| Participation (behavior) [54-57,59,61,63-67,70] |
Authors reported the rates of women who have participated in breast cancer screening by mammography. Participation in breast cancer screening was reported by the authors as “participation rate” [55], “adherence” [57], “uptake” [54,63,70], “receipt” [54,56], “screening behavior” [65], “getting screened” [67], “being up to date” [64,66,67], or “screening gap closure” [59]. Instruments used to measure women’s participation were self-report questionnaires/surveys [54,55,65,70], health records [59,63,67], or both, with self-reports used to supplement missing medical records [56,57,61,64,66]. In 1 study [70], the authors differentiated participation in the breast cancer screening program, which was reported, and opportunistic screening. -
Depending on the study, participation was assessed as follows:
At short-term (up to 6 months): measured at 0.5-2 months [55], at 3 months [59,67,70], at 4 months [61], or at 6 months [54,56,57]. In Krist et al [67], a non–randomized controlled trial, measures were done after each invitation phase. At long-term (12-16 months): measured at 12 months [63,65,66], at 14 months [64], or at 16 months [61]. It must be highlighted that in Champion et al’s [66] study, participation after the 12-month window was classified as “no screening.”
|
| Intention to undergo breast cancer screening [54,55,60,68-74,76] |
Intention, as defined in the Theory of Planned Behavior [89], is assumed to capture the motivational factors that influence a behavior; it indicates how hard people are willing to try or how much effort they are planning to exert to perform the behavior. Intention assessment was reported as the rates of women with intention/willingness to undergo breast cancer screening. Reder and Kolip [70] differentiated the breast cancer screening program and opportunistic screening and reported intention to participate in the breast cancer screening program. In the randomized controlled trials, intention was assessed just after using the e-tool [54,68-74] or 7-10 days after [55]. Generally, intention was assessed through 1 or 2 items in a questionnaire/survey with anchors defined by the authors themselves to be either cross-checked [54,55,68,70-74,76] or evaluated with a Likert scale [60,69]. Intention was formulated either as a general statement [55,73] or as a more specific one and anchored to a specific time or age. Specific time(s) was/were linked to the near future either now [69], in the next 3 months [70], in the coming/next year [60,68], or at different possible times (within 1 month, 3 months, or up to 1 year) [54]. In 1 study it was evaluated for a more distant future (in the next 1-2 years) [74]. When anchored to age, women’s intention was evaluated to start or continue in their forties [71,72], or to start at 50 years [71]. We used the cumulative value of intention to assess Lee et al’s [54] and Seitz et al’s [71] tools in the meta-analysis (see Part A in Appendix S3 in Multimedia Appendix 1): intention “to get screened within 3 months to 1 year” and intention “to start or continue in the forties, or to start at 50 years old” were reported.
|
| Intention not to undergo breast cancer screening (negative intention) [69,70] |
|
| Knowledge [54,55,62-65,67,69,70,72,77] |
Knowledge was assessed about breast cancer screening (most studies) and, in some studies, about breast cancer. With Walsh et al’s [64] and Krist et al’s [67] e-tools, participants were asked to evaluate whether the tool improved their own knowledge; the authors reported rates of participants presenting different degrees of agreement [64,67,77]. With Eden et al’s [72] e-tool, women rated how much new information they learned from the e-tool on a scale from 1 (lowest) to 10 (highest), with the authors reporting median and IQR. In the remaining 7 e-tools (ie, 6 randomized controlled trials [54,55,63,65,69,70] and 1 mixed method study [62]), the instrument used to measure knowledge was more complex and heterogeneous; the instrument was a series of 5-18 items created or adapted by the authors to assess participants’ knowledge accompanied by anchors. The format of the anchors was either multichoice responses [62,63], both multichoice and numerical responses [55,69,70], true/false responses [54], or response scales of 1-4 [65]. Lee et al [54] used a validated knowledge instrument, which was revised to reflect the current American Cancer Society’s breast cancer screening guidelines. Adequate knowledge was reported as a quantitative value to reflect the number of items the participants answered correctly [54,55,62,63,65,69,70]. In some studies, the percentage of women with adequate knowledge was also reported; those were defined as women providing correct answers to at least half plus 1 of the total number of questions [55,69,70].
|
| Attitudes about breast cancer screening [55,58,69,70] |
Refers to the degree to which a woman has a favorable or unfavorable evaluation or appraisal of an undergoing breast cancer screening program [16,89]. In 4 randomized controlled trials [55,58,69,70], the instrument to measure attitudes toward undergoing breast cancer screening was adapted from Marteau et al’s [16] definition and measure. It was either 4 items [55,58,70] or 6 items from Dormandy et al’s scale [69,90]; all these items were linked to a Likert scale. In 1 of those randomized controlled trials [58], the authors reported attitudes score, with higher scores reflecting higher positive attitudes toward undergoing breast cancer screening. In the other 3 randomized controlled trials [55,69,70], Marteau et al’s [16] method was used. A predefined threshold was set to differentiate women with positive attitudes [55,70] or values [69], that is, women scoring higher than the threshold, from women with negative attitudes [55,70] or values [69], and the percentage of women with positive attitude was reported.
|
| Self-efficacy [54,72] |
Confidence in one’s ability to take action as described in the Health Belief Model [91]. Also includes the concept of perceived behavioral control in the Theory of Planned Behavior models, referring to the perceived ease or difficulty of performing the behavior and assuming to reflect past experience as well as anticipated impediments and obstacles [89]. Instruments used to measure self-efficacy were validated instruments. They were either the 8 related items from the Champion’s Health Belief Model Scale [54] or the Decision Self-Efficacy Scale [72]. The level of self-efficacy was reported. One component of the Decision Self-Efficacy scale measures self-confidence or belief in one’s ability to make decisions and participate in shared decision-making.
|
| Worry [63,67,69,71,75,78] |
Assessment of emotional reactions related to breast cancer was exclusively based on assessing changes in anxiety, worry, or fear either in general, “feeling fear and worry” [69,71,78], or specifically breast cancer [63,67,75]. In Krist et al’s [67] study, the level of fear and worry was assessed through a series of questions related to breast cancer prompted by the e-tool. The authors then evaluated whether discussion with health professionals was helpful to reduce women’s fears and worries if fears/worries were identified. For other e-tools, instruments used to measure worry were heterogeneous. It was a validated instrument: either the Lerman Breast Cancer Worry Scale [63] or adapted from the McCaul Breast Cancer Worry Scale [75]; a Likert scale was used to rate the answers, and the means of scoring was reported. In other cases, it was created by the authors themselves and was generally 1-2 questions linked to a Likert scale [69,71,78]. Seitz et al [71,78] adapted a previous instrument.
|
| Risk perceptions [54,63,71,75] |
Personal risk estimation to develop a breast cancer. It could also include “perceived susceptibility” from the Health Belief Model (ie, beliefs about the chances of experiencing a risk or getting a condition or disease) [91]. In the randomized controlled trial by Lee et al [54], risk perception was reported through “perceived susceptibility” from the HBM model. In 3 additional e-tools, all tailored-to-risk (Table 1), the authors evaluated women’s risk perceptions differently [63,71,75]. In Elkin et al’s [75], the participants were all at low-to-average risk based on their personalized risk estimates; the authors assessed the rates of women with this adequate “accurate” perception of their own risk. To evaluate the 2 other e-tools, instruments were adapted from previous studies/scales. Participants were asked to evaluate their chance of getting breast cancer as a percentage (0% being no chance and 100% being sure of getting cancer) [71], or as frequency (ie, a number out of 1000) [63,71]. The difference between the risk perception evaluated by the women itself and objective risk estimates was measured and reported by the authors, with a decrease of difference reflecting an increase in women’s accurate perception of risk.
|
|
Quality of decision
|
|
|
Regret [63,70] |
The effect of the e-tool was evaluated on decision regret [70] or on anticipated regret [63]. Decision regret describes regret associated with self-recrimination around having made a bad decision or regret associated with the knowledge that another choice would have resulted in a better outcome. Anticipated regret measure is future oriented; anticipated inaction regret was associated with engaging in protective health behaviors, while anticipated action regret was associated with nonengaging. Findings suggest that anticipated inaction regret is more strongly felt and has more reliable associations with behavioral intentions and health behaviors.
|
|
Decisional conflict [55,63,70,72,74,75] |
Personal uncertainty about which option to choose. Decisional conflict is a multidimensional outcome measuring either 4 dimensions (ie, being “SUREd”) or 5 (previous dimensions plus “effective decision-making”), and measured with the SURE scale and the Decisional Conflict Scale, respectively. These 2 instruments are validated and negatively correlated scales [92]. With the SURE scale, a score of ≤3 indicates decisional conflict and a score equal to 4 an absence of conflict; the authors reported the number of women with a decisional conflict. This instrument was used in 2 randomized controlled trials [55,70]; whereas in the study by Reder and Kolip [70], the instrument was specifically about the intention to undergo breast cancer screening, this was not the case in the study by Roberto et al’s [55] in which the instrument was more general and about “decision.” The Decisional Conflict Scale used was either the low literacy version (ie, including 10 questions with 3 response categories) [72,74] or the traditional one (ie, including 16 statements with 5 response categories) [63,75]; Decisional Conflict Scale scoring ranged from 0 (no decisional conflict) to 100 (ie, extremely high decisional conflict or poor decision process); means of scoring was reported to reflect the decisional conflict.
|
|
Informed choice [55,69,70] |
Based on Marteau et al’s [16] definition: “an informed choice to undergo a screening test occurs when an individual has a positive attitude towards undergoing a test, has relevant knowledge about the test and undergoes it. An informed choice to decline a test occurs when an individual holds a negative attitude towards undergoing a test, has relevant knowledge about the test and does not undergo it. The choices that occur when individuals do not have relevant knowledge or when their attitudes are not reflected in their behavior, are uninformed.” Following Marteau et al’s [16] definition, only 1 study [70] reported informed choice regarding participation in breast cancer screening. Informed choice was usually reported toward intention to undergo breast cancer screening [55,69,70]. “Informed choice about intention” was then reported in our meta-analysis as the measure of informed choice. Except for Roberto et al [55], results regarding informed choice were reported in the original scientific publications only for participants who had made a decision about their intention to participate, excluding those who remained undecided [69,70]. This was made to strictly follow Marteau et al’s [16] definition (ie, measure of informed choice only among those who made either a positive or a negative decision). In our meta-analysis, we reported informed choice over the total number of women for whom data related to informed choice measurement (ie, knowledge, attitude, and intention) were available.
|
| Shared decisions and communications with health professionals [64,67,72,75] |
Eden et al [72] assessed the readiness of women to make a shared decision using the Graham Preparation for Decision Making scale. A higher level of scoring indicated that the women felt the decision aid prepared them to make a shared decision with clinical providers. Among the 22 e-tools, including the 6 identified as preparing for or facilitating shared decision-making [63,64,72-75], none was evaluated against whether a shared decision-making discussion occurred. Instead, the tools were assessed by measuring women’s readiness to discuss mammography [64], whether this discussion occurred [64,75], or the characteristics of the discussion including its quality [67]. The perspective of the health professionals was also evaluated regarding the 2 latest variables [67,75].
|
| Other outcomes |
Few studies assessed outcomes related to the stage of adoption or stage of decision other than behavior or intention, such as “having made a decision” [70] or “readiness to be tested” [64]; data were not reported here. Perceived quality of life was reported in 1 study [65]. Satisfaction with the decision was not reported. Among implementation outcomes, “Reach” (ie, how the tool reached target populations) was the most reported after efficacy/effectiveness (detailed above) and through a variety of instruments.
|
