Transcript Slides

The Role of the HCC Cancer Registry
in Facilitating Cancer Research
Linda Cope, CTR
HCC Registry Coordinator
[email protected]
Commission on Cancer
DHEC’s SC Central Cancer Registry
CoC facilities collect data in standardized codes
and report to NCDB.
SC Law: CHAPTER 35; SECTION 44-35-5.
Makes Cancer reportable to DHEC’s Central
Cancer Registry.
The MUSC/HCC Cancer Registry
5,340 new cases abstracted during FY2015
10 Staff members (7.4FTE)
Certified Tumor Registrars
One full time Follow Up specialist
MUSC/HCC Cancer Registry
• Commission on Cancer is curating a
nationwide, decades-long big data project
which requires patient data from all its
facilities to be housed in the National Cancer
Data Base (NCDB). More from Shai
• Cancer Registry identifies all patients with a
malignancy at MUSC.
• Follows analytic patients annually
Patients added to MUSC/HCC Cancer Registry
FY15
Diagnosed and or rec'd all
or part of 1st course of
therapy
Consult only
13%
13%
16%
Relationship to MUSC
58%
At MUSC after recurrence
or persistent disease
All other cases-h/o,
diagnosed here referred
back, etc.
Currently following >17,000 cancer survivors
The Commission on Cancer
Patient data
National Cancer Data Base
Rapid Quality System
Accuracy requirements
Coding instructions
Cancer
registry
Accreditation
MUSC
RQRS
Currently CoC requires breast and colon cases
Abstracts data concurrently with treatment
Clinically relevant as check for standard therapy.
Data available sooner
The MUSC/HCC Cancer Registry:
what information is captured?
Patient
presents
with
history of
cancer,
we only
follow
Patient
presents
for
consult
only
Patient
only gets
diagnosis
here and
is then
referred
out
Patient presents
with recurrent or
refractory disease
after first course
was given
elsewhere
A basic census is available, with some demographic
information, type of cancer, date of diagnosis, and
date of first contact with MUSC
Patient receives
some or all of first
course at MUSC:
now “ours” for
purposes of
abstracting and
following
Full abstract with
data fields as
shown next slides
*Outcomes tracked
Data fields for all full abstracts
Patient Demographics
Cancer Disease Information
Staging Information
Treatment
Outcomes
Patient Identifiers behind IRB wall
Name
Medical Record Number
Address
Zip code
County
Phone
Secondary contact
Date of birth (age at diagnosis)
Date of death (if applicable)
Sex
Race
Spanish origin
Tobacco history
Alcohol history
Data fields for all full abstracts
Patient Demographics
Cancer Disease Information
Staging Information
Treatment
Outcomes
Class of case
Site
Sequence
Histology (ICD-0)
Behavior
Grade
Laterality
Date of initial diagnosis
Data fields for all full abstracts
Patient Demographics
Cancer Disease Information
Staging Information
Treatment
Outcomes
Tumor size
Tumor extension
T eval method
Regional nodes examined
Regional nodes positive
N eval method
Mets at diagnosis
M eval method
Derived TNM stage
AJCC Clinical stage
AJCC Pathologic stage
Data fields for all full abstracts
Patient Demographics
Cancer Disease Information
Staging Information
Treatment
Outcomes
Dates and Specific Type
Biopsy
Surgery
Chemotherapy
Hormonal therapy
Immunotherapy
Other
Surgical margins
Sequence of systemic vs surgery
Sequence of radiation vs surgery
Where treatment happened
Data fields for all full abstracts
Patient Demographics
Cancer Disease Information
Staging Information
Treatment
Outcomes
Follow up is annual
Date of last contact
Disease status
Date of first recurrence
Type of first recurrence
Second treatment course
Survival analysis
Kaplan-Meier stratified by
stage
treatment
etc.
Specialized Data Fields
• Site Specific Data Fields
– Defined by Commission on Cancer
– Biomarkers
– Site specific prognostic factors
• Custom Data Fields
– Defined by individual registry, usually for
prospective projects
The MUSC/HCC Registry:
how do I request data?
The MUSC/HCC Cancer Registry
DATA REQUESTS
120
2013 (n=86)
100
2014 (n=150)
2015 (n=205)
80
60
40
20
0
Clincial Trials
Physician
Primary
Investigator
Tumor Bank
Institutional
Planning
Student
Pastoral Care
The Cancer Registry:
what kind of research can I do?
• Hospital registries pool their data in the
National Cancer Data Base, so you can design
and power a study based on a huge number of
patients and limited number of data fields OR
a smaller (local) number of patients with
much deeper data.
• We’ll look at one of each as examples.
The big one: NCDB Participant User File
• Includes many of the same data fields that Linda just
explained. It does NOT include some if they have been
determined to be insufficiently reliable at the national
level for various reasons. Examples: recurrence, tobacco
use, exact chemo regimens
• Includes some additional fields, derived and assigned by
the Commission on Cancer rather than being directly
coded by CTRs: education level, income, distance from
facility
• Well-suited to projects about a national research
question
• Keep in mind: demographics>>stage at diagnosis>>first
course of treatment>>outcome
MUSC’s First NCDB PUF Project
• Research problem: In 2004, two landmark papers were
published and recommended trimodal therapy for
advanced head and neck cancers. Since then, survival
rates have increased nationally and trimodal therapy
has become more common at MUSC. No broad study
had been conducted to evaluate national rates of
adherence to the 2004 recommendations. At the same
time, an epidemic of HPV+ OPSCC with relatively good
outcomes appeared in the USA.
• Question: were increased survival rates in head and
neck cancer due to change in the population or change
in the treatment or both?
Percent of patients receiving
indicated treatment by year
A
8
7
6
C
14
5
4
3
2
1
0
10
8
6
4
2
0
C
S
SC
50
45
40
35
30
25
20
15
10
5
0
R
RC
Percent of patients receiving
indicated treatment by year
none
Percent of patients receiving
indicated treatment by year
Percent of patients receiving
indicated treatment by year
Figure 1: treatment trends
B
D
35
SR
SRC
12
30
25
20
15
10
5
0
Figure 2: survival trends by treatment
2001-2004
70
2005-2008
60
50
40
***
30
20
10
0
Total
Stage IV
*** p <0.001
** p<0.01
B
80
70
2001-2004
2005-2008
C
***
60
**
***
50
40
30
***
20
10
0
none
C
R
RC
S
SC
SR SRC
Percent survival at 5 years
80
Percent survival at 5 years
Percent survival at 5 years
A
80
2001-2004
70
2005-2008
60
***
50
40
30
20
10
0
Adjuvant
SRC
Rise in OP cancers seen in
the aggregate Stage IV
group over time (as
expected)
White-Gilbertson S, et. al, J Registry Manag.
2015 Winter;42(4):146-51
No rise in OP cancers seen
in Stage IV group treated
with trimodal therapy
Geographic distribution of
trimodal therapy widened
over time
White-Gilbertson S, et. al, J Registry Manag.
2015 Winter;42(4):146-51
Figure 2: survival trends by treatment
2001-2004
70
2005-2008
60
50
40
***
30
20
10
0
Total
Stage IV
*** p <0.001
** p<0.01
B
80
70
2001-2004
2005-2008
C
***
60
**
***
50
40
30
***
20
10
0
none
C
R
RC
S
SC
SR SRC
Percent survival at 5 years
80
Percent survival at 5 years
Percent survival at 5 years
A
80
2001-2004
70
2005-2008
60
***
50
40
30
20
10
0
Adjuvant
SRC
HCC Registry Data Project
• Research problem: presentation with late stage breast
cancer has been linked to poor insurance status, although
results are mixed on the difference between lack of
insurance and Medicaid, and this is a difficult thing to
analyze in the NCDB due to typical abstraction workflow
and insurance changes specific to breast cancer diagnoses.
In addition, poor insurance is expected to impact screening
practices, but this is not captured in registry databases,
although we capture it locally for specific studies.
• Question: Would real-time abstracting allow us to track the
relationship between insurance, method of cancer
detection, and stage at diagnosis? If so, we hypothesized
that lack of insurance would predict a poorer disease
course from the beginning.
Figure 1. Insurance vs. Stage with National Data
A 100%
B 100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
IV
III
II
I
in
situ
none
n
%
11188
1.99%
medicaid private medicare
30908
5.49%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
other
314548 201554 5051
55.85% 35.78% 0.90%
IV
III
II
I
in
situ
none
n
%
14626
2.29%
medicaid private medicare
42083
6.60%
other
333416 241130 6775
52.26% 37.79% 1.06%
Figure 2. Insurance vs. Stage with Local Data
p=0.013
ns
ns
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
IV
III
II
I
in
situ
none
n
%
23
5.90%
medicaid
18
4.62%
private
177
45.38%
medicare
157
40.26%
veteran
15
3.85%
Figure 3. Insurance vs. Stage after Exclusions
p=0.027
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Stage IV
Stage III
Stage II
Stage I
insured
not insured
Figure 4: Testing the Set
p<0.001
ns
A 100%
B
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
65
and
over
40-64
insured
not insured
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
unknown
triple neg
ER/PR(-) HER2 (+)
ER/PR (+) HER2(-)
ER/PR (+) HER2 (+)
insured
not insured
Figure 5: Testing the Hypothesis
p=0.010
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
palpation
mammography
insured
not insured
Questions?