Among the neonatal population included, there were 2671 survivors and 93 nonsurvivors. To convert hematrocrit to proportion of 1.0, multiply by 0.01.
eTable. Comparison of Neonates With and Without Preoperative Hematocrit.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Goobie SM, Faraoni D, Zurakowski D, DiNardo JA. Association of Preoperative Anemia With Postoperative Mortality in Neonates. JAMA Pediatr. 2016;170(9):855–862. doi:10.1001/jamapediatrics.2016.1032
Neonates undergoing noncardiac surgery are at risk for adverse outcomes. Preoperative anemia is a strong independent risk factor for postoperative mortality in adults. To our knowledge, this association has not been investigated in the neonatal population.
To assess the association between preoperative anemia and postoperative mortality in neonates undergoing noncardiac surgery in a large sample of US hospitals.
Design, Setting, and Participants
Using data from the 2012 and 2013 pediatric databases of the American College of Surgeons National Surgical Quality Improvement Program, we conducted a retrospective study of neonates undergoing noncardiac surgery. Analysis of the data took place between June 2015 and December 2015. All neonates (0-30 days old) with a recorded preoperative hematocrit value were included.
Anemia defined as hematocrit level of less than 40%.
Main Outcomes and Measures
Receiver operating characteristics analysis was used to assess the association between preoperative hematocrit and mortality, and the Youden J Index was used to determine the specific hematocrit cutoff point to define anemia in the neonatal population. Demographic and postoperative outcomes variables were compared between anemic and nonanemic neonates. Univariate and multivariable logistic regression analyses were used to determine factors associated with postoperative neonatal mortality. An external validation was performed using the 2014 American College of Surgeons National Surgical Quality Improvement Program database.
Neonates accounted for 2764 children (6%) in the 2012-2013 American College of Surgeons National Surgical Quality Improvement Program databases. Neonates inlcuded in the study were predominately male (64.5%), white (66.3%), and term (69.9% greater than 36 weeks’ gestation) and weighed more than 2 kg (85.0%). Postoperative in-hospital mortality was 3.4% in neonates and 0.6% in all age groups (0-18 years). A preoperative hematocrit level of less than 40% was the optimal cutoff (Youden) to predict in-hospital mortality. Multivariable regression analysis demonstrated that preoperative anemia is an independent risk factor for mortality (OR, 2.62; 95% CI, 1.51-4.57) in neonates. The prevalence of postoperative in-hospital mortality was significantly higher in neonates with a preoperative hematocrit level less than 40%; being 7.5% (95% CI, 1%-10%) vs 1.4% (95% CI, 0%-4%) for preoperative hematocrit levels 40%, or greater. The relationship between anemia and in-hospital mortality was confirmed in our validation cohort (National Surgical Quality Improvement Program 2014).
Conclusions and Relevance
To our knowledge, this is the first study to define the incidence of preoperative anemia in neonates, the incidence of postoperative in-hospital mortality in neonates, and the association between preoperative anemia and postoperative mortality in US hospitals. Timely diagnosis, prevention, and appropriate treatment of preoperative anemia in neonates might improve survival.
Neonates have a high incidence of perioperative mortality worldwide.1-5 In a 2015 study, the incidences of 24-hour and 30-day mortality in children were 13.3 per 10 000 operations and 41.6 per 10 000 operations, respectively, while a higher 30-day postoperative mortality rate of 386.5 per 10 000 was reported in the neonatal population.2 In another study, postoperative 24-hour and 30-day mortality in children were reported as 0.98 deaths per 10 000 operations while both the 24-hour (168.7 deaths per 10 000 operations) and 30-day (350.4 deaths per 10 000 operations) mortality was substantially higher in neonates.1 To our knowledge, incidence of postoperative neonatal mortality in US hospitals has never been reported.
This study sought to define the neonatal mortality rate in US hospitals and determine risk factors for postoperative mortality in neonates undergoing noncardiac surgery. Preoperative anemia is an independent risk factor for mortality in adults,6-9 but to our knowledge, this has not been investigated in the pediatric population. We hypothesized that preoperative anemia is an independent risk factor for in-hospital mortality in neonates.
Question Is preoperative anemia associated with increased postoperative neonatal mortality in US hospitals?
Findings Using the American College of Surgeons National Surgical Quality Improvement Program database, preoperative anemia of a hematocrit level of less than 40% (found in 32% of neonates) was associated with increased postoperative in-hospital mortality.
Meaning Timely diagnosis, prevention, and appropriate treatment of preoperative anemia in neonates might improve outcome and survival.
This study was performed using data from the 2012 and 2013 pediatric databases of the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP). The Boston Children’s Hospital instituional review board did not require approval for this study because it was a database analysis. The ACS NSQIP was developed to contribute to nationwide quality improvement and research in the surgical specialties. The ACS NSQIP Pediatric database collects deidentified data on children younger than 18 years undergoing noncardiac surgery and includes 129 variables including preoperative risk factors, intraoperative characteristics, 30-day postoperative outcomes, and mortality in both the inpatient and outpatient settings.10 A systematic sampling strategy is used to avoid bias in case selection and to ensure a diverse surgical case mix. A site’s trained and certified Surgical Clinical Reviewer captures data using a variety of methods including medical record review. Adverse events and comorbidities reported in the database are determined by strict inclusion criteria. To ensure the quality of the data collected, the ACS NSQIP Pediatric conducts interrater reliability audits of selected participating sites.11 The results of the audits completed to date reveal an overall disagreement rate of approximately 2% for all assessed program variables. For the databases, exclusion criteria included patients 18 years and older, trauma cases, solid organ transplantation, and patients undergoing multiple procedures performed by different surgical teams under the same anesthetic. In addition, cases coming from hospitals with an interrater reliability audit disagreement rate greater than 5% or a 30-day follow-up rate less than 80% were excluded.
We included all children recorded in the 2012 and 2013 ACS NSQIP Pediatric databases with a recorded preoperative hematocrit value. Quiz Ref IDExclusion criteria were children who had a preoperative transfusion and neonates with congenital heart disease. Neonates were defined as 0 to 30 days old.
The following demographic variables were included: sex, race/ethnicity, age, American Society of Anesthesiologists physical status classification score (ASA class), body weight, prematurity (<24 weeks, 24-36 weeks, and >36 weeks), type of procedure (elective vs emergent surgery), preoperative mechanical ventilation, preoperative inotropic support, sepsis, need for supplemental oxygen preoperatively, hematocrit level, red blood cell (RBC) and blood product transfusion, surgical procedure category (general surgery, neurosurgery, urology, and orthopedic surgery), and surgical relative value unit as a measure of surgical complexity.12 Postoperative in-hospital mortality was assessed.
Continuous data are presented as median and interquartile range (IQR), and categorical variables are expressed as number and percentage. Owing to the lack of a universally accepted definition of anemia in neonates and also because of the wide reference range of acceptable Quiz Ref IDhematocrit values for this age group, we used preoperative hematocrit as a continuous variable in receiver operating characteristic (ROC) curve analysis to assess the relationship between preoperative hematocrit and mortality and for determining the optimal cutoff (Youden J Index) to define anemia in the neonatal population. Demographic and postoperative outcome variables were compared between anemic and nonanemic neonates using the Wilcoxon rank sum test for continuous variables and Pearson χ2 test for categorical variables. Univariate logistic regression analysis was performed to identify factors associated with in-hospital mortality. Cutoff values for continuous variables were determined using the Youden J index in ROC curve analysis. Multivariable logistic regression using a backward selection procedure with the likelihood ratio test was applied to determine independent predictors. Area under the ROC curve was used to assess model performance in terms of predictive accuracy. A generalized estimating equations approach with a logistic model was used to calculate 95% CIs around the estimated prevalence, odds ratio (OR), and predicted probabilities with respect to in-hospital mortality for unadjusted as well as adjusted analysis (controlling for 4 covariates: body weight ≤2 kg, ASA class 3 or higher, preoperative inotropic support, and preoperative mechanical ventilation).
Multivariable logistic regression was applied using maximum likelihood estimation to determine the probability of in-hospital mortality for anemic and nonanemic patients based on a combination of the presence or absence of the significant predictors (ASA class, body weight, preoperative inotropic support, and preoperative mechanical ventilation). Because there were 5 significant multivariable predictors, including hematocrit level less than 40% (to convert to proportion of 1.0, multiply by 0.01), there were 32 possible combinations, and for clarity purposes we elected to organize these to facilitate clinical interpretation as 2 separate columns of probabilities, 1 for anemic neonates (hematocrit level <40%) and 1 for nonanemic neonates (hematocrit level ≥40%). We used a generalized linear model approach (GENLIN in SPSS) with a binomial distribution and a logit link function for multivariable logistic regression analysis to determine the probability of in-hospital mortality for each combination of predictors in neonates with and without preoperative anemia (ie, hematocrit <40%) with an estimated 95% CI around each probability so that clinicians could get a good sense of the range of possible probabilities of in-hospital mortality for each specific combination of predictors.
The 2014 NSQIP database was used as an independent source to confirm the validity and generalizability of our analysis and to ensure that the clinical predictors found in the 2012 and 2013 analysis remained valid in an independent external cohort. Evidence of good validation was defined as an area under the curve of at least 0.700, the accepted level required as evidence of robust validation in other studies.13
A 2-tailed P value less than .05 was considered statistically significant for all tests. Statistical analyses were performed using IBM SPSS Statistics, version 23.0 (IBM).
Among the 114 395 children included in the 2012 to 2013 ACS NSQIP databases, 40 897 children had a preoperative hematocrit recorded. Neonates accounted for 2764 (7%) and had a high (n = 93 of 2764; 3.4%) postoperative in-hospital mortality compared with 2288 of 38 133 (0.6%) in all age groups (0-18 years). Preoperative hematocrit values for neonates in the database ranged from 24% to 52% (median, 38; IQR, 34-41). Preoperative hematocrit was analyzed as a continuous variable and a preoperative hematocrit level less than 40% was identified as the optimal cutoff point (Youden J index) to predict overall mortality.
The results of the ROC analysis showed that preoperative hematocrit was associated with postoperative in-hospital mortality (area under the curve, 0.747; 95% CI, 0.683-0.800). The incidence of preoperative anemia in neonates using this definition was 32% (892 of 2764). The distribution of hematocrit values for the neonatal population (survivors and nonsurvivors) is shown in the Figure; 825 of 2671 neonatal survivors (31%) had a preoperative hematocrit level of 40% or more, while 67 of 93 of neonates who died (72%) had an hematocrit level of less than 40%. Table 1 compares the univariate factors between anemic and nonanemic neonates.
Univariate analysis of factors associated with in-hospital mortality are shown in Table 2. Significant factors by univariate analysis that were associated with increased in-hospital mortality in neonates included race/ethnicity, ASA class, body weight, prematurity, preoperative mechanical ventilation, preoperative inotropic support, sepsis, oxygen supplementation, preoperative hematocrit level, and RBC transfusion. There was no significant difference in postoperative in-hospital mortality with respect to surgical procedure category nor surgical complexity as measured by relative value unit (both mortality groups relative value unit, 20; IQR, 13-28 vs survivors relative value unit, 19; IQR, 11-26 with mortality; P = .18).
Using multivariable regression analysis, we observed that preoperative anemia was an independent risk factor for postoperative in-hospital mortality in neonates (Table 3). Quiz Ref IDOther independent risk factors included ASA class 3 to 5, body weight 2 kg or less, preoperative mechanical ventilation, and preoperative inotropic support.
Quiz Ref IDThe prevalence of postoperative in-hospital mortality was significantly higher with a preoperative hematocrit level less than 40%; 7.5% (67 of 892; 95% CI, 6.0%-9.4%) vs 1.4% (26 of 1872; 95% CI, 1.0%-2.0%) with a preoperative hematocrit level of 40% or more for the unadjusted analysis. When adjusting for the 4 significant predictors of mortality (body weight 2 kg or less, ASA class 3 or higher, preoperative inotropic support, and preoperative mechanical ventilation), the prevalence of neonatal postoperative mortality remained significantly higher with a preoperative hematocrit level less than 40% (3.5%; 95% CI, 1.3%-9.0% vs 1.3%; 95% CI, 0.4%-4.1% for preoperative hematocrit of 40% or more).
Furthermore, the OR of postoperative in-hospital neonatal mortality was significantly higher with a preoperative hematocrit level less than 40% (5.8; 95% CI, 3.6-9.1) for unadjusted analysis. When adjusting for the 4 significant predictors of mortality (body weight ≤2 kg, ASA class 3 or higher, preoperative inotropic support, and preoperative mechanical ventilation), the OR of neonatal postoperative mortality remained significantly higher with a preoperative hematocrit level less than 40% (2.62; 95% CI, 1.51-4.57). Although more anemic than nonanemic neonates were transfused (25% [219 of 892] vs 6% [116 of 1872] P < .001), RBC transfusion, while significant in univariate analysis as a predictor of mortality, was not an independent risk factor associated with in-hospital mortality in multivariable analysis.
Area under the ROC curve was calculated to assess the strength of the association between the multivariable risk factors and postoperative in-hospital mortality (area under the curve, 0.912; 95% CI, 0.883-0.942).
A neonatal mortality risk probability was designed based on the different multivariable predictors identified including a hematocrit level of less than 40% (Table 4). For all combinations of these risk factors, a hematocrit level of less than 40% was associated with a significant increased probability for postoperative in-hospital mortality in neonates. A neonate with all 4 independent risk factors (ASA class 3-5, body weight ≤2 kg, on inotropic support, and on mechanical ventilation preoperatively) had a 0.234 probability of postoperative mortality with a hematocrit level of 40% or greater vs a 0.431 probability with a hematocrit level of less than 40%.
The 2014 ACS NSQIP database, which included 1384 neonates, was used in validation analysis applying the same inclusion and exclusion criteria. The postoperative in-hospital neonatal mortality rate was 2.0%, accounting for a total of 28 deaths. In the infants who died postoperatively in hospital, there was also a statistically significant increased incidence of anemia (defined as hematocrit level <40%) compared with those who survived. The difference in hematocrit between the nonsurvivors and survivors (n = 1356) was significant, with a median hematocrit level of 36.9% (IQR, 28.6%-42.8%) in nonsurvivors compared with a median hematocrit level of 43.4% (IQR, 37.8%-49.4%) in survivors (P < .001). Furthermore, in this independent cohort, 3.4% (16 of 464) of neonates who had a hematocrit level of less than 40% died, while 1.2% (12 of 970) of neonates who had a hematocrit level of 40% or greater survived (P = .01, Fisher exact test). Finally, the results of an ROC analysis in this independent cohort confirmed that preoperative hematocrit was associated with in-hospital postoperative mortality (area under the curve, 0.706; 95% CI, 0.613-0.800; P < .001).
Because only neonates who had a recorded hematocrit level were included in this analysis, the demographics of those neonates excluded for lack of a recorded preoperative hematocrit level were reported for comparison (see the eTable in the Supplement).
We report that neonates undergoing noncardiac surgery in US hospitals had a higher postoperative in-hospital mortality rate (3.4% [93 of 2764]) than older children in whom the postoperative mortality rate is 0.6% (2288 of 38 133). This mortality rate of 340 per 10 000 operations in neonates is similar to the one previously reported in other countries.1,2,14 The incidence of preoperative neonatal anemia in this database, defined as a hematocrit level less than 40%, is 32% (892 of 2764). More than 70% of neonates who died had a hematocrit level less than 40% compared with 31% of survivors who had a preoperative hematocrit level of 40% or higher. To our knowledge, this is the first study of a strong association between preoperative anemia and postoperative mortality in neonates undergoing noncardiac surgery. Postoperative in-hospital mortality was higher in neonates with preoperative anemia (hematocrit level <40%) when controlling for the other independent risks factors: ASA class, body weight, preoperative inotropic support, or preoperative mechanical ventilation. This association held despite our liberal definition of neonatal anemia, hematocrit level less than 40%, which is considered mild or at most moderate anemia.
Additionally, the association between preoperative hematocrit levels less than 40% and increased mortality in neonates was validated in an independent external cohort; the 2014 ACS NSQIP database. This further generalizes the finding that preoperative anemia is associated with postoperative mortality in neonates.
Anemia is not well defined in neonates. Normal reference ranges are derived from large data sets, using blood drawn in neonates with minor pathologic conditions, and vary based on gestational and postnatal ages.15,16 A mean (SD) hematocrit level for term newborns was first reported in 1986 to be 41.2% (6.02%).17 Among a large population of full-term neonates, the reference range at birth for hematocrit varies between 42% and 65%,15 while an authoritative source states a reference hematocrit level ranges from 41.5% to 56% during the neonatal period.18 It is also widely accepted for newborns that a normal hematocrit level equals the gestational age at birth.16 Owing to the lack of a universally accepted definition of anemia in neonates and also because of the wide reference range of acceptable hematocrit values for this age group, we chose to use preoperative hematocrit as a continuous variable, and a preoperative hematocrit level less than 40% was determined to be the optimal cut point (Youden) to discriminate the risk for neonatal mortality. This definition of neonatal anemia of a hematocrit level less than 40% is supported by a 2016 review.19
Preoperative neonatal anemia is likely to be multifactorial in origin. Iatrogenic blood loss is the most common cause of preoperative premature infants owing to frequent in-hospital blood draws or from invasive procedures.20 Daily phlebotomy blood loss of up to 5% of an infant’s total blood volume is not uncommon in US hospitals.21 Premature infants are the highest risk group because they may have a transient deficiency of erythropoietin and iron deficiency together with blood loss, chronic disease, and infection.22
While data within NSQIP do not support any specific therapeutic intervention, timely diagnosis, prevention, and appropriate treatment of preoperative anemia in neonates might improve outcomes and survival. Prevention is imperative and includes delayed cord clamping at birth, limiting number and volume of frequent blood draws, and reducing wastage of blood during invasive procedures. Advances have led to microtubes, the development of computerized point-of-care testing devices attached in a closed system to intravenous catheters, which perform precise analysis with small volume, rapid turnover times, and less waste. Laboratory phlebotomy loss in neonatal intensive care units in the first 6 weeks of life can average 11 to 22 mL/kg/wk with nearly identical volumes of RBCs transfused over the same period.23,24 Erythropoiesis cannot keep up with this loss. Reducing blood draws directly correlates with reduced transfusion.25
Treatment includes enteral or intravenous iron therapy with folate and vitamin E and recombinant human erythropoietin to further increase RBC production. Transfusion may be a therapeutic modality to consider but may also be associated with risks. Therefore, alternative strategies should be used first and foremost. Little is known about the benefits of transfusing neonates at specific hematocrit targets. Therefore, the practice varies widely in neonatal intensive care units. The evidence to support a restrictive vs liberal transfusion strategy in neonates26 is not nearly as robust as that which exists for older children.27 Transfusion rates for neonates have stayed consistent over the last 10 years, with an incidence of 5.4 per 1000 live births.26 While the highest transfusion rates are seen with hemolytic disorders, congenital abnormalities requiring surgery, and in premature babies, 40% of those transfused are more than 32 weeks’ gestation.28 An estimated 60% to 80% of very low-birth-weight and premature infants receive RBC transfusions to treat anemia, 70% in the first week of life.29 Ultimately, the clinical status of the neonate should be taken into consideration to prevent and treat end organ ischemia using specific protocols.19,20,30,31
Our results call into question the trend of using a restrictive vs a liberal hematocrit strategy in neonatal blood management. Given that mild/moderate anemia in neonates has a strong independent association with mortality, consideration may be given to a target hematocrit level of 40% or more in high-risk infants (ASA status 3-5, weight ≤2 kg, preoperative mechanical ventilation, and inotropic support) to possibly minimize mortality associated with anemia. However, while this study has identified a strong independent association between preoperative anemia and postoperative neonatal mortality, no causative conclusions can be drawn.
Quiz Ref IDThe limitations of this study include those inherent to a large, multicenter clinical database analysis including mistakes or missing data, miscoding, and other inherent inaccuracies in data collection.32 However, the ACS NSQIP database is a rigorously designed and well-controlled database; therefore, errors should be minimal. This study included only neonates from hospitals in the NSQIP database (typically larger academic centers) and therefore our findings may not be generalizable to all hospitals caring for neonates in the United States such as smaller community hospitals. Because the NSQIP data collection excludes herniorraphy, myringotomy tube placement, and other low-risk procedures, it is a selective sample of disproportionately high-risk procedures. Consequently, there is likely undersampling of the lowest-risk patients and selective sample bias. Furthermore, site or center information is not included in the NSQIP database and therefore the potential effect of clustering could not be taken into account. However, given that NSQIP contributing hospitals have a certain degree of homogeneity, being mostly tertiary-care academic centers, we feel that the neonatal cases included in this database should be well distributed. Confounding by indication may also be a factor influencing these results because anemic patients may well be the sicker, higher-risk patients. However, we found no increased incidence of mortality when stratified for surgical type or complexity. Furthermore, we controlled for this using multivariate analysis and a robust ROC curve; a hematocrit level less than 40% held as an independent risk factor for increased mortality.
To our knowledge, this is the first study to define the incidence of preoperative anemia in neonates, the incidence of postoperative in-hospital mortality in neonates, and the association between neonatal anemia and postoperative mortality in US hospitals. The incidence of preoperative neonatal anemia was 32% (892 of 2764), and the incidence of postoperative in-hospital mortality in neonates was 3.4% (93 of 2764). Preoperative anemia (hematocrit level <40%) is independently associated with postoperative in-hospital mortality in neonates (OR, 2.6; 95% CI, 1.5-4.6). Timely diagnosis, prevention, and appropriate treatment of preoperative anemia in neonates might improve outcomes and survival.
Corresponding Author: Susan M. Goobie, MD, Department of Anesthesiology, Perioperative and Pain Medicine, Boston Children’s Hospital, 300 Longwood Ave, Boston, MA 02115 (firstname.lastname@example.org).
Accepted for Publication: April 10, 2016.
Published Online: July 18, 2016. doi:10.1001/jamapediatrics.2016.1032.
Author Contributions: Drs Goobie and Faraoni had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All auithors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: All authors.
Administrative, technical, or material support: Faraoni.
Study supervision: Goobie, DiNardo.
Conflict of Interest Disclosures: None reported.
Funding/Support: Financial and material support for the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript was solely supported by the Department of Anesthesiology, Peri-operative and Pain Medicine, Boston Children’s Hospital, Boston.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.