org355: Coffee and death

It's become a truism that epidemiologic research about the effects of coffee or chocolate, will be reported in the mass media.

Here's something from The Guardian.

1. What is the study design?

2. What are the exposure / comparison groups? How were these data collected? Is this subject to measurement bias?

3. What is the outcome? How was this collected? Is this subject to measurement bias?

4. Explain why 'ethnicity' was part of the analysis? Would it be a problem if the ethnicity data was missing?

5. Explain what is the numerator and denominator for the calculation leading to the statement "29% lower risk of death".

6. Is an RCT necessary to demonstrate causation? Is this feasible?

Re: Coffee and death

by | SIH COLETTE - Tuesday, 31 May 2022, 1:16 PM

Hello Sujit,

As always, this was an interesting read. Here are my suggestions:

1. Study design: prospective cohort study

2. Exposure: Sugar-sweetened, artificially sweetened and unsweetened coffee
Outcome: All-cause, cancer-related, and CVD-related mortality
Comparison groups: Non-coffee drinkers in UK Biobank
Data collection method: Exposures were self-reported and extracted from Biobank data
Measurement bias: It would have been interesting to know how the type of sweetener used in the coffee was ascertained by Biobank and eventually classified by study investigators. Also, the frequency at which exposure data was collected as people could change the exposure status from coffee drinker to non-coffee drinker or change type of sweetener from day to day, week to week and year to year. The above could all have introduced non-differential misclassification of exposure status potentially biasing effect measures towards the null. There may also have been a social desirability bias with people having a tendency to misreport exposure data based on what is considered socially desirable

3. Outcome: All-cause, cancer-related, and CVD-related mortality
Data collection method: Objective measure using death certificates
Measurement bias: Could have occurred depending on how cause of death is recorded for example if a study participant with cancer also had COVID-19 and died, would the cause of death be recorded as COVID or cancer? This type of errors could result in fewer outcomes and lower the power/precision of the study. It would also be great to know if those assessing the outcomes were blinded to exposure status of the participant as this could introduce observer bias leading to an over-estimation of the measure of effect.

4. Ethnicity was considered as a potential confounder of the association between consuming coffee and mortality. Some ethnic groups may be more likely to drink tea (so less coffee) and vice versa. Mortality rates may also differ for different ethnic groups. Ethnicity does not lie on the causal pathway between coffee drinking and death. It would be a problem if ethnicity data is missing depending on the proportion of the missing data and if these missing data are differential by exposure status as it could introduce bias due to missing data and reduce sample size included in final models if missing data are simply excluded from analyses.

5. Numerator: hazard rate of death in unsweetened coffee drinkers consuming 2.5 to 4.5 cups per day
Denominator: hazard rate of death in non-coffee drinkers

6. Even though a cohort study aids in establishing temporality, exposed and unexposed may be very different at baseline on both known and potentially unknown confounders. An RCT provides a higher level of evidence of causation because in addition to establishing temporality, randomization ensures that investigators have exposure groups which are comparable and exchangeable on expectation except for the intervention received and therefore balanced on both known and unknown confounders (provided we have a large enough sample size). In addition, an RCT might be considered ethical as there is equipoise on this question even though the harmful effects of sugar on the health have been clearly established and may raise ethical concerns. However, an RCT may not be feasible for logistical and other reasons.

I look forward to reading your thoughts on this crude first shot.

Kind regards,
Colette

Re: Coffee and death

by | DOUGLAS OLCOTT - Tuesday, 31 May 2022, 8:37 PM

I don't have much to add to Collette's excellent response to Sujit's questions, but since my wife is of Greek and Italian ancestry and regularly drinks coffees from both of those traditions I can say that the study needed to specify not only whether the coffee was sweetened but how it was sweetened, i.e., whether the consumer added refined sugar to the unsweetened ground coffee and how much s/he added, or used artificial sweeteners or coffees that already had artificial sweeteners. For example, an Italian mocha coffee typically has powdered chocolate containing refined sugar added to it, and Greek/Turkish coffee is also usually prepared with refined sugar added.

The American Heart Association has sponsored many studies that show that excessive use of refined sugar leads to generally poorer health and increases the risk of heart disease, which is a leading cause of death. It has specified recommended maximum amounts of refined sugar use per day (grams or teaspoons) for adult men and women (different amounts, incidentally) and also for children (of different ages), though it is not clear from the Guardian article what age groups were included in the study. Other studies have shown that excessive consumption of refined sugar can lead to diabetes (Type 2), which can also be a cause of death. Use of refined sugar affects one's calorie intake, and lowering one's use of sugar is usually part of any weight-loss program.

I would therefore add a diagnosis of heart disease or diabetes as a possible cause of death to the description of the outcome, rather than specifying just "all causes" or "cancer-related".

P.S. Since Guardian articles have a limit on how often you can access them without subscribing and paying money, I would appreciate having the reference to the original Pub-Med publication included.

Re: Coffee and death

by | Sujit Rathod - Wednesday, 1 June 2022, 5:11 PM

Hi Douglas -

Try to google search the headline: Coffee drinkers may be at lower risk of early death, study suggests

and see if you get access.

Here is the original article.

And an accompanying editorial about sugar.

Re: Coffee and death

by | DOUGLAS OLCOTT - Wednesday, 1 June 2022, 9:18 PM

Hi Sujit,

Thanks for the reference to the original article and the accompanying editorial. Reading these I find that the study did examine the association between coffee drinking with and without sugar and heart disease. It used the abbreviation CVD for the latter, for cardiovascular disease. And by using the phrase "a spoonful of sugar" in the headline the editorial implies that the study considered refined sugar for the association. My only objection then would be with the definition of the outcome as "death from any cause" as well as death from heart disease or cancer. The specific cause of diabetes should have been considered, a common result of the over-consumption of refined sugar and a common health problem these days that needs to be highlighted.