Weekend Effect in Children With Stroke in the Nationwide Inpatient Sample
Background and Purpose—Studies have demonstrated differences in clinical outcomes in adult patients with stroke admitted on weekdays versus weekends. The study’s objective was to determine whether a weekend impacts clinical outcomes in children with ischemic stroke and hemorrhagic stroke.
Methods—Children aged 1 to 18 years admitted to US hospitals from 2002 to 2011 with a primary discharge diagnosis of ischemic stroke or hemorrhagic stroke were identified by International Classification of Disease, 9th Revision, codes. Logistic regression estimated odds ratios and 95% confidence intervals for in-hospital mortality and discharge to a nursing facility among children admitted on weekends (Saturday and Sunday) versus weekdays (Monday to Friday), adjusting for potential confounders.
Results—Of 8467 children with ischemic stroke, 28% were admitted on a weekend. Although children admitted on weekends did not have a higher in-hospital mortality rate than those admitted on weekdays (4.1% versus 3.3%; P=0.4), children admitted on weekends had a higher rate of discharge to a nursing facility (25.5% versus 18.6%; P=0.003). After adjusting for age, sex, and confounders, the odds of discharge to a nursing facility remained increased among children admitted on weekends (odds ratio, 1.5; 95% confidence interval, 1.1–1.9; P=0.006). Of 10 919 children with hemorrhagic stroke, 25.3% were admitted on a weekend. Children admitted on weekends had a higher rate of in-hospital mortality (12% versus 8%; P=0.006). After adjusting for age, sex, and confounders, the odds of in-hospital mortality remained higher among children admitted on weekends (odds ratio, 1.4; 95% confidence interval, 1.1–1.9; P=0.04).
Conclusions—There seems to be a weekend effect for children with ischemic and hemorrhagic strokes. Quality improvement initiatives should examine this phenomenon prospectively.
- hospital mortality
- International Classification of Diseases
- intracranial hemorrhages
Childhood stroke is among the top 10 causes of death among children in the United States.1,2 Recent literature indicates that >2000 children with the diagnosis of stroke were discharged from US hospitals in each of the years 2000, 2003, and 2006.3 In North America and Europe, childhood stroke incidence ranges from 2 to 13 per 100 000 children per year in those aged <18 years.1,4–6 Studies provide consistent estimates that 39% to 54% of all childhood strokes are hemorrhagic.5–7 A study from a northern California health system demonstrated hemorrhagic stroke (HS) had an incidence of 1.4 per 100 000 person-years in children aged >28 days.8
In one of the largest studies that analyzed the weekend effect among the top 100 causes of in-hospital mortality, weekend admissions were associated with a 4% increase in mortality for any condition. However, this study did not analyze the weekend effect on admissions because of stroke.9 In adults, the incidence of stroke increases during weekends and other stressful periods.10–12 Previous studies13,14 have demonstrated differences in the quality of care and clinical outcomes in adult patients with stroke admitted on weekdays versus weekends. However, there are no data available on differences in clinical outcomes among pediatric stroke patients admitted on weekends versus weekdays. Knowing whether differences exist in outcomes based on admission period (weekday versus weekend) and then understanding the causes of these different outcomes could lead to improvements in quality of care for children with stroke. Therefore, the objective of this study was to determine whether the weekend effect on clinical outcomes exists in children with ischemic stroke (IS) and HS using a large national sample.
The analysis was based on data files from the National Inpatient Sample (NIS), 2002 to 2011. A comprehensive synopsis on NIS data is available at http://www.hcup-us.ahrq.gov. The NIS is a deidentified database; therefore, no institutional review board approval was required.
Children aged 1 to 18 years admitted because of IS or HS were identified using the International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) primary diagnosis codes 433 to 434 and 436 to 437 for IS and 430 to 432 for HS. Patients were categorized as having been admitted on weekdays (Monday through Friday) versus weekends (Saturday or Sunday), a distinct variable in the NIS database that is calculated from the original admission date. The NIS database does not allow identification of specific admission hours. Patients with missing data on the day of admission were excluded. We performed separate analyses to determine whether the weekend effect exists in children with IS and HS.
Study variables included were age, sex, race/ethnicity, and comorbidities obtained from Agency for Healthcare Research and Quality comorbidity data files, including diabetes mellitus, hypertension, congestive heart failure, renal failure, chronic lung disease, deficiency anemia, valvular heart disease, fluid and electrolyte disorder, liver disease, AIDS, obesity, paralysis, solid tumor without metastasis, metastatic cancer, alcohol abuse, and coagulopathy.
ICD-9-CM secondary diagnosis codes identified secondary diagnoses in children with IS and HS, such as aphasia (784.3), hemiplegia and hemiparesis (342), migraine (346), cerebral artery dissection (443.2), anomalies of the cerebrovascular system, which include arteriovenous malformations of the brain and other congenital anomalies of the cerebral vessels (747.81), coma (780.01), fever (780.60), moyamoya disease (437.5), hydrocephalus (331.3–331.4), seizure (345), meningitis (320.0–322.9), congenital heart disease (745–747), altered mental status (780.97), cardiomyopathy (425), systemic lupus erythematosus (710), pneumonia (486, 481, 482.8, and 482.3), urinary tract infection (599.0 and 590.9), sepsis (995.91, 996.64, 038, 995.92, and 999.3), deep venous thrombosis (451.1, 451.2, 451.81, 451.9, 453.1, 453.2, 453.8, and 453.9), pulmonary embolism (415.1), illicit drug abuse (305.6–305.7), atrial fibrillation (427.3), nicotine dependence (305.1), dyslipidemia (272.0), and sickle cell disease (282.6). We also used ICD-9-CM primary and secondary procedure codes to estimate the percentage of children who underwent in-hospital procedures including ECG (89.5), echocardiography (88.72), computed tomography (87.03), magnetic resonance imaging (88.91), ultrasound head and neck (88.71), cardiac catheterization (88.5), valvular repair (35.0–35.9), cerebral angiography (88.41), thrombolytic therapy (99.10), craniotomy (01.2, 01.24), clipping (39.51 and 39.52), coiling (39.71 and 39.79), dialysis (54.98 and 39.95), cardiopulmonary resuscitation (99.60), intubation (96.04), mechanical ventilation (96.72), transfusion (99.04), and gastrostomy (431.1–431.9).
Hospitals were categorized by US geographic region as Northeast, Midwest, West, and South. The admitting hospitals were classified as teaching or nonteaching; teaching hospitals were those that have an American Medical Association–approved residency program or have membership in the Council of Teaching Hospitals.15 Hospitals were further characterized into small, medium, or large size based on available hospital beds. The definition of large size may vary from exceeding 325 to exceeding 450 acute hospital beds depending on the location and characteristics of the hospital.
We determined the length of stay (LOS) and hospital charges (amount billed for services but not specific amounts received in payment). Discharge status was categorized into home/selfcare, nursing facility (includes skilled nursing facilities and inpatient rehabilitation), unknown, and in-hospital mortality.
SAS 9.3 software (SAS Institute, Cary, NC) was used to convert NIS database data into weighted counts to generate national estimates following Healthcare Cost and Utilization Project recommendations. We performed univariable analyses with χ2 tests for categorical variables and t tests for continuous variables to identify differences in study variables and end points among children admitted with IS and HS during weekends versus weekdays. Two logistic regression models were created. Model 1 included all patients, and logistic regression analysis was used to identify the association between weekend admission and odds of in-hospital mortality. Model 2 included patients who were discharged alive, and logistic regression analysis was used to identify the association between weekend admission and odds of discharge to a nursing facility (versus home/selfcare). Logistic regression models were adjusted for age (as a continuous variable), sex, and confounding factors (as categorical variables) that were significant (P≤0.05) in univariable analyses. We performed subanalyses for patients with IS and HS in which we examined each outcome measure in children aged 1 to 12 years and in those aged 13 to 18 years separately. We chose to dichotomize age at 12/13 years because this is the typical time point at which children undergo puberty. In the subanalyses, factors that were significant for the age group in univariable analyses at the 0.05 level were included in the multivariable model. When a variable had 10 or fewer patients, we reported that variable as <11 in accordance with the Healthcare Cost and Utilization Project data requirements because small numbers of observations (ie, individual discharge records) reported in publications increase the risk for identification of individuals.
Of 8467 children aged 1 to 18 years with IS, 28% were admitted on a weekend. Mean age and sex distribution in the weekend and weekday groups were not different. Comorbid conditions, hospital size, insurance status, and hospital teaching status were not different between groups. The rate of weekend admission was higher in hospitals located in the Western region of the United States (P=0.02) and was lower in hospitals located in the Southern region of the United States (P=0.04; Table I in the online-only Data Supplement).
The rate of seizures was higher on weekends compared with weekdays (7.2% versus 4.3%; P=0.03), whereas moyamoya disease was less frequent in the weekend group compared with the weekday group (5.5% versus 10.5%; P=0.004). In-hospital procedures were not different between the groups. Children admitted on weekends had a higher rate of discharge to nursing facilities (25.5% versus 18.6%; P=0.003). In-hospital mortality was not different among children admitted on weekends versus weekdays (4.1% versus 3.3%; P=0.4). The LOS (8±10 versus 7±12 days; P=0.003) and mean hospital charges ($73 295±$96 223 versus $53 022±$70 913; P<0.0001) were significantly greater among children admitted on weekends versus weekdays (Table 1).
After adjusting for age, sex, and confounding factors, the odds of discharge to a nursing facility was greater among children admitted on weekends than among those admitted on weekdays (odds ratios [ORs], 1.5; 95% confidence interval [CI], 1.1–1.9, P=0.006; Table 2).
In a subanalysis in which we dichotomized children by age (1–12 and 13–18 years), among those aged 1 to 12 years, discharge to a nursing facility was more common among those admitted on weekends versus weekdays (OR, 1.58; 95% CI, 1.02–2.46; P=0.04), after adjusting for seizure, moyamoya, and mechanical ventilation. In-hospital mortality was not significantly different in the younger age group among those admitted on weekends versus weekdays (OR, 1.60; 95% CI, 0.81–3.31; P=0.2), after adjusting for seizure, moyamoya, and mechanical ventilation. Among those aged 13 to 18 years, neither discharge to a nursing facility (OR, 1.41; 95% CI, 0.94–2.12; P=0.09) nor in-hospital mortality (OR, 0.50; 95% CI, 0.14–1.65; P=0.2) was more common among those admitted on weekends versus weekdays, after adjusting for moyamoya.
Of 10 919 children aged 1 to 18 years with HS, 25.3% were admitted on a weekend. Mean age and sex distribution in the weekend and weekday groups were not different. The proportion of children with obesity was significantly higher among children admitted on weekends compared with those admitted on weekdays (2% versus 1%; P=0.04). The rate of weekend admission was higher in hospitals located in the Western region of the United States (P=0.05). Patients with Medicare/Medicaid insurance were more likely to have admissions on weekdays compared with on weekends. Hospital size and hospital teaching status were not different between the groups (Table II in the online-only Data Supplement).
The rate of deep vein thrombosis was lower in children admitted on weekends compared with those admitted on weekdays (0.1% versus 0.9%; P=0.006). The rates of mechanical ventilation, intubation, gastrostomy, and blood transfusion were significantly higher among children admitted on weekends compared with those admitted on weekdays. Children admitted on weekends had a higher in-hospital mortality rate (12% versus 8%; P=0.006). The rate of discharge to a nursing facility was not different among children admitted on weekends versus weekdays (22% versus 20%; P=0.3). LOS (10±13 versus 9±14 days; P=0.0004) and mean hospital charges ($168 069±$191 459 versus $85 013±$119 630; P<0.0001) were significantly greater among children admitted on weekends versus weekdays (Table 3⇓).
After adjusting for age, sex, and confounding factors, in-hospital mortality was higher among children admitted on weekends compared with those admitted on weekdays (OR, 1.4; 95% CI, 1.1–1.9; P=0.04; Table 4⇑).
In a subanalysis in which we dichotomized children by age (1–12 and 13–18 years), among those aged 1 to 12 years, discharge to a nursing facility was not more common among those admitted on weekends versus weekdays (OR, 1.12; 95% CI, 0.76–1.65; P=0.5), after adjusting for mechanical ventilation. In-hospital mortality was not significantly different in the younger age group among those admitted on weekends versus weekdays (OR, 1.46; 95% CI, 0.92–2.31; P=0.09), after adjusting for mechanical ventilation. Among those aged 13 to 18 years, discharge to a nursing facility was not more common in those admitted on weekends versus weekdays (OR, 1.12; 95% CI, 0.80–1.56; P=0.5). However, the in-hospital mortality rate was higher among those admitted on weekends versus weekdays (OR, 1.60; 95% CI, 1.12–2.45; P=0.04).
The weekend effect is a phenomenon that exists for many medical conditions in which patients admitted on the weekend have poorer outcomes compared with patients admitted on weekdays with the same disease process.9 Although several studies have demonstrated a weekend effect among patients with adult stroke,13,14 there are no data available about a possible weekend effect among pediatric patients with stroke. In the present study, we used a large national sample and found that a weekend effect does exist for both pediatric IS and HS.
In our study, 28% of children hospitalized primarily for IS were admitted on a weekend. In-hospital mortality did not differ among children with IS admitted on weekend days versus weekdays. Although a difference in mortality might not exist, it is also possible that the low mortality rate in both groups (4.1% weekends; 3.3% weekdays) did not allow for detection of a difference in this sample. However, children with IS admitted on weekends had a higher rate of discharge to a nursing facility compared with those admitted on weekdays, even after adjusting for age, sex, and other potential confounders. In subanalyses by age group, we found similar trends among those aged 1 to 12 years and 13 to 18 years for discharge to a nursing facility although the finding did not reach statistical significance in the older age group.
A study using the NIS database reported no weekend effect on in-hospital mortality and discharge disposition between weekend and weekday admissions for 599 087 adult patients with IS.16 Another large study conducted in acute care hospitals in California, which included 24 565 adult patients with IS, found no significant difference in the mortality rate between weekend and weekday admissions.17 However, Saposnik et al13 found that adult patients with IS admitted on weekends had an increased rate of mortality and were less likely to be discharged to the same place of residence than those admitted on weekdays. After adjusting for confounding factors, weekend admission increased the risk of mortality by 14%. In the Get with the Guidelines-Stroke Program, off-hour (weekends and weeknights) IS presentation was associated with slightly higher in-hospital mortality.14 Hospital admission for IS outside the United States on weekend days compared with weekdays has been associated with higher mortality in Canada,13,18 Sweden,19 Japan,20 and Taiwan.21
One study22 examined whether comprehensive stroke centers (CSCs) overcome the weekend versus weekday gap in stroke treatment and mortality. No such difference in mortality was observed at CSCs on weekends versus weekdays. In another study of 2 CSCs,23 a total of 2211 adult patients were included. Patients were categorized into 4 groups: an intracerebral hemorrhage group, an arterial IS group not treated with intravenous tissue-type plasminogen activator, an arterial IS treated with tissue-type plasminogen activator, and a transient ischemic attack group. No significant differences were found in any group after adjustment for site, age, admission National Institutes of Health Stroke Scale score, and blood glucose, with respect to the rate of symptomatic intracerebral hemorrhage, modified Rankin Scale score on discharge, discharge disposition, 90-day modified Rankin Scale score, or 90-day mortality when comparing weekend and weekday admissions.23 These 2 studies suggest that organization of stroke care, such as that provided by CSCs, may eliminate the weekend effect. The Thrombolysis in Pediatric Stroke trial led to improved preparedness by participating centers for acute diagnosis and management of children presenting with stroke.24 Initiatives to increase readiness of hospitals that treat children with stroke and to standardize pediatric stroke protocols may help to eliminate the weekend effect in children.
In our study, 25.3% of children hospitalized primarily for HS were admitted on a weekend. Children with HS admitted on weekends had a 1.4-fold increased in-hospital mortality rate compared with children admitted on weekdays even after adjustment for age, sex, and confounders. In subanalyses by age group, we found similar trends among those aged 1 to 12 years and 13 to 18 years for in-hospital mortality although the finding did not reach statistical significance in the younger age group. However, the rate of discharge to a nursing facility was not different among children admitted on weekends versus weekdays.
In adult literature on HS, weekend admission for intracerebral hemorrhage was associated with increased risk-adjusted mortality. After adjusting for potential confounders, weekend admission was associated with an increased risk of in-hospital mortality by 12%.25 Weekend admission, however, has not been associated with mortality among patients hospitalized with subarachnoid hemorrhage.26
In our study, LOS and mean hospital charges in both IS and HS were significantly greater among children admitted on weekends versus weekdays. Also, in our study, among children with HS, rates of mechanical ventilation, intubation, gastrostomy, and blood transfusion were significantly higher among children admitted on weekends compared with those admitted on weekdays. These increased rates of procedures associated with greater illness severity might contribute to the increased LOS and higher total charges observed. Many studies have demonstrated the weekend effect on total hospital charges and LOS in patients with adult stroke.16
One explanation for longer hospital stays and higher charges among patients admitted on weekends is that, although hospitals may be able to administer acute treatments such as intensive care or thrombolytics on weekends, other services such as nutritional services, speech/swallow therapy, physical therapy, occupational therapy, case management, and discharge disposition services may be under staffed on weekends.16
In our study, the rate of weekend admission to teaching versus nonteaching hospitals was not different for children admitted with IS or HS. A large adult study conducted in acute care hospitals in California found that the weekend effect was larger in major teaching hospitals compared with that in nonteaching hospitals.17 In our study, tissue-type plasminogen activator administration was not different on weekends versus weekdays. In adult stroke literature, tissue-type plasminogen activator administration has been reported to be more common on weekends than weekdays.22,27
There are many potential explanations for the weekend effect found in the present pediatric study. Weekend shifts often have fewer physicians and nurses, and practitioners caring for patients on weekends may have decreased familiarity with patients.9,28–30 As previously reported in adults, the weekend effect may also be due, in part, to the availability of interventional procedures because they are often less likely to be performed on weekends.31 For example, sedation services for magnetic resonance imaging and other diagnostic procedures for children may not be available on the weekends at some institutions.
There are several limitations to the present study. The NIS is an administrative clinical database in which no information was available on cause of death, stroke severity, imaging findings, size of stroke or hemorrhage, or graded outcome scales. The study has limitations inherent to the NIS data set, namely, the accuracy of the diagnosis and procedure codes listed in the discharge summaries. We used primary ICD-9-CM codes for identifying patients with IS, which had a true positive rate of up to 84% in previous population-based studies.32 However, the yield may be lower in the pediatric population. In 1 pediatric study from a single tertiary care center, ICD-9 codes for IS had accuracies of 46% to 83%.33 For adults with intracerebral hemorrhage, the sensitivity, specificity, and positive predictive values of ICD-9 codes were 85%, 96%, and 89%, respectively.34 We used primary ICD-9-CM codes for identifying patients with subarachnoid hemorrhage, which had an accuracy of 79%35 and a positive predictive value of 76%36 in previous population-based studies. It is also possible that comorbid conditions, medical complications, and in-hospital procedures were under-reported. The NIS data ascertain whether patients were admitted on a weekend (Saturday or Sunday) but do not identify the specific day of the week that patients were admitted. Thus, it was not possible to perform a sensitivity analysis in which weekend admissions were defined as those that occur on Fridays, Saturday, or Sundays. Furthermore, we were not able to examine differences in outcome for children admitted on holidays. Variations in hospital staffing levels and staff expertise at the time a patient is hospitalized are not measured in this database. Also, it was not possible to determine whether rates of admission to intensive care units versus regular hospital floors were different on weekends versus weekdays because admission unit is not recorded in the NIS. Finally, among those with IS, there was a higher rate of seizures and a lower prevalence of moyamoya among those admitted on weekends versus weekdays, but the reasons for and implications of these findings are unclear, and these findings could even be spurious. Without the ability to review the charts, the authors were unable to explore these results in more detail. However, seizures are a common presenting feature of childhood stroke,37 and moyamoya38 is a common arteriopathic cause of childhood stroke, so these variables were maintained in the multivariable analyses for IS. Additional study is needed to determine whether the presentations or underlying causes of stroke are different among those admitted on weekends versus weekdays and how these features might contribute to the observed weekend effect. Despite limitations, our results were consistent among the outcomes measured and with previous adult literature.13,26,39 Our use of a national database allowed comprehensive coverage of stroke-related hospitalizations across the country among a large sample of pediatric patients with stroke.
Children with IS admitted on weekends had a higher rate of discharge to a nursing facility, and children admitted with HS on weekends had a higher rate of in-hospital mortality. Although additional research is needed to replicate our findings, it is possible that certain disparities or variations in resources, expertise, or the number of healthcare providers working during weekends exist, which may explain the weekend effect observed in this study. The continued establishment of pediatric CSCs and standardized stroke protocols like the ones used by participating centers in the Thrombolysis in Pediatric Stroke trial24 may eliminate the weekend effect just as this phenomenon seems to have been eliminated among adult CSCs. Studies are required to understand factors that affect care and may provide new avenues for implementing quality improvement initiatives. Continued initiatives aimed at expediting treatment of children with acute stroke should be encouraged.
Presented in part at the International Stroke Conference, Nashville, TN, February 11–13, 2015.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.013453/-/DC1.
- Received March 13, 2016.
- Revision received March 17, 2016.
- Accepted March 18, 2016.
- © 2016 American Heart Association, Inc.
- Lynch JK,
- Hirtz DG,
- DeVeber G,
- Nelson KB
- 2.↵National Center for Health Statistics. Leading Causes of Death and Numbers of Deaths, According to Age: United States, 1980 and 2001 CDC Web site.. 2001. http://www.Cdc.Gov/nchs/data/hus/hus04trend.Pdf#032. Accessed January 5, 2016.
- Broderick J,
- Talbot GT,
- Prenger E,
- Leach A,
- Brott T
- Jordan LC,
- Johnston SC,
- Wu YW,
- Sidney S,
- Fullerton HJ
- Saposnik G,
- Baibergenova A,
- Bayer N,
- Hachinski V
- Reeves MJ,
- Smith E,
- Fonarow G,
- Hernandez A,
- Pan W,
- Schwamm LH
- 15.↵Description of Data Elements Nationwide Inpatient Sample (NIS). Healthcare Cost and Utilization Project Web site. https://www.hcup-us.ahrq.gov/db/nation/nis/DataElements_NIS_2001_CORE.pdf. Accessed January 2, 2016.
- Hoh BL,
- Chi YY,
- Waters MF,
- Mocco J,
- Barker FG 2nd
- Janszky I,
- Ahnve S,
- Ljung R
- McKinney JS,
- Deng Y,
- Kasner SE,
- Kostis JB
- Bernard TJ,
- Rivkin MJ,
- Scholz K,
- deVeber G,
- Kirton A,
- Gill JC,
- et al
- Crowley RW,
- Yeoh HK,
- Stukenborg GJ,
- Medel R,
- Kassell NF,
- Dumont AS
- Broderick J,
- Brott T,
- Kothari R,
- Miller R,
- Khoury J,
- Pancioli A,
- et al
- Tirschwell DL,
- Longstreth WT Jr
- Spolaore P,
- Brocco S,
- Fedeli U,
- Visentin C,
- Schievano E,
- Avossa F,
- et al
- Amlie-Lefond C,
- Bernard TJ,
- Sébire G,
- Friedman NR,
- Heyer GL,
- Lerner NB,
- et al