Injury Data and NCHSInjury Data and NCHS
U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICESCenters for Disease Control and PreventionNational Center for Health Statistics
Lois A. FingerhutLois A. FingerhutNCIPC Conference, May 2005NCIPC Conference, May 2005
Focus for this morningFocus for this morning
• NCHS injury-related web pages
• ICD-9 to ICD-10 comparability file
• Frameworks for presenting data
• Poisoning- a recent example of a question of definition
• Injury severity- new collaborative work
Our new injury website Our new injury website pagespages
• One stop shopping for questions/presentations/publications regarding NCHS surveys and data sets that have an injury component
• Links to non-NCHS sources (eg WISQARS)
• Up-to-date information on the International Collaborative Effort (ICE) on Injury Statistics
Injury Mortality DataFrom the National Vital Statistics System
Data Source | Mortality-Injury Summary | Injury Death Codes
Publications |Presentations |Tabulated Data |Public Use Data
Data Tools | Query Systems | Related Links
National Databases
ICD-9 to ICD-10 Comparability
• A Guide to State implementation of ICD-10 for mortality ( 2000)
• Comparability reports going back to ICD-4 to ICD-5
• ICD-9 t o ICD-10 detail
• Downloadable file on ICD-9 to ICD-10 comparability study
• Full file documentation
• SAS statements
http://www.cdc.gov/nchs/datawh/statab/unpubd/comp.htm#A%20guide%20to%20state%20implementation%20of%20ICD-10
ICD-10 on the WHO WebsiteICD-10 on the WHO Website
• WHO Family of International Classifications
• http://www.who.int/classifications/en/
•There is a complete online version of ICD-10
Frameworks for presenting Frameworks for presenting datadata
• External causes
•ICD-9 and ICD-9 CM external cause code matrices
• ICD-10 external cause code matrix
• Injury Diagnoses
•ICD-9 CM Diagnosis codes: Barell Matrix
•ICD-10 injury diagnosis code matrix
External cause matrix-basic External cause matrix-basic structurestructure
Intent of injury
Mechanism Unintentional
Suicide
Homicide Undeter-mined
Other
MV-traffic
Cut
Firearm ICD-9, ICD-9-CM and ICD-10 External cause codes
Poisoning
Struck by/
against
Suffocation
Etc…..
External Cause of Injury Mortality External Cause of Injury Mortality Matrix (ICD-10)Matrix (ICD-10)
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
45,000
50,000
MotorVehicleTraffic
Firearm Poisoning Falls Suffocation Drowning
Selected Mechanisms of injury
Number of deaths
Unintentional Self-harm Assault Undetermined Legal int/other
Injury deaths by matrix: 2002Injury deaths by matrix: 2002
New International New International Recognition!Recognition!
• The Mortality Reference Group (MRG), the group charged with refining and recommending changes to the ICD formally recommended
“Publish External Cause of Injury Mortality Matrix data in addition to standard WHO tabulations to facilitate statistical analysis”
Source: www.who.int/classifications/network/en/icelandexecutifsummary.pdf
Nature of injury Fractures Internal Open Sprains Burns Dislocationwounds & Strains
Site of injuryHead and NeckTraumatic Brain Injury
specific sites
Spinal Cord Injuryspecific sites
Vertebral Column Injuryspecific sites
Torsospecific sites
Extremitiesspecific sites
SYSTEM WIDE
Barell Matrix-basic structureBarell Matrix-basic structure
ICD-9 CM codes
Barell matrix: a standard for Barell matrix: a standard for presenting injury morbidity datapresenting injury morbidity data
• See ICE webpage for full description of the matrix• www.cdc.gov/nchs/about/otheract/ice/barellmatrix.htm
• Barell V, Aharonson-Daniel L, Fingerhut LA, MacKenzie EJ, et al. An introduction to the Barell body region by nature of injury diagnosis matrix. • Injury Prevention 2002;8:91-6.
• National Hospital Discharge Survey: 2002 Annual Summary With Detailed Diagnosis and Procedure Data (table 24)
• www.cdc.gov/nchs/data/series/sr_13/sr13_158.pdf
Nature of injury Fractures Internal Open Amputations Burns Dislocation Bloodorgan wounds vessel
Body region of injuryHead and NeckTraumatic Brain Injury
specific sites
Spine and Upper backspecific sites
Torsospecific sites
Extremitiesspecific sites
Unclassifiable by reionSYSTEM WIDE
ICD-10 Injury Mortality ICD-10 Injury Mortality Diagnosis MatrixDiagnosis Matrix
ICD-10 ‘S’ & ‘T’ codes
Head and neck allTraumatic brain injuryOther headNeckHead and Neck
Spine and upper backSpinal cordVertebral column
TorsoThorax AbdomenPelvis and lower backAbdomen, lower back & pelvisTrunk, other
ExtremitiesUpper extremitiesHipOther lower extremities
Not classifiable by siteMultiple body regionsSystem wideUnspecified
ICD-10 Body regionof injury categoriesfor mortality
Level 1
Level 2
Additional detail is available, butAdditional detail is available, butnot necessarily appropriate for not necessarily appropriate for
mortalitymortality
• For example, Level 2- ‘Other lower extremities’ can be disaggregated to Level 3 categories
• Thigh L3-31• Hip & Thigh L3-32• Upper Leg and thigh L3-33• Knee L3-34• Lower leg L3-35• Foot L3-36• Ankle L3-37• Other and multiple ankle and foot L3-38• Toes L3-39• Other lower limb L3-40
ICD-10 mortalityNature of injuryCategories
FracturesDislocationInternal organ injuriesOpen woundsAmputationsBlood vesselsSuperficial & contusionCrushingBurnsEffects of foreign bodyOther effects of external
causesPoisoningToxic effectsMultiple injuriesOther specifiedSprain or strainMuscle and tendon injuriesNerve injuriesUnspecified
Level 1
Level 2
Total and any mentions of injury Total and any mentions of injury diagnoses by body region: 2002diagnoses by body region: 2002
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
# of mentions
Total mentionsAny mention
Total and any mentions of injury Total and any mentions of injury diagnoses by nature of injury: diagnoses by nature of injury:
20022002
-
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
# of mentions
Total mentions
Any mention
Injuries mentioned in MVT deaths Injuries mentioned in MVT deaths (44,065 deaths and 70,684 injuries) by body (44,065 deaths and 70,684 injuries) by body
region and nature of injury: US, 2002region and nature of injury: US, 2002
- 10.0 20.0 30.0 40.0
Head & Neck
Spine & Back
Torso
Extremities
Not classifiable byregion
Percent of all injuries mentioned
Fracture
Internal
BloodvesselMultiple
Other
Unspecified
• ICD-9 vs ICD-10
• Underlying cause: external cause codes (ICD-10 X & Y codes)
• ICD-10 Multiple cause: T codes for substances
• Mental health “F” codes
• Nondependent abuse
• Dependent abuse
• Alcohol intoxication (not included here)
• Adverse effects codes (not included here)
Definitional issues: poisoning Definitional issues: poisoning mortalitymortality
ICD ‘definitions’ of drug ICD ‘definitions’ of drug poisoningpoisoning
ICD-10
Nondependent abuse of drugs 305.2-.9 F11-16, 18-19 (not .2)
Dependent abuse 304 F11-16,18-19(.2)
Unintentional E850-E858 X40-X44
Suicide E950(.0-5) X60-X64
Undetermined E980(.0-.5) Y10-Y14
Homicide E962.0 X85
[Alcohol intoxication 305.0 F10.0]
ICD-9
ICD-9 vs ICD-10ICD-9 vs ICD-10Substance selectionSubstance selection
• ICD-9 underlying cause codes for poisoning more specific than ICD-10 codes
• ICD-10, to get specific substances
• Literals from the death certificate
•Code the multiple cause data
ICD mortality coding varies by ICD mortality coding varies by countrycountry
ICD-9-UCOD
ICD-10-UCOD
Substance-specific
England and Wales
1990-2000
2001-2002 Literals from text files
Canada 1990-1999
2000-2002 From under-lying cause
United States
1990- 1998
1999-2002 Multiple cause coding
Drug poisoning death rates: US, Drug poisoning death rates: US, 20022002
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
Under15
15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+
Unintentional
Suicide
Undetermined
DAD
NDAD
Deaths per 100,0000 pop
Leading specified substances Leading specified substances mentioned in drug deaths for 35-54 mentioned in drug deaths for 35-54
year olds: US, 2002year olds: US, 2002
1. Cocaine T40.5: 25%
2. Other specified opioids T40.2: 21%
3. Antidepressants T43.0-T43.2: 11%
4. Alcohol T51: 11%
5. Carbon Monoxide T58: 8%
6. Heroin T40.1: 10%
7. Benzodiazepines T42.4: 9%
8. Methadone T40.3 10%
Next StepsNext Steps
• In 2003, the MRG recommended to the URC of the WHO that new rules will apply in January 2006 to the underlying cause coding of certain ICD codes from Mental and Behavioral Disorders (F10-F19)
• If there is any mention of an external cause on the certificate, the code will be to the external cause rather than MBD code
• Codes in the F10-F19 range with a 4th digit of .0 (acute intoxication) will be coded to poisoning codes in the external cause of poisoning section
Injury SeverityInjury Severity
Some new considerations for national data
Acknowledging many of these next slides from
Dr. Ellen MacKenzie, Johns Hopkins University
September 2004 MeetingSeptember 2004 Meeting
• Meeting convened at NCHS bringing together national and international experts in the area of injury severity scoring
• Why? Because the current standards “AIS” and “ICDMAP” are proprietary and many believe that
•There are nonproprietary alternatives
•They should not be proprietary any longer
Injury Severity Indices:Injury Severity Indices:Major Areas of ApplicationMajor Areas of Application
Triage
Prognostic Evaluation
Research and Evaluation Clinical Research
Systems Evaluation
Surveillance and Epidemiology
EM
The Abbreviated Injury Scale The Abbreviated Injury Scale (AIS)(AIS)
A classification of injuries based on anatomic descriptors
A severity score ranging from 1 (minor) to 6 (maximum injury, virtually unservivavle) assigned to each injury
EM
Scores are subjective Scores are subjective assessments assigned by a assessments assigned by a group of experts and implicitly group of experts and implicitly based on based on fourfour criteria: criteria:
Threat to life
Permanent Impairment
Treatment Period
Energy Dissipation
EM
AISAIS
Currently, most widely used severity score based on anatomic descriptors
Official injury data collection tool of NHTSA crash investigation teams
Developed in 1971; 5th revision to be published in 2005
EM
Using AIS for Multiple InjuriesUsing AIS for Multiple Injuries
for predicting survivalfor predicting survival
Injury Severity Score (ISS)
The New Injury Severity Score (NISS)
The Anatomic Profile (AP) and the Anatomic Profile Scale (APS)
EM
ICD-Based Measures ICD-Based Measures
of Injury Severityof Injury Severity
ICD to AIS Conversion
ICISS Family of Measures
EM
ICDMAPICDMAPICD-CM to AIS ConversionICD-CM to AIS Conversion
Converts ICD-9CM coded discharge diagnoses into AIS injury descriptors, AIS scores and computes ISS, NISS, APS
Conservative measure of injury severity - refer to as ICD/AIS scores
Limitations identified; revision needed
EM
ICISSICISS
Based strictly on ICD rubrics
The ICISS score for a given patient is the product of the survival risk ratios (SRRs) associated with each ICD diagnosis
SRRs are calculated by dividing the number of survivors among patients with a specific ICD by the total number of patients with that ICD code
EM
Refining the ICISSRefining the ICISS
Computation of SRRs: based on multiple trauma patients or patients with single injuries?
Database used for calculation of SRRs:
Trauma centers only vs. population based ?
Include ALL deaths, only deaths in ED or hospital or only in-hospital deaths ?
Registry data vs. administrative data ?
Regional/local vs. national data?
Computation of ICISS: use product of SRRs or lowest SRR?
EM
To think about….To think about….
Need to keep in mind the application; severity (case mix?) adjustment for use with hospital discharge data (HDD) – also mortality data, ambulatory care encounter data ?
By necessity – must be based on ICD (but what do we lose – how good can we get without physiology ?)
Age, gender, co-morbidities and mechanism are important in case mix adjustment – and all are measurable using HDD
EM
and . . .and . . .
Are we just interested in measures that predict mortality ?
Need to carefully consider the overall advantages (current and future) of the AIS classification in any recommendations
What are implications of the 2005 revision of the AIS and the ICD-10 (CM??)
EM
What we knowWhat we know
• National trends in injury-related hospital discharges and emergency dept. visits reflects utilization, but not differences in injury severity
• ICD codes alone cannot distinguish severity among injuries
• ICD-10 has provided no real guidance on how to select a main injury among multiple cause of injury mortality data
The “practical problems”The “practical problems”
• ICD-9 CM is still being used for coding morbidity data; annual updates to CM continue
• Most recent version of ICDMAP doesn’t recognize new codes
• ICD-10 CM doesn’t yet have an implementation date and there is no new ICDMAP based on ICD-10
• ICD codes used for mortality data often lack specificity
What was discussedWhat was discussed
• Strengths and weaknesses of different severity scales
• Solutions for administrative data acknowledging the limitations of the source data (e.g., non-specific coding, changes in admission practices)
• Can we measure threat to function as well as threat to life?
What “we” would have liked What “we” would have liked to accomplishto accomplish
• Agree upon a measure of injury severity to add to NCHS survey data
• Incorporate a method to identify the “main injury” in mortality and add it to the mortality file
• Recommend a standard measure to users of administrative databases (e.g., Statewide hospital discharge data sets)
Where might these measures be Where might these measures be used?used?
• Tracking Department’s Healthy People Objectives
• CDC Futures Initiative- Health Protection Goals
• NCHS reports: Health, United States
• Injury Chart book(s)
• Statewide trauma and general injury databases
NCHS Data Sources for NCHS Data Sources for Injury SeverityInjury Severity
• National Hospital Discharge Survey
• National Hospital Ambulatory Medical Care Survey-ED component
• National Health Interview Survey ??
• Mortality data from vital statistics
Other Federal SourcesOther Federal Sources
• Agency for Health Care Research and Quality
•Medical Expenditure Panel Survey (MEPS)
•Healthcare Cost & Utilization Project (HCUP)•Nationwide Inpatient Sample (NIS)•State inpatient databases (SID)•State emergency dept databases (SEDD)
http://www.ahrq.gov/data/hcup/
Discharge Disposition for Discharge Disposition for injury diagnoses: NHDS, injury diagnoses: NHDS,
20022002
62%
26%
9%
2%
1%
Routine/home
Transferred
NS (but known alive)
Died
Not stated
Survival Risk Ratios (SRR’s): Survival Risk Ratios (SRR’s): NHDS, 2002NHDS, 2002
0.860
0.880
0.900
0.920
0.940
0.960
0.980
1.000
Based on all 7 dx fields;Ordered by # of discharges
Discharged alive: all discharges
11stst listed injury dx by nature of listed injury dx by nature of injury grouped by AIS: NHDS, injury grouped by AIS: NHDS,
1999-20021999-2002
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
All Fractures Spr/str Int org Openwounds
Super/cont Burns Unspec
AIS 1-3
AIS 4
AIS 5-6
AIS 9
% of 1st listed dx AIS 1-9
NHAMCS- ED: Percent hospitalized NHAMCS- ED: Percent hospitalized by nature and body region of by nature and body region of
injury, injury, 1999-20021999-2002
0.0
5.0
10.0
15.0
20.0
25.0
% of all injury visits
NHIS: “severity” variables in NHIS: “severity” variables in addition to nature of injuryaddition to nature of injury
• Days out of school
• Days out of work
• Hospitalized
• Any limitations of activity
Mortality data from NVSSMortality data from NVSS• ICD-10: uses all digits; up to 20 listed
diagnoses
• For 2001, range (0-15 injuries listed)
• 1 injury listed 65% of deaths
• 2 injuries 22%
• 3 injuries 8%
• 4 -15 injuries 4%
• How can we select the most severe injury?
• Do we need to include underlying cause of death?
So…So…
• Most hospitalized injuries are not fatal
• Most ED visits don’t result in hospitalization
• Too many mortality records lack detailed diagnosis codes
Optimism….Optimism….
• SRR’s and ICISS can be readily calculated from hospital discharge data – we are creating a file with them that will be on the web
• AIS could theoretically be added also to hospital discharge data file once it is in public domain and ICDMAP is updated
• Main injury (method yet to be determined) will be added to mortality file
Consensus (well almost….)Consensus (well almost….)
• AIS and ICDMAP should be maintained and updated
• BOTH should be non-proprietary
• ICISS is a useful alternative to the current non-updated ICDMAP
•Statistical methods need continued evaluation and improvement•Lowest SRR may be better than
ICISS
Visit us at:Visit us at: www.cdc.gov/nchs/injury.htmwww.cdc.gov/nchs/injury.htm
Email us at: [email protected] us at: [email protected]