HIMSS ASIAPAC 11 CONFERENCE & LEADERSHIP SUMMIT 20 23 SEPTEMBER 2011 MELBOURNE, AUSTRALIA
INTRODUCTION AND APPLICATION OF A CODING QUALITY TOOL PICQ JOE BERRY OPERATIONS AND PROJECT MANAGER, PAVILION HEALTH 2
Why is Coding Quality Important? Good Clinical Documentation Guide Good documentation supports quality patient care Present or future episodes of hospitalisation have access to the records they need The records are accurate, complete and understandable to ensure continuity of care This continuity supports high quality and safe patient care Quality clinical documentation also ensures the information is reliable for many other purposes Research Planning Quality improvement activities Quality documentation quantifies hospital activity Activity based funding (ABF) is focusing attention on the reliability and validity of coded health data 3
Making the Most of a Scarce Resource 2010 AIHW survey: National health reform and Activity Based Funding (ABF) implementation will increase demand for skilled HIMs and CCs Increased requirement for high quality data collection and reporting AIHW projections for the national workforce inflows over the next five years: Low end estimates, 300 new HIMs and 1,492 CCs required (total 1,792) High end estimates, total 3,101! HIMs in demand for non clinical coding work Clinical coding is one of several career options for HIMs CCs can take 12-18 months to meet the level of independent coding required Your coding quality management system needs to support Coding resource performance management to make the most of a scare resource Coding resource education and training to shorten time to independence Clinician education and training to support quality source documentation 4
PICQ TM A Coding Quality Tool PICQ stands for Performance Indicators for Coding Quality PICQ is an auditing tool which identifies records in data sets that may be incorrectly coded PICQ measures coding accuracy by using a set of indicators PICQ is used for benchmarking across health services, hospitals and clinical coders PICQ is used for internal performance management to support: the continuous review of coding quality, and review of amended coded data quality 5
The PICQ TM Concept PICQ is designed to examine admitted patient morbidity data coded using the: International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM), and the Australian Classification of Health Interventions (ACHI) ICD-10-AM and ACHI are the national standard coding systems for admitted patient morbidity episodes in Australia and a number of other countries 6
What does a PICQ TM Indicator do? When an indicator examines a record, it analyses diagnosis and procedure codes: in combination with other codes in combination with National Health Data Dictionary (NHDD) data items in a sequence for their presence or absence for their specificity PICQ TM can be applied at any stage of data collection PAS extract Data warehouse extract PICQ TM indicators are expressed in a standard format allowing comparison of outcomes: over time between facilities between coders 7
What Input does PICQ TM Require? PICQ TM uses the following data fields extracted from a PAS or data warehouse: User episode ID Hospital ID Medical record number Date of admission Date of separation Date of birth Sex Principal diagnosis code Additional diagnosis codes Procedure (intervention) codes Care type Mode of separation DRG Days in ICU Intended length of stay Neonatal admission weight in grams Hours on mechanical ventilation Leave days Mental health legal status Coder User defined fields 8
When to Run PICQ TM PICQ can be used any time after the ICD-10-AM codes have been entered into a system Hospitals and health services run PICQ TM at desired intervals (daily, weekly, monthly ) for continuous review and amendment of coded data to provide feedback to individual coders to benchmark to identify education activities prior to sending data extracts to secondary collectors Secondary and tertiary data collectors run PICQ when: data has been received, and before data sets are finalised analysis can provide benchmark performance over time and between hospitals Researchers use PICQ analysis to measure data quality System support agencies can incorporate PICQ indicators into the clinical coding software systems to provide immediate feedback about potential errors 9
PICQ TM Indicator Degree There are four indicator degrees: F, Fatal Indicator any record found by such an indicator has been coded incorrectly by definition W1, Warning Indicator, 1% threshold records found by a warning indicator indicates that individual codes or combinations of codes or data items are likely to be incorrect W2, Warning Indicator, other records found by a warning indicator indicates that individual codes or combinations of codes or data items are likely to be incorrect (although the record is possibly correct) R, Relative Indicator records found by such an indicator are counted and expressed as a ratio of a larger (usually) group of episodes. These indicators would generally be used to assess the overall quality of coded data rather than identify individual problem records. 10
PICQ TM Indicator Type Each indicator is categorised according to the type of problem the indicator seeks to identify The indicator types are: 1. edit problem codes or code combinations that should have been prevented by basic editing, such as the use of edits incorporated in ICD-10-AM reference (or library) files 2. completeness problem codes are missing 3. redundancy problem unnecessary codes are present 4. specificity problem codes lack specificity or the incorrect code has been selected 5. sequencing problem codes are incorrectly sequenced 11
PICQ TM Reporting Numerator, Denominator and Ratio Data Set Numerator records are the cases the indicator is seeking to identify (problem records); these records are selected from the denominator records Denominator records are the cases in the data set under analysis in which the numerator records (problem records) could occur When the PICQ program processes indicators against a data set the results are expressed as a ratio of numerator to denominator Denominator Records Numerator Records 12
PICQ TM Indicator Examples Ind. Num Indicator Name Indicator Degree Indicator Type Rationale Denominator 100367 External cause code as principal diagnosis Fatal 1 Edit This indicator identifies records with an external cause code as principal diagnosis. ACS 2001 'External cause code use and sequencing' states that external cause codes must never be the principal diagnosis: the principal diagnosis is the injury, poisoning, etc. Total records containing an external cause code 100248 Admit for dialysis as only diagnosis code, without dialysis procedure code Fatal 2 Comp. "This indicator identifies records with 'admission for dialysis' as only diagnosis code but no dialysis procedure code. ACS 1404 'Admission for renal dialysis' states that the principal diagnosis for a patient admitted for a same day or over one night stay for renal dialysis should indicate the admission is for dialysis. If a patient were admitted for renal dialysis but did not undergo the procedure because of contra-indications, a code for 'procedure not carried out' should be assigned in addition or the episode cancelled, dependent on the admission policy of your State. Also run Indicators 100240 and 100241 to verify this principal diagnosis is correct for the length of stay." Total records with an admit for dialysis code as the only diagnosis code 100347 Impaired glucose regulation code with diabetes code Fatal 3 Redun. "This indicator identifies records containing both the impaired glucose regulation code and a diabetes code. The excludes notes at E09, E10, E11, E13 and E14 precludes the use of both codes in one episode of care. Indicators 100347, 101989-101995 check for IGR and the different forms of diabetes assigned in combination. Total records containing the impaired glucose regulation code 13
PICQ TM Indicator Examples Ind. Num Indicator Name Indicator Degree Indicator Type Rationale Denominator 101432 Delivery of twins without code for outcome of twin delivery Fatal 4 Spec. "This indicator identifies records containing the code for twin pregnancy and an outcome of delivery code but not one that specifies twin delivery. 'Code also the outcome of delivery' note at Delivery (O80-O84) and 'Code first the delivery' note at Z37 indicates that, for every delivery, the appropriate outcome of delivery code should be added to the mother's record. Before running this indicator, run Indicators 101407 and 100253 which check for an outcome code in records that require one. Indicators 101432 and 101433 check for outcome code appropriate to the multiple gestation code." Total records containing the twin pregnancy code and an outcome of delivery code 100386 Convalescence code as principal diagnosis followed by rehabilitation code Fatal 5 Sequ. This indicator identifies records with a convalescence code as principal diagnosis and a rehabilitation code. ACS 2104 'Rehabilitation' states that, if both rehabilitation and convalescence care are provided, the rehabilitation code should be sequenced before the convalescence code. Total records with convalescence code as principal diagnosis 14
PICQ TM Reporting Formats Standard PICQ TM reporting formats Summary report Numerator report Benchmark report User defined reports across nine filters Hospital ID Coder ID ICD Disease Chapter ICD Procedure Chapter ACS Number DRG Driven PICQ Indicator Type PICQ Indicator Degree (most widely used, i.e. fatal) PICQ Topic 15
Numerator records Standard PICQ TM Reporting Formats Summary reporting the records the indicator is seeking to identify (problem records) Denominator records the records in the data set under analysis in which the numerator records (problem records) could occur Unique numerator An absolute count of problem records Unique denominator An absolute count of denominator records; should equal the data set record count Overall quality ratio calculation Total numerator records / Total denominator records 6,768 / 114,099 = 5.9% 16
Target Sources of Improvement Opportunities Top PICQ TM indicators and top 5 Local Health Networks Top fatal indicator Top Fatal PICQ Indicators by Top 5 Local Health Networks 102020, Place of occurrence other than health service area with complication of medical or surgical care external cause code Start with 4 LHNs Top warning 1% indicator 250 200 150 216 191 146 131 20.0% 18.0% 16.0% 14.0% 12.0% 10.0% 102063, Dialysis procedure code or dependence on kidney dialysis status code without chronic kidney disease, stage 5 code Start with 2 LHNs Top warning other indicator 100 50-83 LHN 1 LHN 2 LHN 3 LHN 4 LHN 5 8.0% 6.0% 4.0% 2.0% 0.0% 100196, Noninfective gastroenteritis and colitis Numerator Count Ratio (%) Start with 2 LHNs 300 Top W2 PICQ Indicators by Top 5 Local Health Networks 20.0% 50 Top W1 PICQ Indicators by Top 5 Local Health Networks 20.0% 46 265 257 18.0% 45 44 18.0% 250 16.0% 40 16.0% 14.0% 35 14.0% 200 12.0% 30 29 28 12.0% 150 10.0% 25 25 10.0% 130 8.0% 20 8.0% 102 102 100 6.0% 15 6.0% 4.0% 10 4.0% 50 2.0% 5 2.0% - 0.0% - 0.0% LHN 1 LHN 2 LHN 3 LHN 4 LHN 5 LHN 1 LHN 2 LHN 3 LHN 4 LHN 5 Numerator Count Ratio (%) Numerator Count Ratio (%) 17
Standard PICQ TM Reporting Formats Numerator reporting Provides indicator rationale and link to Episode ID Primarily used for investigations and possible corrections Most useful when Filters are used before report generation, or Report is generated in Excel to support filtering 18
Benchmark across: Time Hospital Coder User defined field Standard PICQ TM Reporting Formats Benchmark reporting DATA SOURCE: PICQ 7.0 Result > Previous Month Result < Previous Month Result = Previous Month Not Applicable for Comparison Not Applicable Indicator Number Jul-10 Numerator Counts Jul-10 Denominator Counts Aug-10 Aug-10 Numerator Denominator Counts Counts Sep-10 Sep-10 Numerator Denominator Counts Counts Error Thres Jul-10 Aug-10 Sep-10 Rationale Type Degree hold Ratio Ratio Ratio Total Total FIN Yr YTD 100001 Pharmacotherapy session for neoplasm code as additional diagnosis when same day stay 5 W2 1 10749 0.01% 3 10146 0.03% 1 9688 0.01% 5 30,583 0.02% 100002 Pharmacotherapy session for neoplasm code as principal diagnosis when not same day stay 5 F 0% 3 5 60.00% 5 8 62.50% 3 5 60.00% 11 18 61.11% 100017 HIV disease code with a code from another phase/stage of HIV/AIDS infection 3 F 0% 3 132 2.27% #DIV/0! 2 123 1.63% 5 255 1.96% 100038 Secondary neoplasm site code without primary site code 2 F 0% 9 9281 0.10% 8 8544 0.09% 10 8269 0.12% 27 26,094 0.10% 100062 Use of 'unspecified' diagnosis codes in chapter 4 compared to use of all diagnosis codes in chapter 4 4 R 2032 14242 14.27% 1880 13768 13.65% 1804 12904 13.98% 5,716 40,914 13.97% 100076 DKA (ketoacidosis) with other than Type 1 diabetes mellitus or long term use of insulin 4 W2 21 169 12.43% 30 193 15.54% 21 172 12.21% 72 534 13.48% 100081 Use of 'other' diagnosis codes in chapter 4 compared to use of all diagnosis codes in chapter 4 4 R 1570 14242 11.02% 1473 13768 10.70% 1348 12904 10.45% 4,391 40,914 10.73% 100091 Alzheimer's disease code without dementia code 2 F 0% 7 469 1.49% 2 404 0.50% 4 407 0.98% 13 1,280 1.02% 100139 Acute tonsillitis code with tonsillectomy code 4 W1 1% 7 934 0.75% 9 863 1.04% 12 964 1.24% 28 2,761 1.01% 100161 Right/left cardiac catheterisation code with age >9 years 4 W2 117 2524 4.64% 113 2445 4.62% 95 2041 4.65% 325 7,010 4.64% 100162 Insertion of pacemaker code without insertion of electrodes code 2 W1 1% 4 302 1.32% 1 251 0.40% 1 223 0.45% 6 776 0.77% 100177 Uncomplicated asthma code with ICU hours 4 W2 23 873 2.63% 13 717 1.81% 13 697 1.87% 49 2,287 2.14% 100190 Division of abdominal adhesions code without corresponding diagnosis code 2 F 0% 25 1229 2.03% 15 1231 1.22% 8 945 0.85% 48 3,405 1.41% 100192 Dehydration code as principal diagnosis followed by gastroenteritis code 5 W2 7 655 1.07% 9 616 1.46% 4 602 0.66% 20 1,873 1.07% 100196 Noninfective gastroenteritis and colitis 4 W2 454 3718 12.21% 423 3638 11.63% 375 3283 11.42% 1,252 10,639 11.77% 100218 Concussive injury code or closed head injury code with loss of consciousness code 3 W1 1% 2 424 0.47% 4 397 1.01% 7 422 1.66% 13 1,243 1.05% 100240 Admit for dialysis code as additional diagnosis when same day or one night stay 5 W2 22 25075 0.09% 25 24618 0.10% 25 24289 0.10% 72 73,982 0.10% 100241 Admit for dialysis code as principal diagnosis when multi day stay 5 F 0% 3 25056 0.01% 5 24598 0.02% 2 24266 0.01% 10 73,920 0.01% 100244 Symptom urinary incontinence code with specific urinary incontinence code 3 F 0% 1 308 0.32% 1 308 0.32% 2 616 0.32% 100248 Admit for dialysis as only diagnosis code, without dialysis procedure code 2 F 0% 2 19697 0.01% 1 19341 0.01% 3 39,038 0.01% 100253 Delivery, possible, without outcome of delivery code 2 W2 67 808 8.29% 55 600 9.17% 45 434 10.37% 167 1,842 9.07% 100265 Pregnant state incidental code with code from chapter 15 3 W1 1% 1 9503 0.01% 1 8843 0.01% 2 18,346 0.01% 100284 Newborn affected by complication of labour and/or delivery code without code indicating effect 2 W1 1% 8 93 8.60% 13 89 14.61% 9 116 7.76% 30 298 10.07% 100331 Type of spinal cord lesion code without functional level of spinal cord lesion code 2 F 0% 3 24 12.50% 5 25 20.00% 6 38 15.79% 14 87 16.09% 100345 Burn of external body surface without body surface area (burnt) code 2 F 0% 5 248 2.02% 6 192 3.13% 1 172 0.58% 12 612 1.96% 100346 Body surface area (burnt) code without external body surface burn code 2 F 0% 2 245 0.82% #DIV/0! #DIV/0! 2 245 0.82% 100348 Poisoning code with therapeutic use external cause code 4 F 0% 24 2979 0.81% 16 2823 0.57% 16 2754 0.58% 56 8,556 0.65% 19 100368 External cause code required but not present with chapter 19 code 2 F 0% 1 16082 0.01% #DIV/0! 1 16,082 0.01% Numerator Denominator Overall N/D ratio
Target Sources of Improvement Opportunities PICQ TM Quality Ratio Example Wide range of PICQ TM Quality Ratio outcomes across Local Health Networks 5.1% to 12.6% Opportunities to target, improve 14.0% PICQ Quality Ratio (%) by Local Health Network and narrow outcome range 12.6% Identify and leverage internal best practice Review quality management systems and processes in LANs with the best outcomes Look to target education and training Clinical coders Clinicians 12.0% 10.0% 8.0% 6.0% 5.1% 5.8% 5.8% 6.1% 6.3% 6.4% 6.5% 6.8% 6.8% 6.9% 7.0% 7.3% 7.3% 7.7% 8.9% 9.1% 9.5% 10.1% 11.0% 4.0% 2.0% 0.0% LHN 1 LHN 2 LHN 3 LHN 4 LHN 5 LHN 6 LHN 7 LHN 8 LHN 9 LHN 10 LHN 11 LHN 12 LHN 13 LHN 14 LHN 15 LHN 16 LHN 17 LHN 18 LHN 19 LHN 20 20
Numerator Count PICQ TM Possibilities from a Profile User defined reports 10,000 9,000 8,000 7,000 So, what does this profile tell us? What are the opportunities that would lead to performance improvements in the clinical coding process? PICQ TM Example Analysis Over 550,000 Separation Records 6,000 5,000 4,000 3,000 2,000 1,000 0 1,137 1,829 901 1,804 950 755 3,101 812 8,846 267,347 154 88 177 Edit Completeness Redundancy Specificity Sequencing PICQ TM Indicator Type Fatal Warning 1 Warning 2 Relative 21
Numerator Count PICQ TM Possibilities from a Profile User defined reports 10,000 PICQ TM Example Analysis Over 550,000 Separation Records 9,000 8,000 7,000 6,000 5,000 4,000 Edit PICQ TM indicators typically line up with hospital system edits... Look for education and training themes, as well as, system edit developments 3,000 2,000 1,000 0 1,137 1,829 901 1,804 950 755 3,101 812 8,846 267,347 154 88 177 Edit Completeness Redundancy Specificity Sequencing PICQ TM Indicator Type Fatal Warning 1 Warning 2 Relative 22
Numerator Count PICQ TM Possibilities from a Profile User defined reports 10,000 PICQ TM Example Analysis Over 550,000 Separation Records 9,000 8,000 7,000 6,000 5,000 4,000 3,000 Fatal and Warning 1 PICQ TM indicators should be corrected... Look for education and training themes (coders and/or clinicians) 2,000 1,000 0 1,137 1,829 901 1,804 950 755 3,101 812 8,846 267,347 154 88 177 Edit Completeness Redundancy Specificity Sequencing PICQ TM Indicator Type Fatal Warning 1 Warning 2 Relative 23
Numerator Count PICQ TM Possibilities from a Profile User defined reports 10,000 PICQ TM Example Analysis Over 550,000 Separation Records 9,000 8,000 7,000 6,000 5,000 4,000 Redundancy PICQ TM indicators, look for education and training themes... your scarce clinical coding resources are taking time coding unnecessarily 3,000 2,000 1,000 0 1,137 1,829 901 1,804 950 755 3,101 812 8,846 267,347 154 88 177 Edit Completeness Redundancy Specificity Sequencing PICQ TM Indicator Type Fatal Warning 1 Warning 2 Relative 24
Numerator Count PICQ TM Possibilities from a Profile User defined reports 10,000 PICQ TM Example Analysis Over 550,000 Separation Records 9,000 8,000 7,000 6,000 Sequencing PICQ TM indicators monitor possible principal vs. additional diagnosis corrections. Sequencing supports research and DRG assignments, look for education and training themes (coders and/or clinicians) 5,000 4,000 3,000 2,000 1,000 0 1,137 1,829 901 1,804 950 755 3,101 812 8,846 267,347 154 88 177 Edit Completeness Redundancy Specificity Sequencing PICQ TM Indicator Type Fatal Warning 1 Warning 2 Relative 25
Numerator Count 10,000 9,000 8,000 7,000 6,000 PICQ TM Possibilities from a Profile User defined reports Relative Specificity PICQ TM indicators are a gold mine, dig well... Look at records that trigger multiple indicators for education and training themes (coders and/or clinicians) as well as links to corrections that can change DRG PICQ TM Example Analysis Over 550,000 Separation Records 5,000 4,000 3,000 2,000 1,000 0 1,137 1,829 901 1,804 950 755 3,101 812 8,846 267,347 154 88 177 Edit Completeness Redundancy Specificity Sequencing PICQ TM Indicator Type Fatal Warning 1 Warning 2 Relative 26
Count Target Sources of Improvement Opportunities Education / training by ICD Chapter 250,000 Adversely effects research analysis Relative Specificity Profile by ICD Chapter Total Numerator Count = 585,373 ICD-10-AM Chapter Title 4 Endocrine 10 Diseases of the respiratory system 200,000 195,433 11 Diseases of the digestive system 19 Injury, poisoning and certain other consequences of external causes 150,000 20 External causes of morbidity and mortality 100,000 Largest payback in terms of quality information for patient care and DRG assignment 50,000 50,942 46,618 39,327 38,506 30,878 24,007 21,097 20,889 19,041 17,132-14,933 13,468 12,100 10,187 10,045 9,176 5,927 2,867 2,636 20 18 10 19 11 4 14 1 13 9 5 7 2 6 3 15 16 12 17 8 ICD Chapter 164 8/11/2011 2011 Healthcare Information and Management Systems Society 27
Target Sources of Improvement Opportunities Medical records to target for likely DRG change The more R4 indicators triggered, the more likely a DRG change would occur on correction Historically, the sum of the DRG changes would be up from a cost weight perspective Target medical records that triggered 7 R4 s then, 6 R4 s then, 5 R4 s then... 50,000 45,000 Medical Record Count by Number of Multiple R4 Indicators Triggered (Input Data Set Count = 1.2 million) 46,308 40,000 35,000 30,000 27,315 Multiple R4 Triggered Medical Record Count Percentage of Data Set Records 7 10 0.00% 20,000 6 98 0.01% 15,000 5 1,621 0.13% 4 11,150 0.93% 10,000 3 27,315 2.27% 5,000 2 46,308 3.85% Totals 86,502 7.20% 10 98 25,000-1,621 7 6 5 4 3 2 Medical Record Count 11,150 28
PICQ TM Performance Example Accuracy improvements with time and rapid payback Process A hospital in Victoria with over 110,000 separations per annum PICQ TM used as part of a coding quality management system An average of 800 medical records that had triggered PICQ TM indicators were reviewed each year Outcomes Between 2.5 and 9.2% of the sampled medical records resulted in a WIES change The sample sum of WIES changes (positive and negative) ranged from $ 13k to 84k per annum Conclusions from Data The coding quality management process is driving improvements in accuracy over time (see trend line) The annual cost of PICQ TM is paid back many times each year $90,000 $80,000 $70,000 $60,000 $50,000 $40,000 $30,000 $20,000 $10,000 $0 Medical Records that Triggered PICQ Indicators Average Sample Size = 800 per Annum Percentage of Sample Records with WIES Change (%) and Resulting Dollars ($) FY05 FY06 FY07 FY08 FY09 FY10 WIES Change ($) WIES Change (%) Linear (WIES Change (%)) 10.0% 9.0% 8.0% 7.0% 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% 29
THANK YOU JOE BERRY JOE@PAVILION-HEALTH.COM 30