Coffee: the paper filter is the health intervention

Unfiltered coffee raises cholesterol and filtered coffee barely does, and the difference is a paper filter trapping two compounds called cafestol and kahweol. Across 14 randomised trials the relationship was dose-dependent, stronger in people who already had high lipids, and largely absent with filtered brews. Put on a scale with every other food tested for its effect on LDL, unfiltered coffee is one of the few that pushes the number up, by about as much as oats and plant sterols pull it down. French press, moka pot, Turkish and boiled coffee are all unfiltered.

The numbers, per 100 g

What it does to your health

• Unfiltered coffee raises cholesterol and filtered coffee barely does, and the paper filter is the whole difference. — 14 published trials of coffee consumption (n = 14 trials) Evidence A source
  A dose-response relation was found between coffee consumption and both total and LDL cholesterol (p<0.01); increases were larger in people with hyperlipidaemia and in trials of caffeinated or boiled coffee; trials using filtered coffee showed very little increase

Safety, limits and interactions

• Coffee raises homocysteine by half a micromole, a marker whose importance was overturned years ago. — clinical trials of coffee supplementation in adults Evidence A source
  Homocysteine rose by 0.55 micromol/L (95% CI 0.17 to 0.93, p = 0.005, I2 0%); leptin was unchanged (+1.34 ng/mL, -0.78 to 3.45, p = 0.21, I2 0%); sensitivity analysis was robust for both

What people believe that the data does not support

• Ranked against every other food tested for its effect on LDL cholesterol, unfiltered coffee is one of the few that pushes it up. — 37 guidelines, 108 systematic reviews and 20 randomised controlled trials (n = 165 documents) Evidence A source
  With high evidence, foods high in unsaturated fat, with added plant sterols or stanols, and high in soluble fibre lowered LDL by at least 0.20 to 0.40 mmol/L, while unfiltered coffee caused a moderate to large increase; soy protein, tomatoes, flaxseed and almonds caused small reductions; with moderate evidence avocado and turmeric caused moderate to large reductions
• Europe refused all twelve health claims filed for coffee, the antioxidant one included. — — Evidence E source
  Twelve register entries name coffee and all twelve carry the status Non-authorised; the example claims that antioxidants in coffee help protect cells against free radicals

Who this works differently for

• The Nordic review written for dietary guidelines found moderate coffee harmless long term, and high caffeine in pregnancy not. — published systematic reviews on coffee, tea, sugar-sweetened beverages and low-calorie sweetened beverages Evidence E source
  Moderate coffee and tea consumption do not have long-term adverse health effects; favourable long-term associations for coffee cover endometrial cancer, liver cancer, type 2 diabetes and cardiovascular death; randomised trials show diterpene-rich brews such as boiled coffee raise serum cholesterol; high caffeine intake in pregnancy is associated with higher risk of pregnancy loss, preterm birth and low birth weight

What is still unknown

• Even filtered coffee moved cholesterol when people stopped and started it four times over. — 121 healthy nonsmoking men and women aged 29 to 65 (n = 121) Evidence B source
  The two abstention periods lowered serum cholesterol by 0.22 mmol/L (95% CI -0.31 to -0.13) and 0.36 mmol/L (-0.46 to -0.26); 600 mL a day of filtered coffee raised it by 0.25 mmol/L (0.15 to 0.36) and 0.15 mmol/L (0.04 to 0.26)

The bottom line

Filtered coffee is not quite neutral either. In 121 people who stopped and started four times over, 600 mL a day raised serum cholesterol by 0.15 to 0.25 mmol/L and stopping lowered it by 0.22 to 0.36. Against that, the review written for Nordic dietary guidelines lists lower risks of endometrial cancer, liver cancer, type 2 diabetes and cardiovascular death with moderate coffee, all of it observational. The one line in that review to act on concerns pregnancy: high caffeine intake is associated with pregnancy loss, preterm birth and low birth weight. Coffee also raises homocysteine by half a micromole, a marker whose importance collapsed once trials showed that lowering it changes nothing.

Sources

1. cof-dr-01 · meta-analysis of randomised controlled trials
   Consumption of unfiltered, but not filtered, coffee increases serum levels of total and LDL cholesterol.
   Am J Epidemiol, 2001 · checked 2026-09-22
2. cof-dr-04 · systematic review and meta-analysis of clinical trials
   The meta-analysis demonstrated that coffee administration significantly increases homocysteine levels (WMD: 0.55 μmol/L, 95% CI: 0.17, 0.93, p = 0.005, I2 = 0%) but has no significant changes in leptin concentrations (WMD: 1.34 ng/mL, 95% CI: -0.78, 3.45, p = 0.21, I2 = 0%).
   Curr Pharm Des, 2023 · checked 2026-09-22
3. cof-dr-02 · systematic review of systematic reviews and randomised controlled trials
   With high evidence, foods high in unsaturated and low in saturated and trans fatty acids (e.g. rapeseed/canola oil), with added plant sterols/stanols, and high in soluble fiber (e.g. oats, barley, and psyllium) caused at least moderate (i.e. 0.20-0.40 mmol/L) reductions in LDL cholesterol. Unfiltered coffee caused a moderate to large increase.
   Nutr Metab Cardiovasc Dis, 2021 · checked 2026-09-22
4. cof-dr-06 · regulatory register entry
   POL-HC-6618 Antioxidants in coffee Antioxidants in coffee helps protect our cells against free radicals. Non-authorised
   EU Register on nutrition and health claims, заява POL-HC-6618, знімок 2026-09-21 · checked 2026-09-22
5. cof-dr-03 · scoping review for dietary guidelines
   The long-term favorable effects of coffee consumption are related to reduced risk of endometrial and liver cancer, type 2 diabetes, and cardiovascular deaths. However, results from randomized controlled trials (RCTs) suggest that coffee brews that are rich in diterpenes, such as boiled coffee, increase serum cholesterol concentrations. High caffeine intake in pregnancy is associated with higher risk of pregnancy loss, preterm birth, and low birth weight.
   Food Nutr Res, 2024 · checked 2026-09-22
6. cof-dr-05 · prospective controlled crossover study
   The two coffee abstention periods were associated with a decline in serum cholesterol of 0.22 mmol/l (95% CI -0.31, -0.13) and 0.36 mmol/l (95% CI -0.46, -0.26), respectively.
   Eur J Clin Nutr, 2003 · checked 2026-09-22