Intended for healthcare professionals

CCBYNC Open access

Rapid response to:


Non-alcoholic fatty liver disease and risk of incident acute myocardial infarction and stroke: findings from matched cohort study of 18 million European adults

BMJ 2019; 367 doi: (Published 08 October 2019) Cite this as: BMJ 2019;367:l5367

Rapid Response:

Re: Non-alcoholic fatty liver disease and risk of incident acute myocardial infarction and stroke: findings from matched cohort study of 18 million European adults

We thank Professor Targher for his response and for his interest in our study. The response covers a number of points that, while covered in the Discussion of our paper, we feel would be important to reiterate to be fully objective and to robustly counter his concerns.

The findings of our study do not question the existing data that most patients with NAFLD die from cardiovascular disease (CVD). However, this is also true of many conditions and is not in itself a reason to consider NAFLD as a strong independent risk factor for CVD beyond its associated risk factors. Rather, and in the same way as people with pre-diabetes are at elevated CVD risk, our results strongly suggest that any excess CVD risk in patients with NAFLD can be explained by their associated risk characteristics; greater obesity, diabetes, dyslipidaemia, and higher blood pressure. In addition to the risk factors listed, a key design strength of our study is that we also matched non-exposed controls from the same practice site, thereby also adjusting for a measure of socioeconomic deprivation, a likely important confounder in many prior papers in this area.

While it may be true that we are unable to ascertain the veracity or origin of the recorded diagnoses, these were records extracted from source clinical systems where these data were being used by doctors for clinical decision making. Under the EU General Data Protection Regulations doctors also have a duty to ensure that the clinical data they hold on their patients are correct. In real life clinical practice, doctors can only make disease-relevant decisions for those patients who have been diagnosed with a condition. Thus, even if many more people might have NAFLD than the around 2% currently diagnosed, doctors can only make decisions on such identified individuals and this is the group we therefore studied. There are, yet, no strong arguments for population screening for NAFLD.

When we compared the clinical characteristics of our coded NAFLD/NASH and matched controls, we found levels of obesity, diabetes and hypertension in the NAFLD/NASH group were near identical to those in prior cohorts of patients with NAFLD detected and quite distinct from those in the controls. Therefore, we believe that those identified with NAFLD did indeed have excess liver fat since they have all the hallmarks of the condition. We accept a proportion in the controls will have had undiagnosed NAFLD but these were overwhelmingly diluted out by others without NAFLD: evidenced by the average control characteristics.

Any potential alcohol intake misclassification, as we carefully argued, would not explain our results given near identical findings for stroke and myocardial infarction – we refer readers to our detailed arguments in the paper. To eliminate confounding from alcohol-related liver disease, we excluded adults with alcoholic liver disease and those with a coded diagnosis of alcohol misuse. In addition, in real life clinical practice, doctors can only accept what patients are happy to report about their drinking habits.

Finally, we would respectfully reiterate that we are not saying that there is no association of NAFLD with incident cardiovascular outcomes. Rather, we stand by our conclusion that as in people with pre-diabetes, those diagnosed with NAFLD should not automatically be considered at high CVD risk. The existing body of literature is insufficient to recommend that patients’ CVD risk should be managed differently (e.g. use of statins) by virtue of a diagnostic code of NAFLD/NASH alone. Rather, clinicians should look for and attend to CVD risk factors, including undiagnosed diabetes, using established methods and risk scores. Of course, all should receive supportive behaviour and lifestyle advice with particular focus on diet for sustained weight loss and moderation of alcohol consumption.

Naveed Sattar and Will Alazawi on behalf of all authors

Competing interests: We are lead senior authors of the article being commented upon

15 October 2019
Naveed Sattar
Professor of Metabolic Medicine
William Alazawi
University of Glasgow
University of Glasgow, UK