Objective
Carbon monoxide (CO) in cigarette smoke may be the mechanism by which tobacco use during pregnancy decreases the risk of the development of preeclampsia. We attempted to test this hypothesis by examining the effect of maternal exposure to ambient CO on preeclampsia.
Study Design
Births that occurred between 2004 and 2009 in the Canadian province of Ontario were extracted from the data. Study subjects were divided into 4 groups according to quartiles of CO concentration that were based on maternal residence. Adjusted odds ratio and 95% confidence interval were used to estimate the independent effect of CO on preeclampsia.
Results
Rates of preeclampsia were 2.32%, 1.97%, 1.59%, and 1.26%, respectively, in the first, second, third, and fourth quartile of CO concentration. The inverse association between CO concentration and preeclampsia risk remained the same after adjustment for several important confounding factors.
Conclusion
Maternal exposure to moderate ambient CO is associated independently with a decreased risk of preeclampsia.
Preeclampsia is new-onset hypertension and proteinuria after 20 weeks of gestation that occurs in 5% of pregnant women worldwide. Preeclampsia is a leading cause of maternal death in the industrialized countries and is associated with severe maternal morbidity. Because delivery is the only known cure, preeclampsia is the leading cause of indicated preterm birth, which imposes major burdens on the family and society because of the need to care for the premature infants. Epidemiologic studies have identified a number of risk factors of preeclampsia. However, the cause of preeclampsia remains poorly understood.
For Editors’ Commentary, see Contents
See Journal Club, page 78
Cigarette smoking is a major risk factor for a number of health problems (such as cancers, heart diseases, diabetes mellitus, and adverse pregnancy outcomes [such as placental abruption, preterm birth, and fetal growth restriction]). Ironically, maternal cigarette smoking is associated with lower risk of preeclampsia. The mechanism for the effect of cigarette smoking on preeclampsia is unknown but cannot be explained by nicotine, confounding factors, and changes in placental morphologic or histopathologic characteristics. Moreover, the use of the tobacco product snuff (noncombusted form) is not associated with a decrease in the incidence of preeclampsia, which has led to the hypothesis that carbon monoxide (CO) that is produced by cigarette smoking may be the substance that underlies the negative association between maternal cigarette smoking and preeclampsia. In addition, preeclampsia is a part of a syndrome of major obstetric complications that include defective deep placentation and includes worsening of preexisting hypertension, intrauterine growth restriction, preterm labor and delivery, late spontaneous abortion, and abruption placenta. Several studies found that CO seems to protect against preeclampsia–associated fetal growth restriction.
In this population based cohort study, we first aimed to test the hypothesis that maternal exposure to moderately elevated level ambient CO is associated with lower risk of preeclampsia.
The second aim is to investigate whether maternal exposure to moderately elevated level ambient CO with preeclampsia has protective effect on fetal growth.
Materials and Methods
We used the 2004-2009 provincial birth record system from BORN Ontario (Better Outcomes Registry & Network available at http://www.bornontario.ca ). This real time, internet-based system captures most of the hospital births in the province. From 2004-2009, data capture in hospitals improved from approximately 82-97% of births. The BORN database contains information on maternal and prenatal characteristics, health services, intrapartum interventions, and maternal and infant outcomes. For the current study, the dataset contained birth records from April 2004 to December 2009.
Ambient CO data were maintained by the Ontario Ministry of Environment and were retrieved from the Air Quality Ontario website ( http://www.airqualityontario.com ). Ambient CO was measured continuously at 23 monitoring sites in Ontario from January 2003 to December 2009. The Trace Level Gas Filter Correlation CO Analyzer (model TE 48CTL; Thermo Environmental Instruments Inc, Franklin, MA) and infrared absorption spectrophotometry were used for the measurement of CO concentration in the monitoring stations. The ambient CO database was linked with the perinatal database by postal code of mother’s residence. To ensure the accuracy and precision of exposure measurement, we included only women whose residence was within 10 km of the closest monitoring station. The date on the CO records was matched temporally to the date of pregnancy, which had been calculated by the date of birth and length of gestation. Maternal exposure to ambient CO during pregnancy was estimated by an average (median) of CO concentration of the monitoring station that was located in the residential area of the mother during the entire pregnancy and was used as the exposure variable in the study. There were no changes in the placement of the monitoring stations or the monitoring instruments during the study period.
To adjust for potential confounding by and to explore the potential effect modification of neighborhood level socioeconomic status (SES) on the association between environmental CO exposure and preeclampsia, we abstracted 2 SES variables from the 2006 Canada Census at the postal-code level: median family income and percentage of university degrees among the adult population who were 25-64 years old. These area-based variables have been shown to be reasonable measures of neighborhood level SES. We then assigned the 2 area-based Canada census variables to each study subject.
From the linked records, we excluded those study subjects who had a multiple gestation or a stillbirth or those whose information on preeclampsia or exposure measure or birth date or gestational age were missing. The baby’s date of birth, mother’s date of birth, and postal codes were deleted from the research file after the record linkage for the protection of privacy and confidentiality.
We first described the distribution of CO concentration by monitoring stations during the study period. Study subjects were divided into 4 groups according to quartiles of CO concentration on the basis of maternal residence. Adjusted odds ratio and 95% confidence interval, which were estimated by multilevel logistic regression model with the use of the SAS GLIMMIX procedure (SAS Institute Inc, Cary, NC), were used to estimate the independent effect of CO on preeclampsia, with the lowest quartile of CO concentration as the reference. Preeclampsia (measured at individual level) was the outcome (dependent) variable, and environmental CO (measured at aggregate level) was the exposure (independent) variable. Two sets of confounding variables were used in the regression analyses: variables that were measured at the individual level (included maternal age, cigarette smoking, chronic health problems [chronic hypertension, type I and II diabetes mellitus, and heart disease], previous cesarean section delivery, and parity) and variables that were measured at the aggregate level (median family income and percentage of university degrees among the adult population who were 25-64 years old). We performed 2 sets of multilevel regression analyses with and without confounding variables that were measured at aggregate level. This way will help us to assess whether the neighborhood variables were important confounding factors. There was no clear order of hierarchic relationship between the exposure variable (ie, environmental CO) and the 2 SES variables (ie, neighborhood income and education levels). We repeated the multilevel logistic regression analysis after stratifying study subjects by SES strata. This stratified analysis can help us to further assess potential effect modification on the association between maternal exposure to environmental CO and preeclampsia. So that we could explore the association between maternal exposure to CO preeclampsia and fetal growth, fetal growth was assessed as z-score of birthweight for gestational age (BWGA) with the following formula: Z = (observed birthweight – mean birthweight)/SD, where mean and SD were based on published Canadian population-based standards that were stratified by infant sex and gestational age in completed weeks. A multiple linear regression model was used to test the associations among the measures of maternal exposure to CO, preeclampsia, and the z-score of BWGA. Models were constructed in which the dependent variable was fetal growth that was represented by the z-score of BWGA; the independent variables were preeclampsia and maternal exposure to CO and an interaction term for preeclampsia and CO exposure. The models were further adjusted by maternal age, parity, smoking, previous cesarean section delivery, maternal health problem (chronic hypertension, types 1 and 2 diabetes mellitus, and heart disease), education, and income. A similar model was built to assess the association between maternal smoking, preeclampsia, and fetal growth.
Supplement analyses with the use of different time windows to define exposure (eg, by trimester or month of pregnancy) were performed to assess the robustness of the results.
All analyses were conducted with SAS software (version 9.2; SAS Institute Inc). The study was approved by both the Ottawa Hospital and the Children’s Hospital of Eastern Ontario Research Ethics Boards.
Results
CO measurements were taken from 23 monitoring stations during the study period; the average CO concentration was 0.25 ppm (lowest, 0.18 ppm; highest, 0.58 ppm; Figure ).
A total of 127,370 eligible study subjects were included in the final analysis. Among them, 2186 subjects (1.72%) had a diagnosis of preeclampsia. Advanced maternal age, primaparity, lower SES, and maternal health problems were associated with an increased risk of the development of preeclampsia in our data ( Table 1 ). On the other hand, maternal cigarette smoking was associated with lower risk of the development of preeclampsia ( Table 1 ).
| Characteristic | Study subjects, n | Studies of preeclampsia, n | Incidence of preeclampsia, % | P value | Crude odds ratio (95% CI) |
|---|---|---|---|---|---|
| Total | 127,370 | 2186 | 1.72 | ||
| Maternal age, y | |||||
| <20 | 1532 | 18 | 1.17 | < .01 | 0.72 (0.45–1.14) |
| 20-34 | 93,852 | 1539 | 1.64 | Reference | |
| ≥35 | 31,986 | 629 | 1.97 | 1.20 (1.10–1.32) | |
| Parity, n | |||||
| 0 | 59,219 | 1404 | 2.37 | < .0001 | 2.19 (1.97–2.43) |
| 1 | 42,775 | 470 | 1.10 | Reference | |
| ≥2 | 23,931 | 293 | 1.22 | 1.12 (0.96–1.29) | |
| Smoking | |||||
| No | 105,365 | 1905 | 1.81 | < .01 | Reference |
| Yes | 10,213 | 127 | 1.24 | 0.68 (0.57–0.82) | |
| Median family income, $ | |||||
| Quartile 1: 15,947-47,099 | 30,348 | 530 | 1.75 | .02 | 1.12 (0.99–1.27) |
| Quartile 2: 47,100-64,028 | 30,206 | 534 | 1.77 | 1.16 (1.02–1.31) | |
| Quartile 3: 64,029-84,307 | 30,202 | 564 | 1.87 | 1.21 (1.07–1.37) | |
| Quartile 4: 84,308-499,358 | 30,402 | 469 | 1.54 | Reference | |
| Education a | |||||
| Quartile 1: 0-15.9 | 30,537 | 622 | 2.04 | < .01 | 1.26 (1.12–1.42) |
| Quartile 2: 16-26.9 | 30,781 | 506 | 1.64 | 1.07 (0.95–1.21) | |
| Quartile 3: 27-44.9 | 30,705 | 502 | 1.63 | 1.02 (0.90–1.06) | |
| Quartile 4: 45-100 | 30,734 | 488 | 1.59 | Reference | |
| Previous cesarean delivery | |||||
| No | 110,154 | 181 | 1.14 | < .01 | Reference |
| Yes | 15,938 | 2001 | 1.82 | 1.61 (1.38–1.88) | |
| Maternal health problem | |||||
| Chronic hypertension | |||||
| No | 125,991 | 2040 | 1.62 | < .01 | Reference |
| Yes | 752 | 129 | 17.15 | 12.58 (10.36–15.28) | |
| Diabetes insulin dependent | |||||
| No | 125,787 | 2126 | 1.69 | < .01 | Reference |
| Yes | 956 | 43 | 4.50 | 2.74 (2.01–3.73) | |
| Diabetes mellitus noninsulin dependent | |||||
| No | 126,066 | 2157 | 1.71 | .90 | Reference |
| Yes | 677 | 12 | 1.77 | 1.04 (0.59–1.84) | |
| Heart disease | |||||
| No | 126,162 | 2153 | 1.71 | .05 | Reference |
| Yes | 581 | 16 | 2.75 | 1.63 (0.99–2.69) |
a Percentage of university degrees among an adult population who were 25-64 years old.
The risk of the development of preeclampsia was associated inversely with maternal environmental CO concentration, with a clear linear dose-response relationship: rates of preeclampsia were 2.32%, 1.97%, 1.59%, and 1.26%, respectively, in the first, second, third, and fourth quartile of CO concentration ( Table 2 ). The inverse association between CO concentration and preeclampsia risk remained the same after adjustment for several confounding factors ( Table 2 ) and across populations with different SES ( Table 3 ).
| Group | Average concentration of carbon monoxide, ppm | Study subjects, n a | Incidents of preeclampsia, n | Incidence of preeclampsia, % | Crude odds ratio (95% CI) | Adjusted odds ratio b (95% CI) | Adjusted odds ratio c (95% CI) | Adjusted odds ratio d (95% CI) |
|---|---|---|---|---|---|---|---|---|
| Per 0.1 ppm increase | 121,158 | 2097 | 1.73 | 0.79 (0.76–0.84) | 0.83 (0.77–0.89) | — | — | |
| Quartile 1 | 0.01-0.16 | 27,357 | 635 | 2.32 | Reference | Reference | Reference | Reference |
| Quartile 2 | 0.17-0.22 | 27,988 | 552 | 1.97 | 0.97 (0.87–1.08) | 0.96 (0.86–1.08) | 0.81 (0.72–0.92) | 0.81 (0.71–0.91) |
| Quartile 3 | 0.23-0.28 | 27,953 | 444 | 1.59 | 0.78 (0.70–0.88) | 0.79 (0.70–0.89) | 0.67 (0.59–0.77) | 0.68 (0.59–0.77) |
| Quartile 4 | 0.29-0.60 | 27,877 | 351 | 1.26 | 0.62 (0.55–0.70) | 0.63 (0.55–0.71) | 0.55 (0.46–0.62) | 0.54 (0.47–0.62) |
a Missing value for control variables is 6212;
b Adjusted for maternal age, parity, smoking, previous cesarean section delivery, maternal health problem (chronic hypertension, types 1 and 2 diabetes mellitus, heart disease), income, and education;
c Multilevel model results with education and income;
| Variable | Carbon monoxide exposure quartile | Carbon monoxide concentration, ppm | Study subjects, n | Studies of preeclampsia, n | Incidence of preeclampsia, % | Odds ratio (95% CI) a |
|---|---|---|---|---|---|---|
| Family income quartiles, $ | ||||||
| Quartile 1: 15,947-47,099 | 1 | 0.02-0.16 | 6653 | 128 | 1.92 | Reference |
| 2 | 0.17-0.22 | 7915 | 136 | 1.72 | 0.71 (0.55–0.92) | |
| 3 | 0.23-0.29 | 7199 | 123 | 1.71 | 0.79 (0.60–1.03) | |
| 4 | 0.30-0.60 | 7415 | 107 | 1.44 | 0.62 (0.47–0.83) | |
| Quartile 2: 47,100-64,028 | 1 | 0.01-0.16 | 6489 | 155 | 2.39 | Reference |
| 2 | 0.17-0.22 | 7936 | 164 | 2.07 | 0.95 (0.75–1.21) | |
| 3 | 0.23-0.29 | 6479 | 97 | 1.50 | 0.65 (0.49–0.86) | |
| 4 | 0.30-0.60 | 7170 | 79 | 1.10 | 0.50 (0.37–0.67) | |
| Quartile 3: 64,029-84,307 | 1 | 0.02-0.16 | 6944 | 197 | 2.84 | Reference |
| 2 | 0.17-0.21 | 7008 | 145 | 2.07 | 0.80 (0.63–1.00) | |
| 3 | 0.22-0.28 | 6224 | 90 | 1.45 | 0.62 (0.48–0.82) | |
| 4 | 0.29-0.60 | 6131 | 77 | 1.26 | 0.52 (0.39–0.69) | |
| Quartile 4: 84,308-499,358 | 1 | 0.02-0.15 | 7063 | 136 | 1.93 | Reference |
| 2 | 0.16-0.23 | 5163 | 99 | 1.92 | 0.91 (0.69–1.20) | |
| 3 | 0.24-0.29 | 7027 | 101 | 1.44 | 0.70 (0.54–0.92) | |
| 4 | 0.30-0.60 | 6562 | 71 | 1.08 | 0.54 (0.40–0.73) | |
| Education quartiles b | ||||||
| Quartile 1: 0-15.9 | 1 | 0.03-0.16 | 5852 | 151 | 2.58 | Reference |
| 2 | 0.17-0.21 | 10,084 | 223 | 2.21 | 0.88 (0.71–1.10) | |
| 3 | 0.22-0.28 | 5520 | 103 | 1.87 | 0.84 (0.63–1.11) | |
| 4 | 0.29-0.60 | 7013 | 105 | 1.50 | 0.60 (0.45–0.79) | |
| Quartile 2: 16-26.9 | 1 | 0.03-0.17 | 5343 | 119 | 2.23 | Reference |
| 2 | 0.18-0.22 | 8036 | 121 | 1.51 | 0.73 (0.56–0.96) | |
| 3 | 0.23-0.29 | 6482 | 110 | 1.70 | 0.84 (0.63–1.12) | |
| 4 | 0.30-0.60 | 6957 | 92 | 1.32 | 0.65 (0.49–0.87) | |
| Quartile 3: 27-44.9 | 1 | 0.01-0.15 | 7140 | 161 | 2.25 | Reference |
| 2 | 0.16-0.23 | 6047 | 120 | 1.98 | 0.84 (0.65–1.09) | |
| 3 | 0.24-0.28 | 7097 | 104 | 1.47 | 0.59 (0.45–0.77) | |
| 4 | 0.29-0.60 | 6496 | 65 | 1.00 | 0.42 (0.31–0.57) | |
| Quartile 4: 45-100 | 1 | 0.02-0.12 | 9225 | 191 | 2.07 | Reference |
| 2 | 0.13-0.24 | 4107 | 82 | 2.00 | 0.94 (0.71–1.25) | |
| 3 | 0.25-0.28 | 8288 | 99 | 1.19 | 0.57 (0.45–0.74) | |
| 4 | 0.29-0.60 | 7238 | 80 | 1.11 | 0.57 (0.43–0.74) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
