Identifying cases & Quality Assurance: Data Completeness & Accreditation The Trauma Audit & Research Network (TARN) Data Collection session

Identifying cases: 2 case studies

System 1: Retrospective Data capture Clinical (ICD10) codes  Clinical coding departments use a coding system called ICD10  ICD10: International Classification of Diseases  ICD10 codes document: Admission reason (Injury, Medical, Elective, Complication)  ICD10 codes that begin with S or T indicate injury e.g.  S82.2: fracture to shaft of tibia  S82.21: open shaft of tibia  S82.20: closed shaft of tibia  T055: Traumatic amputation of both legs

 Trust IT dept. can therefore generate weekly spreadsheet showing:  Patients discharged previous week with any S or T ICD10 code  Filtering out:  <3 days stay, discharge destination = home  65+ isolated NOF  65+ isolated pubic rami fracture  Minor injuries  Full list of all applicable ICD10 codes:  Result: List of potential TARN patients  Check imaging reports to ensure inclusion System 1: Retrospective Data capture Clinical (ICD10) codes

System 1: Retrospective Data capture Clinical (ICD10) codes

Example ICD10 spreadsheet Age Admission Date Discharge DateDisch. DestinationLoSDiag. 1 ICDDiag. 1 Text Diag. 2 ICDDiag. 2 Text Diag. 3 ICD Diag. 4 ICD Diag. 5 ICD Diag. 6 ICD 8405/06/201016/07/2010NHS NURSING41S0650 TRAUMATIC SUBDURAL HAEMATOMAK709ALCOHOLIC LIVER DAMAGE UNSPECIFIEDL031B181F102N /12/201006/01/2011USUAL RESIDENCE23S825 FRACTURE OF MEDIAL MALLEOLUSS526 CLOSED FRACTURE OF RADIUS AND ULNA, LOWER ENDV99X 8226/11/201001/12/2010PATIENT DIED5S063 CLOSED HINDBRAIN CONTUSIONW199 [X]UNSPECIFIED FALL, OCCURRENCE AT UNSPECIFIED PLACEN390I48XI120I /06/201016/07/2010USUAL RESIDENCE45S323 CLOSED FRACTURE OF ILIUM, UNSPECIFIEDN390 URINARY TRACT INFECTION, SITE NOT SPECIFIED NOSI10XF339F319E /07/201009/08/2010USUAL RESIDENCE36S220 CLOSED FRACTURE THORACIC VERTEBRAW190 [X]UNSPECIFIED FALL, OCCURRENCE AT HOMEN390J22XI639I10X 7519/09/201005/10/2010 NON-NHS RUN RESID. CARE HOME16S422 CLOSED FRACTURE PROXIMAL HUMERUS, NECKS721 CLOSED FRACTURE OF FEMUR, INTERTROCHANTERICW194N390D649E /10/201003/11/2010 NHS NURSING HOME/31S327 MULTIPLE FRACTURES OF LUMBAR SPINE AND PELVISS499 [X]UNSPECIFIED INJURY OF SHOULDER AND UPPER ARMW109N390R32XI10X 7503/11/201021/12/2010USUAL RESIDENCE48Z501[X]OTHER PHYSICAL THERAPYS327 MULTIPLE FRACTURES OF LUMBAR SPINE AND PELVISS499I350R32XI10X  Patients potentially have multiple ICD10 (Diagnosis) codes  Ensure your Trust reviews the first 5 diagnosis codes  Reviewing Primary Diagnosis code only – will definitely result in missed cases

System 1: Retrospective Data capture Clinical (ICD10) codes: Advantages Captures patients who bypass ED (transfers in, GP admissions) TARN can liaise with IT to help set this up: SQL script Limited staff resource required Used by most Trauma Units who employ Retrospective data capture Used as backup to “live” data capture by most Major Trauma Centres

 Only admitted patients are assigned an ICD10 code  Transfers out from ED  Deaths in ED  Separate system required to capture these  Accuracy of Trust ICD10 coding  Use of NOS (not otherwise specified) codes can increase potential cases  Cases admitted for Rehabilitation only – not easily identified  Delay between discharge and clinical coding System 1: Retrospective Data capture Clinical (ICD10) codes: Disadvantages

System 2: Live Data capture Employed by Royal Stoke University Hospital Major Trauma Centre North West Midlands & North Wales Major Trauma Network

System 2: Live Data capture Employed by Royal Stoke University Hospital Major Trauma Centre  Diary in Resuscitation completed by ED Consultant.  Data Co-ordinators pagers that receive Trauma Calls.  ED system search: Patients who were imaged & admitted.  Daily contact with relevant Wards: Notification of any trauma patients admitted.  Attend morning Orthopaedic/Neurosurgical case discussion meetings.  Liaise with Rehabilitation Co-ordinators re: Trauma patients seen on wards When eligible cases identified:  Early care data collected on forms by TNC & Data coordinators whilst still inpatient.  Later care data (Operation, ICU, Imaging, Ward, Discharge data) taken from online systems.

System 2: Live Data capture: Advantages Employed by Royal Stoke University Hospital Major Trauma Centre  Very little backlog; patients often dispatched to TARN day after discharge.  Case notes only required for “missed” cases identified post discharge.  Regular communication between Coordinators & Clinicians.  Rehabilitation Co-ordinators involved very early on.  Deaths data: Available from Bereavement office quickly, no waiting for inquest.

System 2: Live Data capture: Disadvantages Employed by Royal Stoke University Hospital Major Trauma Centre  Greater staff resource required  Post discharge ICD10 report required to capture “missed” patients.

Collecting the data  Retrospective Data Entry: Post discharge  Access to electronic Imaging and Theatre systems required  Enter data directly from notes or use Pro-forma  CORE PROFORMA IN REPORTS SECTION OF WEBSITE  DO NOT have to start and finish a submission in one session

Quality Assurance Data Completeness (quantity) & Data Accreditation (quality)

Data completeness % Measure of Expected v Submitted number of cases  HES Data used to calculate expected no. of cases per Trust  HES = Hospital Episode Statistics database  HES data contains ICD10 codes assigned by Trusts in previous year: 2013  TARN Inclusion criteria applied to HES data  No. of Expected cases then derived  Expected number of cases used as a guide only. Not a 100% target!  All eligible cases should be submitted to TARN

Data completeness % Measure of Expected v Submitted number of cases  % of expected v submitted cases shown on TARN website  Updated every 4 months: End of  March  July  November  Shown as Trust and individual Hospital figures  Should always be viewed alongside Hospital Survival rate

Data Completeness calculation Data Completeness 76.1% Submissions: NUMERATOR238 Expected submissions: DENOMINATOR313 Data completeness76.1% HES dataset/ Denominator TARN submissions/Numerator 1-49%: May not reflect true practice 50-79%: May not reflect true practice 80%+: View with confidence

Improving Data completeness HES v TARN 2013 Data comparison exercise

HES v TARN 2013 comparison exercise Comparison spreadsheet produced:  Green: Cases appear in both datasets (Submission ID shown)  Black: Cases appear in HES dataset only (not submitted in TARN) ‘Not TARN eligible’ field: Completed by Trust & Fed-back to TARN  Missing cases: Issues identified & entered: Increase in Numerator  Ineligible cases: Removed from expected no. of cases: Decrease in Denominator

HES v TARN 2013 comparison exercise results Commonly missed groups:  Aged 65+ Hip fractures with other (non hand or foot) fracture  Traumatic SDH admitted to medical wards  Aged 65+ Pelvic fractures  Spinal fractures  Patients whose LOS is exactly 3 days Common reasons for the variance between the HES and TARN:  Inaccuracies in ICD10 coding  Old injuries being coded.  Use of NOS (Not Specified) codes, which varies from trust to trust  Rehabilitation only admissions  Elective admissions  45 Trust participated in exercise  >3,700 cases removed from Expected no. of cases  18% reduction & comparable Data Collection increase

Data Accreditation % (quality) Measure of frequency of CORE data field completion CURRENT ACCREDITATION FIELDS Glasgow Coma Score or Intubation/ventilation (Pre Hospital or ED) Incident or Call 999 Date/time Arrival time Transfer: Reason & Date of Referral CT time Operation: Start time, grade & speciality of surgeon & grade of anaesthetist ED Doctors: Time seen, grade & speciality Injury detail – proportion of NFS codes

NEW ADDITIONS TO ACCREDITATION FIELDS *Pre-existing conditions **Pupil reactivity for patients with AIS 3+ (Serious) head injuries Data Accreditation%: Recent additions From December 2014 onwards * New Probability of Survival (Ps)14 model ‘Other’ and ‘Not Known’ detrimental to Data Accreditation **Future Probability of Survival model

Data Accreditation  Data Accreditation report available on TARN  Missing fields highlighted

Questions?