HCUP: Research Applications and Tools

1
HCUP Research Applicationsand Tools

• Claudia Steiner, M.D., M.P.H.

2
HCUP Tools

• Software to assist in using the data
• SAS and SPSS load programs
• Labels (HCUP data elements, DRGs, CCS)
• Statistics
• HCUPnet
• Clinical Classifications Software
• HCUP Quality Indicators

3
National Statistics

• HCUP Research Notes (www.ahcpr.gov/data/hcup)
• Hospital Inpatient Statistics, 1996, 1995 and
  1992
• Diagnosis and Procedure Combinations in Hospital
  Inpatient Data, 1996 and 1993
• Descriptive statistics by Insurance Status

4
HCUPnet

• Interactive on-line access to HCUP data
• Users generate tables of outcomes by diagnoses
  and procedures
• Can be cross-classified by patient and hospital
  characteristics

5
Types of Output from HCUPnet

• Rank order conditions/procedures by LOS, charges,
  deaths
• Examine LOS, charges, mortality, discharge status
  for conditions/ procedures by
• patient characteristics (age, sex, payer, median
  income of ZIP Code)
• hospital characteristics (owner, teaching,
  location, bedsize, region)

6
HCUPnet Can Answer Questions Such As

• What percentage of hospitalizations for children
  are uninsured, by state?
• What are the most expensive conditions treated in
  U.S. hospitals?
• What is the trend in admissions for depression?
• Will I have enough cases to do my analysis?
• How do my estimates compare with HCUPnet?
  (validation)

7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
Chartbooks

• Detailed stats on web . . . but what does it all
  mean?
• Provide easily accessible information based on
  HCUP data

24
(No Transcript)
25
Chartbooks (contd)

• Overview of hospital care
• Future chartbooks
• Procedures in U.S. hospitals
• Childrens hospitalizations
• Hospitalization of women
• Hospitals that care for patients covered by
  Medicaid and the uninsured

26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
Clinical Classifications Software (CCS)

• Aggregates ICD-9-CM diagnosis and procedure codes
• Creates smaller number of clinically meaningful
  clusters
• Reporting tool
• Can be applied to any ICD-based administrative
  data

30
Two Related Classification Systems

• Single level CCS
• Mutually exclusive categories
• 259 diagnosis categories
• 231 procedure categories

• Multi-level CCS
• Multi-level hierarchical system
• Aggregates single-level CCS into broader
  categories
• Splits CCS into more detailed categories

31
(No Transcript)
32
(No Transcript)
33
How can CCS be Used?

• Rank hospitalizations by type of condition -- N
  of cases, LOS, total charges
• Develop clinically based utilization profiles
• Explore data by grouping conditions and
  procedures conveniently
• Predict future resource use (Cowen et al., Med
  Care. 1998361108-1113.

34
HCUP Quality Indicators

• Use with administrative data
• Multiple dimensions of quality
• Diverse populations
• High tech and ordinary care

35
33 Indicators in 3 Areas

• Potentially avoidable adverse hospital outcomes
• Potentially inappropriate utilization of hospital
  procedures
• Potentially avoidable admissions or ambulatory
  care sensitive conditions

36
Relatively Few, Common Variables

• Diagnoses
• Procedures
• Age
• Sex
• Admission source
• Discharge status
• Procedure dates

37
HCUP QIs The Next Generation

• Contract with UCSF-Stanford
• Expand / refine current QIs
• pediatric, chronic medical, ambulatory sensitive
  conditions
• Analytic / methods development
• population-based denominators for some QIs
• risk adjustment
• separate patient safety indicators (PSIs) module
• literature review and empirical validation

38
Using the SID An Example

• Use of in-hospital diagnostic and therapeutic
  procedures
• Comparing Hispanics and non-Hispanic whites
• Examine disparities explore potential
  explanations

39
Procedure Use

• Examined vector of procedures (PR1-PRn)
• Identified all procedures used by more than 5 of
  discharges
• CT scan of head
• Diagnostic ultrasound of head
• Echocardiogram
• MRI of head
• Cerebral arteriogram
• Carotid endarterectomy

40
ResultsUse of Non-Invasive Diagnostic Procedures
41
ResultsUse of Invasive Procedures
42
Results - Logistic Regression (NY)Odds ratios
for Hispanic vs. Non-Hispanic whites
43
Summary of Findings

• Hispanics had higher use of noninvasive tests and
  lower use of invasive procedures.
• Controlling for hospitals experience with
  Hispanic patients eliminated or greatly reduced
  disparities in procedure use.
• Disparities are lower in CA than in NY

44
Use the AS for

• Compare inpatient and outpatient surgery
• Small area variation research
• Study surgeries that only occur outpatient
• Compare state-specific trends in use of
  outpatient setting

45
Using the AS An Example

• Use of outpatient setting for major surgery
• current public policy concern
• Examine use of outpatient setting for procedures
  grouped by body system
• compare across states

46
Ranking of Rates for Major OR Procedure
Groupings, Outpatient, 1997
47
NIS An Example

• Charges for Childhood Asthma by Hospital
  Characteristics. Pediatrics, Dec 1998
• John Meurer, Medical College of Wisconsin
• Uses NIS 1994 data, linked to Hospital Weights
  file
• Applies All Patient Refined-Diagnosis Related
  Group (APDRG) severity classification

48
Meurer et al. Results

• Study finds asthma severity did not differ
  significantly by hospital location or teaching
  status
• Nonprofit hospitals treated higher proportion of
  children with extreme severity asthma
• Urban teaching hospitals treated more children
  from low-income neighborhoods

49
NIS An Example

• Patients Undergoing Major Surgery in Hospitals
  with Limited Nursing Staffs Suffer more
  Complications Image, Nov 1998
  
• Christine Kovner, New York University
• Uses NIS 1993 data, linked to AHA Survey
• Applies Quality Indicator software to the NIS

50
Kovner Results

• Study finds that the fewer FTE registered nurses
  per inpatient day ?
• higher incidence of
• urinary tract infection
• pneumonia
• thrombosis
• pulmonary compromise

51
(No Transcript)
52
Conclusion

• HCUP provides multiple resources
• Databases for study
• Research reports
• Tools for analysis
• Statistics to support policy decisions, provide
  national comparisons

53
For More Information - How to Reach us

• E-mail
• hcup_at_ahrq.gov
• Phone
• 301-594-3075
• Website
• http//www.ahrq.gov/data/hcup
• HCUPnet
• http//www.ahrq.gov/data/hcup/hcupnet.htm