Proportion of patients admitted following injury

1
Proportion of patients admitted following injury
setting a frame for analysis of variation
with diagnosis, calendar year, age and
sex?JM.Lauritsen Accident/Injury
AnalysisGroup. Dept. of Orthopaedics Odense
University Hospital, Odense Denmark.
Contributions by Thomas Foged.
2
Setting

• Denmark Europe
• Odense University Hospital AE dept.
• Public financing (tax) no fee for contact
• Data Recording Patient system
• Population 225000
• Trauma center level 1 (one fifth of Denmark
  about 1 mio)
• Previous study showed that ICISS figures are
  comparable btw. Denmark and Australia

3
ICISS severity grading of emergency room contacts
Are Danish values comparable to the original
AUS/NZ values ?

• Overall comparability between Denmark and
  Australia.
• Although there are differences in diagnosis
  specific survival proportions btw. Danish and
  Australia the good story is that when
  calculating overall survival probability (ICISS)
  differences are minor.
• This indicates that possibly ICISS can be
  calculated based on SRRs derived from data from
  other parts of the world.
• The question is then about patterns of diagnosis
  and admission

4
Material for study

• Period Jan 1st 1994 to June 30th 2006
• Contacts Injury general (n421283)
  Violence (n12104)
• Excluding Medical cause and attempted suicide
• Data completenessNo diagnosis coded n1
  (whole period)No admission status n5 (whole
  period)No cause of contact registration lt 50
  records per year

5
Material in analysis

• Period Jan 1st 1994 to June 30th 2006
• From 1 5 S/T diagnoses per patient
• Only contacts with at least one S/T diagnosis
  are included in analysis n 410139 patients
  nd 481778 diagnosis codings ndc 1024
  different S/T codes (3 digit) 108 (2 digit)

6
Analysis phases

• Phase 1 Variation in admission by age, sex and
  period (patient level)
• Phase 2 Investigation of positional stability
  of diagnosis
• Phase 3 Proportion of mortality known
• Phase 4 Analysis of variation by diagnosis

7
Phase 1 Percentage of admission by age, sex and
period (patient level)

• Age Percent 95 CIlt 20 5.4 5.3- 5.5 20-40
  6.8 6.7- 6.9 41-64 10.9 10.7-11.2
  65Males 25.0 24.2-25.8 Females 30.0 29.5-3
  1.0
• No Period effect

8
Phase 2 Position of diagnosis

• From 1- 6 diagnoses coded.
• For 1 of patients S/T was not the first.
• These patients had 15.8 of all S/T diagnosis
• Admission percentage
• S/T was not first 22.3 (CI 22.0-22.6)
• S/T was first 9.0 (CI 8.9-9.1)
• No period effect, but large position effect -
  OR 2.6 (CI 2.5-2.7)

9
Phase 3Variation in proportion of dead on
arrival and all mortality

• Died as inpatient N736
• Dead on arrival N382 total 1118Diagnostic
  problem Only 1/3 autopsy
• Time pattern percentage dead on arrival 40
  in 1994 26 at end of period (Highly
  significant trend)
• Consequence Registerfollow-up to determine e.g.
  7 day, 30 day and 1 year mortality, plus
  alternate sources.

10
The problem is what we do not see (unreported
cases)
11
Conclusion

• Phase1 Variation in admission levels with - sex
  and age, but not with period.
• Phase 2 Variation with position of first S/T
  diagnosis, but no period effect.
• Phase 3 Highly significant variation (trend) in
  composition of mortality known at hospital
  (Underreporting and insufficient diagnostics).
• Phase 4 Before starting we need clear
  definitions of who to include, how to handle
  mortality, age and sex issues.