Thanks for the valuable comments by Mokou and colleagues. We agree that our results differed from a few previous research findings. This is mainly due to the differences in terms of that study used single outcome, we emphasized the dynamic progression trajectory of CVD. We have provided a detailed explanation of our methods in our previous response to the reviewers and other comments. On the other hand, the different populations that were included in the two analyses could also partly explain the inconsistency. In our study, we excluded subjects who were diagnosed with AF, heart failure, MI, stroke, and cancer at baseline. And the other study excluded subjects with CVD (MI and stroke) and cancer at baseline.
with great interest we read the manuscript entitled “Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study” by Chen et al., published in BMJmedicine [1]. The authors reported that dietary supplement of fish oil does not appear to result in a positive impact on the risk of cardiovascular disease, based on data from 415737 individuals, enrolled in the UK biobank study. The authors conclude “Regular use of fish oil supplements might be a risk factor for atrial fibrillation and stroke among the general population but could be beneficial for progression of cardiovascular disease from atrial fibrillation to major adverse cardiovascular events, and from atrial fibrillation to death.”
However, in another study reported 2020 in BMJ [2], the authors, apparently using an almost identical dataset (427 678 datasets from the UK Biobank study, likely most of the datasets overlapped) reported on apparent benefits of fishoil supplement, coming to the conclusion “Habitual use of fish oil seems to be associated with a lower risk of all cause and CVD mortality and to provide a marginal benefit against CVD events among the general population.”
We are surprised that this apparent disagreement between the two reported studies was not discussed by the authors of the BMJmedicine paper, as this seems to be quite relevant. We respectfully ask for an explanation, and an in depth discussion of the apparent inconsistency....
with great interest we read the manuscript entitled “Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study” by Chen et al., published in BMJmedicine [1]. The authors reported that dietary supplement of fish oil does not appear to result in a positive impact on the risk of cardiovascular disease, based on data from 415737 individuals, enrolled in the UK biobank study. The authors conclude “Regular use of fish oil supplements might be a risk factor for atrial fibrillation and stroke among the general population but could be beneficial for progression of cardiovascular disease from atrial fibrillation to major adverse cardiovascular events, and from atrial fibrillation to death.”
However, in another study reported 2020 in BMJ [2], the authors, apparently using an almost identical dataset (427 678 datasets from the UK Biobank study, likely most of the datasets overlapped) reported on apparent benefits of fishoil supplement, coming to the conclusion “Habitual use of fish oil seems to be associated with a lower risk of all cause and CVD mortality and to provide a marginal benefit against CVD events among the general population.”
We are surprised that this apparent disagreement between the two reported studies was not discussed by the authors of the BMJmedicine paper, as this seems to be quite relevant. We respectfully ask for an explanation, and an in depth discussion of the apparent inconsistency.
We very much look forward to your reply,
Sincerely
Marika Mokou1, Agnieska Latosinska1 and Harald Mischak1,2
1mosaiques-diagnostics GmbH, Rotenburger Str. 20, D-30659 Hannover GERMANY,
2University of Glasgow, 126 University Avenue, Glasgow, G12 8TA
REFERENCES
[1] Chen, G., Qian, Z. M., Zhang, J., Zhang, S., Zhang, Z., Vaughn, M. G., Aaron, H. E., Wang, C., Lip, G. Y., and Lin, H. (2024) Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJ Med., 3: e000451.
[2] Li, Z. H., Zhong, W. F., Liu, S., Kraus, V. B., Zhang, Y. J., Gao, X., Lv, Y. B., Shen, D., Zhang, X. R., Zhang, P. D., Huang, Q. M., Chen, Q., Wu, X. B., Shi, X. M., Wang, D., and Mao, C. (2020) Associations of habitual fish oil supplementation with cardiovascular outcomes and all cause mortality: evidence from a large population based cohort study. BMJ, 368: m456.
Thanks for the valuable comments by Ortega, Harris, Hands, Viana and colleagues. Ortega and Hands suggested that this study is an observational study and fish oil use is not a randomized allocation, which is lacking of causal conclusions. We acknowledge that randomized controlled trials could theoretically provide evidence of a higher standard than observational studies. However, it is widely acknowledged that observational studies also provide high-quality evidence by encompassing generalizable populations, controlling for various covariates, and utilizing appropriate statistical methods, allowing an assessment of the potential effects of fish oil supplementation in real-world practice.
Among these, the multistate model is a prevalent model for describing longitudinal survival data1-3. The multistate model is defined as a stochastic process model where the process takes one of several discrete states. In medical research, these states can be health, disease, disease complications, and death. Changes in states are referred to as transitions or events, such as the onset of disease, the development of disease complications, and death. States are classified as transient or absorbing. Transient states can transition to other states, while absorbing states represent the end of the process, with no further transitions possible. The multistate model provides a visual representation of the state structure, clarifying the possible transitions between states. Ther...
Thanks for the valuable comments by Ortega, Harris, Hands, Viana and colleagues. Ortega and Hands suggested that this study is an observational study and fish oil use is not a randomized allocation, which is lacking of causal conclusions. We acknowledge that randomized controlled trials could theoretically provide evidence of a higher standard than observational studies. However, it is widely acknowledged that observational studies also provide high-quality evidence by encompassing generalizable populations, controlling for various covariates, and utilizing appropriate statistical methods, allowing an assessment of the potential effects of fish oil supplementation in real-world practice.
Among these, the multistate model is a prevalent model for describing longitudinal survival data1-3. The multistate model is defined as a stochastic process model where the process takes one of several discrete states. In medical research, these states can be health, disease, disease complications, and death. Changes in states are referred to as transitions or events, such as the onset of disease, the development of disease complications, and death. States are classified as transient or absorbing. Transient states can transition to other states, while absorbing states represent the end of the process, with no further transitions possible. The multistate model provides a visual representation of the state structure, clarifying the possible transitions between states. Therefore, this methodology has the potential to explain some of the associations observed in observational studies, contributing to a more comprehensive understanding of disease progression and outcomes.
In addition, to address the issue of non-randomized use of fish oil, we employed the inverse probability weighting method (IPW)4-6 to validate the robustness of the results. IPW as a causal modeling, seeks to make data from an observational study closely approximate that from a randomized controlled trial whose causal interpretation is widely accept. Furthermore, we undertook a series of analyses such as adjusting difference covariates to test the robustness of our findings. Of course, as Hands pointed out, there may still be some limitations to this study, such as residual confounding, lack of dosage and formula information for fish oil supplements. Nevertheless, we have reported the association and believe it has the potential to guide future research in this area, which requires further research to confirm.
We agree with Viana and co-authors that the number needed to harm (NNH) might enhance shared decision-making processes around the use of fish oil supplements by providing more information to the general public and medical professionals on the hazards associated with such supplements. In the following research, we will be focusing on the use of this approach. Furthermore, we take note of Dr. Fladerer's insight that one of the most significant risk factors for new-onset atrial fibrillation (AF) is age, with older individuals having a larger risk than younger ones. We appreciate his viewpoint and have made modifications in our research, such as adjusting age components in final models and doing age-subgroup analysis. These findings are consistent with other research7 and showed that regular fish oil intake may be a risk factor for AF.
1. Neumann JT, Thao LTP, Callander E, et al. A multistate model of health transitions in older people: a secondary analysis of ASPREE clinical trial data. Lancet Healthy Longev. Feb 2022;3(2):e89-e97. doi:10.1016/s2666-7568(21)00308-1
2. Crowther MJ, Lambert PC. Parametric multistate survival models: Flexible modelling allowing transition-specific distributions with application to estimating clinically useful measures of effect differences. Stat Med. Dec 20 2017;36(29):4719-4742. doi:10.1002/sim.7448
3. Cook ME, Bruun NH, Davidsen L, Drewes AM, Olesen SS. Multistate Model of the Natural History of Inflammatory Pancreatic Diseases: A Nationwide Population-based Cohort Study. Gastroenterology. Dec 2023;165(6):1547-1557.e4. doi:10.1053/j.gastro.2023.08.042
4. Mansournia MA, Altman DG. Inverse probability weighting. Bmj. Jan 15 2016;352:i189. doi:10.1136/bmj.i189
5. Liu T, Hogan JW. Unifying instrumental variable and inverse probability weighting approaches for inference of causal treatment effect and unmeasured confounding in observational studies. Stat Methods Med Res. Mar 2021;30(3):671-686. doi:10.1177/0962280220971835
6. Hernán MA, Robins JM. Estimating causal effects from epidemiological data. J Epidemiol Community Health. Jul 2006;60(7):578-86. doi:10.1136/jech.2004.029496
7. Gencer B, Djousse L, Al-Ramady OT, Cook NR, Manson JE, Albert CM. Effect of Long-Term Marine ɷ-3 Fatty Acids Supplementation on the Risk of Atrial Fibrillation in Randomized Controlled Trials of Cardiovascular Outcomes: A Systematic Review and Meta-Analysis. Circulation. Dec 21 2021;144(25):1981-1990. doi:10.1161/circulationaha.121.055654
The study of Chen et al. (2024) presents a prospective cohort study conducted within the UK Biobank, aiming to evaluate the effects of fish oil supplements on the clinical course of cardiovascular disease, including transitions from a healthy state to atrial fibrillation, major adverse cardiovascular events, and death (1). While the study is ambitious and large-scale, involving 415,737 participants with a median follow-up of 11.9 years, several critical issues may compromise the validity of its conclusions.
One major concern is the potential confounding effect of age, which is not adequately addressed in the study. Age is a well-established and significant risk factor for atrial fibrillation (AF) and other cardiovascular diseases (2). Additionally, it is notable that participants who regularly used fish oil supplements in this study were significantly older than those who did not (p < 0.0001). Given that the risk of AF and adverse cardiovascular events increases with age, differences in age distribution between those who do not regularly use fish oil supplements and those who do could significantly bias the results. This age-related bias likely contributes to the different risks observed, as the older age of supplement users naturally predisposes them to higher rates of AF, independent of fish oil supplement use.
The reported absolute risk for AF was 4.24% for participants who did not use fish oil supplements and 4.75% for those who regularly used fish oil s...
The study of Chen et al. (2024) presents a prospective cohort study conducted within the UK Biobank, aiming to evaluate the effects of fish oil supplements on the clinical course of cardiovascular disease, including transitions from a healthy state to atrial fibrillation, major adverse cardiovascular events, and death (1). While the study is ambitious and large-scale, involving 415,737 participants with a median follow-up of 11.9 years, several critical issues may compromise the validity of its conclusions.
One major concern is the potential confounding effect of age, which is not adequately addressed in the study. Age is a well-established and significant risk factor for atrial fibrillation (AF) and other cardiovascular diseases (2). Additionally, it is notable that participants who regularly used fish oil supplements in this study were significantly older than those who did not (p < 0.0001). Given that the risk of AF and adverse cardiovascular events increases with age, differences in age distribution between those who do not regularly use fish oil supplements and those who do could significantly bias the results. This age-related bias likely contributes to the different risks observed, as the older age of supplement users naturally predisposes them to higher rates of AF, independent of fish oil supplement use.
The reported absolute risk for AF was 4.24% for participants who did not use fish oil supplements and 4.75% for those who regularly used fish oil supplements. These risks fall within the normal prevalence range of 2.1% to 6.4% for individuals aged 55 to 69 years. This suggests that the observed difference in AF incidence among supplement users might not be clinically significant and is within expected variability for this age group.
In conclusion, while the current study provides valuable data on the potential impacts of fish oil supplements, its results and conclusions may be misleading due to inadequate control for age, a critical confounder, and the significant age differences between supplement users and non-users. The reported absolute risk for AF among supplement users falls within the expected prevalence range for their age group, indicating that the findings may not represent a clinically meaningful increase in risk. Further research with meticulous adjustment for age is necessary to elucidate the true effects of fish oil supplements on cardiovascular health.
(1) Chen G, Qian Z(, Zhang J, et alRegular use of fish oil supplements and course of cardiovascular diseases: prospective cohort studyBMJ Medicine 2024;3:e000451. doi: 10.1136/bmjmed-2022-000451.
(2) Kjerpeseth LJ, et al. Heart 2021;107:201–207. doi:10.1136/heartjnl-2020-316624.
To the editors of the BMJ Medicine
Dr. Emma Rouke and Dr. Sophie Cook
About the study: Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study
I read with great interest the article published by Chen G. et al. The supplement industry is a multimillion-dollar business, and the use of supplements without any criteria occurs worldwide. Most supplements are sold without prescription, and many consumers think that using supplements involves no health risks.
Chen et al. demonstrate that the use of fish oil supplements increases the incidence of atrial fibrillation and stroke in a prospective study conducted in the UK Biobank. It would be very informative to know the number needed to harm (NNH) to better inform the general population and health professionals about the risks of such supplementation.
This data could also be compared with the number needed to treat (NNT) from another study conducted in the same population (UK Biobank), which shows that habitual fish oil supplementation is associated with a 13% lower risk of all-cause mortality, a 16% lower risk of CVD mortality, and a 7% lower risk of CVD events among the general population.
Such information could lead to better shared decision-making regarding the use of fish oil supplements.
Bibliography:
Chen et al. Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJMED 2024;3:e000...
To the editors of the BMJ Medicine
Dr. Emma Rouke and Dr. Sophie Cook
About the study: Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study
I read with great interest the article published by Chen G. et al. The supplement industry is a multimillion-dollar business, and the use of supplements without any criteria occurs worldwide. Most supplements are sold without prescription, and many consumers think that using supplements involves no health risks.
Chen et al. demonstrate that the use of fish oil supplements increases the incidence of atrial fibrillation and stroke in a prospective study conducted in the UK Biobank. It would be very informative to know the number needed to harm (NNH) to better inform the general population and health professionals about the risks of such supplementation.
This data could also be compared with the number needed to treat (NNT) from another study conducted in the same population (UK Biobank), which shows that habitual fish oil supplementation is associated with a 13% lower risk of all-cause mortality, a 16% lower risk of CVD mortality, and a 7% lower risk of CVD events among the general population.
Such information could lead to better shared decision-making regarding the use of fish oil supplements.
Bibliography:
Chen et al. Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJMED 2024;3:e000451
Li et al. Associations of habitual fish oil supplementation with cardiovascular outcomes and all cause mortality: evidence from a large population based cohort study. BMJ 2020;368:m456
Chen et al. (2024) find that fish oil supplementation exerts several adverse effects in “healthy” users without baseline history of cardiovascular disease (CVD). In particular, the authors note that the risk of conversion from healthy status to atrial fibrillation (HR 1.13 p<0.001) and stroke (HR 1.05, p=0.05) were increased in healthy users without a pre-existent diagnosis of CVD (1).
While we applaud the authors for their research, we note several major flaws with their results that warrant further exploration.
First, the authors claim that stroke risk is increased in “healthy” users of fish oil. The definition of “healthy” user is heterogenous and misleading as the authors use a mix of electronic medical records and self-reported survey data to derive this definition. Further, fish oil use is not a randomized allocation. Consequently, fish oil users may have some indication for elevated CVD or stroke risk at baseline that prompts its consumption, which would indicate reverse causation. Therefore, the reliability of survey data for such indications is weak at best and cannot be used to extrapolate a truly CVD risk-free state. Moreover, although hypertension is the most significant modifiable risk factor for stroke, family history, inflammatory parameters, and polygenic risk scores may inform personalized CVD risk to a similar or greater extent, and were not adequately controlled in this analysis (2).
Chen et al. (2024) find that fish oil supplementation exerts several adverse effects in “healthy” users without baseline history of cardiovascular disease (CVD). In particular, the authors note that the risk of conversion from healthy status to atrial fibrillation (HR 1.13 p<0.001) and stroke (HR 1.05, p=0.05) were increased in healthy users without a pre-existent diagnosis of CVD (1).
While we applaud the authors for their research, we note several major flaws with their results that warrant further exploration.
First, the authors claim that stroke risk is increased in “healthy” users of fish oil. The definition of “healthy” user is heterogenous and misleading as the authors use a mix of electronic medical records and self-reported survey data to derive this definition. Further, fish oil use is not a randomized allocation. Consequently, fish oil users may have some indication for elevated CVD or stroke risk at baseline that prompts its consumption, which would indicate reverse causation. Therefore, the reliability of survey data for such indications is weak at best and cannot be used to extrapolate a truly CVD risk-free state. Moreover, although hypertension is the most significant modifiable risk factor for stroke, family history, inflammatory parameters, and polygenic risk scores may inform personalized CVD risk to a similar or greater extent, and were not adequately controlled in this analysis (2).
Second, the reports of stroke risk in healthy users suffers, too, when assessing the totality of the evidence, i.e. several meta-analyses have found opposite or null associations between fish oil use and stroke (3-5). Similarly, omega-3 fatty acid blood levels (nutritional status) are inversely related to stroke risk in prior analyses (6-8). Moreover, this point is particularly misleading in that major adverse cardiovascular events (MACE) risk, including myocardial infarction and heart failure, were actually reduced in healthy users. Thus, the use of fish oil in healthy users may actually be supported by this analysis, though the authors claim otherwise. In addition, that the lower bound of a supposedly significant elevated stroke risk (p=0.05) includes the line of no effect (1.00) is a dubious result. Additionally, the authors similarly did not report any significant correction, Bonferroni nor Tukey, to address these concerns for adjustment. The net results of these issues prompts concern for an erroneous finding.
Third, heterogeneity in the quality of fish oil supplements substantially limits analysis of the “effects” of its supplementation (9,10). As a non-controlled substance, fish oil quality has been demonstrated to vary significantly with respect to oxidation, pollutants (PCBs, dioxins, furans), and omega-3 (EPA, DHA) concentration. For example, it has been demonstrated that oxidized fish oil demonstrated null or adverse effects on several markers of CVD risk (9,10). Similarly, without more information about the omega-3 concentration of the supplements used, it is difficult to conclude that omega-3 fatty acids are the true culprit for the supposed result, as most fish oil supplements fall well-below the “pharmaceutical” standardized concentration (80-90% omega-3 concentration) reportedly contained by pharmaceutical omega-3 preparations, which have demonstrated significant risk reductions across a range of MACE including ischemic stroke (8). Thus, without more information about the quality of fish oil in use, this analysis may falsely conclude that supplemental omega-3 fatty acids are detrimental.
In conclusion, while Chen et al.’s piece is intriguing, it may distort, if not falsely report risk associated with supplemental fish oil use and omega-3 fatty acid broadly.
References
1. Chen G, Qian Z(, Zhang J, et alRegular use of fish oil supplements and course of cardiovascular diseases: prospective cohort studyBMJ Medicine 2024;3:e000451. doi: 10.1136/bmjmed-2022-000451
3. Li Z, Zhong W, Liu S, Kraus V B, Zhang Y, Gao X et al. Associations of habitual fish oil supplementation with cardiovascular outcomes and all cause mortality: evidence from a large population based cohort study BMJ 2020; 368 :m456
4. Wang C, Harris WS, Chung M, et al. n-3 Fatty acids from fish or fish-oil supplements, but not alpha-linolenic acid, benefit cardiovascular disease outcomes in primary- and secondary-prevention studies: a systematic review. Am J Clin Nutr. 2006;84(1):5-17. doi:10.1093/ajcn/84.1.5
5. Khan SU, Lone AN, Khan MS, et al. Effect of omega-3 fatty acids on cardiovascular outcomes: A systematic review and meta-analysis. EClinicalMedicine. 2021;38:100997. Published 2021 Jul 8. doi:10.1016/j.eclinm.2021.100997
6. Iso H, Rexrode KM, Stampfer MJ, et al. Intake of Fish and Omega-3 Fatty Acids and Risk of Stroke in Women. JAMA. 2001;285(3):304–312. doi:10.1001/jama.285.3.304
7. Qin, ZZ., Xu, JY., Chen, GC. et al. Effects of fatty and lean fish intake on stroke risk: a meta-analysis of prospective cohort studies. Lipids Health Dis 17, 264 (2018). https://doi.org/10.1186/s12944-018-0897-z
8. O'Keefe JH, Tintle NL, Harris WS, et al. Omega-3 Blood Levels and Stroke Risk: A Pooled and Harmonized Analysis of 183 291 Participants From 29 Prospective Studies [published correction appears in Stroke. 2024 Mar;55(3):e109]. Stroke. 2024;55(1):50-58. doi:10.1161/STROKEAHA.123.044281
9. Hands, J. M., Anderson, M. L., Cooperman, T., & Frame, L. A. (2024). A Multi-Year Rancidity Analysis of 72 Marine and Microalgal Oil Omega-3 Supplements. Journal of Dietary Supplements, 21(2), 195–206. https://doi.org/10.1080/19390211.2023.2252064
10. HANDS, Jacob M.; FRAME, Leigh A. Omega-3 Fatty Acid Therapy: A Review of Study Design Flaws, Quality, and Composition. Medical Research Archives, [S.l.], v. 12, n. 4, apr. 2024. ISSN 2375-1924. Available at: <https://esmed.org/MRA/mra/article/view/5273>. Date accessed: 26 may 2024. doi: https://doi.org/10.18103/mra.v12i4.5273.
In Chen and colleagues’ observational report of use of fish oil supplements (FOS) and cardiovascular outcomes in the UK Biobank study (1), their stated conclusions suggest equipoise of observed relationships, i.e., relatively balanced risks vs. benefits. However, only one adverse association achieved statistical significance - onset of atrial fibrillation (AF) - while five protective associations did so, including onset of heart failure, mortality after heart failure, and (after onset of AF) risk of myocardial infarction, major adverse cardiovascular events, and death (Figure 1). While important, AF also remains a relatively less severe clinical outcome compared with heart failure and myocardial infarction. These new findings provide evidence to support general cardiovascular benefits of FOS use.
Ten prior studies - all from UK Biobank - have assessed observational relationships between FOS use and health conditions, ranging from fractures to liver cancer to CV disease to dementia to death (2-11). Across these, 18 relationships were both statistically significant and favorable for FOS users (including total mortality); and only 1 showed an adverse association (AF). Thus, consistent with the new report, the overall findings from the UK Biobank support important net health benefits of FOS use.
These observational studies have advantages of large, fairly generalizable populations and extended follow-up, allowing an assessment of potential long-term effects of...
In Chen and colleagues’ observational report of use of fish oil supplements (FOS) and cardiovascular outcomes in the UK Biobank study (1), their stated conclusions suggest equipoise of observed relationships, i.e., relatively balanced risks vs. benefits. However, only one adverse association achieved statistical significance - onset of atrial fibrillation (AF) - while five protective associations did so, including onset of heart failure, mortality after heart failure, and (after onset of AF) risk of myocardial infarction, major adverse cardiovascular events, and death (Figure 1). While important, AF also remains a relatively less severe clinical outcome compared with heart failure and myocardial infarction. These new findings provide evidence to support general cardiovascular benefits of FOS use.
Ten prior studies - all from UK Biobank - have assessed observational relationships between FOS use and health conditions, ranging from fractures to liver cancer to CV disease to dementia to death (2-11). Across these, 18 relationships were both statistically significant and favorable for FOS users (including total mortality); and only 1 showed an adverse association (AF). Thus, consistent with the new report, the overall findings from the UK Biobank support important net health benefits of FOS use.
These observational studies have advantages of large, fairly generalizable populations and extended follow-up, allowing an assessment of potential long-term effects of FOS use in real-world practice. Of course, limitations exist, in particular residual confounding which could be in unpredictable directions, given the competing possibilities of both a healthy user effect (i.e., people with healthier lifestyles choosing to take FOS) versus confounding by indication (i.e., people who have greater disease risk electing to take FOS).
Yet, meta-analyses of randomized clinical trials (RCTs) generally support these observational findings, with net benefits of FOS for major cardiovascular events as well as risk after heart failure, coupled with modestly higher risk of AF (12, 13).
In sum, the findings from Chen et al. provide new evidence on regular FOS use supporting a higher risk of AF, together with benefits for major cardiovascular events, heart failure, and mortality.
References
1. Chen G, Qian ZM, Zhang J, Zhang S, Zhang Z, Vaughn MG, et al. Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJ Med 2024;3(1):e000451.
2. Liu X, Li Y, Wan X, Zhuang P, Wu Y, Zhang L, et al. Association of Fish Oil Supplementation with Risk of Coronary Heart Disease in Individuals with Diabetes and Prediabetes: A Prospective Study in the UK Biobank. Nutrients 2023;15(14).
3. Gan X, Liu M, He P, Ye Z, Xiang H, Zhou C, et al. Habitual fish oil supplementation, genetic susceptibility of kidney stones and the risk of new-onset kidney stones. J Clin Lipidol 2024;18(1):e116-e24.
4. Zhang J, Cai A, Chen G, Wang X, Cai M, Li H, et al. Habitual fish oil supplementation and the risk of incident atrial fibrillation: findings from a large prospective longitudinal cohort study. European journal of preventive cardiology 2022;29(14):1911-20.
5. Jiang W, Li FR, Yang HH, Chen GC, Hua YF. Relationship Between Fish Oil Use and Incidence of Primary Liver Cancer: Findings From a Population-Based Prospective Cohort Study. Front Nutr 2021;8:771984.
6. Ma T, He L, Luo Y, Li J, Zhang G, Cheng X, et al. Associations of baseline use of fish oil with progression of cardiometabolic multimorbidity and mortality among patients with hypertension: a prospective study of UK Biobank. European journal of nutrition 2022:1-10.
7. Ma H, Zhou T, Li X, Heianza Y, Qi L. Use of fish oil supplements is differently related to incidence of all-cause and vascular dementia among people with the distinct APOE ε4 dosage. Clinical nutrition (Edinburgh, Scotland) 2022;41(3):731-6.
8. Liu X, Zhuang P, Li Y, Wu F, Wan X, Zhang Y, et al. Association of fish oil supplementation with risk of incident dementia: A prospective study of 215,083 older adults. Clinical nutrition (Edinburgh, Scotland) 2022;41(3):589-98.
9. Huang X, Li Y, Zhuang P, Liu X, Zhang Y, Zhang P, et al. Habitual Fish Oil Supplementation and Risk of Incident Inflammatory Bowel Diseases: A Prospective Population-Based Study. Front Nutr 2022;9:905162.
10. Mei Z, Chen GC, Hu J, Lin C, Sun Z, Liu C, et al. Habitual use of fish oil supplements, genetic predisposition, and risk of fractures: a large population-based study. Am J Clin Nutr 2021;114(3):945-54.
11. Li Z-H, Zhong W-F, Liu S, Kraus VB, Zhang Y-J, Gao X, et al. Associations of habitual fish oil supplementation with cardiovascular outcomes and all cause mortality: evidence from a large population based cohort study. BMJ (Clinical research ed) 2020;368:m456.
12. Laukkanen JA, Bernasconi AA, Lavie CJ. Bringing the Potential Benefits of Omega-3 to a Higher Level. Mayo Clinic proceedings 2024;99(4):520-3.
13. Rizos EC, Markozannes G, Tsapas A, Mantzoros CS, Ntzani EE. Omega-3 supplementation and cardiovascular disease: formulation-based systematic review and meta-analysis with trial sequential analysis. Heart 2021;107(2):150-8.
In a recent publication, Chen et al. (1) show the potential negative effects of fish oil supplementation on incident atrial fibrillation (AF) and subsequent effects on cardiovascular events (CVE) and death in the UK Biobank. Even though the authors start with a well-defined question (i.e., risk of AF had everyone taken taking fish oil supplements for primary prevention vs. had they not), moving into the conclusion, they state: “Regular use of fish oil supplements […] could be beneficial for progression of cardiovascular disease from AF to major adverse CVE, and from AF to death”. We reflect on the aspects for which the study was not fit for this conclusion.
To illustrate the point, imagine that the authors had instead conducted a randomized trial to estimate their well-defined question – ‘Does fish oil supplementation reduce the incidence of AF and CVE?’. The investigators would recruit a population free of AF and CVE at baseline, and randomly assign them to regular fish oil supplement or placebo. Follow-up would then start until AF, CVE, or death. The investigators would determine if the intervention prevented AF, CVE or death by counting the number of cases at the end of the follow-up period, or by comparing survival curves.
Now suppose the investigators of this trial conduct a secondary analysis, among those who develop AF during the first half of the follow up period. Among these individuals, the investigators then count the incidence of further CVE and st...
In a recent publication, Chen et al. (1) show the potential negative effects of fish oil supplementation on incident atrial fibrillation (AF) and subsequent effects on cardiovascular events (CVE) and death in the UK Biobank. Even though the authors start with a well-defined question (i.e., risk of AF had everyone taken taking fish oil supplements for primary prevention vs. had they not), moving into the conclusion, they state: “Regular use of fish oil supplements […] could be beneficial for progression of cardiovascular disease from AF to major adverse CVE, and from AF to death”. We reflect on the aspects for which the study was not fit for this conclusion.
To illustrate the point, imagine that the authors had instead conducted a randomized trial to estimate their well-defined question – ‘Does fish oil supplementation reduce the incidence of AF and CVE?’. The investigators would recruit a population free of AF and CVE at baseline, and randomly assign them to regular fish oil supplement or placebo. Follow-up would then start until AF, CVE, or death. The investigators would determine if the intervention prevented AF, CVE or death by counting the number of cases at the end of the follow-up period, or by comparing survival curves.
Now suppose the investigators of this trial conduct a secondary analysis, among those who develop AF during the first half of the follow up period. Among these individuals, the investigators then count the incidence of further CVE and stroke. Surprisingly, they find a lower incidence of CVE among those assigned to supplementation and conclude that fish oil supplementation is beneficial for secondary prevention. However, by focusing on these individuals, supplementation can no longer be considered randomly assigned. That is, the individuals for whom supplementation caused AF are different to those who developed AF in the control group, who likely had a higher risk of CVE in the first place. The apparent protective effect of supplementation among these individuals is biased and should not have been included in the manuscript.
Estimating the effect of fish oil supplementation for secondary prevention of CVE and death after AF would have required a different experiment. An appropriate design would recruit a population with AF at baseline, and randomly assign these individuals to supplementation or not. In the previous example, selection bias is introduced when selecting on an event occurring after time-zero (2). The multistate analysis used by Chen et al. encounters a similar problem, that is the transition state from AF to CVE selects only those who experience AF, and the estimate cannot be given a causal interpretation.
A similar issue arises when targeting the effect of fish oil supplementation on AF in the presence of competing events (i.e., CVE and death). The cause-specific hazard ratios may be biased since they compare the probability of AF in treated who did not experience the competing event vs untreated who did not experience the competing event (3). This, once more, compares populations that are selected on a post randomization characteristic (i.e., no competing event).
We reinforce the need for careful and explicit target trial specification to avoid unjustified analyses and statements, especially when trying to answer causal questions. Editors and reviewers must ask, if such analysis decisions are so obviously unacceptable as part of the analysis of a randomized experiment, why is it allowed in an observational study? It is unfortunate that this message must be relayed to such a journal as BMJ Medicine, particularly after the article has been picked up by close to 200 news outlets according to BMJ Medicine metrics. These issues should not pass peer-review.
1. Chen G, Qian Z, Zhang J, Zhang S, Zhang Z, Vaughn MG, et al. Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJ Medicine. 2024;3(1):e000451.
2. Hernán MA, Hernández-Díaz S, Robins JM. A structural approach to selection bias. Epidemiology. 2004;15(5):615-25.
3. Young JG, Stensrud MJ, Tchetgen Tchetgen EJ, Hernán MA. A causal framework for classical statistical estimands in failure-time settings with competing events. Stat Med. 2020;39(8):1199-236.
The umbrella review by the research team of Szczerba and colleagues offers a perspective of the potential of scrutinizing the utility of personalized nutrition for impacting type 2 diabetes management and is a contribution to the field. By synthesizing evidence from RCTs that lasted at least 12 weeks, the authors not only attempt to offer a robust assessment of various dietary interventions, but also employ strict methodological rigor through the re-calibration of meta-analyses and application of the GRADEpro approach.
One of the work’s strengths lies in its comprehensive scope, meticulously evaluating the certainty of evidence and identifying gaps for future research. This approach not only signals the reliability of findings due to rigor and quality of studies, but also shows a way to scrutinize the potential of personalized nutrition strategies.
Although authors state and acknowledge limitations, such as the exclusion of the most recent RCTs and the absence of subgroup and sensitivity analyses, which could refine the study’s conclusions further, the study stands out for its dedication to precision and methodological excellence and voluminous work, offering valuable insights into the complex relationship between diet and type 2 diabetes management.
Findings underscore the importance of personalized nutrition in managing cardiometabolic health, exploring new standards for future studies in this vital area. It’s a significant step forward in our unders...
The umbrella review by the research team of Szczerba and colleagues offers a perspective of the potential of scrutinizing the utility of personalized nutrition for impacting type 2 diabetes management and is a contribution to the field. By synthesizing evidence from RCTs that lasted at least 12 weeks, the authors not only attempt to offer a robust assessment of various dietary interventions, but also employ strict methodological rigor through the re-calibration of meta-analyses and application of the GRADEpro approach.
One of the work’s strengths lies in its comprehensive scope, meticulously evaluating the certainty of evidence and identifying gaps for future research. This approach not only signals the reliability of findings due to rigor and quality of studies, but also shows a way to scrutinize the potential of personalized nutrition strategies.
Although authors state and acknowledge limitations, such as the exclusion of the most recent RCTs and the absence of subgroup and sensitivity analyses, which could refine the study’s conclusions further, the study stands out for its dedication to precision and methodological excellence and voluminous work, offering valuable insights into the complex relationship between diet and type 2 diabetes management.
Findings underscore the importance of personalized nutrition in managing cardiometabolic health, exploring new standards for future studies in this vital area. It’s a significant step forward in our understanding of how tailored dietary strategies can contribute to the management of type 2 diabetes, considering the diverse dietary habits, genetic predispositions, and lifestyle factors across different demographics.
We address points made in the two letters written by Rice, Lambert, and their colleagues.
Point 1: Unrealistic premise / Confuses clinical researchers and the public.
We show that polygenic risk scores are poor predictors of disease whether used alone (as they are in direct-to-consumer genetic tests and have been in previous publications), or in combination with other risk factors [1]. If they perform poorly on their own, as we show, they cannot have useful incremental value when used with other risk factors, as we also show. Use of the appropriate metrics, as we employed in our paper, clarifies the position; it is the use of inappropriate metrics that causes needless confusion.
Point 2: Multifactorial models incorporating PRS improve risk prediction and are cost effective.
Any potential screening marker that offers negligible screening performance when used alone or in a risk model cannot be cost effective. In the case of cardiovascular disease, risk factor models offer poor discrimination, with or without polygenic risk scores. Table 1 in our paper [1] shows that over 5000 people need to be genotyped to prevent one additional cardiovascular event. The cost-effectiveness analysis cited by the correspondents [2] did not compare the effectiveness of a multi risk factor model based strategy with alternatives such as an aged based strategy [3]. All risk models (with or without polygenic risk scores) add cost and impose a barrier to accessing ef...
We address points made in the two letters written by Rice, Lambert, and their colleagues.
Point 1: Unrealistic premise / Confuses clinical researchers and the public.
We show that polygenic risk scores are poor predictors of disease whether used alone (as they are in direct-to-consumer genetic tests and have been in previous publications), or in combination with other risk factors [1]. If they perform poorly on their own, as we show, they cannot have useful incremental value when used with other risk factors, as we also show. Use of the appropriate metrics, as we employed in our paper, clarifies the position; it is the use of inappropriate metrics that causes needless confusion.
Point 2: Multifactorial models incorporating PRS improve risk prediction and are cost effective.
Any potential screening marker that offers negligible screening performance when used alone or in a risk model cannot be cost effective. In the case of cardiovascular disease, risk factor models offer poor discrimination, with or without polygenic risk scores. Table 1 in our paper [1] shows that over 5000 people need to be genotyped to prevent one additional cardiovascular event. The cost-effectiveness analysis cited by the correspondents [2] did not compare the effectiveness of a multi risk factor model based strategy with alternatives such as an aged based strategy [3]. All risk models (with or without polygenic risk scores) add cost and impose a barrier to accessing effective, safe, and inexpensive preventive medication such as statins and blood pressure lowering drugs. What is needed now is more prevention and less prediction of cardiovascular disease [3].
Point 3: Polygenic risk scores improve the effectiveness of cancer screening.
Since polygenic risk scores offer little effective discrimination between people later affected or unaffected by diseases as shown in Figures 1, 3, 4, 5 and 6 of our paper[1], they can only marginally improve the effectiveness of cancer or any other screening. The detailed modellng by Huntley and colleagues [4], using methods from our paper, shows that the use of polygenic risk scores to target screening for breast and other common cancers does little to improve screening performance. About one-third of women undergoing mammography screening will avoid dying of breast cancer [5] a medically significant advantage to which polygenic risk scores add little benefit but increase cost.
Point 4: Pooling performance of polygenic risk scores is not recommended.
The median DR5 values used to illustrate the performance of polygenic risk scores for coronary artery disease (DR5 11%) and breast cancer (DR5 10%) are representative of the actual DR5 values of polygenic risk scores developed using some of the largest genome wide association studies for these conditions. We evaluated specific polygenic risk scores (Polygenic Score Catalog identifier PGS000018 for coronary artery disease, DR5 13%; and PGS000004 for breast cancer, DR5 12%) when examining their use along with conventional risk factors or screening tests. The breast cancer polygenic risk score PGS000004 that we used for illustration has been adopted in breast cancer risk models [6] and, to our knowledge, has not been superseded by a substantially higher performing score. The currently ‘best performing’ polygenic risk score for coronary artery disease (PGS003725) [7], reported after our paper was accepted for publication, produces a negligible increase in performance over PGS000018 [8] that was used in our illustrative analysis (DR5 increases from 13% to 14%).
Point 5: The DR5 is usually used for diagnostic tests.
This is incorrect. The detection rate for a given false positive rate has been widely used in antenatal [9] and prostate cancer screening [10] as well as for diagnostic tests. The DR5 metric is both appropriate and informative. We reported performance in terms of the detection rate for a 5% false positive rate (DR5) but also provided a calculator in the supplementary tables and via a weblink to enable readers to calculate detection rate for any user-defined false positive rate or vice versa (https://www.medicalscreeningsociety.com/rsc.asp). We also provided the likelihood ratio and odds of being affected in screening, risk prediction and risk stratification, which are additional metrics established for evaluating predictive tests in medicine.
Point 6: Polygenic risk scores for autoimmune diseases may perform better than for other diseases.
This is true. We commented in our paper that polygenic risk scores for a few autoimmune diseases perform better than for other conditions due to large effect HLA variants. In the example of type 1 diabetes cited by the correspondents [11] using UK Biobank data, the area under the curve (AUC) was 0.92 (DR5 about 63%) [11]. This combines large effect DNA variants at the HLA locus (AUC 0.90, DR5 about 57%) and smaller effect non-HLA variants (AUC 0.75, DR5 about 24%). For most autoimmune diseases, the issue is that screening is not worthwhile in the absence of a safe long term preventative intervention. A possible exception is type 1 diabetes, where targeted monitoring of islet cell antibodies in those at high risk [12], and immunosuppressive treatment [13] soon after the onset of symptoms might become a realistic prospect. This untested possibility in any case does not affect our assessment that disproportionate claims of the value of polygenic risk scores have been made in common disorders such as cardiovascular diseases and common cancers.
The correspondents also refer to a very specific example of polygenic risk scores being helpful in differentiating type 1 from type 2 and monogenic diabetes in adults with a pre-existing diagnosis of diabetes [14]. This shows the value of precision medicine in selecting the best treatment for different types of diabetes, but it is a separate issue from screening or prediction and was not the topic of our paper.
In summary, our analysis provides a clear, comprehensive, and reliable presentation of the performance of polygenic risk scores in screening and prediction. The two latest responses to our article do not alter the conclusions of our paper.
Aroon Hingorani and Nicholas Wald on behalf of all the authors.
References
1 Hingorani AD, Gratton J, Finan C, et al. Performance of polygenic risk scores in screening, prediction, and risk stratification: secondary analysis of data in the Polygenic Score Catalog. BMJ Med. 2023;2:e000554.
2 Kiflen M, Le A, Mao S, et al. Cost-Effectiveness of Polygenic Risk Scores to Guide Statin Therapy for Cardiovascular Disease Prevention. Circ Genomic Precis Med. 2022;15:e003423.
3 Wald NJ, Hingorani AD, Vale S, et al. Comparing the effectiveness of the NHS Health Check and the Polypill Prevention Programmes in the primary prevention of heart attacks and strokes. medRxiv. 2023;2023.10.06.23296215.
4 Huntley C, Torr B, Sud A, et al. Utility of polygenic risk scores in UK cancer screening: a modelling analysis. Lancet Oncol. 2023;24:658–68.
5 Wald NJ, Chamberlain A HA. Report of the European Society of Mastology Breast Cancer Screening Evaluation Committee. The Breast. 1993;2:209–16.
6 Lee A, Mavaddat N, Wilcox AN, et al. BOADICEA: a comprehensive breast cancer risk prediction modelincorporating genetic and nongenetic risk factors. Genet Med. 2019;21:1708–18.
7 Patel AP, Wang M, Ruan Y, et al. A multi-ancestry polygenic risk score improves risk prediction for coronary artery disease. Nat Med. 2023;29:1793–803.
8 Inouye M, Abraham G, Nelson CP, et al. Genomic Risk Prediction of Coronary Artery Disease in 480,000 Adults: Implications for Primary Prevention. J Am Coll Cardiol. 2018;72:1883–93.
9 Wald, N. J. and, Hackshaw A. Tests Using Multiple Markers. Wald and Leck Antenatal and Neonatal Screening. Oxford University Press 2000.
10 Wald NJ, Bestwick JP, Morris JK. Multi-marker risk-based screening for prostate cancer. J Med Screen. 2022;29:123–33.
11 Sharp SA, Rich SS, Wood AR, et al. Development and Standardization of an Improved Type 1 Diabetes Genetic Risk Score for Use in Newborn Screening and Incident Diagnosis. Diabetes Care. 2019;42:200–7.
12 Ziegler AG, Rewers M, Simell O, et al. Seroconversion to multiple islet autoantibodies and risk of progression to diabetes in children. JAMA. 2013;309:2473–9.
13 Waibel M, Wentworth JM, So M, et al. Baricitinib and β-Cell Function in Patients with New-Onset Type 1 Diabetes. N Engl J Med. 2023;389:2140–50.
14 Luckett AM, Weedon MN, Hawkes G, et al. Utility of genetic risk scores in type 1 diabetes. Diabetologia. 2023;66:1589–600.
Thanks for the valuable comments by Mokou and colleagues. We agree that our results differed from a few previous research findings. This is mainly due to the differences in terms of that study used single outcome, we emphasized the dynamic progression trajectory of CVD. We have provided a detailed explanation of our methods in our previous response to the reviewers and other comments. On the other hand, the different populations that were included in the two analyses could also partly explain the inconsistency. In our study, we excluded subjects who were diagnosed with AF, heart failure, MI, stroke, and cancer at baseline. And the other study excluded subjects with CVD (MI and stroke) and cancer at baseline.
Dear Editor,
with great interest we read the manuscript entitled “Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study” by Chen et al., published in BMJmedicine [1]. The authors reported that dietary supplement of fish oil does not appear to result in a positive impact on the risk of cardiovascular disease, based on data from 415737 individuals, enrolled in the UK biobank study. The authors conclude “Regular use of fish oil supplements might be a risk factor for atrial fibrillation and stroke among the general population but could be beneficial for progression of cardiovascular disease from atrial fibrillation to major adverse cardiovascular events, and from atrial fibrillation to death.”
Show MoreHowever, in another study reported 2020 in BMJ [2], the authors, apparently using an almost identical dataset (427 678 datasets from the UK Biobank study, likely most of the datasets overlapped) reported on apparent benefits of fishoil supplement, coming to the conclusion “Habitual use of fish oil seems to be associated with a lower risk of all cause and CVD mortality and to provide a marginal benefit against CVD events among the general population.”
We are surprised that this apparent disagreement between the two reported studies was not discussed by the authors of the BMJmedicine paper, as this seems to be quite relevant. We respectfully ask for an explanation, and an in depth discussion of the apparent inconsistency....
Thanks for the valuable comments by Ortega, Harris, Hands, Viana and colleagues. Ortega and Hands suggested that this study is an observational study and fish oil use is not a randomized allocation, which is lacking of causal conclusions. We acknowledge that randomized controlled trials could theoretically provide evidence of a higher standard than observational studies. However, it is widely acknowledged that observational studies also provide high-quality evidence by encompassing generalizable populations, controlling for various covariates, and utilizing appropriate statistical methods, allowing an assessment of the potential effects of fish oil supplementation in real-world practice.
Among these, the multistate model is a prevalent model for describing longitudinal survival data1-3. The multistate model is defined as a stochastic process model where the process takes one of several discrete states. In medical research, these states can be health, disease, disease complications, and death. Changes in states are referred to as transitions or events, such as the onset of disease, the development of disease complications, and death. States are classified as transient or absorbing. Transient states can transition to other states, while absorbing states represent the end of the process, with no further transitions possible. The multistate model provides a visual representation of the state structure, clarifying the possible transitions between states. Ther...
Show MoreThe study of Chen et al. (2024) presents a prospective cohort study conducted within the UK Biobank, aiming to evaluate the effects of fish oil supplements on the clinical course of cardiovascular disease, including transitions from a healthy state to atrial fibrillation, major adverse cardiovascular events, and death (1). While the study is ambitious and large-scale, involving 415,737 participants with a median follow-up of 11.9 years, several critical issues may compromise the validity of its conclusions.
Show MoreOne major concern is the potential confounding effect of age, which is not adequately addressed in the study. Age is a well-established and significant risk factor for atrial fibrillation (AF) and other cardiovascular diseases (2). Additionally, it is notable that participants who regularly used fish oil supplements in this study were significantly older than those who did not (p < 0.0001). Given that the risk of AF and adverse cardiovascular events increases with age, differences in age distribution between those who do not regularly use fish oil supplements and those who do could significantly bias the results. This age-related bias likely contributes to the different risks observed, as the older age of supplement users naturally predisposes them to higher rates of AF, independent of fish oil supplement use.
The reported absolute risk for AF was 4.24% for participants who did not use fish oil supplements and 4.75% for those who regularly used fish oil s...
To the editors of the BMJ Medicine
Dr. Emma Rouke and Dr. Sophie Cook
About the study: Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study
I read with great interest the article published by Chen G. et al. The supplement industry is a multimillion-dollar business, and the use of supplements without any criteria occurs worldwide. Most supplements are sold without prescription, and many consumers think that using supplements involves no health risks.
Chen et al. demonstrate that the use of fish oil supplements increases the incidence of atrial fibrillation and stroke in a prospective study conducted in the UK Biobank. It would be very informative to know the number needed to harm (NNH) to better inform the general population and health professionals about the risks of such supplementation.
This data could also be compared with the number needed to treat (NNT) from another study conducted in the same population (UK Biobank), which shows that habitual fish oil supplementation is associated with a 13% lower risk of all-cause mortality, a 16% lower risk of CVD mortality, and a 7% lower risk of CVD events among the general population.
Such information could lead to better shared decision-making regarding the use of fish oil supplements.
Bibliography:
Show MoreChen et al. Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJMED 2024;3:e000...
Chen et al. (2024) find that fish oil supplementation exerts several adverse effects in “healthy” users without baseline history of cardiovascular disease (CVD). In particular, the authors note that the risk of conversion from healthy status to atrial fibrillation (HR 1.13 p<0.001) and stroke (HR 1.05, p=0.05) were increased in healthy users without a pre-existent diagnosis of CVD (1).
While we applaud the authors for their research, we note several major flaws with their results that warrant further exploration.
First, the authors claim that stroke risk is increased in “healthy” users of fish oil. The definition of “healthy” user is heterogenous and misleading as the authors use a mix of electronic medical records and self-reported survey data to derive this definition. Further, fish oil use is not a randomized allocation. Consequently, fish oil users may have some indication for elevated CVD or stroke risk at baseline that prompts its consumption, which would indicate reverse causation. Therefore, the reliability of survey data for such indications is weak at best and cannot be used to extrapolate a truly CVD risk-free state. Moreover, although hypertension is the most significant modifiable risk factor for stroke, family history, inflammatory parameters, and polygenic risk scores may inform personalized CVD risk to a similar or greater extent, and were not adequately controlled in this analysis (2).
Second, the reports of stroke risk in healthy...
Show MoreIn Chen and colleagues’ observational report of use of fish oil supplements (FOS) and cardiovascular outcomes in the UK Biobank study (1), their stated conclusions suggest equipoise of observed relationships, i.e., relatively balanced risks vs. benefits. However, only one adverse association achieved statistical significance - onset of atrial fibrillation (AF) - while five protective associations did so, including onset of heart failure, mortality after heart failure, and (after onset of AF) risk of myocardial infarction, major adverse cardiovascular events, and death (Figure 1). While important, AF also remains a relatively less severe clinical outcome compared with heart failure and myocardial infarction. These new findings provide evidence to support general cardiovascular benefits of FOS use.
Ten prior studies - all from UK Biobank - have assessed observational relationships between FOS use and health conditions, ranging from fractures to liver cancer to CV disease to dementia to death (2-11). Across these, 18 relationships were both statistically significant and favorable for FOS users (including total mortality); and only 1 showed an adverse association (AF). Thus, consistent with the new report, the overall findings from the UK Biobank support important net health benefits of FOS use.
These observational studies have advantages of large, fairly generalizable populations and extended follow-up, allowing an assessment of potential long-term effects of...
Show MoreIn a recent publication, Chen et al. (1) show the potential negative effects of fish oil supplementation on incident atrial fibrillation (AF) and subsequent effects on cardiovascular events (CVE) and death in the UK Biobank. Even though the authors start with a well-defined question (i.e., risk of AF had everyone taken taking fish oil supplements for primary prevention vs. had they not), moving into the conclusion, they state: “Regular use of fish oil supplements […] could be beneficial for progression of cardiovascular disease from AF to major adverse CVE, and from AF to death”. We reflect on the aspects for which the study was not fit for this conclusion.
To illustrate the point, imagine that the authors had instead conducted a randomized trial to estimate their well-defined question – ‘Does fish oil supplementation reduce the incidence of AF and CVE?’. The investigators would recruit a population free of AF and CVE at baseline, and randomly assign them to regular fish oil supplement or placebo. Follow-up would then start until AF, CVE, or death. The investigators would determine if the intervention prevented AF, CVE or death by counting the number of cases at the end of the follow-up period, or by comparing survival curves.
Show MoreNow suppose the investigators of this trial conduct a secondary analysis, among those who develop AF during the first half of the follow up period. Among these individuals, the investigators then count the incidence of further CVE and st...
The umbrella review by the research team of Szczerba and colleagues offers a perspective of the potential of scrutinizing the utility of personalized nutrition for impacting type 2 diabetes management and is a contribution to the field. By synthesizing evidence from RCTs that lasted at least 12 weeks, the authors not only attempt to offer a robust assessment of various dietary interventions, but also employ strict methodological rigor through the re-calibration of meta-analyses and application of the GRADEpro approach.
One of the work’s strengths lies in its comprehensive scope, meticulously evaluating the certainty of evidence and identifying gaps for future research. This approach not only signals the reliability of findings due to rigor and quality of studies, but also shows a way to scrutinize the potential of personalized nutrition strategies.
Although authors state and acknowledge limitations, such as the exclusion of the most recent RCTs and the absence of subgroup and sensitivity analyses, which could refine the study’s conclusions further, the study stands out for its dedication to precision and methodological excellence and voluminous work, offering valuable insights into the complex relationship between diet and type 2 diabetes management.
Findings underscore the importance of personalized nutrition in managing cardiometabolic health, exploring new standards for future studies in this vital area. It’s a significant step forward in our unders...
Show MoreWe address points made in the two letters written by Rice, Lambert, and their colleagues.
Point 1: Unrealistic premise / Confuses clinical researchers and the public.
We show that polygenic risk scores are poor predictors of disease whether used alone (as they are in direct-to-consumer genetic tests and have been in previous publications), or in combination with other risk factors [1]. If they perform poorly on their own, as we show, they cannot have useful incremental value when used with other risk factors, as we also show. Use of the appropriate metrics, as we employed in our paper, clarifies the position; it is the use of inappropriate metrics that causes needless confusion.
Point 2: Multifactorial models incorporating PRS improve risk prediction and are cost effective.
Show MoreAny potential screening marker that offers negligible screening performance when used alone or in a risk model cannot be cost effective. In the case of cardiovascular disease, risk factor models offer poor discrimination, with or without polygenic risk scores. Table 1 in our paper [1] shows that over 5000 people need to be genotyped to prevent one additional cardiovascular event. The cost-effectiveness analysis cited by the correspondents [2] did not compare the effectiveness of a multi risk factor model based strategy with alternatives such as an aged based strategy [3]. All risk models (with or without polygenic risk scores) add cost and impose a barrier to accessing ef...
Pages