Combining all prenatal methamphetamine exposure into one category can hide differences in timing and in the comparison population. Wright and colleagues’ 2015 study of a Honolulu prenatal clinic illustrates both problems. Its findings concern a historical clinical cohort, not a randomized test of exposure or a prediction for an individual pregnancy.
What population did the study follow?
The investigators retrospectively analyzed clinical data collected prospectively during prenatal care from 2007 through 2013. The live-birth analysis included 251 infants: 144 with recorded prenatal methamphetamine exposure and 107 without it. A subgroup of 59 infants had no recorded drug exposure. The 59 were part of the comparison population, not an additional independent group to add to 251.
The clinic provided prenatal care, addiction-related services and practical support. Many participants experienced housing insecurity and socioeconomic disadvantage. Women entering this clinic, and particularly those who remained in care, may differ from people not seeking similar services.
Which birth-weight comparison was null?
Mean birth weight was 3,159 g in the methamphetamine-exposed group and 3,168 g in the non-methamphetamine group (p=0.9). That broad comparison did not demonstrate a difference. Against the no-drug subgroup, the exposed mean was lower: 3,159 versus 3,321 g (p=0.04).
These findings are compatible because the control groups differ. “No methamphetamine exposure” could include other drug or tobacco exposure and prior addiction history. Reporting only the significant comparison would conceal an important null finding and alter what the study actually tested.
What did timing add?
Among the 144 exposed infants, recorded exposure ended in the first trimester for 50, continued into the second for 45, and continued into the third but was negative at delivery for 29. Twenty were classified as having positive toxicology at delivery.
The study associated continued use through delivery with shorter gestation and smaller infants. The delivery-positive subgroup’s mean birth weight was 2,932 g, compared with 3,300 g after first-trimester-only exposure (p=0.01). Those are group observations, not estimates of how many grams a specified dose would change an individual infant’s weight.
The source’s exposure terminology needs care: its abstract mentions meconium or urine in describing birth toxicology, while the methods define the delivery-positive classification using maternal or neonatal urine because meconium can reflect earlier exposure. The classification should not be treated as a validated exact last-use clock.
What did adjustment change?
The adjusted analysis associated persistent exposure with preterm delivery, but not with small-for-gestational-age classification. Persistent smoking, rather than methamphetamine, was associated with that latter outcome. Birth weight, gestational age and size relative to gestational age are different endpoints.
The authors also found that continuing users were more likely to have inadequate prenatal care. Statistical adjustment addresses measured covariates; it does not make the groups exchangeable as though exposure cessation had been randomized.
What can be concluded about stopping use?
Participants who stopped use had more favorable observed birth outcomes than those continuing through delivery. That is clinically meaningful historical evidence, but cessation occurred alongside differences in care engagement and potentially other unmeasured factors. The study cannot isolate how much of the association came from stopping methamphetamine alone.
Self-report, incomplete alcohol information, absent maternal weight-gain data and possible dating error in pregnancies entering care late limit interpretation. The study’s null results do not establish safety of prenatal exposure. Its useful lesson is to retain timing, comparator composition and the particular outcome before drawing a causal conclusion.
Source
Wright TE, Schuetter R, Tellei J, Sauvage L. Methamphetamines and Pregnancy Outcomes. Journal of Addiction Medicine. 2015;9(2):111–117. doi:10.1097/ADM.0000000000000101.
Leave a Reply