Modelling attrition and nonparticipation in a longitudinal study of prostate cancer

Table 2 Comparison of respondents and drop-outs in wave 2 by their characteristics^b

Characteristic	Respondents (%)	Drop-Outs (%)	p-value^{d, c}	Attrition Rate (%)
Age at Diagnosis			0.009
40–59	33	31		36
60–69	45	40		35
70–79	22	29		45
Race			< 0.001
AA^a	47	58		43
EA^a	53	43		33
Study Phase			0.010
Post-K^a	85	79		36
Pre-K^a	15	21		46
Income			< 0.001
$0–30,000	35	53		47
$30,001–70,000	34	32		36
≥ $70,001	32	16		23
Education			< 0.001
≤ High School	42	61		47
> High School	58	39		29
REALM ^a Score			< 0.001
≤ 6th Grade	19	34		52
7th or 8th Grade	21	20		37
High School	60	46		32
Cancer Aggressiveness			< 0.001
Low	54	44		33
Intermediate	32	28		34
High	14	28		55
PPC ^a Score			0.011
Mean (SD)	3.88 (0.79)	3.75 (0.89)		38
CCI ^a			0.030
Mean (SD)	0.97 (1.41)	1.15 (1.47)		38

^aAA african american, EA european american, Post-K post-katrina, Pre-K pre-katrina, REALM rapid estimate of adult literacy in medicine, PPC patient provider communication, CCI charlson comorbidity index
^bMissing data: Income (n = 155), Education (n = 2), REALM score (n = 2), Tumor aggressiveness (n = 83), PPC score (n = 16), CCI (n = 4)
^cPearson chi-square tests for categorical variables and two-sample t-tests for continuous variables were used to assess associations between characteristics and participation status. Note that for the PPC score, the p-value for equality of variances was 0.004, thus Satterhwaite’s approximation was used; for the CCI the p-value for equality of variances was 0.2878
^dSignificant p-values at Type I error 0.05 are bolded

ISSN: 1471-2288