This study was initiated when local medical providers expressed concerns about potential adverse health effects of maternal smokeless tobacco use during pregnancy. In particular, some providers worried that use of smokeless tobacco might result in an increased risk of preeclampsia and/or placental abruption. CDC investigators met with local providers and tribal health center board members to discuss these concerns. Based on the high prevalence of tobacco use, availability of exposure data in the medical record, and severity of the outcomes of concern, both organizations agreed to jointly conduct a study of pregnancy outcomes in smokeless tobacco users. In addition, local medical center staff expressed an interest in the prevalence of these and other adverse outcomes in their community. The current analysis focuses on the latter topic. The collaborative research team included local, state, and CDC professionals.

The study region is in a remote area of western Alaska with a large indigenous population that is relatively homogenous with respect to socioeconomic status and culture. This region has its own health care system consisting of a single main outpatient and hospital facility and individual village clinics. The village clinics are staffed by lay health care workers who treat minor and some major illnesses based on clinical guidelines. Nearly all pregnant Alaska Native women in the study region (96 %) receive their health care through this system. Women from the study region who are experiencing pregnancy complications or who otherwise have high risk pregnancies are referred to the Alaska Native Medical Center in Anchorage (ANMC) or to Providence Alaska Medical Center in Anchorage (PAMC) for specialty care during pregnancy and/or for delivery. Of approximately 500 deliveries a year, <15 per year occur at home or in villages.

Singleton deliveries were identified using discharge codes recorded in the regional hospital/medical center’s and the Alaska Native Medical Center’s electronic Resource and Patient Management System (RPMS). The research team searched for hospitalizations of Alaska Native women residing in the study region with the International Classification of Diseases, 9th Revision (ICD9-CM) discharge codes indicating delivery of an infant (liveborn or stillborn): any V27.0 or V27.1 or any 656.4 (intrauterine death after 22 weeks gestation). Because ICD9-CM code lists for hospitalizations are sometimes truncated, the following procedure codes were also used: 73.5, 74.0–74.2, 74.4, 74.99 (manually assisted delivery or cesarean section). Deliveries were excluded if the ICD9-CM codes indicated a multifetal gestation, including 651 (multiple gestation delivery, regardless of the 4th or 5th digit extension) or 652.61 (multiple gestation delivery with malpresentation of 1 or more fetus). Deliveries at PAMC to women residing in the study region were identified from hospital transfer records, as this hospital does not utilize the RPMS system.

As previously described [4], the research team developed a study protocol, and created and piloted abstraction forms based on standard clinical and hospital forms used by the medical facilities. CDC staff held intensive training sessions with two chart abstractors (a nurse practitioner and a hospital data manager) at the beginning of the study and reviewed their completed abstraction forms several times throughout the study period to ensure that abstractors understood and adhered to abstraction procedures and that data were being abstracted from medical record sources in a consistent manner. CDC staff developed a data entry system using Epi Info and ensured that each data entry screen closely resembled the corresponding abstraction form page. Abstractors first examined medical records for all potential study deliveries to verify that they met inclusion criteria; deliveries were excluded if the mother was not a resident of the study region during the pregnancy of interest or was not Alaska Native, the pregnancy had more than one fetus, the delivery occurred before 22 weeks’ gestation. In addition, we excluded any pregnancies indicating that the mother used alcohol in late pregnancy or used cocaine, amphetamines, or opioids at any time during pregnancy, or if the infant was born with a major congenital anomaly. Exact numbers of women excluded for prenatal drug exposure or congenital anomalies are not reported in order to protect the privacy of the study population and the community.

Once inclusion status was confirmed, additional data were abstracted from medical records, including: maternal demographic factors (age at delivery, marital status, highest level of education); gravidity; parity; trimester of prenatal care initiation; maternal height, prepregnancy weight, medical conditions (including type 1 or type 2 diabetes mellitus and chronic hypertension); tobacco exposure; birth weight; pregnancy complications [including gestational diabetes mellitus (GDM) and pregnancy associated hypertension (PAH)]; mode of delivery; and selected birth outcomes.

As part of routine data management procedures and to identify additional training needs to ensure data quality and accuracy, CDC investigators reviewed for accuracy abstracted data for 23 selected variables in approximately 10 % of the study deliveries. Each original response recorded on the abstraction form was reviewed to confirm that it could be verified in the medical record. Errors in abstractions were noted and then reviewed with the abstractors and, when necessary, retraining was provided. Frequent study meetings were held by phone to address questions about the abstraction process. The percentage of variables selected for review in which abstracted information could be verified ranged from 88 to 100 %.

Once primary data were analyzed, the results and the study’s implications were presented to local providers and the hospital executive board and to perinatologists in Anchorage. During these presentations, local providers and board members gave input as to the implications of the findings. During the drafting of the manuscript, Alaska collaborators provided input as to the cultural appropriateness of the text. In addition, they provided feedback on the processes at the hospital and in the community. Based on the findings, local providers are now able to more accurately counsel patients about health risks, based on data from their own community. Importantly, the study was able to provide data to providers to address their specific concerns and provides baseline data that can be used for many purposes, including resource planning and grant applications. The original case–control study has been published in a peer-reviewed journal [4].

Our study population consisted of a population-based random sample of approximately 10 % of all singleton deliveries to Alaska Native women who resided in the study region during pregnancy, used health care services at the regional hospital/medical center, and who delivered their infants at the regional hospital or referral hospital (ANMC or PAMC) between January 1, 1997 and December 31, 2005. As previously stated, data for this prevalence study came from a control population from a larger case–control study of pregnancy outcomes in smokeless tobacco users, previously described elsewhere [4].

The research team classified maternal prepregnancy BMI using categories of the National Heart, Lung, and Blood Institute: <25 kg/m² (underweight and normal), 25–29.9 (overweight), 30–34.9 (class I obesity), 35–39.9 (class II obesity), and 40+ (class III obesity). High parity was defined as one or more previous births for adolescents (<18 years), 3 or more previous births for 18–21 years, 4 or more previous births for 22–24 years, or 5 or more previous births for 25 and older [5]. Tobacco users were categorized into four mutually exclusive categories: tobacco non-users, cigarette smokers only, smokeless tobacco users only (iq’mik or commercial chew), or users of both cigarettes and smokeless tobacco. Iq’mik is a handmade mixture that includes leaf tobacco and ash from burned punk fungus, willow bush, or driftwood [6]. Pregnancies in which women quit using tobacco during pregnancy were categorized according to the type of tobacco used before cessation.

Maternal prenatal conditions included chronic diabetes, GDM, PAH, anemia, urinary tract infection (UTI), and pyelonephritis. Diabetes case definitions were as follows: GDM cases were identified using laboratory data and Carpenter-Coustan criteria [7]. We used physician or provider diagnosis of preexisting diabetes to identify chronic diabetes. Five mutually exclusive diabetes categories were developed: GDM, possible GDM (1-h screen was>140 and an incomplete 3 h glucose tolerance test), chronic diabetes, no diabetes, and unknown. PAH was defined as clinician-diagnosed pregnancy induced hypertension, gestational hypertension, PAH, preeclampsia, eclampsia, or HELLP syndrome. Chronic hypertension was defined as clinician-diagnosed chronic hypertension, pregnancy induced hypertension superimposed on chronic hypertension, or preeclampsia superimposed on chronic hypertension. Anemia was defined as hemoglobin<10 g/dL during pregnancy.

Labor and delivery outcomes included cesarean delivery, polyhydraminos, oligohydramnios, fetal distress, and abruption; all were based on clinician diagnoses as recorded in the medical record. Adverse birth outcomes included shoulder dystocia, preterm delivery, admission into the neonatal intensive care unit, low birth weight, and macrosomia. We defined preterm delivery as delivery at <37 completed weeks gestation based on the provider’s estimated date of confinement. Low birth weight was defined as birth weight<2,500 g. Macrosomia was defined both as birth weight >4,000 g and as birth weight >4,500 g.

After limiting our study sample to the randomly-selected control group, we calculated the proportion of pregnancies with selected maternal characteristics, maternal medical conditions, and adverse birth outcomes with corresponding 95 % confidence intervals. We excluded pregnancies ending in a stillbirth or fetal death when calculating the percentage of pregnancies with adverse birth outcomes. Statistical analyses were performed using SAS software V.9 (SAS Institute Inc., Cary, NC) for Windows. The study proposal and data collection tools were reviewed and approved by the institutional review boards (IRBs) of the CDC and the Mayo Clinic, and by the Alaska Area IRB.

Our analysis has several limitations. As previously stated, this study only includes singleton deliveries among women who did not use alcohol in late pregnancy or illicit drugs at any time during pregnancy and who did not give birth to an infant with a major congenital anomaly. Thus our results cannot be generalized to all Alaska Native women in the study region. Our exclusions likely resulted in an underestimation of some adverse health outcomes, such as pre-term delivery and low birth weight. In addition, the size of our random sample was based on requirements for the main case control analysis and it was too small to generate stable estimates for rare outcomes such as meconium aspiration, placenta previa, sepsis, and very low birth weight. Therefore, these outcomes were not included in our analysis.