Transcript Evaluation Report to the Governor’s Task Force on Cancer

Evaluation and Implementation
of State Comprehensive Cancer
Control Plans: Evolving Lessons
APHA 2005 Annual Meeting
Epidemiology Section
Session 3187.0
12:30–2:00 PM
Monday, December 12, 2005
Assessing cancer burden:
Estimating and utilizing
prevalence
Presented by:
Judith B. Klotz, DrPH
UMDNJ-School of Public Health
Co-authors of this presentation include:
Stanley H. Weiss, MD
Xiaoling Niu, MS
Jung Y. Kim, MPH
Daniel M. Rosenblum, PhD
3
Context: Capacity and Needs
Assessment at County Level

Focus on seven NJ-CCCP priority cancers
 Breast, Cervical, Colorectal, Lung, Melanoma,
Oral/Oropharyngeal, Prostate

Need for estimates of burden of cancer in the
population
 Prevalence = number of people living with a
disease at a point in time
 Cancer prevalence estimates are useful
supplements to incidence and mortality
statistics, and help determine the level of cancer
control efforts needed
4
Uses of Prevalence Data Include
Assessing current burden of disease
 Predicting future burden of disease
 Planning of health services
 Allocation of medical resources
 Planning and administering health care
facilities
 Guiding health care research programs

5
A Limitation of Incidence and
Mortality Statistics

Adjusted rates do not reflect actual
burden of disease or number of persons
affected
6
Institute of Medicine (IOM)
2006 Report on Cancer Survivors
Over 10,000,000 prevalent cases today
in U.S.
 Dearth of coordinated clinical and
support follow-up services for patients
and their families

7
IOM Report: Follow up needs for
patients and families






Rehabilitation and quality-of-life issues
Psychological stresses
 e.g. potential for recurrence
Acute or chronic pain or other side effects from
cancer treatment
Risks of additional cancer from radiation/
chemotherapy
Needs for continuing treatment and/or screening
Insurance issues
8
Types of Prevalence

Prevalence (count): Number of people living
with the disease at a point in time

Prevalence rate: Number of prevalent cases
divided by the total population
9
Types of Prevalence, cont.

“Complete Prevalence”
 Usually preferred for cancer
 Includes all survivors, regardless of years
since diagnosis
 Rationale: long-term needs of patients and
families for medical and psychosocial services

“Limited Duration Prevalence”
 Includes those who were diagnosed within
specified number of years (e.g., 2, 5, 10, 20)
 So does NOT include those who survive after
the number of years at which follow-up is
truncated
10
Typical Sources of Prevalence Data
Population Surveys
 Estimation from combination of incidence
and survival data
 Cannot simply combine mortality and
incidence data in a particular year because
they pertain to different, specific persons:

 Most people who die in a particular year were
diagnosed in an earlier year
11
Current Availability of
Total Prevalence Estimates

Conducted by NCI for U.S. based on longest
cancer registries and complex modeling
 Connecticut Tumor Registry: since 1935
 New models developed in Italy and adapted by NCI

New SEER*Stat program
 Newly available as of August 2005 (after C/NA was
completed)
 Provides counting method for limited duration
prevalence
 Years since diagnosis depend on inception of state
cancer registry
 New utility, “COMPREV” estimates complete
prevalence from limited-duration prevalence
12
Capacity and Needs Assessment
Prevalence Estimates for Counties
Basic method for C/NA, 2003–2004
 This method was reviewed and approved by the
Evaluation Committee of Governor’s Task Force
 Use the ratio of prevalence rate to crude
incidence rate from national (NCI SEER) data,
 By specific cancer
 By gender

Apply this ratio to county-specific crude incidence
rate
 Wide variability among counties expected due to
variations in population size and demographics
13
Capacity and Needs Assessment
Prevalence Estimates for Counties
Source Data
a. Total populations for each county (by gender),
from the 2000 Census
b. Incidence counts for 1996–2000 for each county,
as provided by the NJ State Cancer Registry
These were used to calculate crude incidence rate,
separately for each gender:
Counts of new cases
Crude incidence rate =
x 100,000
Total population
14
Simplifying Assumptions

County survival rates assumed to resemble
national survival rates by gender, for each
cancer, whereas these may in fact vary

Migration in and out of counties assumed not to
affect prevalence counts, whereas migration
after diagnosis could alter the true number of
affected people still living in a given county
15
Simplifying Assumptions, cont.

Racial and ethnic distributions assumed
not to alter county survival rates, whereas
these demographic differences could
affect numbers of survivors in any county

Crude incidence is approximated by
1996–2000 data, whereas current
incidence may now differ
16
Prevalence to Incidence Ratios*
Of the prevalence to incidence ratios for
the 7 NJ-CCCP priority cancers,
Lowest ratio: Lung cancer (males) = 1.4
Highest ratio: Cervical cancer
= 17.0
Interpretation: There are about 17 times as
many living women who have been
diagnosed with cervical cancer as have been
newly diagnosed during one year.
* Ratio of national estimated complete prevalence rate to
national incidence rate
17
Calculated from NCI Data:
Prevalence/Crude Incidence Ratios
Ratios of Prevalence to Crude Incidence
Cancer Site
Males Females
All cancer combined
5.6
8.8
Breast
-11.4
Cervical
-17.0
Colorectal
5.4
7.8
Lung and bronchus
1.4
2.2
Melanoma of the skin
10.2
16.3
Oral/oropharyngeal
6.6
8.7
Prostate
7.3
-18
SEER*Stat Prevalence Estimates
for NJ and its Counties

Calculated a 20-year duration limited prevalence
 NJ State Cancer Registry began 1979, so that there
is data for more than 20 years
 Data currently available through 2003
 Used January 1, 1999 as the sample point in time
 These prevalence statistics have not yet been
published by NJ Dept of Health and Senior Services
19
SEER*Stat Prevalence Estimates
for NJ and its Counties

For long-survival cancers,
SEER*Stat count estimates were markedly lower
than C/NA complete prevalence estimates
 e.g. Limited/complete ratio for cervical cancer
State:
0.64
Counties: 0.56–0.84

Note: It is to be expected that estimates for counties
will vary markedly from each other
20
SEER*Stat Prevalence Estimates
for NJ and its Counties

For short-survival cancers,
SEER*Stat limited duration estimates were
in closer agreement with C/NA complete
prevalence estimates, both statewide and
for many counties:
 e.g. Limited/complete ratio for lung cancer
State:
1.1
Counties: 0.93–1.3
21
Ratios of Estimate Counts for
SEER*Stat Limited Prevalence to
C/NA Complete Prevalence
Cancer
Site
Lung
Colorectal
Oral
Melanoma
Breast
Cervical
Prostate
Gender
Ratio for NJ
statewide
Male
Female
Male
Female
Male
Female
Male
Female
Female
Female
Male
1.1
0.84
1.2
0.87
0.79
0.71
0.89
0.73
0.89
0.64
0.9
Range of ratios
in counties
Results discussed above are highlighted in yellow
0.93 – 1.3
0.73 – 1.1
1.1 – 1.4
0.75 – 1.0
0.63 – 1.0
0.54 – 0.85
0.63 – 1.2
0.59 – 0.9
0.77 – 0.95
0.56 – 0.84
0.77 – 0.97
22
Comparison with SEER*Stat
Estimates for NJ Counties, cont.

Gender differences in prevalent case estimates
for colorectal cancer were shown by C/NA
method but not SEER*Stat
 perhaps related to longer lifespan of women

Limitations of prevalence estimates currently
available from State Cancer Registries using
SEER*Stat
 Duration depends on year of inception of Registry
 For 15 states: less than 10 years available
23
Comparisons of Estimated Counts–
Some Examples
Cancer
Site
Cervical
Oral*
Location
New Jersey
Passaic County
Hudson County
New Jersey
Ocean County
Mercer County
SEER*Stat
Limited
Duration
Prevalence
C/NA
Complete
Prevalence
5,337
331
469
4,949
353
217
8,377
560
702
6,550
545
331
* Male + female combined
24
Comparison with SEER*Stat
Estimates for NJ Counties, cont.
Future analyses:

We anticipate using SEER’s new COMPREV
to estimate Complete prevalence from the
Limited-Duration prevalence, and then to
compare these results to the C/NA method
used in 2003–2004
25
Use of County Prevalence
Estimates to date
County cancer control planners and county
cancer coalitions have found prevalence
estimates useful for:
Estimation of relative burden of disease
among county populations of different
cancers
 Recommendations for priority actions

26
County Use of Prevalence Data in
Assessing Needs for Cancer Control
An Example:
 "The four most prevalent NJ-CCCP priority
cancers in Somerset County are breast,
prostate, colorectal cancer, and melanoma....
[and] the goals and strategies in the NJ-CCCP
that are of highest priority for Somerset County
are outlined below for each of these four
cancers.”
Source: Somerset County, Capacity and Needs Assessment
Executive Summary 2003
27
Acknowledgments and Websites

We acknowledge:
 Cancer Epidemiology Services, New Jersey
Department of Health and Senior Services:
Lisa Roché, PhD
Betsy Kohler, MS, CTR
 County Evaluators of the NJ-CCCP Capacity
and Needs Assessment
NCI SEER*Stat website:
http://srab.cancer.gov/comprev/
 Evaluation Committee website:
http://www.umdnj.edu/evalcweb/

28