Omega-3: “You need fish”

Forms and dietary sources

The three principal dietary omega-3 fatty acids are ALA, EPA, and DHA (NIH ODS, 2025). Flaxseed and flaxseed oil, chia seeds, walnuts, soybean oil, and canola oil provide ALA, while long-chain EPA and DHA are found in fish and in supplements derived from fish or algae (NIH ODS, 2025). For people who do not eat fish, algae-derived products provide a direct dietary source of DHA, and some formulations also contain EPA (NIH ODS, 2025). Fish is therefore one source of long-chain omega-3 fatty acids, not a biological requirement for obtaining them (NIH ODS, 2025).

Conversion of ALA

Humans can convert ALA to EPA and then to DHA, although the pathway is not highly efficient (Burdge and Calder, 2005; NIH ODS, 2025). A review of tracer and feeding studies found that increasing ALA raises EPA in several blood and tissue pools, while DHA generally does not rise and can decline in some pools at high ALA intakes (Burdge and Calder, 2005). Conversion was greater in women than in men in the reviewed tracer studies, indicating that sex and hormonal factors contribute to variability (Burdge and Calder, 2005).

A review focused on vegetarian diets similarly concluded that ALA conversion is slow, inefficient, and influenced by genetics, sex, age, and dietary composition (Saunders et al., 2013). It reported lower EPA and DHA biomarkers in vegetarians than in non-vegetarians while noting that the clinical significance of those differences was unknown (Saunders et al., 2013). A low biomarker concentration should therefore not be described automatically as a clinical deficiency, and the absence of a dietary EPA or DHA source should not be described as proof of harm (Saunders et al., 2013; NIH ODS, 2025).

Evidence for algal DHA

Algae are the primary producers of DHA in the marine food chain and can be used directly in supplements (NIH ODS, 2025). A systematic review identified four randomized trials and two prospective cohort studies in vegetarian populations (Craddock et al., 2017). Every included study reported increased DHA concentrations or omega-3 indices with algal DHA, although the small evidence base did not establish a clear time-response or dose-response relationship (Craddock et al., 2017). This supports the narrower conclusion that algal supplementation changes biomarkers; it does not by itself establish a reduction in cardiovascular events, cognitive decline, or other clinical outcomes (Craddock et al., 2017).

The distinction between biomarkers and clinical outcomes is material (Craddock et al., 2017; Abdelhamid et al., 2020). A Cochrane review of long-chain and plant-derived omega-3 interventions found small or uncertain effects across cardiovascular outcomes, with clearer evidence for lowering triglycerides than for preventing major events (Abdelhamid et al., 2020). Evidence that a supplement increases blood DHA is therefore stronger than evidence that every person who avoids fish will obtain a measurable clinical benefit from taking it (Craddock et al., 2017; Abdelhamid et al., 2020).

Pregnancy and guideline variation

Some professional guidance gives special attention to long-chain omega-3 intake during pregnancy or in people with little or no fish intake (Agnoli et al., 2017; Axelsson et al., 2024). The Italian Society of Human Nutrition position paper advises algae-based supplementation for pregnant and breastfeeding vegetarians and for young children because ALA conversion is limited (Agnoli et al., 2017). This is not a universal guideline position: a Danish obstetric guideline concluded that it found no evidence supporting DHA supplementation during vegetarian or vegan pregnancy (Axelsson et al., 2024).

Accordingly, the inherited claim that guidelines often suggest considering algae-derived DHA or EPA is supportable, particularly for pregnancy and low-fish diets, but it should be presented as consideration rather than a universal requirement (Agnoli et al., 2017; Axelsson et al., 2024). Evidence quality is moderate for conversion and biomarker effects and lower or mixed for broad clinical-outcome claims (Burdge and Calder, 2005; Craddock et al., 2017; Abdelhamid et al., 2020).