Tracking Cancer Mortality and Screening Procedure Use

1
Tracking Cancer Mortality and Screening
Procedure Use

• Manon A. Boudreault, MPH and Richard J. Klein,
  MPH
• November 17, 2003

2
Screening/Early Detection

• Pap smear
• FOBT
• Lower endoscopic exam
• Mammography
• Oral exam

3
Early cancer detection proceduresby family
income level, 2000
Near poor
Age-adjusted percent
Poor
Middle/High
Mammogram Females 40
Pap test Females 18
Lower endoscopy Adults 50 (ever)
Fecal occult blood test Adults 50
Includes sigmoidoscopy, colonoscopy, proctoscopy.
I 95 confidence interval
Note Age adjusted to the 2000 standard
population. Source National Health Interview
Survey (NHIS), NCHS, CDC.
4
Cervical and breast cancer screening, by family
income level, and race/ethnicity, 2000
Hispanic
Black, not Hispanic
White, not Hispanic
Mammogram past 2 years, ages 40
Pap test, past 3 years, ages 18
2010 Target
2010 Target
Age-adjusted percent
Near poor
Poor
Middle/ High
Poor
Near poor
Middle/ High
I 95 confidence interval
Note Data are age adjusted to the 2000 standard
population. Source National Health Interview
Survey (NHIS), NCHS, CDC.
5
Colorectal cancer screening, adults age 50 and
older by family income level, and race/ethnicity,
2000
White, not Hispanic
Hispanic
Black, not Hispanic
Sigmoidoscopy (ever received)
Fecal occult blood test (past 2 years)
2010 Target
2010 Target
Age-adjusted percent
Near poor
Poor
Middle/ High
Poor
Near poor
Middle/ High
I 95 confidence interval
Note Data are age adjusted to the 2000 standard
population. Source National Health Interview
Survey (NHIS), NCHS, CDC.
6
Lower Endoscopy, Ever, 2000
Domain
Estimate
Z
Comparison group
Race
Income level
Health Insurance (Ages 50-65)
Values comparison group.
7
Cancer screening and mortality, 2000
Age-adjusted percent
Endoscopy (ever)
Pap smear past 3 yrs ages 18
Oropharyngeal annual exam (1998) ages 40
FOBT past 2 yrs
Mammogram past 2 yrs ages 40
ages 50
lower includes sigmoidoscopy, colonoscopy,
proctoscopy Note Data are age adjusted to the
2000 standard population. Source National Health
Interview Survey (NHIS), NCHS, CDC.
I 95 confidence interval
8
Colorectal Screening, 2000Preliminary Analysis
95 confidence interval Note FOBT (Fecal Occult
Blood Test), lower endoscopy includes
sigmoidoscopy, colonoscopy, and proctoscopy.
Age adjusted to the 2000 standard
population. Source National Health Interview
Survey (NHIS), NCHS, CDC.
9
Colorectal Screening, 2000Preliminary Analysis
95 confidence interval Note FOBT (Fecal Occult
Blood Test), lower endoscopy includes
sigmoidoscopy, colonoscopy, and proctoscopy.
Age adjusted to the 2000 standard
population. Source National Health Interview
Survey (NHIS), NCHS, CDC.
10
Reduction in Cancer Deaths

• Colorectal
• Oropharyngeal
• Prostate
• Melanoma

• All (malignant)
• Lung
• Breast
• Cervix uteri

11
Distribution of the HP2010 Malignant Cancer
Mortality Objectives, 2000
Remaining malignant cancer mortality
Specific HP2010 objective
Source National Vital Statistics
System-Mortality (NVSS-M), CDC, NCHS.
12
Cancer mortality by site, 2001
Age-adjusted rate per 100,000 standard population
All ( malignant)
Lung
Prostate (male)
Breast (female)
Colorectal
Cervical
Oro- pharyngeal
Melanoma
Note Age adjusted to the 2000 standard
population. Source National Vital Statistics
System-Mortality (NVSS-M), CDC, NCHS.
13
All cancer mortality by sex and race/ethnicity,
2001
Age-adjusted rate per 100,000 standard population
2010 Target
Hispanic
American Indian/ Alaska Native
Asian/ Pacific Islander
Black
White
Total
Female
Male
Not Hispanic
Note ICD-10 codes C00-C97. American Indians
or Alaska Natives, Asian or Pacific Islanders may
be of Hispanic origin. Hispanics may be of any
race. Data are age adjusted to the 2000 standard
population. Source National Vital Statistics
System Mortality (NVSS-M), NCHS, CDC.
14
Evaluation of 2010 Target Attainment
Percent of Malignant Cancer Mortality
Required Change to Meet 2010 Target
Cancer Site
EAPC
Estimated Annual Percent Change. Based on the
EAPC (1996-2000), found at www.seer.cancer.gov. No
te Data are age adjusted to the 2000 standard
population. Sources National Vital Statistics
System, NCHS, CDC National Cancer Institute,
NIH.
15
Annual Percent Change (APC) The Annual Percent
Change (APC) is calculated by fitting a least
squares regression line to the natural logarithm
of the rates using the calendar year as a
regressor variable.              
                  
rates
r
y
Ln ( r )
x
calendar year
y
mx b
100 x (em - 1)
EAPC
16
Evaluation of 2010 Target Attainment All
Malignant Cancer Mortality
Age-adjusted rate per 100,000 standard population
Actual
Projected Annual Change
(-0.9)
Required Annual Change
(-2.3)
2010 Target
0
Based on EAPC, found at www.seer.cancer.gov,
Fast Stats. The most current group of years
(1996-2000) was used for the projection. Note
Data are age adjusted to the 2000 standard
population. Sources National Vital Statistics
System, NCHS, CDC National Cancer Institute,
NIH.
17
Evaluation of 2010 Target AttainmentColorectal
Cancer
Age-adjusted rate per 100,000 standard population
Projected annual change
Required annual change
Black Population
(-0.3)
(-6.9)
White Population
(-1.4)
(-3.7)
2010 Target
0
Based on EAPC, found at www.seer.cancer.gov,
Fast Stats. The most current group of years
(1996-2000) was used for the projection. Note
Data are age adjusted to the 2000 standard
population. Sources National Vital Statistics
System, NCHS, CDC National Cancer Institute,
NIH.
18
International Classification of Diseases (ICD)
History

• Since 1900, ICD revised about
• every 10-20 years
• All revisions have had an impact on comparability
• Revisions cause statistical discontinuities in
  cause-of-death trends

19
History.

• Beginning with deaths occurring in 1999, ICD-10
  implemented
• Many systematic differences between ICD-9 and
  ICD-10
• Between 1979 and 1998, ICD-9 in use

20
Comparability Study Measure Effect of Revision

• Same death certificates coded to each ICD version
• Comparability Ratio (derived from dual
  classification)
• ICD-10 coded deaths divided by
• ICD-9 coded deaths

21
Comparability Ratio

• Di,ICD-10
• Di,ICD-9

Ci
Measures discontinuity related solely to a
revised ICD
22
Malignant Neoplasms

• Comparability ratio 1.0068
• Net number of deaths due to malignant neoplasms
  remained relatively stable across revisions
• Nevertheless,
• Substantial number of deaths shifted in and out
  of the total malignant neoplasm category and
  site-specific subcategories

23
Section Cut Adapted from
24
Shifts Within Malignant Neoplasm Category Across
Revisions

• Changes in the rule governing the selection of
  the primary site in ICD-10
• Example 1
• Lung secondary to many other cancers
  (CR.98),
• Example 2
• Two primary sites listed, no specific site
  coded
• All other and unspecified malignant neoplasms
  (CR1.1251)
• (order of entry is not used to make neoplasms
  primary or secondary)

Coded into
25
Shifts In and Out of Malignant Neoplasm Category
Across Revisions
ICD-9
ICD-10
Shift In Pneumonia Deaths Malignant
Neoplasms (fewer causes for which pneumonia is
considered a direct consequence)
Shift Out Malignant Neoplasms HIV
(any malignant neoplasm can be due to
HIV) Malignant Neoplasms In situ, benign,
and neoplasms
of unknown behavior (mass malignancy)
(mass disease of the site)
26
Selected Estimated Preliminary CRs
Preliminary comparability ratio calculated using
large sample of 1996 deaths, except for HIV
deaths (1998 deaths). Source National Vital
Statistics System-Mortality, NCHS, CDC.
27
Comparability Ratio CAUTION!!!!

• CR of 1 does not necessarily mean cause of death
  was totally unaffected by revision
• Unknown if age-, race-, sex-, or State-specific
  CRs are appropriate
• In theory CRs should be constant across
  subgroups, but in practice, differences in
  cause-of-death distribution may produce some
  variation in CRs

28
Recommendations for Analysis

• Trend data, show a break in trend line at 1999
• Use CR to measure the effect of the break
• Official recommendation dont adjust long-term
  trends using the CR

29
(No Transcript)
30
(No Transcript)
31
Comparability-Modified Deaths (or Rate)
.

• Di,ICD-9 Di,ICD-9 Ci

CM
Use to calculate percent change between 1998 and
1999
32
(No Transcript)
33
Recommendations for Analysis

• Where coding rules may have affected specific
  malignant neoplasm of interest, evaluate
• multiple-cause-of-death file
• Any mention on death certificate (Parts I II)
• rather than single underlying cause reported in
• standard mortality tables

Capture
34
http//www.cdc.gov/nchs/hphome.htm