Decreasing Laboratory Testing for Neonatal Jaundice Through Revision of a Clinical Practice Pathway Free

Approximately 100 to 150 patients are admitted annually with neonatal hyperbilirubinemia to our free-standing children’s hospital. Our PHM service had an existing CPP for the treatment of neonatal hyperbilirubinemia that was created in 2011 on the basis of the 2004 American Academy of Pediatrics guideline.⁷  In 2019, a group of 3 PHM physicians, in collaboration with local neonatologists, revised the existing CPP for neonatal hyperbilirubinemia to incorporate new evidence and standardize laboratory ordering practices (Supplemental Fig 3).

Edits to the CPP included guidance on limiting diagnostic laboratory tests to only patients with rapidly rising serum bilirubin levels, risk factors for hemolysis, or jaundice unexplained by history or physical examination^6–8 ; grouping these diagnostic laboratory tests with the next bilirubin level; increasing the interval of time between bilirubin checks for patients >2 mg/dL below their exchange transfusion level^9–11 ; and increasing the maximum bilirubin level to consider discharge from 4 to 6 mg/dL below phototherapy treatment threshold to ≥3 mg/dL below phototherapy treatment threshold.^12–14  These changes were made after review of recent literature and publicly available institutional guidelines from other academic pediatric hospitals and after discussion with local experts. The revised CPP was published internally on the hospital’s intranet on April 1, 2019 (Supplemental Fig 3). The pathway changes were reviewed with all PHM providers and the pediatric residents before publication. Finally, modifications were made to an electronic health record (EHR) order set to align with the CPP, and an electronic link was added for ease of reference.

We performed a retrospective cohort study to compare the number of laboratory tests and blood draws in patients with neonatal hyperbilirubinemia before and after implementation of a revised CPP. The study population included infants <14 days old admitted to the PHM service after their birth hospitalization with a primary diagnosis of neonatal hyperbilirubinemia between April 2017 and March 31, 2019 (prepathway implementation group), and April 1, 2019, and October 2019 (postpathway implementation group). Patients who did not receive phototherapy during their admission, were born at <35 weeks gestational age, were admitted to the NICU, were treated for sepsis, or had a previous admission (after birth hospitalization) for phototherapy were excluded. Patients were identified using International Classification of Diseases, Tenth Revision, codes. A manual chart review was performed to extract variable data, which were recorded in a REDCap database. This study was determined to not be human subjects research and was exempt by our institutional review board.

Patient demographic information extracted from the EHR included sex, age at admission, gestational age at birth, and race. Race is caregiver reported and entered in the EHR at the time of registration. Race is reported as Asian; Black; White, or other, which includes American Indian, Alaskan native, or no race or “other” race entered in the EHR.

Primary outcome measures included the total number of blood draws per patient, the total number of laboratory tests obtained per patient, and LOS. The number of blood draws was determined by the time of laboratory acquisition, as documented in the EHR. Each unique time of acquisition was considered a separate blood draw. Inpatient LOS was calculated as the difference between electronic time stamps for time of arrival to the acute care floor and time of discharge from the hospital.

Secondary outcome measures included 7-day readmission rate, charges, and total bilirubin level at discharge. Readmission was defined as any admission within 7 days of discharge. Hospital charges per hospitalization for neonatal jaundice requiring phototherapy were obtained through an institutional health system portal. Discharge total bilirubin level was defined as the last recorded total serum bilirubin level before discharge.

Data were summarized as number (%), median and interquartile range (IQR), or mean and SD. Categorical characteristics were compared between prepathway implementation group and postpathway implementation group using χ² test or Fisher’s exact test, whereas continuous variables were compared using t test or Mann-Whitney-Wilcoxon test. Poisson regression or quasi-Poisson model were used to investigate associations between number of laboratory tests or number of blood draws and different groups. P < .05 was considered statistically significant. SAS version 9.4 (SAS Institute, Cary, NC) was used for all analyses.

During the study period, 375 patients with neonatal hyperbilirubinemia were admitted to the PHM service. Of these patients, 68 were excluded. Therefore, 307 patients were included in the analysis (225 patients in the prepathway implementation group, 82 patients in the postpathway implementation group). There were no differences in patient characteristics between the 2 groups (Table 1).

The median number of blood draws per patient decreased after pathway implementation to 2 (IQR, 2–3) compared with 3 (IQR, 2–3) before implementation (Poisson model–based estimated mean difference, 1.1; 95% confidence interval, 1.0–1.3; P = .018). The distribution of the proportion of patients who received each discrete number of blood draws before and after pathway implementation is shown in Fig 1.

The median number of laboratory tests per patient decreased after pathway implementation to 3 (IQR, 2–4) compared with 4 (IQR, 3–6) before implementation (Poisson model–based estimated mean difference, 1.3; 95% confidence interval, 1.2–1.5; P < .0001). The distribution of the proportion of patients who received each discrete number of laboratory tests before and after pathway implementation is shown in Fig 2.

There was no difference in LOS between the 2 groups. The median LOS before pathway implementation was 19.88 hours (range, 17.60–22.08 hours) compared with 19.25 hours (range, 16.88–21.27 hours) after pathway implementation (P = .11).

There were 9 readmissions within 7 days for hyperbilirubinemia in our study population. Six (2.67%) occurred in the prepathway implementation group, and 3 (3.70%) occurred in the postpathway implementation group (P = .70).

There was no change in hospital charges per hospitalization for phototherapy after pathway implementation. Median charge per hospitalization for phototherapy before pathway implementation was $4818 (IQR, $4570–$5172) compared with $4790 (range, $4624–$5033) after pathway implementation (P = .98). There was no difference in discharge bilirubin level before pathway implementation (mean, 11.53; SD, 1.88) compared with after pathway implementation (mean, 11.57; SD, 1.60; P = .84).

Implementation of a revised CPP was associated with a decrease in the median blood draws and laboratory tests per patient admitted with neonatal hyperbilirubinemia. However, although our results are statistically significant, we do not know whether a decrease of 1 blood draw is clinically significant to an infant or caregiver, especially considering that the number of blood draws reported in this study is likely an underestimation of total phlebotomy attempts, as it often takes multiple attempts to obtain neonatal blood samples.

No change in LOS or discharge bilirubin level for hyperbilirubinemia admissions was found after implementation of the CPP. Although the revised CPP increased the recommended total serum bilirubin threshold for discharge, there was no change in the actual discharge bilirubin levels after the pathway was implemented. We suspect that providers either continued to use their previous bilirubin threshold for discharge or the routine practice of obtaining laboratory tests in the morning and discharging patients after rounds resulted in similar LOS and discharge bilirubin levels between groups.

We also did not see a decrease in hospital charges per hyperbilirubinemia admission as previous studies have revealed.⁶  This finding is not surprising because laboratory tests constitute only a small proportion of overall hospital charges and the implementation of our pathway did not decrease LOS. We were unable to determine actual laboratory charges per patient because of limited access to these data.

This study was a single-center design, which limits overall generalizability. Additionally, although we intended to continue data collection into the winter months, patients with neonatal hyperbilirubinemia were diverted to the NICU because of a high census on acute care floors, so we do not know whether the effect has been sustained. Diversion of patients to the NICU also contributed to small sample size, particularly in the postpathway implementation group, which makes the findings vulnerable to type II error. Future directions will include continued monitoring for sustained effect.

In conclusion, a CPP can be effective in minimizing unnecessary laboratory services. By implementing a CPP that included recommendations on laboratory testing frequency and diagnostic screening, we were able to decrease the number of blood draws and laboratory tests without affecting LOS or readmissions.

The authors thank Fatima Anibaba, MS, for assistance with data collection and Kelsey Porada, MA, for assistance with preparing the manuscript.