Acarbose br www ajpmonline org br McConnon

www.ajpmonline.org
McConnon et al / Am J Prev Med 2019;56(4):487−493 491 Table 2. Cancer Screening Status of People Released From Provincial Correctional Facilities and the General Population in Ontario, Canada
Cancer screening status Corrections group General population
Colorectal
Overdue on index datea
Overdue on index datea
aIndex date is the date of the admission leading to the initial release in 2010 in the corrections group or July 1, 2010 for the general population group. bRR=unadjusted relative risk. cARR=adjusted relative risk, which is the relative risk adjusted for neighborhood income quintile.
Table 3. Primary Care Encounters for People Released From Provincial Correctional Facilities in Ontario, Canada, by Cancer Screening Status
Primary care encounter
Primary care encounter
in 3 years before index datea
in 3 years after index datea
Screening status
Colorectal
Breast
a Index date is the date of the admission leading to the initial release in 2010 in the corrections group.
may be due in part to the use of administrative data in this study rather than self-reported data, which could overesti-mate cancer screening.16,17 Because differences in screening rates between this study and these U.S. studies were large, the authors propose that at least part of the difference in 
screening rates is likely real. The lower screening rate in this study is surprising given universal healthcare coverage in Ontario. This large difference and low absolute screening rates suggest that substantial barriers to screening exist for this population and that lack of universal health insurance
is not the sole barrier to screening in the U.S. context. Also of note, this study’s findings of the percentage overdue in the general population in Ontario were similar to those in provincial reports.15
Limitations
This study has several potential limitations. Although the rate of data Acarbose was high, some linkages may have been incorrect. This was likely uncommon, given the high proportion of direct or deterministic linkage. Some people in the general population group would have experienced imprisonment, such as a person released from a provincial correctional facility in 2009 but not also released in 2010. This exposure misclassifi-cation would likely have had a small effect given the large size of the general population group, and any bias would have been conservative. People eligible for screen-ing on the basis of clinical indication, such as family his-tory, were not identified, and it was not possible to differentiate tests done for screening or diagnostic pur-poses. Screening tests that took place in a hospital, out-side of Ontario, on First Nations, or in federal prison were not captured. Though these potential sources of bias may have differentially affected the corrections and general population groups, it is unlikely that they would have substantially affected the results. The breast cancer screening interval in this study was 3 years,13 whereas current provincial guidelines in Ontario recommend an interval of 2 years.18 In the context of conflicting guid-ance, the relatively long period was selected to provide a conservative estimate of testing participation. This study reports screening status between 2008 and 2010, and over the subsequent 3 years. Although the proportion of people accessing screening may have improved since these data were collected, the difference in screening rates likely persists between people in correctional facili-ties and the general population because no specific policy or program has been implemented in provincial correc-tional facilities. Finally, this study examined screening rates and did not assess for an independent association between imprisonment and screening or between imprisonment status and screening. The RR of being overdue for screening remained positive after adjusting for neighborhood income quintile, but in the corrections group there is overrepresentation of other factors that may be associated with not accessing cancer screening and for which the authors did not adjust, such as black and Indigenous race and mental illness. These unad-justed (and partially adjusted) data showing relatively low screening rates highlight the need for a focus on this population and this setting for cancer prevention activities.