Desserts: what the trials measured, and what they left out

According to the CDC, desserts and sweet snacks sit next to sweetened drinks as the leading sources of added sugar in the American diet. The trials that pull desserts apart from other sugary foods are few, and they rarely test a slice of cake. The best-studied result, though still uncertain, is this one. Across 47 controlled feeding trials, when sugars from sweets and desserts replaced other carbohydrate at the same calories, blood uric acid (a waste product the body makes when it breaks food down) went up, in the same direction as with sugary drinks. The rise was small. For all sugars pooled it was 0.16 mg/dL, and the review rated certainty for sweets and desserts moderate to very low. Everything else is thinner. The review of 64 trials that saw desserts lower IL-6, a signalling protein the body makes during inflammation, was looking at dark chocolate, which tells you about dark chocolate and very little about pudding. Liver fat rose in trials where extra calories came from added sugars in general, and for sweets and desserts the same review could not say either way. The clearest practical number comes from the plate itself. When 34 adults ate a familiar dessert freely after lunch, the most energy-dense version added about 240 kcal compared with the lighter ones.

The numbers, per 100 g

What a real portion looks like

• In a randomised crossover experiment in 34 adults eating a familiar dessert freely at one lunch, the high energy-density version (over 3 kcal/g) added about 240 kcal compared with lower-density versions. — 34 adults, 62% women, mean age 37 (n = 34) Evidence B source
  About +240 kcal with high energy-density dessert (p < 0.001)
• The CDC names desserts and sweet snacks, together with sugar-sweetened beverages, as the leading sources of added sugars in the US diet. — US population Evidence E source
  Desserts among the leading sources of added sugars

What your body absorbs

• In MRI measurements in 4 volunteers, gastric volume fell at 0.33 mL/min with an 8% fat chocolate dessert and rose at 0.09 mL/min with chocolate bars, and 8 patients allergic to peanut or hazelnut were challenged in the same two matrices. — 4 volunteers for gastric emptying, 8 patients allergic to peanut or hazelnut (n = 12) Evidence D source
  Gastric volume fell 0.33 mL/min with dessert and rose 0.09 mL/min with bars in 4 volunteers, no reaction threshold result in the quote

What it does to your health

• A meta-analysis of 47 controlled feeding trials (N = 2763) found that sweets and desserts, like sugar-sweetened beverages, raised blood uric acid when their sugars replaced other carbohydrate at equal energy. — 47 trials of at least 7 days, mostly healthy adults of mixed weight (n = 2763) Evidence A source
  Total fructose-containing sugars +0.16 mg/dL uric acid in substitution trials, sweets and desserts increased it
• A meta-analysis of 64 controlled feeding trials (n = 4094) found no effect of most sugar sources on inflammation markers, and in added-energy trials the sweets and desserts stratum (dark chocolate) lowered IL-6. — 64 trials, median 6 weeks, healthy or overweight adults (n = 4094) Evidence A source
  Sweets and desserts (dark chocolate) decreased IL-6 in addition trials

Safety, limits and interactions

• In 100 samples of farm-made ice cream and sorbet from 48 Swiss farms, no Salmonella, STEC or Listeria monocytogenes was found, and Bacillus cereus group bacteria were present in 15%. — 100 frozen dessert samples, 48 farms, June to September 2025 (n = 100 samples) Evidence C source
  0 major pathogens, B. cereus group in 15% at 2.00-3.57 log CFU/g (about 100 to 3,700 CFU per gram)

What is still unknown

• A meta-analysis of 51 controlled trials (n = 2059) of 10 food sources of fructose-containing sugars found that varying uncertainty remains about the lack of effect of other important food sources on liver fat. — 51 trials of at least 7 days in mostly healthy younger adults (n = 2059) Evidence A source
  IHCL SMD 1.72 (95% CI 1.08 to 2.36) for total fructose-containing sugars pooled in addition (excess energy) trials, not a result for any single food source, and an uncertain lack of effect for other sources
• A dose-response meta-analysis of 15 prospective cohorts (930677 people) found that the apparent protective link between dairy desserts and hypertension did not hold in the dose-response analysis. — 15 prospective cohorts, 930677 participants, 363459 cases (n = 930677) Evidence C source
  Pairwise protective association for dairy desserts not supported by linear dose-response, certainty very low
• In a randomised trial in 30 healthy adults eating one Thai dessert with breakfast, the high-GI dessert group showed lower Bifidobacterium and Collinsella after 24 hours than the low- and medium-GI groups, and the authors call the effects possibly temporary. — 30 healthy adults aged 18-45, three groups of 10, 24 hours (n = 30) Evidence B source
  Relative abundance changes in several genera after 24 h, no clinical outcome

The bottom line

If you want dessert to cost you less, change the recipe before you change the habit. In one laboratory lunch, the 240 kcal gap came from the same familiar dessert made at different energy densities, above 3 kcal per gram at the top end, and people simply ate what was in front of them. What they ate later that day was not measured, so nobody knows whether they made up for it. Of everything on this page, a lighter version of the dessert you already like is the one step tested directly. Two hopeful findings did not hold up. Pooling 15 cohorts and 930677 people, dairy desserts first seemed to go with less high blood pressure, and the link vanished once the authors checked it against how much people actually ate. They rated that evidence very low in certainty. In a trial of 30 adults, one high-GI Thai dessert at breakfast (high GI meaning it raises blood sugar quickly) shifted a few gut bacteria within 24 hours. Nobody measured whether anyone felt or fared any differently, and the authors say the change may be temporary, so it is too early to read anything into it. If you have a peanut or hazelnut allergy, this page cannot tell you anything about safe amounts. Farm-made ice cream and sorbet from 48 Swiss farms carried none of the major food poisoning bacteria across 100 samples, and 15 percent held Bacillus cereus group bacteria at about 100 to 3,700 per gram. That is one summer survey in one country. This survey says nothing about whether farm or soft-serve ice cream is safe in pregnancy. We have no evidence cards on desserts in pregnancy, for children, or for people managing diabetes or taking medicines, so this page cannot tell those groups anything specific. That question belongs with the person treating you.

Sources

1. dsb3-05 · RCT (crossover)
   In experiment 1 (n = 34, 62% female, mean age 37.4 years), participants were served a lunch including a familiar low, medium or high ED dessert to eat ad libitum...The high ED food was associated with an increased intake of approximately 240 kcals compared to medium (p < 0.001, Cohen's d = 2.31) and low (p < 0.001, Cohen's d = 4.42) ED foods.
   Appetite, 2024 · checked 2026-09-23
2. dsb3-09 · agency position
   The leading sources of added sugars in the U.S. diet are sugar-sweetened beverages, desserts, and sweet snacks.
   CDC, Get the Facts: Added Sugars, updated 29 April 2026 · checked 2026-09-23
3. dsb3-06 · human physiology study
   Analysis showed a mean rate of decrease in gastric volume with an 8% fat dessert was 0.33 ± 0.09 mL/min compared to an average rate of increase in gastric volume of 0.09 ± 0.10 mL/min for the chocolate bars. In parallel, eight allergic patients were challenged for either peanut or hazelnut in the same two matrices and doses using a standardized protocol.
   Mol Nutr Food Res, 2012 · checked 2026-09-23
4. dsb3-01 · meta-analysis
   We included 47 trials (85 comparisons; N = 2763) assessing 9 food sources...There was evidence of an interaction by food source: SSBs and sweets and desserts increased uric acid levels in the substitution design
   J Nutr, 2021 · checked 2026-09-23
5. dsb3-03 · meta-analysis
   We identified 64 controlled trials (91 trial comparisons, n = 4094) assessing 12 food sources...addition trials (fruit decreased CRP and TNF-α; sweets and desserts (dark chocolate) decreased IL-6)
   Nutrients, 2022 · checked 2026-09-23
6. dsb3-08 · cross-sectional
   Neither Salmonella, STEC, nor L. monocytogenes were detected...Members of the B. cereus group (B. mosaicus subsp. cereus, B. mosaicus, B. mycoides, B. cereus s.s., B. toyonensis) were identified in 15% of samples, at levels ranging from 2.00 to 3.57 log CFU/g.
   J Food Prot, 2026 · checked 2026-09-23
7. dsb3-02 · meta-analysis
   We included 51 trials (75 trial comparisons, n = 2059) of 10 food sources...Varying uncertainty remains for the lack of effect of other important food sources of fructose-containing sugars at different levels of energy control.
   Nutrients, 2022 · checked 2026-09-23
8. dsb3-04 · meta-analysis of cohorts
   The pair-wise protective association of dairy desserts was not supported by linear dose-response analysis. Fruit drinks or sweet snacks were not associated with hypertension. Certainty of the evidence was "low" for sugar-sweetened beverages, 100% fruit juice, fruit, and yogurt and "very low" for fruit drinks, sweet snacks, and dairy desserts.
   J Am Heart Assoc, 2019 · checked 2026-09-23
9. dsb3-07 · RCT
   This open-label, parallel randomized clinical trial involved 30 healthy individuals aged 18 to 45 years...In the high-GI dessert group, the abundance of Collinsella and Bifidobacterium decreased compared to the low- and medium-GI groups...though these effects may be temporary
   Nutrients, 2024 · checked 2026-09-23