In this study, we examined the association between neighborhood socioeconomic status (defined by percent of households below the federal poverty level) and use of colonoscopy in an insured population. A unique characteristic of the population was that, within each study site, subjects were in the same health plan and served by the same clinical provider network, and thus the ability to acess a usual place of care. The health systems included in this study provide colonoscopy as a covered benefit to members and also had systems to encourage members to use preventive health services. These characteristics of the healthcare environment would be expected to mitigate barriers to the use of colonoscopy for persons in low-socioeconomic status and thus lessen disparities.
We found significant socioeconomic differences in the use of colonoscopy. Persons residing in the lowest SES neighborhoods were 16% less likely to undergo any colonoscopy relative to those in the highest SES neighborhoods. This association was even stronger (30%) when only screening colonoscopy was considered. Socioeconomic differences in colonoscopy use observed in the general population could be attributed to the fact that people receive care from diverse clinical provider networks and have differing types of health insurance coverage 
. However, our results suggest that simply having health insurance and a usual place of care are not enough to eliminate socioeconomic disparities in the use of colonoscopy use, and that other factors related to poverty limit or restrict colonoscopy use. Elimination of disparities concerning colorectal cancer for socioeconomically disadvantaged populations will require measures that also address other economic, social and cultural barriers to receipt of health care services.
Our findings, together with existing research, suggest the need for effective patient navigation or outreach programs in integrated healthcare delivery systems to address disparities in colonoscopy use. Also, performance incentives based on Healthcare Effectiveness Data and Information Set (HEDIS) measures, as practiced in some health care systems, could be effective means to address the disparities we found 
. However, the effectiveness of such programs in eliminating socioeconomic disparities in colonoscopy use among people receiving their health care under the auspices of healthcare delivery systems is unproven and needs to be studied.
There are few published studies on area-level variations in the use of colonoscopy; very few, if any, of such studies have been conducted within integrated healthcare delivery system settings. Our findings, however, are consistent with studies on Medicare populations showing disparities based on individual-level measures of socioeconomic status 
. Klabunde et al., using 2008 US National Health Interview Survey data, found that about 34% of screening-eligible adults in families at ≤200% of federal poverty level had colonoscopy compared to 58% among those at ≥500% of federal poverty level, a 1.7-fold difference 
. A study using Missouri Behavioral Risk Factor Surveillance Survey data also reported that the use of colorectal cancer screening varied across zip-code areas as well as by zip-code poverty levels; socioeconomic differences remained even after adjustment for health insurance type and having a primary care provider 
. Our study examined socioeconomic differences in colonoscopy use in a smaller area of aggregation and found both geographic and socioeconomic variations.
The population we studied was comprised predominantly of employed persons from varying socioeconomic backgrounds. Socioeconomic status is a predictor of both where people live and colorectal cancer testing. Therefore, the findings from this study are a reflection of the socioeconomic diversity among the members of the respective health care systems, which might influence screening colonoscopy use in several possible ways in integrated healthcare delivery systems. People from low-socioeconomic groups may be late adopters 
of colonoscopy which has been used increasingly for routine screening in recent years, and they may have greater ambivalence about the balance of risks and benefits associated with the test. Factors related to poverty such as resource deprivation as posited in the Deprivation-Amplification hypothesis 
, cultural barriers, and difficulty navigating healthcare systems, may act separately or together to restrict access to colonoscopy 
. The need for transportation to and from the procedure may disproportionately impact those from lower socioeconomic groups, despite having health insurance. People in low socioeconomic status may also experience a greater negative impact from the time and preparation required for colonoscopy and the burden of taking time off from work.
Although the enrollees were insured, the health plans provide a variety of insurance programs with varying levels of coverage for colonoscopy. The burden of out-of-pocket expenses may disproportionally impact low-income people in the health plans studied. The co-pay for colonoscopy in the health plans across all coverage types during the study period was between $0 and $200. A study of 106 health plans in the United States found that out-of-pocket costs of $300 or greater negatively affect colonoscopy use 
. This suggests that potential differences in co-pay among study subjects are unlikely to explain our findings. Further, analyses stratified on health plan or age confirmed the results.
There may also be socioeconomic differences in the patient-physician communication 
around colorectal cancer screening, including differences in physician recommendation for colonoscopy, that may account for some of the differences observed. It is also possible that people of a lower socioeconomic status may experience higher levels of mistrust of the medical care system and may have greater difficulties gaining access to health care systems despite having health insurance 
. Other barriers may include embarrassment, lack of knowledge and cultural factors 
. Analyses of these potential barriers for insured populations are beyond the scope of our study, but warrant further investigation.
Although we were not able to determine if our findings were solely the result of patient, provider or healthcare system factors, the findings do suggest the need to pay greater attention to the preventive care needs of all people who reside in socioeconomically deprived neighborhoods regardless of whether or not they have health insurance. Area-based socioeconomic measures are readily accessible and can be utilized to guide the implementation of patient navigator programs and reminder systems 
Our study has other limitations. We relied on codes in electronic administrative and clinical databases to ascertain colonoscopy utilization and did not have precise measurements of screening colonoscopy. This might have led to a non-differential misclassification of the outcome, thus attenuating differences. A more accurate measurement may have found larger socioeconomic disparities in colonoscopy use.
We did not have individual-level measures of socioeconomic status determinants such as education, income or occupation. Therefore, the observed neighborhood effects cannot be interpreted as being independent of individual-level socioeconomic factors. However, given the challenges of collecting information on individual-level socioeconomic data, our findings reinforce the value of area-level socioeconomic data as being a suitable approach for assessing socioeconomic disparities in colorectal cancer screening. Further, while neighborhood poverty level may not fully reflect the poverty level of individuals within an area and its effect on their use of colonoscopy, the contextual factors captured by neighborhood measures provide information beyond characteristics of individuals alone. Prior research also shows that neighborhood socioeconomic measures have similar predictive power as individual measures 
. Another limitation was that over one-half of the subjects had missing data on race. We were therefore unable to account for potential confounding by race on the associations studied. However, prior research suggests that inclusion of race in the analyses would not substantially alter our results 
Finally, we did not follow subjects for 10 years, which is the recommended interval for screening colonoscopy. This may have resulted in an underestimation of the true socioeconomic effect if subjects from high-SES neighborhoods had continued to have higher rates of colonoscopy use past the 8 years of follow-up on this study.
In conclusion, this study found that, among insured persons receiving care in integrated health care delivery systems, those residing in poor neighborhoods were less likely to have had a colonoscopy compared to persons in high-SES neighborhoods, despite receiving care from a common clinical provider network. Therefore, providing health insurance or even free colonoscopy services, a sound public health policy, may not eliminate socioeconomic disparities in colonoscopy use without attention to other barriers. Future studies of financial and non-financial barriers to colonoscopy use are needed to identify effective approaches to eliminate disparities in colonoscopy use in insured populations.