Trends in Colorectal Cancer Incidence Rates by Race, Age and Indices of Access to Medical Care in th

1
Trends in Colorectal Cancer Incidence Rates by
Race, Age and Indices of Access to Medical Care
in the U.S., 1995-2004

• Yongping Hao, PhD1
• Ahmedin Jemal, PhD1
• Xingyou Zhang, PhD2
• Elizabeth Ward, PhD1
• June 11, 2008
• NAACCR

1 American Cancer Society 2 American Academy of
Family Physicians
2
Outline

• Background
• Objectives
• Data sources and analyses
• Results
• Conclusions
• Limitations

3
Background

• Colorectal cancer (CRC) incidence rates decreased
  rapidly since 1998, following a longer-term
  declining trend from 1985-1995 and a 3-year
  stable trend in 1995-1998
• The long-term and recent declines in incidence
  are thought to largely reflect trends in CRC
  screening which prevents cancer through detection
  and removal of precancerous polyps
• There are disparities in CRC screening by
• race, rurality, poverty, and insurance coverage
• --- Coughlin et al 2004 Schootman et al 2006
  Meissner et al 2006

4
Background (contd.)

• Two studies evaluated trends in CRC incidence
  rates
• -Singh et al, 2003 Siegel et al, 2008 (in press)
• SEER 9 data
• Race
• Poverty

5
Objective

• Explore the disparities in
• Trends in CRC incidence rates
• Changes in CRC endoscopic screening rates
• Race
• Age group
• Indices of access to medical care
• measured by county level primary care
  physician (PCP) supply, poverty, uninsuranced
  rate, and metro status

6
Data Sources Analyses

• Incidence data,1995-2004, were from NAACCR
  (784,111 cases)
• California, Colorado, Connecticut, Florida,
  Hawaii, Idaho, Illinois, Iowa, Kentucky,
  Louisiana, Maine, Detroit, Nebraska, New Jersey,
  New York, Rhode Island, Texas, Utah, and
  Washington

7
Data Sources Analyses

• Endoscopic screening data,1995-97 and 2002-04,
  from BRFSS (336,819 respondents)
• CRC screening rates were expressed as proportion
  screening within past 5 years
• ---sigmoidoscopy or colonoscopy / proctoscopy
• Responses coded as "don't know/not sure" or
  "refused" were excluded
• Analyses were restricted to those counties met
  the minimum of 50 participants required by CDC
  for confidentiality

8
Data Sources and Analyses (contd.)

• Age
• Less than 50 years
• 50 - 64 years (recommended for CRC screening)
• 65 years and old (eligible for Medicare coverage)

9
Data Sources and Analyses (contd.)

• Race
• African Americans
• White (by sex)

10
Data Sources and Analyses (contd.)

• Primary care physician (PCP)
• Low lt 4 per 10,000
• Moderate 4 7 per 10,000
• High gt 7 per 10,000
• Area resource file

11
Data Sources and Analyses (contd.)

• Poverty
• Low lt 10
• Moderate 10 19.9
• High gt 20
• US Census 2000, available in NAACCR database

12
Data Sources and Analyses (contd.)

• Uninsured rate (under 65 years)
• Low lt 15
• Moderate 15 - 19.9
• High gt 20
• Current Population Survey

13
Data Sources and Analyses (contd.)

• Metro status
• Metro 1 - 3
• Nonmetro 4 - 9
• USDA 2003 rural urban continuum code, available
  in NAACCR database

14
Data Sources and Analyses (contd.)

• The analysis focused on 692,548 cases
• in age gt50, accounting for 82 of total cases

15
Data Sources and Analyses (contd.)

• Indices of access to medical care were linked to
  
• Incidence data
• Edoscopic screening data

16
Data Sources and Analyses (contd.)

• Incidence rates were calculated for each category
  of age, race, and indices of access to medical
  care
• Endoscopic screening rates were calculated for
  the same categories

17
Data Sources and Analyses (contd.)

• Trends in CRC incidence rates (1995-2004),
  measured by annual percent change (APC), were
  analyzed by Joinpoint regression
• Absolute changes in screening rates between
  1995-97 and 2002-04 were calculated

18
Results
19
Trends in CRC incidence rates by age and race,
1995-2004
APC Annual percent change p lt 0.05
20
Trends in CRC incidence rates by age, race, and
PCP supply, 1995-2004
21
Trends in CRC incidence rates by age, race, and
Poverty, 1995-2004
22
Trends in CRC incidence rates by age, race, and
uninsured rate, 1995-2004
23
Trends in CRC incidence rates by age, race, and
metro status, 1995-2004
24
Absolute change (AC) in CRC screening rates by
age, race, and indices of access to medical care
(age gt65)
25

• Theses results are consistent with the previous
  findings regarding the positive impact of
  Medicare coverage policy changes on CRC screening
  
• Meissner et al 2006 CDC MMWR 2006,2008 Seeff et
  al 2004 Gross et al 2006
• At least two studies found that lack of physician
  recommendation was an important contributing
  factor to the underutilization of CRC screening
  among Medicare consumers
• Klabunde et al 2005,2006

26

• These results may reflect the lack of a
  significant increase in CRC screening in areas
  with less access to medical care and the overall
  lower CRC screening rates among African American
  patients compared to white patients in this age
  group
• Cooper et al 1999 Ko et al 2005

27
Absolute change (AC) in CRC screening rates by
age, race, and indices of access to medical care
(age 50-64)
28

• Similarly, the lack of decrease in CRC incidence
  rates in 50-64 years African Americans in general
  and white men and women residing in the most
  disadvantaged areas may in part reflect lack of
  access to CRC screening and early detection
  services

29

• Several studies have shown that CRC screening
  rates are significantly lower in medically
  uninsured than insured persons, in rural than
  urban areas, and in poor than affluent areas
• Matthews et al 2005 Coughlin et al 2004
  Schootman et al 2006
• Studies have also shown that primary care
  availability to be associated with increased
  screening rates and early detection of CRC
• Roetzheim et al 1999, 2001 Rutledge et al 2006

30
Conclusions

• Individuals residing in poorer communities with
  lower access to medical care have not experienced
  the reduction in CRC incidence rates that have
  benefited more affluent communities
• This is likely explained in part by lower
  utilization of CRC screening even in older
  populations with coverage through Medicare

31
Limitations

• we used county-level indicators of access to
  medical care because of lack of individual level
  data on health insurance status, socioeconomic
  status or other measures of access to medical
  care
• Limitations of BRFSS data include variable
  response rates and reliance on self-report

32
Thanks

• Comments and questions?