1. Non-fatal injury indicators Colin Cryer and Rolf Gedeborg International Collaborative Effort on Injury Statistics, Swansea, Wales, September 2010 What are the pros and con’s of different approaches to measuring serious non-fatal injury?

2. Program The session now A ‘basket’ of ICD diagnoses ICISS-based definitions Tomorrow (8:30 – 10:30 am) Non-fatal indicators work Facilitators Colin Cryer & Rolf Gedeborg

3. Aim To produce a draft specification of a serious non-fatal injury indicator for use in international comparisons. To produce a draft specification of a serious non-fatal injury indicator for use in international comparisons. Today’s presentations feed into the discussion that will occur tomorrow - where we aim to agree a (partial) draft specification. Today’s presentations feed into the discussion that will occur tomorrow - where we aim to agree a (partial) draft specification. 3Colin Cryer, IPRU, University of Otago

4. Main Issue Operational definition of serious non-fatal injury Operational definition of serious non-fatal injury Two main themes Two main themes “A ‘basket’ of ICD diagnoses” vs “ICISS-based definitions” “A ‘basket’ of ICD diagnoses” vs “ICISS-based definitions” At the heart of the debate. At the heart of the debate. Talks between now and 10:00 aimed at informing that discussion tomorrow. Talks between now and 10:00 aimed at informing that discussion tomorrow. 4Colin Cryer, IPRU, University of Otago

5. We want valid indicators - indicators that measure what they intend to measure Indicators point Good indicator Bad indicator 5Colin Cryer, IPRU, University of Otago

6. International comparison of serious non-fatal injury. Colin Cryer Injury Prevention Research Unit University of Otago, New Zealand International Collaborative Effort on Injury Statistics, Swansea, Wales, September 2010

7. International comparison of serious non-fatal injury. Colin Cryer Injury Prevention Research Unit University of Otago, New Zealand International Collaborative Effort on Injury Statistics, Swansea, Wales, September 2010 Definition of serious non-fatal injury using a ‘basket’ of ICD diagnoses.

8. International comparison of serious non-fatal injury. Colin Cryer Injury Prevention Research Unit University of Otago, New Zealand International Collaborative Effort on Injury Statistics, Swansea, Wales, September 2010 Definition of serious non-fatal injury using a ‘basket’ of ICD diagnoses. Minimising health service effects in international comparisons.

9. Potential biases in international comparisons - 1 Serious non-fatal injury Serious non-fatal injury Source of data? Source of data? Hospital inpatient / discharge / separations data Hospital inpatient / discharge / separations data Assumption: Assumption: Most ubiquitous source collected by countries? Most ubiquitous source collected by countries? Most accurate source w.r.t. diagnosis of injury and external cause. Most accurate source w.r.t. diagnosis of injury and external cause. Major problem Major problem Variations in place and time in who gets admitted to hospital. Variations in place and time in who gets admitted to hospital. Eg. health service provision and access. Eg. health service provision and access. 9Colin Cryer, IPRU, University of Otago

10. Potential biases in international comparisons - 2 Major problem Major problem Variations in place and time in who gets admitted to hospital. Variations in place and time in who gets admitted to hospital. Eg. health service provision, policy and access. Eg. health service provision, policy and access. Eg. head injury – hospital A has scanning facilities available in O/P so minor head injury not admitted vs hospital B has not so minor head injury routinely admitted for observation. Eg. head injury – hospital A has scanning facilities available in O/P so minor head injury not admitted vs hospital B has not so minor head injury routinely admitted for observation. Want to remove this health service effect Want to remove this health service effect Option for operational definition Option for operational definition Injuries that have a high probability of admission (PrA) Injuries that have a high probability of admission (PrA) Others? Others? 10Colin Cryer, IPRU, University of Otago

11. Potential biases in international comparisons - 3 Direct method Direct method Estimate diagnosis-specific probabilities of admission (Prob of Admission project) Estimate diagnosis-specific probabilities of admission (Prob of Admission project) Select only those (for our operational definition) that have a high probability of admission. Select only those (for our operational definition) that have a high probability of admission. Alternative: ICISS-based method Alternative: ICISS-based method 11Colin Cryer, IPRU, University of Otago

12. Probability of admission (PrA) Project Thanks to collaborators Thanks to collaborators Soufiane Boufous, Senior Research Fellow, Injury Division, The George Institute for International Health, Australia; Li-Hui Chen, Office of Analysis and Epidemiology, National Center for Health Statistics, Maryland, USA; Nick Dessypris, Department of Hygiene and Epidemiology, Athens University Medical School, Greece; Lois Fingerhut, L A Fingerhut Consulting, Washington, DC, USA; Vicki Kalampoki, Department of Hygiene and Epidemiology, Athens University Medical School, Greece; Jens Lauritsen, Consultant, Orthopedic Dpt., Accident Analysis Group, Odense Universitetshospital, Sdr., Denmark ; Bruce Lawrence, Pacific Institute for Research and Evaluation, Calverton, Maryland, USA; Alison Macpherson, School of Kinesiology and Health Science, York University, Toronto, Canada; Ted Miller, Pacific Institute for Research and Evaluation, Calverton, Maryland, USA; Catherine Perez, Agència de Salut Pública de Barcelona, Spain; Eleni Petridou, Department of Hygiene and Epidemiology, Athens University Medical School, Greece; Margie Warner, Office of Analysis and Epidemiology, National Center for Health Statistics, Maryland, USA Soufiane Boufous, Senior Research Fellow, Injury Division, The George Institute for International Health, Australia; Li-Hui Chen, Office of Analysis and Epidemiology, National Center for Health Statistics, Maryland, USA; Nick Dessypris, Department of Hygiene and Epidemiology, Athens University Medical School, Greece; Lois Fingerhut, L A Fingerhut Consulting, Washington, DC, USA; Vicki Kalampoki, Department of Hygiene and Epidemiology, Athens University Medical School, Greece; Jens Lauritsen, Consultant, Orthopedic Dpt., Accident Analysis Group, Odense Universitetshospital, Sdr., Denmark ; Bruce Lawrence, Pacific Institute for Research and Evaluation, Calverton, Maryland, USA; Alison Macpherson, School of Kinesiology and Health Science, York University, Toronto, Canada; Ted Miller, Pacific Institute for Research and Evaluation, Calverton, Maryland, USA; Catherine Perez, Agència de Salut Pública de Barcelona, Spain; Eleni Petridou, Department of Hygiene and Epidemiology, Athens University Medical School, Greece; Margie Warner, Office of Analysis and Epidemiology, National Center for Health Statistics, Maryland, USA 12Colin Cryer, IPRU, University of Otago

13. Methods 6 countries involved 6 countries involved Agreed protocol data supply Agreed protocol data supply Submitted Submitted Checked and analysed by IPRU Checked and analysed by IPRU 13Colin Cryer, IPRU, University of Otago

14. Probability of admission (PrA) Results / Issues / Problems Summary of results on spreadsheet Summary of results on spreadsheet Small number of diagnoses show consistently high estimated PrA Small number of diagnoses show consistently high estimated PrA Lower 95% CI for PrA >0.75 Lower 95% CI for PrA >0.75 Fractured shaft and neck of femur Fractured shaft and neck of femur Wide confidence intervals for many diagnoses Wide confidence intervals for many diagnoses Diagnoses with potentially consistently high PrA – ie. Upper 95% CI >0.75 Diagnoses with potentially consistently high PrA – ie. Upper 95% CI >0.75 see over for list see over for list 14Colin Cryer, IPRU, University of Otago

15. S052 – Ocular laceration and rupture with prolapse and loss of intraocular tissue. S052 – Ocular laceration and rupture with prolapse and loss of intraocular tissue. S063 – Focal brain injury S063 – Focal brain injury S272 - Traumatic haemopneumothorax S272 - Traumatic haemopneumothorax S360 – Injury of spleen S360 – Injury of spleen S361 – Injury of liver and gall bladder S361 – Injury of liver and gall bladder S364 – Injury of small intestine S364 – Injury of small intestine 15Colin Cryer, IPRU, University of Otago

16. PrA Project - Issues ICD-10 ICD-10 Used only 4-character -> Lack of specificity Used only 4-character -> Lack of specificity Can we infer high PrA ICD-10 diagnoses from ICD-9 results - Eg. open long bone fractures; brain haemorrhage / laceration Can we infer high PrA ICD-10 diagnoses from ICD-9 results - Eg. open long bone fractures; brain haemorrhage / laceration Use of only 1st diagnosis listed eg. for head injury Use of only 1st diagnosis listed eg. for head injury Inconsistent results. Inconsistent results. Surprising for certain diagnoses Surprising for certain diagnoses Eg. traumatic subdural haemorrhage (v low PrA for 1country). Eg. traumatic subdural haemorrhage (v low PrA for 1country). Combining ICD-9 and ICD-10 results Combining ICD-9 and ICD-10 results Possible for some diags (eg. fractured neck & shaft of femur) Possible for some diags (eg. fractured neck & shaft of femur) Less obvious for others Less obvious for others 16Colin Cryer, IPRU, University of Otago

17. Conclusions In theory, using a ‘basket’ of diagnoses is a solution to reducing health service effects on international comparisons. In theory, using a ‘basket’ of diagnoses is a solution to reducing health service effects on international comparisons. Creating an operational definition of high PrA diagnoses requires some judgement. Creating an operational definition of high PrA diagnoses requires some judgement. My proposed set includes My proposed set includes Fractured neck and shaft of femur Fractured neck and shaft of femur Those with UCL>0.75 for all available countries Those with UCL>0.75 for all available countries Long bone open fractures Long bone open fractures Brain laceration and haemorrhage Brain laceration and haemorrhage Spinal cord lesion Spinal cord lesion Intra-thoracic and intra-abdominal injury (excl. bladder & urethra) Intra-thoracic and intra-abdominal injury (excl. bladder & urethra) This is a starting point for discussion This is a starting point for discussion 17Colin Cryer, IPRU, University of Otago