J. Lee Annest, Ph.D. Director, Office of Statistics and Programming

1
Using FARS data to Classify Unspecified
MV-traffic Deaths by Person Type in NVSS data
J. Lee Annest, Ph.D. Director, Office
of Statistics and Programming National Center for
Injury Prevention and Control
2
The Question

• How many MV-traffic deaths are occupants in the
  vehicle?

3
The Problem

• In 2001, almost 35 of 42,443 MV-traffic deaths
  in the NVSS were unspecified for person type
  (occupant, motorcyclist, pedal cyclist,
  pedestrian, other)
• Current death certificate used in most states not
  set up to routinely capture person type
• New death certificate has a separate data item
  that should help classify MV traffic deaths by
  person type (needs to be assessed)

4
One Possible Solution

• Use data from the Fatal Analysis Reporting System
  (FARS) to allocate unspecified MV-traffic deaths
  into specified person type categories (occupant,
  motorcyclist, pedal cyclist, pedestrian, other)

5
Method

• Use the known distribution of FARS deaths for
  specified person type to determine where to
  allocate the unspecified MV-traffic deaths in the
  NVSS.

6
Allocation Process

• First, allocate unspecified deaths into
  specified person type categories other than
  occupant
• Then, put the remaining unspecified MV-traffic
  deaths into occupant.

7
Allocation Rules Applied

• Distribute MV-traffic deaths into person type
  categories within specific age-by-sex groupings
• Allocate unspecified deaths into person type
  categories other than occupant as follows
• gt If the number of specified death is higher in
  FARS than NVSS, then change NVSS to the higher
  number
• gt If the number of specified deaths is higher in
  NVSS than FARS, then do not change NVSS

8
Example

• Pedal cyclist (Males 15-19 years)
• Deaths
• Data Source Observed Allocated
• FARS 66 --
• NVSS 57 66
• --------------------------------------------------
  --
• Redistributed unspecified 9

9
Example

• Pedestrian (Females 1-4 years)
• Deaths
• Data Source Observed Allocated
• FARS 35 --
• NVSS 56 56
• --------------------------------------------------
  --
• Redistributed unspecified 0

10
Allocation Rules Applied

• 3. Allocate the remaining unspecified deaths to
  the occupant category
• 4. Sum deaths across age-by-sex groupings to get
  final numbers overall and by broader age and sex
  groups.
• 5. After allocation, the distributions of
  MV-traffic deaths by person type should be very
  similar for FARS and NVSS

11
Conclusions

• Allocation procedure works well for obtaining
  national estimates
• Procedure not tested for state and local
  estimates need to consider
• gt FARS deaths reported by where the fatal
  crash occurred
• gt NVSS deaths often reported by place of
  residence rather than place of occurrence