Kids' Inpatient Database: Empowering Scientific Discovery (Text Version)

On September 28, 2010, Jay Berry made this presentation at the 2010 Annual Conference. Select to access the PowerPoint® presentation (350 KB).

---

Slide 1

Kids' Inpatient Database: Empowering Scientific Discovery

Jay G. Berry MD MPH

Complex Care Service, Cerebral Palsy Program, Program for Patient Safety and Quality
Division of General Pediatrics, Children's Hospital Boston, Boston, MA

Slide 2

2:30 am, December 14th, 2003

• First KID Exposure:
  • Setting:
    • Pediatrics resident, on-call
    • Admitting patients with Dr. Raj Srivastava
  • Patient with hypoplastic left heart syndrome:
    • Surgical complication
    • Related to surgical team inexperience

Slide 3

2:30 am, December 14th, 2003

• First KID Exposure:
  • ICD-9 code book
  • KID 1997
  • Modeling: mortality variation among hospitals

Slide 4

2:30 am, December 14th, 2003

• First KID Exposure:
  • ICD-9 code book
  • KID 1997
  • Modeling: mortality variation among hospitals

Higher mortality rates for children undergoing surgery for hypoplastic left heart syndrome in low volume and non-teaching hospitals.

Slide 5

KID: Personal Impact

• Power of large, administrative datasets to study outcomes and utilization of children with rare diseases.
• Entrance into the world of pediatric quality of care and health services research.

Slide 6

 

My Clinical and Research Interest

• Children with medical complexity:
  • Chronic health conditions
  • Multiple co-morbidities
• Surgery collaboration:
  • Predict outcomes
  • Develop care plans

Image: A child on crutches with a dog is shown.

Slide 7

Predicting Outcomes

• Children with medical complexity:
  • Attributes:
    • Low prevalence
    • Unique combinations of co-morbid conditions
  • Existing outcome evidence:
    • Single institution-based
    • Longitudinal data, over multiple decades

Slide 8

Improving Outcome Prediction

• Kids' Inpatient Database:
  • Powered to study rare conditions:
    • Robust stratified sample
    • Nationally-representative
  • Elements for co-morbidity-outcome analyses:
    • Diagnoses, procedures
    • Mortality, complications
    • Inpatient utilization

Slide 9

Tracheotomy in Children

• Indication:
  • Overcome life-limiting respiratory compromise
• Rise in patient complexity:
  • Multiple co-morbid conditions
• Major caregiving burden:
  • Life disruption

Image: A baby with a tracheotomy tube is shown.

Slide 10

Mortality Following Tracheotomy

• Existing evidence:
  • 1-3% early mortality rate
  • Single institutions
• Mortality may be under-estimated:
  • Rising patient complexity
  • Presence of tenuous co-morbid conditions

Slide 11

Tracheotomy Mortality Analysis

• Kids' Inpatient Database 2006:
  • Tracheotomy ICD-9 codes
  • In-hospital mortality
• Partition, Regression Tree Modeling:
  • Demographic and co-morbid conditions
  • Characteristic combinations and mortality

Slide 12

KID 2006
4,751 Tracheotomy Hospitalizations

Clinical Condition	Percentage
Neurological Impairment	46%
Chronic Lung Disease	44%
Upper Airway Anomaly	28%
Congenital Heart Disease	19%
Prematurity	13%
≥ 2 Clinical Conditions	67%

 

Slide 13

Tracheotomy Mortality Regression Tree

Image: A tree chart depicts the different mortality rates that children undergoing tracheotomy experience, depending on which co-morbid conditions they possess. It consist of the following:

• All Patients Mortality = 9%
  • Congenital Heart Disease (-) 6%
    • Prematurity (-) 5%
      • Airway Anomaly (+) 2%
      • Airway Anomaly (-) 6%
    • Prematurity (+) 17%
  • Congenital Heart Disease (+) 19%
    • Airway Anomaly (+) 7%
    • Airway Anomaly (-) 27%
      • Age ≥ 1 year 12%
      • Age < 1 year 30%

Slide 14

Tracheotomy Mortality Regression Tree

Image: A tree chart depicts the different mortality rates that children undergoing tracheotomy experience, depending on which co-morbid conditions they possess. The tree chart is simular to the chart in Slide 13, but sections are darkened and muted. Only the following sections are highlighted to be visible:

• All Patients Mortality = 9%
  • Congenital Heart Disease (-) 6%
    • Prematurity (-) 5%
      • Airway Anomaly (+) 2%

Slide 15

Tracheotomy Mortality Regression Tree

Image: A tree chart depicts the different mortality rates that children undergoing tracheotomy experience, depending on which co-morbid conditions they possess. The tree chart is simular to the chart in Slide 13, but sections are darkened and muted. Only the following sections are highlighted to be visible:

• All Patients Mortality = 9%
  • Congenital Heart Disease (+) 19%
    • Airway Anomaly (-) 27%
      • Age < 1 year 30%

Slide 16

KID Impact for Children Undergoing Tracheotomy

• Bringing evidence to the bedside:
  • Individualizing outcome prediction
  • Counseling families of risk and benefit
  • Increased attention to at-risk patients

Slide 17

Keep it coming!

• The HCUP-KID empowers scientific discovery that is leading to improvements in care for children!
• Future data element expansion will enhance its power!

Slide 18

Thank you

• AHRQ and the HCUP team
• Pamela Owens and Anne Elixhauser
• Raj Srivastava and Don Goldmann