Seven days of low-plastic eating cut three chemical markers in urine

Most of what is known about plastic-associated chemicals and human health comes from observational studies: measure a metabolite in urine, follow people, look for associations. That design cannot tell anyone whether changing behaviour changes exposure. The PERTH trial — Plastic Exposure Reduction Transforms Health — tried to answer the narrower, more tractable question by randomising it.

The study, published in Nature Medicine on April 21, has two parts: an observational cohort of 211 Australian participants and a seven-day pilot randomised controlled trial in 60 participants [s1].

The design

Intervention groups received combinations of three things: plastic-free kitchenware, low-plastic personal-care products, and food sourced from more than 100 producers chosen to minimise plastic touchpoints from farm to plate [s1]. The control group received no intervention [s1].

The primary outcome was a reduction in urinary levels of plastics-associated chemicals [s1]. The trial is registered with the Australian and New Zealand Clinical Trials Registry as ACTRN12622001252707 [s1].

One design detail matters more than it might appear: participants' daily energy intake was maintained across the intervention [s1]. Swapping the food supply without changing how much people eat isolates the packaging variable from the diet-quality variable, at least partially.

What changed

Over seven days, the dietary intervention decreased plastic exposure (P < 0.001) and reduced urinary levels of three specific chemicals [s1]:

• mono-n-butyl phthalate, down 37.5% (95% CI −55.6 to −12.0; P = 0.007)
• monobenzyl phthalate, down 53.5% (95% CI −72.7 to −20.6; P = 0.005)
• bisphenol A, down 59.7% (95% CI −82.5 to −6.87; P = 0.033)

The widest effect on excretion came from the groups given food with minimal to no plastic contact [s1]. Replacing low-plastic personal-care products alone produced an independent decrease in urinary mono-n-butyl phthalate compared with no intervention [s1].

Note the confidence intervals. The bisphenol A interval runs from −82.5% to −6.87% — the direction is consistent but the magnitude is not pinned down, which is what a 60-person pilot over seven days should be expected to produce.

The cohort half

In the observational cohort of 211 people, highly processed, plastic-packaged and canned foods emerged as important modifiable contributors to urinary metabolite levels [s1]. The researchers also observed negative associations between cardiometabolic biomarkers and higher urinary metabolites of di(2-ethylhexyl) phthalate [s1].

That association is observational. It is a correlation within a cohort of 211 people, not a demonstrated effect, and it belongs in the hypothesis column.

What this does not establish

The trial measured chemicals in urine. It did not measure health. Nobody's blood pressure, liver function, insulin sensitivity, or long-term risk was shown to improve, and seven days is nowhere near long enough to test any of that.

The reductions also came from a bundled intervention that included sourcing food from more than 100 producers [s1] — an arrangement no household can replicate. Isolating which specific change did the work is beyond what this design can support, though the comparison across intervention groups points at food contact rather than personal-care products as the larger contributor [s1].

Urinary metabolites of these compounds clear quickly, which cuts both ways. It is why a seven-day intervention could move them at all, and it is why the phrase "despite constant plastic exposures" appears in the authors' own summary: exposure through non-food routes continued throughout [s1].

The seven-day window

Seven days is short for a trial and about right for these particular molecules. Phthalate and bisphenol metabolites have short biological half-lives, which is why a week-long change in exposure can show up in urine at all — and why a single urine measurement is a poor guide to what someone was exposed to a month earlier. The design leans on that property rather than fighting it.

It also means the result says nothing about whether the reductions persist. The trial ran for seven days [s1]. Nobody was followed afterwards to see whether metabolite levels rebounded once the supplied food and kitchenware went away, and in a real household they would.

What it adds

The value of PERTH is not the effect size. It is that the effect exists in a randomised design at all. Before this, the plausible objection to any advice about food packaging was that background exposure is so pervasive that individual changes are swamped. Over one week, in one pilot, in 60 people, they were not [s1].

Whether that translates into anything a body notices is the next question, and this trial does not answer it.

This article is informational and is not medical advice.

Sources

• Low-plastic diet and urinary levels of plastic-associated phthalates and bisphenols: the randomized controlled PERTH Trial — Nature Medicine, 2026-04-21