Objective. To determine the mortality rate, during the first 2 years of life, in infants who were exposed to cocaine, opiate, or cannabinoid during gestation.
Methods. For a period of 11 months, a large group of infants were enrolled and screened at birth for exposure to cocaine, opiate, or cannabinoid by meconium analysis. Death outcome, within the first 2 years after birth, was determined in this group of infants using the death registry of the Michigan Department of Public Health.
Results. A total of 2964 infants was studied. At birth, 44% of the infants tested positive for drugs: 30.5% positive for cocaine, 20.2% for opiate, and 11.4% for cannabinoids. Compared to the drug negative group, a significantly higher percentage (P < .05) of the drug positive infants had lower weight and smaller head circumference and length at birth and a higher percent of their mothers were single, multigravid, multiparous, and had little to no prenatal care. Within the first 2 years of life, 44 infants died: 26 were drug negative (15.7 deaths per 1000 live births) and 18 were drug positive (13.7 deaths per 1000 live births). The mortality rate among cocaine, opiate, or cannabinoid positive infants were 17.7, 18.4, and 8.9 per 1000 live births, respectively. Among infants with birth weight ≤2500 g, infants who were positive for both cocaine and morphine had a higher mortality rate (odds ratio = 5.9, confidence interval [CI] = 1.4 to 24) than drug negative infants. Eleven infants died from the sudden infant death syndrome (SIDS); 58% were positive for drugs, predominantly cocaine. The odds ratio for SIDS among drug positive infants was 1.5 (CI = 0.46 to 5.01) and 1.9 (CI = 0.58 to 6.2) among cocaine positive infants.
Conclusion. We conclude that prenatal drug exposure in infants, although associated with a high perinatal morbidity, is not associated with an overall increase in their mortality rate or incidence of SIDS during the first 2 years of life. However, a significantly higher mortality rate was observed among low birth weight infants (≤2500 g) who were positive for both cocaine and opiate.
The effect of antenatal drug exposure on infant mortality has not been studied. Although the infant mortality rate appears increased in areas with a high prevalence of illicit drug use,1 the causal relationship between these two events has not been established for the following reasons: (1) the study requires a large population of control and drug-exposed infants and (2) the methods used to detect gestational drug exposure in newborn infants, eg, maternal history or urine toxicology, are not sensitive enough to identify the true (drug-exposed) population at risk.
We report on the mortality rate of a large (N = 2964), prototype urban population of infants who were screened at birth for exposure to cocaine, opiate, or cannabinoid by meconium analysis and whose death outcome was determined during the first 2 years of life.
From November 1988 to September 1989, we anonymously screened every other infant who was delivered at Hutzel Hospital (a tertiary perinatal center in Detroit) for drugs (cocaine, opiate, and cannabinoid) by meconium analysis. Details of this study population and the sensitivity of the meconium drug test have been previously published.2-4 Thereafter, we determined the death outcome (within 1 to 2 years after birth) of this group of infants, using the death registry of the Michigan Department of Public Health. Infants in the registry who died between November 1988 to September 1990 were matched with the infants in our study population by name, date of birth, and sex. A 2-year mortality surveillance was designed to allow the last set of enrolled subjects to attain their first birthday. Confidentiality of drug use, which was required in the study, was achieved by maintaining and matching two separate sets of databases, one that contained the name, date of birth, sex, and identification number of the infants and the other that contained the infants' identification number and the drug test results. It was further observed that 37% of the infants had a change in their surnames on their discharge from the hospital, principally due to an acknowledgement of paternity by the father. Fortunately, the hospital keeps a registry of the infants' names as they appear on their birth certificate and this registry was used to correct for any change in their names. Data analysis included Pearson χ2 analysis, odds ratio, and its 95% confidence interval (CI) on cross-tabulated data and Student's t test for comparison of independent sample means. The relationship between cocaine, opiate, or cannabinoid in meconium to death outcome in the infants were analyzed by logistic regression, using the method of Hosmer and Lemshow.5P values are given without adjustment. Data analysis was done using SPSS for Windows software (Version 6.1). The study was approved by the Human Investigations Review Board of Wayne State University.
A total of 3879 infants was eligible for the study; however, only 3010 (78%) were enrolled because meconium was not tested in 869 (22%) infants for the following reasons: no meconium collected due to infant death, transfer, or early discharge (N = 131), meconium lost or not collected by the mother (N = 185), and insufficient meconium obtained (N = 553). Data on the latter group are not available. Among the 3010 infants tested, only 2964 were eligible for this study because the names of 45 infants were not found in the hospital's birth certificate registry.
The weight distribution of the study population (N = 2964) was as follows: ≤1000 g (0.3%), 1001 to 1500 g (1.3%), 1501 to 2000 g (3.7%), 2001 to 2500 g (10.7%) and >2500 g (84%) and their gestational age distribution was: ≤30 weeks (0.8%), 31 to 34 weeks (3.4%), 35 to 37 weeks (14.9%), and ≥38 weeks (80.8%). The socioeconomic status of the study population is inferred from their health care coverage: 48.9% were Medicaid recipients, 20.3% were health maintenance organization participants, 21.8% had private insurance, and 9% had no medical insurance.
The results of the meconium drug test at birth were as follows: 1306 (44.1%) infants were positive for drug(s) wherein 903 (30.5%) were positive for cocaine, 599 (20.2%) were positive for opiate and 338 (11.4%) were positive for cannabinoid. Polydrug exposure was present in 479 (16.2%) infants: 424 (14.3%) were positive for two drugs, predominanantly cocaine and opiate (10.1%) and 55 (1.9%) were positive for three drugs. A fourfold increase in the incidence of gestational drug exposure was detected by meconium analysis as compared to maternal history (44.1% vs 11.1%). About 9.8% of the infants had a positive maternal history of drug use and a positive meconium drug screen, 34.4% negative maternal history and a positive meconium drug screen, and 54.4% negative maternal history and negative meconium drug screen.
Clinical profiles of the mother and infant based on the drug test results are shown in Table 1. A significantly (P < .05) higher percentage of women in the drug positive group were service (nonprivate) patients, single, multigravid (≥3), and multiparous (≥3) and had little to no prenatal care (defined as having none or one prenatal visit during the entire pregnancy) compared to the drug negative group. Similarly, the weight, head circumference, and length at birth were significantly lower among the drug positive, compared to drug negative neonates. However, for exposure to a single drug (except cocaine), eg, opiate or cannabinoid, there were fewer significant differences noted between the drug positive and drug negative groups.
Within 1 to 2 years after birth, 44 infants died. Except for a higher frequency of Apgar score ≤6 at 5 minutes in the infants who died (4.5% vs 1.1%, P < .03), there were no significant differences (P > .10) in the marital status, race (P > .3), service, gravidity and parity of the mother, and gestational age (P > .3), Apgar score at 1 minute, sex, weight, length, and head circumference (P > .6) between infants who died compared to those who lived. The cause of death, available in 42 infants, are shown in Table 2.
The infant mortality rates are shown in Table 3. Among the 44 infants who died, 26 were drug negative (15.7 per 1000 live births) and 18 were drug positive at birth (13.7 per 1000 live births). There were no significant differences (P > .10) in the marital status, race, service, gravidity and parity of the mother, and gestational age, Apgar score at 1 minute, sex, weight, length, and head circumference between infants who were drug positive vs drug negative. Mortality rates between the drug-positive group or specifically, the cocaine-positive, morphine-positive, or cannabinoid-positive groups were not significantly different from the drug-negative groups (P > .3). In the group exposed to a single drug, there was also no significant difference in the mortality rate between cocaine only, morphine only, or cannabinoid vs the drug negative group (P > .1). Similarly, no significant difference (P > .05) was found in mortality rate between cocaine only vs cocaine plus other drugs, morphine only vs morphine plus other drugs, and cannabininoid only vs cannabinoid plus other drugs. The odds ratio for mortality in drug positive infants was 0.8 (CI = 0.47 to 1.61); in infants positive for cocaine = 1.3 (CI = 0.71 to 2.43), in infants positive for morphine = 1.3 (CI = 0.67 to 2.63) and positive for cannabinoid = .56 (CI = 0.17 to 1.83). By birth weight category, the odds ratios for death occurring in drug-positive infants are shown in Table 4. Among infants with birth weight ≤2500 g, a trend toward a high mortality rate was found among infants who were positive for cocaine (odds ratio = 4.83, CI = 0.96 to 24.17) and a significantly higher mortality rate (odds ratio = 5.9, CI = 1.4 to 24) among infants who were positive for both cocaine and morphine. In this group, there was also a significant (P < .05) difference in the incidence of vaginal delivery, gravidity >3 and parity >3 between the cocaine-positive vs drug-negative groups or cocaine- and morphine-positive vs drug-negative groups. However, there were no significant differences (P > .1) in their Apgar scores at 1 and 5 minutes, gestational age, sex, weight, length, head circumference, and maternal marital status and service.
Of the variables in the logistic regression analysis, eg, marital status, gravidity, parity, race, sex of infant, Apgar score at 5 minutes, birth weight, length, head circumference, gestational age, single drug exposure, or a combination of two or more, only gravidity, Apgar score at 5 minutes, and taking of a combination of two or more drugs were significant at P < .25 and were considered with interactions, for a stepwise inclusion with the likelihood ratio criteria. When interactions were significant, the data were examined to see if there was any meaningful result. However, none of the variables yielded any useful result, the fit was bad, and the prediction gave; at best, only 1/29 predicted as dying suggesting that either relevant variables were not included or that there was no relationship. For those with birth weight <2500 g, a similar situation arose except that the presence of taking two drugs was significantly higher in those who died and for a higher Apgar score (≥7) at 5 minutes, drug taking was more likely to be present for those who died (P< .018, two-sided). This was also true for the entire population (P < .09, two-sided).
Most of the infant deaths occurred within the first month of life (n = 28 or 63.3%): 19 were drug negative and 9 were drug positive (3 were positive for cannabinoid, 8 positive for cocaine, and 6 positive for morphine). There was no significant difference (P > .10) in the marital status, service, gravidity, and parity of the mother and Apgar scores ≤6 at 1 and 5 minutes, gestational age, sex, birth weight, head circumference, and length between the drug-positive vs drug-negative infants.
Among the drug-positive infants, there was no relationship between the concentrations of cocaine, opiate, and cannabinoid in meconium and infant death. Furthermore, with meconium drug concentrations grouped into <1000> ng/mL for cocaine or opiate and <100> ng/mL for cannabinoid, there was no significant difference in the infant mortality rate based on grouped drug concentrations (P > .17).
Sudden infant death syndrome (SIDS) occurred in 11 infants: 5 were drug negative (3.0 deaths per 1000 live births) and 6 (4.9 deaths per 1000 live births) were drug positive. The odds ratio for SIDS among the drug positive infants was 1.5 (CI = 0.46 to 5.01). A total of 54% of infants with SIDS was positive for drugs, predominantly cocaine. The incidence of SIDS in cocaine positive infants was 5.5 per 1000 live births (odds ratio = 1.9, CI = 0.58 to 6.3) and in opiate positive infants, 6.7 deaths per 1000 live births (odds ratio = 2.3, CI = 0.66 to 7.76).
Many studies have documented the immediate and long-term morbidity of antenatal exposure of infants to illicit drugs6-15; however, none has reported on its effect on infant mortality. Several reasons may account for this. The need for a large study population and an accurate method to detect the infant population at risk are foremost obstacles. These drawbacks were overcome in this study by the enrollment of a large (N = 2964) number of infants who were inborn and prototype urban and who were screened for drugs by meconium analysis, a method more sensitive and specific in detecting gestational drug exposure in infants than maternal history or urine toxicology.3,16-19
Some potential limitations of this study are recognized. First, the death outcome of the infants relied exclusively on the death registry of the State of Michigan Department of Public Health. Although the death registry in Michigan is accurate because Michigan law mandates the reporting of all deaths within the state, the exception will be if the infant dies outside of Michigan. High mobility is characteristic of maternal addicts in our population as evidenced by the frequent change of their telephone numbers or home addresses. However, our experience has also been that the mobility of the maternal addicts is confined principally to within the city, rather than out of the state,20 for the following reasons: (1) most have few to no families to go to, (2) they are familiar with the local support system and feel insecure elsewhere, and (2) there is very little motivation to leave. Second, by virtue of the study design, the period of follow-up of infants for their death outcome ranged from 1 to 2 years and with more years of follow-up encountered in those who were enrolled early in the study. However, the impact of this discrepancy was minimal because most deaths, if they occurred, were observed within the first 11 months of age.
There is also some underrepresentation of infants of very low birth weight or gestation in the study. At the time of the study, infants who weighed <1000 g or who were <30 weeks of gestation constituted about 1.8% and 3.0%, respectively, of the total live births in our hospital, in contrast to 0.3% and 0.8%, respectively, in the study population. The main reason for failure to include more infants in this group was the insufficient amount of meconium that could be obtained from the very low birth weight infants for drug analysis. Underrepresentation of the very low birth weight or very premature infants may result in a falsely low infant mortality rate. However, we have also shown that about 50% of infants in our neonatal intensive care unit are drug positive, principally from cocaine.21Thus, the very low birth weight and very premature infants, although underrepresented in the study, may be assumed to be equally distributed into both the drug-positive and drug-negative groups.
The third potential problem that was encountered in the study was the high rate (37%) of name change among infants discharged from the hospital. Fortunately, our hospital, cognizant of this practice, has established a registry containing both the original and revised (referred to as aka or “also known as”) surnames of the infants based on information in their birth certificate. By virtue of this registry, we were able to verify the discharge names of 98% of the infants in the study population.
Our study demonstrated a high percentage (34.4%) of infants who were found to be positive for drugs by meconium drug screening but whose mothers denied the history of illicit drug use. This emphasizes the importance of using a sensitive screening method to identify the infants at risk, particularly if outcome measures are studied. Many factors contribute to maternal denial of drug use foremost of which is the fear of the consequences stemming from such admission. The history of drug use was obtained from the mother as part of her routine admission history interview. However, it has been shown that drug history taking obtained by a more structured interview and in a nonthreatening environment will significantly improve the rate of maternal admission to illicit drug use.22
Overall, our study did not demonstrate a significant increase in mortality, within 1 to 2 years after birth, in infants who were antenatally exposed to drugs (13.7 vs 15.7 deaths per 1000 live births). Our findings seem consistent with several follow-up studies showing that the long-term morbidity of gestationally drug-exposed infants is not as serious as was originally expected.13,22Nonetheless, the likelihood of a type II error in our observation cannot be excluded. To achieve at least 80% power to detect the difference between 0.137 and 0.157 that was seen in our study, at a 5% significance level, 5000 subjects would have been required in each part of the study. However, to study such a large population of infants, which includes drug testing, would be impractical and costly. However, our sample size was adequate to detect with 80% power at the 5% significance level, a difference exceeding 4% between proportions of the magnitude we saw; ie, if the proportions were 0.117 and 0.157 or 0.137 and 0.177.
A higher mortality rate was observed among low birth weight (≤2500 g) infants who were positive for both cocaine and opiate. However, both low birth weight and prematurity are known consequences of illicit drug use during pregnancy and are also important determinants of infant morbidity and mortality. Whether drug (cocaine or opiate) exposure, prematurity, and low birth weight either alone or in combination contributed to the higher mortality in this group of infants cannot be ascertained from this study due to the small number of infants who died.
In contrast to what has been reported in the literature,24,25 the incidence of SIDS was not significantly increased among the drug-positive infants. This may be due, in part, to the wide spectrum of drug-exposed infants that could be detected by the meconium drug test. Among opiate-exposed neonates, the risk to SIDS has been shown to relate to the severity of the drug withdrawal that in turn, correlated to the degree of maternal drug use.24 Past studies that have demonstrated a high association of SIDS to drug abuse were based on drug detection by maternal history or urine toxicology.24,25 We have shown, by quantitative meconium drug analysis, that women who admit to drug use during pregnancy (positive maternal history) are commonly heavy or consistent users of drugs.3 Thus, the detection of a wide range of drug exposure in the infants, by meconium analysis, can lead to a comparatively lower incidence of SIDS.
We conclude that prenatal drug exposure in infants, although associated with high perinatal morbidity, is not associated with an increase in their mortality rate during the first 2 years of life. However, among low birth weight infants (≤2500 g) exposed to both cocaine and opiate, a significantly higher mortality rate was observed.
This study was supported in part by a grant from the Michigan Department of Public Health and a grant from the National Institute on Drug Abuse (1 RO1 DA06821–01A1). We thank the Michigan Department of Public Health for its invaluable assistance in this study.
- Received July 26, 1996.
- Accepted November 14, 1996.
Reprint requests to (E.M.O.) Department of Pediatrics, Hutzel Hospital, 4707 St Antoine, Detroit, MI 48201.
- CI =
- confidence interval •
- SIDS =
- sudden infant death syndrome
- ↵Office of the State Registrar and Center for Health Statistics, Michigan Department of Public Health, 1993
- Ostrea EM,
- Brady M,
- Gause S,
- Raymundo A,
- Stevens M
- Ostrea EM,
- Parks P,
- Brady M
- ↵Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons, Inc; 1989:82–173
- Wilson GS,
- McCreary R,
- Kean J,
- Baxter J
- Chasnoff IJ,
- Hatcher R,
- Burns WJ
- Ostrea EM,
- Martier S,
- Welch R,
- Brady M
- Bandstra ES,
- Stele BW,
- Chitwood DD,
- et al.
- Delaney-Black V,
- Covington C,
- Templin T,
- Sokol R,
- Ager J,
- Martier S
- Ostrea EM,
- Lizardo E,
- Tanafranca M
- Chasnoff IJ,
- Griffith DR,
- Freier C,
- Murray J
- Pierson PS,
- Howard P,
- Kleber HD
- Copyright © 1997 American Academy of Pediatrics