Abstract
Aims
We aimed to determine the association between stressful life events (SLEs) in the year prior to childbirth with (1) pre-pregnancy cannabis use, (2) cessation of cannabis use during pregnancy and (3) postpartum relapse to cannabis use.
Design
We used data from the Pregnancy Risk Assessment Monitoring System (PRAMS) 2016, a cross-sectional, population-based surveillance system.
Setting
Mailed and telephone surveys conducted in five states—Alaska, Colorado, Maine, Michigan and Washington—in the United States.
Participants
Women (n = 6061) who delivered a live infant within the last 6 months and had data on cannabis use.
Measurements
Self-reported data included SLEs (yes/no response for 14 individual events in the 12 months prior to childbirth) and cannabis use [yes/no prior to pregnancy, during pregnancy, and at the time of the survey (approximately 2–6 months postpartum)]. The associations between SLEs and cannabis use (primary outcomes) were examined in logistic regression models adjusted for maternal demographics (e.g. age, race, education), geography (i.e. state of residence) and cigarette smoking.
Findings
Pre-pregnancy, 16.4% (997/6061) of respondents endorsed using cannabis, with 36.4% (363/997) continuing cannabis use during pregnancy. Among the 63.6% (634/997) who did not report use during pregnancy, 23.2% (147/634) relapsed to cannabis use during the postpartum. Nine of the 14 possible SLEs were associated with increased odds of pre-pregnancy cannabis use [e.g. husband/partner or mother went to jail, adjusted odds ratio (aOR) = 2.16, 95% confidence interval (CI) = 1.30–3.62] and four were associated with increased odds of continued cannabis use during pregnancy (e.g. husband/partner lost job, aOR = 2.19, 95% CI = 1.21–3.96). The odds of postpartum relapse to cannabis were significantly associated with two SLEs (husband/partner said they did not want pregnancy, aOR = 2.86, CI = 1.10–7.72; husband/partner or mother went to jail, aOR = 0.37, 95% CI = 0.13–1.00).
Conclusions
Stressful life events during the year prior to childbirth appear to be linked to greater odds of women’s cannabis use during the perinatal period, especially during pre-pregnancy.
Keywords: Cannabis, epidemiology, postpartum, PRAMS, pregnancy, stress
INTRODUCTION
Perinatal cannabis use has increased in recent years [1]. Specifically, national estimates in the United States indicate that 4.2% report use of cannabis during pregnancy and 6.8% report use during the postpartum period [1]. Prevalence is higher in those of lower socio-economic status (e.g. lower education, lower income) in urban areas and who use other substances during pregnancy (e.g. tobacco, alcohol) [1,2].
While there is some debate about the health effects of perinatal cannabis use [3], a recent review indicates that cannabis use during pregnancy is associated with greater odds of low birth weight and admission to the neonatal intensive care unit [4]. Further, there is some suggestion cannabis use may cause both short- (e.g. miscarriage, stillbirth) [2] and long-term negative health effects (e.g. poor executive functioning, emotional and behavioral problems) [5,6]. Therefore, the current clinical recommendations are that clinicians should advise women to quit or reduce their cannabis use during pregnancy [4,5,7]; however, relatively little is known about risk factors for use of cannabis during the perinatal period.
Stressful life events (SLEs), such as adverse childhood experiences, have been linked to increased risk of substance use disorders [8]. The impact of SLEs is also extended to other processes and phases of addiction, including maintenance of substance use disorders and vulnerability for relapse [9,10]. Less is known about the more proximal SLEs. We recently reported that SLEs in the year prior to childbirth were associated with increased odds of perinatal cigarette smoking, especially cigarette smoking during pregnancy [11]. Those who reported cannabis use during pregnancy had a greater number of SLEs in the year prior to childbirth in 2009–11 [1]; however, this research was limited to three states (Alaska, Hawaii, Vermont) and did not examine the relationship between SLEs and cessation of cannabis during pregnancy nor with postpartum cannabis relapse. Therefore, in this study we aimed to explore the association between SLEs that occurred in the year prior to childbirth with perinatal cannabis use in a population-based surveillance system—the Pregnancy Risk Assessment Monitoring System (PRAMS), 2016. Specifically, we aimed to determine the association between SLEs with (1) pre-pregnancy cannabis use, (2) continued cannabis use during pregnancy and (3) postpartum relapse to cannabis use. We hypothesized that greater SLEs would be related to higher odds of pre-pregnancy cannabis use, continued cannabis use during pregnancy and postpartum cannabis relapse.
METHODS
Data source and study design
This cross-sectional study utilized data from the PRAMS, a population-based ongoing surveillance system that is developed between state health departments and the Centers for Disease Control and Prevention (CDC). PRAMS uses state birth certificate files to conduct probability sampling among women who have had a live birth (between 1300 and 3400 women per year per participating site), with oversampling of certain subpopulations (e.g. racial/ethnic minorities) [12]. Selected women are contacted within 6 months of childbirth via mailed questionnaires. Those who do not respond to the mailed questionnaires are contacted via telephone. The PRAMS data are weighted to adjust for the complex survey design, non-coverage and non-response, and are representative of women delivering a live infant in the respective state. The PRAMS protocol is approved by the CDC Institutional Review Board. Additional and detailed information about the PRAMS methodology and surveys can be found elsewhere [12]. For the present analyses (which were not pre-registered), we used Phase 8 (2016) data. We restricted these data to five states (Alaska, Colorado, Maine, Michigan and Washington) that included questions on SLEs and cannabis use, and also met a 65% response threshold as required by the CDC.
Survey respondents
Survey respondents were excluded if they had missing data on stress, cannabis use variables or maternal demographic covariates of interest (maternal age, race, Hispanic ethnicity, education, state of residence and cigarette smoking) (Fig. 1). These demographic covariates were selected a priori based on existing literature as confounders to be controlled for in analyses. In the Aim 2 analyses, we also excluded respondents who were defined as non-users (i.e. those who reported not using cannabis prior to pregnancy). In the Aim 3 analyses, we further restricted our analysis to only those respondents who reporting using any cannabis during the 12 months before becoming pregnant but who quit use during pregnancy.
Figure 1.
Study sample selection using Pregnancy Risk Assessment Monitoring System (PRAMS) Phase 8 (2016) data set
Measures
Stressful life events (SLEs), the primary exposure of interest, were measured in PRAMS using a 14-item subset of the Modified Life Events Inventory (see Table 2 for a list of the items; Cronbach’s alpha = 0.69) [13,14]. Respondents endorsed each SLE that occurred during the 12 months prior to delivery. We assessed each stressful item as a binary event (yes/no) as well as a cumulative SLE score where the presence of each item contributed one point (0–14 points).
Table 2.
Adjusted odds ratios and 95% confidence intervals between stressful life events and cannabis use at different time-points.a
| Cannabis use before pregnancyb | Continued cannabis use during pregnancyb | Postpartum cannabis relapsec | |
|---|---|---|---|
|
|
|||
| n = 6061 | n = 997 | n = 634 | |
|
| |||
| Cumulative stressful life eventsa (0–14), effect for + 1 SLE | 1.16 (1.09, 1.22) | 1.15 (1.05, 1.26) | 1.04 (0.90, 1.20) |
| Cumulative stressful life eventsa (0–14), categorized | |||
| 0 | Reference | Reference | Reference |
| 1–2 | 1.18 (0.87, 1.60) | 1.23 (0.62,2.43) | 0.74 (0.33, 1.67) |
| 3–5 | 1.91 (1.37, 2.66) | 1.96 (0.99,3.86) | 1.24 (0.50, 3.03) |
| 6 or more | 2.83 (1.85, 4.35) | 2.64 (1.21, 5.77) | 1.13 (0.38, 3.38) |
| Stressful life eventsa (yes versus no) | |||
| A close family member was very sick and had to go into the hospital | 1.37 (1.07, 1.77) | 1.11 (0.69, 1.77) | 0.62 (0.30, 1.29) |
| I got separated or divorced from my husband or partner | 1.41 (0.89, 2.24) | 1.36 (0.66,3.14) | 0.77 (0.34, 1.72) |
| I moved to a new address | 1.32 (1.04, 1.67) | 1.36 (0.86,2.15) | 0.80 (0.44, 1.47) |
| I was homeless or had to sleep outside, in a car, or in a shelter | 1.68 (0.96, 2.95) | 1.50 (0.74, 3.06) | 1.43 (0.36, 5.68) |
| My husband or partner lost their job | 1.81 (1.27, 2.57) | 2.19 (1.21, 3.96) | 0.85 (0.29, 2.46) |
| I lost my job even though I wanted to go on working | 1.34 (0.93, 1.93) | 1.79 (0.99,3.21) | 1.25 (0.55, 2.85) |
| My husband, partner or I had a cut in work hours or pay | 1.39 (1.04, 1.85) | 1.30 (0.77, 2.21) | 1.43 (0.68, 3.03) |
| I was apart from my husband or partner due to military deployment or extended work-related travel | 0.88 (0.54, 1.45) | 0.59 (0.21, 1.66) | 1.86 (0.61, 5.70) |
| I argued with my husband or partner more than usual | 1.32 (1.01, 1.72) | 1.06 (0.65, 1.72) | 1.29 (0.69, 2.40) |
| My husband or partner said they did not want me to be pregnant | 1.77 (1.14, 2.74) | 1.41 (0.65, 3.08) | 2.86 (1.10, 7.42) |
| I had problems paying the rent, mortgage, or other bills | 1.27 (0.94, 1.71) | 1.89 (1.12, 3.19) | 1.86 (0.88, 4.00) |
| My husband, partner or I went to jail | 2.16 (1.30, 3.62) | 1.92 (0.94, 3.92) | 0.37 (0.13, 1.00) |
| Someone very close to me had a problem with drinking or drugs | 1.59 (1.20, 2.10) | 1.60 (0.94, 2.74) | 1.85 (0.86, 3.98) |
| Someone very close to me died | 1.52 (1.15, 2.00) | 1.96 (1.20, 3.21) | 0.71 (0.35, 1.42) |
OR = odds ratio; CI = confidence interval; SLE = stressful life event.
14-item stressful life events a subset ofModified Life Events Inventory;
ORs and 95% CIs have been weighted to account for the PRAMS survey design and the statistical weighting of the data, and adjusted for maternal age, race, Hispanic ethnicity, education, state of residence and cigarette smoking during the same time-period. ORs significant at the 0.05 level are shown in bold type;
ORs and 95% CIs have been weighted to account for the PRAMS survey design and the statistical weighting of the data and adjusted for maternal age and state of residence. ORs significant at the 0.05 level are shown in bold type.
Cannabis use, the primary outcome of interest, was assessed differently by state. Of the five states in the analysis, Alaska, Colorado, Maine and Michigan asked: ‘During any of the following time-periods did you use marijuana or hash in any form?’ Respondents were classified as users if they endorsed use at the following time-points: (a) before pregnancy: ‘During the 12 months before I got pregnant’ (Alaska, Maine and Michigan only) (b) during pregnancy: ‘During my most recent pregnancy’ and (c) postpartum: ‘Since my new baby was born’. Colorado used a modified version of the above question only for the time-period before pregnancy that asked responders to indicate marijuana use in the 3 months before pregnancy. In contrast, Washington asked a modified version of the question asked in Alaska, Colorado, Maine and Michigan for all time-periods for marijuana use only (i.e. not hash). Respondents in Washington were classified as users if they endorsed use at the following time-points: (a) before pregnancy: ‘12 months before pregnancy’ (b) during pregnancy: ‘During pregnancy’ and (c) postpartum: ‘Since the baby was born’.
The following variables were included as covariates: maternal age (≤ 17, 18–19, 20–24, 25–29, 30–34 and ≥ 35 years), race (white, black, Asian/Pacific Islander, American Indian/Alaskan Native, other), ethnicity (Hispanic, non-Hispanic), highest education completed [less than high school diploma (0–11 years of education), high school diploma (12 years of education) and at least some college education (13+ years of education)], marital status (married, other), state of residence (Alaska, Colorado, Michigan, Maine, Washington) and cigarette smoking before, during and after pregnancy (yes, no). Additionally, the following variables were included as sample descriptors: gestational age at delivery (< 37 weeks, ≥ 37 weeks), birth weight (< 2500 g, ≥ 2500 g), small for gestational age (based on 10th percentile), ever breastfed, weeks of breastfeeding, parity (primiparous, multiparous), type of insurance (Medicaid or private insurance/other), physical abuse during pregnancy by partner (yes, or no/not reported) and mode of survey participation (mail, telephone).z
Statistical analysis
The unweighted distributions of demographic characteristics and SLEs within the study sample were described using percentages, frequency, means and standard deviations, stratified by reported cannabis use. Separate logistic regression models were used to estimate adjusted associations between SLEs (a continuous cumulative score, a categorized cumulative score and each of the 14 SLEs separately) and cannabis use outcomes. Linearity of a continuous cumulative SLE score was assessed using restricted cubic spline functions to perform Wald χ2 tests of linearity [15–20]. If linearity was not met, only cumulative scores categorized using pre-established definitions recommended by the PRAMS Working Group (0, 1–2, 3–5, 6 or more SLEs) were used. All logistic regression analyses were weighted for the complex survey design [21]. Adjusted associations were determined by including a priori covariates in the model based on theoretical significance. Adjusted odds ratios (aORs) between SLEs and cannabis use and the 95% confidence intervals (CIs) were adjusted for all covariates in the model. As part of the main analyses, the cross-sectional associations between SLEs and cannabis use before pregnancy (Aim 1) were adjusted for cigarette smoking during the same time-period, maternal characteristics (age, race, ethnicity, level of education) and state. Associations between SLEs and continued cannabis use during pregnancy (Aim 2) were adjusted for the same six covariates as in Aim 1. As the number of reports of cannabis use relapse after the baby was born was anticipated to be reduced compared to cannabis reports during other time-periods, associations between SLEs and cannabis use relapse after the baby was born (Aim 3) were only adjusted for maternal age and state of residence, as we considered the most important confounders of the six potential confounders considered in Aims 1 and 2. In supplemental analyses, the cross-sectional associations between SLEs and cannabis use during and after pregnancy, adjusted for the same six covariates of Aims 1 and 2, were described.
A sensitivity analysis was performed to determine the influence of missing responses to cannabis usage measures. Self-reported drug use has been reported as an unreliable measure, especially when stigma and discrimination are feared, such as during pregnancy [22]. Logistic regression models were re-run, re-categorizing all missing responses as a positive report of cannabis usage at each time-point (before, during or after baby was born). Results significantly different than the primary results were reported. All statistical tests were performed in 2018–19 with SAS version 9.4, used a two-tailed significance level of 0.05 and accounted for the PRAMS complex sampling strategy.
RESULTS
Study sample
A total of 31 642 respondents were included in PRAMS Phase 8 in 2016 (Fig. 1). Of these, 6061 met our eligibility criteria and were included in analyses. Overall, regardless of time-point, cannabis users tended to be younger, less educated, multiparous, were more likely to smoke cigarettes and experience physical abuse during pregnancy (Table 1). Wald χ2 tests for linearity indicated that the cumulative number of SLEs was linearly associated with cannabis use (Supporting information, Table S1).
Table 1.
Distribution of demographics and experiences among women stratified by cannabis use during different time-periods, Pregnancy Risk Assessment Monitoring System (PRAMS) 2016.
| Cannabis use before pregnancy |
Continued cannabis use during pregnancy |
Postpartum cannabis relapse |
||||
|---|---|---|---|---|---|---|
| Reported use | Reported no use | Reported use | Reported no use | Reported use | Reported no use | |
|
|
||||||
| n (%) | n (%) | n (%) | n (%) | n (%) | n (%) | |
|
| ||||||
| Sample | 997(16.4) | 5064 (83.6) | 363 (36.4) | 634(63.6) | 147 (23.2) | 487 (76.8) |
| Cumulative stressful life eventsa (0–14), mean (SD) | 2.98 (2.51) | 1.67 (1.86) | 3.45 (2.64) | 2.71 (2.40) | 2.88 (2.54) | 2.66 (2.35) |
| Cumulative stressful life eventsa (categorized) | ||||||
| 0 | 165 (16.6) | 1658 (32.7) | 48 (13.2) | 117 (18.4) | 28(19.0) | 89 (18.3) |
| 1–2 | 341 (34.2) | 2124 (41.9) | 104(28.6) | 237(37.4) | 47(32.0) | 190(39.0) |
| 3–5 | 325 (32.6) | 1047 (20.7) | 132 (36.4) | 193 (30.4) | 50 (34.0) | 143 (29.4) |
| 6 or more | 166 (16.6) | 235 (4.6) | 79 (21.8) | 87(13.7) | 22 (15.0 | 65 (13.4) |
| Age (years) | ||||||
| < 17 | 14(1.4) | 47 (0.9) | 4(1.1) | 10(1.6) | 0 | 10(2.0) |
| 18–19 | 74 (7.4) | 143 (2.8) | 15 (4.1) | 59 (9.3) | 16 (10.9) | 43 (8.8) |
| 20–24 | 243 (24.4) | 824(16.3) | 98 (27.0) | 145 (22.9) | 38 (25.8) | 107 (22.0) |
| 25–29 | 309 (31.0) | 1584 (31.3) | 119(32.8) | 190(30.0) | 44 (29.9) | 146 (30.0) |
| 30–34 | 238 (23.9) | 1486 (29.3) | 88 (24.2) | 150(23.7) | 31 (21.1) | 119 (24.4) |
| 35+ | 119 (11.9) | 980 (19.4) | 39 (10.7) | 80(12.6) | 18 (12.2) | 62 (12.7) |
| Highest education completed | ||||||
| Less than high school diploma | 128 (12.8) | 445 (8.8) | 55 (15.2) | 73 (11.5) | 17(11.6) | 56 (11.5) |
| High school diploma | 330 (33.1) | 1098 (21.7) | 144(39.7) | 186 (29.3) | 57 (38.8) | 129 (26.5) |
| At least some college | 539 (54.1) | 3521 (69.5) | 164 (45.2) | 375 (59.2) | 73 (49.7) | 302 (62.0) |
| Marital status | ||||||
| Married | 405 (40.6) | 3497 (69.1) | 119(32.8) | 286 (45.1) | 65 (44.2) | 221 (45.4) |
| Other | 592 (59.4) | 1558 (30.8) | 244 (67.2) | 348 (54.9) | 82 (55.8) | 266 (54.6) |
| Missing | 0 | 9 (0.2) | 0 | 0 | 0 | 0 |
| Race | ||||||
| White | 581 (58.3) | 3117 (61.6) | 204 (56.2) | 377 (59.5) | 87 (59.2) | 290 (59.6) |
| Black | 163 (16.4) | 808 (16.0) | 46 (12.7) | 117 (18.4) | 25 (17.0) | 92 (18.9) |
| API | 21 (2.1) | 343 (6.8) | 7(1.9) | 14(2.2) | 1 (0.7) | 13 (2.7) |
| AI/AN | 128 (12.8) | 340 (6.7) | 62 (17.1) | 66 (10.4) | 18 (12.2) | 48 (9.9) |
| Other | 104 (10.4) | 456 (9.0) | 44 (12.1) | 60 (9.5) | 16 (10.9) | 44 (9.0) |
| Hispanic ethnicity | ||||||
| No | 917 (92.0) | 4425 (87.4) | 337(92.8) | 580(91.5) | 137 (93.2) | 443 (91.0) |
| Yes | 75 (7.5) | 606 (12.0) | 25 (6.9) | 50(7.9) | 9(6.1) | 41 (8.4) |
| Unknown | 5 (0.5) | 33 (0.6) | 1 (0.3) | 4 (0.6) | 1 (0.7) | 3 (0.6) |
| State | ||||||
| Alaska | 196 (19.7) | 811 (16.0) | 92 (25.3) | 104(16.4) | 31 (21.1) | 73 (15.0) |
| Colorado | 232 (23.3) | 1277 (25.2) | 83 (22.9) | 149 (23.5) | 30 (20.4) | 119 (24.4) |
| Maine | 164 (16.4) | 607 (12.0) | 66 (18.2) | 98 (15.5) | 31 (21.1) | 67 (13.8) |
| Michigan | 227 (22.8) | 1406 (27.8) | 61 (16.8) | 166 (26.2) | 36 (24.5) | 130(26.7) |
| Washington | 178 (17.8) | 963 (19.0) | 61 (16.8) | 117 (18.4) | 19(12.9) | 98 (20.1) |
| Smoked cigarettes up to 12 months before pregnant | ||||||
| Yes | 494 (49.6) | 757 (15.0) | 229 (63.1) | 265 (41.8) | 75 (51.0) | 190(39.0) |
| No | 503 (50.4) | 4307 (85.0) | 134(36.9) | 369 (58.2) | 72 (49.0) | 297 (61.0) |
| Smoked cigarettes during pregnancy | ||||||
| Yes | 231 (23.2) | 370 (7.3) | 135 (37.2) | 96 (15.1) | 34(23.1) | 62 (12.7) |
| No | 766 (76.8) | 4694 (92.7) | 228 (62.8) | 538 (84.9) | 113 (76.9) | 425 (87.3) |
| Smoked cigarettes since baby was born | ||||||
| Yes | 335 (33.6) | 515 (10.2) | 177 (48.8) | 158 (24.9) | 61 (41.5) | 97 (19.9) |
| No | 662 (66.4) | 4549 (89.8) | 186 (51.2) | 476 (75.1) | 86 (58.5) | 390(80.1) |
| Parity | ||||||
| Primiparous | 491 (49.2) | 1950(38.5) | 151 (41.6) | 340 (53.6) | 69 (46.9) | 271 (55.7) |
| Multiparous | 504(50.6) | 3107 (61.4) | 212 (58.4) | 292 (46.1) | 78 (53.1) | 214 (43.9) |
| Missing | 2 (0.2) | 7(0.1) | 0 | 2 (0.3) | 0 | 2 (0.4) |
| Method of payment | ||||||
| Medicaid | 595 (59.7) | 1865 (36.8) | 249 (68.6) | 346 (54.6) | 81 (55.1) | 265 (54.4) |
| Other | 394(39.5) | 3174(62.7) | 109 (30.0) | 285 (45.0) | 65 (44.2) | 220 (45.2) |
| Missing | 8 (0.8) | 25 (0.5) | 5 (1.4) | 3 (0.5) | 1 (0.7) | 2 (0.4) |
| Physical abuse during pregnancy | ||||||
| Yes | 42 (4.2) | 49 (1.0) | 22 (6.1) | 20 (3.2) | 5 (3.4) | 15 (3.1) |
| No/not reported | 955 (95.8) | 5015 (99.0) | 341 (93.4) | 614 (96.8) | 142 (96.6) | 472 (96.9) |
| Gestational age at delivery | ||||||
| < 37 weeks | 196(19.7) | 999 (19.7) | 77 (21.2) | 119 (18.8) | 28 (19.1) | 91 (18.7) |
| ≥ 37 weeks | 799 (80.1) | 4057(80.1) | 284(78.2) | 515 (81.2) | 119 (81.0) | 396 (81.3) |
| Missing | 2 (0.2) | 8 (0.2) | 2 (0.6) | 0 | 0 | 0 |
| Birthweight | ||||||
| < 2500g | 290(29.1) | 1281 (25.3) | 125 (34.4) | 165 (26.0) | 41 (27.9) | 124(25.5) |
| ≥ 2500 g | 707 (70.9) | 3784(74.7) | 238 (65.6) | 469 (74.0) | 106 (72.1) | 363 (74.5) |
| Small for gestational age (based on 10th percentile) | ||||||
| Yes | 191 (19.2) | 780 (15.4) | 90 (24.8) | 101 (15.9) | 26 (17.7) | 75 (15.4) |
| No | 765 (76.7) | 4029 (79.6) | 258 (71.1) | 507 (80.0) | 113 (76.9) | 394(80.9) |
| Missing | 41 (4.1) | 255 (5.0) | 15 (4.1) | 26 (4.1) | 8 (5.4) | 18 (3.7) |
| Ever breastfed | ||||||
| Yes | 848 (85.1) | 4553 (89.9) | 280 (77.1) | 568 (89.6) | 129 (87.8) | 439 (90.1) |
| No | 110(11.0) | 408 (8.1) | 60(16.5) | 50 (7.9) | 10(6.8) | 40 (8.2) |
| Missing | 39 (3.9) | 103 (2.0) | 23 (6.3) | 16 (2.5) | 8 (5.4) | 8(1.7) |
| Weeks of breastfeeding among women who ever breastfed and are currently not breastfeeding, mean (SD) | 6.68(5.4) | 7.38 (6.0) | 6.08 (5.0) | 7.06 (5.6) | 7.09 (5.6) | 7.04(5.6) |
| Mode of participation | ||||||
| 716 (71.8) | 3665 (72.4) | 242 (66.7) | 474 (74.8) | 101 (68.7) | 3 73 (76.6) | |
| Phone | 281 (28.2) | 1399 (27.6) | 121 (33.3) | 160 (25.2) | 46 (31.3) | 114(23.4) |
SD = standard deviation; n = frequency; API = Asian/Pacific Islander; AI/AN = American Indian/Alaskan Native. Percentages may not sum to 100 due to rounding.
14-item stressful life events are a subset of Modified Life Events Inventory.
Stressful life events and cannabis use before pregnancy
Overall, 16.4% (997/6061) reported using cannabis prior to pregnancy. Those who reported use had, on average, 1.31 more cumulative SLEs compared to those who did not report use (2.98 ± 2.51 versus 1.67 ± 1.86). Generally, experiencing greater numbers of cumulative SLEs (3–5 or ≥ 6) compared to no SLEs was associated with increased odds of cannabis use during pregnancy (OR = 1.91, 95% CI = 1.37–2.66; OR = 2.83, 95% CI = 1.85–4.35, respectively) (Table 2). Further, having experienced nine of the 14 specific SLEs in the year before childbirth (family member was sick, moved to new address, husband/partner lost job, husband/partner or the mother had reduced work or pay, argued with husband/partner more than usual, husband/partner did not want the pregnancy, husband/partner or the mother went to jail, someone close to the mother had alcohol/drug problem, someone close to the mother died) was associated with higher odds of cannabis use before pregnancy. This ranged from 32% higher odds of cannabis use for those who moved to a new address to 116% higher odds of cannabis use for those with a husband/partner or who themselves went to jail (Table 2).
Stressful life events and continued cannabis use during pregnancy
Among those who reported cannabis use prior to pregnancy, 36.4% continued to use cannabis during pregnancy (363/997). Those who maintained cannabis use during pregnancy had, on average, 0.74 more cumulative SLEs compared to those who quit using cannabis during pregnancy (3.45 ± 2.64 versus 2.71 ± 2.40). Compared to those who reported experiencing no SLEs, those who experiencing six or more had increased odds of continued cannabis use (OR = 2.64, 95% CI = 1.21, 5.77). Further, among those who used cannabis prior to pregnancy, those who experienced any of three of 14 SLEs in the year prior (husband/partner lost their job, problems paying bills, someone close to the mother died) had higher odds of continued cannabis use during pregnancy. This ranged from 89% higher odds associated with mother paying bills to 119% higher odds associated with partner job loss.
Stressful life events and postpartum cannabis relapse
Among those who reported cannabis use prior to pregnancy and abstained during pregnancy, 23.2% relapsed to cannabis use after pregnancy (147/634). Cumulative SLE scores were similar between those who maintained abstinence postpartum and those who relapsed (n = 634; 2.66 ± 2.35 versus 2.88 ± 2.54). One SLE (husband/partner said they did not want the pregnancy) was associated with postpartum cannabis relapse (186% higher odds of cannabis relapse).
Supplemental analyses
Among all respondents, regardless of cannabis use prior to pregnancy, a total of 6.4% (385/6016) reported use during pregnancy and 7.8% (473/6061) reported cannabis use during the postpartum period (Supporting information, Table S2). Those who experienced any of 11 of the 14 SLEs had significantly higher odds of using cannabis during pregnancy, ranging from 49% higher odds associated with a family member being hospitalized to 192% higher odds associated with husband/partner or the mother herself going to jail (Supporting information, Table S3). Postpartum cannabis use was associated with six of the 14 SLEs, ranging from 61% higher odds of use among those who reported a cut in work hours or pay to 147% increased odds of use if someone very close to them had a problem with drinking or drugs.
Sensitivity analysis
Of the 136 respondents who were removed from the primary analyses due to at least one missing response to the cannabis use questions, 99 respondents had missing responses for all time-points and 37 respondents had a non-missing response to at least one of the other cannabis use questions (Fig. 1). The missing responses were recoded to endorse cannabis use. All logistic models were re-run under this new assumption (Supporting information, Table S4). The directions of effect of these aORs were similar to that reported in primary analysis, although the magnitudes were attenuated. Specifically, for the odds of cannabis use during pregnancy, three SLEs lost their significance (moved to a new address, cut in work hours or pay, argued with husband/partner more) and one gained significance (problems paying the rent, mortgage or other bills). For the odds of continued cannabis use during pregnancy, two SLEs lost their significance (husband/partner lost job, someone very close died) and one gained significance (husband/partner or mother herself went to jail). For the odds of postpartum cannabis relapse, two SLEs gained significance (someone very close to mother had a problem with drinking or drugs, husband/partner or mother went to jail).
DISCUSSION
Using a cross-sectional, population-based surveillance system, we observed that numerous SLEs were significantly associated with perinatal use of cannabis. Specifically, of 14 SLEs examined, nine were significantly associated with odds of pre-pregnancy cannabis use, four with continued cannabis use during pregnancy and one with postpartum cannabis relapse. Respondents with six or more SLEs were at 183 and 164% higher odds of reporting the use of cannabis before pregnancy, and continued use during pregnancy, respectively, compared to those who reported experiencing no SLEs. However, those who endorsed one to two SLEs did not significantly differ from those who experienced no SLEs, suggesting that there may be a threshold effect. There were no differences in odds of postpartum cannabis relapse among those who endorsed six or more SLEs compared to those who endorsed none. Further, in supplemental analyses, examining associations between SLEs and cannabis use during and after pregnancy revealed results similar to what was reported for cannabis use before pregnancy. Overall, these observations indicate that SLEs experienced in the year prior to childbirth are associated with perinatal cannabis use, perhaps especially pre-pregnancy cannabis use and, less so, use during pregnancy. This observation is similar to our prior publication examining the link between SLEs and cigarette smoking during the perinatal period [11].
Although a relatively small percentage (6.4%) of women used cannabis during pregnancy, this number has increased from previous observations. Specifically, Ko and colleagues utilized the same surveillance system to estimate cannabis use during pregnancy in 2009–11, observing a prevalence of 4.2% [1]. This suggests that cannabis use during pregnancy has increased by approximately 35% during the past 5–7 years; which concurs with prior literature indicating that, overall, in the general population cannabis use has become more common [23]. Therefore, it will become increasingly important for women and families to be informed of the adverse health consequences of in-utero exposure to cannabis as recommended [4,5,7]. Also, treatment options need to be provided for women who have difficulty abstaining from cannabis during pregnancy, such as women who report stressful life events, as our data appear to suggest. First, health-care providers should counsel and educate women about the potential harms of cannabis use during pregnancy [7,24] and recommend against its use during pregnancy. However, to date, there is a dearth of literature on how to best treat cannabis use during pregnancy. Secondly, evidence from the tobacco literature demonstrates that compared to psychotherapy, providing financial incentives for smoking cessation (e.g. contingency management) early during pregnancy increases the odds of cigarette smoking abstinence before delivery (OR = 1.55, 95% CI = 1.19, 2.02) [25]. This approach may be especially salient to cannabis use during pregnancy, as the two SLEs that focused on financial hardships (i.e. ‘reduced hours/pay’ and ‘had bills could not pay’) were associated with a 54 and 91% higher odds of use during pregnancy. Similar observations were made for the two SLEs that asked about husband/partner or mother job loss; indicating a 143 and 196% higher odds of endorsing cannabis use. Further, three of the six SLEs associated with greater risk of cannabis use during the postpartum period are also associated with financial hardships (i.e. ‘partner lost job’, ‘reduced hours/pay’ and ‘had bills that she could not pay’). Overall, counseling and educating women about cannabis use during pregnancy and providing treatment programs, perhaps in the form of contingency management, may prevent poor infant health outcomes by preventing in-utero exposure to cannabis.
There are a number of limitations that should be noted. First, all data are based on self-report, which is subject to recall bias, social stigma bias and other forms of error. However, the overall results of our sensitivity analysis generally support our primary analyses in that SLEs are associated with odds of perinatal cannabis use; associations between use and individual SLEs differed slightly. This suggests that our results are robust and the missing cannabis responses did not significantly impact our overall conclusion. Validity of self-reported measures of substance use among pregnant women is low; however, sensitivity has been reported to be higher for cannabis use compared to other illicit drug use and also for ‘socially acceptable’ drugs (such as cannabis) compared to highly stigmatized drugs such as heroin and cocaine [22,26,27]. Secondly, given the cross-sectional nature of this study, both cannabis use and SLEs were assessed at the same time-point, making them susceptible to a number of biases including recall bias and survivor bias. This may have influenced the accuracy of reporting. Thirdly, we treated each SLE as if they were equal when we calculated our summary score. In reality, some SLEs may be less stressful than others (e.g. ‘I moved to a new address’ versus ‘Someone close to me died’). This lack of weighting may have altered the observed relationship between the summary SLE score and odds of cannabis use. Fourthly, statistical issues of multiple testing and power should be considered. The regression analyses that were performed did not adjust for multiple comparisons, which increases the likelihood of false positives. For analyses of cannabis relapse, the number of events of reported relapse was smaller compared to other outcomes, and may have affected our power to detect associations between SLEs (some of which are generally uncommon) and cannabis relapse. Fifthly, the generalizability of these observations may be limited. Only five states were included in this work. Specifically, cannabis was legalized for recreational use in Colorado and Washington in 2012, followed by Alaska in 2014, then Maine in 2016 (during data collection), then Michigan in 2018 (after collection). It is possible that the differences in legalization of cannabis for recreational use may have resulted in differences in self-reported use and changes in behavior. The validity of self-reported cannabis uses among pregnant women living in areas where cannabis has been legalized (for both medical and social purposes) has not been previously reported, although it is likely that accuracy of self-reported use is higher in areas where it is legalized. Further, due to limited sample size and/or data, we were unable to explore other potentially important confounders (e.g. breastfeeding, adverse birth outcomes) and contexts (e.g. pregnancy loss). Next, SLEs reported were based on the year prior to childbirth. Given the cross-sectional nature of the survey, it is impossible to know the temporality and causality between SLEs and cannabis use. Relatedly, the SLE focused on jail during the past year may be especially prone to error, given that it does not distinguish between mother versus husband/partner, and it may have acted as a protective factor simply because of a lack of cannabis access if mother was jailed. Lastly, these analyses were not pre-registered and the results should be considered exploratory.
In conclusion, SLEs during the year prior to childbirth are related to increased odds of cannabis use during the perinatal period, especially pre-pregnancy use. Additional research is needed to explore the causality of this relationship. Further research is needed to explore how cannabis cessation interventions may target SLEs to decrease perinatal cannabis use.
Supplementary Material
Table S1 Results of tests for linear associations performed using restricted cubic spline functions generated using the %RCS SAS macroa
Table S2 Distribution of demographics and experiences among women who reported cannabis use during and since pregnancy stratified by cannabis use, PRAMS 2016.
Table S3 Adjusted odds ratios and 95% confidence intervals between stressful life events and cannabis use during and since baby was borna,b, N = 6061.
Table S4 Sensitivity Analysis- Adjusted odds ratios and 95% confidence intervals between stressful life events and cannabis usage, recoding missing cannabis use question as ‘yes‘ response (use before, continued use during or postpartum relapse) if non-missinga.
Acknowledgments
The authors received no financial support for the research. We would like to extend our thanks to the PRAMS Working Group for granting us access to this data, as well as their review of this manuscript. Additional thanks to Stephanie Mallahan for her assistance with editing and proofreading.
Footnotes
Declaration of interest
None.
Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.
References
- 1.Ko JY, Tong VT, Bombard JM, Hayes DK, Davy J, Perham-Hester KA Marijuana use during and after pregnancy and association of prenatal use on birth outcomes: a population-based study. Drug Alcohol Depend 2018; 187: 72–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Coleman-Cowger VH, Oga EA, Peters EN, Mark K Prevalence and associated birth outcomes of co-use of cannabis and tobacco cigarettes during pregnancy. Neurotoxicol Teratol 2018; 68: 84–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Merlob P, Stahl B, Klinger G For debate: does cannabis use by the pregnant mother affect the fetus and newborn? Pediatr Endocrinol Rev 2017; 15: 4–7. [DOI] [PubMed] [Google Scholar]
- 4.Gunn JKL, Rosales CB, Center KE, Nuñez A, Gibson SJ, Christ C, et al. Prenatal exposure to cannabis and maternal and child health outcomes: a systematic review and meta-analysis. BMJ Open 2016; 6: e009986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Jaques SC, Kingsbury A, Henshcke P, Chomchai C, Clews S, Falconer J, et al. Cannabis, the pregnant woman and her child: weeding out the myths. J Perinatol 2014; 34: 417–24. [DOI] [PubMed] [Google Scholar]
- 6.El Marroun H, Brown QL, Lund IO, Coleman-Cowger VH, Loree AM, Chawla D, et al. An epidemiological, developmental and clinical overview of cannabis use during pregnancy. Prev Med (Balt) 2018; 116: 1–5. [DOI] [PubMed] [Google Scholar]
- 7.American College of Obstetricians and Gynecologists (ACOG). ACOG Committee Opinion. Washington, DC: American College of Obstetricians and Gynecologists. 2017. https://www.acog.org/Clinical-Guidance-and-Publications/Committee-Opinions/Committee-on-Obstetric-Practice/Marijuana-Use-During-Pregnancy-and-Lactation [Google Scholar]
- 8.LeTendre ML, Reed MB The effect of adverse childhood experience on clinical diagnosis of a substance use disorder: results of a nationally representative study. Subst Use Misuse 2017; 52: 689–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.al’Absi M Stress response pathways, appetite regulation, and drug addiction. Biol Psychol 2018; 131: 1–4. [DOI] [PubMed] [Google Scholar]
- 10.al’Absi M, Lemieux A, Westra R, Allen S Early life adversity influences stress response association with smoking relapse. Psychopharmacology (Berl) 2017; 234: 3375–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Allen AM, Jung AM, Lemieux AM, Alexander AC, Allen SS, Ward KD, et al. Stressful life events are associated with perinatal cigarette smoking. Prev Med (Balt) 2019; 118: 264–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Shulman HB, Gilbert BC, Msphbrenda CG, Lansky A The pregnancy risk assessment monitoring system (PRAMS): current methods and evaluation of 2001 response rates. Public Health Rep 2006; 121: 74–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Newton RW, Webster PA, Binu PS, Maskrey N, Phillips AB Psychosocial stress in pregnancy and its relation to the onset of premature labour. BMJ 1979; 2: 411–3; http://www.ncbi.nlm.nih.gov/pubmed/486966 (accessed 26 October 2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lu MC, Chen B Racial and ethnic disparities in preterm birth: the role of stressful life events. Am J Obstet Gynecol 2004; 191: 691–9. [DOI] [PubMed] [Google Scholar]
- 15.Durrleman S, Simon R Flexible regression models with cubic splines. Stat Med 1989; 8: 551–61. [DOI] [PubMed] [Google Scholar]
- 16.Harrell FE Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York: Springer; 2001. [Google Scholar]
- 17.Fang J, Austin PC, Tu JV Test for Linearity between Continuous Confounder and Binary Outcome First, Run a Multivariate Regression Analysis Second. Available at: www.meduniwien.ac.at/msi/biometrie/programme/Rcs.htm (accessed 22 January 2020).
- 18.Kendzerska T, Gershon AS, Hawker G, Leung RS, Tomlinson G Obstructive sleep apnea and risk of cardiovascular events and all-cause mortality: a decade-long historical cohort study. PLOS Med 2014; 11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Croxford R Restricted Cubic Spline Regression: A Brief Introduction. Available at: http://pages.cs.wisc.edu/~deboor/draftspline.html (accessed 22 January 2020).
- 20.Desquilbet L, Mariotti F Dose–response analyses using restricted cubic spline functions in public health research. Stat Med 2010; 29; 10.1002/sim.3841. [DOI] [PubMed] [Google Scholar]
- 21.Cdc-drh-prams. Describing the PRAMS Sample Design for SUDAAN, SAS Complex Survey, SPSS Complex Samples Modules, and STATA. Available at: https://www.cdc.gov/prams/pdf/researchers/SUDAAN_SAS_SPSS_STATA_2013.pdf (accessed 8 March 2019).
- 22.Grekin ER, Svikis DS, Lam P, Connors V, Lebreton JM, Streiner DL, et al. Drug use during pregnancy: validating the drug abuse screening test against physiological measures. Psychol Addict Behav 2010; 24: 719–23; 10.1037/a0021741. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Frieden TR, Jaffe HW, Rasmussen SA, Azofeifa A, Mattson ME, Schauer G, et al. Morbidity and Mortality Weekly Report National Estimates of Marijuana Use and Related Indicators-National Survey on Drug Use Centers for Disease Control and Prevention MMWR Editorial and Production Staff (Serials) MMWR Editorial Board. 2016. Available at: https://www.cdc.gov/mmwr/volumes/65/ss/pdfs/ss6511.pdf (accessed 8 March 2019).
- 24.Holland CL, Rubio D, Rodriguez KL, Kraemer KL, Day N, Arnold RM, et al. Obstetric health care providers’ counseling responses to pregnant patient disclosures of marijuana use. Obstet Gynecol 2016; 127: 681–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wilson SM, Newins AR, Medenblik AM, Kimbrel NA, Dedert EA, Hicks TA, et al. Contingency management versus psychotherapy for prenatal smoking cessation: a meta-analysis of randomized, controlled trials. Womens Health Issues 2018; 28: 514–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Garg M, Garrison L, Leeman L, Hamidovic A, Borrego M, Rayburn WF, et al. Validity of self-reported drug use information among pregnant women. Matern Child Health J 2016; 20: 41–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Golub A, Liberty HJ, Johnson BD The variation in Arrestees’ disclosure of recent drug use across locations, drugs, and demographic characteristics. J Drug Issues 2005; 35: 917–40; Available at: http://www.ncbi.nlm.nih.gov/pubmed/18852842 (accessed 8 March 2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1 Results of tests for linear associations performed using restricted cubic spline functions generated using the %RCS SAS macroa
Table S2 Distribution of demographics and experiences among women who reported cannabis use during and since pregnancy stratified by cannabis use, PRAMS 2016.
Table S3 Adjusted odds ratios and 95% confidence intervals between stressful life events and cannabis use during and since baby was borna,b, N = 6061.
Table S4 Sensitivity Analysis- Adjusted odds ratios and 95% confidence intervals between stressful life events and cannabis usage, recoding missing cannabis use question as ‘yes‘ response (use before, continued use during or postpartum relapse) if non-missinga.