Article Text

Original research
Effect of the gut microbiome in glaucoma risk from the causal perspective
  1. Yaxuan Wu1,2,3,
  2. Ronghua Shi1,2,3,
  3. He Chen1,2,3,
  4. Zicheng Zhang1,2,3,
  5. Siqi Bao1,2,3,
  6. Jia Qu1,2,3,
  7. Meng Zhou2
  1. 1School of Biomedical Engineering, School of Information and Communication Engineering, Hainan University, Haikou, People's Republic of China
  2. 2National Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, People's Republic of China
  3. 3Hainan Institute of Real World Data, Qionghai, People's Republic of China
  1. Correspondence to Dr Meng Zhou; zhoumeng{at}wmu.edu.cn; Dr Jia Qu; qujia{at}eye.ac.cn

Abstract

Objective Evidence from observational studies has reported possible associations between the gut microbiome (GM) and glaucoma. However, the causal effect of GM on glaucoma risk remains to be determined.

Methods and analysis We conducted two-sample bidirectional Mendelian randomisation (MR) analyses to explore the causal association between GM and glaucoma. Genome-wide association study summary statistics of 196 GM taxa (n=18 340) and glaucoma (18 902 cases and 358 375 controls) were obtained from MiBioGen and FinnGen Consortium. Inverse variance weighted, MR-Egger, weighted median, weighted mode, Mendelian Randomisation Pleiotropy Residual Sum and Outlier, MR-Egger intercept and Cochran’s Q statistical analyses were used to supplement MR results and sensitivity analysis. An independent cohort from the Medical Research Council (MRC) Integrative Epidemiology Unit at the University of Bristol (MRC-IEU) Consortium (1715 cases and 359 479 controls) was used to validate causal effects.

Results Results of the MR analysis suggested that the family Oxalobacteraceae (OR 0.900, 95% CI 0.843 to 0.961, p=0.002) and the genus Eggerthella (OR 0.881, 95% CI 0.811 to 0.957, p=0.003) had a negative effect on glaucoma, whereas the genus Bilophila (OR 1.202, 95% CI 1.074 to 1.346, p=0.001), LachnospiraceaeUCG010 (OR 1.256, 95% CI 1.109 to 1.423, p=0.0003) and Ruminiclostridium 9 (OR 1.258, 95% CI 1.083 to 1.461, p=0.003) had a positive effect on glaucoma. Among these, the positive causal effect of LachnospiraceaeUCG010 (OR 1.002, 95% CI 1.000 to 1.004, p=0.033) on glaucoma was replicated in an independent cohort.

Conclusion This MR analysis from large population studies demonstrated the causal effect of GM on glaucoma risk and supported the role of GM in influencing glaucoma susceptibility.

  • glaucoma
  • microbiology

Data availability statement

Data are available in a public, open access repository. Data sets analysed in this study were publicly available from the IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/), MiBioGen Consortium (https://mibiogen.gcc.rug.nl/) and FinnGen Consortium (https://www.finngen.fi/en/access_results).

http://creativecommons.org/licenses/by-nc/4.0/

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: http://creativecommons.org/licenses/by-nc/4.0/.

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

  • The existence of the gut-retina axis has been established, indicating that alterations in the gut microbiome (GM) may affect retinal health.

  • Significant correlations have been found between GM dysbiosis and several eye diseases.

  • Observational studies have suggested a possible association between GM dysregulation and glaucoma.

WHAT THIS STUDY ADDS

  • Genetically elevated levels of the family Oxalobacteraceae and the genus Eggerthella are associated with a decreased risk of glaucoma.

  • Genetically elevated levels of the genus Bilophila, LachnospiraceaeUCG010 and genus Ruminiclostridium 9 are associated with an increased risk of glaucoma.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • This study has uncovered an initial potential association between GM and glaucoma.

  • The findings highlight modulation of the gut microbiome as a novel avenue for optic neuroprotective therapy in glaucoma.

Background

Glaucoma, a common sight-threatening eye disease, is characterised by progressive and irreversible damage to the optic nerve and retinal nerve fibre layer, resulting in significant visual impairment.1 Due to its insidious nature, symptoms often do not manifest until later stages of life, contributing to severe and permanent visual impairment.2 Although various systemic factors, such as obesity,3 diabetes4 and dyslipidaemia,5 have been associated with glaucoma, its underlying causes remain incompletely understood.

Recently, a gut-retina axis has been demonstrated, with observed changes in the gut microbiome (GM) having discernible effects on retinal health.6 7 A growing body of evidence highlights the significant link between GM dysbiosis and a variety of ocular diseases6 8–10 and neurodegenerative conditions.11 Notably, observational studies have shown variations in the bacterial composition of individuals with glaucoma, and intriguingly, transplantation of glaucoma-associated faecal microbiota into mice has been shown to activate microglial cells and induce retinal inflammation, suggesting an underlying gut-retina axis in disease pathogenesis.12 Another study demonstrated the significant role of gut-homing β7+ CD4+ T cells in retinal ganglion cell damage in glaucoma in both human and mouse models.13 14 This finding highlights the complex interplay of the gut-retina axis in the pathophysiology of glaucoma. While emerging evidence from some observational studies suggests an association between GM dysregulation and glaucoma, whether there is a causal relationship between GM composition and glaucoma risk remains to be established.

Mendelian randomisation (MR) is a robust epidemiological approach that uses genetic variation as an instrumental variable to explore potential causal relationships between a particular exposure of interest and an outcome of interest.15 Unlike randomised controlled trials, MR minimises biases in observational studies, including confounding factors and reverse causation.16 In this study, we perform a bidirectional MR analysis to investigate the potential causal effects of GM on glaucoma risk. We then validate the causal relationship using summary statistics from another large publicly available genome-wide association study (GWAS). Based on these findings, we aimed to elucidate the role of GM taxa in glaucoma development, uncover potential biological mechanisms, improve our understanding of the gut-retina axis and lead to innovative therapeutic approaches for the treatment of ocular diseases.

Methods

Study design and data sets

The two-sample MR method was used to assess the causal relationship between 196 GM taxa and glaucoma susceptibility. We also used a validation set to verify the previously obtained results. The design of the study is shown in figure 1.

Figure 1

Schematic representation of the study. (A) Instrumental variable (IV) assumption. (B) Schematic representation of the analytical workflow. GWAS, genome-wide association study; MR, Mendelian randomisation; SNP, single-nucleotide polymorphism.

Summary statistics for genetic associations related to GM were obtained from the largest GWAS summary data from the MiBioGen Consortium.17 The MiBioGen Consortium conducted a genome-wide meta-analysis of associations between autosomal genetic variants and the GM in 18 340 individuals from 24 cohorts. Ultimately, 211 taxa of the GM at five levels were identified. We removed unknown GM taxa to obtain more accurate results and finally included 196 taxonomic units (9 phyla, 16 classes, 20 orders, 32 families and 119 genera) in our analysis.

GWAS summary statistics for glaucoma in the primary analyses were derived from the FinnGen Consortium18 (https://www.finngen.fi/en/access_results, releases 9), involving 18 902 cases and 358 375 controls. Glaucoma was defined based on the end points of the International Classification of Diseases, Tenth Revision, and included primary open-angle glaucoma, primary closed-angle glaucoma and other glaucoma cases. Summary statistics of genetic associations for glaucoma in the validation phase (GWAS ID: ukb-d-H40) were sourced from the IEU Open GWAS project (https://gwas.mrcieu.ac.uk/datasets/ukb-d-H40/),19 comprising 1715 cases and 359 479 controls.

Instrument variables selection

We chose a locus-wide significance threshold (p<1×10−5)20 for assessing the association between single-nucleotide polymorphisms (SNPs) and the exposure variable in forward MR analysis. For reverse MR analysis, instrument variables (IVs) with strong correlation with exposure were selected based on a genome-wide significance threshold (p<5×10−8). Independent SNPs were obtained through linkage disequilibrium analysis based on European samples from the 1000 Genomes Project (r2<0.001 and clumping window size=10 000 kb).21 SNPs with strong correlations with confounding factors (such as myopia, obesity, intraocular pressure, etc) were removed using the phenoscanner package.22 Heterozygous SNPs were filtered out using the RadialMR package.23 The F-statistic Embedded Image17 was used to assess the strength of the SNPs, and we excluded those with F<10 to reduce the bias associated with weak instrumental variables. N represents the sample size, k denotes the number of instrumental variables and R2 represents the proportion of variance in the exposure variable explained by genetic variance. Finally, to prevent incorrect inference of causal direction, Steiger filtering was performed to eliminate SNPs with reverse causal effects.24

Statistical analysis

All statistical analyses were performed using R software (V.4.3.1), and related R packages, including TwoSampleMR (V.0.5.7), phenoscanner (V.1.0), RadialMR (V.1.1), gwasglue (V.0.0.0.9000), qvalue (V.2.15.0), devtools (V.2.4.5), MRPRESSO (V.1.0). Prior to MR analyses, we harmonised the direction of SNP alleles for exposure and outcome. We chose inverse variance weighted (IVW) as the primary analysis method for this study.25 Four other methods were used as supplements to balance the robustness of the pleiotropy assumption, including MR-Egger, weighted median, simple mode and weighted mode. We used three different statistical methods, including Cochran’s Q statistic,26 MR-Egger regression’s intercept test27 and Mendelian Randomisation Pleiotropy Residual Sum and Outlier (MR-PRESSO),28 to assess heterogeneity and horizontal pleiotropy in our MR results. We performed false discovery rate (FDR) correction29 using an FDR of a q value <0.1. Finally, leave-one-out analyses were conducted to confirm the stability and reliability of the results.

Results

Estimated potential causal effects of GM on glaucoma risk

According to the selection criteria for IVs, we obtained 1916 SNPs as IVs for 196 GM taxa. The number of IVs for each GM taxon ranged from 2 to 18, with F-statistics ranging from 20.190 to 23.817 (online supplemental files 1; 2). The results of IVW analysis revealed causal associations between specific GM taxa and glaucoma risk (figure 2). As shown in figure 3, the genetically enhanced family Oxalobacteraceae (OR 0.900, 95% CI 0.843 to 0.961, p=0.002, q=0.075) and the genus Eggerthella (OR 0.881, 95% CI 0.811 to 0.957, p=0.003, q=0.083) were correlated with a decreased risk of glaucoma, whereas the genetically increased genus Bilophila (OR 1.202, 95% CI 1.074 to 1.346, p=0.001, q=0.075), LachnospiraceaeUCG010 (OR 1.256, 95% CI 1.109 to 1.423, p=0.0003, q=0.049) and genus Ruminiclostridium 9 (OR 1.258, 95% CI 1.083 to 1.461, p=0.003, q=0.083) correlated with an increased risk of glaucoma. In addition, 10 GM taxa were associated with glaucoma risk. IVW analyses indicated that class Bacilli (OR 1.123, 95% CI 1.012 to 1.245, p=0.028, q=0.316), class Deltaproteobacteria (OR 1.129, 95% CI 1.004 to 1.270, p=0.042, q=0.415), genus RuminococcaceaeUCG009 (OR 1.114, 95% CI 1.023 to 1.212, p=0.013, q=0.312), genus Senegalimassilia (OR 1.177, 95% CI 1.016 to 1.364, p=0.030, q=0.316), genus Streptococcus (OR 1.143, 95% CI 1.015 to 1.287, p=0.028, q=0.316) and order Bacillales (OR 1.078, 95% CI 1.008 to 1.154, p=0.029, q=0.316) are associated with an increased risk of glaucoma. However, family Family XI (OR 0.919, 95% CI 0.854 to 0.988, p=0.022, q=0.316), genus Ruminococcus torques group (OR 0.781, 95% CI 0.635 to 0.960, p=0.019, q=0.316), genus Clostridium sensus tricto 1 (OR 0.781, 95% CI 0.626 to 0.973, p=0.028, q=0.316) and genus LachnospiraceaeUCG008 (OR 0.908, 95% CI 0.838 to 0.984, p=0.019, q=0.316) were associated with a reduced risk of glaucoma (figure 3).

Figure 2

Causal analyses of each gut microbial component taxon and glaucoma based on five Mendelian randomisation (MR) analyses (p<1×10−5) from the primary analysis stage (FinnGen Consortium). From the outside to the inside are the taxon name, p value of inverse variance weighted (IVW), p value of MR-Egger, p value of the weighted median (WM), p value of the simple mode (SM), p value of weighted mode (WMODE) and OR value based on IVW results.

Figure 3

Causality at the primary analysis stage (FinnGen Consortium). Forest plot shows significant causal associations. Five two-sample Mendelian randomisation (MR) methods (inverse variance weighted, MR-Egger, weighted median, simple mode and weighted mode) were used to estimate causal effects. False discovery rate-corrected p<0.1 was considered significant. SNP, single-nucleotide polymorphism.

To ensure the robustness of causality, we performed a series of sensitivity analyses on the MR results of 196 GM taxa (online supplemental table S2). In significant causal relationships, Cochran’s Q test, MR-PRESSO global test and MR-Egger intercept collectively indicated no significant heterogeneity (p>0.05) and horizontal pleiotropy (p>0.05). Leave-one-out analysis consistently demonstrated the stability of the MR results, with no outliers detected in the overall analysis (online supplemental figure S1).

Reverse MR of glaucoma on GM

In the reverse MR analysis, glaucoma had no significant causal effect on the abundance of GM taxa (online supplemental figure S2). However, we observed a suggestive association between glaucoma and genus Phascolarctobacterium (OR 0.926, 95% CI 0.863 to 0.994, p=0.034) (online supplemental figure S3). However, this association was no longer significant after the FDR adjustment. No significant causal association was found between glaucoma and other GM taxa. Sensitivity analyses also confirmed the reliability of the causal effects assumed in the reverse MR results (online supplemental table S3).

The causal relationship between GM and glaucoma in the validation cohort

In the validation cohort (MRC-IEU Consortium), 2052 SNPs were selected as IVs for GM taxa, with the number of IVs for each GM taxon varying from 3 to 21 and F-statistics ranging from 20.379 to 27.739 (online supplemental table S4). During the validation, we examined the causal effect of 196 GM taxa on glaucoma risk based on the MRC-IEU Consortium. IVW analyses confirmed a potential causal association of 18 GM taxa with glaucoma risk (figure 4). However, only the genus Clostridium sensu stricto 1 and LachnospiraceaeUCG010 were further validated for glaucoma risk. As in the primary analysis, the IVW estimate suggested that genus LachnospiraceaeUCG010 (OR 1.002, 95% CI 1.000 to 1.004, p=0.033) is a risk factor for glaucoma. Genus Clostridium sensu stricto 1 (OR 0.997, 95% CI 0.995 to 0.999, p=0.011) had a protective effect against glaucoma, although not corrected by FDR adjustment in the primary analysis (figure 5). Sensitivity analyses showed no significant heterogeneity or pleiotropy, further validating the causal effect of genus LachnospiraceaeUCG010 on glaucoma (online supplemental table S5). The leave-one-out analysis yielded similar results, with no abnormal SNPs being found (online supplemental figure S4).

Figure 4

Causal analyses of each gut microbial component taxon and glaucoma based on five Mendelian randomisation (MR) analyses (p<1×10−5) in the validation stage (MRC-IEU Consortium). From outside to inside are the taxon name, the p value of inverse variance weighted (IVW), p value of MR-Egger, p value of the weighted median (WM), p value of the simple mode (SM), p value of weighted mode (WMODE) and OR value based on IVW results.

Figure 5

Causality in the validation analysis stage (MRC-IEU Consortium). Forest plot showing significant causal associations. Five two-sample Mendelian randomisation (MR) methods (inverse variance weighted, MR-Egger, weighted median, simple mode and weighted mode) were used to estimate causal effects. False discovery rate-corrected p<0.1 was considered significant. SNP, single-nucleotide polymorphism.

Discussion

Despite the fact that observational studies have suggested the potential importance of the gut-retina axis in the pathogenesis of glaucoma,12 13 the exact causal relationship between GM and glaucoma risk remains uncertain. In this study, we conducted a bidirectional two-sample MR analysis to explore this relationship. Our results indicate that genetically elevated levels of the family Oxalobacteraceae and the genus Eggerthella are associated with a decreased risk of glaucoma. In contrast, genetically elevated levels of the genus Bilophila, LachnospiraceaeUCG010 and genus Ruminiclostridium 9 are associated with an increased risk of glaucoma. In general, these results suggest some evidence for a causal influence of ecological dysbiosis of GM on glaucoma pathogenesis. However, the causal relationship between other GM taxa and glaucoma needs to be further investigated.

Our MR design effectively reduces the potential bias from confounding and reversing causation that often occurs in observational studies. Leveraging GWAS summary statistics from large samples increases the reliability of our results. Although our MR analysis did not reveal significant differences between GM taxa compared with previous observational glaucoma studies, this may be due to the limitations of observational studies, or the fewer SNP instrumental variables used for these GM taxa. Our MR studies further support the concept of a gut-retina axis by investigating the role of the genus LachnospiraceaeUCG010 in ocular disease. Although the precise causative mechanism of glaucoma remains unclear, findings from both our primary and validation phases consistently indicate that a higher abundance of the genetically predicted genus LachnospiraceaeUCG010 correlates with an increased risk of glaucoma. However, further investigation is needed to elucidate the underlying mechanism. GM taxa may influence the development of glaucoma through several pathways. For example, the Lachnospiraceae family members, a genus known for producing short-chain fatty acids (SCFAs), generate SCFAs through fermentation.30 SCFAs are commonly recognised as beneficial microbial metabolites for human health.31 However, recent research suggests a potential link between SCFAs and the development of glaucoma. Our discoveries align with previous studies,12 suggesting the involvement of the gut microbiome in glaucoma pathogenesis. Their research highlights that administering intestinally metabolised SCFAs exacerbated retinal cell loss. The impact of intestinal microorganisms and their SCFAs metabolites appears to influence neuroinflammatory responses by activating retinal microglia through the microRNA network. Moreover, another study32 demonstrated that SCFAs treatment promoted microglial activation, heightened neuroinflammation and exacerbated Parkinson’s pathophysiology in mouse models. One plausible hypothesis suggests that the genus LachnospiraceaeUCG010 produces SCFAs, subsequently facilitating full maturation and inflammatory potential of microglia. As such, both the genus LachnospiraceaeUCG010 and its SCFAs metabolites might emerge as novel targets for preserving optic nerve function in glaucoma. These findings offer promising avenues for the development of innovative glaucoma treatments.

Our findings further indicate a potential association between the genus Bilophila and an elevated risk of glaucoma. Bilophila, identified as an opportunistic pathogen, has been linked to exacerbating intestinal inflammation.33 Studies reveal that increased levels of Bilophila wadsworthia, especially in conjunction with a high-fat diet, intensify inflammatory reactions, disrupt intestinal barrier function and perturb bile metabolism, thereby contributing to irregular glucose metabolism and the onset of fatty liver development.34 In this study, we observed an interesting finding related to the family Oxalobacteraceae, which may reduce the risk of glaucoma. However, the existing literature on the Oxalobacteraceae family is limited. The gut microbiome also serves in the breakdown of detrimental substances, with Oxalobacteraceae notably playing a crucial role in the catabolism of oxalate.35 Oxalobacteraceae metabolise oxalate in the gastrointestinal tract, maintaining healthy oxalate homeostasis in the gastrointestinal tract.35 Imbalances in oxalate metabolism can lead to various health risks, including localised and systemic inflammation, progressive renal disease and cardiovascular complications.36–38 The association and underlying mechanisms of the Oxalobacteraceae family warrant further investigation and understanding. Future studies will shed light on these mechanisms and elucidate the potential involvement of the Oxalobacteraceae family in glaucoma. Clostridium sensu stricto 1 may also reduce the risk of glaucoma. However, at the primary stage, Clostridium sensu stricto 1 did not survive FDR correction; however, at the validation stage, it reiterated results consistent with the discovery stage. Genetically predicted elevated levels of Clostridium sensu stricto 1 were associated with a decreased risk of glaucoma. While prior research has indicated that members of Clostridium sensu stricto 1 might be pathogenic and considered markers of a less healthy microbiome,39 the potential intricate interactions among gut microbiome could explain the disparity between genetic predictions and clinical observations. Further validation through prospective randomised controlled trials might be necessary to clarify this discrepancy. Future investigations are encouraged to uncover the potential role of Clostridium sensu stricto 1 in reducing the risk of glaucoma. As for Ruminiclostridium 9, studies in mice have highlighted its involvement in regulating lipid metabolism, mitigating inflammation and bolstering intestinal barrier function.40 However, its specific implications in glaucoma remain unexplored. Similar to genus Eggerthella, both necessitate additional scrutiny within the context of glaucoma.

Despite the preliminary findings suggesting a possible causal relationship between specific GM taxa and glaucoma, there was considerable variation among individuals, which is likely to be due to the unique composition of their GM, genetic factors and lifestyle. Therefore, future investigations should explore the underlying mechanisms to determine whether tailored interventions targeting GM may benefit specific individuals. Based on the results of this study, the potential role of GM in treating glaucoma needs to considered. There are opportunities to influence GM alterations that may influence glaucoma development through strategies such as dietary modification, probiotics and prebiotics. Nevertheless, the safety and efficacy of these interventions require further thorough research and validation.

A recent study investigated the potential association of GM with age-related macular degeneration and glaucoma.41 This study has several strengths over previous research. The use of MR analysis enabled the identification of causal relationships between GM and glaucoma, effectively mitigating confounding and reverse causation for precise causal inference. The robustness of the analytical tools used was supported by the inclusion of GWAS summary data from a comprehensive meta-analysis, the largest in its category. Rigorous checks for horizontal pleiotropy and heterogeneity were performed using MR-PRESSO, MR-Egger regression intercept term test and Cochran’s Q test. In addition, an independent validation set was included to confirm previously established causal relationships. This significantly increased the reliability of the findings.

Nevertheless, several limitations of this study must be considered when interpreting the results. First, the sample size in this study was relatively small, the duration of the study was short and the number of SNPs strongly associated with GM was limited. Therefore, larger GWAS data and long-term studies are essential to fully substantiate the observed effects. Second, the genetic data used in this study were obtained from individuals of European ancestry, which precludes the generalisation of our findings to other populations. Finally, this study focused exclusively on the influence of GM on glaucoma without considering the potential impact of other factors, such as lifestyle and genetics, which may also significantly influence disease development. Future research should focus on elucidating the mechanisms linking GM and glaucoma, translating these findings into practical clinical treatment strategies, exploring associations with other ocular diseases and understanding the broader implications of GM on systemic health that warrant further in-depth investigation.

In conclusion, our study has uncovered an initial potential link between GM and glaucoma and highlights modulation of the gut microbiome as a novel avenue for optic neuroprotective therapy in glaucoma. These findings provide preliminary insights for future research and potential treatment strategies. However, comprehensive investigations are essential to deepen our understanding of this association and to effectively translate these discoveries into viable clinical applications.

Data availability statement

Data are available in a public, open access repository. Data sets analysed in this study were publicly available from the IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/), MiBioGen Consortium (https://mibiogen.gcc.rug.nl/) and FinnGen Consortium (https://www.finngen.fi/en/access_results).

Ethics statements

Patient consent for publication

Acknowledgments

We thank the IEU OpenGWAS database, FinnGen Consortium and MiBioGen Consortium for providing statistical data. We also appreciate BioRender (biorender.com) for assistance in creating figure 1.

References

Supplementary material

  • 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

  • YW and RS are joint first authors.

  • Contributors MZ and JQ designed this study and act as guarantor. YW and RS conducted analyses and drafted the manuscript. HC, RS, SB and ZZ contributed to the interpretation of data. MZ and JQ revised the manuscript draft.

  • Funding This study was supported by grants from the Real World Study Project of Hainan Boao Lecheng Pilot Zone (Real World Study Base of NMPA) (HNLC2022RWS002).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • 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.