Nasheen Naidoo, Cynthia Chen, Salome A Rebello, Karl Speer, E Shyong Tai, Jeanette Lee, Sandra Buchmann, Isabelle Koelling-Speer, Rob M van Dam
8 min
Certain coffee diterpenes, primarily cafestol and kahweol, are known to increase serum cholesterol and triglyceride levels. These compounds are lipid-soluble and are typically retained by paper filters. However, their concentration varies significantly based on how coffee is prepared. While Western drip-filtered coffee contains minimal diterpenes, unfiltered methods—such as Scandinavian boiled or French press—result in much higher concentrations. This study investigated whether traditional Asian coffee preparation methods, such as the Singaporean 'sock' method, the Indian 'filter' method, and the Indonesian 'tubruk' (unfiltered) method, present similar cardiovascular risks.
The researchers analyzed coffee samples from Singapore, India, and Indonesia. They found that the traditional 'sock' method used in Singapore and the metal mesh 'filter' method used in India resulted in very low levels of cafestol, comparable to standard paper-filtered coffee. These levels are unlikely to impact blood lipid concentrations. In contrast, Indonesian coffee samples showed higher variability. While some Indonesian methods also yielded low diterpene levels, their unfiltered 'tubruk' method contained high amounts of cafestol, similar to European boiled or French press coffee, which is associated with a predicted increase in serum lipids.
In a cross-sectional study of 3,000 multi-ethnic residents in Singapore, the researchers examined whether coffee consumption was associated with lipid profiles (LDL-cholesterol, triglycerides, and total cholesterol). After adjusting for major confounders such as body mass index (BMI), smoking, education, and diet, they found no statistically significant association between coffee intake and serum lipid levels. This result supports the finding that the dominant coffee preparation methods in Singapore do not pose a meaningful risk for elevating cholesterol.
Understanding regional differences in coffee preparation is crucial for cardiovascular risk assessment. While filtered coffee is generally considered neutral regarding lipid levels, the study highlights that popular traditional brewing methods in some regions (like the Indonesian unfiltered method) may still contribute to elevated cholesterol if consumed in high quantities. This provides actionable information for local public health guidance in Asia regarding coffee preparation and cardiovascular health.
BACKGROUND: To measure the content of cholesterol-raising diterpenes in coffee sold at the retailer level in Singapore, Indonesia and India and to determine the relationship of coffee consumption with lipid levels in a population-based study in Singapore. METHODS: Survey and cross-sectional study in local coffee shops in Singapore, Indonesia and India to measure the diterpene content in coffee, and a population-based study in Singapore to examine the relationship of coffee consumption and blood lipid levels. Interviews and coffee samples (n=27) were collected from coffee shops in Singapore, Indonesia and India. In addition, 3000 men and women who were Chinese, Malay, and Indian residents of Singapore participated in a cross-sectional study. RESULTS AND DISCUSSION: The traditional 'sock' method of coffee preparation used in Singapore resulted in cafestol concentrations comparable to European paper drip filtered coffee (mean 0.09±SD 0.064 mg/cup). This amount would result in negligible predicted increases in serum cholesterol and triglyceride concentrations. Similarly low amounts of cafestol were found in Indian 'filter' coffee that used a metal mesh filter (0.05±0.05 mg/cup). Coffee samples from Indonesia using the 'sock' method (0.85±0.41 mg/cup) or a metal mesh filter (0.98 mg/cup) contained higher amounts of cafestol comparable to espresso coffee. Unfiltered coffee from Indonesia contained an amount of cafestol (4.43 mg/cup) similar to Scandinavian boiled, Turkish and French press coffee with substantial predicted increases in serum cholesterol (0.33 mmol/l) and triglycerides (0.20 mmol/l) concentrations for consumption of 5 cups per day. In the Singaporean population, higher coffee consumption was not substantially associated with serum lipid concentrations after adjustment for potential confounders [LDL-cholesterol: 3.07 (95% confidence interval 2.97-3.18) for <1 cup/week versus 3.12 (2.99-3.26) for ≥3 cups/day; p trend 0.12]. CONCLUSIONS: Based on the low levels of diterpenes found in traditionally prepared coffee consumed in Singapore and India, coffee consumption in these countries does not appear to be a risk factor for elevation of serum cholesterol, whereas samples tested from Indonesia showed mixed results depending on the type of preparation method used.
Sam: That captures real retail practice, rather than a perfectly standardized laboratory brew. But it also means the cup reflects several choices at once.
Alex: Yes: the filter, the grounds, the recipe, and the brewing process. The researchers freeze-dried the samples and used chromatography, a laboratory method that separates and measures compounds. They quantified both cholesterol-raising compounds in the finished coffee.
Sam: What did they find in the Singapore samples, where the fabric sock was the usual method?
Alex: Average cafestol content was about 0.09 milligrams per cup. That was comparable to earlier measurements of European paper-filtered coffee. The South Indian metal-filter samples also contained very little.
Sam: Metal filters are not all identical, though. Did the Indonesian samples fit that low-content pattern?
Alex: They did not. Indonesian coffee made with a sock or metal mesh contained amounts comparable to espresso in the earlier European research. The unfiltered Indonesian sample contained 4.43 milligrams of cafestol per cup, much higher than the Singapore sock-filtered average.
Sam: So “filtered” was not a sufficient description. Did they establish why the same broad methods gave different results?
Alex: They proposed explanations rather than testing them. The material’s pore size could matter, as could repeated filtration and how tightly the grounds were packed. Those possibilities make the preparation details worth recording, but they are not demonstrated mechanisms from this study.
Sam: The amount in a cup is useful chemistry. How did they translate that into a claim about cholesterol?
Alex: They used predictions based on earlier human trials of the compounds. For the unfiltered Indonesian sample, drinking five cups daily was predicted to raise total cholesterol by about a third of a millimole per litre. That is a predicted blood concentration change, not an observed change in these customers.
Sam: And that distinction matters: they did not assign people these coffees and watch their cholesterol rise.
Alex: No. The measured result was the coffee’s chemical content. The predicted rise came from prior trial evidence, using the standardized cup size in this study. For the low-content Singapore and South Indian samples, the same approach predicted negligible increases.
Sam: Let us move to the people, then. What was the independent check on whether this chemistry mattered in Singapore?
Alex: The authors analysed coffee intake and fasting blood samples from 3,000 Chinese, Malay, and Indian residents. This was a cross-sectional analysis: coffee habits and blood fats were assessed without establishing which came first. Blood fats included cholesterol and triglycerides, another type of fat circulating in blood.
Sam: Coffee drinking often comes with other habits. Did the heavier coffee drinkers differ in ways that could also affect those measurements?
Alex: They did. Higher consumption went with more smoking, higher body mass index, and differences in education and diet, among other characteristics. With adjustment only for age, sex, and ethnicity, higher coffee intake was associated with higher levels of several blood fats.
Sam: So the initial association could have been a coffee effect, or it could have reflected who drank more coffee. What happened with fuller adjustment?
Alex: The associations were no longer statistically significant. Body mass index weakened the triglyceride association. For total cholesterol and low-density lipoprotein cholesterol—the cholesterol fraction abbreviated as LDL—the combination of body mass index, education, and smoking weakened the associations.
Sam: That supports the low-content measurements, but a non-significant association is not proof of no effect. How far do the authors take it?
Alex: They conclude that traditionally prepared coffee in Singapore and South India does not appear to be an important risk factor for elevated blood fats. But only Singapore had the population analysis. The South Indian conclusion rests on the sampled coffee’s low compound content, not a corresponding population study.
Sam: There is another link I would want checked. Did they know whether each Singapore participant actually drank sock-filtered coffee?
Alex: They did not assess individual preparation methods. That is a central limitation: the chemical samples and population observations are separate strands of evidence, not matched cups and drinkers. Their consistency is reassuring, but it cannot establish a preparation-specific effect in individuals.
Sam: Could people with high cholesterol already have reduced their coffee intake, making the association look weaker?
Alex: The authors excluded participants with known abnormal blood fats, partly to reduce that possibility. They also excluded several other conditions that could affect the analysis. Still, the cross-sectional design cannot establish temporal order, and measurement error or residual confounding could conceal small effects.
Sam: And the Indonesian results need particular restraint, because the evidence is a small collection of coffees rather than a national survey.
Alex: The authors explicitly caution about the few Indonesian samples per preparation method. The Singapore and Indian measurements were more consistent. My reading is that the study supports preparation-specific investigation, not a country-wide ranking of coffee safety.
Sam: Who should read the full document, and where would you send them first?
Alex: Nutrition epidemiologists and researchers studying coffee preparation or cardiovascular prevention should read it. Start with the preparation-method results, especially the contrast between Singaporean and Indonesian filtered coffee. Then read the adjusted population analysis and the discussion of its limitations together.
Sam: And what should everyone else remember without turning these findings into a broader health claim?
Alex: This study concerns blood fats, not coffee’s overall health effects. Its practical lesson is narrower: when studying coffee and cholesterol, count the brewing details, not just the cups.
Sam: A useful distinction to keep with your next cup.