BACKGROUND: Screening for illicit drugs in newborns has privacy, social, and legal risks for families of the infants. Established drug-screening criteria may be applied in a manner that considers nonproven risk factors such as race in addition to evidence-based factors.
OBJECTIVE: The goal of this study was to determine if race was used as a criterion for screening infants for intrauterine cocaine exposure. We hypothesized that infants of black mothers would be more likely to be screened regardless of whether they met the standard criteria for screening of our institution's NICU.
METHODS: We used the electronic medical records of newborn infants and their mothers to determine which mother-infant pairs had documented evidence of meeting the criteria for screening infants for prenatal exposure to illicit drugs set forth in the guidelines of our NICU. We then assessed the rates of drug screening to determine the strongest predictors of whether an infant would be screened.
RESULTS: We assessed 2121 mother-infant pairs. Infants born to black mothers were more likely than those born to white mothers to have screening performed whether they met screening criteria (35.1% vs 12.9%; P < .001) or did not (5.3% vs 1.2%; P < .001). In a logistic regression analysis, black race remained independently associated (odds ratio: 2.17 [95% confidence interval: 1.25–3.79]) with drug screening even when we controlled for our standard screening criteria and income, insurance status, and maternal education.
CONCLUSION: Providers seemed to have used race, in addition to recognized risk criteria, as a factor in deciding whether to screen an infant for maternal illicit drug use.
WHAT'S KNOWN ON THIS SUBJECT:
Several studies have shown disparities in the use of race by providers as a factor in determining whether to test infants in the peripartum setting for exposure to illicit drugs.
WHAT THIS STUDY ADDS:
Our study results demonstrated that race was used as a factor for determining whether infants should be screened for illicit drugs, even at an institution with a standard protocol that did not include race as a screening factor.
One of the causes of high-risk and premature infant deliveries is maternal abuse of illicit drugs such as cocaine during the prenatal period. Several prenatal factors, such as limited prenatal care and placental abruption, have been associated with a higher likelihood of drug abuse in mothers.1,–,4 Similarly, factors identified during the infant's neonatal period have also been associated with maternal drug abuse.5,–,8 Because of the reported high incidence of intrauterine cocaine exposure (IUCE) among infants admitted to NICUs 9,10, including our own11, our NICU uses specific guidelines for the screening of infant urine for a cocaine metabolite, a finding that suggests cocaine exposure in utero.
Drug screening in newborns entails significant privacy, social, and legal risks for families. In Monroe County, New York, IUCE may be used as grounds for reports to child protective services. Such reports, in the presence of other social risk factors, may result in removal of the child from parental care. In other states, positive results of drug screening have been used as a basis for alerting law-enforcement authorities, which has led to the arrest and imprisonment of mothers of newborns.12 In Pinellas County, Florida, which requires that findings of illicit drug use during pregnancy be reported to health authorities, discrepancies in reporting of findings that indicate maternal drug use have been described and attributed to the use of race as a factor in determining which mother-infant pairs to screen for illicit drugs. During a 6-month period, the proportion of white women reported for drug use was 1.1%, whereas that for black women was 10.7%. This discrepancy occurred despite similar rates of positive drug-screening results among white and black women (15.4% vs 14.1%) when universal drug screening was performed.13
Although screening for in utero drug exposure may serve a legitimate purpose of facilitating identification of high-risk infants, the application of drug screening in a prejudicial manner places an unfair burden on patients subjected to screening by exposing them disproportionately to child-protective and law-enforcement actions. We pursued further research in this area by investigating the screening practices in an institution at which specific screening guidelines were in place. We explored the possibility that drug screening in our NICU might be applied in a manner that considers factors not included in our screening criteria, such as race, to determine whether an infant should be screened. We hypothesized that infants born to black mothers would be more likely than those born to white mothers to be screened for illicit drugs even if they did not meet the criteria for screening that have been specifically delineated at our institution.
The University of Rochester Medical Center NICU has ∼1100 patient admissions per year. To include a minimum of 2000 mother-infant pairs in our study, we used data for all infants admitted to our NICU during the years 2005 and 2006. Because our institution performs standard maternal data collection on all patients, we were able to include mother-infant data for infants born at our institution (inborn) and for infants born at surrounding hospitals and then transferred to our NICU (outborn).
The protocol for record review was approved by the University of Rochester research subjects review board, which waived the requirement for patient consent. We used our NICU's clinical database to identify all infants admitted from January 1, 2005, through December 31, 2006. This relational database contained demographic information concerning the mother-infant pair, including delivery statistics (including infant gestational age and weight), pregnancy and delivery complications, neonatal diagnoses, admission and discharge notes, and daily updates of the clinical course.
Maternal medical information was obtained from our clinical obstetrical database. The database contained maternal obstetric and demographic data on all obstetrical patients admitted to our hospital. For inborn patients, socioeconomic and clinical data were collected before delivery. Similar data for mothers of outborn patients were collected when these infants were admitted to our institution. Variables for which data were obtained included maternal race, level of maternal education, insurance status, number of prenatal visits, maternal drug and alcohol use, maternal history of sexually transmitted diseases, and delivery complications. Race information entered into the database was the choice selected by the mother from a menu on a questionnaire administered at the time of her admission to the hospital's obstetrics unit.
Data on newborn and maternal toxicology screening were obtained from the hospital's clinical information system. We examined only screening for cocaine use, because we wanted to eliminate possible false-positive results for opiates and benzodiazepines attributable to maternal medication and because marijuana reporting was not required by child protective services.
Infant urine was screened, and the presence of cocaine was confirmed by identification of the metabolite benzoylecgonine. Benzoylecgonine screening was accomplished by using a cloned enzyme-donor immunoassay (CEDIA) (Microgenics Corporation, Freemont, CA) running on a Bayer Advia 2400 (Diamond Diagnostics, Holliston, MA) platform, and confirmed with liquid chromatography–tandem mass spectrophotometry. The CEDIA benzoylecgonine screening assay cutoff point was 300 ng/mL. Only results for urine specimens with confirmed benzoylecgonine were reported as positive.
Study Design and Measures
This study was designed as a retrospective cohort investigation that used race of the mother as the independent variable. The primary outcome variable was whether the infant was screened for in utero drug exposure. The mother-infant pairs were stratified on the basis of whether documented evidence in their electronic record indicated that they met criteria for illicit drug screening set forth in the University of Rochester's 2005 to 2006 guidelines for housestaff in the NICU.
The guidelines listed criteria that should trigger screening of the newborn infant, even in the presence of just a single criterion. The maternal criteria included history of drug abuse, limited prenatal care (<5 prenatal visits), placental abruption, history of sexually transmitted diseases (hepatitis B and C, AIDS, syphilis, gonorrhea), precipitous labor, myocardial infarction, severe mood swings, cerebrovascular accidents, and hypertensive episodes. The infant criteria included unexplained neurologic complications, evidence of possible drug withdrawal (eg, irritability), unexplained intrauterine growth restriction (IUGR), urogenital anomalies, and necrotizing enterocolitis within the first 1 to 2 days after birth.
For the purpose of this study, a mother was considered to have a history of drug use that warranted testing if she had documented evidence of using cocaine, heroin, or amphetamines at any time or of abusing alcohol or marijuana during the current pregnancy. In addition, because of the unreliable coding of IUGR, we chose to use birth weight less than the third percentile as a surrogate marker for IUGR. Maternal race was classified according to pre-2000 census definitions (ie, including Hispanic as a race rather than an ethnicity). Race categories used were white, black, Hispanic, Asian, and other.
For all mother-infant pairs, we searched the electronic records of the delivery and neonatal period for documented evidence that criteria for screening had been met. On the basis of the screening-criteria guidelines, we divided the mother-infant pairs into 2 groups: those who had documented evidence of meeting screening criteria and those who had no documented evidence of meeting screening criteria. We compared the proportion of infants screened, as documented by laboratory records, between the 2 groups. This comparison allowed us to determine if individual mother-infant pairs were appropriately screened or not screened.
All data were entered into and analyzed by a statistical computer program (Stata 7 [Stata Corp, College Station, TX]). We used χ2 or Fisher's exact tests, as appropriate, for analyzing categorical variables, and a value of P < .05 to determine statistical significance. For our first analysis we compared the rates of drug screening among the mother-infant pairs of different races, who were categorized according to whether they had documented evidence of meeting screening criteria. To examine the associations among various factors and their use as independent variables in determining whether an infant was screened, we applied a multivariate logistic regression model as our second analysis. A list of variables was established and entered in a backward step-wise logistic regression model. The list that was developed a priori included, in addition to the screening criteria delineated in the NICU guidelines, maternal race, education, income, and insurance status. All races were compared independently against white race. Two higher levels of maternal education were compared independently against education level less than a high school diploma. Income was grouped according to the national quartile of the median household income for the maternal residence postal code.14 A P value of <.1 was set for inclusion of a variable in the regression, and a P value of ≥.20 was set for the removal of a variable from the model. Our final analysis examined the likelihood that a positive screening result would be obtained in both the absence and presence of specific factors (eg, race, screening criteria). Because of the limited number of positive results, we performed only a bivariate analysis for these data.
Between January 2005 and December 2006, 2121 infants were admitted to our NICU and generated an electronic record in our database. We included all of these infants in our study. We had maternal information for all admissions and complete infant data for 1732 (82%) of the admitted infants. Drug-screening data were available for all 2121 infants. The demographic data of the study population are shown in Table 1. Approximately two-thirds of the mothers were white, and approximately one-fifth were black. Race was not documented for 21 mothers in the study population, and these mothers were excluded from any analysis that used race as an independent variable. Of the infants evaluated, 153 (7.2%) were screened for exposure to an illicit drug, and test results were positive for 13 (8.5%).
Table 2 lists the individual screening criteria and the number of mother-infant pairs who had documented evidence of meeting each specific criterion. The number of infants whose urine was actually screened for drugs when the criterion was present is also shown. The highest rates of screening were found in infants with maternal drug-use history (47% screened) and limited prenatal care (39% screened). The presence of factors such as repeated spontaneous abortions in the mother and neurologic complications and urogenital anomalies in the infant rarely resulted in screening of the infant.
We identified 565 mother-infant pairs as having documented evidence of meeting at least 1 screening criterion (Table 3). For those mother-infant pairs that met screening criteria, 117 (20.7%) of the infants were actually screened. Infants born to black mothers were threefold more likely than those born to white mothers to be screened if they met screening criteria. Thirty-four (2.2%) of the mother-infant pairs who did not have documented evidence of meeting screening criteria were screened. Among pairs that did not meet screening criteria, infants born to black mothers were fourfold more likely to be screened than those born to white mothers.
The results of the multivariate analysis (Table 4) demonstrated several characteristics independently associated with infant drug screening. Among them, race was independently associated with screening of an infant. After we adjusted for other factors, infants born to black or Hispanic mothers were more than twice as likely to be screened for illicit drug use. Other factors independently associated with screening included maternal history of drug use, limited prenatal care, placental abruption, and IUGR. Conversely, as the level of maternal education increased, the likelihood of her infant being screened for drug exposure decreased.
Among our study population, 153 infants were screened for drug exposure, and positive test results indicating exposure to cocaine were found in 13 (8.5%). Among those infants whose urine test results were positive, 7 (12.3% of white infants screened) were white, 3 (4.2%) were black, and 3 (14.3%) were Hispanic (Table 5). Of the 13 infants whose test results were positive, 11 had mothers with a history of drug use. The other 2 positive results were obtained from infants who were screened even though they did not have documented evidence of meeting any screening criteria. Bivariate analyses demonstrated that maternal drug history, limited prenatal care, and precipitous labor were statistically associated with a positive screening-test result.
Our results support the hypothesis that infants born to black mothers were more likely than those born to white mothers to be screened for illicit drugs regardless of whether they met our NICU guidelines for screening. In those mother-infant pairs who met screening criteria, the infant of a black mother was more likely to be screened than a similar white infant who also met screening criteria. These findings suggested that race may have been used as a factor in determining whether an infant was screened, even if other criteria were met. Despite discrepancies across all races in the application of screening criteria, criteria indicating that screening should be performed seemed to be selectively ignored by providers more frequently for infants born to white women. In mother-infant pairs who did not have documented evidence of meeting screening criteria, infants with black mothers were more likely to be screened than infants with white mothers.
Other study results have also shown racial disparities in testing for illicit drug use in the peripartum setting.15,16 These studies revealed that infants of black mothers had 1.5 to 1.9 times the likelihood of being screened. These observations were made at institutions in which there were no protocols for drug testing and decisions to test relied solely on the discretion of the provider. Our study results extend these findings, because they suggest that race was used as an independent criterion for screening even at an institution in which an established, apparently objective, screening protocol that did not include race as a factor was in place.
Previous studies, including an investigation performed in our institution, have revealed that when universal drug testing was done, infants with black mothers had higher rates of positive results than those with white mothers, which suggested that higher screening rates among black infants may have been justified.11 Our current results, however, do not support that conclusion. Screening methods used in our study did not detect any such disparity in the incidence of positive screening results and, therefore, did not justify the higher rates of screening performed in infants with black mothers. If anything, we found the rate of positive screens to be lower among the black population.
The results of our study raise questions about the causes of the racial disparity in infant drug screening. Is it a matter of the providers not following screening criteria appropriately, or are there fundamental flaws in the screening criteria themselves? Other institutions with similar criteria for screening have also reported difficulties in effecting their proper application. Oral and Strang17 found that among infants born in 2002–2003 at the University of Iowa Hospitals and Clinics, 40.2% of the infants who should have been screened on the basis of risk indicators recognized by the hospital protocol were not screened. The investigators suggested that this finding may have been attributable to the fact that this institution, as well as other Iowa hospitals with screening protocols,18 did not have in-service training programs for staff members who performed screening of infants.
We found that even the established criteria in our institutional guidelines were not being applied equally in determining which infants were screened. Maternal drug history was the factor most commonly present among infants who were screened. Other factors associated with high rates of screening were limited prenatal care, placental abruption, and IUGR. However, other factors included in screening criteria, such as infant urogenital anomalies, almost never triggered screening.
This study had several limitations. Urine testing was the method used for detecting IUCE in infants at our institution. Meconium testing is a more accurate way to detect cocaine exposure, and some reports have suggested that urine screens may miss up to 40% of positive results.11,19 However, unless we assume that in our study meconium testing of infants would have differentially revealed false-negative screens for black patients, the type of testing would not affect our conclusion regarding factors associated with screening.
The use of electronic records and the lack of complete electronic data on some infants may also be considered a weakness of this study. Because we obtained data exclusively by searching electronic records we may have missed screening indications found only in patient data recorded on paper. However, the use of uniform coding and ability to perform field text searches allowed us to more completely evaluate a larger number of records than if we had reviewed paper charts. Unless incomplete data were distributed unevenly among racial groups, the inclusion of data from this additional source would not have affected our conclusions. Furthermore, the largest disparity in our study was seen with providers who did not screen infants of mother-infant pairs who were documented to have met screening criteria, a finding unaffected by incomplete risk data.
The ultimate goal of screening for IUCE in the newborn is to effectively identify and treat those infants who are at the highest risk for complications. This goal might best be accomplished if only specific risk factors that have been proven to predict both positive screening results and poorer outcomes were applied strictly and universally. Results of some studies have suggested that only a few screening criteria are actually useful in the prediction of poorer neonatal outcomes.19,20 Similarly, certain studies have shown that only a few screening criteria are associated with positive drug-screening results.15,16,21,–,23 The 2 most commonly identified criteria that may be useful for predicting both poorer outcomes and positive screens are maternal history of drug use and limited prenatal care.16,19,20 Additional research is needed to better identify specific criteria that would help providers more accurately achieve the aims of neonatal drug screening. This goal might be most effectively accomplished by use of protocols that include consistent screening of infants who meet defined, limited criteria and the use of both urine and meconium testing to detect drug exposure.
Screening infants for IUCE is an important component of optimal care for at-risk infants. However, we found that at our institution many infants who met specified criteria were not being screened and that differences in screening decisions made by providers were associated with maternal race. Lack of training and inclusion of several largely ignored (and possibly clinically insignificant) criteria in the screening protocol may have contributed to inequitable application of the screening criteria. A more effective screening protocol might contain fewer criteria, with an emphasis on maternal drug history and prenatal care, factors that are associated with high risks of poor neonatal outcomes.
This work was supported in part by a Strong Children's Research Center summer fellowship.
We thank Ken Edel for help in obtaining the maternal medical records and Dr Sanjiv Amin for assistance with the statistical software.
- Accepted February 8, 2010.
- Address correspondence to Carl T. D'Angio, MD, Golisano Children's Hospital at Strong, University of Rochester Medical Center, 601 Elmwood Ave, Box 651, Rochester, NY 14642. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- IUCE =
- intrauterine cocaine exposure •
- IUGR =
- intrauterine growth restriction
- Telsey AM,
- Merrit TA,
- Dixon SD
- Chiriboga CA,
- Brust JC,
- Bateman D,
- Hauser WA
- Behnke M,
- Eyler FD,
- Conlon M,
- Casanova OQ,
- Woods NS
- 14.↵The Sourcebook of County Demographics. Vienna, VA: ESRI Business Information Solutions; 2003
- Oral R,
- Mueller R,
- Ahmed W
- Ostrea EM Jr.,
- Brady M,
- Gause S,
- Raymundo AL,
- Stevens M
- Copyright © 2010 by the American Academy of Pediatrics