Secondary Logo

Journal Logo

Perinatal: Review Article

Behavioral Influences on Preterm Birth

A Review

Savitz, David A.a; Murnane, Pamelab

Author Information
doi: 10.1097/EDE.0b013e3181d3ca63


The goal of this review is to summarize the epidemiologic evidence on maternal behaviors and preterm birth, evaluate the quality and strength of the evidence, and make specific recommendations regarding research methods that are likely to be productive in future studies. Epidemiologic research on the etiology of preterm birth has expanded considerably in the 15 years since the last comprehensive review.1 Although overviews and commentaries have appeared, including a recent report from the Institute of Medicine,2 to our knowledge there has been no comprehensive presentation and evaluation of the epidemiologic literature on behavioral risk factors for preterm birth since the early 1990s.

Clinicians and public health practitioners recognize that early delivery has a profound impact on infant morbidity and mortality.3 However, despite extensive investigation, we have made little progress in identifying modifiable behavioral causes of preterm birth. The risk of preterm birth appears to be rising modestly but steadily over time.4 Part of the increase is attributable to increased use of assisted reproductive technology leading to more multiple births (at notably higher risk of preterm delivery). Risks are also increasing among singleton births, in part due to medical intervention.5

The operational definition of preterm birth is birth prior to the completion of 37 weeks' gestation. However, gestational age estimates are subject to error, with varying reliance on last menstrual period dates, ultrasound at different times in gestation, and algorithms for combining information from the 2 sources. Multiple approaches have been used to subdivide preterm birth, including clinical presentation (preterm labor, preterm rupture of chorioamniotic membranes, medical indication for early delivery), etiologic pathway (eg, infection/inflammation, vascular compromise, presence of specific complications), and severity of prematurity (based on number of weeks' gestation). The proportion of studies considering subsets of preterm births appears to be increasing, yet with little consistency with regard to which subsets are considered and precisely how they are defined.

Etiologic influences may act at different times in relation to pregnancy: prior to the onset of pregnancy (through maternal genetic or lifestyle influences), early in pregnancy during placentation, and close to the time of delivery. Since behaviors often vary over time, judgment is required regarding the appropriate time(s) for measurement. Beyond the general concerns with accurate assessment of behaviors, there are special concerns with the potential for reverse causality if symptoms associated with future preterm delivery result in behavioral changes, or if previous adverse pregnancy outcomes affect behaviors in the subsequent pregnancies. Isolating specific behaviors from closely correlated ones is always a challenge in epidemiology, with socioeconomic deprivation strongly associated with an array of potentially harmful behaviors. Self-reported exposures are susceptible to response biases from social desirability, which may be particularly strong in relation to pregnancy and vulnerable to reporting biases when exposures are recalled after the outcome of pregnancy is known.

In contrast to the limited success in identifying strong behavioral influences on preterm birth, there are clear social patterns, with markedly increased risk among African-Americans compared with non-Hispanic Whites.4 Elevated risk is less consistently observed for other minorities in the United States, particularly Mexican-Americans,6,7 but generally found among Puerto Ricans and Filipinos.8 Women who are less educated or poorer often manifest elevated risk of preterm birth, though not with the same strength or consistency as is found for racial disparities.1,9 As increasing proportions of preterm birth are due to medical indications,5 which depend in part on intensity of medical care, women who have insufficient medical care may experience a lower risk of preterm birth due to induction, once other influences on risk are taken into account.

Several biologic markers of impending preterm birth have been identified that are predictive in the days or weeks before delivery. These are not true causes in the sense that manipulating the marker would not prevent the outcome, but they are informative and potentially useful in allowing clinical interventions to postpone delivery or to help prepare the fetus for postnatal life through the use of antenatal steroids.10 Fetal fibronectin has been shown to be highly predictive of impending very early preterm birth when measured at 22–24 weeks' gestation.11 Rises in corticotrophin-releasing hormone and cortisol also precede preterm birth.12

With the continued, accelerating pursuit of the behavioral causes of preterm birth through epidemiologic inquiry, it is timely to take stock and identify the more promising methods and risk factors that should be pursued, as well as those that have stagnated and appear to be unworthy of continued effort without fundamentally new approaches. Although social and biologic processes associated with behaviors are recognized, they have not been fully integrated into the epidemiologic studies thus far. Despite the logistical challenges of conducting randomized behavioral intervention trials, such research needs to be considered when the observational studies are sufficiently promising.


We conducted PubMed literature searches to locate all relevant articles. A highly sensitive search strategy was used to retrieve literature comprehensively back to 1992 (a 15-year period, approximately the date at which the previous comprehensive review1 left off) through January 2008. We used the following key words: “obstetric labor, premature”; “premature birth”; “infant, premature”; “pregnancy outcome”; “pregnancy complications”; “preterm”; “premature”; and “prematurity.” Similarly, a broad search strategy was used for each behavior of interest. For example, “smoking or smoker or smokers or tobacco” was used to retrieve all smoking literature. We then combined the outcome retrieval set with the behavior retrieval set using the Boolean AND operator. Retrieval was limited to English language articles. References were cross-checked for completeness using review articles. One of the authors (P.M.) reviewed each article for inclusion, the requirement being an assessment of a dichotomous outcome of preterm birth, rather than a continuous gestational age outcome. For topics in which the literature was sparse, we calculated relative risks from available data; otherwise, we omitted the few articles that did not provide relative risks.

The following key features of the methods and results were tabulated: geographic source of the population, study design, outcome definition(s), method of measurement of gestational age, exposure ascertainment methods, specific definition and timing of exposure, and the resulting risk ratio (RR) or odds ratio (OR) for each index of exposure with 95% confidence intervals (CIs). We considered preterm birth in the aggregate (all births <37 weeks' gestation) and subsets of preterm birth defined by clinical presentation, severity, complications, or maternal attributes (eg, parity), as information was available.

Within each behavioral category, the literature was first divided into groups with comparable reported outcomes, allowing some flexibility (eg, <37-week and <38-week endpoints were combined), exposure levels (eg, 1–9 cigarettes and 1–10 cigarettes), and reference levels (eg, no caffeine and <10 mg caffeine per day). Preterm births were subdivided based on severity and clinical presentation (spontaneous, medically indicated) where sufficient numbers of studies provided such information. After generating subgroups of literature based on shared outcome definitions and shared exposure assignment, for those with sufficient numbers of studies, we stratified by several methodologic features of the studies: method of gestational age assessment, prenatal or postpartum exposure assessment, clinic versus population source, and geographic location of study. Summary estimates were generated from subgroups containing at least 3 studies sharing these common measures. Estimates were calculated with both random effects and fixed effects methods using R software (; results of random effects analysis are presented here. Individual studies that did not present 95% CIs for the relative risk estimates were excluded from summary estimates.

From the summary estimates, forest plot figures were generated to present the results. All tabulated studies are included in the eAppendix ( In behavioral categories where the literature was too sparse to generate any summary estimates, the relevant literature is described in the text and the full list of identified articles has been included in the electronic appendix.


Tobacco is by far the most extensively evaluated behavior-related influence on preterm birth, with 58 relevant studies identified (eFigure 1, This literature allows for summary estimates to address dose-response gradients, some indicators of study quality, and potential effect-modifiers of the smoking/preterm birth relationship (Fig. 1).13–43 There is evidence of increased risk even in the lowest dose range (for >0–10 cigarettes per day, RR = 1.2 [95% CI = 1.1–1.3]) with a modest increase for the higher dose groups but still with summary ORs of <1.5 and imprecision in the highest dose ranges. Both medically indicated and spontaneous preterm births were associated with smoking, but the latter more strongly (relative risks of 1.5 vs. 1.2 for 10+ cigarettes per day). No meaningful differences were found for prenatal versus postpartum assessment, clinic versus population-based studies, or European versus North American studies. Method of gestational age assessment, grouped for simplicity into those that were and were not “cleaned” by considering information in addition to the date of the last menstrual period (LMP) (using ultrasound or birth-weight), showed no evidence for enhanced associations with refinement. Consideration of earlier preterm birth, <32 or <33 weeks' gestation, yielded somewhat stronger associations than were found for preterm births in the aggregate. Despite the precision resulting from the large number of studies, there is little variation in the summary estimates, ranging only from 1.0 to 1.5, so that “stronger” means only relative risks of ∼1.4–1.5 and “weaker” denotes relative risks of ∼1.1–1.2.

Summary estimates of the association between smoking and preterm birth: number of studies, relative risks, and confidence intervals by dose, preterm birth subsets, and methodologic characteristics.

Smokeless tobacco has been evaluated to a very limited extent (eFigure 2, due to the relative rarity of exposure among women, at least in western countries. The 2 studies that addressed this issue found stronger associations with preterm birth than have been found for active smoking.35,44 Because smokeless tobacco exposes the women to nicotine but not the combustion products of tobacco, it seems unlikely that smokeless tobacco could be more harmful than smoking.

Environmental tobacco smoke exposure has been examined over the last 15 years, with 9 pertinent publications (eFigure 2). Exposure indicators vary widely across studies, including self-reported hours per day of exposure, number of smokers in the household, smoke exposure at work, and urinary or hair cotinine. In the aggregate, the studies suggest an association comparable in magnitude to that observed for active smoking—RRs of 1.2–1.4, with little impact of refinements to gestational age assessment and a slightly stronger association for population-based versus clinic-based studies (Fig. 2).21,29,43,45–50 The comparability of relative risks for active and passive smoking, despite markedly differing doses, raises questions about potential biases, perhaps confounding or publication bias.

Summary estimates of the association between environmental tobacco smoke and preterm birth: number of studies, relative risks, and confidence intervals by methodologic characteristics.

Alcohol has been evaluated in 25 epidemiologic studies (eFigure 3,, with most studies indicating no association with preterm birth (Fig. 3)14,16,17,22,39,42,51–65 The exposure categories vary markedly across studies, depending on the overall distribution of alcohol use in the study population. Comparing studies that provided relative risks for doses of >0–6 drinks per week and 7 or more drinks per week (less than daily vs. at least daily, on average) identified a modestly increased risk in the higher dose group. Otherwise, population source, timing of ascertainment, and study location all generated null findings for the association between alcohol and preterm birth.

Summary estimates of the association between alcohol and preterm birth: number of studies, relative risks, and confidence intervals by dose and methodologic characteristics.

There have been few studies of marijuana use and preterm birth (6 identified) (eFigure 4,, generating imprecise evidence of a modest increase in risk of preterm birth (summary RR = 1.4 [0.8–2.3]).39,58,66–68 Studies of cocaine (eFigure 4) yielded slightly stronger associations based on self-report (summary RR = 1.8 [1.2–2.7] based on 7 studies)22,39,66,67,69–71 or urine screening (1.6 [0.9–2.6], based on 4 studies).66,71–73 Studies that identified cocaine use through self-report or urine screening were likely to isolate the heaviest users, with accompanying adverse social conditions and associated behaviors. Studies that used more sensitive biomarkers found little or no support for an association. Other drugs, including opiates and mixed drug use, yielded RR estimates around 2.0 (eFigure 4), with insufficient homogeneity in methods to generate summary estimates. Isolating the biologic effects of illicit drugs from their social and behavioral context poses a serious challenge, in addition to the potential for incomplete and selective identification of exposure based on suspicion followed by query or measurement of biomarkers, rather than systematic assessment of drug use among all women.74

The potential effect of caffeine on preterm birth has been examined in 9 studies (eFigure 5,, all based on self-reported exposure. The absolute levels of caffeine that were considered vary depending on the study population's range, making it difficult to generate summary estimates that address absolute dose level (Fig. 4).14,15,48,58,75–79 There is no support for an adverse effect at lower doses and some suggestion of an increased risk associated with 3 or more cups of coffee per day (RR = 1.2 [1.0–1.5]) as well as among the studies based on clinic populations (RR = 1.2 [1.0–1.4]). Overall, these studies offer very little support for an association between caffeine and preterm birth.

Summary estimates of the association between caffeine and preterm birth: number of studies, relative risks, and confidence intervals by dose and methodologic characteristics.

Although several studies of leisure time physical activity (6 identified) suggest a fairly strong reduction in risk of preterm birth (eFigure 6,, it is difficult to combine results across studies given the markedly differing definitions of “physically active.” Nonetheless, the summary relative risk is very close to the null (RR = 0.9 [0.7–1.1]) (Fig. 5)80–83 and driven closer to the null due to one large, negative study.82 Isolating direct effects of the physical activity from the social conditions that create the opportunity to be active (having leisure time, having access to places for exercise) and the behavioral determinants (being free of medical contraindications, feeling motivated to exercise) is challenging in observational studies.

Summary estimates of the association of exercise and employment with preterm birth: number of studies, relative risks, and confidence intervals.

Only 3 studies were identified that evaluated sex ual activity in relation to preterm birth (eFigure 7, with divergent findings. One reported a strong positive association,84 one a strong inverse association,85 and one with results close to the null.86 Isolating sexual behavior from the determinants of sexual behavior, as would be done in a hypothetical randomized trial, is very difficult, given that sustained sexual activity requires the presence of a partner, absence of medical contraindications, and the desire to engage in this behavior, all of which may be associated with preterm birth through other pathways.

A sizable number of studies (27 identified) (eFigure 8, have evaluated the relationship between employment (vs. unemployment) and many aspects of work in relation to the risk of preterm birth. Employment per se was not associated with preterm birth, with slightly increased risks in studies with more refined gestational age assessment and those with prenatal assessment (Fig. 5).18,48,57,87–108 Focusing on specific aspects of work, we found little indication that jobs involving standing, lifting, or shift work were associated with preterm birth, whereas those requiring night work (RR = 1.3 [0.8–2.2]) and physically strenuous work (RR = 1.3 [1.1–1.6]) were weakly associated with increased risk.


The volume of epidemiologic research on behavioral influences on preterm birth is substantial, but shows little progress in reaching firm conclusions regarding causal effects. Tobacco smoking is consistently associated with a modest increased risk of preterm birth, and it appears that the association is stronger for higher number of cigarettes per day and earlier preterm birth, consistent with a causal interpretation. Environmental tobacco smoke, though studied less extensively, tends to yield associations of a similar magnitude despite the markedly lower exposure levels than for active smoking; these results are thus of questionable etiologic significance. A range of addictive behaviors, notably higher levels of alcohol, cocaine, and other illicit drugs (behaviors that are among the least socially acceptable), appear to be associated with preterm birth. Two plausible explanations are that a range of pharmacologically active agents are all capable of causing preterm birth or that all these associations are confounded by behaviors correlated with the use of these substances (poor nutrition, sexually transmitted infections) or by adverse socioeconomic conditions more generally. Studies that measure a range of exposures with accuracy, a major challenge for stigmatized behaviors, and that consider not just the agent of interest but the most plausible confounders, would help to isolate any independent causal influence of these behaviors. By contrast, caffeine appears to be unrelated to preterm birth.

Some (but not all) studies of recreational physical activity have suggested a substantially reduced risk of preterm birth, plausibly related to a direct physiological effect as well as to confounding by the social circumstances that promote physical activity and other favorable health behaviors that may accompany it. A more careful evaluation of the underlying proclivity and opportunity for physical activity is essential to allow observational studies to come closer to simulating a randomized trial. Measurement of activity is notoriously difficult and requires some form of objective assessment in addition to self-report.

Employment and preterm birth show only sporadic associations, with the most support for an association between jobs requiring physical exertion or night work and increased risk of preterm birth. Even though work-related physical demands point in the opposite direction as recreational physical activity, the same possibilities remain: an adverse physiologic effect or confounding by the social and behavioral correlates of holding physically demanding jobs.

The constellation of findings and the failure to show much progress with the marked expansion in the volume of research in the last 15 years calls for a radical rethinking of the value of further research on the etiologic influence of these behaviors on preterm birth. Had the large number of reasonably well-designed observational studies yielded a strong signal, a case could be made for proceeding with large behavioral intervention trials. However, except for the modest impact of smoking (with the attributable fraction lower than in the past, due to the success in smoking cessation efforts), no strong signal has been found. For every one of the more promising directions, additional studies of similar quality using traditional methods will yield only small increments to the collective picture, with a slightly narrower confidence interval on the summary estimate. Instead, we need to tackle the real limiting factors: (1) the antecedents of the behaviors, (2) the measurement of the behaviors, (3) and the biologic events that result from the behaviors.


To address the causal impact of the behaviors of concern, we need a better understanding of the social context and determinants of those behaviors. This applies to stigmatized behaviors such as cocaine use or heavy alcohol consumption, to health-motivated behaviors such as recreational physical activity, and to the life circumstances that result in jobs with varying physical demands. To the extent that we must rely on observational studies, those studies need to approximate randomized trials in attempting to isolate the individuals whose self-assignment to exposure groups is sufficiently free of major confounding to contribute to causal inference. As illustrated so dramatically by our inability to isolate postmenopausal estrogen use from associated risks for cardiovascular disease, this is a major challenge but one that can be addressed. Where the social disparities associated with the behavior are profound, (eg, cocaine use, physically demanding jobs), there is a need to restrict the study population to women who are broadly comparable in education and income. Where the medical course of the pregnancy may affect engagement in the behavior, (eg, contraindications to physical activity or continued sexual activity), those women must be excluded to generate a more valid estimate of the independent, causal effect of the behavior. Conventional approaches to measuring and adjusting for confounders are pertinent, but a more systematic, careful dissection of the proclivity to engage in the behavior should be undertaken, isolating those who faced a more random choice or at least one that is free of profound distortion by other obvious determinants of preterm birth. Propensity scores109 may be useful in refining the estimate of the causal impact, but the setting and structure of the study need to be selected with this problem in mind. Measuring these sources of confounding and demonstrating the impact of their removal through restriction or control would enhance our confidence in the causal relevance of what remains.


Past research clearly shows that even sophisticated approaches to the study of behaviors and preterm birth do not yield strong or consistent evidence of associations. These approaches include carefully designed interviews and standardized instruments, with appropriate attention to data quality. Within the range of available studies, there is little basis for confidence that modest refinements (more carefully worded questions, better-trained interviewers) would yield much benefit. It is of course possible that there truly is no impact of the behavior on preterm birth, but there is such extensive opportunity for measurement error for these behaviors (eg, caffeine, alcohol, physical activity, work demands) that it remains possible that the measured relative risks of 1.0 to 1.5 would be substantially higher if misclassification were markedly reduced. Where suitable technologies exist, such as monitoring of physical activity or biomarkers reflective of exposures over the time course of interest, new studies will at minimum yield informative findings that can shape our understanding of the causal impact of the behavior and thus encourage or discourage further investigation. While refinements to questionnaire-based methods could in theory yield radically different exposure assignment, the more likely direction is in qualitatively different technologies, such as hair nicotine or cocaine metabolites or prospectively applied instruments to measure physical activity. Perinatal epidemiologists need to watch for breakthroughs in exposure measurement that could move the field forward.

Biologic Pathways

The lack of biologic specificity inherent in the clinical entity of preterm birth is well appreciated,2 with subsets of cases resulting from different etiologic pathways even though the specific pathways have yet to be clearly defined. Moreover, the appropriate axes for division may have to be addressed on an exposure-by-exposure basis, ie, the subset of cases plausibly affected by tobacco or by alcohol. This might be manifested either by a subset of preterm births that are more strongly associated with the exposure or with an effect-measure modifier defining a group in which the impact of the exposure is enhanced. The literature provides some clues, suggesting, for example, a greater impact of smoking on earlier preterm birth and spontaneous as compared with medically indicated preterm birth. Where feasible, division of cases along those conventional lines is encouraged, although doing so requires very large studies.

A more promising avenue for elucidating the potential causal impact of behaviors on preterm birth is through evaluation of near-term biologic consequences that affect the pathways thought to contribute to preterm birth. While the mechanisms resulting in preterm birth are incompletely understood, there is a strong justification to focus on a subset of biologic responses that are most likely to be relevant, including inflammation, vascular compromise, and neuroendocrine changes. In each of those pathways (and other suspected ones, including oxidative stress and hypoxia), there are biologic markers of response that could help to strengthen and clarify the link between behaviors and clinical outcomes. For example, a great deal is known about the impact of physical activity on cardiovascular health in the nonpregnant state. These markers of effect through lipid profile, blood pressure, and inflammatory markers are pertinent to pregnancy as well. It would be feasible to carry out observational studies that monitor physical activity over shorter periods of time (and thus permit precise assessment of activity), combined with short-term biologic response in the form of altered inflammatory markers or lipids.

The feasibility of randomized trials requires postulating the relevant impact of the behavior of concern, being able to manipulate the behavior (which can be a major logistical or ethical challenge), and examining the near-term biologic consequences. It would be daunting to modify behavior throughout pregnancy to assess impact on clinical outcomes such as preterm birth, whereas short-term manipulation of behavior to study proximal biologic changes is not out of reach. While such evidence constitutes only one link in the chain from behavior to preterm birth, the tightness of the question and clarity of the answer would help to encourage or discourage further efforts. Such results could also help to pinpoint the subset of preterm births most deserving of attention in relation to the behavior.


For many of the behaviors of concern, conventional epidemiologic approaches have been applied with little in the way of conclusive data other than the small effect of smoking on preterm birth. The absence of clear associations for other intensively studied behaviors offers some guidance for clinical and public health application, but they do little to help in designing more ambitious efforts to prevent preterm birth. Prevention requires identification of clear etiologic relationships, and reaching that level of certainty calls for a markedly different approach to the examination of the antecedents of the behaviors of interest, novel approaches to assessment, and examination of near-term biologic consequences of the behaviors of concern, focusing on pathways thought to result in preterm birth.


We thank Stephanie Engel and Anna Maria Siega-Riz for their helpful review of an earlier draft.


1. Berkowitz GS, Papiernik E. Epidemiology of preterm birth. Epidemiol Rev. 1993;15:414–443.
2. Behrman RE, Butler AS. Preterm Birth: Causes, Consequences, and Prevention. Washington, DC: National Academies Press; 2007.
3. Callaghan WM, MacDorman MF, Rasmussen SA, Qin C, Lackritz EM. The contribution of preterm birth to infant mortality rates in the United States. Pediatrics. 2006 ;118 :1566–1573.
4. Demissie K, Rhoads GG, Ananth CV, et al. Trends in preterm birth and neonatal mortality among blacks and whites in the United States from 1989 to 1997. Am J Epidemiol. 2001 ;154 :307–315.
5. Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal mortality among singletons: United States, 1989 through 2000. Obstet Gynecol. 2005;105(5 Pt 1):1084–1091.
6. Crump C, Lipsky S, Mueller BA. Adverse birth outcomes among Mexican-Americans: are US-born women at greater risk than Mexico-born women ? Ethn Health. 1999 ;4 :29–34.
7. Buekens P, Notzon F, Kotelchuck M, Wilcox A. Why do Mexican Americans give birth to few low-birth-weight infants ? Am J Epidemiol. 2000 ;152 :347–351.
8. Singh GK, Yu SM. Adverse pregnancy outcomes: differences between US- and foreign-born women in major US racial and ethnic groups. Am J Public Health. 1996 ;86 :837–843.
9. Kaufman JS, Dole N, Savitz DA, Herring AH. Modeling community-level effects on preterm birth. Ann Epidemiol. 2003 ;13 :377–384.
10. Merrill JD, Ballard RA. Antenatal hormone therapy for fetal lung maturation. Clin Perinatol. 1998 ;25 :983–997.
11. Goldenberg RL, Mercer BM, Meis PJ, Copper RL, Das A, McNellis D. The preterm prediction study: fetal fibronectin testing and spontaneous preterm birth. NICHD Maternal Fetal Medicine Units Network. Obstet Gynecol. 1996;87(5 Pt 1):643–648.
12. Korebrits C, Ramirez MM, Watson L, Brinkman E, Bocking AD, Challis JR. Maternal corticotropin-releasing hormone is increased with impending preterm birth. J Clin Endocrinol Metab. 1998 ;83 :1585–1591.
13. Kramer MS, McLean FH, Eason EL, Usher RH. Maternal nutrition and spontaneous preterm birth. Am J Epidemiol. 1992 ;136 :574–583.
14. McDonald AD, Armstrong BG, Sloan M. Cigarette, alcohol, and coffee consumption and prematurity. Am J Public Health. 1992 ;82 :87–90.
15. Williams MA, Mittendorf R, Stubblefield PG, Lieberman E, Schoenbaum SC, Monson RR. Cigarettes, coffee, and preterm premature rupture of the membranes. Am J Epidemiol. 1992 ;135 :895–903.
16. Verkerk PH, Buitendijk SE, Verloove-Vanhorick SP. Differential misclassification of alcohol and cigarette consumption by pregnancy outcome. Int J Epidemiol. 1994 ;23 :1218–1225.
17. Verkerk PH, Zaadstra BM, Reerink JD, Herngreen WP, Verloove- Vanhorick SP. Social class, ethnicity and other risk factors for small for gestational age and preterm delivery in The Netherlands. Eur J Obstet Gynecol Reprod Biol. 1994 ;53 :129–134.
18. Meis PJ, Michielutte R, Peters TJ, et al. Factors associated with preterm birth in Cardiff, Wales. II. Indicated and spontaneous preterm birth. Am J Obstet Gynecol. 1995 ;173 :597–602.
19. Nordentoft M, Lou HC, Hansen D, et al. Intrauterine growth retardation and premature delivery: the influence of maternal smoking and psychosocial factors. Am J Public Health. 1996 ;86 :347–354.
20. Wisborg K, Henriksen TB, Hedegaard M, Secher NJ. Smoking during pregnancy and preterm birth. Br J Obstet Gynaecol. 1996 ;103 :800–805.
21. Horta BL, Victora CG, Menezes AM, Halpern R, Barros FC. Low birthweight, preterm births and intrauterine growth retardation in relation to maternal smoking. Paediatr Perinat Epidemiol. 1997 ;11 :140–151.
22. Kramer MS, Platt R, Yang H, et al. Secular trends in preterm birth: a hospital-based cohort study. JAMA. 1998 ;280 :1849–1854.
23. Kyrklund-Blomberg NB, Cnattingius S. Preterm birth and maternal smoking: risks related to gestational age and onset of delivery. Am J Obstet Gynecol. 1998 ;179 :1051–1055.
24. Wu T, Buck G, Mendola P. Can regular multivitamin/mineral supplementation modify the relation between maternal smoking and select adverse birth outcomes ? Ann Epidemiol. 1998 ;8 :175–183.
25. Ancel PY, Saurel-Cubizolles MJ, Di Renzo GC, Papiernik E, Breart G. Very and moderate preterm births: are the risk factors different ? Br J Obstet Gynaecol. 1999 ;106 :1162–1170.
26. Cnattingius S, Granath F, Petersson G, Harlow BL. The influence of gestational age and smoking habits on the risk of subsequent preterm deliveries. N Engl J Med. 1999 ;341 :943–948.
27. Moore ML, Zaccaro DJ. Cigarette smoking, low birth weight, and preterm births in low-income African American women. J Perinatol. 2000 ;20 :176–180.
28. Pollack H, Lantz PM, Frohna JG. Maternal smoking and adverse birth outcomes among singletons and twins. Am J Public Health. 2000 ;90 :395–400.
29. Windham GC, Hopkins B, Fenster L, Swan SH. Prenatal active or passive tobacco smoke exposure and the risk of preterm delivery or low birth weight. Epidemiology. 2000 ;11 :427–433.
30. Savitz DA, Dole N, Terry JW Jr, et al. Smoking and pregnancy outcome among African-American and white women in central North Carolina. Epidemiology. 2001 ;12 :636–642.
31. Wisborg K, Henriksen TB, Secher NJ. Maternal smoking and gestational age in twin pregnancies. Acta Obstet Gynecol Scand. 2001 ;80 :926–930.
32. Newburn-Cook CV, White D, Svenson LW, Demianczuk NN, Boff N, Edwards J. Where and to what extent is prevention of low birth weight possible ? West J Nurs Res. 2002 ;24 :887–904.
33. Ohmi H, Hirooka K, Mochizuki Y. Fetal growth and the timing of exposure to maternal smoking. Pediatr Int. 2002 ;44 :55–59.
34. Ahern J, Pickett KE, Selvin S, Abrams B. Preterm birth among African American and white women: a multilevel analysis of socioeconomic characteristics and cigarette smoking. J Epidemiol Community Health. 2003 ;57 :606–611.
35. England LJ, Levine RJ, Mills JL, Klebanoff MA, Yu KF, Cnattingius S. Adverse pregnancy outcomes in snuff users. Am J Obstet Gynecol. 2003 ;189 :939–943.
36. Burguet A, Kaminski M, Abraham-Lerat L, et al. The complex relationship between smoking in pregnancy and very preterm delivery. Results of the Epipage study. BJOG. 2004 ;111 :258–265.
37. Jaakkola JJ, Gissler M. Maternal smoking in pregnancy, fetal development, and childhood asthma. Am J Public Health. 2004 ;94 :136–140.
38. Wenman WM, Joffres MR, Tataryn IV. A prospective cohort study of pregnancy risk factors and birth outcomes in Aboriginal women. CMAJ. 2004 ;171 :585–589.
39. Bada HS, Das A, Bauer CR, et al. Low birth weight and preterm births: etiologic fraction attributable to prenatal drug exposure. J Perinatol. 2005 ;25 :631–637.
40. Kyrklund-Blomberg NB, Granath F, Cnattingius S. Maternal smoking and causes of very preterm birth. Acta Obstet Gynecol Scand. 2005 ;84 :572–577.
41. Nabet C, Ancel PY, Burguet A, Kaminski M. Smoking during pregnancy and preterm birth according to obstetric history: French national perinatal surveys. Paediatr Perinat Epidemiol. 2005 ;19 :88–96.
42. Dew PC, Guillory VJ, Okah FA, Cai J, Hoff GL. The effect of health compromising behaviors on preterm births. Matern Child Health J. 2007 ;11 :227–233.
43. Fantuzzi G, Aggazzotti G, Righi E, et al. Preterm delivery and exposure to active and passive smoking during pregnancy: a case-control study from Italy. Paediatr Perinat Epidemiol. 2007 ;21 :194–200.
44. Gupta PC, Sreevidya S. Smokeless tobacco use, birth weight, and gestational age: population based, prospective cohort study of 1217 women in Mumbai, India. BMJ. 2004;328:1538.
45. Ahluwalia IB, Grummer-Strawn L, Scanlon KS. Exposure to environmental tobacco smoke and birth outcome: increased effects on pregnant women aged 30 years or older. Am J Epidemiol. 1997 ;146 :42–47.
46. Hanke W, Kalinka J, Florek E, Sobala W. Passive smoking and pregnancy outcome in central Poland. Hum Exp Toxicol. 1999 ;18 :265–271.
47. Jaakkola JJ, Jaakkola N, Zahlsen K. Fetal growth and length of gestation in relation to prenatal exposure to environmental tobacco smoke assessed by hair nicotine concentration. Environ Health Perspect. 2001 ;109 :557–561.
48. Tough SC, Newburn-Cook CV, White DE, et al. Do maternal characteristics and past pregnancy experiences predict preterm delivery among women aged 20 to 34 ? J Obstet Gynaecol Can. 2003 ;25 :656–666.
49. Kharrazi M, DeLorenze GN, Kaufman FL, et al. Environmental tobacco smoke and pregnancy outcome. Epidemiology. 2004 ;15 :660–670.
50. Kim YJ, Lee BE, Park HS, Kang JG, Kim JO, Ha EH. Risk factors for preterm birth in Korea: a multicenter prospective study. Gynecol Obstet Invest. 2005 ;60 :206–212.
51. Feldman JG, Minkoff HL, McCalla S, Salwen M. A cohort study of the impact of perinatal drug use on prematurity in an inner-city population. Am J Public Health. 1992 ;82 :726–728.
52. Lazzaroni F, Bonassi S, Magnani M, et al. Moderate maternal drinking and outcome of pregnancy. Eur J Epidemiol. 1993 ;9 :599–606.
53. Verkerk PH, van Noord-Zaadstra BM, Florey CD, de Jonge GA, Verloove-Vanhorick SP. The effect of moderate maternal alcohol consumption on birth weight and gestational age in a low risk population. Early Hum Dev. 1993;32:121–129.
54. Adams MM, Sarno AP, Harlass FE, Rawlings JS, Read JA. Risk factors for preterm delivery in a healthy cohort. Epidemiology. 1995 ;6 :525–532.
55. Windham GC, Fenster L, Hopkins B, Swan SH. The association of moderate maternal and paternal alcohol consumption with birthweight and gestational age. Epidemiology. 1995 ;6 :591–597.
56. Lundsberg LS, Bracken MB, Saftlas AF. Low-to-moderate gestational alcohol use and intrauterine growth retardation, low birthweight, and preterm delivery. Ann Epidemiol. 1997 ;7 :498–508.
57. Meis PJ, Goldenberg RL, Mercer BM, et al. The preterm prediction study: risk factors for indicated preterm births. Maternal-Fetal Medicine Units Network of the National Institute of Child Health and Human Development. Am J Obstet Gynecol. 1998;178:562–567.
58. Wright SP, Mitchell EA, Thompson JM, Clements MS, Ford RP, Stewart AW. Risk factors for preterm birth: a New Zealand study. N Z Med J. 1998 ;111 :14–16.
59. Kesmodel U, Olsen SF, Secher NJ. Does alcohol increase the risk of preterm delivery ? Epidemiology. 2000 ;11 :512–518.
60. Tough SC, Svenson LW, Johnston DW, Schopflocher D. Characteristics of preterm delivery and low birthweight among 113,994 infants in Alberta: 1994–1996. Can J Public Health. 2001 ;92 :276–280.
61. Parazzini F, Chatenoud L, Surace M, et al. Moderate alcohol drinking and risk of preterm birth. Eur J Clin Nutr. 2003 ;57 :1345–1349.
62. Wang CS, Chou P. Differing risk factors for premature birth in adolescent mothers and adult mothers. J Chin Med Assoc. 2003 ;66 :511–517.
63. Albertsen K, Andersen AM, Olsen J, Gronbaek M. Alcohol consumption during pregnancy and the risk of preterm delivery. Am J Epidemiol. 2004 ;159 :155–161.
64. Gao W, Paterson J, Carter S, Percival T. Risk factors for preterm and small-for-gestational-age babies: a cohort from the Pacific Islands Families Study. J Paediatr Child Health. 2006 ;42 :785–792.
65. Jaddoe VW, Bakker R, Hofman A, et al. Moderate alcohol consumption during pregnancy and the risk of low birth weight and preterm birth. The generation R study. Ann Epidemiol. 2007 ;17 :834–840.
66. Kliegman RM, Madura D, Kiwi R, Eisenberg I, Yamashita T. Relation of maternal cocaine use to the risks of prematurity and low birth weight. J Pediatr. 1994 ;124 (5 Pt 1):751–756.
67. Shiono PH, Klebanoff MA, Nugent RP, et al. The impact of cocaine and marijuana use on low birth weight and preterm birth: a multicenter study. Am J Obstet Gynecol. 1995 ;172 (1 Pt 1):19–27.
68. Burns L, Mattick RP, Cooke M. The use of record linkage to examine illicit drug use in pregnancy. Addiction. 2006 ;101 :873–882.
69. Kistin N, Handler A, Davis F, Ferre C. Cocaine and cigarettes: a comparison of risks. Paediatr Perinat Epidemiol. 1996 ;10 :269–278.
70. Miller JM Jr, Boudreaux MC. A study of antenatal cocaine use-chaos in action. Am J Obstet Gynecol. 1999 ;180 (6 Pt 1):1427–1431.
71. Savitz DA, Henderson L, Dole N, et al. Indicators of cocaine exposure and preterm birth. Obstet Gynecol. 2002 ;99 :458–465.
72. Miller JM Jr, Boudreaux MC, Regan FA. A case-control study of cocaine use in pregnancy. Am J Obstet Gynecol. 1995 ;172 (1 Pt 1):180–185.
73. Sprauve ME, Lindsay MK, Herbert S, Graves W. Adverse perinatal outcome in parturients who use crack cocaine. Obstet Gynecol. 1997 ;89 (5 Pt 1):674–678.
74. Holzman C, Paneth N. Maternal cocaine use during pregnancy and perinatal outcomes. Epidemiol Rev. 1994 ;16 :315–334.
75. Fortier I, Marcoux S, Beaulac-Baillargeon L. Relation of caffeine intake during pregnancy to intrauterine growth retardation and preterm birth. Am J Epidemiol. 1993 ;137 :931–940.
76. Pastore LM, Savitz DA. Case-control study of caffeinated beverages and preterm delivery. Am J Epidemiol. 1995 ;141 :61–69.
77. Eskenazi B, Stapleton AL, Kharrazi M, Chee WY. Associations between maternal decaffeinated and caffeinated coffee consumption and fetal growth and gestational duration. Epidemiology. 1999 ;10 :242–249.
78. Bracken MB, Triche EW, Belanger K, Hellenbrand K, Leaderer BP. Association of maternal caffeine consumption with decrements in fetal growth. Am J Epidemiol. 2003 ;157 :456–466.
79. Chiaffarino F, Parazzini F, Chatenoud L, et al. Coffee drinking and risk of preterm birth. Eur J Clin Nutr. 2006 ;60 :610–613.
80. Hatch M, Levin B, Shu XO, Susser M. Maternal leisure-time exercise and timely delivery. Am J Public Health. 1998 ;88 :1528–1533.
81. Evenson KR, Siega-Riz AM, Savitz DA, Leiferman JA, Thorp JM Jr. Vigorous leisure activity and pregnancy outcome. Epidemiology. 2002 ;13 :653–659.
82. Leiferman JA, Evenson KR. The effect of regular leisure physical activity on birth outcomes. Matern Child Health J. 2003 ;7 :59–64.
83. Hegaard HK, Hedegaard M, Damm P, Ottesen B, Petersson K, Henriksen TB. Leisure time physical activity is associated with a reduced risk of preterm delivery. Am J Obstet Gynecol. 2008;198:180.e1–180.e5.
84. Ekwo EE, Gosselink CA, Woolson R, Moawad A, Long CR. Coitus late in pregnancy: risk of preterm rupture of amniotic sac membranes. Am J Obstet Gynecol. 1993 ;168 (1 Pt 1):22–31.
85. Sayle AE, Savitz DA, Thorp JM Jr, Hertz-Picciotto I, Wilcox AJ. Sexual activity during late pregnancy and risk of preterm delivery. Obstet Gynecol. 2001 ;97 :283–289.
86. Read JS, Klebanoff MA. Sexual intercourse during pregnancy and preterm delivery: effects of vaginal microorganisms. The Vaginal Infections and Prematurity Study Group. Am J Obstet Gynecol. 1993;168:514–519.
87. Henriksen TB, Savitz DA, Hedegaard M, Secher NJ. Employment during pregnancy in relation to risk factors and pregnancy outcome. Br J Obstet Gynaecol. 1994 ;101 :858–865.
88. Xu X, Ding M, Li B, Christiani DC. Association of rotating shiftwork with preterm births and low birth weight among never smoking women textile workers in China. Occup Environ Med. 1994 ;51 :470–474.
89. Fortier I, Marcoux S, Brisson J. Maternal work during pregnancy and the risks of delivering a small-for-gestational-age or preterm infant. Scand J Work Environ Health. 1995 ;21 :412–418.
90. Olsen P, Laara E, Rantakallio P, Jarvelin MR, Sarpola A, Hartikainen AL. Epidemiology of preterm delivery in two birth cohorts with an interval of 20 years. Am J Epidemiol. 1995 ;142 :1184–1193.
91. Mercer BM, Goldenberg RL, Das A, et al. The preterm prediction study: a clinical risk assessment system. Am J Obstet Gynecol. 1996;174:1885–1893; discussion 1893–1895.
92. Savitz DA, Olshan AF, Gallagher K. Maternal occupation and pregnancy outcome. Epidemiology. 1996 ;7 :269–274.
93. Brett KM, Strogatz DS, Savitz DA. Employment, job strain, and preterm delivery among women in North Carolina. Am J Public Health. 1997 ;87 :199–204.
94. Savitz DA, Brett KM, Dole N, Tse CK. Male and female occupation in relation to miscarriage and preterm delivery in central North Carolina. Ann Epidemiol. 1997 ;7 :509–516.
95. Misra DP, Strobino DM, Stashinko EE, Nagey DA, Nanda J. Effects of physical activity on preterm birth. Am J Epidemiol. 1998 ;147 :628–635.
96. Tuntiseranee P, Geater A, Chongsuvivatwong V, Kor-anantakul O. The effect of heavy maternal workload on fetal growth retardation and preterm delivery. A study among southern Thai women. J Occup Environ Med. 1998 ;40 :1013–1021.
97. Bodin L, Axelsson G, Ahlborg G Jr. The association of shift work and nitrous oxide exposure in pregnancy with birth weight and gestational age. Epidemiology. 1999 ;10 :429–436.
98. Hanke W, Saurel-Cubizolles MJ, Sobala W, Kalinka J. Employment status of pregnant women in central Poland and the risk of preterm delivery and small-for-gestational-age infants. Eur J Public Health. 2001 ;11 :23–28.
99. Nguyen N, Savitz DA, Thorp JM. Risk factors for preterm birth in Vietnam. Int J Gynaecol Obstet. 2004 ;86 :70–78.
100. Saurel-Cubizolles MJ, Zeitlin J, Lelong N, Papiernik E, Di Renzo GC, Breart G. Employment, working conditions, and preterm birth: results from the Europop case-control survey. J Epidemiol Community Health. 2004 ;58 :395–401.
101. Zhu JL, Hjollund NH, Olsen J. Shift work, duration of pregnancy, and birth weight: the National Birth Cohort in Denmark. Am J Obstet Gynecol. 2004 ;191 :285–291.
102. Magann EF, Evans SF, Chauhan SP, et al. The effects of standing, lifting and noise exposure on preterm birth, growth restriction, and perinatal death in healthy low-risk working military women. J Matern Fetal Neonatal Med. 2005 ;18 :155–162.
103. Pompeii LA, Savitz DA, Evenson KR, Rogers B, McMahon M. Physical exertion at work and the risk of preterm delivery and small-for-gestational-age birth. Obstet Gynecol. 2005 ;106 :1279–1288.
104. Al-Dabbagh SA, Al-Taee WY. Risk factors for pre-term birth in Iraq: a case-control study. BMC Pregnancy Childbirth. 2006;6::13.
105. Arafa MA, Amine T, Abdel Fattah M. Association of maternal work with adverse perinatal outcome. Can J Public Health. 2007;98:217–221.
106. Bell JF, Zimmerman FJ, Diehr PK. Maternal work and birth outcome disparities. Matern Child Health J. 2008 ;12 :415–426.
107. Meyer JD, Warren N, Reisine S. Job control, substantive complexity, and risk for low birth weight and preterm delivery: an analysis from a state birth registry. Am J Ind Med. 2007 ;50 :664–675.
108. Rodrigues T, Barros H. Comparison of risk factors for small-for-gestational-age and preterm in a Portuguese cohort of newborns. Matern Child Health J. 2007 ;11 :417–424.
109. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998 ;17 :2265–2281.

Supplemental Digital Content

© 2010 Lippincott Williams & Wilkins, Inc.