ORAL ABSTRACTS: CANCER SCREENING: PATIENT DECISIONS, PREFERENCES AND SUPPORT
Scott LaJoie, PhD
University of Louisville, Department of Health Promotion and Behavioral Sciences
Method: We recruited 1208 men age 40 and older (M=56, SD=10; range=40-86, 77% white) to complete an online survey. First, participants received basic information about PSA screening and then reported whether they would get screened (yes/no). Next, participants were given detailed information about PSA screening, including pictographs depicting the number of lives it saves and the rate of false positives and overdiagnosis. Screening acceptance was then reassessed. Finally, participants were randomly assigned to 1 of 4 narrative conditions: (1) a narrative that described physical harm, (2) a narrative that described feelings of uncertainty following screening, (3) a narrative that described overdiagnosis, and (4) all 3 narratives. After this, desire for screening was assessed for a third time.
Result: After receipt of basic information, 76.8% of men wanted PSA screening, which dropped to 56.4% after the presentation of risk data, Z=-14.37, p<.001. Narratives further reduced screening acceptance to a small, but statistically significant, degree, to 52.3%, Z=-4.63, p<.001. Compared to screening acceptance just prior to the narrative, only the harm narrative and all 3 narratives together further reduced screening acceptance (both ps<.01). Neither the overdiagnosis or uncertainty narratives influenced test acceptance (both ps>.29).
Conclusion: Providing detailed information about PSA risks and benefits significantly reduced men’s interest in PSA testing. Narratives had an additional effect on test acceptance, but this effect appeared to be specific to respondents who read about physical harms associated with testing and follow-up. By contrast, uncertainty and overdiagnosis narratives had no effect on test acceptance.
Method: In a nationally representative U.S. survey, 1601 men and women age 40-70 (M=55.96, SD=8.40) considered a screening test for either prostate or breast cancer that “unquestionably does not extend life or reduce the chance of death from cancer”. The description of screening harms was randomized: Half of participants were told that harms included “unnecessary treatment,” whereas others were given a detailed list of the harms of unnecessary treatment (e.g., anxiety, financial debt, infection, hospitalization, death). Participants indicated whether they would want to get this test (yes/no), and whether they believed that the test saved lives (regardless of what they were told). Participants also reported cancer anxiety, perceived cancer risk, perceived test risk, and demographics.
Result: Overall, 43% of participants indicated that they would want to get the described screening test. 51% of participants wanted the test when the harms were “unnecessary treatment,” whereas 34% wanted the test when the harms were described in greater detail, χ2(1)=44.74, p<.001. Only 54% of the sample believed that the test did not save lives, but even of these participants, 33% stated that they wanted the screening test. Decisions to get this test were associated with cancer anxiety (r=.21, p<.001), perceived test risk (r=-.32, p<.001) and belief that the test saved lives (r=.21, p<.001).
Conclusion: Almost half of U.S. participants wanted cancer screening in the absence of any survival benefit, and approximately one-third want screening even when the test had a long list of possible harms. Reasons for wanting this screening test included cancer anxiety, low perceived riskiness of screening, and inability to accept that the test lacked benefits.
Purpose: The objective of this study was to elucidate the role of tolerance of ambiguity and response to stochastic, statistical, and evidentiary uncertainty in the valuation of benefits and harms in the context of lung cancer screening (LCS) decision making.
Method: Structured interviews were conducted among a sample of current or former smokers, aged 40-75, recruited from primary care and pulmonary practices in Philadelphia. An interview guide included prompts related to benefits, harms, response to general ambiguity and specific constructs of uncertainty (stochastic, statistical, and evidentiary) associated with LCS. Graphics were used to convey relevant outcomes from the National Lung Screening Trial including elaborations to depict statistical uncertainty (Figure). Interviews were audio-recorded, transcribed, and independently coded by 2 investigators. An inductive approach was undertaken with themes and categories allowed to emerge from iterative review of the data.
Result: The study population (n=22) was diverse in age, gender, and education with the majority of participants African American. Four salient themes emerged from the study: 1) variation was observed in tolerance for ambiguity in the context of LCS decision making, 2) tolerance for ambiguity impacts the way that people balance the various benefits and harms of LCS, 3) understanding of random, statistical and evidentiary uncertainty influenced attitudes towards LCS, and 4) provider communication was able to mitigate concerns about ambiguity. The desire to avoid or maintain ambiguity emerged as a motivating factor in decision making for lung cancer screening. Some sought to avoid ambiguity through seeking additional information from screening, additional imaging, and second opinions; “I fear what I don't know because what I know I feel like I can control.” In contrast, others sought to avoid ambiguity associated with false positive findings or overdiagnosis by declining LCS. Some were more tolerant of ambiguity, recognizing it as part of everyday life. There was general understanding of the magnitude of expected outcomes conveyed in pictographs and statistical imprecision but random uncertainty remained a difficult concept to grasp. Poor understanding of the natural history of cancer confounded the relationship between tolerance of ambiguity and LCS decisions.
Conclusion: A communication strategy that addresses the specific type of uncertainty concerning to patients and provides context and information pertaining to that concern will best support an effective shared decision making process.
Method: Internet survey data was collected from 2017 Japanese adults in their 40s to 60s who made decisions on cancer screening within the last three years (mammography: 626; colonoscopy: 1226 (male: 652; female: 574); PSA test: 165). The questions covered topics such as the importance of early detection of cancer, precision, cost and discomfort associated with the screening, their assessment of the health care provider's explanation regarding screening, and their satisfaction with their decision.
Result: Logistic regression analysis was applied to determine the factors related to those who were satisfied/confident with their decision and those who were not (Japanese/US adults). The common variable among the cancer types determining satisfaction with their decision was the health care provider's explanation (mammography: OR=2.068, p<0.01; colonoscopy: male OR=23.404, p<0.01; female OR=1.416, p<0.01; PSA test: OR=1.420, p<0.01), while confidence in their decision was related to early detection of the cancer (mammography: OR=1.302, p<0.01; colonoscopy: male OR=1.238, p<0.05, female OR=1.416,p<0.01; PSA test: OR=1.420, p<0.01). There was a common trend between Japanese and US adults regarding the factors affecting decision-making to undergo mammography: the important factors for Japanese subjects were the health care provider's explanation, precision of screening, early detection of cancer, and discomfort related with screening, while for US subjects they were the early detection of cancer, fear of getting cancer, shared decision-making, and discomfort related to screening.
Conclusion: In conclusion, comparison of decision-making for cancer screening in different countries would help improve cancer screening rates: the effectiveness of early detection of cancer should be emphasized for Japanese, while shared decision-making could potentially encourage US females to undergo screening colonoscopies, considering the differences in the factors that affect the confidence in decision-making regarding undergoing mammography and colonoscopy in the US.
Method: A two-armed randomized controlled trial funded by the National Cancer Institute (R01CA152413) was performed in twelve community and three university-based primary care practices in Metro Detroit. Participants were patients in these practices, men and women aged 50 to 75 years not current on CRC screening. Patients arriving to the practices for the scheduled visit were randomized, just prior to their face-to-face encounter with their clinician, to either a DA with interactive patient risk assessment and preference clarification tools in selecting a CRC screening test option (Intervention Arm) or to a DA that matched in content without interactive tools (Control Arm). Patient surveys were taken immediately before and after the DA use, and immediately after their clinician encounter. All web use and data collection was done in office. Primary outcome was medical record documentation of CRC screening six months after the office visit. Secondary outcomes were patient decision quality (knowledge, preference clarification, intent) obtained from the abovementioned surveys.
Result: Among 549 participants (Intervention: 275; Control: 274), mean age was 57.7 years, 56.5% were women, and 54.5% were white and 37.3% black. Baseline knowledge, attitude, self-efficacy, test preference, and intent were similar between the two arms. At six months, CRC screening rate was 36.1% (95% confidence interval: 30.5-42.2%) in the Intervention Arm and 40.5% (95% confidence interval: 34.7-46.6%) in the Control Arm (p=0.29). Knowledge, attitude, self-efficacy, test preference, and intent increased in both arms after the DA use compared to the baseline. There was no significant difference between the two arms.
Conclusion: A DA that incorporated interactive patient risk assessment and preference clarification tools did not increase the CRC screening rate at six months compared to a DA that matched in content but did not provide the interactive tools.
Method: We recruited participants from outpatient clinics at three tertiary care centers. Eligible participants were African American, aged 49-75 years, due for a CRC screening, fluent in English, and had a scheduled office visit. The intervention participants viewed an entertainment-education decision aid with culturally tailored information about CRC screening options and theory-based support in decision making (e.g., role modeling). The Ottawa Decision Support Framework, Integrated Model of Behavior, and the Edutainment Decision Aid Model guided the development of the intervention decision aid. Control participants viewed a video about hypertension, which lacked the culturally tailored information and theory-based support in decision making. Participants met with their clinician and then completed follow-up questionnaires assessing their knowledge, decisional conflict, self-advocacy, attitudes, perceived social norms, and intentions. At three months, completion of screening was assessed by chart review. Analysis included descriptive statistics and ANCOVA.
Result: On average, participants (N=89) were aged 57.6 years (standard deviation [SD]=6.9) and married or in a long-term relationship (49%). They were primarily female (68%), college educated (57%), and privately insured (75%). Viewing the culturally-tailored decision aid significantly increased participants’ knowledge of CRC screening recommendations and options (Intervention: mean ± SD=8.9±3.1 to 11.6±2.4; Control: mean ± SD=9.2±2.9 to 9.6±2.5; p-value<0.01). Intervention participants’ reported lower decisional conflict (Intervention: mean ± SD=11.0±16.7; Control: mean ± SD=39.6±27.7); p-value<0.01) and higher self-advocacy (Intervention: mean ± SD=1.6±0.3; Control: mean ± SD=1.8±0.3; p-value=0.01) compared to control participants. No significant differences were observed in participants’ attitudes (Intervention: mean ± SD=9.7±2.0 to 9.4±2.2; Control: mean ± SD=8.9±1.7 to 8.6±2.3; p-value=0.49), norms (Intervention: mean ± SD=10.5±2.2 to 10.6±1.9; Control: mean ± SD=11.6±2.0 to 10.6±2.2; p-value=0.49), or intentions (Intervention: mean ± SD=12.5±2.1 to 13.0±1.5; Control: mean ± SD=12.1±2.6 to 12.8±1.6; p-value=0.69). At three months, 23% of patients had completed a colonoscopy regardless of group assignment.
Conclusion: Designing targeted, engaging decision aids for groups that have sub-optimal screening shows promise for improving knowledge, decisional conflict, and self-advocacy. Additional research is warranted to investigate the effectiveness of these decision aids in clinical practices on patient downstream behaviors, such as repeat testing.