Appendix 2: Description and details
of included studies
Appendix 1: Search strategies used in the review
Search strategies were developed from earlier reviews in the field,1 2 tailored to the relevant databases, and used as appropriate mainly subject headings (for example, MEDLINE, CINAHL, CancerLit) or the equivalent translated textwords (for example, EMBASE). "Explode" functions were used on all subject headings. The search strategies (and output numbers) were as follows.
(a) Cochrane consumers and communication review group specialised register (searched March 2001)
A
risk.tw
risk taking/
risk factors/
health behavior/
life style/
B
communication/
counselling/
health promotion/
health education/
patient education/
patient compliance/
genetic counselling/
C
screen.tw
mass screening/
Take (A or B) and C
(b) Medline (1985-2001)
(1) risk/ or logistic models/ or risk assessment/ or risk factors/ or risk taking/ (276698)
(2) communication/ or persuasive communication/ or counseling/ or genetic counseling/ or health promotion/or patient education/ or knowledge, attitudes, practice/ (118266)
(3) mass screening/ or genetic screening/ or neonatal screening/ or mammography/ or vaginal smears/ or occult blood/ or prostate specific antigen/ or sigmoidoscopy/ or colonoscopy/ or precancerous conditions/ (16922)
(4) 1 or 2 (385157)
(5) 4 and 3 (14172)
(6) Limit 5 to (human and english language and yr=1985-2001 and randomized controlled trial) (352)
(c) EMBASE (1985-2001)
(1) risk/ or risk assessment/ or risk benefit analysis/ or risk management/ or risk factor/ or cancer risk/ or cardiovascular risk/ or genetic risk/ or fetus risk/ or population risk/ or high risk population/ or high risk patient/ (202993)
(2) interpersonal communication/ or persuasive communication/ or verbal communication/ or counseling/ or genetic counseling/ or parent counseling/ or patient counseling/ or health education/ or health promotion/ or patient education/ or knowledge, attitudes, practice/ (70417)
(3) screening/ or screening test/ or cancer screening/ or genetic screening/ or mass screening/ or newborn screening/ or prenatal screening/ or mammography/ or vaginal smear/ or occult blood/ or prostate specific antigen/ or sigmoidoscopy/ or colonoscopy/ or precancer/ (73660)
(4) 1 or 2 (265098)
(5) 3 and 4(12816)
(6) Limit 5 to (human and english language and yr=1985-2001 and(article or journal) and randomised controlled trial) (195)
(d) CancerLit (1985 to 2001)
(1) risk/ or risk adjustment/ or risk assessment/ or risk factors/ or risk-taking/ or logistic models/ (45510)
(2) communication/ or persuasive communication/ or counseling/ or genetic counseling/ or health promotion/ or patient education/ or knowledge, attitudes, practice/ (6633)
(3) mass screening/ or genetic screening/ or neonatal screening/ or mammography/ or vaginal smears/ or occult blood/ or prostate-specific antigen/ or colonoscopy/ or sigmoidoscopy/ or precancerous conditions/ (38529)
(4) 1 or 2 (51005)
(5) 3 and 4 (6166)
(6) Limit 5 to (human and English language and yr=1985-2001 and randomized controlled trial) (216)
(e) CINAHL (1985 to 2001)
(1) risk.mp or risk assessment/ or risk factors/ or "risk identification: childbearing family (iowa nic)"/ or "risk control (iowa noc)"/ or "risk identification (iowa nic)"/ or risk taking behavior/ or pregnancy, high risk/ or cardiovascular risk factors/ (39664)
(2) communication/ or counseling/ or "counseling (iowa nic)"/ or genetic counseling/ or "genetic counseling (iowa nic)"/ or "health teaching, guidance and counseling (omaha)"/ or patient education/ or health education/ or health promotion/ or attitude to health/ or health beliefs/ or health knowledge/ or health behavior
(3) health screening/ or cancer screening/ or genetic screening/ or cervical smears/ or mammography/ or Occult Blood/ or Prostate-Specific Antigen/ colonoscopy/ or sigmoidoscopy/ or Precancerous Conditions(7501)
(4) 1 or 2 (74130)
(5) 3 and 4 (2608)
(6) Limit 5 to ( english and yr=1985-2001 and clinical trial) (26)
(f) ClinPSYC (1989 to 2001)
(1) risk analysis/ or risk perception/ or risk taking/ or risk factors/ or logistic models/ (3373)
(2) communication/ or persuasive communication/ or counseling/ or genetic counseling/ or health promotion/ or client education/ or health attitudes/ or health behavior/ or health education/ or health knowledge/
(3) screening/ or cancer screening/ or health screening/ or mammography / or screening tests/ (2503)
(4) 1 or 2 (12162)
(5) 3 and 4 (406)
(6) limit 5 to (human and english language and [follow up study or longitudinal study or prospective study or treatment outcome study or clinical trial]) (56)
(g) Science Citation Index Expanded (searched March 2002)
No limits were applied.
(1) tailored risk or
(2) tailored print or
(3) individualised risk or
(4) tailored interventions or
(5) tailored print communications or
(6) (personal* risk and screen*) or
(7) (risk communication and screening)
A manual follow up of references from key publications and journals, and of key authors was undertaken. This was informed by the output of the electronic searches, targeting the publications or authors encountered most frequently. Thus one journal was selected: Preventive Medicine (1712 articles from 1985-2002). Seven authors were selected on the basis of many publications in the field: B K Rimer(141 refs), C Lerman (117 refs), M D Schwartz (89 refs), V Champion (60 refs), M W Kreuter (52 refs), C S Skinner (30 refs) and R Bastani (28 refs). These yielded no more included studies, but provided a wealth of background information. Other prominent reviews or review protocols in the field were accessed to see if they had any further relevant work (published or unpublished).2 3 4 One article was identified that represents the closest this field has to a prospective trials register,5 documenting 11 US research programmes in the field. Citation index searches for these authors were also undertaken.
Appendix 2: Description and details of included studies
The following sections provide brief descriptions of the 13 included studies. (Please note that the reference list at the end includes further references used in the text and includes two publications arising from the same study (Lerman et al, 1995 and Schwartz et al, 1999).)
(1) Bastani et al described a tailored risk notification programme for women over 30 with a family history of breast cancer, identified via their first degree relatives diagnosed with the disease.1 Women from the same family were randomised as a unit into either a control group, or an intervention group in which each person received:
(2) Champion et al 1994 studied women aged 35 and older, with no history of breast cancer.2 They were randomised into four groups: control group, belief intervention (involving counselling based on the variables of the Health Belief Model), information intervention (information about mammography and correct screening intervals), and belief+information interventions.
Women in the second and fourth groups received counselling from graduate nursing research assistants, which entailed a discussion of individual risk factors if baseline data showed that they had low susceptibility scores (that is, low perceived risks). For the purposes of this review, results for the second and fourth (which contained the risk communication component) were combined and compared with those of the first and third groups. One year after the intervention, 117/147 (79.6%) of people receiving the belief interventions were "compliant" with recommendations for screening compared with 103/153 (67.3%) in the other groups. (NB: Women aged under 41 were regarded as compliant even if they had not had mammography.) Perceived susceptibility to breast cancer was the only belief variable not to undergo a significant change after the intervention, although this was the focus of the risk communication intervention. The method score was 16/22 (73%).
(3) Champion and Huster reported a similar study, but on women aged 40 and over.3 In addition, they examined movement across "stages of change."4 Again, no statistically significant differences in susceptibility scores were identified. The belief interventions increased mammography compliance one year after intervention (odds ratio 1.96 for the belief only group and 2.26 for the belief and information group (raw data not reported)). There was also a net movement from lower to higher stages of change which was greatest in the belief and belief and information groups. The method score was 12/22 (55%).
(4) Curry et al also used a four group design to examine the effect on mammography uptake of a risk factor questionnaire and personal risk invitation on women aged 50 and over, who were enrolled at a health maintenance organisation.5 For the purposes of this review, we compared the results of the two groups that received the questionnaire, one followed by a general risk invitation, giving a generic list of risk factors—the "control"—and the other by the personal risk invitation (including a list of personal risk factors). Uptake of mammography at one year after the intervention in the personal risk group was 162/413 (39.2%), compared with 161/428 (37.6%) in the general risk group. Participation rates were appreciably higher in the individualised group among women with a family history of breast cancer (66.7% versus 42.9% in the control group). The method score was 18/22 (82%).
(5) Hutchison et al investigated the effect on cholesterol screening uptake of a postal questionnaire appraising the risk of coronary heart disease.6 People (aged 20-69) in the intervention group were advised to go for a cholesterol test if their answer scores were above a certain level. The control group received a health questionnaire that determined whether they were at risk without identifying the risk factors as related to coronary heart disease. The rate of cholesterol screening during three months of follow up was 75/1544 (4.9%) in the intervention group compared with 27/1603 (1.7%) in the control. Specifically, screening uptake for those who met the criteria, and were thus advised to have the tests, was 45/421 (10.7%; intervention) versus 9/504 (1.8%; control). The method score was 14/22 (64%).
(6) Kreuter et al used a three group design to investigate the effectiveness of health risk assessments on facilitating a range of behaviour changes including uptake of mammography, pap smears and cholesterol tests.7 The "typical health risk assessment" intervention consisted of feedback of personal risk information, graphically and numerically. People in an "enhanced health risk assessment" group also received individually tailored behaviour change information. For the purposes of this review, we used the data from the typical assessment versus control, to assess the influence of the individualised risk communication, but the data from the enhanced assessment are valuable for further interpretation. Analysis only included people (aged 18-75) who were non-compliant at baseline but were at least contemplating changing their behaviour. In the typical health risk assessment group, 10/36 (27.8%), 24/46 (52.2%) and 19/33 (57.6%) were compliant after intervention for cholesterol, cervical, and breast cancer screening respectively. This compares with 16/40 (40%), 21/32 (65.6%) and 17/31 (54.8%) in the control group (non-significant changes). However, there were some indications that enhanced health risk assessment may have had a greater effect (compared with control) with, for example, cholesterol screening increasing to 16/30 (53.3%) after this intervention. The method score was 15/22 (68%).
(7) Lee conducted a study with US federal employees aged 40 and above as subjects.8 The intervention group received a risk appraisal for colorectal cancer that included a statement of the individual’s risk (categorised as high, moderate, or normal) compared to their peer group, as well as general facts about colorectal cancer and the importance and availability of the faecal occult blood test. The control group received a simple information letter on the availability of the faecal occult blood test at the worksite clinic. Outcomes were measured three months after the intervention. No statistically significant differences were observed between groups in terms of knowledge and beliefs regarding colorectal cancer and the faecal occult blood test (data not given). A small rise in the proportion of people in the intervention group taking the test was observed (12/139 (8.6%) versus 6/139 (4.3%) in control group). Additionally, 62.6% (87/139) in the intervention group compared with 36.2% (50/139) in the control group (odds ratio 3.18) described an intention to get a faecal occult blood test within the next year. The method score was 16/22 (73%).
(8) Lerman et al investigated the effects of individualised breast cancer counselling, involving a discussion of personal risk factors and presentation of individualised risk estimates in women aged 35 and older with a family history of the disease.9 The control group received general health counselling. Three months after the intervention, the subjects were assessed for an improvement in risk comprehension. Those in the intervention group were more likely to improve, 26% (23/90) versus 17% (19/110) (odds ratio 3.5, 95% confidence interval 1.3 to 9.5); and this was more marked among African American women. However, in both groups, about two thirds of women continued substantially to overestimate their lifetime risks. The method score was 14/22 (64%). In an accompanying paper,10 the authors also report on the impact of the risk counselling on breast-cancer specific distress and general distress. Risk counselling succeeded in significantly reducing the former (P<0.01), especially in women with less formal education. The counselling did not affect general mood distress. Schwartz determined mammography uptake in the risk counselling group compared with the general counselling group one year after intervention.11 The groups did not differ significantly at baseline or follow up. Among the less educated women, those receiving risk counselling showed reduced mammography use (odds ratio 0.44, 0.23 to 0.83). The method score for this element of the study was 13/22 (59%).
(9) Lerman et al also evaluated the impact of education and counselling (more specifically addressing knowledge and decision-making regarding BRCA1 testing) among women with a family history of breast or ovarian cancer.12 In addition to information on the benefits, limitations, and risks of testing, the educational intervention entailed a qualitative discussion of individual risk factors. Women on the waiting list served as control subjects. Perceived risk and testing intentions were measured at a one month follow up. Education succeeded in reducing risk perceptions from an average score of 2.12 to 1.88 (range 1 to 4) compared with controls (P<0.04). No significant effect of education on intent to undergo testing was found, with 57% (73/128) versus 53% (95/180) in the control group stating that they would definitely or probably take the test. The method score was 16/22 (73%).
(10) Myers et al compared the effects of two interventions on prostate specific antigen (PSA) "adherence" in a sample of African American men aged 40-70 years.10 The minimal intervention group ("control" for the purposes of this review) received a letter of invitation for free screening, as well as reminder calls and letters. The enhanced intervention group received in addition a culturally sensitive educational booklet and a form with tailored (personal) risk factors and symptoms, based on information given by the men in a baseline survey. Fifty one per cent (98/192) of the risk intervention group were defined as "adherers" after one year follow up, compared with 29% (64/221) in the control group. The method score was 15/22 (68%).
(11) Rimer et al investigated the effects of tailored print and counselling compared with usual care on mammography uptake as well as perceptions of breast cancer risk, with women members of a health maintenance organisation aged in their 40s and 50s.14 The tailored print intervention included a section on personal risk of developing breast cancer in the next ten years presented graphically and numerically. Screening uptake one year after in the tailored print group was 52% (204/392) compared with 56% (231/412) in the control group and 28% in the tailored print group had an accurate risk perception (defined as not overestimating risk) compared with 25% in the control group. People in the group that received tailored print reinforced with tailored counselling fared better in terms of accurate risk perception (42%) and were also found to have higher mammography rates of 61% (197/323). The method score was 18/22 (82%).
(12) Saywell et al compared the cost effectiveness of five interventions to increase mammography screening.15 For the purposes of this review, the combined results of the telephone and in-person counselling groups (in which personal susceptibility was addressed depending on a woman’s initial response) were compared with the control "no counselling" group. One month after the intervention, the mammography "compliance" rates were 29% (68/237) in the combined intervention group, counselled either by telephone or in person, compared with 18.2% (20/110) in the control group. The method score was 12/22 (55%).
(13) Skinner et al examined the effects of tailored letters (addressing personal risk status, as well as beliefs and barriers, and taking into account people’s stage of change) compared with standardised letters, on recall and readership of information, and mammography stage movement.16 The tailored letters were framed to sound as if they were meant for general audiences, not specifically for the individual concerned, but were still based on risk calculation for the individual woman. Women (aged 40-65) who received tailored letters were more likely to remember them (P<0.05). However no significant main effects were found for stage movement by letter type. Forty four per cent (33/76) of the tailored letter recipients who were due for screening at baseline had a mammogram, compared with 31% (24/76) in the standard letter group. The intervention was most effective for African American and low income (<$26 000 annual income) women. The method score was 15/22 (68%).