Alternative models for Tables 3 and 4 allowing for interaction

Table 3a: Views of patients at follow-up on information booklet.

Significant factors (and p-values) from general linear model (intervention factors shown in italics) and percentages (numbers) from cross-tabulations for significant factors.

 

Factors significant

(p<0.05) in GLM (p)

Cross-tabulations of significant factors from GLM showing percentage (number) of patients

Found booklet very useful

Automatic (0.030)

67% (102) automatic vs 54% (77) interactive

Said booklet definitely told them something new

Personal (0.006)

Sex (0.002)

AMA*sex (0.035)

52% (81) of personal vs 41% (58) general

60% (55) men vs 41% (84) women

69% (35) men with no AMA vs

50% (20) men with AMA vs

43% (43) women with AMA

38% (41) women with no AMA

Thought information definitely relevant

Sex (0.030)

86% (77) men vs 71% (146) women

Possibly or definitely were able to find information easily

Meredith (attitude to information) (0.022)

 

Automatic*monitor (0.048)

91% (31) do not want information or only want good news vs

80% (202) want all information

89% (73) monitor with automatic vs

85% (45) blunter with interactive vs

79% (69) monitor with interactive vs

78% (56) blunter with automatic

Possibly or definitely overwhelmed

Newspaper read (0.001)

30% (48) tabloid vs 12% (16) broadsheet

Possibly or definitely too technical

Newspaper read (0.020)

11% (18) tabloid vs 4% (5) broadsheet

Too limited

Automatic (0.015)

36% (51) interactive vs 24% (36) automatic

Information expected but not there

Automatic

(0.045)

23% (31) interactive vs 17% (24) automatic

Information possibly or definitely changed their ideas

Sex (0.017)

47% (41) men vs 29% (58) women

Questionnaires for 325 patients were included in analyses but there were missing data from 27 (8%) to 34 (11%) of these questions. Predictor variables included in the GLMs were the three intervention factors (automatic/interactive, personal/general, anxiety management advice (AMA) or none), age and length of diagnosis (continuous variables), gender, newspaper read, Miller coping style (monitor/blunter), Meredith (attitude to information), and used a computer before (all categorical variables). Interaction terms for gender and intervention factors and coping style and intervention factors were included.
 

Table 4A: Analysis of 224 patients who said they had shown the booklet to confidant.

Significant factors (p<0.05) from a general linear model are shown with percentages and c 2 values from cross-tabulations. (Intervention factors are shown in italics).

Change in social support score

Significant factors (p<0.05) from GLM (p value)

Percentages (numbers) deteriorating

Informational

Newspaper read (0.002)

44% (48) tabloid readers vs

18% (18) broadsheet readers

Instrumental

Anxiety management advice (0.046)

27% (26) with AMA vs

13% (15) with no AMA

Emotional

Anxiety at baseline (0.029)

 

 

Miller*AMA (0.044)

 

 

 

Sex*AMA (0.028)

31% (20) with anxiety >=8 at baseline vs 10% (15) with anxiety <8 at baseline

54% (19) monitor with AMA vs

52% (17) blunter with AMA vs

49% (16) blunter with no AMA vs

46% (16) monitor with no AMA

24% (16) women with AMA vs

17% (5) men with AMA vs

16% (7) women with no AMA vs

10% (7) men with no AMA

Negative interactions

Miller*Personal information (0.028)

47% (30) monitors with personal vs

35% (19) blunters with personal vs

34% (11) blunters with general vs

19% (11) monitors with general

Social support scores were grouped as deterioration, no change, and improvement. Predictor variables included in the GLMs were the three intervention factors (automatic/interactive, personal/general, anxiety management advice or none), age and length of diagnosis (continuous variables), gender, newspaper read, Miller coping style (monitor/blunter), Meredith (attitude to information), and whether case/probable case anxiety at recruitment. Interaction terms between gender and intervention factors and Miller coping style and intervention factors were included.