Introduction

ABO hemolytic disease is a relatively common cause of early hyperbilirubinemia (before the infant leaves the nursery), but it is a relatively rare cause of hyperbilirubinemia in infants who have been discharged and readmitted.1 About 45% of newborns are born to mothers with blood type O.2 Of the infants born to mothers with blood type O, about 15 to 20% of these infants will have blood type A or B, creating a set up for ABO incompatibility that produces a spectrum of hemolytic disease, commonly a mild disease requiring no therapy.3 About 33% of these infants will have a positive direct antiglobulin test (Coombs test), indicating they have anti-A or anti-B antibodies attached to their red cells.1 However, the overall frequency of significant ABO hemolytic disease of the newborn is only about 0.02 to 1.4%.4 Clinical practice guidelines recommend that if the maternal blood is group O, Rh-positive, it is an option to test the cord blood for the infant's blood type and direct antibody test, but it is not required provided there is appropriate surveillance, risk assessment before discharge, and follow-up.5 Despite such published guidelines, some institutions routinely do cord-blood typing and Coombs testing on all infants born to blood type O women.6 Previous studies have shown that selective newborn cord-blood testing can decrease the use of resources and costs without additional patient morbidity from hemolytic disease.7 The Blood Bank at Loyola University Medical Center (LUMC) recently changed their policy regarding routine cord-blood typing and Coombs’ testing in newborns. The Blood Bank no longer does automatic cord-blood typing and Coombs’ testing on newborns born to O+ mothers. Rather, it performs these tests for only those patients in which testing is ordered by the physician caring for the infant. The objectives of our study included:

(1) to determine whether infants born to O+ mothers who had selective cord-blood testing after this policy change would have higher rates of clinically significant hyperbilirubinemia compared with those newborns that had routine cord-blood testing before the policy change; (2) to determine the amount of cost savings to our hospital by implementing a policy of selective cord-blood testing in newborns born to O+ mothers.

Methods

A retrospective chart review was performed on all infants in the newborn nursery born at Loyola to blood type O+ women between 1 April 2008 and 1 April 2009. LUMC Blood Bank changed the policy regarding routine cod blood typing and Coombs testing in 1 April 2008, but it did not get implemented until 1 October 2008. The patients in our study were divided into two cohorts—one group of patients included infants born within 6 months of implementation of the new policy (routine cord-blood testing from 1 April 2008 until 30 September 2008) and the other group included infants born up to 6 months after the policy change (selective cord-blood testing from 1 October 2008 until 31 March 2009). Data were collected for each of these two groups regarding clinically significant hyperbilirubinemia, including rate of cord-blood testing, average 24 h serum bilirubin, number of patients with a serum bilirubin above 14 mg dl−1 during birth hospitalization, rate of phototherapy during birth hospitalization, number of patients readmitted in the first week of life for hyperbilirubinemia and average peak bilirubin at readmission. Of note, our hospital performs antibody testing on all mothers as part of the prenatal labs and it also performs a 24-h serum bilirubin level on all infants.

The study was approved by the Loyola University Medical Center Institutional Review Board. Statistical analysis was performed using Stata 10.1 (Stata Corp, College Station, TX, USA). Fisher's exact test and a two-tailed t-test were used to analyze the data and calculate P-values.

Results

A total of 414 newborns born to blood type O+ women were included in the chart review. Of these, 250 infants were in the routine testing cohort and automatically received cord-blood typing and Coombs testing. Also, 164 infants were included in the selective testing cohort and had cord-blood typing and Coombs testing only if ordered by the managing physician. Table 1 compares the demographic data between the two groups and shows that both groups were similar, including potential risk factors for developing significant hyperbilirubinemia. A difference in the number of overall patients in the two groups was noted. This was most likely due to a change in the structure of the Obstetrics Department around the time of this study that resulted in fewer deliveries.

Table 1 Demographic data for study groups before and after policy change
Full size table

When comparing routine vs selective testing, there was no statistically significant difference in the 24-h serum bilirubin, proportion of patients with a bilirubin level greater than 14 mg dl−1 during the birth hospitalization, rate of phototherapy during the birth hospitalization, rate of readmission for hyperbilirubinemia or peak serum bilirubin level at readmission (Table 2).

Table 2 Rate of clinically significant hyperbilirubinemia before and after policy change
Full size table

One hundred percent of the newborns in the routine testing group had cord-blood testing done, while in the 6 months after the new policy was implemented and selective testing was done, only 25.6% of infants had cord-blood testing. Almost 75% reduction of cord-blood typing and Coombs testing was based on the overall testing rate in the 6 months following the policy change. If this 6-month period is divided into two 3-month blocks, we obtain a cord-testing rate of 43.2% in the first 3 months and only 8.4% in the last 3 months. Of the 42 infants in the selective testing group that had cord-blood testing, 27 (65%) had a 24-h serum bilirubin in the high–intermediate or high-risk zone based on the nomogram developed by Bhutani et al.8 This finding may have prompted the managing physician to order the cord-blood testing on these patients, as our lab stores the cord blood for 7 days.

Assuming the rate of testing remains about 8.4%, this 91.6% reduction in testing would mean tremendous savings to the hospital and to the patients. At $11 per cord-blood testing for hospital cost, $50 per test for patient charges, and 15 min per test for technician time, annually a total of $4100 would be saved to our hospital, $18 900 would be saved to our patients, and 95 h of technician time would be saved. Extrapolating these findings to all births in Illinois would mean even further savings. In 2008 there were about 176 600 births in Illinois of which about 45% of them would be born to O+ mothers. If selective cord-blood testing at a rate of 8.4% was done on these 79 500 newborns, that would mean an annual savings of $3.6 million to the patients and $800 000 to Illinois hospitals, assuming that all of these hospitals are testing all infants.

Discussion

In 2004, the American Academy of Pediatrics Subcommittee on Hyperbilirubinemia issued recommendations on management of hyperbilirubinemia. In these guidelines, they recommended if the maternal blood is group O, Rh-positive, it is an option to test cord blood for the infant's blood type and direct antibody test, but it is not required provided that there is appropriate surveillance, risk assessment before discharge, and follow-up.5 Despite these recommendations our own hospital was routinely testing cord blood for blood type and Coombs on all babies born to O+ mothers. The policy at LUMC regarding routine cord-blood testing in newborns born to O+ mothers was changed and ultimately selective cord testing was performed beginning in October 2008. It was interesting to note that our institution had been doing this routine cord-blood testing despite the fact we were doing a thorough risk assessment before discharge. Since 2007 we have been checking the serum bilirubin level at 24 h of age in all newborns. We then assess the risk of subsequent clinically significant hyperbilirubinemia using the percentile-based nomogram developed by Bhutani et al.8

Leistikow et al.7 had shown over 15 years ago that selective newborn cord-blood testing decreases the use of resources and costs without apparent additional patient morbidity from hemolytic disease of the newborn. Another study a few years earlier had also found selective cord-blood testing to be safe and cost saving.9 Both of these studies were done before the publication of the 2004 AAP guidelines.

The findings in our study were similar to the above two studies and reassert that routine cord-blood testing is not necessary for infants born to O+ mothers. The cohort of patients in our study that underwent selective testing after our new policy was implemented had a similar 24 h serum bilirubin level compared with those patients that had routine testing before the policy change. No difference was seen in the proportion of patients with a bilirubin level greater than 14 mg dl−1 during the birth hospitalization. In addition, the number of serum bilirubin lab tests done between each group was similar (18.0% vs 17.7%), as well as the length of stay. The selective group did not have more bilirubin tests ordered or have a delay in discharge in order to be monitored clinically for another day, as both of these actions would have cut into any economic savings. There was no statistically significant difference between these two groups when analyzing the rate of phototherapy during the birth hospitalization, rate of readmission for hyperbilirubinemia, or peak serum bilirubin level at readmission. We might expect a higher proportion of readmissions after the policy change if we were missing the diagnosis of severe ABO incompatibility in these patients at the time of initial discharge. Any patients readmitted with a missed diagnosis of hemolytic disease of the newborn would be expected to have a higher peak bilirubin at the time of readmission. We did not observe any of these findings. Instead we found similar results between our two study cohorts.

These findings suggest that performing cord-blood typing and Coombs testing on select neonates born to O+ mothers is a safe and reasonable practice as it does not appear to place infants at increased risk for unrecognized hemolytic disease of the newborn. Performing these tests on selective patients rather than doing them routinely does not appear to place these infants at increased risk of morbidity from ABO incompatibility. This practice of selective cord-blood testing appears to be even more practical in the setting of routine bilirubin screening before discharge from the newborn nursery, as performed in our institution with a 24-h serum bilirubin level. Eggert et al.10 showed that prehospital discharge bilirubin screening, coupled with the evaluation of serum bilirubin concentrations using a percentile-based nomogram, can identify neonates who are at risk for severe hyperbilirubinemia. Mah et al.11 further demonstrated that universal predischarge neonatal bilirubin screening significantly reduces the subsequent development of bilirubin levels that are known to place newborns at risk for bilirubin encephalopathy.

Implementation of the new policy recommending selective cord-blood testing took several months. To achieve this nearly 92% reduction in testing, several strategies were used to educate the physicians and nursing staff regarding the policy change. The medical director of the newborn nursery sent an email in October 2008 informing the attending physicians that provide coverage in the nursery of this policy change. The new policy was discussed at a faculty meeting later that month. The medical director also sent an email in October 2008 to all of the pediatric residents educating them about the change and the importance of adhering to the new policy. The nursing staff was informed at their October monthly staff meeting. The email explaining the new policy was printed and several copies were hung up at strategic locations, including the resident work area in the nursery and the main nursing station in the nursery. Monthly reminders for the first three months were then emailed by the medical director to the attending physicians, pediatric residents, and the nurse manager for the nursery. All of these measures allowed for successful implementation of the new policy and triggered the substantial reduction in testing over a 6-month period.

There were a few limitations in our study. The main one was potential loss of follow-up. It is possible that some of our patients who were discharged from our nursery were admitted to another hospital for hyperbilirubinemia, but we were unable to determine that from a retrospective chart review from our own institution. We would need to call the patients’ parents and ask whether they were admitted to another hospital. Small sample size may be another limitation of our study. The statistically insignificant differences in readmission rate (1.2% vs 2.4%) may be a consequence of an underpowered study. Another limitation was that we reviewed charts only out to 6 months following implementation of the policy change. It is possible the rate of testing may have started to increase as we move further away from the date of the policy change. We would need to do another chart review to determine whether the 8.4% testing rate continued beyond March 2009, the date of the last chart review. The final limitation was related to cost analysis that was based on the cost of this testing at LUMC. Other institutions likely have different costs for these tests that would affect the overall cost savings to the hospital and to the patient.

Conclusion

Routine cord-blood typing and Coombs testing of infants born to O+ mothers is not necessary, especially at institutions that routinely perform pre-hospital-discharge bilirubin screening. Furthermore, implementing a policy of selective newborn cord testing for these infants can decrease the use of resources and costs without increasing the risk of clinically significant hyperbilirubinemia from hemolytic disease of the newborn.