Table 3.

Measures of agreement for risk factors, cancer screening, and behavioral counseling among participants in SCOPE, New Jersey, 2006-2007

ItemMedical record:
Yes
Yes
No
No
Agreement
MR as gold standard
PS as gold standard
Patient survey:
Yes
No
Yes
No
Sensitivity
Specificity
Sensitivity
Specificity
Total nnnnn% (95% CI)*% (95% CI)*% (95% CI)*% (95% CI)*% (95% CI)*
Cancer-related risk factors
    Personal history of cancer72352191164196 (94-97)73 (49-83)98 (97-99)83 (57-90)97 (96-98)
    Family history of cancer72371508551781 (78-84)59 (50-67)86 (83-88)46 (38-53)91 (89-93)
    Current smoker72364221462395 (93-96)74 (64-83)98 (96-99)82 (72-89)97 (95-98)
Cancer screening
    Mammography434146915412562 (58-67)94 (89-97)45 (34-58)49 (43-54)93 (87-98)
    Pap testing30813361145561 (53-67)96 (91-98)33 (26-40)54 (43-60)90 (80-96)
    FOBT723464413050376 (73-79)5179 (76-82)2692 (89-94)
    Colorectal endoscopy321156134011283 (79-87)92 (87-96)74 (66-80)80 (73-85)90 (82-98)
    PSA testing2899637777961 (51-67)72 (64-79)51 (43-58)56 (39-63)68 (59-76)
Behavioral counseling
    Recommendation for mammography134522109776 (68-83)19 (0-28)91 (84-95)33 (0-56)82 (74-88)
    Recommendation for Pap testing613625087 (45-90)33 (0-99)96 (86-99)6089 (49-93)
    Recommendation for CRC screening2195125459868 (62-74)67 (58-87)69 (57-76)53 (43-63)80 (72-86)
    Diet or weight counseling7231858224820854 (51-58)69 (64-75)46 (41-50)43 (38-47)72 (67-82)
    Exercise counseling7231496127823553 (49-57)71 (65-77)46 (42-50)35 (31-40)79 (74-84)
    Smoking counseling7823442941 (19-52)85 (34-94)18 (9-31)35 (11-46)69 (41-88)
    Formulasabcd(a+d)/(a+b+c+d)a/(a+b)d/(c+d)a/(a+c)d/(b+d)
  • NOTE: History of cancer includes breast, cervical, colorectal or prostate cancer. Colorectal endoscopy includes sigmoidoscopy and colonoscopy procedures.

    Abbreviations: MR, medical record; PS, patient survey; 95% CI, 95% confidence interval; CRC, colorectal cancer.

  • * 95% Confidence intervals calculated based on pseudo-likelihood estimation for a hierarchical logistic regression model.

  • 95% Confidence intervals were not estimable due to small sample sizes or extreme clustering within practices.

  • In patients who reported they did not ever receive the test.