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

Disclosures

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

In This Article

Methods

Data

This descriptive cross-sectional study used the Global Burden of Disease (GBD) 2016 adult >15 years dataset (both sexes)[22] to illustrate the potential impact of increasing per capita coffee consumption to more than two and also more than four cups per day on the number of liver-related deaths globally. Annual estimates or absolute number of deaths (both genders) due to all liver disease and hepatitis B specifically were extracted for every country in the GBD 2016 dataset. We restricted the dataset to those aged 15 years or older, as those under 15 years are unlikely to be consuming multiple cups of coffee. For regional sub-analyses, World Health Organisation (WHO) regional definitions were used.[23] To estimate annual per capita coffee consumption for each country, public-available data were used from the International Coffee Organisation, an international commercial and market interest organisation that records annual estimates of national per capita coffee consumption (kilograms) based on national coffee bean import and export data.[24] Per capita estimates (kilograms/person/year) were converted to grams/day and then into a categorical estimate of number of cups consumed per day, based upon standardised measures of coffee content (grams) per cup used in previous publications (7 grams ground coffee per cup).[14,16,19,24] For countries where coffee consumption data were not available, we ascribed a per capita coffee intake of less than two cups per day, a conservative estimate based upon the absence of recorded import and export markets.

Statistical Analyses

To estimate the impact of increased coffee consumption on liver-related mortality in 2016, we used the risk reduction ratios published by Setiawan and colleagues.[14] These were chosen because they were derived from the largest prospective study to date on the impact of coffee consumption on liver-related mortality, drawn from the US Multiethnic Cohort Study which included 162 022 people from multiple ethnicities and with diverse aetiology of liver disease; and adjusted for key confounders including BMI, obesity, alcohol, age, gender and smoking. In this study, past or current HCC was excluded at baseline and a detailed validated analysis of coffee consumption was conducted prior to measurement of the outcome of liver-related mortality, minimising temporal ambiguity and recall bias. Lag time analysis was also performed to minimise (but not remove) the impact of reverse causation. To estimate the predicted effect on global liver-related mortality in 2016 if per capita coffee consumption were increased globally to more than two cups per day, or four or more cups per day, we applied the risk reduction ratios for liver-related survival calculated by Setiawan and colleagues[14] for drinking more than two cups of coffee per day (RR 0.54, 95% CI 0.42–0.69) and for drinking four or more cups of coffee per day (RR 0.29, 95% CI 0.17–0.50) to the GBD 2016 dataset (both genders, adults > 15 years). We then calculated the predicted number of liver-related deaths and liver related mortality rate for each country and each WHO region in 2016, if per capita coffee intake had been greater than two cups per day, or four or more cups per day.

Only countries with per capita coffee intake less than two cups per day in 2016 would derive a survival benefit from increasing coffee intake to more than 2 cups of coffee per day. Therefore, for countries where per capita coffee consumption is already more than two cups per day, a risk reduction ratio of 1.00 (no change) was used for the analysis assessing impact of increased coffee consumption to more than two cups of coffee per day on mortality. Similarly, for countries with per capita coffee consumption of four or more cups of coffee per day at baseline, an RR of 1.0 (no change) was used for the analysis assessing impact of increased coffee consumption on mortality.

The mortality benefit from coffee intake per cup was not perfectly linear in the Setiawan et al paper,[14] ie: the risk reduction from four cups of coffee or more per day (RR 0.29) was not twice the risk reduction of two or more cups of coffee per day (RR 0.54). Therefore, countries with baseline per capita coffee consumption of more than two cups of coffee per day would have a lower expected survival benefit from increasing their coffee consumption to four or more cups of coffee per day, compared with countries with baseline per capita coffee consumption of less than two cups of coffee per day. We accounted for this by using a risk ratio of 0.75 (1.00−0.25, the difference in RR for two (RR 0.54) and four cups (RR 0.29) of coffee per day) to calculate the potential number of lives saved by increasing coffee consumption from two to four or more cups per day.

Finally, to calculate the number of predicted lives saved in each country if their baseline per capita coffee intake was increased to more than two cups, or four or more cups of coffee per day, we subtracted the number of predicted deaths from the estimated number of actual deaths in 2016 in the GBD dataset.

For WHO regional analyses, the median per capita coffee intake (cups per day) was estimated using country per capita consumption levels in each region. A baseline coffee intake of two or more cups per day for the Western Europe region was assumed, whereas for all other regions the baseline median per capita coffee intake was less than two cups of coffee per day.

Sensitivity Analyses

A sensitivity analysis was performed to estimate the upper and lower number of predicted lives saved with increased coffee intake, with the lower estimate for predicted lives saved calculated using the lower limit of the 95% CI for number of deaths and the lower limit of the 95% CI for the risk reduction ratio; conversely an upper limit was obtained using the upper limit of the 95% CI for number of liver-related deaths and the upper limit of the 95% CI for the risk ratio. As some studies have suggested that there may not be a strong protective effect of coffee among people with hepatitis B- related liver disease,[15] a second calculation was performed excluding all HBV related liver deaths from the analysis. We also performed additional sensitivity analyses to test our assumptions, restricting the dataset in turn to only those aged 55–79 years; and presuming that coffee intake in countries with no reported import and export coffee data already drink four or more cups per capita per day, therefore would not benefit from an increase in coffee consumption. All analyses were performed using Excel for Mac v 15.29 (Microsoft, CA USA) and Stata v14.1 (StataCorp, College Station, Texas USA).

processing....