The National Academy of Sciences (NAS) is a non-profit, independent organization of scholars. Their reviews are generally good, rigorous and thorough and this is no exception.
If I simply reviewed 10,000 abstracts, I would generally agree with their interpretation of the literature. However, having followed this body of literature for more than 10 years, this is my analysis of the findings that made headlines.
Read the report here
Reviewed 10,000 abstracts
Abstracts are advertisements, they represent what the authors want you to take away from their data. An evaluation of auto safety by looking through advertisements is not as good as seeing and evaluating the cars themselves. In my experience, much of the marijuana research here in the US is funded by the National Institute of Drug Abuse and that perspective is often seen in how the data are represented in the abstract.
In addition, this review is not evaluating 10,000 separate and distinct studies as many of the 10,000 abstracts come from three longitudinal studies. These studies, by their authors own admission, are limited by the confounders of tobacco use, race and poverty. The report itself mentions this limitation to the body of evidence they reviewed.
“Substantial evidence for low birth weight”
Low birth weight (LBW) is particularly affected by these variables – a point that has been documented in recent independent reviews of the marijuana literature (see S. Conner et al., AmJObGyn, 2016 and Leemaqz et al., Repro.Tox, 2016). These reviews found that when tobacco use is accounted for, marijuana is not an independent risk factor for LBW (SGA). In my opinion, the substantial risk cited in this report is more attributable to cigarette smoking than marijuana use alone.
I certainly know stories of folks who smoked marijuana prenatally and had LBW babies. But generally they were heavy users (daily or more) and/or they had another risk factor for LBW, suggesting that marijuana is part of a risk analysis for LBW, but should not be considered the only element.
But if you simply read the abstracts, you would indeed come to this conclusion.
“Some evidence for NICU admission”
I struggle with NICU admission as a birth outcome data point. This is, in part, a result of my 25 years of clinical research experience and how I see newborns handled after birth in hospitals with NICUs. Unfortunately, NICU admission is clinical parameter that is very dependent on individual practitioners and is not well standardized. Policies for NICU admission vary across hospitals and providers and, in my experience are also multi-factorial.
If a LBW baby is born, they will most likely end up in the NICU for observation, even if nothing is wrong. Does that NICU admission reflect the harm of marijuana or reflect the way the system works? Further, I have heard stories of babies being observed in the NICU because their parent refused a drug screen upon admission to the hospital in labor. While some of those babies indeed had been exposed to marijuana prenatally, their NICU admission reflects medical concern, not health outcomes.
Finally, because I am a physiology junkie, I know that NICU admissions are sometimes the result of a deviation from physiologic, mammalian birth. When a just born human mammal is separated from their caregivers, they decompensate physiologically. Their heartrate increases, their adrenal response spikes, their oxygen saturation decreases (increasing their respiratory rate) and they are less able to regulate their temperature. In my experience, when you add the layer of suspected substance use, you begin a cycle that can increase NICU admissions.
I hear this story regularly: a baby is suspected of being exposed to marijuana so they are removed to the warmer for immediate evaluation. While some babies’ physiology is resilient enough to maintain their vitals in this state, some mammals really do require immediate extragestation after birth to maintain their physiology. So the baby who needs extragestation decompensates because of the separation, presents with abnormal vitals and heads to the NICU. What is the underlying cause of this NICU admission?
Biologic outcomes related to marijuana use that require an increased level of health care or is this an indication that the system response to suspected or confirmed substance use ITSELF increases NICU admissions? In a post-liberalization climate of Colorado, we must pay very close attention to the bias inherent in our health care systems so that we do not create harm to families.
The schizophrenia element
I have done some research on this relationship over the years because I am asked about the impact of marijuana on schizophrenia regularly. I do believe that marijuana use – particularly regular use – can impact mental health in positive and negative ways. With respect to schizophrenia specifically, my understanding is that schizophrenia is usually associated with a reduced endocannabinoid system. Since schizophrenia is quite genetically linked, marijuana use might very well (and indeed does) run in those families. The comment by NAS that they didn’t account for genetics is good truth-telling and allows us to interpret their data accordingly. I would think there is a possibility that folks with a history or increased risk of schizophrenia might be more likely to self-medicate with marijuana. My reading of the literature does not support that the use of marijuana increases schizophrenic symptoms per se because I do not know if people with increased symptoms are more likely to use marijuana or if increased use increases symptoms.
The problem of the null hypothesis
What doesn’t make news are the non-significant findings and there were some biggies in this review. Two of the questions I am asked most regularly is whether prenatal marijuana use increases the rate of SIDS and/or decreases academic performance in adolescents. The NAS literature review found there was “insufficient evidence to support or refute a statistical association between maternal cannabis smoking and later outcomes in the offspring (e.g., SIDS, cognition/academic achievement, and later substance use).”
This received little to no media coverage and is a clear example of the bias against the null hypothesis. We humans love to demonstrate differences because we struggle to explain similarities. But in this case, the lack of findings are important and should be noted by the press and health care providers. I do not believe that this report will change many minds that already believe that prenatal cannabis use increases SIDS or decreases academic performance.
{PSA: To be clear, I don’t think developing teenage brains should use marijuana and that THAT DOES cause problems for some. And poverty also increases risks for these outcomes. But the perception that prenatal use affects adolescent behavior is very strong and currently accepted as truth among health care providers and policy makers. At least in Colorado.}
In sum, this report makes clear the danger of reading just abstracts to interpret a body of literature. It also highlights our inherent bias against the null hypothesis, which makes changing minds and policies very difficult. My hope is that future policies and reports are built on a deeper reading of the literature.