Effects of Smoking Reduction and Cessation on Birth Outcomes in a Scheduled Gradual Reduction Cessation Trial

Abstract

Introduction

Smoking during pregnancy can affect infant birthweight. We tested whether an intervention that promoted scheduled gradual reduction improved birth outcomes among pregnant women who smoked. We also examined race differences in birth outcomes.

Methods

We conducted a 2-arm randomized controlled trial where pregnant women who smoked received either SMS text-delivered scheduled gradual smoking reduction (SGR) program plus support texts or support messages only throughout their pregnancy. The outcomes for this paper were birth outcomes including birth weight and gestational age obtained from chart review. Analyses were conducted using chi-square and t-tests in SAS.

Results

We approached 2201 pregnant women with smoking history. Of the 314 women recruited into the study, 290 completed a medical release form (92%). We did not find any significant differences in birth outcomes by arm or race. The majority of participants reduced smoking by the 80%. Women who reduced more than 50% of their baseline cigarettes per day had a birth weight increase of 335 g compared to those that did not (p = 0.05). The presence of alcohol/drug use in prenatal visit notes was associated with low infant birth weight (p = 0.05).

Discussion

The scheduled gradual reduction intervention did not improve birth outcomes. Additional research is needed to help improve birth outcomes for pregnant women who engage in tobacco and illicit substance use.

Introduction

In 2016, one in 14 women who gave birth in the United States reported smoking during pregnancy (Drake et al., 2018; Kondracki, 2019). Smoking during pregnancy is related to complications and poor birth outcomes, including miscarriage, low birth weight, preterm birth, stillbirth, and neonatal death (US Department of Health and Human Services (USDHHS), 2014). For example, the prevalence of low birthweight pregnancies among smokers is almost double at 14% compared to the national average of 8% (Child Trends, https:www.childtrends.org). While the rate of smoking in pregnancy in the U.S. has declined over recent decades, women in vulnerable populations are more likely to continue to smoke, namely women with less education, lower socioeconomic status, and substance misuse (Drake et al., 2018; Oga et al., 2018; Scherman et al., 2018; US Department of Health and Human Services (USDHHS), 2014 ). There are also links between education level and socioeconomic status and poor birth outcomes in the United States (Mohlman & Levy, 2016). Further, there are known race differences in birth outcomes, which may be partly attributable to smoking in pregnancy with non-Hispanic Black women being at the highest risk, independent of socioeconomic status (Moore & Zaccaro, 2000).

Previous psychosocial interventions, including behavioral counseling and financial incentives, resulted in increased rates of smoking cessation and reduced the proportion of low birthweight babies (Chamberlain et al., 2017; Zhang et al., 1992, 2017). However, overall the quit rate in these studies remains low (Chamberlain et al., 2017). A higher proportion of pregnant smokers reduced their cigarette consumption if they were unable to quit (Chamberlain et al., 2017). Studies exploring the impact of reducing smoking on birth outcomes, specifically birthweight and gestational age, are mixed. One investigator identified a dose–response relationship between the number of cigarettes per day and birthweight deficits and recommended pregnant smokers reduce to less than six cigarettes per day (Kataoka et al., 2018). Others have demonstrated that daily low cigarette consumption is still associated with reduced birthweight compared to cessation (Berlin et al., 2017; England et al., 2001a). Further, numerous studies have shown women who smoke beyond the second trimester remain at an increased risk of having an infant born low birth weight or preterm compared to those who quit (England et al., 2001b; Kataoka et al., 2018; Wallace et al., 2017).

Perhaps, because of these mixed findings, few studies have examined the impact of reduction-based approaches to promoting smoking cessation on birth outcomes. We designed a trial to both promote cessation and reduction via scheduled gradual reduction (SGR). SGR has been shown to increase rates of smoking cessation in non-pregnant smokers by gradually reducing nicotine dependence and unlinking cue-induced smoking. We reduced the number of daily cigarettes smoked on a pre-determined schedule over three to five weeks. We texted them at scheduled times when we asked them to smoke and asked them not to smoke unless we texted them. We delivered the intervention via text messages, which provided a simple, scalable way to send participants their smoking schedule and pregnancy-appropriate support messages in real-time throughout their pregnancy (Pollak et al., 2019).

One aim of Baby Steps was to determine whether a scheduled gradual reduction intervention improved birth outcomes. We were promoting reduction, and given women who smoke < 5 cigarettes per day have similar birth outcomes, (Kataoka et al., 2018) we wanted to test whether our intervention could improve outcomes even when women did not quit altogether. We also examined the impact of pregnancy smoking behaviors, including smoking reduction, and other known risk factors on birth outcomes.

Methods

Participants and Procedures

The methods and outcomes are presented in detail elsewhere. Briefly, this was a two-arm randomized controlled trial to promote smoking cessation among pregnant women. A secondary outcome was birth outcomes. We recruited participants for the Baby Steps trial from 14 prenatal clinics in Central North Carolina March 2014 and December 2017 (Pollak et al., 2019). The study was approved by the Duke University Health System IRB. We consented women who were between 10 and 30 weeks pregnant; were currently smoking ≥ 3 cigarettes per day; had smoked at least 100 cigarettes in their lifetime; aged 18 years and older; were enrolled in prenatal care; interested in quitting smoking; spoke English; with a capacity to consent. We excluded women who were currently using NRT and/or had evidence of unstable cognitive health problems. Women were asked to complete a medical release form at the time of consent for birth outcomes. Of the 314 women recruited into the study (Fig. 1), 290 completed a medical release form (92%) with singleton births. We were unable to complete medical release forms for 19 women (5 SGR + SMS and 14 SMS), and 5 women with multiple births were also removed (1 SGR + SMS and 4 SMS). We randomized enrolled participants into two arms: support messages only vs. support messages and SGR (See Fig. 1). The intervention procedures are described elsewhere (Pollak et al., 2019).

Figure 1.

Smoking Status

Participants completed questionnaires at enrollment during the screener and baseline survey, in the last part of their third trimester, and at three months postpartum. These questionnaires collected pre-pregnancy smoking behavior, “On average how many cigarettes per day did you smoke before you discovered you were pregnant?” as well as their current smoking behavior, “Have you smoked in the past 7 days?” We biochemically validated smoking cessation in the late third trimester and at 3-months postpartum using salivary cotinine. To assess 50% reduction, we assessed the absolute difference in number of cigarettes smoked from prepregnancy to end pregnancy, which is consistent with other SGR studies (Lindson et al., 2010).

Birth Information

We abstracted birth outcomes including infant sex, weight (pounds and ounces/ grams), length (inches/centimeters), adverse events (i.e., pre-eclampsia, preterm delivery, congenital malformation), prenatal diagnosis, postnatal diagnosis, type of delivery, gestational age (EGA) (weeks and days), Apgar scores, and NICU admissions from all available medical records for the pregnant mother and the infant (or infants) following delivery. We also recorded evidence of alcohol/drug use documented in the patient’s chart from pregnancy prenatal care and/or labor & delivery (yes/no). Women completed a baseline questionnaire to share demographics and pregnancy history. We abstracted birth weight as pounds and ounces was converted to grams. Low birth weight was defined as less than 2500 g. Preterm birth was defined as earlier than 37 weeks gestation (yes/no). We derived small for gestational age from infant birth weight and gestational age with gender specific intrauterine growth curves. We included only singleton births in the analyses of birth outcomes. Due to distribution EGA was dichotomized at the median, < 39 (1) and ≥ 39 (0) weeks.

Statistical Analysis

We tested the relationships between five birth outcomes (low birth weight (LBW), preterm, small for gestational age (SGA), birth weight, and estimated gestational age (EGA)) with smoking reduction, cessation at late third trimester (LTT) and substance abuse. We only counted women quit if they had biochemically verified smoking abstinence. We used chi-square and t-tests. We coded LBW, preterm, and SGA as no (0) vs. yes (1) and EGA as < 39 weeks (0) vs. ≥ 39 weeks (1). Unadjusted and adjusted simple logistic regression was used for dichotomous outcomes (low birth weight preterm, EGA and SGA). We ran 3 adjusted logistic models which included arm, race, substance/drugs use, cessation and 50% reduction and prior LBW and preterm where appropriate. Proc Mixed used for continuous outcome (birth weight). Normality was assessed with plots and skewness and kurtosis statistic. EGA (in weeks) was dichotomized due to non-normality at the median (< 39/ ≥ 39). Baseline characteristics and loss to follow-up difference by arm were tested. SAS V9.4 was used for these analyses.

Results

Among the birth outcomes cohort, most women were partnered, unemployed, and multiparous. The mean birth weight was 2962.9 (SD = 647.98), and 20% had a low birth weight infant. The average gestational age was 38 (SD = 4.46) weeks, and 17% delivered prematurely (earlier than 37 weeks). 50% of the singleton births were male. Table 1 presents the distribution of baseline characteristics of the study population by intervention arm. 39% of the cohort had evidence of alcohol/drug use from nurse notes.

Table 1.

Baseline characteristics by arm

Baseline characteristics	Total (N = 290) Mean (SD)/n(%)	SGR + SMS (N = 148) Mean (SD)/n(%)	SMS (N = 142) Mean (SD)/n(%)
Age	28(5)	27 (5)	28 (6)
Education
High school or less	165(57)	81(55)	84(59)
Race
Black	157(54)	78(53)	79(56)
Non-black	133(46)	70(48)	63(44)
Employment
Employed	196(68)	91(61)	105(74) +
Financial security
Money for special things	54(19)	26(18)	28(20) ±
Pay bills, little spare money	120(42)	75(52)	45(32)
Pay bills, cut back on things	45(16)	20(14)	25(18)
Difficulty paying bills	64(230)	23(16)	41(30)
First pregnancy	35(12)	21(14)	14(10)
Partner smokes*	210(76)	110(77)	100(75)
Times pregnant median (IQR) (range0–13)	2(1–4)	2 (1–4)	2 (1–4)
Women with prior live birth	N = 220	N = 104	N = 116
Previous Pre-term Birth	62(28)	31(30)	31(27)
Previous LBW infant	28(13)	13(13)	15(13)

⁺x² = 5.13, p = 0.02

^±x² = 13.10, p = 0.004

^*14 women not partnered (6 SGR + SMS and 8 SMS)

At pre-pregnancy, the sample smoked an average of 16.9 cigarettes per day (95% CI [15.9, 17.7]. Of the women with chart review, 10% had biochemically verified smoking abstinence, and 80% reduced their cigarettes per day from pre-pregnancy to late third-trimester. Women, who reduced smoking, smoked an average of 4.4 cigarettes per day (95% CI [3.8, 4.9]) at late third-trimester survey. No significant arm differences were found across the five birth outcomes (0.14 ≤ p ≤ 0.61). Overall, the mean birth weights by arm were 2996 and 2928, for the intervention arm and control arm, respectively (p = 0.38).

We also explored race differences due to known race disparities in low birth weight. Mean unadjusted birth weight was lower for black pregnant women (M = 2886.3, SD = 661) compared to non-black pregnant women (M = 3054.1, SD = 621.9), p = 0.03. Race was not associated with LBW (24% Black vs. 15% Non-black, p = 0.08), preterm (18% Black vs. 15% Non-black, p = 0.63), SGA (15% Black vs. 12% Non-black, p = 0.61) and EGA less than 39 weeks (48% Black vs. 44% Non-black, p = 0.0.64).

We also explored alcohol/drug use as a potential confounder because alcohol and drug use during pregnancy is a predictor for low weight and pre-term births. Results for unadjusted and adjusted models were similar and displayed in Table 2. Cessation was only associated with birth weight. For women who quit, infants on average weigh 317 g more than those women who did not quit (p = 0.02) We also examined the association between reduction in smoking and birth outcomes. Women who reduced 50% or more was associated with a birth weight increase of 335 g (0.02), while controlling for study arm, race, and substance use. Women with documented alcohol/drug use were more likely to have LBW infants (OR 2.41, 95% CI 1.15–5.01, p = 0.02) and infants with EGA less than 39 weeks (OR 1.96, 95% CI 1.07–3.60, p = 0.03). Infant weight on average was 254 g less for women with alcohol/drug use (p = 0.005).

Table 2.

The relationships of smoking cessation, reduction by 50% or more and drug use with birth outcomes: low birth weight (LBW), preterm birth, estimated gestational age (EGA) and birth weight (BW) controlling for arm and race

	LBWa OR (95% CI)		Preterm birthb OR (95% CI)		EGA < 39 weeks OR (95% CI)		BW (grams) Mean (95% CI)
Cessation	Unadj	Adj	Unadj	Adj	Unadj	Adj
No	1.0	1.0	1.0	1.0	1.0	1.0	2786(2648,2924) ++
Yes	0.13 ± (0.02,1.00)	0.13 (0.02,1.06)	0.55 (0.16,1.90)	0.76 (0.20,2.66)	0.78 0.35,1.73	1.01 (0.42,2.42)	3103 (2821,3385)
Smoking reduction by 50% or more
No	1.0	1.0	1.0	1.0	1.0	1.0	2777(2487,3067)
Yes	0.51 (0.21,1.25)	0.71 (0.21,1.94)	0.52 (020,1.32)	0.63 (0.22,1.79)	0.56 (0.25,1.27)	0.60 (0.24,1.46)	3112(2979,3245) +
Substance use
No	1.0	1.0	1.0	1.0	1.0	1.0	3071(2886,3256)
Yes	2.12 + (1.14,4.20)	2.41 + (1.15,5.01)	1.92 (0.97,3.81)	1.91 (0.89,4.12)	1.75 +++ (1.03,2.97)	1.96 ++++ (1.07,3.60)	2817(2603,3033)

Sample reduced to N = 239 due to lost to follow-up at LTT assessment

^aControlled for prior LBW

^bControlled for prior preterm birth

^±p = 0.05

⁺p = 0.02

⁺⁺p = 0.007

⁺⁺⁺p = 0.04

⁺⁺⁺⁺p = 0.03

Discussion

This study is the first to explore the impact of a scheduled gradual reduction program delivered via text messages on birth outcomes. Due to the similar quit and reduction rates in each arm, we did not observe arm differences in any birth outcomes. Thus, it is not surprising that we did not find arm differences in birth outcomes because we did not find differences in smoking cessation or reduction, the drivers of birth outcomes in this population.

However, consistent with other studies, we observed significant differences in mean birthweight as well as a higher mean infant birth weight among women who reduced smoking as well as race differences. Furthermore, we identified an association between documented alcohol or drug use during pregnancy and an increased risk of an infant born low birth weight.

This study contributes to the growing literature on text-based as well as reduction-based interventions for pregnant smokers. Previous trials using a text-based, cessation intervention did not include birth outcomes as study endpoints. The few that reported birth outcomes did not find any significant arm differences (Forinash et al., 2018; Naughton et al., 2012, 2017). In this trial, most smokers reduced but did not quit smoking by the late-third trimester survey. Within this large group of reducers, we found statistically significant differences between the birth weights and estimate gestational age for infants born to women who reduced by 50% or more and those who did not. These findings support other harm reduction studies on pregnant smokers that suggest a dose–response effect of smoking on birth outcomes (Kataoka et al., 2018; Seybold et al., 2012).

We also explored the influences of substance and alcohol use given the known risks and substantial portion of the study cohort that had drug and/or alcohol use during pregnancy documented in their chart. We found that pregnant women with a presence of alcohol or drug use in their chart were more likely to have an LBW infant even when controlling for smoking. Most studies restrict their interventions to targeting cigarette smoking or substance use during pregnancy (Jackson et al., 2019). Further, smoking cessation trials often control for exposure to alcohol or illicit drugs through exclusion criteria or sample analysis (Bailey et al., 2012). Previous studies have demonstrated a link between substance use during pregnancy and infant gestational age and weight (Mayes et al., 2003; Pinto et al., 2010; Swamy et al., 2009). However, some investigators contest that smoking during pregnancy impacts birth outcomes more than substance use (Bailey et al., 2012). The findings support the impact of both cigarettes and other substances on infant weight.

These results should be viewed in light of some limitations. Unfortunately, we have limited information on the participants’ alcohol and drug use. We obtained this data through retrospective chart review, which relies on the clinical staff’s documentation and self-report of substance use from the patient (Swamy et al., 2009). We do not know whether participants used alcohol, marijuana vs hard illicit drugs, or a combination. Further, participants were coded based on any presence of documentation of drug or alcohol use during a prenatal visit, so it is unclear the duration, amount, or timing of use. We are also limited by unmeasured covariates (maternal BMI and history of chronic disease). The sample also had higher rates or prior preterm and LBW births, which are risk factors for adverse birth outcomes (Valero de Bernabé et al., 2004). Further, the definition of reduction is based on two time-points, baseline and late-third trimester. Also, we did not have a comparison group of pregnant women who did not smoke. Study limitations in regards to the primary smoking endpoints are described elsewhere (Pollak et al., 2019). Even with these limitations, substance or alcohol use was a predictor of negative birth outcomes.

Many of the women who continue to smoke during pregnancy are not able to quit abruptly and eliminate their final cigarettes (Adams et al., 1992; Scherman et al., 2018). Additionally, most women who use drugs or alcohol during pregnancy also smoke cigarettes (Bailey et al., 2012; Oga et al., 2018). Given these challenges, this population may need interventions throughout pregnancy to address both substance use and smoking. While this study only attempted to target smoking cessation through scheduled gradual reduction via text messages, we were able to present findings that support further research on the co-use of substances.

In conclusion, this study supports previous findings that reducing cigarette consumption by 50% can improve birth-weight. Nevertheless, pregnant women should be encouraged to quit completely.

Significance.

Previous trials have demonstrated smoking cessation improves infant birth weight outcomes. However, few text-based trials have included birth outcomes as study endpoints.

Data Availability

N/A.