Adult Comorbidity Evaluation27 ACE27

1
Adult Co-morbidity Evaluation-27 (ACE-27)

• Presentation to the
• National Health Service
• Information Authority
• Risk Stratification Seminar
• March 7, 2003
• Jay F. Piccirillo, MD, FACS
• Washington University School of Medicine
• St. Louis, Missouri

2
Overview of Co-Morbidity Research
3
Co-Morbidity Education Program

• As part of a NCI-sponsored cancer education
  grant, certified tumour registrars taught to code
  co-morbidity
• Entire education program lasted 10 hours
• Training video
• The Whole Picture Coding Co-morbidity
• Training manual, documentation book, and 55
  clinical examples
• Johnston et al J Registry Management
  200128125-131

4
Nationwide Co-Morbidity Network
5
Videoclip
The Whole Picture Coding Co-morbidity
6
Adult Co-Morbidity Evaluation-27

• 27-item co-morbidity index for patients with
  cancer
• Developed through modification of the
  Kaplan-Feinstein Comorbidity Index
• Modifications were made through discussions with
  clinical experts and a review of the literature
• Completed by health care professionals
• Also validated in Chronic Obstructive Pulmonary
  Disease
• J. Chronic Disease 1974 27387-404

7
Medical Record Approach

• Co-morbidity severity can be assigned to a
  majority of patients within tumour registry
• Very accurate assessment of co-morbidity
• Co-morbidity coding added approximately
  3 additional work effort

8
(No Transcript)
9
ExampleCongestive Heart Failure

• Mild Exertional or paroxysmal dyspnea which has
  responded to treatment
• Moderate Hospitalized more than six months ago
• Severe Hospitalized within last 6 months or
  ejection fraction lt 20

10
Overall Co-Morbidity ScoreNone, Mild, Moderate,
or Severe

• Algorithm developed by Kaplan and Feinstein
• Highest ranked single ailment
• In cases where two or more Moderate ailments
  occur in different organ systems, the Overall
  Co-Morbidity Score should be designated as Severe

11
Example
12
Example
13
ACE-27 On-Line Form
http//oto.wustl.edu/clinepi/calc.html
14
Burden of Coding Co-Morbidity

• Amount of time required to abstract complete
  medical record, including co-morbidity
• Before training program, cancer registrar
  estimated time required to abstract complete
  medical record
• After training program, cancer registrar
  estimated time required to abstract complete
  medical record, including co-morbidity

15
Abstraction Time
Maximum
Abstraction Time (mins)
Median
Minimum
With co-morbidity
Without co-morbidity
16
Qualitative Assessment of Coding Co-morbidity

• Coding co-morbidity is no problem!
• We are already reviewing the medical record for
  cancer information
• Many of us had been collecting this information
  already as open-text, now we have a way to
  collect it and use it in our reports
• The education program is excellent

17
Reliability

• Medical record review of 190 patients
• Two reviewers
• Reviewer A first-year medical student
• reviewed 190 charts
• Reviewer B trained research assistant
• Reviewed 190 charts and re-reviewed 112 charts 5
  months later
• Weighted kappa statistic used to measure inter
  and intraobserver variability

18
Reliability

• Interobserver variability
• 0.80 (95CI 0.72 to 0.88)
• Intraobserver variability
• 0.93 (95 CI 0.88 to 0.98)
• Interpretation of Kappa
• 0.61 -- 0.80 Substantial Agreement
• 0.81 1.0 Almost Perfect
• Landis Koch, 1977

19
Data Collection

• Since 1999, 9,092 newly diagnosed patients with
  cancer have been enrolled
• 600-800 new patients are enrolled each month
• Adult Co-morbidity Evaluation-27 (ACE-27)
• Comorbid health has been linked to standard data
  elements contained in tumour registry

20
Statistical Analysis

• To develop and validate cancer-specific case-mix
  and risk adjustment models that incorporate
  comorbid health information and are
  methodologically sound in their development and
  statistically rigorous in their validation

21
All SitesN9092
22
Impact of Co-Morbidity on SurvivalAll
SitesN9092
None
Mild
Moderate
Severe
Log Rank ?2 379.24, p lt 0.0001
23
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
Death All Sites (N9092)
24
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
Death All Sites (N9092)
25
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
Death All Sites (N9092)
26
Prostate CancerN 1457
27
Impact of Co-Morbidity on SurvivalProstate
CancerN1457
None
Mild
Moderate
Severe
Log Rank ?2 108.82, plt 0.0001
28
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
Death Prostate Cancer (N1457)
29
Breast CancerN1397
30
Impact of Co-Morbidity on SurvivalBreast
CancerN1397
None
Mild
Moderate
Severe
Log Rank ?2 29.65, plt 0.0001
31
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
Death Breast Cancer (N1397)
32
Lung (NSCLC) CancerN1196
33
Impact of Co-Morbidity on SurvivalLung (NSCLC)
CancerN1196
Moderate
None
Severe
Mild
Log Rank ?2 7.91, p 0.0478
34
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
DeathLung (NSCLC) Cancer (N1196)
35
Colorectal CancerN 1123
36
Impact of Co-Morbidity on SurvivalColorectal
CancerN1123
None
Mild
Moderate
Severe
Log Rank ?2 33.34, plt 0.0001
37
Results of Cox-Proportional Hazards
ModelingImpact of Co-morbidity on the Risk of
Death Colorectal Cancer (N1123)
38
UK Experience

• Pilot Project January 2002 to June 2002
• South Tees
• Royal Orthopaedic Hospital Birmingham
• Christie Hospital, Manchester

• Aims of Pilot Project
• Skills required
• Retrospective collection
• Process of collection
• Who, how, when
• Lessons learned
• Time burden
• Perform validation checks
• Ease of use

39
Time to Collect

• South Tees for Head and Neck Patients
• Patient-based questionnaire took patients 8.3
  minutes
• Doctors performing retrospective review 16.8
  minutes
• Royal Orthopaedic Hospital for Sarcoma Patients
• No time reported
• Christie Hospital, Manchester for Women with
  Endometrial Cancer
• 5-10 minutes

40
Problems Encountered

• Various co-morbidities not included
• Laboratory values not in UK units conversion
  mandatory
• Renal system has extended definitions confusing
• Pancreas co-morbidity form varies from coding
  book
• Differences in terminology (e.g., s/p instead
  of previous )

41
Omitted Co-Morbidities

• Valvular Heart Disease
• Thyrotoxicosis and hypothyroidism
• Epilepsy
• Neurofibromatosis
• Osteoporosis

42
Additional Feedback from Users

• South Tees for Head and Neck Patients
• Very positive
• Comorbidity added to presentations and
  publications
• Royal Orthopaedic Hospital
• ACE-27 is easy to use
• Training needed to be more in-depth
• Christie Hospital
• Relationship between co-morbidity and survival
  significant
• ACE-27 has important omissions and must be
  adapted to UK

43
Cancer Dataset Project
Pilot Lessons Learned Report Version 5a
Co-morbidity http//www.nhsia.nhs.uk/cancer/pages
/dataset/docs/cdp_lessons_learned_comorbidity.pdf
44
Cancer Prognostics

• The goal of this project to develop an
  interactive prognostigram program based on the
  new prognostic models
• The prognostigram program creates individualized
  survival curves based on the Cox Proportional
  Hazards model of survival data from Barnes-Jewish
  Hospital (BJH) Oncology Data Services (ODS) and
  SEERStat

45

• BJH ODS has been collecting co-morbid health
  status information since 1995
• To date, for over 25,000 patients
• SEERStat does not
• We determined adjusted hazard ratios for
  co-morbidity from the BJH ODS database
• Co-morbidity-adjusted SEER survival curves are
  generated which take into account the impact of
  co-morbid health information

46
Example

• 55-year old woman with newly diagnosed regional
  breast cancer
• Based on SEER data, the observed overall 3-year
  survival rate is 83
• Expected 3-year survival rate could vary from
• 90 for women with no co-morbidity
• 75 for women with severe co-morbidity
• This 15 difference is both clinically impressive
  and statistically significant

47
Prognostigram Program
48
Survival According to Mode of Therapy Regional
Breast Cancer
Chemotherapy
CURRENT SITUATION Recommendations based on
composite results
Surgery
RadiationTherapy
Survival Rate
No Treatment
Survival Duration
FUTURE REALITYTailored individual therapy
Radiation Therapy
No Treatment
Surgery
Chemotherapy
49
Conclusions

• Co-morbidity is important in the selection of
  treatment, prognosis, and evaluation of quality
  of care
• The ACE-27 is a valid instrument to collect
  co-morbid information
• Web-based program exists to train cancer
  registrars and other health professionals to code
  co-morbidity

50
Conclusions

• Continued exclusion of co-morbidity impedes the
  scientific study of cancer and the humanistic
  care of patients
• Co-morbidity should be added as a required data
  element to hospital-based and central cancer
  registries

51
Clinical Outcomes Research Web Site

• http//oto.wustl.edu/clinepi/

52
Web-Based Co-morbidity Education Program

• http//cancercomorbidity.wustl.edu/index.html