancreatic status and date of death. Clinical measurements include lung function, nutritional status, sputum bacteriology, number of hospitalisations or home intravenous antibiotic therapy, date of transplant as well as organ transplanted.

Variable definitions

Variable definitions and method of data collection in the four registries were evaluated by the research team to create a harmonised dataset for analysis. Date of CF diagnosis was imputed using date of birth plus 30 days when missing and categorised as age at diagnosis as <2 and ≥2 years of age.8 CF genotype was classified as Phe508del homozygous, Phe508del heterozygous, other or missing. Lung function, specifically the forced expiratory volume in 1 s (FEV1), was recorded in litres and interpreted as percent predicted (ppFEV1) calculated using Global Lung Function Initiative reference equations.9 The best annual ppFEV1 was selected for each individual each year. Body mass index (BMI) was calculated for adults over 19 years of age as weight (kg)/height2 (m). BMI percentiles were calculated for individuals ages 2–19 years of age by using the Centre for Disease Control growth charts.10 Annual BMI or BMI centiles were calculated using data on the closest date within 3 months of the best lung function measurement. Individuals were categorised as underweight (BMI <19 kg/m2 or BMI %ile ≤12%), adequate weight (BMI between ≥19 kg/m2 and ≤24.9 kg/m2 or BMI %ile between >12% and <85%) or overweight (BMI >24.9 kg/m2 or BMI %ile ≥85%).10 11 Severe pulmonary exacerbations were defined as the administration of intravenous antibiotics in the hospital and/or at home and were recorded as the number of events per year. Pancreatic insufficiency (PI) was determined by pancreatic enzyme use recorded in the registries. The presence of Burkholderia cepacia (B. cepacia) complex or Pseudomonas aeruginosa (P. aeruginosa) was considered positive if these bacteria were recorded at least once each year12.

Statistical analysis

Median and IQR were used to summarise continuous variables whereas frequency and proportion were used to summarise categorical variables. A period approach was used to estimate median age of survival over a 5-year window from 2015 to 2019. The Kaplan-Meier (KM) method was used to calculate the probability of survival in each of the four countries. KM survival curves were generated for the following subgroups: PI, pancreatic sufficiency and those who were homozygous deltaF508. Observations were left truncated on 1 January 2015 or the date of diagnosis if the diagnosis occurred within the time window. The follow-up period ended on 31 December of the last observation recorded or death date.

Differences in survival between the countries were further explored by adjusting for demographic and clinical characteristics using a multivariable Cox proportional hazards (PH) model. In New Zealand, only 21 deaths were recorded, and thus it was not possible to include this country in the multivariable analysis. The multivariable model, which included both transplanted and non-transplanted individuals, was adjusted for time-independent characteristics specifically sex, age at diagnosis, NBS, pancreatic status, transplant status, baseline ppFEV1 and baseline BMI. ‘Baseline’ was defined as the last available measurement for those who were not transplanted, or before death or the end of the follow-up period. For those who were transplanted, baseline is defined as before transplant if transplant occurred in 2013 or later, otherwise baseline pre-transplant data were imputed using multiple imputation for transplanted recipients. Individuals <6 years of age were excluded as this age group does not reliably perform lung function testing. Missing data were imputed using multiple imputation with chained equations using the R package mice (V.3.7.0). All variables except for post-transplant specific variables were included in the imputation model for predicting the missing values. Ten imputed datasets were created and the results were pooled as outlined in Rubin.13 The PHs assumption was checked by testing the independence between the Schoenfeld residuals and time variable and the assumption of linearity was checked by plotting the Martingale residuals. All statistical analysis was done using the R software (V.3.6.2).14 All p values are two-sided and assessed for significance at p<0.05 unless otherwise stated.


Comparison of study populations between countries

The cohort flow diagram for individuals included in the analysis can be seen in figure 1. Demographic and clinical and characteristics of each CF population are reported in table 1. Between 2015 and 2019, there were a total of 14 842 subjects included in the analysis: 3537 individuals with CF included from Australia, 4434 in Canada, 6411 in France and 460 in New Zealand. The percentage of individuals who were diagnosed through newborn screening was the lowest in Canada (13.8%) compared with the other countries. France had the lowest proportion of individuals who were homozygous for Phe508del (42%) and the highest percentage of people with ‘other’ mutations (13.9%). The proportion of individuals who were underweight was the highest in France (18.5%) compared with the other countries whereas the median ppFEV1 was lowest in Canada.

Figure 1

Cohort creation 2015–2019. KM, Kaplan-Meier.

Table 1

Characteristics of individuals in Canada, France, Australia and New Zealand at the most recent measurement within the study window (2015–2019; n=14 842)*

Transplanted patients were included in the study window but the transplant itself may have occurred prior to or during the study period of 2015–2019. Of the individuals followed in the study period, 413 (11.7%) had received a lung transplant in Australia compared with 589 (13.3%) in Canada, 1061 (16.5%) in France and 31 (6.7%) in New Zealand (table 2). Age at transplant and age at death were the lowest in France. The proportion of deaths without transplant was higher than the proportion of death post-transplant for all countries except for France where the opposite was observed (table 2).

Table 2

Survival, deaths and transplants in Canada, France, Australia and New Zealand (2015–2019; n=14 842)*

Survival comparisons

The KM survival curves for each country and by subgroups are shown in figure 2 and online supplemental figure S1. The median age of survival was highest in France (65.9 years; 95% CI: 59.8 to 76.0 years) compared with Australia (53.3 years; 95% CI: 49.7 to 59.8), Canada (55.4 years; 95% CI: 51.3 to 59.2) and New Zealand (54.8 years; 95% CI: 40.7 to NA) (figure 2, table 2).

Supplemental material

Figure 2

Overall survival probability in cystic fibrosis between Canada, France, Australia and New Zealand (2015–2019; n=14 842). AUS, Australia; CAN, Canada; FRA, France; NZ, New Zealand.

The Cox proportional modelling analysis excluded New Zealand therefore, included a total of 14 382 subjects with 178 deaths in Australia, 257 deaths in Canada and 241 deaths in France. The univariable analysis showed the risk of death was higher in Canada (HR 1.35, 95% CI: 1.13 to 1.62; p<0.001) and Australia (HR 1.45, 95% CI: 1.2 to 1.76; p<0.001) compared with France. After adjusting for sex, age at diagnosis, NBS, pancreatic status and lung transplant status, baseline ppFEV1 and BMI category, the risk of death was significantly higher in Canada (HR 1.79, 95% CI: 1.44 to 2.22; p<0.001) and Australia (HR 2.04, 95% CI: 1.61 to 2.59; p<0.001) compared with France. There was no evidence of a difference in survival between Canada and Australia (HR 0.88, 95% CI: 0.68 to 1.13; p=0.27). (online supplemental figure S2)


Using contemporary CF data, our study identified a survival gap between people living in high-income countries with similar healthcare systems. France had the highest recorded median age of survival of the four countries, surpassing 60 years in the contemporary time window between 2015 and 2019. Although there were some notable differences in individual characteristics between the countries, after adjusting for these factors, the risk of death remained higher in Canada and Australia compared with France despite the countries having national healthcare systems. The exceptional survival in France is despite having the highest proportion of underweight individuals which has consistently been found in the literature to be strong predictor of worse outcomes in CF. Furthermore, these data were collected prior to the availability of elexacaftor–tezacaftor–ivacaftor which has been shown to improve lung function and is predicted to increase future survival estimates.15

Through international comparisons, differences in health outcomes raise questions about the reasons for such disparities. It is unlikely that a single factor can explain why one country has better survival than another country. Although universal, government-funded healthcare allows broad access to medical care, this does not necessarily translate into full access to chronic medications known to improve health and there are distinct differences in this regard between the countries studied. France has an extensive, government-funded medication programme that minimises barriers to accessing CF medications. In Canada, drug coverage plans vary between provinces with some provinces having more extensive coverage of CF medications while others are more restrictive. Canadians with private drug insurance through employment may have coverage for a broad list of medications at no additional cost to them. While others without drug insurance would have to pay out-of-pocket to access certain CF medications thus limiting access. Access to medication in Australia and New Zealand is, in essence, a hybrid between Canada and France with the majority of medications being covered by healthcare funders with a nominal cost to patients. Such differential access to medications could explain differences in health outcomes as therapies have been linked to improved lung function and reduced exacerbations, both of which are closely associated with survival.16 17 Unfortunately, we did not have complete medication data for the study period across all registries that would allow a robust comparison to answer this question, but this would be an important line of investigation for future research.

Given that the majority of deaths in CF are due to progressive lung disease, lung transplantation can dramatically and quickly alter the trajectory of the disease. Transplantation explained 30% of the survival gap between Canada and the USA, again highlighting the importance of this potentially modifiable therapeutic option as a means to increase survival for individuals with CF.7 Our data showed that France had the highest proportion of transplant recipients in the study window. Furthermore, France was the only country that had a lower proportion of deaths without transplant of all the countries compared with deaths that occurred after the opportunity for transplant. These data suggest that a significant proportion of individuals in need of transplant in France were able to receive this life-saving therapy. Interestingly, in 2007 France implemented a high emergency lung transplant (HELT) programme at a national level by the French Agence de la Biomedicine. When a patient corresponds to the criteria of HELT, their file is submitted to the Agency and examined by a panel of experts. If the experts confirm the eligibility to HELT, the patient is given national priority for 7 days; it can be extended by 7 days (total time 14 days) if no organ transplant was made available during the initial 7 days.18 Since this was implemented, the number of individuals dying on the transplant list in France dramatically decreased resulting in more transplants for those who needed them. This, in turn, decreased the number of individuals who died without transplant while the number of transplant recipients increased.19 Further, although patients transplanted from the HELT programme were critically ill, there was no negative impact on overall post-transplant survival compared with the survival prior to the implementation of the program.19 The HELT programme is unique to France and although medical need influences prioritisation on the transplant list in Canada, Australia and New Zealand, the process is less formalised when compared with the system adopted in France.

Unlike transplantation, the impact of NBS on survival estimates would not be evident for several decades until babies screened at birth reached an age where they would be at risk for death. NBS was introduced at different times depending on the country: France (~1989 for two regions, nationally in 2002), Australia (1981 in New South Wales, nationally in 2001) and New Zealand (~1983) versus Canada (~2007 in most provinces, nationally in 2018)20–23 and this is reflected in the proportion of individuals diagnosed through NBS by country. Furthermore, differential algorithms for NBS exist so it is possible that milder mutations may be identified in some countries compared with others which may impact survival estimates. However, when mutations were classified by functional class, the proportions were similar between the countries. Given paediatric mortality is not very prevalent, and most deaths occurred in adults, it is unlikely that NBS already had an effect of survival. This effect, if it exists, will need more time to be observed.

There is some evidence, although not consistently shown, that higher healthcare spending as determined by the percentage of gross domestic product (GDP), is associated with better health outcomes.24 25 According to 2019 statistics, France spent 11.1% of GDP on healthcare compared with 10.8% for Canada, 9.8% for Australia and 9.1% for New Zealand. Such differences are small and are unlikely to explain such a large gap in survival as seen in our study. Furthermore, spending more on healthcare does not guarantee better outcomes. The USA spent the highest percentage of GDP on healthcare (~17%) in 2019, almost twice that of Canada, yet prior literature showed that Canadians with CF live on average 10 years longer than their US counterparts again suggesting that the reasons for international differences in survival are complex and multifactorial.

Our study has several strengths and limitations to consider. The detailed clinical information available in national registries allowed a methodologically rigorous approach to addressing international differences regarding survival and health outcomes in CF. A harmonised approach to variable definitions within the registries ensured appropriate and robust comparisons to allow adjustment for important confounding variables. The registries capture a national picture of individuals with CF with few individuals lost to follow-up. It is possible that individuals with CFTR-related disease were included within the registries but not clearly identified; however, this would represent a small number and would not be unique to any single country therefore minimising any potential bias. Certain data elements were either unavailable in all registries (ie, not captured at all) or were incompletely recorded specifically medication, microbiology and socioeconomic status thus we were unable to account for these factors in our analyses. Future analyses may take into consideration such factors since data collection in registries has expanded with enhanced granularity in recent years. As with all registries, there is a degree of missing data; however, where possible, original source documents were obtained to fill in the gaps and multiple imputation was used to inclusion of all available subjects in the multivariable model. Furthermore, our results were unchanged when we conducted a complete case analysis compared with those where we used multiple imputation. Finally, the variable availability of earlier CFTR modulator therapies (ie, lumacaftor–ivacaftor, tezacaftor–ivacaftor) coupled with inconsistent data capture of these elements across all four registries made it impossible to accurately summarise the data across the study period. Although ivacaftor was used in Canada, Australia and France for the study period, the proportion of patients on this drug was low (ranging from 2.6% in France to 4.6% in Australia) and unlikely to explain the survival gap identified. Furthermore, these analyses were conducted prior to the widespread availability of elexacaftor–tezacaftor–ivacaftor. In the future, differential access to CFTR modulators may alter the trajectory of disease and further exacerbate differences in health outcomes in CF. CF communities worldwide need to continue their advocacy efforts to maximise access to all therapies to reduce health disparities such as those identified in this study for individuals living with this disease. For multiple reasons, which will need to be further explored, CF care results in better prognosis in France as compared with Canada and Australia.


In conclusion, our study has shown significantly higher survival in France compared with countries with comparable healthcare systems. France survival estimate exceeded 60 years of age which may be due to differences in availability of transplant for those who need it most.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants. The research ethics board approval for this study was obtained from Unity Health Toronto (REB #17-312) and approval for use of the proposed registry data was granted by Monash University for Australia, Cystic Fibrosis Canada, Vaincre la Mucoviscidose for France, and Cystic Fibrosis New Zealand. Participants gave informed consent to participate in the study before taking part.


We would like to acknowledge and thank all of the patients with CF and families in Canada, Australia, France, New Zealand who consent to be part of their respective national CF patient registries as well as the CF clinic staff who spend many hours inputting the data. Without their efforts, this study would not be possible.


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 @CysFibMD

  • Contributors ALS, SS, PRB, SCB contributed to study concept and design. ALS, RR, LL, CD, JT, CAB contributed to acquisition of data. ALS, SS. PRB, AC, XM, JS, RR, LL, CD, JT, CAB, SCB contributed to analysis and interpretation of data. AC, ALS contributed to drafting of the manuscript. ALS, SS. PRB, AC, XM, JS, RR, LL, CD, JT, CAB, SCB contributed to critical revision of the manuscript for important intellectual content. XM, JS, SS, AC contributed to statistical analysis. ALS, SS, PRB, SCB obtained funding. ALS contributed to study supervision. ALS is the guarantor of the study, therefore accepts full responsibility for the work and/or the conduct of the study, had access to the data, and controlled the decision to publish.

  • Funding This study was funded by Cystic Fibrosis Canada (Grant #609767). AC was supported by a Post-Doctoral Fellowship from the Canadian Lung Association.

  • Disclaimer The study sponsor did not participate in the design and conduct of the study; collection, management, analysis and interpretation of the data; or preparation, review or approval of the manuscript; decision to submit the paper for publication.

  • Competing interests SS is a co-applicant of the CF Canada grant for this study, and received a grant from Vertex pharmaceuticals and the European Respiratory Society, unrelated to this study. SS also received consulting fees from Chiesi pharmaceuticals, unrelated to this study. CAB received support from Cystic Fibrosis New Zealand (for this study) and grants from HRC, Flu Lab and APP, unrelated to this study. CAB was also a trustee for the Bronchiectasis Foundation and chair of the clinical advisory panel for cystic fibrosis as well as the PORT CF NZ registry. PRB received grants from GSK, Vertex and Boehringer Ingelheim, unrelated to this study. PRB has also received consulting fees from Astra Zeneca, Chiesi, Insmed and Vertex as well as payment from Astra Zeneca, Chiesi, GSK, Novartis, Pfizer, Vertex and Zambon, unrelated to this study. PRB also received support for attending meetings and/or travels from Astra Zeneca and Zambon, unrelated to this study. SCB is a co-applicant of the CF Canada grant for this study, in addition to receiving grants from the CF Foundation, NHMRC and MRFF, unrelated to this study. SCB was chair of the Data Safety monitoring Board (DSMB) of the Phase IIA/IIB RCT Ataxia telangiectasia project and also participated in the DSMB of the RCT Inhaled hypertonic saline in people with bronchiectasis project, unrelated to this study. SCB was also a board member of Health Translation Qld Board, Gallipoli Medical Research Foundation and the European CF Society, unrelated to this study. ALS has received a grant from CF Canada for this study, as well as the CF Foundation and the Canadian institute for Health research, unrelated to this study. ALS received consulting fees from CF Canada (or her role as the Medical Director of the Canadian CF Registry) and payment form Vertex Pharmaceuticals, unrelated to this study. ALS has also participated in the DSMB of Vertex pharmaceuticals and Horizon Pharma, unrelated to this study.

  • 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.