Association of Weekend Admission With Hospital Length of Stay, Time to Chemotherapy, and Risk for Respiratory Failure in Pediatric Patients With Newly Diagnosed Leukemia at Freestanding US Children’s Hospitals

Importance  In adult patients with leukemia, weekend admission is associated with increased inpatient mortality. It is unknown whether weekend diagnostic admissions in pediatric patients with leukemia demonstrate similar adverse outcomes.

Objective  To estimate adverse clinical outcomes associated with weekend admission in the first hospitalization of pediatric patients with newly diagnosed leukemia.

Design, Setting, and Participants  This retrospective cohort study from 1999 to 2011 featured index hospital admissions identified from the Pediatric Health Information System database. Participants were children with newly diagnosed acute lymphoid leukemia or acute myeloid leukemia.

Exposures  Weekend (Saturday and Sunday) or weekday index admission.

Main Outcomes and Measures  Inpatient mortality, length of inpatient stay, time to chemotherapy, and organ-system failure in index admission.

Results  A total of 10 720 patients with acute lymphoid leukemia and 1323 patients with acute myeloid leukemia were identified; 2009 patients (16.7%) were admitted on the weekend. While the total daily number of patients receiving intensive care unit–level care was constant regardless of the day of admission, these patients represented a larger percentage of total admissions on weekends. In adjusted analyses, patients admitted on the weekend did not have an increased rate of mortality during the first admission (odds ratio, 1.0; 95% CI, 0.8-1.6). Patients whose initial admission for leukemia occurred during a weekend had a significantly increased length of stay (1.4-day increase; 95% CI, 0.7-2.1), time to initiation of chemotherapy (0.36-day increase; 95% CI, 0.3-0.5), and risk for respiratory failure (odds ratio, 1.5; 95% CI, 1.2-1.7) after adjusting for demographics, severity of illness, and hospital-level factors.

Conclusions and Relevance  While pediatric patients with newly diagnosed leukemia admitted on weekends do not have higher mortality rates, they have a prolonged length of stay, increased time to chemotherapy, and higher risk for respiratory failure. Patients who are severely ill at presentation represent a higher proportion of weekend index admissions. Optimizing weekend resources by increasing staffing and access to diagnostic and therapeutic resources may help to reduce hospital length of stay across all weekend admissions and may also ensure the availability of comprehensive care for those weekend admissions with higher acuity.

Leukemia is the most common pediatric malignancy, representing approximately 30% of all pediatric cancer diagnoses.¹ At presentation, children with acute lymphoblastic leukemia (ALL) or acute myeloid leukemia (AML) can be severely ill with hyperleukocytosis, tumor lysis syndrome, and other complications leading to organ dysfunction.^2,3 Given the high acuity at diagnosis, the timing of presentation can influence early outcomes. A large retrospective cohort study demonstrated increased inpatient mortality in adults with leukemia whose index admission commenced on a weekend.⁴ A single-center retrospective study found that weekend admission in adult patients with AML was associated with delayed central access catheter placement.⁵ Such adult data suggest that weekend admission may adversely influence outcomes in patients with leukemia and that hospital-specific resources may mediate these effects.^6-9

To our knowledge, there is a paucity of publications investigating the association of weekend admission with mortality, time to care, and specific adverse clinical outcomes for pediatric patients with leukemia. One single-center study of 207 pediatric patients with ALL found that Friday admission was not associated with delayed chemotherapy but did not examine other weekend days.¹⁰ Additionally, chemotherapy delay was not associated with relapse or death. Other potentially important factors were not studied including patient acuity and staffing quality.¹¹ Understanding whether weekend admission increases the risk for adverse outcomes in pediatric patients with leukemia and potential modifying factors could support altering hospital staffing models and increasing availability of diagnostic and therapeutic procedures on weekends.

A cohort of pediatric patients with ALL or AML treated at 43 children’s hospitals contributing data to the Pediatric Health Information System (PHIS) database was assembled to investigate the consequences of weekend vs weekday admission.^12,13 Mortality rate, hospital length of stay (LOS), time to chemotherapy, and risk for organ failure were compared between patients with a weekend index admission and those with a weekday index admission. Additionally, the association of patient-level factors, such as severe illness at presentation and hospital-level factors, such as fellowship program and nursing quality as measured by Magnet status, with described outcomes was explored. The identification of clinically relevant differences between index weekend and weekday admissions for pediatric patients with leukemia could provide insight into efficient methods of resource allocation for leukemia care across all days of the week.

A retrospective cohort study of patients with newly diagnosed ALL or AML treated at freestanding pediatric hospitals that contribute data to the PHIS was performed. The PHIS is an administrative database containing inpatient billing data from 43 not-for-profit, tertiary children’s hospitals affiliated with the Child Health Corporation of America (Overland Park, Kansas). Contributing hospitals are located in 17 major metropolitan areas and account for 85% of admissions to US freestanding children’s hospitals. Data from the PHIS consist of inpatient information including patient demographics, dates of service, discharge disposition, and International Classification of Diseases, Ninth Revision (ICD-9) diagnosis and procedure codes (up to 41 per admission). Additionally, the database contains daily billing data for specific resources including medications, imaging, and other clinical services. The quality of PHIS data is overseen by the Children’s Hospital Association, Truven Health Analytics (data processing partner, Ann Arbor, Michigan), and participating hospitals. Data for each admission are deidentified at submission and subject to numerous audits for valid entries (eg, valid ICD-9 diagnosis codes) and reasonable patient information (eg, birth weight).

In accordance with the Common Rule (45 CFR 46.102[f]) and the policies of The Children’s Hospital of Philadelphia institutional review board, this study using a deidentified data set was not considered human patients research.

Patients with new-onset ALL or AML diagnosed between January 1, 1999, to December 31, 2011, were identified from previously defined and validated pediatric cohorts.^12,13 Data from the index ALL or AML admission for each patient in the cohort were considered for this analysis. Patients were followed up from admission to discharge or death, whichever occurred first.

The outcomes of interest included inpatient death at any time during the index leukemia admission, LOS, time to chemotherapy, and presence of organ failure. Time to chemotherapy was defined as days from the admit date of index admission to the date of first systemic chemotherapy. Date of first chemotherapy was determined by manual review of pharmacy billing data. First intrathecal chemotherapy was not included as systemic chemotherapy. Specific organ failures were defined by using a composite of ICD-9 diagnosis codes, ICD-9 procedure codes, and resource use billing codes as previously described.¹⁴ Sepsis was defined by a combination of previously described ICD-9 diagnosis codes.^14-16 Organ failure and sepsis were classified as present or absent within each admission.

Weekday admissions were defined as admissions on Monday, Tuesday, Wednesday, Thursday, or Friday. Weekend admissions were defined as admissions on Saturday or Sunday. This classification was chosen because the PHIS does not provide admission hour and because the Saturday-Sunday definition is better correlated with periods of reduced hospital staffing.¹⁷

Demographic data and insurance status were defined at the time of the index ALL or AML admission. Age was collected in years and categorized into the following groups: younger than 1 year, 1 year to younger than 5 years, 5 years to younger than 10 years, 10 years to younger than 15 years, and 15 years or older. Other demographic variables included sex and race/ethnicity, dichotomized into white or nonwhite (black, Asian/Pacific Islander, Native American, other, and unknown). Insurance status was categorized into private, public (Medicaid, Medicare, other government, and Title V), and other (self-pay and other). Severe illness status was defined as the need for intensive care unit (ICU)–level care within the first 2 days of admission and was dichotomized as present or absent. Intensive care unit–level care was defined as the presence of at least 1 ICD-9 procedure and/or clinical resource, as previously defined. Examples of ICU-level care include use of bilevel positive airway pressure, mechanical ventilation, dialysis, leukopheresis, or vasopressors.^14,18 A patient was considered treated at a hospital with an active fellowship program if the hospital had a pediatric hematology-oncology fellowship program in the year of admission, as reported by publically available information from the Accreditation Council for Graduate Medical Education.¹⁹ Magnet status is an American Nurses Credentialing Center recognition program that identifies centers with high-quality nursing care as well as high nurse staffing and skill level.²⁰ A patient was considered treated at a hospital with a positive Magnet status if the hospital was determined to have Magnet status for greater than half of the years in the study period, as reported by publically available information. Additional hospitalwide variables were studied including the percentage of public payers per all admissions per year per hospital and the total number of inpatient oncology admissions per year per hospital as defined by complex chronic condition codes.²¹ These were studied as continuous variables.

Descriptive statistics were used to summarize patient demographics, as well as patient- and hospital-level covariates. χ² Tests were used for bivariate comparison of the distribution of each demographic and hospital-associated variable among the patients with an index admission on a weekday vs weekend. The rates of organ failure, sepsis, and mortality in the first admission were described and compared using χ² tests for patients admitted on a weekday vs weekend. The LOS and time to chemotherapy are reported as mean (SD). The t test was used to compare these outcomes for the weekday vs weekend admissions. Multivariate logistic regressions were performed to evaluate the association of weekend admission with mortality, organ failure, and sepsis, while adjusting for demographics, severity of illness at presentation, and hospital-level factors. Similarly, multivariate linear regressions were used to assess the adjusted effect of weekend admission on LOS and time to chemotherapy.

Sensitivity analyses were performed to evaluate an alternative definition of weekend admission where the weekend included Friday, Saturday, and Sunday. Sensitivity analyses were also performed to analyze holiday weekends as a separate category, defining a weekend as a holiday weekend if the adjacent Friday or Monday was New Year’s Day, Memorial Day, July 4th, Labor Day, or Christmas.

Between January 1, 1999, and December 31, 2011, 10 720 patients with new-onset ALL and 1323 patients with new-onset AML were identified. Patients admitted on a weekend comprised 16.7% of the cohort (Figure, A). Table 1 compares baseline demographic information of weekday and weekend admissions. Patients with weekend admissions were similar to patients with weekday admissions with respect to disease and sex. Weekend admissions had significantly higher percentages of patients who were younger than 5 years of age, of nonwhite race/ethnicity, and were publicly insured; however, these differences are seen in the context of a large sample size and so may not be clinically significant. Although the absolute number of patients designated as severely ill was similar throughout the week, severely ill patients represented a higher percentage of total admissions on the weekend (Figure, B). Overall, patients admitted on the weekend were significantly more likely to require ICU-level care within the first 2 days of index admission (4.8% vs 3.1%, P < .001).

Table 2 presents unadjusted analyses of the association of weekend admission with mortality, LOS, and time to chemotherapy. Inpatient mortality during the first admission was similar for weekday (0.8%) and weekend (1.0%) admissions (Table 2). Patients admitted on weekends had significantly longer hospital stays (17.8 vs 15.8 days, P < .001) and longer mean times to chemotherapy (3.9 vs 3.5 days, P < .001). Finally, patients admitted on weekends had an associated increased risk for cardiovascular failure (6.4% vs 5.0%, P = .01), respiratory failure (8.5% vs 5.7%, P < .001), and renal failure (10.2% vs 8.3%, P = .01) in the index admission (eTable 1 in the Supplement).

Results of the adjusted analyses of the association of weekend admission with mortality, LOS, and time to chemotherapy are presented in Table 3. After adjustment for demographics (disease, sex, age, race/ethnicity, and insurance), severe illness at admission, and hospital-level factors (presence of a fellowship program, Magnet status, percentage of public payers per admissions per hospital per year, and number of oncology admissions per hospital per year), weekend admission was not associated with increased mortality. Weekend admission was associated with prolonged LOS by 1.4 days (95% CI, 0.7-2.1 days). In disease-specific adjusted analyses for LOS, weekend admission was associated with significantly prolonged LOS by 0.8 days (95% CI, 0.2-1.5) for patients with ALL and by 4.9 days (95% CI, 1.5-8.2) for patients with AML. Weekend admission was also associated with increased time to chemotherapy (0.4 days, 95% CI, 0.3-0.5 days) in adjusted analyses.

Multivariate analyses also identified AML, age younger than 1 year or 10 years and older, female sex, nonwhite race/ethnicity, public insurance, severe illness at presentation, and presence of a fellowship program as independently associated with prolonged LOS (eTable 2 in the Supplement). Acute myeloid leukemia, age younger than 5 years or 10 years and older, nonwhite race/ethnicity, and public insurance remained independently associated with a longer time to chemotherapy (eTable 2 in the Supplement). The presence of a fellowship program was associated with increased time to chemotherapy (0.3 days; 95% CI, 0.2-0.4), whereas hospital Magnet status was associated with a decreased time to chemotherapy (−0.3 days; 95% CI, −0.4 to −0.2).

Table 4 presents individual multivariate analyses for the association of weekend vs weekday admission with cardiovascular, respiratory, and renal failure adjusted for demographics, severe illness at presentation, and hospital-level factors. Weekend index admission was associated with an increased risk for respiratory failure (odds ratio, 1.5; 95% CI, 1.2-1.7) but not cardiovascular or renal failure. Acute myeloid leukemia disease, age younger than 1 year and older than 15 years, nonwhite race/ethnicity, and public insurance were associated with respiratory failure (eTable 3 in the Supplement).

To assess the sensitivity of the definition of weekend admission, the same multivariate analyses were performed defining Friday, Saturday, and Sunday admissions as weekend admissions. With this definition of weekend, significant differences in LOS persisted. Cardiovascular and respiratory failure neared statistical significance (eTable 4 in the Supplement). Analyzing holiday weekends separately demonstrated significantly increased LOS but no significant increase in cardiovascular or respiratory failure (eTable 5 in the Supplement).

These results suggest that weekend index admission in pediatric patients with leukemia is associated with increased LOS, time to chemotherapy, and risk for respiratory failure when controlling for patient acuity and hospital-specific characteristics. Unlike prior literature, this study investigated clinical outcomes in a national multicenter cohort of pediatric patients with ALL and AML, thus, these results may better describe national patterns of outcomes in pediatric patients with leukemia admitted on the weekend.

Although prior literature has described an association between weekend admission and increased mortality, we did not observe this association in our cohort of pediatric patients presenting for diagnostic leukemia admission. Increased mortality has been observed in weekend admissions in adults with cancer, myocardial infarction, gastrointestinal hemorrhage, or pulmonary embolism, as well as in neonates and ICU-hospitalized children.^4-8,22,23 As the only prior pediatric study focused on the association between weekend admission and care delay, to our knowledge, this study represents the first evidence that weekend admission is not associated with increased mortality in pediatric patients with leukemia.¹⁰

The association of weekend admission with increased LOS, risk for respiratory failure, and time to chemotherapy suggests consequences to the variation in care provision between weekends and weekdays. Increased hospital LOS leads to increased health care cost, potential exposure to unnecessary health care–associated pathogens, and has implications on quality of life.²⁴ The increased likelihood of respiratory failure despite adjustments for patient acuity suggests that respiratory issues may be less well managed in children presenting on the weekend. This is important because respiratory failure, often secondary to mediastinal mass, hyperleukocytosis, or infection, is a common complication in children with newly diagnosed leukemia.²⁵ The clinical relevance of the statistically significant increase in the time to chemotherapy in weekend index admissions is questionable because this delay was less than 1 day. Additionally, a previous study in pediatric patients with ALL did not associate chemotherapy delay with adverse outcomes.¹⁰ Delays in other supportive care interventions among weekend admissions could have a more profound negative impact on clinical outcomes.

A potential source for these observed variations in patient outcomes and time to chemotherapy is patient acuity.²⁶ Although similar absolute numbers of severely ill patients were admitted across each weekday in our cohort, high-acuity patients represent a higher proportion of overall weekend admissions. After adjusting for patient acuity at the time of presentation, variation in LOS and the risk for respiratory failure persisted between weekend and weekday admissions. Studies in neonates have also shown that weekend admissions have an increased rate of adverse outcomes despite adjustment for patient acuity level.²²

These data suggest that other hospital-specific characteristics contribute to the observed variations in patient outcomes by day of admission. Staff number and skill mix, physician cross coverage, and availability of diagnostic and therapeutic resources have each been implicated as a source for increased adverse outcomes in weekend admissions.¹¹ Previous adult studies found associations between hospital teaching status, physician staffing and skill level, and nurse staffing and skill level and patient outcomes.^27-32 These analyses identified the presence of pediatric hematology-oncology fellowship during year of admission as a hospital factor that was independently associated with increased hospital LOS (0.8 days; 95% CI, 0.1 to 1.5). Magnet status did not significantly impact LOS or the risk for respiratory failure. Payer mix and oncology patient volume did not influence studied outcomes to a clinically relevant degree. These variables are relatively nonspecific descriptors of a hospital’s care capacity and may not capture important nuances likely to influence clinical outcomes.

Taken collectively, a number of modifiable factors exist that should be further explored to improve the observed adverse outcomes in pediatric patients with leukemia with weekend index admission. For instance, it would be useful to better understand why an increased proportion of patients presenting on the weekend are acutely ill. Potential explanations for this finding include delayed referral from primary care physicians or delay in parents bringing their children to medical attention early owing to socioeconomic constraints. Additionally, hospitals can implement a systematic process to assess their weekend availability of pertinent staff and diagnosis services relative to weekday availability. This process could help hospitals rectify areas with insufficient resources that are necessary to appropriately address care needs of the higher-acuity patient.¹¹ Of course, increasing the availability of resources on the weekends in anticipation of higher-acuity patients may result in increased hospital expenditures without necessarily decreasing negative patient outcomes. Previous literature suggests mixed findings regarding the cost-effectiveness of increasing nurse staffing on the weekends to improvement in patient outcomes.³¹ Additional research is necessary to determine the most clinically and cost-effective combination of hospital resources to reduce the LOS of pediatric patients with leukemia and to improve their clinical outcomes.

These data need to be interpreted with the understanding of several limitations. First, because inclusion in the cohort required review of billed chemotherapy, children who died before receiving chemotherapy were not included. Such patients represent less than 1% of patients with ALL and less than 5% of patients with AML; however, omission of these patients may underestimate the impact of weekend admission on mortality.³³ Second, organ failure and the severity of illness were defined by ICD-9 diagnosis codes and resource use rather than by laboratory data or physician documentation. Previously validated ICD-9 codes were used to the extent possible.^14,18 Because the same method was used for determining organ failure and severe illness in all patients, this limitation should have little impact on comparisons between weekend and weekday admissions. Finally, because weekend was defined as Saturday or Sunday admission, the impact of Friday afternoon admission on patient outcomes was not captured. Given the sensitivity analyses suggested similar findings, the chosen weekend definition did not likely bias the final study conclusions.

Children with newly diagnosed leukemia with their index admission on the weekend had an increased LOS and risk for respiratory failure despite controlling for patient acuity and hospital-level factors. Given the increasing need for cost-effective care in medically complex children, these findings highlight a potential area for improvement in patient care and in cost reduction.³⁴ Hospitals should consider the increased acuity level of index admissions of pediatric patients with leukemia when determining allocation of weekend staff and clinical resources. Optimizing weekend resources may not only help to reduce hospital LOS across all weekend admissions but may also ensure the availability of comprehensive care for those weekend admissions with higher acuity.

Corresponding Author: Elizabeth K. Goodman, BA, Division of Oncology, The Children’s Hospital of Philadelphia, 4018 CTRB, 3501 Civic Center Blvd, Philadelphia, PA 19104 (egoodm@mail.med.upenn.edu).

Accepted for Publication: May 12, 2014.

Published Online: August 25, 2014. doi:10.1001/jamapediatrics.2014.1023.

Author Contributions: Ms Goodman and Dr Aplenc 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: Goodman, Reilly, Fisher, Fitzgerald, Seif, Bagatell, Aplenc.

Acquisition, analysis, or interpretation of data: Goodman, Fisher, Fitzgerald, Li, Seif, Huang, Bagatell, Aplenc.

Drafting of the manuscript: Goodman, Aplenc.

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

Statistical analysis: Fisher, Li, Huang.

Obtained funding: Aplenc.

Administrative, technical, or material support: Goodman, Fitzgerald, Bagatell, Aplenc.

Study supervision: Reilly, Fisher, Bagatell, Aplenc.

Conflict of Interest Disclosures: Dr Fisher receives research funding from Pfizer unrelated to this project. No other disclosures were reported.

Funding/Support: This study was supported by grant R01CA165277 from the National Institutes of Health and by grant sponsor Hematologic Research Malignancies Fund at the Children’s Hospital of Philadelphia.

Role of the Funder/Sponsor: The funders 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.