. Author manuscript; available in PMC: 2017 Sep 12.

Published in final edited form as: JAMA. 2017 Jan 10;317(2):207–209. doi: 10.1001/jama.2016.17383

Table.

Trends in prevalence of marijuana use in pregnant and non-pregnant women, 2002–2014 (NSDUH)^a

	Adjusted Prevalence^b						Prevalence Ratio (95% CI)^f		Test for difference in prevalence ratios^g
	2002 N=15,284^c			2014 N=15,318^d					Test for difference in prevalence ratios^g
	n	%	(95% CI)	n	%	(95% CI)			p-value
Past-month Marijuana use^e
Pregnant Women	40	2.37	(1.85, 3.04)	43	3.85	(2.87, 5.18)	1.62	(1.09, 2.43)	0.64
Non-Pregnant Women	1531	6.29	(6.02, 6.57)	1673	9.27	(8.90, 9.65)	1.47	(1.38, 1.58)	0.64
Past-year Marijuana use^e
Pregnant Women	134	8.64	(7.32, 10.19)	115	11.63	(9.78, 13.82)	1.35	(1.05, 1.72)	0.73
Non-Pregnant Women	2809	12.37	(12.05, 12.70)	2824	15.93	(15.48, 16.40)	1.29	(1.23, 1.35)	0.73

Data are from the U.S. National Survey on Drug Use and Health (NSDUH)

Adjusted prevalence estimates are from the “linear predicted” prevalence model described in footnote “a” of the Figure.

Sample sizes in 2002: pregnant women, n=797; non-pregnant women, n=14,487.

Sample sizes in 2014: pregnant women, n=735: non-pregnant women, n=14,583.

Past-month marijuana use was defined as responding “within the past 30 days” to the question, “How long has it been since you last used marijuana or hashish?”. Past-year marijuana use was defined as responses of “within the past 30 days” or “more than 30 days ago but within the past 12 months”. Pre-processing of missing variables by Predictive Mean Neighborhood imputation and recoding is done prior to public release of the NSDUH datasets.⁵ Because the analyses used the NSDUH’s imputed variables, there were no missing data. N’s represent the actual number of participants in the survey and are not weighted values.

Prevalence ratios are the ratio of the adjusted prevalence estimates from 2014 divided by the adjusted prevalence estimates from 2002; Ratios and 95% confidence intervals are from log-Poisson regressions. Confidence intervals for prevalence ratios that do not include 1.00 within the lower and upper levels indicate statistically significant increasing trends in marijuana use.

The test for difference in prevalence ratios is the p-value of the pregnancy*year interaction in the log-Poisson regression. This test indicates whether the ratio of the prevalence ratios for pregnant versus non-pregnant women differs significantly from 1.00. Non-significant (p>0.05) p-values indicate insufficient evidence to conclude that the prevalence ratios differ.