200 likes | 292 Views
Assessing the Concordance of Coded Morbidity and Mortality Data for In-Hospital Trauma-Related Deaths. Presenter: Bridget Allison Research Team: Kirsten McKenzie, Sue Walker, Leanne Aitken, Andrea Besenyei, Deirdre McDonagh
E N D
Assessing the Concordance of Coded Morbidity and Mortality Data for In-Hospital Trauma-Related Deaths Presenter: Bridget Allison Research Team: Kirsten McKenzie, Sue Walker, Leanne Aitken, Andrea Besenyei, Deirdre McDonagh Affiliations: National Centre for Classification in Health, QUT Institute of Health and Biomedical Innovation, QUT Queensland Trauma Registry, UQ
National Centre for Classification in Health • Mission:The National Centre for Classification in Health (NCCH) is the Australian centre of excellence in health classification theory and an expert centre in coding systems. The NCCH is dedicated to supporting our clients in their use of health classifications and related products.
Background • In-hospital mortality rate key indicator of trauma system effectiveness • Few researchers have investigated the concordance of causes of death and causes of hospital admission • Are the same causes of trauma listed on death certificate as documented in hospital records?
Factors affecting concordance of morbidity and mortality data • Different coding guidelines and selection rules • Differences in classification versions used • Coding errors • Documentation differences • Autopsy and certification processes
Different coding guidelines, selection rules, and classification versions used • Principal Diagnosis(ICD-10-AM) • The principal diagnosis is considered to be chiefly responsible for occasioning the patient's episode of care in hospital • Underlying Cause of Death (ICD-10) • The underlying cause of death is defined as (a) the disease or injury which initiated the train of morbid events leading directly to death, or (b) the circumstances of the accident or violence which produced the fatal injury • Cause of injury is the UCOD for deaths due to trauma • Multiple Causes of Death • All other conditions, including the injuries resulting from the external cause, are coded as multiple causes of death • Part I and Part II of Death Certificate
Coding errors • Incomplete or inaccurate coding affects data quality • Previous research has identified errors in injury coding and external cause coding affecting up to 28% of medical records
Documentation differences • Morbidity coders have complete hospital record to code from while mortality coders largely confined to death certificate • Previous research found that concordance of hospital records and death certificates varies by: • Principal diagnosis • Type and number of co-morbidities of the patient • Time from admission to death • Acuity of the condition • Details available regarding the diagnoses
Autopsy and certification processes • Certifier may be unaware of prior injury • Certifier may not consider the injury to be a contributory factor towards the death • Certifier may not document the injury and/or external cause on the death certificate • Autopsy results may not be available for coding of hospital records and/or death certificate
Research Questions • What was the in-hospital mortality rate for patients admitted to hospital for trauma? • Was trauma recorded on the death certificate of patients who died in hospital? • If trauma was recorded, was there concordance in the coded data between the morbidity and mortality collections for trauma patients who died in hospital?
Research Methodology • Participants = 1672 patients admitted to hospital for >24hrs with PDx of injury • Procedure: • Data matched to NDI using probabilistic matching based on demographic variables (name, sex, DoB etc) • NDI reported specificity 98.5% and sensitivity 89.2% (Kelman, ANZJPH, 2000, 24 (2) pp. 201-203) • Matched cases formed sample for this research
Results • Of 1672 trauma admissions, 3.6% died in hospital (n=60) • Medical vs Traumatic COD • 89% had trauma coded in NDI though 18% of these had medical condition as UCOD • 11% did not have trauma coded in NDI • Age by medical vs traumatic COD • All but one person <65 had trauma as UCOD • Only 50% of >65 year olds had trauma as UCOD
Results • Concordance of cause of injury between morbidity and mortality data: • 36% same cause of injury • 14% more defined cause in NDI • 22% less defined cause in NDI • 28% no match • Age by concordance of cause of injury data • 67% <65 had matched or more detailed NDI data • Only 27% >65 year olds had matched NDI data
Discussion • Despite being admitted and treated for trauma related injuries • 18% of cases did not have a trauma-related UCOD and • 11% of cases did not have a trauma-related code in the National Death Index • People over 65 years old less likely to have trauma documented on death certificate, though trauma arguably increases risk of dying from co-morbidities
Discussion • Where trauma was documented on death certificate, causes of injury not concordant between morbidity and mortality data with 50% less defined or no match in NDI • Use of unspecified codes on NDI(e.g. Exposure to unspecified factor)compared with more detailed coded hospital morbidity data(e.g. fall from bed) -> Unable to determine that injury caused by fall if using mortality data alone
Recommendations • Use linked hospital morbidity data in conjunction with national mortality data to afford greater detail for trauma outcome research • Consider aims of your research and determine most appropriate source of coded data • Understand the underlying constraints inherent in the production of national morbidity and mortality datasets and how this affects data quality
Future Developments • Study conducted using data from a single hospital and a single trauma registry and using a small sample • Expanding to include all QTR sites for 2003, ~ 12,000 trauma admissions
Further information Bridget Allison Health Information Manager National Centre for Classification in Health b.allison@qut.edu.au Ph. 07 3864 3911