Does a Positive Maternal Urine THC Result Prove THC Is Present in Breast Milk One Week After Delivery? A 13-Woman Study

Short answer: No. Maternal urine and breast milk are different biological specimens, and this study measured different cannabis compounds in them. Among eight women reporting no cannabis use after delivery who supplied complete specimens about one week postpartum, urinary THC-COOH remained quantifiable in 7 of 8, while breast-milk delta-9-tetrahydrocannabinol (Δ9-THC) was quantifiable in 0 of 8 at the study’s analytical threshold. The result demonstrates a concrete mismatch between measured matrices in a small observed group; it does not establish that the milk contained no THC molecules, that breastfeeding was risk-free, or that one week’s abstinence guarantees the same finding in another person.

The underlying original research is Sandonís and colleagues’ 2026 longitudinal observational study, “Breast milk Δ9-tetrahydrocannabinol in cannabis users during the postpartum period: correlation between breast milk, maternal urine and saliva samples during early lactation”. Its evidentiary value lies in keeping the analyte, specimen, sampling time, and limit of quantification together instead of treating a generic “THC positive” result as a universal finding.

What did the investigators actually measure?

The researchers obtained three kinds of maternal samples: expressed breast milk, passively drooled saliva, and urine. Laboratory liquid chromatography–tandem mass spectrometry (LC-MS/MS) targeted parent Δ9-THC in milk and saliva, but 11-nor-9-carboxy-Δ9-THC (THC-COOH), an inactive metabolite, in urine. All numerical results were reported in nanograms per milliliter (ng/mL). The same measurement unit does not turn the three specimen compartments or two chemical targets into interchangeable evidence.

For clinical or litigation interpretation, this is an important distinction from venous whole-blood THC, plasma THC, an immunoassay screening class result, or an estimated dose. No contemporaneous venous blood or plasma THC concentration was measured in this research. It did not test maternal driving, neurocognitive impairment, actual infant absorbed dose, or a specific adverse infant outcome.

Who was studied, and why do the denominators matter?

The Barcelona hospital enrolled postpartum women over 18 who screened positive for cannabis in urine at delivery and intended to breastfeed. Fifteen were initially considered; one was excluded after the delivery screen proved false positive on quantitative follow-up, and another provided specimens two weeks postpartum, outside the prespecified window. Thus the final analyzed sample was 13 women, not 15.

Eleven women reported remaining abstinent after delivery. Two reported using cannabis between the second and third sampling stages and were therefore presented separately for relevant follow-up analysis. Only seven participants completed all three scheduled kits, one missed the first kit but supplied the next two, and five did not complete the third kit. The abstinence-stratum count consequently varied: 10 complete kits around 24 hours, 11 around 48 hours, and only eight at approximately one week. These are changing sample counts within one longitudinal study, not 29 independent mothers.

The protocol used collections around 24 hours, 48 hours, and seven days (±48 hours) after birth. Urine, milk, and saliva were generally obtained together at each stage when a complete kit was available. An observation at seven days is not equivalent to seven days after a precisely observed last cannabis dose: the researchers could not establish exact pre-delivery last-use times, and abstinence after birth was self-reported.

How were the samples collected and analyzed?

Approximately 10 mL of urine was collected by spontaneous voiding, approximately 5 mL of saliva by passive drooling into polypropylene tubes, and expressed milk by the mother with trained staff guidance. Milk from both breasts was pooled when available; small early colostrum volumes meant a full 5 mL was not always obtainable. Samples were held refrigerated, shipped to a reference laboratory on the same day, and tested by LC-MS/MS using an Agilent 1290 chromatographic system and 6430 triple-quadrupole mass spectrometer.

For parent THC in milk and saliva, the study stated a 5 ng/mL limit of quantification (LOQ). For urinary THC-COOH, the LOQ was 2 ng/mL. These were the research measurement thresholds. Separately, the clinical maternal urine screen at delivery employed a 50 ng/mL cutoff; that screening criterion is not the same thing as the later LC-MS/MS urine LOQ of 2 ng/mL. Mixing these thresholds would materially change the interpretation.

A specimen labeled “<LOQ” could contain analyte at a concentration that the method did not reliably quantify. It should not be silently converted to a true analytical zero, an absent molecule, or an assurance of no possible exposure. Readers evaluating the laboratory record should consider the general principles of biological-matrix interpretation and the specific reporting limits of the assay actually performed.

What happened to quantifiable results at 24 hours, 48 hours, and one week?

The following evidence table preserves the study’s changing denominator for women who reported abstinence after delivery. It reports the number of specimens at or above the applicable quantification threshold, not a proportion of infants exposed and not the percentage of mothers whose milk was necessarily free of THC.

The principal numerical point is 7/8 versus 0/8 at about one week. A detectable urinary metabolite did not equate to a simultaneously quantifiable parent-THC milk specimen at the study’s 5 ng/mL LOQ. Conversely, the absence of a quantifiable milk result does not authorize a universal statement that a maternal urine finding is always unrelated to milk concentrations; at the first stage, some positive values co-occurred.

The original specimen-level entries also show why a simple monotonically declining graph can conceal variation. In one abstinence-stratum participant with two milk samples, concentrations were 34 ng/mL at the first collection and 81 ng/mL at the second; another had 40 then 49 ng/mL. Neither supplied a third-stage complete kit. One woman had a milk value of 5 ng/mL at the first stage and an unquantifiable second-stage sample. The change in which people contributed each stage, colostrum and milk fat variation, and imprecise exposure timing all constrain time-course conclusions.

What does a woman with quantifiable urine THC-COOH but <LOQ milk THC represent?

It represents two different answers from two different assays on different samples. Urinary THC-COOH supports prior cannabinoid exposure under the assay’s identification criteria. The breast-milk result means parent Δ9-THC was not quantifiable at 5 ng/mL in that particular collected sample. There is no valid arithmetic that turns an arbitrary urine THC-COOH concentration into a current breast-milk Δ9-THC concentration without a validated specimen-specific prediction model. This small study did not establish such a model.

This distinction matters when records are condensed into phrases such as “tested positive for marijuana,” without identifying the analyte or specimen. A precise forensic statement names the sample, compound, method, unit, reporting limit, and collection time. The specimen-collection record also matters because lactation stage, volume expressed, handling and collection timing affect what material the laboratory actually analyzed.

Were saliva and urine reliable substitutes for breast-milk testing?

No diagnostic substitution was validated. Early paired measures showed significant within-dataset Spearman correlations, but the paper did not derive or externally validate an equation for predicting an individual milk concentration from maternal saliva or urine. At approximately 24 hours, the authors reported correlations between milk parent THC and urinary THC-COOH (ρ = 0.770, p < .01) and between milk and salivary parent THC (ρ = 0.653, p < .05). At 48 hours, the corresponding milk–urine coefficient was 0.501 and milk–saliva 0.435, without the table’s significance markers.

Below-LOQ observations were common in saliva and milk. Many tied or low measurements and a tiny selected cohort complicate interpretation of rank correlations. By one week, milk and saliva values were below LOQ for all eight complete abstinence-stratum kits, so correlation could not be calculated from those essentially constant quantified outcomes. An early correlation is not a later predictive performance measure, and it does not furnish diagnostic sensitivity, specificity, negative predictive value, or a legally defensible individual classifier.

What did the two participants who resumed cannabis use show?

They were not silently combined with the 11 reported-abstinent participants for interpreting cessation. At the first two stages, both had provided samples. One had urine THC-COOH of 31 and 74 ng/mL at 24 and 48 hours, then 407 ng/mL at about one week; at that later collection her saliva THC was 108 ng/mL and milk THC 13 ng/mL. The other had urine THC-COOH of 78 then 263 ng/mL, but did not provide a complete third kit. These observations are consistent with the importance of intervening use and repeat collection, but only one complete third-stage kit was available in this small subgroup.

Neither those isolated values nor the abstinence-stratum numbers estimate infant intake, nor do they reveal the precise timestamp or dose of the resumed use. Even a detailed single subject timeline requires a defensible dosing history, full analytical record, and appropriate pharmacokinetic context before individualized inferences could be attempted.

Can these findings justify a fixed seven-day “clearance” rule?

No. “None quantifiable at the final collection” is a statement about eight available milk specimens with an LOQ of 5 ng/mL. The paper was not a controlled dose-clearance experiment. It did not collect every feeding, continuously assay expressed milk, determine each person’s last-dose time, establish a milk elimination half-life for each mother, or directly measure infant exposure. The observation should not be translated into a universal seven-day waiting period or a claim about clinical safety.

Because cannabis compounds are lipophilic, differences between colostrum and mature milk and between lower-fat and higher-fat fractions could affect apparent concentrations. The investigators themselves noted that some milk values increased from the first to the second stage and that another published cohort reported longer detection. Those contextual papers were discussed by the authors, but the numerical conclusions here are drawn only from this original 13-person investigation. Decisions about breastfeeding and infant care require individualized clinical assessment and current professional guidance, not a forensic inference from these limited counts.

What other design limitations constrain testimony?

Selection started from mothers whose clinical urine tests were positive at delivery and who intended to breastfeed; it was not a representative sample of all lactating mothers or cannabis users. The sample was small, stage-three attrition was substantial, and the two reported-resumption participants were too few to compare statistically with the abstinent group. Cannabis cessation was based on self-report, and exact last use before delivery was unknown. There was no controlled randomization of exposure, and no contemporaneous maternal plasma or venous whole-blood THC measurement.

Moreover, breast milk composition changes sharply during early lactation. Expression volume and timing, pooled breasts, potential differences in fat content, and refrigerated handling are relevant to interpretation even when the analytical instrument performs as reported. Correlation analyses involved several selected clinical and laboratory measures, repeated sampling and numerous below-LOQ entries. The article did not establish an adjustment for all multiplicity or generate a clinical classification operating-characteristic curve. Source Table 3’s correlation labels should not be treated as a validated individual prediction model.

What should be requested when this study is cited in litigation or a medical record dispute?

Start with the original paper and its tables, not an abbreviated claim that urine THC “means” milk THC. Identify whether the contested evidence is a delivery-admission urine immunoassay, a subsequent quantitative urine THC-COOH assay, direct milk parent-THC LC-MS/MS, or merely a clinical note. Obtain sampling chronology, laboratory identification/quantification criteria, actual reporting limits and any confirmatory results, collection and storage documentation, and a reliable history of postpartum use. Consider whether milk was collected in early colostrum or later lactation and whether any specimen was missing.

The resulting statement should match the proof. The study can support a limited methodological point: **a urine metabolite result may remain quantifiable when parent THC is not quantifiable in sampled milk**. It cannot by itself establish a child’s dose, health effect, negligence, protective fitness, impaired driving, intent, or cause of an alleged event. For analytical questions, method-specific confirmatory testing is a separate inquiry from deciding whether any biological result explains a clinical or legal outcome.

Seven frequently asked questions

1. Does maternal urine THC positivity prove contemporaneous THC in breast milk?

No. The paper measured urine THC-COOH and milk parent THC with different LOQs. At about one week, urine was quantifiable in 7/8 complete abstinence-stratum kits while milk THC was quantifiable in 0/8. That is evidence against treating the two test results as interchangeable.

2. Does “below the limit of quantification” mean the milk was THC-free?

No. The 5 ng/mL milk LOQ describes the method’s ability to quantify a concentration reliably; it is not a proof that the specimen contained zero molecules or that exposure could not occur at a lower concentration.

3. Did the study confirm that stopping cannabis for seven days makes breastfeeding safe?

No. The investigators observed a limited cohort, could not establish a uniform last-use clock, sampled a subset at one week, and did not measure infant absorbed dose or clinical safety outcomes. The data do not define an individualized waiting interval.

4. Was the urine screening cutoff the same as the laboratory LOQ?

No. Maternal clinical screening at delivery used a 50 ng/mL criterion. Later quantitative LC-MS/MS research measured urinary THC-COOH with a 2 ng/mL LOQ, while milk and saliva parent THC each had a 5 ng/mL LOQ.

5. Could saliva replace breast milk as the tested specimen?

Not based on these data. An early correlation was reported, but the stage-two association was not significant, and all eight final-stage saliva specimens were below the assay LOQ. No individual prediction algorithm or classification error rates were validated.

6. How many women were actually represented at one week?

Eight who reported postpartum abstinence provided complete third-stage kits, not all eleven in that stratum and not all thirteen women in the final cohort. One of two women reporting interval cannabis use also had a complete third-stage kit and was reported separately.

7. Can these maternal tests determine an infant’s actual THC dose or impairment?

No. The study did not quantify infant ingestion or systemic exposure, measure clinical impairment, or assess individual injury causation. Maternal specimen concentrations do not automatically establish the amount delivered to or absorbed by a child.

Source and bottom line

Primary original research: Sandonís M, Brik M, Ramirez C, et al. (2026), Frontiers in Psychiatry 17:1816611. DOI: 10.3389/fpsyt.2026.1816611. The original Methods, Results, Tables 1–3, and Discussion support the measurements and limits described here.

Bottom line: This 13-woman study illustrates why the specimen and the exact molecule matter more than a broad statement that a person is “THC-positive.” A positive urine THC-COOH result cannot be automatically equated with quantifiable parent Δ9-THC in breast milk. Neither a quantitative milk result nor an unquantifiable one independently determines infant risk, clinical advice, or legal causation.