Efficacy of Subthreshold Newborn Phototherapy During the Birth Hospitalization in Preventing Readmission for Phototherapy

Key Points

Question  How effective is subthreshold neonatal phototherapy during the birth hospitalization in preventing readmissions for phototherapy?

Findings  In this cohort study of 25 895 newborns born at 35 or more weeks’ gestation from 2010 through 2014 with total serum bilirubin levels just below the threshold for phototherapy, receipt of phototherapy during the birth hospitalization decreased readmission for phototherapy, but often with large numbers needed to treat. Formula feeding was also protective against readmission for phototherapy.

Meaning  Although an effective treatment, subthreshold phototherapy during the birth hospitalization requires treatment of many newborns unnecessarily for each readmission prevented.

Importance  Treatment of jaundiced newborns with subthreshold phototherapy (phototherapy given to newborns with bilirubin levels below those recommended in American Academy of Pediatrics [AAP] guidelines) is common. However, the use of subthreshold phototherapy may have risks and increase costs, and, to date, it has not been systematically studied in newborns.

Objectives  To estimate the efficacy of subthreshold phototherapy for newborns with total serum bilirubin (TSB) levels from 0.1 to 3.0 mg/dL below the appropriate AAP phototherapy threshold during the birth hospitalization in preventing readmissions for phototherapy, and to identify predictors of readmission for phototherapy.

Design, Setting, and Participants  Retrospective cohort study of 25 895 newborns born at 35 or more weeks’ gestation, born in 1 of 16 Kaiser Permanente Northern California hospitals from January 1, 2010, through December 31, 2014, with at least 1 TSB level from 0.1 to 3.0 mg/dL below the appropriate AAP phototherapy threshold and not exceeding the threshold during the birth hospitalization. Data were analyzed from November 1, 2015, to November 28, 2017.

Exposure  Subthreshold phototherapy during the birth hospitalization.

Main Outcomes and Measures  Readmission for phototherapy.

Results  Among 25 895 newborns with qualifying TSB levels from 0.1 to 3.0 mg/dL below the appropriate AAP phototherapy threshold, 4956 (19.1%) received subthreshold phototherapy and 241 of these (4.9%) were readmitted for phototherapy compared with 2690 of 20 939 untreated newborns (12.8%) (unadjusted odds ratio [OR], 0.35; 95% CI, 0.30-0.40). In a logistic regression model, adjustment for confounding variables, including gestational age, race/ethnicity, formula feedings per day, and the difference between the TSB level and the phototherapy threshold, strengthened the association (OR, 0.28; 95% CI, 0.19-0.40). Estimated numbers needed to treat ranged from 60.8 in the lowest quintile of predicted risk to 6.3 in the highest quintile. Newborns who received formula feedings had lower adjusted odds of readmission for phototherapy compared with exclusively breastfed newborns (OR, 0.58; 95% CI, 0.47-0.72 for >0 to <2 formula feedings per day; OR, 0.24; 95% CI, 0.21-0.27 for ≥6 formula feedings per day). Subthreshold phototherapy was associated with a 22-hour longer length of stay (95% CI, 16-28 hours).

Conclusions and Relevance  Subthreshold phototherapy during the birth hospitalization is effective in preventing readmissions for phototherapy; however, for each readmission prevented, many newborns require phototherapy who would otherwise not need it.

Phototherapy is commonly used to treat neonatal hyperbilirubinemia.¹ The American Academy of Pediatrics (AAP) has established thresholds for when phototherapy should be used based on total serum bilirubin (TSB) level, gestational age, chronological age, and neurotoxicity risk factors.² Newborns with TSB levels below these thresholds are sometimes treated,³ not to prevent exchange transfusions or kernicterus but to prevent future readmissions for hyperbilirubinemia and their associated copays and chagrin. Because phototherapy is generally considered safe,^1,2 clinicians and parents may consider subthreshold phototherapy to have a favorable risk-benefit profile for newborns with TSB levels close to the threshold for treatment.

To our knowledge, no data presently exist regarding the efficacy of subthreshold phototherapy for newborns to prevent future readmissions for phototherapy. In this study, we sought to estimate the efficacy of using phototherapy to treat newborns with TSB levels from 0.1 to 3.0 mg/dL (to convert to micromoles per liter, multiply by 17.104) below the appropriate AAP phototherapy threshold during the birth hospitalization in order to prevent readmissions for phototherapy. In addition, we sought to identify positive and negative predictors of readmission for phototherapy to identify predictors of the number needed to treat (NNT) for this therapy.

This was a retrospective cohort study of newborns born at 35 or more weeks’ gestation in any of 16 Kaiser Permanente Northern California hospitals from January 1, 2010, to December 31, 2014. We excluded newborns with 7 or more days of birth hospitalization length of stay. We included newborns who had at least 1 TSB level from 0.1 to 3.0 mg/dL below the appropriate 2004 AAP phototherapy threshold² but whose level never exceeded the threshold during the birth hospitalization. As in previous studies,^3,4 we determined the AAP phototherapy threshold based on chronological age, gestational age, and direct antiglobulin test (DAT) result of the newborn. Although other factors (sepsis, acidosis, asphyxia, significant lethargy, temperature instability, hypoalbuminemia, and glucose-6-phosphate dehydrogenase deficiency) also increase the risk of neurotoxicity,² we did not include these factors when determining AAP thresholds because of rarity, lack of explicit definitions in the guidelines, and/or lack of availability in the electronic medical records.

The study was approved by the institutional review board at Kaiser Permanente Northern California and the Committee on Human Research at the University of California, San Francisco, which also determined that patient consent was not required because deidentified records were used.

The primary predictor variable was receipt of inpatient subthreshold phototherapy during the birth hospitalization. As previously described,⁵ we classified newborns to have received subthreshold phototherapy if they had either nursing documentation of phototherapy in the electronic medical record or both a procedure code and a physician order for phototherapy.

We obtained the following potential confounders directly from the electronic medical record: sex, gestational age to the final completed week, birth weight, race/ethnicity, delivery mode (spontaneous vaginal, cesarean, or assisted vaginal), DAT positivity, year of birth, and hospital facility of birth.

We obtained data about formula feeding from nursing documentation in the electronic medical record. We included a variable for the number of formula feedings per day during the birth hospitalization obtained by dividing the total number of formula feedings during the birth hospitalization by the length of stay in days. We created a variable to indicate a family history of phototherapy; the variable was coded yes if a child identified from a previous pregnancy of the newborn’s mother was in the birth cohort and had received phototherapy. We also created a variable for home phototherapy within 72 hours of discharge from the birth hospitalization that was based on the date of delivery of a phototherapy unit.

During the time of this study, all TSB levels were measured using the Vitros BuBc Neonatal Bilirubin method⁶ (Ortho Clinical Diagnostics). As previously reported,⁷ Ortho Clinical Diagnostics issued new calibrators for their bilirubin testing instruments in May 2012, which resulted in a significant decrease in measured TSB levels and the use of phototherapy. To account for this recalibration, we included a dichotomous variable indicating whether the newborn was born before or after June 1, 2012, in addition to the indicator variables for year of birth in the multivariable models.

This study was designed to evaluate readmissions after subthreshold phototherapy use during the birth hospitalization. We restricted our analysis to newborns with TSB levels just below, but not exceeding, the appropriate phototherapy threshold during the birth hospitalization. We defined the qualifying TSB level as the first TSB level from 0.1 to 3.0 mg/dL below the AAP phototherapy threshold and entered this level as a difference from the phototherapy threshold in categories of 1.0 mg/dL, as previously described (0.1 to 1.0, >1.0 to 2.0, and >2.0 to 3.0 mg/dL below the threshold).⁴ We used the first TSB level from 0.1 to 3.0 mg/dL below the phototherapy threshold (instead of the TSB level closest to the phototherapy threshold) as an inclusion criterion to mimic the entry criteria for a clinical trial (the TSB level closest to the phototherapy threshold would only be known after randomization). Only TSB measurements were used in this study; transcutaneous bilirubin measurements were not used.

We categorized the age of the newborn in 24-hour increments at the time that the qualifying TSB level was obtained. Weight loss was estimated using the birth weight and the most recent weight documented at the time that the qualifying TSB level was obtained. Two continuous variables were changed to dichotomous variables (birth weight at 2500 g and weight loss at 7%) for simplicity. Including cubic spline terms for these continuous variables did not improve the fit of the models or alter the coefficient for phototherapy.

The primary outcome was a readmission in the first 2 weeks after birth (<15 days) at which phototherapy was administered. We evaluated the following secondary outcomes occurring after discharge: (1) a TSB level that exceeded the appropriate AAP phototherapy threshold by any amount; (2) a TSB level that exceeded the appropriate AAP phototherapy threshold by 2.0 mg/dL or more; and (3) a TSB level that exceeded the appropriate AAP exchange transfusion threshold.² We also compared the length of stay by receipt of subthreshold phototherapy with and without adjustment for confounding variables.

From November 1, 2015, to November 28, 2017, we performed basic analyses using odds ratios (ORs) and 2-tailed t tests, with P < .05 indicating statistical significance. We used ORs instead of risk ratios to facilitate comparison between crude and adjusted results. Multivariable logistic regression was used to control for potential confounding variables while accounting for clustering by hospital facility using generalized estimating equations with robust SEs and an independent working correlation structure. We used multivariable-adjusted linear regression to estimate the effect of subthreshold phototherapy on length of stay.

Only a small number of newborns had TSB levels that exceeded the exchange transfusion threshold; therefore, multivariable analyses of that outcome required collapsing the variable of gestational age (<38 weeks vs ≥38 weeks) and the variable of age at qualifying TSB level (<24 hours, 24 to <48 hours, or ≥48 hours) and omitting the variable of weight loss less than 7% (not statistically significant in the primary adjusted model) to avoid dropping newborns with covariate patterns that perfectly predicted no outcomes from the model.

We calculated multivariable-adjusted marginal rates for each outcome. For these analyses, we set the indicator variable for post-recalibration to 1 because we believed the phototherapy rates observed after recalibration were more likely to apply going forward. We estimated NNT as reciprocals of marginal risk differences (1 divided by the estimated mean risk difference). These marginal risk differences estimate the (mean) effect if all newborns vs no newborns with a particular covariate level (eg, gestational age of 37 weeks) were treated with phototherapy during the birth hospitalization; in contrast, standard risk differences are dependent on the values of other covariates (eg, the NNT for 37-week newborns would also depend on factors such as sex, race/ethnicity, and formula feeding). We also divided the newborns into quintiles of predicted risk of readmission (based on variables that were associated with the odds of readmission in this study, excluding phototherapy) and estimated the marginal NNT by quintile. All analyses were performed using Stata, version 13 (StataCorp).

Of the 171 777 newborns born at 35 or more weeks’ gestation at Kaiser Permanente Northern California hospitals from 2010 to 2014, 29 488 had a qualifying TSB level. After exclusions for a length of stay of 7 days or longer (n = 940), having a TSB measurement that exceeded the AAP phototherapy threshold during the birth hospitalization (n = 2528), and missing data (sex [n = 1], birth weight [n = 2], delivery mode [n = 8], maternal identification [n = 8], and formula feeding data [n = 106]), there remained 25 895 newborns.

In the remaining cohort of 25 895 newborns with TSB levels from 0.1 to 3.0 mg/dL below the appropriate AAP phototherapy threshold during the birth hospitalization, 4956 (19.1%) received subthreshold phototherapy during the birth hospitalization and 2931 (11.3%) were readmitted for phototherapy. Those who received subthreshold phototherapy had a mean length of stay that was 25.4 hours longer than those who did not (mean [SD] length of stay, 75.8 [26.2] hours vs 50.4 [20.2] hours; P < .001). After adjustment for sex, race/ethnicity, gestational age, year of birth, low birth weight, delivery mode, DAT positivity, family history of phototherapy, qualifying TSB level, age at qualifying TSB level, weight loss at the time of the qualifying TSB level, and TSB level recalibration, subthreshold phototherapy was associated with a 22-hour increase in length of stay (95% CI, 16-28 hours).

A total of 241 of 4956 newborns (4.9%) treated with subthreshold phototherapy during the birth hospitalization were readmitted for phototherapy compared with 2690 of 20 939 untreated newborns (12.8%). Receipt of subthreshold phototherapy during the birth hospitalization was associated with significantly decreased odds of readmission for phototherapy (crude OR, 0.35; 95% CI, 0.30-0.40) (Table 1). In an adjusted analysis, receipt of subthreshold phototherapy during the birth hospitalization was even more strongly associated with decreased odds of readmission rates for phototherapy (OR, 0.28; 95% CI, 0.19-0.40) (Table 2). The multivariable-adjusted marginal decrease in readmission rates for phototherapy after receipt of subthreshold phototherapy during the birth hospitalization was 7.1% (95% CI, 5.3%-8.9%), with a marginal NNT of 14.1 (Table 3).

Several variables were associated with increased adjusted odds of readmission for phototherapy: male sex, lower gestational age, Asian race/ethnicity, assisted vaginal delivery, family history of phototherapy, a qualifying TSB level closer to the phototherapy threshold, being of lower chronological age at the time of the qualifying TSB level, and exclusive breastfeeding (Table 2). Conversely, low birth weight, cesarean delivery, DAT positivity, and home phototherapy were associated with decreased odds of readmission for phototherapy. The association between formula feeding and readmission for phototherapy was dose dependent, with increased use of formula being associated with decreased odds of readmission for phototherapy (Tables 2 and 3). After combining predictors, the mean NNT of newborns in the quintile at highest risk of readmission was only 6.3 (95% CI, 5.1-8.3) compared with 60.8 (95% CI, 44.5-95.8) for newborns in the quintile at lowest risk (Table 3). The risk of readmission was more than an order of magnitude greater in the highest risk quintile than the lowest risk quintile (27.0% vs 2.3%). Despite this, subthreshold phototherapy use was not much higher in the highest risk quintile than the lowest risk quintile (21.4% vs 18.2%).

As demonstrated in Table 4, of the 25 895 newborns in the cohort, 4166 (16.1%) developed a TSB level at or above the phototherapy threshold after discharge. Of these, 2595 (62.3%) were readmitted for phototherapy. Subthreshold phototherapy during the birth hospitalization decreased the odds of having a TSB level at or above the phototherapy threshold after discharge (adjusted OR, 0.16; 95% CI, 0.12-0.21). There were 859 newborns (3.3% of the cohort) who developed a TSB level that was 2.0 mg/dL or more above the phototherapy threshold after discharge, of whom 730 (85.0%) were readmitted for phototherapy. Subthreshold phototherapy again decreased the odds of having a TSB level that exceeded the phototherapy threshold after discharge by 2.0 mg/dL or more (adjusted OR, 0.24; 95% CI, 0.16-0.34). Only 40 newborns (0.2% of the cohort) developed a TSB level that was at or above AAP exchange transfusion levels; this led to wide confidence intervals around the estimates for efficacy to prevent this outcome (adjusted OR, 0.58; 95% CI, 0.20-1.66; P = .31). Notably, of the 2931 newborns readmitted for phototherapy, 336 (11.5%) never had a TSB level that met or exceeded the phototherapy threshold.

We performed an analysis to investigate the unexpected protective effect of a positive DAT using the cohort of 171 777 newborns before exclusions. In this analysis, 23.0% of DAT-positive newborns had TSB levels that exceeded the phototherapy threshold during their birth hospitalization compared with 2.7% of DAT-negative or untested newborns.

Our study found that newborns who received subthreshold phototherapy during the birth hospitalization had 72% lower adjusted odds of readmission for phototherapy with a mean NNT of 14.1, although the NNT values varied widely depending on other covariates.

At first glance, the direction of some of the crude associations in our study may be surprising. Table 1 shows high-risk newborns (eg, low gestational age, >7% weight loss, low birth weight, DAT positive, and other variables) to be at decreased odds of being readmitted for phototherapy. However, as seen in Table 2, after adjustment for potentially confounding variables, low gestational age became associated with the expected increase in the risk of readmission for phototherapy, and the association between a more than 7% weight loss and readmission for phototherapy was no longer statistically significant.

Higher-risk infants may be at decreased risk of readmission for phototherapy because they are more likely to have longer lengths of stay. Longer lengths of stay reduce the likelihood of readmission for phototherapy because TSB levels are more likely to peak during the birth hospitalization, resulting in newborns being more likely to receive phototherapy during that time. High-risk newborns may also be more likely to receive additional interventions during the birth hospitalization (eg, formula supplementation, subthreshold phototherapy, and other interventions) that lead to decreased readmissions.

Positive DAT results appeared to be protective against readmission for phototherapy in this study because of the exclusion of newborns with TSB levels that exceeded the phototherapy threshold during the birth hospitalization. By excluding these newborns, the lower-risk group of newborns with DAT-positive results remained in the cohort. A prior study from New Zealand showed that only 23% of newborns with positive DAT results received phototherapy,⁸ which is consistent with our data.

Phototherapy is generally considered a low-risk intervention.² Still, it can cause physical separation of the mother and the newborn, potentially interfere with breastfeeding and bonding, increase inpatient hospitalization costs, and increase the hospital length of stay. Data are emerging that phototherapy may be associated with a small risk of adverse effects.^5,9-11 In developed countries, the risk of bilirubin-related encephalopathy is low, even at levels well above present exchange transfusion thresholds.^12-18 It is, therefore, important to consider both the risks and benefits of phototherapy, especially when used in newborns with TSB levels below the recommended treatment thresholds.

The NNT values for subthreshold phototherapy in preventing readmission for phototherapy varied greatly based on the covariates. Even for newborns in the highest quintile of risk, the NNT to prevent readmission was 6.3. The most appropriate NNT when treating infants with hyperbilirubinemia will vary based on individual risk factors. Although many families may find prolonged hospitalization to be less disruptive than readmission, it is not judicious to treat large numbers of newborns with an unnecessary intervention in order to prevent 1 readmission. For the high-risk infants for whom subthreshold phototherapy is most reasonable, such as those born at lower gestational ages, those with TSB levels close to the phototherapy threshold, or those with uncertain or difficult follow-up, it is reasonable to include the families in the decision-making process.

Breastfed newborns have higher TSB levels^19,20 and are at higher risk of developing TSB levels of 25.0 mg/dL or more^21,22 than their formula-fed counterparts. In our study, formula feeding was associated with an expected decrease in odds of readmission for phototherapy, with newborns who received 6 or more formula feedings per day having 76% lower odds of readmission for phototherapy than those who were exclusively breastfed. Receiving 4 to 6 formula feedings per day had a similar benefit in preventing readmission to that of receiving subthreshold phototherapy during the birth hospitalization.

Breastfeeding confers numerous health benefits to both mothers and infants,²³ and the AAP recommends that infants exclusively breastfeed for 6 months and continue breastfeeding for at least 1 year.²³ Formula supplementation may decrease the duration of breastfeeding.^24-26 One small study showed that limited amounts of formula supplementation after breastfeeding for a limited time may not have an adverse effect on breastfeeding.²⁷ The extent to which this limited formula supplementation might reduce TSB levels is an important topic for further investigation.

In the US health care system, hospital readmissions soon after discharge are often considered a negative indicator of quality.²⁸ However, because 80% of newborns develop jaundice²⁹ and TSB levels often peak after newborns are discharged, some readmissions for hyperbilirubinemia should be expected and should not be considered a negative quality indicator. Efforts to avoid these readmissions could lead to overtreatment of hyperbilirubinemia during the birth hospitalization and to more newborns receiving phototherapy overall.

This study has several limitations. The cohort was a predominantly insured population in Northern California, which may make the results less generalizable to other populations. The study design was observational, not experimental. In addition, we had no data for formula use after discharge and only limited data for formula use during the birth hospitalization because data for formula type and quantity were not available.

While subthreshold phototherapy during the birth hospitalization prevents readmission for phototherapy, it results in the unnecessary treatment of many newborns.

Accepted for Publication: December 7, 2017.

Corresponding Author: Andrea C. Wickremasinghe, MD, Department of Pediatrics, Kaiser Permanente Santa Clara Medical Center, 700 Lawrence Expressway, Department 302, Santa Clara, CA 95051 (andrea.wickremasinghe@kp.org).

Published Online: February 26, 2018. doi:10.1001/jamapediatrics.2017.5630

Author Contributions: Drs Wickremasinghe and Newman had full access to all 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: Wickremasinghe, Kuzniewicz, Newman.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Wickremasinghe.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Wickremasinghe, McCulloch, Newman.

Obtained funding: Newman.

Administrative, technical, or material support: Kuzniewicz.

Study supervision: Newman.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by grant R01HS020618 from the Agency for Healthcare Research and Quality.

Role of the Funder/Sponsor: The funding organization played no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.