Does glyphosate exposure during pregnancy raise the chance of delivering early?
Yes, at least at one specific point in pregnancy. In 1,450 pregnant women in New York City, higher glyphosate levels measured between 18 and 25 weeks were linked to about 35% higher odds of preterm birth (odds ratio 1.35). For births where labor started on its own, the odds were roughly double (odds ratio 2.01).
Glyphosate is the active ingredient in many common weed killers. It is the most widely used herbicide in the world, and small amounts end up in food, water, and dust. Most people in the United States have some in their urine on any given day. That was true here too: exposure was close to universal in this group, just at low levels.
What makes this study different is that researchers did not guess at exposure from where people lived or what they said they ate. They measured the chemical directly in urine, twice, at two different stages of pregnancy. Then they followed each pregnancy to see how long it lasted.
What the data show
The study followed 1,450 pregnant women in the NYU Children’s Health and Environmental Study between 2016 and 2019. Overall, 7% of the births were preterm, meaning they happened before 37 weeks. That works out to 103 early births out of the full group.
The timing of the exposure mattered a great deal. Urine samples collected early, between 4 and 18 weeks, showed no link to preterm birth at all. Samples collected later, between 18 and 25 weeks, told a different story. Higher glyphosate at that second window was tied to about 35% higher odds of any preterm birth (odds ratio 1.35, 95% CI 1.04 to 1.76). Read plainly, that confidence range means the true increase is very likely somewhere between 4% and 76%, so the direction is fairly clear even though the size is not pinned down.
The signal was stronger, and more specific, for spontaneous preterm birth. That is the kind where labor or the water breaking starts on its own, without a doctor deciding to deliver early. There, higher glyphosate at 18 to 25 weeks was linked to roughly double the odds (odds ratio 2.01, 95% CI 1.40 to 2.90), or very likely somewhere between 40% and 190% higher. Medically indicated preterm births, the ones a care team chooses for reasons like preeclampsia, showed no association.
The researchers also measured AMPA, the main chemical that glyphosate breaks down into in the environment. AMPA was not linked to preterm birth in any of their models.
Dr. Kumar’s Take
The split between spontaneous and medically indicated births is the most useful detail here. If glyphosate were simply a marker for being poorer, sicker, or more stressed, I would expect to see the risk rise across both types. Instead it showed up only where labor started by itself. That is the pattern you would predict if the chemical were nudging an inflammatory process that helps trigger labor, rather than tracking someone’s underlying health.
The second-trimester timing fits that idea too. The window between 18 and 25 weeks is when a lot of the machinery that holds a pregnancy in place is under the most strain.
The limits are real. This is an observational cohort, so it cannot prove glyphosate caused these early births. Urine glyphosate clears fast, so two samples are a rough snapshot of a nine-month exposure. And this was one urban American population, so I would not assume the same numbers hold in a farming community with much higher exposure or in a country with different food rules. What it does do is measure the actual chemical in actual people and find a consistent, biologically sensible signal. That is a stronger starting point than most environmental research gets.
How the study was done
This was a prospective birth cohort, which means the researchers enrolled women during pregnancy and collected the urine samples before anyone knew how the pregnancies would turn out. That ordering matters. It rules out the problem where people who had a bad outcome remember their exposures differently.
Glyphosate levels were adjusted for creatinine, a standard correction that accounts for how diluted a urine sample is. Without it, someone who drank a lot of water that morning would look falsely low. The analysis used a method that handles two measurements from the same person, so both time points could be compared inside one model rather than treated as unrelated.
What this means if you are pregnant
There is no version of this study that lets me tell you to hit a specific glyphosate number, and no clinical test that would be useful to order. Exposure at these low levels is nearly universal, which means it is not really a personal failing to have some. Still, most exposure in a non-farming population comes through food and drink, and that is the part you have some say over.
Practical Takeaways
- Washing produce well and peeling when practical removes some surface herbicide residue, and it is a low-cost habit that carries no downside during pregnancy.
- Certified organic grains and legumes are grown without glyphosate, so if you are going to spend more on a few organic items, oats, wheat, and beans are reasonable places to start.
- Skip using glyphosate-based weed killers around your own home and garden during pregnancy, since direct handling produces far higher exposure than diet does.
- Bring this up with your obstetrician if you have already had a spontaneous preterm birth, because that history is the strongest known predictor and deserves a plan of its own regardless of chemical exposure.
Related Studies and Research
- Everyday chemicals in pregnancy linked to early birth
- PEPFAR put 20 million people on HIV treatment and prevented nearly 8 million infected births
- Long daytime naps linked to higher death risk in older adults
- Your favorite music helps you exercise nearly 20 percent longer
FAQs
Where does glyphosate exposure come from if you do not farm?
For most people it arrives through food. Glyphosate is sprayed on crops like wheat, oats, and legumes, and in some cases it is applied shortly before harvest to dry the crop evenly. Residues survive into flour, cereal, and packaged foods. Smaller amounts come from drinking water and from household weed killers used on lawns and driveways. This study looked at an urban New York population with no agricultural work involved, and glyphosate still showed up in nearly everyone.
Why did the first-trimester samples show nothing?
Nobody knows yet, but there are two reasonable explanations. One is biological: the processes that lead to spontaneous early labor may only become sensitive to an inflammatory push later in pregnancy, so an exposure at 10 weeks lands on a system that is not yet vulnerable. The other is technical: glyphosate leaves the body within a day or two, so any single urine sample captures only recent exposure, and a first-trimester reading may simply be a poorer proxy for what happened months later. Finding an effect in one window and not another is a real result, but it is not the same as proving the first trimester is safe.
What is the difference between spontaneous and medically indicated preterm birth?
Spontaneous preterm birth means the body starts the process itself, through early contractions or membranes rupturing before term. Medically indicated preterm birth means a doctor recommends delivering early because continuing the pregnancy is riskier than ending it, often for conditions like severe high blood pressure or a baby who has stopped growing. The two have different causes and different prevention strategies. That is exactly why it is informative that glyphosate tracked with one and not the other, since a chemical acting on inflammation would be expected to affect the spontaneous kind.
Bottom Line
Glyphosate measured directly in the urine of pregnant women at 18 to 25 weeks was linked to about 35% higher odds of preterm birth, and roughly double the odds of preterm birth where labor began on its own. First-trimester levels showed no link, and neither did glyphosate’s breakdown product AMPA. This is an observational study and cannot prove cause, but it is one of the few to measure the chemical itself rather than estimate it, and the specific timing and subtype pattern point toward a real biological effect worth taking seriously.
