Estimates of the Global Reduction in Liver Disease-related Mortality With Increased Coffee Consumption

An Analysis of the Global Burden of Disease Dataset

Paul Gow; Tim Spelman; Sarah Gardner; Margaret Hellard; Jessica Howell


Aliment Pharmacol Ther. 2020;52(7):1195-1203. 

In This Article


Liver disease is responsible for greater than one million deaths globally each year,[2,26] with the greatest burden in low-middle income countries due to viral hepatitis. Therefore, cost-effective, affordable and accessible initiatives to implement at an international population level to reduce this figure are urgently needed. The last two decades have seen a significant number of important publications that have consolidated the evidence that coffee drinking plays an important and powerful role in reducing the risk of death from cirrhosis and/or HCC.[19,27,28] This current paper estimates that nearly half a million lives may be able to be saved annually if moderate coffee drinking became embraced globally as a health initiative to reduce the risk of premature death from liver disease.

Projections of potential lives saved if a two cup/day coffee intake could be widely implemented varies markedly on a geographical basis. Our projections estimate that there would be few if any lives saved in Europe with this policy, as in this region the current mean coffee intake is already two or more cups daily.[24] However, outside of Europe rates of coffee drinking are relatively low. In South-East Asia alone we estimate that greater than 150 000 lives could be saved annually with the implementation of a 2 cup/day policy. Similar large numbers of lives could be saved in other global regions where coffee drinking is low and rates of underlying liver disease are relatively high. Notably, these countries most likely to derive liver-related mortality benefit on a population level are also coffee producers (South-East Asia, Sub-Saharan Africa, Latin America).[24]

Our data also estimate that greater than 720 000 lives could be saved annually if worldwide coffee intake increased to four or more cups daily. The paper by Setiawan[14] and other published series[25,27,29] have demonstrated a dose-response effect for coffee intake and protection from mortality from both cirrhosis and HCC. The magnitude of the reduction in mortality rates from both cirrhosis and HCC of course varies between case series. In this current paper we chose to use the risk reduction estimates generated from the paper by Setiawan and colleagues.[14] Their cohort study was prospective and included the largest data set with information for more than 162 000 individuals and was ethnically mixed. Although their data were derived from an exclusively US population with a median age of 60 years which may not be generalisable to the global population, the study included patients with diverse aetiology of liver disease. The relative risks derived from the study were also appropriately adjusted for confounding by gender, age, alcohol, smoking, BMI, race, education level and diabetes status.[14] Moreover, there has not been a systematic review and meta-analysis on the impact of coffee drinking on liver-related mortality to our knowledge from which we could derive a pooled RR to use for our study.

The data from Setiawan and colleagues[14] revealed that compared to noncoffee drinkers, those who drank 2–3 cups daily has a 38% reduction in HCC risk and a 46% reduction in risk of death from chronic liver disease (CLD). If individuals drank four or more cups daily the risk reduction for HCC was 41% and for death from CLD was 71%. Other cohort studies have reported risk reductions of similar magnitude.[5,15,27,29] In the Singapore Chinese Health study involving data from 63 257 middle aged Chinese, Goh and colleagues reported a risk reduction in death from nonviral CLD of 38% in those drinking two or more cups of coffee daily.[15] Lai et al[25] also reported mortality reduction in patients with liver disease from increased coffee consumption in a Norwegian adult population-based cohort, with an adjusted RR of 0.55 (95% CI 0.48–0.63) per extra cup consumed per capita, however this cohort was exclusively male smokers from Finland, therefore the generalisability of the results beyond this population is unclear. Klatsky et al also reported a dose-response effect between coffee intake and cirrhosis-related mortality in a US population-based cohort.[5] A further meta-analysis by Kennedy and colleagues[29] explored the impact of coffee consumption on the risk of cirrhosis, but not chronic liver disease-related mortality. They found that the risk of cirrhosis was reduced by 41% (pooled RR0.59, 95% CI 0.41–0.76). There have also been several case-control studies supporting the association between coffee drinking and improved liver-related survival,[9,30] though causation cannot be established from these studies. Collectively, these data have been considered compelling enough that for the first time the European Association for the Study of the Liver (EASL) Clinical Practice Guidelines for both hepatocellular carcinoma (2018)[31] and non-alcoholic fatty liver disease (2015)[32] recommend increased coffee consumption in people with chronic liver disease to reduce liver disease progression, HCC incidence and liver-related mortality. Moreover, coffee appears a safe intervention for people with chronic liver disease: as highlighted in the EASL HCC CPG, adverse events in people with chronic liver disease have not been identified in the studies conducted to date. An umbrella review of several meta-analyses by Poole and colleagues found minimal adverse events reported for usual consumption levels including in patients with chronic liver disease.[33]

However, the paper by Goh and colleagues[15] raises the important point that it is not clear if the benefits of coffee on liver mortality risk reduction are uniformly spread between all aetiologies of liver disease.[15,17] In this paper, the positive impact of coffee on liver related mortality was exclusively in those with nonviral hepatitis related liver disease. However, only two of 63,257 subjects had hepatitis C-related liver mortality and these were excluded from the analysis, therefore the majority of those with viral hepatitis were hepatitis B infected, suggesting that coffee may not have a benefit in those with hepatitis B. However, other papers have not supported this finding.

In the Setiawan paper,[14] the protective effect of coffee on both HCC risk reduction and CLD mortality was not specifically examined in patients with underlying viral hepatitis as a subgroup analysis. However, they did not find evidence to support interaction between coffee drinking and viral hepatitis aetiology of liver disease. The only other publication to specifically explore the benefits of coffee drinking in people with hepatitis B found coffee to be protective against HCC development, but did not examine CLD mortality as an endpoint.[17] Importantly in our study, there were still significant benefits associated with coffee drinking even after people living with hepatitis B were excluded from our global mortality estimates.

Although the data supporting the potential benefits of coffee for reducing HCC risk are compelling,[14,17] cancer registry data are unreliable in many parts of the world, particularly in low-middle income countries with high prevalence of HCC and low coffee intake, such as Africa and Asia.[34] Therefore, we did not explore the potential impact of increased coffee consumption on risk of HCC.

The mechanism by which coffee offers protection from liver disease has not been clearly demonstrated, but it appears that caffeine is not the protective chemical.[15] Of the many compounds in coffee, diterpenes and chlorogenic acids are the most studied in liver disease.[21] High coffee consumption has been correlated with improved insulin sensitivity, suggesting that coffee may exert protective effects through attenuating insulin-induced hepatic fibrosis and/or NAFLD as a co-factor in liver disease progression.[35] Coffee intake has also been associated with lower rates of biochemical inflammation in human cross-sectional studies and in animal studies, liver enzymes and pro-inflammatory cytokines IFN-γ and TNF-α were lower with coffee intake.[21]

This current paper and projections of lives saved has many potential biases. First, this is a hypothetical study that models the potential lives saved based on published risk reduction ratios from one study, in the absence of published risk reduction ratios derived from a systematic review and meta-analysis. Moreover, a common limitation of cohort studies (including the one by Setiawan and colleagues) is that while evaluation of coffee consumption (exposure) occurred prior to the outcome of interest, the impact of severe chronic liver disease on coffee intake has not been taken into consideration, therefore reverse causation cannot be excluded. The paper by Setiawan and colleagues was able to exclude prior or current history of HCC at baseline, however, detail of underlying liver disease and severity at study entry was limited. Overall, the ecological study design means derived data are population-based and cannot be extrapolated to the individual—this is ecological fallacy. The predicted numbers of liver-related deaths averted by the modelling used here have wide ranges from best case to worst case scenario. Although we project that over 450 000 lives may be saved annually with more than 2 cups of coffee daily, this number may be as high as 630 000 or as low as 169 000 based on our sensitivity analysis. Similar wide confidence intervals are also predicted for greater than 4 cups daily and with respect to lives saved from liver-related mortality reduction. Although the racial spectrum included in the Setiawan paper was quite broad the numbers of Asian and African (non-American) participants were negligible. Whether the estimates of protection from liver mortality reported from the Setiawan paper can be predicted to all ethnic groups remains unclear. The Setiawan data were also developed from a population of people living in the USA, where there are high background rates of obesity, non-alcoholic fatty liver disease and alcohol use[36]—whether these estimates can be broadened to communities in other counties with different liver disease risk factors also remains unanswered. Although adjustment of the models has been made to account for obvious confounding and interaction, residual confounders for the apparent association between coffee intake and liver related mortality need careful exclusion. For example, wealth was measured at baseline in the Setiawan paper but often changes over time, which may have impacted the results through unmeasured factors associated with coffee consumption and liver-related mortality. Data on the underlying mean coffee intake in many counties was also not available. In this situation we presumed the coffee intake to be less than 2 cups daily. This may have led to an overestimate of the predicted effects of coffee intake on mortality in these populations. However, when a sensitivity analysis was performed to instead assume that countries without available coffee import and export data have higher coffee consumption (four or more cups per day), there was still over 400,000 deaths averted by more than two cups of coffee consumed per capita per day and over 700,000 lives saved if four or more cups of coffee were consumed per capita per day. Similarly, some data suggest that only those aged over 57 years may derive mortality benefit from increased coffee consumption[25] and chronic disease related mortality increases with increasing age. While data on the duration of consumption needed to derive survival benefits from coffee are unavailable, we performed a further sensitivity analysis to estimate the number of lives potentially saved in 2016 with increased coffee consumption, restricted to those aged 55–79, which showed that there were still hundreds of thousands of liver-related deaths that could be averted with increased coffee consumption. We therefore conclude that even with adjustments for confounding and assumptions, an increase in coffee intake has the potential to save hundreds of thousands of lives on a global scale. As liver related mortality data are of poor quality in many counties due to inaccuracy of recording cause of death data, it is likely that our data may under report the benefits of coffee as many of the counties with limited quality death data are in regions with high rates of viral hepatitis.

Despite these caveats, this population-based ecological study has demonstrated the potential positive impact of global increased coffee consumption on liver related mortality. Our results suggest that increased coffee drinking may reduce worldwide liver disease related mortality. However, to date the key reported studies have been observational cohorts and until a mortality benefit is observed in randomised controlled trials enthusiasm to embrace coffee consumption policy should be tempered. Our findings suggest further research into the protective mechanisms of coffee in liver disease and data from a randomised controlled trial would be beneficial, particularly given its low cost, wide accessibility and good safety profile. Public health intervention studies are also warranted to explore how best to deliver the potential impact that coffee could have in terms of global reduction in the burden of liver disease.