• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
StudyDesignPrimary Reasons for Poor Quality Rating
Chan and Vernon, 20082RCTHigh potential for bias. Study had high overall attrition and high differential attrition between the control and intervention groups.
Cronan et al., 2008 #2836}Cross-sectionalHigh potential for selection bias. Reporting of enrollment was inadequate and distinguishing between screening and diagnostic testing was not possible.
Erban et al., 20013Cross-sectionalHigh potential for confounding bias. Baseline differences were not accounted for and reporting of statistical analysis was inadequate.
Farmer et al., 20084Cross-sectionalHigh potential for selection bias. Study had low response rates and high refusal rates. Screening behaviors and frequency of routine check-ups differed between responders and nonresponders
Fisher et al., 20065Modeling studyHigh potential that the model assumptions were invalid. Indirect costs were not included, future increased costs for use of fecal occult blood tests were not included.
Fitzgibbon et al., 20076; Ferreira et al., 20057; Wolf et al., 20058RCTCointerventions were not avoided and not assessed, making it difficult to determine the actual effect of either intervention. Reporting of randomization and blinding was inadequate
Friedman et al., 20019RCTHigh potential for selection bias. Study failed to randomize subjects adequately. Serious baseline differences between the groups were highly likely but difficult to assess with information provided.
Friedman and Borum, 200710Uncontrolled Experimental studyHigh potential for selection and confounding biases. Selection methods were inadequate. Study failed to control for confounding in the statistical analysis.
Ganz et al., 200511RCTReporting was inadequate for method of randomization and whether outcome assessors were blinded. Article does not report baseline screening status or change over time. Statistical methods were inadequate and included no adjustment for baseline screening status.
Goldberg et al., 200412RCTUnable to determine potential for selection bias. Major elements of the study were not reported, including information on randomization process and allocation concealment, blinding of outcome assessors, and comparability of the groups at followup.
Green and Kelly, 200413Descriptive/CorrelationalHigh potential for selection bias. Results were potentially biased because the statistical analysis did not include important potential confounders.
Greiner et al., 200514Prospective observationalHigh potential for selection bias. Attrition rate was high, measures were not validated, and the statistical analysis did not include important potential confounders. Comparability of the groups at followup was not reported.
Harewood et al, 200215Case-controlHigh potential for selection and confounding biases. Reporting of baseline differences between the groups was inadequate and inadequate controlling for confounding in the analysis.
Honda, 200416Cross-sectionalHigh potential for selection bias. Response rate was low and the statistical analysis did not include important potential confounders.
James et al., 200217Cross-sectionalHigh potential for selection bias. Statistical methods inadequate to determine which factors are related to CRC screening adherence. Reporting inadequate.
James et al., 2008 #1369}Prospective cohortHigh potential for selection bias. Reporting of response rates and baseline differences was inadequate. Statistical methods were inadequate to determine which factors are related to outcomes.
Juon et al., 200318Retrospective cohortHigh potential for bias. Reporting was inadequate such that determining whether sample characteristics met inclusion criteria was not possible.
Lane et al., 200819RCTHigh potential for selection bias. Baseline differences of groups were not reported adequately and the statistical methods used to control for potential differences at baseline were inadequate. The attrition rate was high.
Lawsin et al., 200720Cross-sectionalHigh potential for selection bias. Outcomes measures were not validated. Reporting of statistical methods was inadequate to determine which factors were related to outcomes.
Marcus et al., 200521RCTHigh potential for selection bias. Reporting was inadequate for numerous elements: randomization process, allocation concealment, blinding, and baseline characteristics of sample. Attrition rate was high.
Matthews et al., 200522Cross-sectionalHigh potential for selection bias. Reporting was inadequate and outcomes were not validated.
Miller et al., 200523RCTHigh potential for selection bias. Reporting was inadequate. Attrition rate was high.
Patel, 200424Cross-sectionalHigh potential for selection bias. Reporting was inadequate. Potential bias at baseline but difficult to assess with information provided. Outcomes not validated and statistical methods did not adequately control for potential confounders. Attrition rate was high.
Powe et al., 200425; Powe, 200226RCTHigh potential for selection bias. Numerous elements were not reported, including randomization method and attrition rate.
Samuel et al., 200927Cross-sectionalHigh potential for selection bias. Reporting was inadequate on issues such as characteristics of sample population.
Sarfaty and Feng28Prospective cohortHigh potential for selection bias. Study had no comparison group. Reporting of measures and statistical methods was inadequate. Important potential confounders were not considered.
Teng et al., 200629Observational, cross-sectionalHigh potential for selection bias. Statistical analysis did not include important potential confounders. Outcomes measures (scales) were not validated in Chinese Americans and have not been pilot tested.
Tessaro et al., 200630Cross-sectionalHigh potential for selection bias. Reporting was inadequate, and baseline differences were not described adequately. Statistical methods were inadequate to determine which factors are related to CRC screening adherence.
Thompson et al., 200631RCTThis study was included in KQ2 as a fair quality study but received a poor quality rating as a KQ3 include because there was high potential for selection bias. Reporting was inadequate, Baseline differences were not described adequately.
Walsh et al., 200532RCTHigh potential for bias because of contamination. Randomization was done at the physician level not the patient level. No allocation concealment was done. Several elements were not reported, including blinding of outcome assessors, information on comparability of the groups at the patient level, and attrition.
Wong, 33Cross-sectionalHigh potential for measurement bias, selection bias and confounding. Reporting was inadequate. Important confounders were not accounted for in the statistical analysis.
Wolf et al., 200134Cross-sectionalHigh potential for selection bias. Analysis did not control for important potential confounding variables.
Wolf et al., 200635ObservationalHigh potential for selection bias and confounding. High potential for measurement bias (outcome measures not valid and reliable, and outcome assessors not masked). Insufficient baseline data reported. Statistical analysis did not account for potentially important confounders.
Yepes-Rios et al., 200636Cross-sectionalPotential for selection bias. Response rate was very low.

From: Appendix F, Characteristics of Studies with Poor Internal Validity

Cover of Enhancing the Use and Quality of Colorectal Cancer Screening
Enhancing the Use and Quality of Colorectal Cancer Screening.
Evidence Reports/Technology Assessments, No. 190.
Holden DJ, Harris R, Porterfield DS, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.