- Email: [email protected]
The cognitive structuring of patient delay in breast cancer
ARTICLE IN PRESS
Social Science & Medicine 63 (2006) 3137–3149 www.elsevier.com/locate/socscimed
The cognitive structuring of patient delay in breast cancer Noreen C. Facionea,, Peter A. Facioneb a
University of California San Francisco, USA b Loyola University Chicago, USA Available online 25 September 2006
Abstract The reasons women give for delaying diagnosis of breast cancer symptoms are numerous and striking. Yet none prove reliable as indicators of those who will delay, and most women overcome all barriers to seek immediate diagnosis. This study looks more deeply into the reasoning of symptomatic women sustaining confidence in a decision to delay diagnosis of self-discovered breast symptoms. Using argument and heuristic analysis, we examined the structure and soundness of the reasoning in interviews with 28 women from the San Francisco Bay area monitoring breast symptoms. Fifteen women were sustaining decisions to delay seeking diagnosis. Their arguments’ structure and soundness, and their dependence on heuristic strategies, were compared with those of women who did not delay. Prompt diagnosis-seekers used vivid stories of other women with breast cancer to explain their diagnosis seeking, and the others used similar stories to justify on-going decisions to delay. Diagnosis-seekers offered more arguments for doing so than for delay. Delayers offered fewer arguments for seeking diagnosis and many more for delay. Delayers abandoned sound and usually compelling arguments to seek diagnosis, relying instead on false information, poorly reasoned arguments, and self-created dominance structures around decisions to delay. Decisions to delay were resilient, yet required maintenance to sustain. Intervention studies aimed at decreasing patient delay should address the thinking process by questioning reliance on mistaken claims of control over possibly advancing cancer, satisficing (corner-cutting to arrive at a minimally adequate solution to achieve a goal) when scheduling diagnostic visits, simulating a benign diagnosis rather than the prevention of late-staged cancer, prioritizing fear control over protection of life. Interventions might also include challenging mistaken analogies and the too facile abandonment of sound arguments for seeking prompt diagnosis. r 2006 Elsevier Ltd. All rights reserved. Keywords: Patient delay; Breast cancer; Logical analysis; Heuristic reasoning; Dominance structure; Denial; USA
Introduction Patient delay refers to a lagtime in the presentation of self-discovered cancer symptoms for clinical diagnosis. A seminal literature review of patient delay in cancer (Pack & Gallo, 1938) defined Corresponding author. Tel.: +1 650 743 8631;
fax: +1 650 692 0141. E-mail addresses: [email protected] (N.C. Facione), [email protected] (P.A. Facione). 0277-9536/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2006.08.014
‘‘undue patient delay’’ in breast cancer as 3 months or more from symptom discovery to initial clinical examination, allowing researchers to track patient delay in breast cancer over time. Facione and colleagues proposed a model for predicting the decision to delay in breast cancer, separating the factors influencing symptom appraisal from those influencing the decision to delay diagnosis (Facione, Miaskowski, Dodd, & Paul, 2002). This and other studies support the observation that even women who appraise symptoms as likely to be cancer
ARTICLE IN PRESS 3138
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
related frequently make confident judgments to delay (Facione & Dodd, 1995; Facione, Dodd, Holzemer, & Meleis, 1997; Lauver, Coyle, & Panchmatia, 1995; Lierman, 1988). Delays of at least 3 months occur in one-third of all women diagnosed with self-discovered breast cancer, and 6-month patient delays occur in as many as 25% of cases (Facione, 1993; Richards, Westcombe, Love, Littlejohns, & Ramirez, 1999). Most studies link patient delay with more advanced staged breast cancer at diagnosis (Arndt et al., 2002; Harirchi, Ghaemmaghami, Karbakhsh, Moghimi, & Mazaherie, 2005; Nosarti et al., 2000). Decreased survival related to patient delay was quantified in one metastudy of 87 published reports (101,954 breast cancer patients) as a 12–19% decrease in 5-year survival in those women with delays of 3 months or more versus those with shorter times to diagnosis (Richards et al., 1999). A decision to delay diagnosis, particularly if one suspects cancer, might seem highly irrational. Clinicians commonly consider delay as a manifestation of denial, a minimization of threat to manage symptom-related anxiety (Cousins, 1982), a conclusion suggested by psychiatric studies of prolonged delay (Magarey, Todd, & Blizard, 1977; Wool, 1986). But Cousins (1982) questioned whether the term, was being used too liberally, or too judgmentally. Subsequent qualitative studies of patient delay have described numerous other factors related to patient delays of months or even years. The variables associated with patient delay are strikingly similar across cancer sites, involving similar social and cultural factors internationally. Poor access to health care, lack of preventive health care habits, being older and married, perceiving prejudice in care delivery, and having childcare/ elder care obligations are all significantly associated with patient delay (Arndt et al., 2002; Burgess, Hunter, & Ramirez, 2001; Harirchi et al., 2005; Ramirez et al., 1999; Rozniatowski et al., 2005). Similarly, intrapersonal factors including: benign attribution, poor education and lack of knowledge, misperception of risk, lower anxiety/optimistic bias, embarrassment, fear of chemotherapy or breast loss, concern about being an hypochondriac, and pessimism about survival are associated with patient delay (Bain & Campbell, 2002; Burgess et al., 2001; De Nooijer, Lechner, & De Vries, 2000; Facione & Dodd, 1995; Harirchi et al., 2005; Nosarti et al., 2000; Rozniatowski et al., 2005; Smith, Pope, & Botha, 2005). Hypothesized individual differences
(pessimism, anxiety and depression, external locus of control) have not been consistently observed in group comparison studies of patient delay (Bain & Campbell, 2002; Ramirez et al., 1999). However, this list of factors, aggregated by observing people who delayed cancer diagnosis, is neither unique to those who delay nor predictive of patient delay. The majority of women overcome varying clusters of these factors, deciding to seek immediate diagnosis (Richards et al., 1999). Women engage potential breast cancer symptoms much as they would engage other potential health threats that permit cognitive assessment. That is, they assess risk and weigh arguments for action. Some decide to delay even when they worry that the symptom is breast cancer (Andersen, Cacioppo, & Roberts, 1995; Burgess et al., 2001; Facione & Dodd, 1995; Smith et al., 2005). This study was undertaken to analyze the reasoning involved when symptomatic women weigh the need to seek a diagnosis, and achieve and sustain confidence in decisions to delay. Determining whether and when to seek diagnosis of breast symptoms is a decision that is purposeful, more or less well informed, reflective and of high consequence. Theories that attempt to explain this type of naturalistic decision model involve the interaction of two overlapping systems of reasoning that are active in human decision-making (Gilovich, Griffin, & Kahneman, 2002; Kahneman & Miller, 2002; Sloman, 2002; Zsambok, 1997). One is intuitive, non-reflective, associative, and holistic (sometimes referred to as System 1), and the other (System 2) is more deliberative, reflective, and rule governed. System 1 thinking is believed to rely heavily on heuristics (logical shortcuts), key situational cues, and salient memories to arrive quickly and confidently at judgments, particularly when situations are familiar and immediate action is required. System 2 thinking is believed to be useful for judgments in unfamiliar situations, for processing abstract concepts, and for deliberating when there is time for planning and more comprehensive consideration. Humans use heuristic shortcuts in System 2 as well, often as part of their argument making. Both systems are believed valuable, functioning interactively and in parallel to address problems. We suspend further discussion of this model due to space limitations, and direct the reader to a recent excellent anthology (Gilovich et al., 2002). Whether there are two reasoning systems or one complexly integrated reasoning system, it is clear
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
that we need to analyze and understand how humans use both argument making and heuristic thinking to make decisions such as the ones women make when they discover a change in their breast and consider whether it might be a signal of breast cancer. Risk and uncertainty complicate this decision, upping the potentially severe consequences of a faulty judgment. Studies report that women who fail to present self-discovered breast cancer for diagnosis prematurely dismiss their symptoms as benign or monitor them for a time as unclassified breast changes, most seeking diagnosis only when symptoms exacerbate (Coates et al., 1992; Schleicher & Ammon, 1998). There are no longitudinal studies documenting this decision process over time, but there are studies that suggest how such a potentially risky judgment might be sustained. Montgomery (1989) described the thinking process humans use to sustain risky and uncertain decisions. People form early preferences for how they would like to respond to a high stakes dilemma. They structure their arguments for the value of one chosen alternative while minimizing the values of the other alternatives not to be chosen. They may seek selectively for facts that support their choice. While this natural process may fall short on objectivity, there is increasing evidence that this type of maneuver is needed in order to find sufficient confidence to act. The ‘dominance structure’ supporting the chosen alternative bolsters and sustains confidence in the uncertain judgment, even when the negative consequences of error are high (Montgomery, 1989; Zsambok, 1997). Here we examine the on-going decision to delay a diagnostic visit in women with self-discovered breast symptoms to determine if there is evidence of dominance structuring in their decision process. This would seem a reasonable hypothesis because of the confidence level needed to sustain this high-risk decision given the uncertainty and possible risk posed by the symptoms. In this study, we examined symptomatic women’s reasoning for seeking diagnosis versus making an on-going judgment to delay. Our goal was to better understand how women could feel confident about judgments to delay the evaluation of breast symptoms that they themselves feared might be signals of breast cancer. Methods Symptomatic women from the San Francisco Bay area, a sub-sample of women recruited to a
3139
community-based survey study of breast health behavior (Facione et al., 2002) were recruited for in depth interviews regarding symptom-related decision-making. Data was gathered over a 6-month period in 1999–2000, after completion of the survey study. ‘Symptomatic’ was defined as an affirmative response to at least one of two survey items: ‘‘There is an area in my breast y (1) that I’m not sure is normal; or (2) that I worry may be breast cancer’’. All data for this study were collected and analyzed in accordance with institutional review board approval and with assurance of protection of human subjects. Thirty-nine women in the parent study fit the criterion of ‘symptomatic’. We successfully interviewed 28 (72%). Table 1 displays the characteristics of our interview sample. Women we failed to contact were more likely to be Spanish-speaking, not college educated, and born outside USA. Our interviewed sample contains an ample sub-sample of Spanish speaking subjects to provide cultural breadth in our findings. Spoken language and national origin were assumed to be neutral in Table 1 Sample characteristics Demographics Age Annual income
Language Education
Race/ethnicity
Sexual orientation Health care coverage Employment: Relationship status
Mean 42.34 years (s.d 14.1), range 24–72 years o$9,999 (n ¼ 7); $10,000–30,000 (n ¼ 8); $35,000–50,000 (n ¼ 4); $60,000–80,000 (n ¼ 4); 4$80,000 (n ¼ 5) English (n ¼ 18); Spanish (n ¼ 10) Grade school (n ¼ 3); high school/ vocational school (n ¼ 8); college (n ¼ 13); graduate school (n ¼ 4) White (n ¼ 12); Black (n ¼ 4); Latino (n ¼ 10); Asian (n ¼ 1); multi-ethnic (n ¼ 1) Heterosexual (n ¼ 22); lesbian or bisexual (n ¼ 3); missing (n ¼ 3) Insured (n ¼ 22); uninsured (n ¼ 6) Employed (n ¼ 16); unemployed or retired (n ¼ 12) Married or partnered (n ¼ 14); single or widowed (n ¼ 14)
Cancer-related variables Breast lump Yes (n ¼ 14); No (n ¼ 13); missing (n ¼ 1) Cancer symptom Yes (n ¼ 14); No (n ¼ 12); missing attribution (n ¼ 2) Close friend of Yes (n ¼ 11); No (n ¼ 17) relative with cancer
ARTICLE IN PRESS 3140
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
relationship to reasoning ability. Education is widely believed to increase the factual knowledge that one might used to inform a judgment. Our sample contains a majority of women with college education (61%), and thus the results we report may represent better overall knowledge about breast cancer than might be the case in a less well-educated sample. In-depth interviews were conducted in women’s homes or in conference rooms of community agencies. Women discussed their ideas about breast cancer early detection in general, and their current breast symptom(s) in particular. The interviews were conducted by trained research assistants in the woman’s native language. Each woman was asked about her thinking process without introducing her to new factual information, correcting faulty assumption or inferences, or evaluating her stated reasons or conclusions. This ‘talk-aloud’ methodology is described in detail elsewhere (Svenson, 1989). Women were reimbursed $30 for participating in the 90 min interview. At the completion of the interview, all were offered assistance in obtaining clinical breast examinations. All but three declined. Two of these women underwent biopsy and one was later treated for breast cancer. All interviews were transcribed for analysis by native English or Spanish-speaking research staff. The 10 Spanish language interviews were translated by two native Spanish-speaking researchers, and all data were analyzed in English for this decisionmaking study. Our data analysis method, described in Table 2, is grounded in analytical research describing applied logic, heuristic thinking and argument structure (Facione & Facione, 2001, 2006; Toulmin, 1969) and behavioral research describing heuristic reasoning (Montgomery, 1989; Kahneman, Slovic, & Tversky, 1982; Zsambok, 1997). This method examines the decision-making process embedded in participant interviews. Standard methods were used to demarcate the data to identify and retain the decision-making content. Typically, there is non-linearity in the oral presentation of human reasoning. For instance someone might say ‘‘I need to leave. My son is waiting, and my parking meter is running out’’. In this case the conclusion comes first, and the reasons follow. There may even be more reasons for ‘needing to leave’ given later in the dialogue. To analyze the complete argument accurately, these data segments need to be reordered. Arguments that are repeated later in the interview are not treated as
Table 2 Protocol for argument and heuristic data analysis Step 1 Prepares data for analysis
Step 2 Identifies main conclusions
Step 3 Identifies all relevant arguments
Step 4 Evaluates arguments for logical soundness
Step 5 Identifies possible heuristic errors
Step 6 Identifies dominance structuring
The data is parsed, retaining its ordered chronology, into units of analysis (typically a single communication: assertion, belief, comment, proposition, pause/silence, question). These units are complete sentences, parts of sentences or incomplete sentences, each contextually understandable even if not entirely grammatical. Identify the dilemma that is central to the thinking process and each of the alternative conclusions being considered. Identify the final conclusion, if the individual comes to a final judgment in the interview. All interview data asserting arguments in support of conclusions are ordered into their most logical format (argument strands). Argument strands are examined for redundancy and related content. Each argument strand is retained, regardless of how underdeveloped it may appear to be. This process continues until all relevant data is accounted for as a component of an argument supporting one of the identified alternative conclusions. Each argument is then analyzed for soundness (quality of reasoning process and accuracy of information content). Sound arguments rely on accurate premises and assumptions and warranted inferences to reach justified conclusions. Unsound arguments rely on false statements, misinformation, unwarranted inferences, misapplied analogies, fallacies, mistaken estimates or other errors to reach unjustified conclusions. Each reasoning strand is re-examined for use of heuristic strategies. These are often identified in the previous steps, but the data should be comprehensively searched again at this point. The appropriateness of relying on the heuristic to make the decision is evaluated in each instance. Argument strands are clustered into similar content areas to create sets of related reasons supporting each of the main conclusions. All argument strands are visually mapped to permit a holistic examination of the flow and progression of the decision-process. This map is examined for evidence of a dominance structuring around the final conclusion. Watershed claims are identified (statements that lead to arguments for more than one decision/conclusion.
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
new arguments, but if new content is introduced this content is used to elaborate the argument already advanced. The chronology of the data is retained to assist with the analysis of the overall structure of the decision process (Step 6). We examined the network of arguments, which women presented as the reasons (premises, hypotheses, assumptions) and claims (decisions, conclusions) for how they had responded to their selfdiscovered breast symptoms. We evaluated each argument for soundness in two distinct ways: first, for truth or falsity of the informational content of the reason given; and second for whether the claim being made was logically entailed, justified, or warranted by the reason given. Arguments are determined to be ‘warranted’ when they rely on an appropriate generalization (community norm or widely held belief) (Toulmin, 1969). Using contemporary argument evaluation categories, we labeled those arguments that passed these two tests as ‘‘sound’’, meaning that the informational content in the reason given was in fact true, and that the reason either warranted, justified, or logically implied the claim being made. But if the argument was based on misinformation or if the reasoning was not valid, the argument was evaluated as unsound. Finally we identified the women’s use of heuristic reasoning (their logical shortcuts). Concept mapping software (Inspiration Software, 1997) was used to visualize the argument structure within each interview. This software allows manipulation of data segments to designate their membership in argument strands, demarcation of the chronology of the data, and indication of where heuristic reasoning occurs (using colors, shapes, and directional arrows). We analyzed each interview and constructed maps of the arguments in each woman’s judgment process. Each woman’s decision map was examined to assess the presence of a dominance structure bolstering her decision to either delay or to seek diagnosis. Each of the two analysts began independently to develop an analysis of the argument strands contained in each interview and to construct an initial argument map for each case. The independent analyses were then examined collaboratively, case by case, for completeness, accuracy, and congruity. All data provided by the participant regarding symptom appraisal and decision about seeking a diagnostic visit is accounted for in the listing of argument strands and appears in each woman decision map. Finally the results of these
3141
analyses were collected across cases for presentation here. Evidence for the quality of the initial independent analyses was seen in the high congruity of the initial case analyses, but what is also required is a consensus in the completed logical, heuristic and structural analysis of each case interview. Comparative analyses resulted in the identification of a few additional argument strands and a clearer interpretation of how these clustered within each case. Consensus was reached on the basis of demonstrating clear evidence in the data transcriptions for each of the identified argument strands. In two cases investigators returned to the audiotapes to resolve ambiguities. Results Thirteen women (46%) reported having sought an immediate diagnostic visit. In nine cases, providers were monitoring their symptoms and four others had had benign biopsies. The remaining 15 women continued to monitor their symptoms, but had not been examined by a provider. All 15 had symptoms for at least 3 months and eight reported symptoms for a year or more (delayers). Nine of these (60%) had told no one of their symptoms. None of the delayers reported a lack of confidence in their on-going decision not to seek a diagnosis. Women reported having breast lumps (n ¼ 14, 50%), nipple discharge, firm areas, an enlarging breast, pain and itching. Women with breast lumps more often delayed (10 of 14, 71.4%) compared to those with other symptoms (4 of 13, 31%). Fourteen women (50%) reported thinking that their symptoms were probably symptoms of breast cancer, and eight of these were sustaining decisions to delay. Women who delayed were at least as well educated (66% college educated) as women who sought diagnosis (54% college educated). The mean income of both groups was $40,000 per year, and both groups were similarly distributed by race/ ethnicity. Women who delayed were on average 2 years younger (mean 41.7, standard deviation 714.5) than those seeking diagnosis (mean 43.1, 714.1). Heuristic analyses: Table 3 lists the names and descriptions of the heuristic shortcuts identified in these interviews. All but one woman in our sample demonstrated the availability heuristic, telling vivid stories of other women with breast symptoms and breast cancer to explain their own situation. These
ARTICLE IN PRESS 3142
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
Table 3 Heuristic reasoning shortcuts identified in these interviews Heuristic strategy
Description of cognitive maneuver
Potential consequences of misuse of this heuristic in the case of judgments to delay
Affect
Provides a strong and immediate sense of whether the issue at hand is ‘‘good’’ or ‘‘bad’’. Allows the thinker to take a stance for further examination of the issue. Vivid or personally salient past event is used as a basis to (over)estimate the likelihood of the event’s reoccurrence.
Strong negative feelings about breast cancer treatment or mastectomy make seeking diagnosis fearsome and repugnant.
Availability
Representativeness (associational) Representativeness (analogical), also called the Similarity heuristic
Associations of two linked terms or events. This cognitive shortcut might be used to orient further thinking about the topic. Similarities, whether superficial or fundamental, are relied upon as sufficiently representational for purposes of expecting similar outcomes or categorizations.
Satisficing
Deciding on an alternative, which is seen as adequate if not optimal.
Simulation
Imagining or projecting scenarios of ‘‘what will happen’’ if one makes alternative decisions or takes alternative actions as a way of predicting possible outcomes. The level of one’s involvement in an event is used to estimate one’s degree of control over the eventual outcome. Negative association of change with loss and risk. Decisions are aimed at conserving the status quo rather than suffering a loss.
Illusions of control
Loss aversion
stories were always told early in the interviews (first third), and the meanings that women attached to these stories became arguments in support of their ultimate decisions. Distressing stories of breast cancer death were recounted by seven women who sought immediate diagnosis. They emphasized unfortunate suffering and needless breast loss because the woman in their story delayed diagnosis or the physician initially misdiagnosed the cancer. They ended these stories by concluding that women need to seek diagnosis of symptoms as early as possible. Seven women who were delayers told stories that emphasized the pain of treatment and the inevitable death associated with a breast cancer diagnosis. They sometimes returned to these stories later in the interview and elaborated on them, adding additional arguments for the perceived benefit of delay. Use of the associational representativeness heuristic was seen in 14 (50%) women. Also occurring
False judgments as to the likelihood of consequences (e.g. a benign biopsy) can result in risky choices (to forego diagnosis of a new symptom). Associations like ‘‘breast cancer: death’’ or ‘‘breast cancer: breast loss’’ can result in a strong resistance to diagnosis. Over confidence in the apparent similarity of two cases (e.g. discovering a breast lump just as your friend did)) can lead to the belief that similar cases will have inevitably similar conclusions (e.g. death, even with diagnosis and treatment). One’s best option can be lost while one relies upon an approach that is a judged ‘‘good enough’’ (e.g. I can wait until my next scheduled visit). Imagining good (bad) outcomes can lead to the belief that the imagined outcomes will actually occur (e.g. Imagining the doctor laughing at my concern about this lump). Overestimated belief in one’s actual ability to control outcomes (e.g. Belief in the control of symptoms through healthy eating habits). Decisions that fail to foresee the consequences of inaction (e.g. acting to maintain control or avoid the sick role can result in greater morbidity).
early in the interviews, use of this heuristic occurred at the first mention of the related topic (cancer, treatments, breasts). Women seeking diagnosis saw cancer as representative of ‘death’ (n ¼ 3); lumps represented ‘cancer’ (n ¼ 3), treatments represented ‘suffering’ (n ¼ 1) and breasts represented ‘womanhood’ (n ¼ 1). These representations provided the energy to act on their symptom discovery by seeking diagnosis. Women who delayed also saw cancer as representative of ‘death’ (n ¼ 4), ‘ugly/ickiness’ (n ¼ 2) and ‘suffering’ (n ¼ 3), and one described her lump as ‘cancer’. However, these women did not conclude that they should seek diagnosis. Rather they tended to pause, recall episodic memories of prior provider visits, project negative consequences when simulating a future provider visit, and then simply say again that they did not need to seek a diagnosis. This cognitive activity was apparent from subsequent comments, such as: ‘‘I’ve had my symptoms dismissed as frivolous twice’’; ‘‘I’ve had
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
no relief from seeing a physician’’. And ‘‘Treatment means hospitalization, y pain’’. About half of the women used the loss aversion heuristic (Kahneman, Knetsch, & Thaler, 1991), making arguments to avoid loss of a breast, loss of a partner or spouse, or the loss of social standing as one’s normal, non-sick self. An example of this heuristic can be seen in the comment, ‘‘People take you as a poor, skinny victim’’. Heuristic thinking about loss aversion was seen both in those who delayed and in those who sought diagnosis. Misuse of the analogical representativeness heuristic was seen in the reasoning of four women who delayed. They claimed that friends’ or relatives’ experiences with benign biopsies were relevant to their own decision to delay: ‘My lump is just like hers, benign’. In contrast, in women who sought diagnosis, positive stories of well-treated cancer and benign biopsies were used to explain the advantages of seeking diagnosis. Two young women recounted stories of breast cancer in other women of their own age. Both correctly drew the analogy to their own vulnerability to breast cancer. One used the story as her explanation for ‘‘having no question about seeking immediate diagnosis’’. The other rather quickly discounted what she appeared to regard as a powerful heuristic insight, overriding this insight with 16 subsequent arguments for delay, 11 of them logically unsound. The affect heuristic was observable universally. All but one woman spontaneously spoke of cancer in strong negative terms such as ‘‘terrifying’’,
3143
‘‘ugly’’, ‘‘extremely frightening’’. Both groups of women said that discovering their symptoms was a fearful experience, and both were actively engaged in fear avoidance. Women who sought diagnosis used the fear to motivate seeking a diagnostic visit and described the visit as relieving uncertainty and anxiety. Women who delayed described experiencing anxiety, sadness and depression as well as fear. These feelings were associated with concerns about possible biopsy and breast loss. They voiced no expectation of emotional control or expected relief from a provider visit, and in some cases verbally denied this possibility, concluding that these negative emotions were best avoided by delaying. Logical analyses: Fig. 1 displays the analysis of arguments contained in the interviews by group. The distribution of sound and unsound arguments differed greatly between women who sought diagnosis and those who sustained a decision to delay. Those who sought diagnosis provided more than twice the number of arguments in support of this decision than women who delayed. These arguments were typically based on true beliefs and good reasoning. Women who delayed made nearly five times more arguments for delay than did women who sought immediate diagnosis. Many (24.1%) of the arguments for delay were sound, however, the vast majority were unsound, relying on misinformation (29.5%), poor reasoning (30.7%) or both (15.7%). Often women displayed confidence in poorly reasoned arguments containing a true statement. ‘‘A breast cancer lump is often painless. My lump is
Fig. 1. Argument analysis for the sample as a whole.
ARTICLE IN PRESS 3144
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
a bit uncomfortable. It can’t be cancer’’. This is a rather classic error of confusing ‘often’ with ‘always’. On occasion women would hear their unsound reasoning and offer self critique. More often, however, they appeared not to see the weakness of these arguments. Table 4 summarizes the most commonly advanced logical arguments both for seeking diagnosis and delay, grouped into the four categories of the logical analysis: sound; unsound because they rely on misinformation; unsound because they rely on poor reasoning; and unsound because they rely both on misinformation and poor reasoning. At first glance, some arguments coded as sound might appear to be unsound. For example, the argument ‘‘My husband/partner would leave me if I had a mastectomy’’, summarizes six women’s stories of conversations with male partners about breast biopsies, breast cancer, or mastectomy. The stories tell of negative effects of a breast cancer diagnosis on male–female relationships. Unspoken but clearly implied premises in each argument for delay were: ‘‘Being diagnosed with breast cancer would destroy my relationship with this husband/partner’’ and ‘‘This relationship is so important to me that I would risk death to maintain it’’. Taken together, these statements, if true in their particular circumstances, constitute a sound argument for a benefit inherent in delay. Similar premises, both spoken and implicit, were seen to be operative in many of the other sound arguments for delay listed in Table 4. For example, ‘‘I would risk death yto keep my breast intact, yto put off painful treatments, yto avoid incurring a large cost’’. Notice that while many of these arguments pass the logical test of soundness, they do not reflect what most people would take as wisdom in the overall judgment. It seems unwise to risk forfeiting one’s life to secure these proposed benefits. As a rule these arguments, while sound, were underdeveloped. Every interview contained arguments that relied on false information. Most women made inflated estimates of their personal risk of breast cancer. Every woman who had a relative with any type of cancer claimed to be at high risk for breast cancer, yet only three had first-degree relatives (mothers) with breast cancer, and none had high relative risk. These unsound arguments, because they supported seeking diagnosis, support the idea that people can make a good choice for a bad reason. False information about the causes of breast cancer
(abortions, bruising of the breast, wearing underwire bras, stress, and thinking of cancer) was used by delayers to argue that their own symptoms were therefore probably not cancer (for instance, never had an abortion, cannot remember being bruised) and therefore they could delay. Six women who delayed falsely asserted that cancerous lumps would be larger, or that cancer would produce a nipple discharge. Erroneous information about cancer protective behaviors involved breast feeding (I’ve heard it takes 14 years of breast feeding to be safe. I have two to go) and having children (I have no children, so I’ll probably get it). One common argument made by delayers was that they would be thought of as hypochondrical or foolish for presenting their breast symptoms for diagnosis. Some paused to relate stories about frustrating encounters with providers, and others rehearsed aloud (simulation) how they would call appointment schedulers, triage nurses, and physicians or nurse practitioners, be criticized or ridiculed about their breast symptom concerns, and then they concluded that it would be impossible to see their provider. Three women who did not delay also expressed concern about provider encounters, but one counter-argued that her health ‘should be more important than her doctor’s attitudes’ or ‘whether (the doctor) is in a hurry’. Several women misapplied normative generalizations (warrants) to bolster their arguments to delay. The most common misapplied warrant was: ‘Do good unto others’. Women who delayed applied this warrant as though it directed them to risk ignoring the potential threat posed by breast lumps and nipple discharge in favor of performing normal childcare and other family duties (n ¼ 4). Two women who sought diagnosis conversely applied this warrant correctly, saying that they had to seek diagnosis for the sake of their children: ‘‘They need their mother to be there’’; ‘‘I don’t want them to see me die’’. Five women misapplied the warrant ‘Do not burden others unnecessarily’ to justify delay, arguing that seeking diagnosis would worry family and friends. They ignored the potential for added burden on family if their symptoms proved cancerous and advanced. Others argued ‘‘I can wait until my next annual exam to check this symptom’’. Here the satisficing heuristic (corner-cutting to arrive at a minimally adequate solution to achieve a goal) worked in tandem with a misapplied warrant
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
3145
Table 4 The most commonly made arguments for seeking diagnosis and for delay Examples of most commonly made sound arguments
Sought diagnosis (n/%)
Seek diagnosis If you treat cancer early you can survive. My provider will competently manage this breast problem. Getting reassurance from a provider would be a relief. If this (breast symptom) is cancer I could die. I could have breast cancer. I would have a mastectomy to save my life. Relatives/friends with cancer have benefited from early detection.
11(85) 10(77) 8(73) 7(54) 6(46) 5(38) 4(31)
1(6) 4(27) 1(6) 4(27) 8(53) 2(13) 2(13)
4(31) 7(54) 0(0) 2(15)
11(73) 6(40) 5(33) 4(27)
3(23) 1(8) 0(0) 2(15) 0(0) 0(0) 1(8)
10(67) 10(67) 8(53) 7(47) 6(40) 5(33) 4(27)
6(46)
5(33)
7(54) 4(31) 2(15) 0(0) 0(0)
14(93) 14(93) 11(73) 8(53) 4(27)
2(15) 1(8) 0(0) 0(0) 1(8) 0(0) 0(0)
13(87) 10(67) 9(60) 8(53) 7(47) 7(47) 5(33)
Delay Treatments (surgery and chemo) are terrible, painful. Losing a breast is terrible to contemplate. Treatment is costly. I can’t afford medical care. My husband/partner will leave me if I have a mastectomy.
Delayed (n/%)
Examples of poorly reasoned arguments Delay I’ve had bad experiences talking with my doctor/disrespect/rudeness. I try to block thinking about it because it makes me upset. I can’t talk to my provider about most things. I just know that I won’t die from this. This isn’t how I’ll die. If you keep the symptom to yourself, you won’t be nagged about it. It’s just typical of me never to take care of myself. My lump is probably benign like my mother’s/my friend’s. Examples of arguments based on misinformation Seek diagnosis I’m at much higher risk than most women. Delay No one related to me has breast cancer. I’m not at risk. I can manage this by diet, exercise, smoking and drinking habits. Cancer is a death sentence. Once you find a lump it’s too late. I can leave it until my next health promotion visit. Thinking about cancer can make it happen. Examples of arguments based on poor reasoning and misinformation Delay Providers are incompetent. They will make the wrong diagnosis. I don’t want to be thought foolish or a hypochondriac. The symptom isn’t bad enough yet for me to worry about it. Everything can be possible with God’s help. I’ll try prayer instead. Scheduling an appointment is too difficult, intimidating, a hassle. Work and family is why I delay. I need to take care of these first. I don’t want to burden other people. Seeking diagnosis will do that.
(misapplication of the guideline ‘Have an annual clinical breast examination’). Conversely, women who sought diagnosis were not mistaken about what was minimally required, and pointed to the correct guideline: ‘If you find a lump, see your doctor immediately’.
At times women appeared to reason backwards from a norm (Stop smoking because it puts you at risk for breast cancer) to claims about controlling their symptoms. They argued: ‘‘Cancer occurs as a result of smoking, eating a poor diet, or not exercising. So I can solve this problem by stopping
ARTICLE IN PRESS 3146
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
smoking, eating bettery’’. Women seemed to be using reports of possible causal relationships between alcohol, tobacco, a poor diet and the development of breast cancer, drawing the false inference that they could reverse/eliminate a developing breast cancer by altering or ceasing these behaviors. They claimed control over the symptoms by: improving their diet (n ¼ 11), starting to exercise (n ¼ 4), stopping smoking (n ¼ 3), decreasing/eliminating alcohol intake (n ¼ 2), decreasing their stress (n ¼ 2), using homeopathy (n ¼ 3), and meditating (n ¼ 2). These arguments are examples of what has been termed ‘backwards thinking’, a quite common phenomenon that includes counterfactual thinking about how events might have been otherwise if one had only behaved as the norm would recommend (Kahneman & Miller, 2002). Counter-factual thinking becomes logically unsound when one intimates that events would certainly have been different in particular ways if the norm was followed, and in this case the women’s backwards reasoning also violates the logic of temporal causality. Others have described this type of thinking as a misapplication of the ‘illusion of control’ heuristic. Perhaps because the claim of causal control was rarely made explicitly, the unsound logic of the backwardleaping inference went unnoticed and instead bolstered decisions to delay in nine (60%) of the women who delayed.
Dominance structuring: Women described thinking again and again about their symptoms, and reaffirming their decision to delay each time the worrisome symptoms came to mind, which was in most cases a daily occurrence. They demonstrated this process in the interview as they thought aloud about their symptoms. Four manifestations of dominance structuring were observed in the data. First, no women were undecided about whether or not they would seek diagnosis of their breast symptom. All were either firm in their decision to seek diagnosis, and in fact had done so, or were sustaining their judgment to delay. Second, women tended to dismiss as unimportant, refute, or simply abandon strands of reasoning which appeared to be leading to the alternative decision that they did not embrace. Third, examination of the chronology of the data demonstrated the occurrence of numerous, less well elaborated, bolstering arguments later in the interview. Fourth, dominance structuring could be seen visually by examining the argument maps. Fig. 2 is a map of one woman’s interview, each oval representing a summarized strand of argument. She abandoned three sound arguments for seeking diagnosis, and took time to provide a counterargument to two additional sound arguments (watershed claims). In all, she makes 18 arguments for delay, 17 of them unsound. None of these arguments clearly grounds her decision to delay, but
Fig. 2. One woman’s summarized argument map.
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
taken together there is confidence in this choice as dominant. Discussion This study is the first to analyze the arguments sustaining a decision to delay diagnosis made by women currently monitoring potential breast cancer symptoms. In general, women’s reasons for delay were not new or surprising, many having been previously reported as attitudes or beliefs associated with patient delay (Bish, Ramirez, Burgess, & Hunter, 2005; De Nooijer et al., 2000; Facione & Giancarlo, 1998; Facione et al., 2002; Lauver et al., 1995; Ramirez et al., 1999). This leads us to believe that our sample was a good source of information about women’s decisions to delay. Here we show how these variables are used in arguments to delay, the magnitude of arguments made, and how these arguments are used together to sustain confidence in the decision to delay. Through the use of argument and heuristic analysis we have described how decisions to seek diagnosis differ from decisions to delay. Delayers exhibited many more logical errors, including drawing faulty inferences, claiming false facts to be true, and misapplying rules and guidelines. Delayers were also less than appropriately attentive to their poor reasoning and critical self-commentary. Common heuristic thinking errors included the misuse of availability, representativeness, loss aversion, satisficing, and the illusion of control heuristics. Both types of errors put these women at greater risk for advanced cancer at diagnosis should any symptom prove malignant. Both groups of women made arguments for delay that passed the logical test for soundness, but women who delayed were observed to rely on these arguments and to abandon or superficially dismiss other sound arguments for the necessity of seeking prompt diagnosis. In contrast to those who sought diagnosis, they added numerous sound and unsound (bolstering) arguments to support their decision to delay, creating a dominance structure to support their risky and uncertain decision. Some might call this denial, but there was nothing passive or subconscious about this cognitive process in the women we interviewed. Rather, dominance structuring around the decision to delay required considerable on-going effort. That they elected to participate in this breast health study about their decision process, and that they continued to
3147
consider sound arguments to seek diagnosis, may be an important finding suggesting the potential for targeted, guided reasoning interventions. But the resilience of their decisions to delay should not be underestimated (Montgomery, 1989). We observed many abandoned sound arguments for seeking diagnosis and a vast network of arguments supporting risky judgments for delay. The question of whether the dominance structure supporting a judgment to delay becomes more elaborate over time as women work to sustain their delay of diagnosis cannot be answered by the current study. We observed that the women more frequently repeated arguments for delay, but we cannot say whether these repeated arguments were more important. Perhaps they initially anchored women as ‘likely to delay’ versus ‘likely to seek diagnosis’, but they may only represent arguments not confronted by others or arguments the women believed would be persuasive to others. To attain our goal of describing the complete argument for a judgment to delay, our study design precluded challenging women’s unsound arguments or correcting their misuse of heuristics. As a result, we cannot determine whether women would have held tightly to particular arguments, reformulated arguments, or even rescinded their judgments to delay as a result of being specifically asked to reexamine their thinking process. These questions invite intervention studies. The dominance structures around judgments to delay should not be discounted as the self-serving rationale of women who are deliberately refusing medical management. Many women in this sample believed themselves to have thought well about whether to seek diagnosis, and to have arrived at a decision to delay reasonably. It is their sense that they had made a good decision that should be worrisome to health professionals. We found women’s stories of alienation from the health care system, and their concerns about being judged hypochondrical to be particularly disturbing. This is not the first study documenting anxiousness surrounding the presentation of self-discovered symptoms to providers (Bish et al., 2005; LaVeist, Rolley, & Diala, 2003; Reece, 2003), and directly addressing this issue might improve early detection. The women in our well-educated sample often lacked specific breast cancer symptom knowledge, and were able to abandon sound arguments for seeking prompt diagnosis. When a person uses false information to evaluate how they should respond to
ARTICLE IN PRESS 3148
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
a dilemma they are likely to arrive at an unsound conclusion that rings of subjective truth. Reaching women who are currently monitoring breast symptoms with effective media messages may require targeting one or more of the arguments she makes for delaying symptom evaluation. We hypothesize that framing these messages in relation to her thinking strategies might better impact the dominance structure she has likely created around her decision to delay. Our suggestions for future intervention research aimed at decreasing patient delay would include studies targeting multiple aspects of the dominance structure by: (1) challenging misapplied warrants about the relative importance of cancer early detection, (2) correcting an array of misinformation with accurate facts about cancer symptoms and treatment, (3) overcoming fears of possible negative encounters with providers when seeking help, (4) exposing common reasoning mistakes that diminish the consequences of delay, (5) reframing mistaken stories of other women’s experience with breast cancer, and (6) encouraging a more reflective self critique regarding illusions of control, and satisficing in relationship to scheduling a diagnostic visit. References Andersen, B. L., Cacioppo, J. T., & Roberts, D. C. (1995). Delay in seeking a cancer diagnosis: Delay stages and psychophysiological comparison processes. British Journal of Social Psychology, 34, 33–52. Arndt, V., Sturmer, T., Stegmaier, C., Ziegler, H., Dhom, G., & Brenner, H. (2002). Patient delay and stage of diagnosis among breast cancer patients in Germany—a population based study. British Journal of Cancer, 86(7), 1034–1040. Bain, N. S., & Campbell, N. C. (2002). Treating patients with colorectal cancer care: A qualitative study of rural and urban patients. Family Practice, 19, 360–374. Bish, A., Ramirez, A., Burgess, C., & Hunter, M. (2005). Understanding why women delay in seeking help for breast cancer symptoms. Journal of Psychosomatic Research, 58, 321–326. Burgess, C., Hunter, M. S., & Ramirez, A. J. (2001). A qualitative study of delay among women reporting symptoms of breast cancer. British Journal of General Practice, 51, 967–971. Coates, R. J., Bransfield, D. D., Wesley, M., Hankey, B., Eley, J. W., Geenberg, R., et al. (1992). Differences between Black and White women with breast cancer in time from symptom recognition to medical consultation. Journal of the National Cancer Institute, 84, 938–950. Cousins, N. (1982). Commentary: Denial. Journal of the American Medical Association, 248(2), 210–212. De Nooijer, J., Lechner, L., & De Vries, H. (2000). A qualitative study on detecting cancer symptoms and seeking medical
help: An application of Anderson’s model of total patient delay. Patient Education and Counseling, 42(2), 145–157. Facione, N. C. (1993). Delay versus helpseeking for breast cancer symptoms: A critical review of the literature on patient and provider delay. Social Science & Medicine, 36(12), 1521–1534. Facione, N. C., & Dodd, M. J. (1995). Women’s narratives of helpseeking for breast cancer. Cancer Practice, 3(4), 219–225. Facione, N. C., Dodd, M. J., Holzemer, W., & Meleis, A. I. (1997). Helpseeking for self-discovered breast symptoms: Implications for cancer early detection. Cancer Practice, 5(4), 220–227. Facione, N., & Facione, P. (2001). Analyzing explanations for seemingly irrational choices: Linking argument analysis and cognitive science. International Journal of Applied Philosophy, 15(2), 267–286. Facione, N. C., & Facione, P. A. (2006). Argument and Heuristic Analysis: Method, Theory and Application. (pp. 35–48). Millbrae, USA: California Academic Press. Facione, N. C., & Giancarlo, C. A. (1998). Narratives of breast symptom discovery: Implications for early detection. Cancer Nursing, 21(6), 430–440. Facione, N. C., Miaskowski, C., Dodd, M. J., & Paul, S. M. (2002). The self-reported likelihood of patient delay in breast cancer: New thoughts for early detection. Preventive Medicine, 34, 397–407. Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and biases: The psychology of intuitive judgment. Cambridge, UK: Cambridge University Press. Harirchi, I., Ghaemmaghami, F., Karbakhsh, M., Moghimi, R., & Mazaherie, H. (2005). Patient delay in women presenting with advanced breast cancer: An Iranian study. Public Health, 119(10), 885–891. Inspiration Software (1997). Inspiration Software, Inc., Version 6, www.inspiration.com. Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). The endowment effect, loss aversion, and status quo bias. Journal of Economic Perspectives, 5, 193–20644. Kahneman, D., & Miller, D. T. (2002). Norm theory: Comparing reality to its alternatives. In T. Glovitch, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 348–366). Cambridge, UK: Cambridge University Press. Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under uncertainty: Heuristics and biases. Cambridge, UK: Cambridge University Press. Lauver, D., Coyle, M., & Panchmatia, B. (1995). Women’s reasons for and barriers to seeking care for breast symptoms. Women’s Health Issues, 5(1), 27–35. LaVeist, T. A., Rolley, N. C., & Diala, C. (2003). Prevalence and patterns of discrimination among US health care consumers. International Journal of Health Services, 33(2), 331–344. Lierman, L,M. (1988). Discovery of breast changes. Cancer Nursing, 11(6), 352–361. Magarey, C. J., Todd, P. B., & Blizard, P. J. (1977). Psychosocial factors influencing delay and breast self-examination in women with breast symptoms of breast cancer. Social Science & Medicine, 11, 229–232. Montgomery, H. (1989). From cognition to action: The search for dominance in decision making. In H. Montgomery, & O. Svenson (Eds.), Process and structure in human decisionmaking (pp. 23–49). Chichester, UK: Wiley.
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149 Nosarti, C., Craford, T., Roberts, J. V., Elias, E., McKenzie, K., & David, A. S. (2000). Delay in presentation of symptomatic referrals to a breast clinic: Patient and system factors. British Journal of Cancer, 82(3), 742–748. Pack, G. T., & Gallo, J. S. (1938). The culpability for delay in the treatment of cancer. American Journal of Cancer, 33, 443–462. Ramirez, A. J., Westcombe, A. M., Burgess, C. C., Sutton, S., Littlejohns, P., & Richards, M. A. (1999). Factors predicting delayed presentation of symptomatic breast cancer: A systematic review. Lancet, 353(9159), 1127–1131. Reece, M. (2003). HIV-related mental health care: Factors influencing dropout among low-income, HIV-positive individuals. AIDS Care, 15(5), 707–716. Richards, M. A., Westcombe, A. M., Love, S. B., Littlejohns, P., & Ramirez, A. J. (1999). The influence of delay on survival in patients with breast cancer: A systematic review. Lancet, 353(9159), 1119–1126. Rozniatowski, O., Reich, M., Mallet, Y., Penel, N., Frounier, C., & Lefebvre, J. L. (2005). Psychosocial factors in delayed consultation by patients with head and neck cancer. Head & Neck, 27(4), 274–280. Schleicher, U. M., & Ammon, J. (1998). Mode of breast cancer detection: A study from the German part of the Maas-Rhine-
3149
EUREGIO. European Journal of Cancer Prevention, 1, S41–S46. Sloman, S. A. (2002). Two systems of reasoning. In T. Glovitch, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 379–396). Cambridge, UK: Cambridge University Press. Smith, L. K., Pope, C., & Botha, J. L. (2005). Patients’ helpseeking experiences and delay in cancer presentation: A qualitative synthesis. Lancet, 366(9488), 825–831. Svenson, O. (1989). Eliciting and analyzing verbal protocols in process studies of judgment and decision-making. In H. Montgomery, & O. Svenson (Eds.), Process and structure in human decision-making (pp. 65–81). Chichester, UK: Wiley. Toulmin, S. (1969). The uses of argument. Cambridge, UK: Cambridge University Press. Wool, M. S. (1986). Extreme denial in breast cancer patients and capacity for object relations. Psychotherapy and Psychosomatics, 46, 196–204. Zsambok, C. E. (1997). Naturalistic decision-making: where are we now? In C. E. Zsambok, & G. Klein (Eds.), Naturalistic decision making (pp. 3–16). Mahwah, NJ: Lawrence Erlbaum Associates.
Social Science & Medicine 63 (2006) 3137–3149 www.elsevier.com/locate/socscimed
The cognitive structuring of patient delay in breast cancer Noreen C. Facionea,, Peter A. Facioneb a
University of California San Francisco, USA b Loyola University Chicago, USA Available online 25 September 2006
Abstract The reasons women give for delaying diagnosis of breast cancer symptoms are numerous and striking. Yet none prove reliable as indicators of those who will delay, and most women overcome all barriers to seek immediate diagnosis. This study looks more deeply into the reasoning of symptomatic women sustaining confidence in a decision to delay diagnosis of self-discovered breast symptoms. Using argument and heuristic analysis, we examined the structure and soundness of the reasoning in interviews with 28 women from the San Francisco Bay area monitoring breast symptoms. Fifteen women were sustaining decisions to delay seeking diagnosis. Their arguments’ structure and soundness, and their dependence on heuristic strategies, were compared with those of women who did not delay. Prompt diagnosis-seekers used vivid stories of other women with breast cancer to explain their diagnosis seeking, and the others used similar stories to justify on-going decisions to delay. Diagnosis-seekers offered more arguments for doing so than for delay. Delayers offered fewer arguments for seeking diagnosis and many more for delay. Delayers abandoned sound and usually compelling arguments to seek diagnosis, relying instead on false information, poorly reasoned arguments, and self-created dominance structures around decisions to delay. Decisions to delay were resilient, yet required maintenance to sustain. Intervention studies aimed at decreasing patient delay should address the thinking process by questioning reliance on mistaken claims of control over possibly advancing cancer, satisficing (corner-cutting to arrive at a minimally adequate solution to achieve a goal) when scheduling diagnostic visits, simulating a benign diagnosis rather than the prevention of late-staged cancer, prioritizing fear control over protection of life. Interventions might also include challenging mistaken analogies and the too facile abandonment of sound arguments for seeking prompt diagnosis. r 2006 Elsevier Ltd. All rights reserved. Keywords: Patient delay; Breast cancer; Logical analysis; Heuristic reasoning; Dominance structure; Denial; USA
Introduction Patient delay refers to a lagtime in the presentation of self-discovered cancer symptoms for clinical diagnosis. A seminal literature review of patient delay in cancer (Pack & Gallo, 1938) defined Corresponding author. Tel.: +1 650 743 8631;
fax: +1 650 692 0141. E-mail addresses: [email protected] (N.C. Facione), [email protected] (P.A. Facione). 0277-9536/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2006.08.014
‘‘undue patient delay’’ in breast cancer as 3 months or more from symptom discovery to initial clinical examination, allowing researchers to track patient delay in breast cancer over time. Facione and colleagues proposed a model for predicting the decision to delay in breast cancer, separating the factors influencing symptom appraisal from those influencing the decision to delay diagnosis (Facione, Miaskowski, Dodd, & Paul, 2002). This and other studies support the observation that even women who appraise symptoms as likely to be cancer
ARTICLE IN PRESS 3138
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
related frequently make confident judgments to delay (Facione & Dodd, 1995; Facione, Dodd, Holzemer, & Meleis, 1997; Lauver, Coyle, & Panchmatia, 1995; Lierman, 1988). Delays of at least 3 months occur in one-third of all women diagnosed with self-discovered breast cancer, and 6-month patient delays occur in as many as 25% of cases (Facione, 1993; Richards, Westcombe, Love, Littlejohns, & Ramirez, 1999). Most studies link patient delay with more advanced staged breast cancer at diagnosis (Arndt et al., 2002; Harirchi, Ghaemmaghami, Karbakhsh, Moghimi, & Mazaherie, 2005; Nosarti et al., 2000). Decreased survival related to patient delay was quantified in one metastudy of 87 published reports (101,954 breast cancer patients) as a 12–19% decrease in 5-year survival in those women with delays of 3 months or more versus those with shorter times to diagnosis (Richards et al., 1999). A decision to delay diagnosis, particularly if one suspects cancer, might seem highly irrational. Clinicians commonly consider delay as a manifestation of denial, a minimization of threat to manage symptom-related anxiety (Cousins, 1982), a conclusion suggested by psychiatric studies of prolonged delay (Magarey, Todd, & Blizard, 1977; Wool, 1986). But Cousins (1982) questioned whether the term, was being used too liberally, or too judgmentally. Subsequent qualitative studies of patient delay have described numerous other factors related to patient delays of months or even years. The variables associated with patient delay are strikingly similar across cancer sites, involving similar social and cultural factors internationally. Poor access to health care, lack of preventive health care habits, being older and married, perceiving prejudice in care delivery, and having childcare/ elder care obligations are all significantly associated with patient delay (Arndt et al., 2002; Burgess, Hunter, & Ramirez, 2001; Harirchi et al., 2005; Ramirez et al., 1999; Rozniatowski et al., 2005). Similarly, intrapersonal factors including: benign attribution, poor education and lack of knowledge, misperception of risk, lower anxiety/optimistic bias, embarrassment, fear of chemotherapy or breast loss, concern about being an hypochondriac, and pessimism about survival are associated with patient delay (Bain & Campbell, 2002; Burgess et al., 2001; De Nooijer, Lechner, & De Vries, 2000; Facione & Dodd, 1995; Harirchi et al., 2005; Nosarti et al., 2000; Rozniatowski et al., 2005; Smith, Pope, & Botha, 2005). Hypothesized individual differences
(pessimism, anxiety and depression, external locus of control) have not been consistently observed in group comparison studies of patient delay (Bain & Campbell, 2002; Ramirez et al., 1999). However, this list of factors, aggregated by observing people who delayed cancer diagnosis, is neither unique to those who delay nor predictive of patient delay. The majority of women overcome varying clusters of these factors, deciding to seek immediate diagnosis (Richards et al., 1999). Women engage potential breast cancer symptoms much as they would engage other potential health threats that permit cognitive assessment. That is, they assess risk and weigh arguments for action. Some decide to delay even when they worry that the symptom is breast cancer (Andersen, Cacioppo, & Roberts, 1995; Burgess et al., 2001; Facione & Dodd, 1995; Smith et al., 2005). This study was undertaken to analyze the reasoning involved when symptomatic women weigh the need to seek a diagnosis, and achieve and sustain confidence in decisions to delay. Determining whether and when to seek diagnosis of breast symptoms is a decision that is purposeful, more or less well informed, reflective and of high consequence. Theories that attempt to explain this type of naturalistic decision model involve the interaction of two overlapping systems of reasoning that are active in human decision-making (Gilovich, Griffin, & Kahneman, 2002; Kahneman & Miller, 2002; Sloman, 2002; Zsambok, 1997). One is intuitive, non-reflective, associative, and holistic (sometimes referred to as System 1), and the other (System 2) is more deliberative, reflective, and rule governed. System 1 thinking is believed to rely heavily on heuristics (logical shortcuts), key situational cues, and salient memories to arrive quickly and confidently at judgments, particularly when situations are familiar and immediate action is required. System 2 thinking is believed to be useful for judgments in unfamiliar situations, for processing abstract concepts, and for deliberating when there is time for planning and more comprehensive consideration. Humans use heuristic shortcuts in System 2 as well, often as part of their argument making. Both systems are believed valuable, functioning interactively and in parallel to address problems. We suspend further discussion of this model due to space limitations, and direct the reader to a recent excellent anthology (Gilovich et al., 2002). Whether there are two reasoning systems or one complexly integrated reasoning system, it is clear
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
that we need to analyze and understand how humans use both argument making and heuristic thinking to make decisions such as the ones women make when they discover a change in their breast and consider whether it might be a signal of breast cancer. Risk and uncertainty complicate this decision, upping the potentially severe consequences of a faulty judgment. Studies report that women who fail to present self-discovered breast cancer for diagnosis prematurely dismiss their symptoms as benign or monitor them for a time as unclassified breast changes, most seeking diagnosis only when symptoms exacerbate (Coates et al., 1992; Schleicher & Ammon, 1998). There are no longitudinal studies documenting this decision process over time, but there are studies that suggest how such a potentially risky judgment might be sustained. Montgomery (1989) described the thinking process humans use to sustain risky and uncertain decisions. People form early preferences for how they would like to respond to a high stakes dilemma. They structure their arguments for the value of one chosen alternative while minimizing the values of the other alternatives not to be chosen. They may seek selectively for facts that support their choice. While this natural process may fall short on objectivity, there is increasing evidence that this type of maneuver is needed in order to find sufficient confidence to act. The ‘dominance structure’ supporting the chosen alternative bolsters and sustains confidence in the uncertain judgment, even when the negative consequences of error are high (Montgomery, 1989; Zsambok, 1997). Here we examine the on-going decision to delay a diagnostic visit in women with self-discovered breast symptoms to determine if there is evidence of dominance structuring in their decision process. This would seem a reasonable hypothesis because of the confidence level needed to sustain this high-risk decision given the uncertainty and possible risk posed by the symptoms. In this study, we examined symptomatic women’s reasoning for seeking diagnosis versus making an on-going judgment to delay. Our goal was to better understand how women could feel confident about judgments to delay the evaluation of breast symptoms that they themselves feared might be signals of breast cancer. Methods Symptomatic women from the San Francisco Bay area, a sub-sample of women recruited to a
3139
community-based survey study of breast health behavior (Facione et al., 2002) were recruited for in depth interviews regarding symptom-related decision-making. Data was gathered over a 6-month period in 1999–2000, after completion of the survey study. ‘Symptomatic’ was defined as an affirmative response to at least one of two survey items: ‘‘There is an area in my breast y (1) that I’m not sure is normal; or (2) that I worry may be breast cancer’’. All data for this study were collected and analyzed in accordance with institutional review board approval and with assurance of protection of human subjects. Thirty-nine women in the parent study fit the criterion of ‘symptomatic’. We successfully interviewed 28 (72%). Table 1 displays the characteristics of our interview sample. Women we failed to contact were more likely to be Spanish-speaking, not college educated, and born outside USA. Our interviewed sample contains an ample sub-sample of Spanish speaking subjects to provide cultural breadth in our findings. Spoken language and national origin were assumed to be neutral in Table 1 Sample characteristics Demographics Age Annual income
Language Education
Race/ethnicity
Sexual orientation Health care coverage Employment: Relationship status
Mean 42.34 years (s.d 14.1), range 24–72 years o$9,999 (n ¼ 7); $10,000–30,000 (n ¼ 8); $35,000–50,000 (n ¼ 4); $60,000–80,000 (n ¼ 4); 4$80,000 (n ¼ 5) English (n ¼ 18); Spanish (n ¼ 10) Grade school (n ¼ 3); high school/ vocational school (n ¼ 8); college (n ¼ 13); graduate school (n ¼ 4) White (n ¼ 12); Black (n ¼ 4); Latino (n ¼ 10); Asian (n ¼ 1); multi-ethnic (n ¼ 1) Heterosexual (n ¼ 22); lesbian or bisexual (n ¼ 3); missing (n ¼ 3) Insured (n ¼ 22); uninsured (n ¼ 6) Employed (n ¼ 16); unemployed or retired (n ¼ 12) Married or partnered (n ¼ 14); single or widowed (n ¼ 14)
Cancer-related variables Breast lump Yes (n ¼ 14); No (n ¼ 13); missing (n ¼ 1) Cancer symptom Yes (n ¼ 14); No (n ¼ 12); missing attribution (n ¼ 2) Close friend of Yes (n ¼ 11); No (n ¼ 17) relative with cancer
ARTICLE IN PRESS 3140
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
relationship to reasoning ability. Education is widely believed to increase the factual knowledge that one might used to inform a judgment. Our sample contains a majority of women with college education (61%), and thus the results we report may represent better overall knowledge about breast cancer than might be the case in a less well-educated sample. In-depth interviews were conducted in women’s homes or in conference rooms of community agencies. Women discussed their ideas about breast cancer early detection in general, and their current breast symptom(s) in particular. The interviews were conducted by trained research assistants in the woman’s native language. Each woman was asked about her thinking process without introducing her to new factual information, correcting faulty assumption or inferences, or evaluating her stated reasons or conclusions. This ‘talk-aloud’ methodology is described in detail elsewhere (Svenson, 1989). Women were reimbursed $30 for participating in the 90 min interview. At the completion of the interview, all were offered assistance in obtaining clinical breast examinations. All but three declined. Two of these women underwent biopsy and one was later treated for breast cancer. All interviews were transcribed for analysis by native English or Spanish-speaking research staff. The 10 Spanish language interviews were translated by two native Spanish-speaking researchers, and all data were analyzed in English for this decisionmaking study. Our data analysis method, described in Table 2, is grounded in analytical research describing applied logic, heuristic thinking and argument structure (Facione & Facione, 2001, 2006; Toulmin, 1969) and behavioral research describing heuristic reasoning (Montgomery, 1989; Kahneman, Slovic, & Tversky, 1982; Zsambok, 1997). This method examines the decision-making process embedded in participant interviews. Standard methods were used to demarcate the data to identify and retain the decision-making content. Typically, there is non-linearity in the oral presentation of human reasoning. For instance someone might say ‘‘I need to leave. My son is waiting, and my parking meter is running out’’. In this case the conclusion comes first, and the reasons follow. There may even be more reasons for ‘needing to leave’ given later in the dialogue. To analyze the complete argument accurately, these data segments need to be reordered. Arguments that are repeated later in the interview are not treated as
Table 2 Protocol for argument and heuristic data analysis Step 1 Prepares data for analysis
Step 2 Identifies main conclusions
Step 3 Identifies all relevant arguments
Step 4 Evaluates arguments for logical soundness
Step 5 Identifies possible heuristic errors
Step 6 Identifies dominance structuring
The data is parsed, retaining its ordered chronology, into units of analysis (typically a single communication: assertion, belief, comment, proposition, pause/silence, question). These units are complete sentences, parts of sentences or incomplete sentences, each contextually understandable even if not entirely grammatical. Identify the dilemma that is central to the thinking process and each of the alternative conclusions being considered. Identify the final conclusion, if the individual comes to a final judgment in the interview. All interview data asserting arguments in support of conclusions are ordered into their most logical format (argument strands). Argument strands are examined for redundancy and related content. Each argument strand is retained, regardless of how underdeveloped it may appear to be. This process continues until all relevant data is accounted for as a component of an argument supporting one of the identified alternative conclusions. Each argument is then analyzed for soundness (quality of reasoning process and accuracy of information content). Sound arguments rely on accurate premises and assumptions and warranted inferences to reach justified conclusions. Unsound arguments rely on false statements, misinformation, unwarranted inferences, misapplied analogies, fallacies, mistaken estimates or other errors to reach unjustified conclusions. Each reasoning strand is re-examined for use of heuristic strategies. These are often identified in the previous steps, but the data should be comprehensively searched again at this point. The appropriateness of relying on the heuristic to make the decision is evaluated in each instance. Argument strands are clustered into similar content areas to create sets of related reasons supporting each of the main conclusions. All argument strands are visually mapped to permit a holistic examination of the flow and progression of the decision-process. This map is examined for evidence of a dominance structuring around the final conclusion. Watershed claims are identified (statements that lead to arguments for more than one decision/conclusion.
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
new arguments, but if new content is introduced this content is used to elaborate the argument already advanced. The chronology of the data is retained to assist with the analysis of the overall structure of the decision process (Step 6). We examined the network of arguments, which women presented as the reasons (premises, hypotheses, assumptions) and claims (decisions, conclusions) for how they had responded to their selfdiscovered breast symptoms. We evaluated each argument for soundness in two distinct ways: first, for truth or falsity of the informational content of the reason given; and second for whether the claim being made was logically entailed, justified, or warranted by the reason given. Arguments are determined to be ‘warranted’ when they rely on an appropriate generalization (community norm or widely held belief) (Toulmin, 1969). Using contemporary argument evaluation categories, we labeled those arguments that passed these two tests as ‘‘sound’’, meaning that the informational content in the reason given was in fact true, and that the reason either warranted, justified, or logically implied the claim being made. But if the argument was based on misinformation or if the reasoning was not valid, the argument was evaluated as unsound. Finally we identified the women’s use of heuristic reasoning (their logical shortcuts). Concept mapping software (Inspiration Software, 1997) was used to visualize the argument structure within each interview. This software allows manipulation of data segments to designate their membership in argument strands, demarcation of the chronology of the data, and indication of where heuristic reasoning occurs (using colors, shapes, and directional arrows). We analyzed each interview and constructed maps of the arguments in each woman’s judgment process. Each woman’s decision map was examined to assess the presence of a dominance structure bolstering her decision to either delay or to seek diagnosis. Each of the two analysts began independently to develop an analysis of the argument strands contained in each interview and to construct an initial argument map for each case. The independent analyses were then examined collaboratively, case by case, for completeness, accuracy, and congruity. All data provided by the participant regarding symptom appraisal and decision about seeking a diagnostic visit is accounted for in the listing of argument strands and appears in each woman decision map. Finally the results of these
3141
analyses were collected across cases for presentation here. Evidence for the quality of the initial independent analyses was seen in the high congruity of the initial case analyses, but what is also required is a consensus in the completed logical, heuristic and structural analysis of each case interview. Comparative analyses resulted in the identification of a few additional argument strands and a clearer interpretation of how these clustered within each case. Consensus was reached on the basis of demonstrating clear evidence in the data transcriptions for each of the identified argument strands. In two cases investigators returned to the audiotapes to resolve ambiguities. Results Thirteen women (46%) reported having sought an immediate diagnostic visit. In nine cases, providers were monitoring their symptoms and four others had had benign biopsies. The remaining 15 women continued to monitor their symptoms, but had not been examined by a provider. All 15 had symptoms for at least 3 months and eight reported symptoms for a year or more (delayers). Nine of these (60%) had told no one of their symptoms. None of the delayers reported a lack of confidence in their on-going decision not to seek a diagnosis. Women reported having breast lumps (n ¼ 14, 50%), nipple discharge, firm areas, an enlarging breast, pain and itching. Women with breast lumps more often delayed (10 of 14, 71.4%) compared to those with other symptoms (4 of 13, 31%). Fourteen women (50%) reported thinking that their symptoms were probably symptoms of breast cancer, and eight of these were sustaining decisions to delay. Women who delayed were at least as well educated (66% college educated) as women who sought diagnosis (54% college educated). The mean income of both groups was $40,000 per year, and both groups were similarly distributed by race/ ethnicity. Women who delayed were on average 2 years younger (mean 41.7, standard deviation 714.5) than those seeking diagnosis (mean 43.1, 714.1). Heuristic analyses: Table 3 lists the names and descriptions of the heuristic shortcuts identified in these interviews. All but one woman in our sample demonstrated the availability heuristic, telling vivid stories of other women with breast symptoms and breast cancer to explain their own situation. These
ARTICLE IN PRESS 3142
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
Table 3 Heuristic reasoning shortcuts identified in these interviews Heuristic strategy
Description of cognitive maneuver
Potential consequences of misuse of this heuristic in the case of judgments to delay
Affect
Provides a strong and immediate sense of whether the issue at hand is ‘‘good’’ or ‘‘bad’’. Allows the thinker to take a stance for further examination of the issue. Vivid or personally salient past event is used as a basis to (over)estimate the likelihood of the event’s reoccurrence.
Strong negative feelings about breast cancer treatment or mastectomy make seeking diagnosis fearsome and repugnant.
Availability
Representativeness (associational) Representativeness (analogical), also called the Similarity heuristic
Associations of two linked terms or events. This cognitive shortcut might be used to orient further thinking about the topic. Similarities, whether superficial or fundamental, are relied upon as sufficiently representational for purposes of expecting similar outcomes or categorizations.
Satisficing
Deciding on an alternative, which is seen as adequate if not optimal.
Simulation
Imagining or projecting scenarios of ‘‘what will happen’’ if one makes alternative decisions or takes alternative actions as a way of predicting possible outcomes. The level of one’s involvement in an event is used to estimate one’s degree of control over the eventual outcome. Negative association of change with loss and risk. Decisions are aimed at conserving the status quo rather than suffering a loss.
Illusions of control
Loss aversion
stories were always told early in the interviews (first third), and the meanings that women attached to these stories became arguments in support of their ultimate decisions. Distressing stories of breast cancer death were recounted by seven women who sought immediate diagnosis. They emphasized unfortunate suffering and needless breast loss because the woman in their story delayed diagnosis or the physician initially misdiagnosed the cancer. They ended these stories by concluding that women need to seek diagnosis of symptoms as early as possible. Seven women who were delayers told stories that emphasized the pain of treatment and the inevitable death associated with a breast cancer diagnosis. They sometimes returned to these stories later in the interview and elaborated on them, adding additional arguments for the perceived benefit of delay. Use of the associational representativeness heuristic was seen in 14 (50%) women. Also occurring
False judgments as to the likelihood of consequences (e.g. a benign biopsy) can result in risky choices (to forego diagnosis of a new symptom). Associations like ‘‘breast cancer: death’’ or ‘‘breast cancer: breast loss’’ can result in a strong resistance to diagnosis. Over confidence in the apparent similarity of two cases (e.g. discovering a breast lump just as your friend did)) can lead to the belief that similar cases will have inevitably similar conclusions (e.g. death, even with diagnosis and treatment). One’s best option can be lost while one relies upon an approach that is a judged ‘‘good enough’’ (e.g. I can wait until my next scheduled visit). Imagining good (bad) outcomes can lead to the belief that the imagined outcomes will actually occur (e.g. Imagining the doctor laughing at my concern about this lump). Overestimated belief in one’s actual ability to control outcomes (e.g. Belief in the control of symptoms through healthy eating habits). Decisions that fail to foresee the consequences of inaction (e.g. acting to maintain control or avoid the sick role can result in greater morbidity).
early in the interviews, use of this heuristic occurred at the first mention of the related topic (cancer, treatments, breasts). Women seeking diagnosis saw cancer as representative of ‘death’ (n ¼ 3); lumps represented ‘cancer’ (n ¼ 3), treatments represented ‘suffering’ (n ¼ 1) and breasts represented ‘womanhood’ (n ¼ 1). These representations provided the energy to act on their symptom discovery by seeking diagnosis. Women who delayed also saw cancer as representative of ‘death’ (n ¼ 4), ‘ugly/ickiness’ (n ¼ 2) and ‘suffering’ (n ¼ 3), and one described her lump as ‘cancer’. However, these women did not conclude that they should seek diagnosis. Rather they tended to pause, recall episodic memories of prior provider visits, project negative consequences when simulating a future provider visit, and then simply say again that they did not need to seek a diagnosis. This cognitive activity was apparent from subsequent comments, such as: ‘‘I’ve had my symptoms dismissed as frivolous twice’’; ‘‘I’ve had
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
no relief from seeing a physician’’. And ‘‘Treatment means hospitalization, y pain’’. About half of the women used the loss aversion heuristic (Kahneman, Knetsch, & Thaler, 1991), making arguments to avoid loss of a breast, loss of a partner or spouse, or the loss of social standing as one’s normal, non-sick self. An example of this heuristic can be seen in the comment, ‘‘People take you as a poor, skinny victim’’. Heuristic thinking about loss aversion was seen both in those who delayed and in those who sought diagnosis. Misuse of the analogical representativeness heuristic was seen in the reasoning of four women who delayed. They claimed that friends’ or relatives’ experiences with benign biopsies were relevant to their own decision to delay: ‘My lump is just like hers, benign’. In contrast, in women who sought diagnosis, positive stories of well-treated cancer and benign biopsies were used to explain the advantages of seeking diagnosis. Two young women recounted stories of breast cancer in other women of their own age. Both correctly drew the analogy to their own vulnerability to breast cancer. One used the story as her explanation for ‘‘having no question about seeking immediate diagnosis’’. The other rather quickly discounted what she appeared to regard as a powerful heuristic insight, overriding this insight with 16 subsequent arguments for delay, 11 of them logically unsound. The affect heuristic was observable universally. All but one woman spontaneously spoke of cancer in strong negative terms such as ‘‘terrifying’’,
3143
‘‘ugly’’, ‘‘extremely frightening’’. Both groups of women said that discovering their symptoms was a fearful experience, and both were actively engaged in fear avoidance. Women who sought diagnosis used the fear to motivate seeking a diagnostic visit and described the visit as relieving uncertainty and anxiety. Women who delayed described experiencing anxiety, sadness and depression as well as fear. These feelings were associated with concerns about possible biopsy and breast loss. They voiced no expectation of emotional control or expected relief from a provider visit, and in some cases verbally denied this possibility, concluding that these negative emotions were best avoided by delaying. Logical analyses: Fig. 1 displays the analysis of arguments contained in the interviews by group. The distribution of sound and unsound arguments differed greatly between women who sought diagnosis and those who sustained a decision to delay. Those who sought diagnosis provided more than twice the number of arguments in support of this decision than women who delayed. These arguments were typically based on true beliefs and good reasoning. Women who delayed made nearly five times more arguments for delay than did women who sought immediate diagnosis. Many (24.1%) of the arguments for delay were sound, however, the vast majority were unsound, relying on misinformation (29.5%), poor reasoning (30.7%) or both (15.7%). Often women displayed confidence in poorly reasoned arguments containing a true statement. ‘‘A breast cancer lump is often painless. My lump is
Fig. 1. Argument analysis for the sample as a whole.
ARTICLE IN PRESS 3144
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
a bit uncomfortable. It can’t be cancer’’. This is a rather classic error of confusing ‘often’ with ‘always’. On occasion women would hear their unsound reasoning and offer self critique. More often, however, they appeared not to see the weakness of these arguments. Table 4 summarizes the most commonly advanced logical arguments both for seeking diagnosis and delay, grouped into the four categories of the logical analysis: sound; unsound because they rely on misinformation; unsound because they rely on poor reasoning; and unsound because they rely both on misinformation and poor reasoning. At first glance, some arguments coded as sound might appear to be unsound. For example, the argument ‘‘My husband/partner would leave me if I had a mastectomy’’, summarizes six women’s stories of conversations with male partners about breast biopsies, breast cancer, or mastectomy. The stories tell of negative effects of a breast cancer diagnosis on male–female relationships. Unspoken but clearly implied premises in each argument for delay were: ‘‘Being diagnosed with breast cancer would destroy my relationship with this husband/partner’’ and ‘‘This relationship is so important to me that I would risk death to maintain it’’. Taken together, these statements, if true in their particular circumstances, constitute a sound argument for a benefit inherent in delay. Similar premises, both spoken and implicit, were seen to be operative in many of the other sound arguments for delay listed in Table 4. For example, ‘‘I would risk death yto keep my breast intact, yto put off painful treatments, yto avoid incurring a large cost’’. Notice that while many of these arguments pass the logical test of soundness, they do not reflect what most people would take as wisdom in the overall judgment. It seems unwise to risk forfeiting one’s life to secure these proposed benefits. As a rule these arguments, while sound, were underdeveloped. Every interview contained arguments that relied on false information. Most women made inflated estimates of their personal risk of breast cancer. Every woman who had a relative with any type of cancer claimed to be at high risk for breast cancer, yet only three had first-degree relatives (mothers) with breast cancer, and none had high relative risk. These unsound arguments, because they supported seeking diagnosis, support the idea that people can make a good choice for a bad reason. False information about the causes of breast cancer
(abortions, bruising of the breast, wearing underwire bras, stress, and thinking of cancer) was used by delayers to argue that their own symptoms were therefore probably not cancer (for instance, never had an abortion, cannot remember being bruised) and therefore they could delay. Six women who delayed falsely asserted that cancerous lumps would be larger, or that cancer would produce a nipple discharge. Erroneous information about cancer protective behaviors involved breast feeding (I’ve heard it takes 14 years of breast feeding to be safe. I have two to go) and having children (I have no children, so I’ll probably get it). One common argument made by delayers was that they would be thought of as hypochondrical or foolish for presenting their breast symptoms for diagnosis. Some paused to relate stories about frustrating encounters with providers, and others rehearsed aloud (simulation) how they would call appointment schedulers, triage nurses, and physicians or nurse practitioners, be criticized or ridiculed about their breast symptom concerns, and then they concluded that it would be impossible to see their provider. Three women who did not delay also expressed concern about provider encounters, but one counter-argued that her health ‘should be more important than her doctor’s attitudes’ or ‘whether (the doctor) is in a hurry’. Several women misapplied normative generalizations (warrants) to bolster their arguments to delay. The most common misapplied warrant was: ‘Do good unto others’. Women who delayed applied this warrant as though it directed them to risk ignoring the potential threat posed by breast lumps and nipple discharge in favor of performing normal childcare and other family duties (n ¼ 4). Two women who sought diagnosis conversely applied this warrant correctly, saying that they had to seek diagnosis for the sake of their children: ‘‘They need their mother to be there’’; ‘‘I don’t want them to see me die’’. Five women misapplied the warrant ‘Do not burden others unnecessarily’ to justify delay, arguing that seeking diagnosis would worry family and friends. They ignored the potential for added burden on family if their symptoms proved cancerous and advanced. Others argued ‘‘I can wait until my next annual exam to check this symptom’’. Here the satisficing heuristic (corner-cutting to arrive at a minimally adequate solution to achieve a goal) worked in tandem with a misapplied warrant
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
3145
Table 4 The most commonly made arguments for seeking diagnosis and for delay Examples of most commonly made sound arguments
Sought diagnosis (n/%)
Seek diagnosis If you treat cancer early you can survive. My provider will competently manage this breast problem. Getting reassurance from a provider would be a relief. If this (breast symptom) is cancer I could die. I could have breast cancer. I would have a mastectomy to save my life. Relatives/friends with cancer have benefited from early detection.
11(85) 10(77) 8(73) 7(54) 6(46) 5(38) 4(31)
1(6) 4(27) 1(6) 4(27) 8(53) 2(13) 2(13)
4(31) 7(54) 0(0) 2(15)
11(73) 6(40) 5(33) 4(27)
3(23) 1(8) 0(0) 2(15) 0(0) 0(0) 1(8)
10(67) 10(67) 8(53) 7(47) 6(40) 5(33) 4(27)
6(46)
5(33)
7(54) 4(31) 2(15) 0(0) 0(0)
14(93) 14(93) 11(73) 8(53) 4(27)
2(15) 1(8) 0(0) 0(0) 1(8) 0(0) 0(0)
13(87) 10(67) 9(60) 8(53) 7(47) 7(47) 5(33)
Delay Treatments (surgery and chemo) are terrible, painful. Losing a breast is terrible to contemplate. Treatment is costly. I can’t afford medical care. My husband/partner will leave me if I have a mastectomy.
Delayed (n/%)
Examples of poorly reasoned arguments Delay I’ve had bad experiences talking with my doctor/disrespect/rudeness. I try to block thinking about it because it makes me upset. I can’t talk to my provider about most things. I just know that I won’t die from this. This isn’t how I’ll die. If you keep the symptom to yourself, you won’t be nagged about it. It’s just typical of me never to take care of myself. My lump is probably benign like my mother’s/my friend’s. Examples of arguments based on misinformation Seek diagnosis I’m at much higher risk than most women. Delay No one related to me has breast cancer. I’m not at risk. I can manage this by diet, exercise, smoking and drinking habits. Cancer is a death sentence. Once you find a lump it’s too late. I can leave it until my next health promotion visit. Thinking about cancer can make it happen. Examples of arguments based on poor reasoning and misinformation Delay Providers are incompetent. They will make the wrong diagnosis. I don’t want to be thought foolish or a hypochondriac. The symptom isn’t bad enough yet for me to worry about it. Everything can be possible with God’s help. I’ll try prayer instead. Scheduling an appointment is too difficult, intimidating, a hassle. Work and family is why I delay. I need to take care of these first. I don’t want to burden other people. Seeking diagnosis will do that.
(misapplication of the guideline ‘Have an annual clinical breast examination’). Conversely, women who sought diagnosis were not mistaken about what was minimally required, and pointed to the correct guideline: ‘If you find a lump, see your doctor immediately’.
At times women appeared to reason backwards from a norm (Stop smoking because it puts you at risk for breast cancer) to claims about controlling their symptoms. They argued: ‘‘Cancer occurs as a result of smoking, eating a poor diet, or not exercising. So I can solve this problem by stopping
ARTICLE IN PRESS 3146
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
smoking, eating bettery’’. Women seemed to be using reports of possible causal relationships between alcohol, tobacco, a poor diet and the development of breast cancer, drawing the false inference that they could reverse/eliminate a developing breast cancer by altering or ceasing these behaviors. They claimed control over the symptoms by: improving their diet (n ¼ 11), starting to exercise (n ¼ 4), stopping smoking (n ¼ 3), decreasing/eliminating alcohol intake (n ¼ 2), decreasing their stress (n ¼ 2), using homeopathy (n ¼ 3), and meditating (n ¼ 2). These arguments are examples of what has been termed ‘backwards thinking’, a quite common phenomenon that includes counterfactual thinking about how events might have been otherwise if one had only behaved as the norm would recommend (Kahneman & Miller, 2002). Counter-factual thinking becomes logically unsound when one intimates that events would certainly have been different in particular ways if the norm was followed, and in this case the women’s backwards reasoning also violates the logic of temporal causality. Others have described this type of thinking as a misapplication of the ‘illusion of control’ heuristic. Perhaps because the claim of causal control was rarely made explicitly, the unsound logic of the backwardleaping inference went unnoticed and instead bolstered decisions to delay in nine (60%) of the women who delayed.
Dominance structuring: Women described thinking again and again about their symptoms, and reaffirming their decision to delay each time the worrisome symptoms came to mind, which was in most cases a daily occurrence. They demonstrated this process in the interview as they thought aloud about their symptoms. Four manifestations of dominance structuring were observed in the data. First, no women were undecided about whether or not they would seek diagnosis of their breast symptom. All were either firm in their decision to seek diagnosis, and in fact had done so, or were sustaining their judgment to delay. Second, women tended to dismiss as unimportant, refute, or simply abandon strands of reasoning which appeared to be leading to the alternative decision that they did not embrace. Third, examination of the chronology of the data demonstrated the occurrence of numerous, less well elaborated, bolstering arguments later in the interview. Fourth, dominance structuring could be seen visually by examining the argument maps. Fig. 2 is a map of one woman’s interview, each oval representing a summarized strand of argument. She abandoned three sound arguments for seeking diagnosis, and took time to provide a counterargument to two additional sound arguments (watershed claims). In all, she makes 18 arguments for delay, 17 of them unsound. None of these arguments clearly grounds her decision to delay, but
Fig. 2. One woman’s summarized argument map.
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
taken together there is confidence in this choice as dominant. Discussion This study is the first to analyze the arguments sustaining a decision to delay diagnosis made by women currently monitoring potential breast cancer symptoms. In general, women’s reasons for delay were not new or surprising, many having been previously reported as attitudes or beliefs associated with patient delay (Bish, Ramirez, Burgess, & Hunter, 2005; De Nooijer et al., 2000; Facione & Giancarlo, 1998; Facione et al., 2002; Lauver et al., 1995; Ramirez et al., 1999). This leads us to believe that our sample was a good source of information about women’s decisions to delay. Here we show how these variables are used in arguments to delay, the magnitude of arguments made, and how these arguments are used together to sustain confidence in the decision to delay. Through the use of argument and heuristic analysis we have described how decisions to seek diagnosis differ from decisions to delay. Delayers exhibited many more logical errors, including drawing faulty inferences, claiming false facts to be true, and misapplying rules and guidelines. Delayers were also less than appropriately attentive to their poor reasoning and critical self-commentary. Common heuristic thinking errors included the misuse of availability, representativeness, loss aversion, satisficing, and the illusion of control heuristics. Both types of errors put these women at greater risk for advanced cancer at diagnosis should any symptom prove malignant. Both groups of women made arguments for delay that passed the logical test for soundness, but women who delayed were observed to rely on these arguments and to abandon or superficially dismiss other sound arguments for the necessity of seeking prompt diagnosis. In contrast to those who sought diagnosis, they added numerous sound and unsound (bolstering) arguments to support their decision to delay, creating a dominance structure to support their risky and uncertain decision. Some might call this denial, but there was nothing passive or subconscious about this cognitive process in the women we interviewed. Rather, dominance structuring around the decision to delay required considerable on-going effort. That they elected to participate in this breast health study about their decision process, and that they continued to
3147
consider sound arguments to seek diagnosis, may be an important finding suggesting the potential for targeted, guided reasoning interventions. But the resilience of their decisions to delay should not be underestimated (Montgomery, 1989). We observed many abandoned sound arguments for seeking diagnosis and a vast network of arguments supporting risky judgments for delay. The question of whether the dominance structure supporting a judgment to delay becomes more elaborate over time as women work to sustain their delay of diagnosis cannot be answered by the current study. We observed that the women more frequently repeated arguments for delay, but we cannot say whether these repeated arguments were more important. Perhaps they initially anchored women as ‘likely to delay’ versus ‘likely to seek diagnosis’, but they may only represent arguments not confronted by others or arguments the women believed would be persuasive to others. To attain our goal of describing the complete argument for a judgment to delay, our study design precluded challenging women’s unsound arguments or correcting their misuse of heuristics. As a result, we cannot determine whether women would have held tightly to particular arguments, reformulated arguments, or even rescinded their judgments to delay as a result of being specifically asked to reexamine their thinking process. These questions invite intervention studies. The dominance structures around judgments to delay should not be discounted as the self-serving rationale of women who are deliberately refusing medical management. Many women in this sample believed themselves to have thought well about whether to seek diagnosis, and to have arrived at a decision to delay reasonably. It is their sense that they had made a good decision that should be worrisome to health professionals. We found women’s stories of alienation from the health care system, and their concerns about being judged hypochondrical to be particularly disturbing. This is not the first study documenting anxiousness surrounding the presentation of self-discovered symptoms to providers (Bish et al., 2005; LaVeist, Rolley, & Diala, 2003; Reece, 2003), and directly addressing this issue might improve early detection. The women in our well-educated sample often lacked specific breast cancer symptom knowledge, and were able to abandon sound arguments for seeking prompt diagnosis. When a person uses false information to evaluate how they should respond to
ARTICLE IN PRESS 3148
N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149
a dilemma they are likely to arrive at an unsound conclusion that rings of subjective truth. Reaching women who are currently monitoring breast symptoms with effective media messages may require targeting one or more of the arguments she makes for delaying symptom evaluation. We hypothesize that framing these messages in relation to her thinking strategies might better impact the dominance structure she has likely created around her decision to delay. Our suggestions for future intervention research aimed at decreasing patient delay would include studies targeting multiple aspects of the dominance structure by: (1) challenging misapplied warrants about the relative importance of cancer early detection, (2) correcting an array of misinformation with accurate facts about cancer symptoms and treatment, (3) overcoming fears of possible negative encounters with providers when seeking help, (4) exposing common reasoning mistakes that diminish the consequences of delay, (5) reframing mistaken stories of other women’s experience with breast cancer, and (6) encouraging a more reflective self critique regarding illusions of control, and satisficing in relationship to scheduling a diagnostic visit. References Andersen, B. L., Cacioppo, J. T., & Roberts, D. C. (1995). Delay in seeking a cancer diagnosis: Delay stages and psychophysiological comparison processes. British Journal of Social Psychology, 34, 33–52. Arndt, V., Sturmer, T., Stegmaier, C., Ziegler, H., Dhom, G., & Brenner, H. (2002). Patient delay and stage of diagnosis among breast cancer patients in Germany—a population based study. British Journal of Cancer, 86(7), 1034–1040. Bain, N. S., & Campbell, N. C. (2002). Treating patients with colorectal cancer care: A qualitative study of rural and urban patients. Family Practice, 19, 360–374. Bish, A., Ramirez, A., Burgess, C., & Hunter, M. (2005). Understanding why women delay in seeking help for breast cancer symptoms. Journal of Psychosomatic Research, 58, 321–326. Burgess, C., Hunter, M. S., & Ramirez, A. J. (2001). A qualitative study of delay among women reporting symptoms of breast cancer. British Journal of General Practice, 51, 967–971. Coates, R. J., Bransfield, D. D., Wesley, M., Hankey, B., Eley, J. W., Geenberg, R., et al. (1992). Differences between Black and White women with breast cancer in time from symptom recognition to medical consultation. Journal of the National Cancer Institute, 84, 938–950. Cousins, N. (1982). Commentary: Denial. Journal of the American Medical Association, 248(2), 210–212. De Nooijer, J., Lechner, L., & De Vries, H. (2000). A qualitative study on detecting cancer symptoms and seeking medical
help: An application of Anderson’s model of total patient delay. Patient Education and Counseling, 42(2), 145–157. Facione, N. C. (1993). Delay versus helpseeking for breast cancer symptoms: A critical review of the literature on patient and provider delay. Social Science & Medicine, 36(12), 1521–1534. Facione, N. C., & Dodd, M. J. (1995). Women’s narratives of helpseeking for breast cancer. Cancer Practice, 3(4), 219–225. Facione, N. C., Dodd, M. J., Holzemer, W., & Meleis, A. I. (1997). Helpseeking for self-discovered breast symptoms: Implications for cancer early detection. Cancer Practice, 5(4), 220–227. Facione, N., & Facione, P. (2001). Analyzing explanations for seemingly irrational choices: Linking argument analysis and cognitive science. International Journal of Applied Philosophy, 15(2), 267–286. Facione, N. C., & Facione, P. A. (2006). Argument and Heuristic Analysis: Method, Theory and Application. (pp. 35–48). Millbrae, USA: California Academic Press. Facione, N. C., & Giancarlo, C. A. (1998). Narratives of breast symptom discovery: Implications for early detection. Cancer Nursing, 21(6), 430–440. Facione, N. C., Miaskowski, C., Dodd, M. J., & Paul, S. M. (2002). The self-reported likelihood of patient delay in breast cancer: New thoughts for early detection. Preventive Medicine, 34, 397–407. Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and biases: The psychology of intuitive judgment. Cambridge, UK: Cambridge University Press. Harirchi, I., Ghaemmaghami, F., Karbakhsh, M., Moghimi, R., & Mazaherie, H. (2005). Patient delay in women presenting with advanced breast cancer: An Iranian study. Public Health, 119(10), 885–891. Inspiration Software (1997). Inspiration Software, Inc., Version 6, www.inspiration.com. Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). The endowment effect, loss aversion, and status quo bias. Journal of Economic Perspectives, 5, 193–20644. Kahneman, D., & Miller, D. T. (2002). Norm theory: Comparing reality to its alternatives. In T. Glovitch, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 348–366). Cambridge, UK: Cambridge University Press. Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under uncertainty: Heuristics and biases. Cambridge, UK: Cambridge University Press. Lauver, D., Coyle, M., & Panchmatia, B. (1995). Women’s reasons for and barriers to seeking care for breast symptoms. Women’s Health Issues, 5(1), 27–35. LaVeist, T. A., Rolley, N. C., & Diala, C. (2003). Prevalence and patterns of discrimination among US health care consumers. International Journal of Health Services, 33(2), 331–344. Lierman, L,M. (1988). Discovery of breast changes. Cancer Nursing, 11(6), 352–361. Magarey, C. J., Todd, P. B., & Blizard, P. J. (1977). Psychosocial factors influencing delay and breast self-examination in women with breast symptoms of breast cancer. Social Science & Medicine, 11, 229–232. Montgomery, H. (1989). From cognition to action: The search for dominance in decision making. In H. Montgomery, & O. Svenson (Eds.), Process and structure in human decisionmaking (pp. 23–49). Chichester, UK: Wiley.
ARTICLE IN PRESS N.C. Facione, P.A. Facione / Social Science & Medicine 63 (2006) 3137–3149 Nosarti, C., Craford, T., Roberts, J. V., Elias, E., McKenzie, K., & David, A. S. (2000). Delay in presentation of symptomatic referrals to a breast clinic: Patient and system factors. British Journal of Cancer, 82(3), 742–748. Pack, G. T., & Gallo, J. S. (1938). The culpability for delay in the treatment of cancer. American Journal of Cancer, 33, 443–462. Ramirez, A. J., Westcombe, A. M., Burgess, C. C., Sutton, S., Littlejohns, P., & Richards, M. A. (1999). Factors predicting delayed presentation of symptomatic breast cancer: A systematic review. Lancet, 353(9159), 1127–1131. Reece, M. (2003). HIV-related mental health care: Factors influencing dropout among low-income, HIV-positive individuals. AIDS Care, 15(5), 707–716. Richards, M. A., Westcombe, A. M., Love, S. B., Littlejohns, P., & Ramirez, A. J. (1999). The influence of delay on survival in patients with breast cancer: A systematic review. Lancet, 353(9159), 1119–1126. Rozniatowski, O., Reich, M., Mallet, Y., Penel, N., Frounier, C., & Lefebvre, J. L. (2005). Psychosocial factors in delayed consultation by patients with head and neck cancer. Head & Neck, 27(4), 274–280. Schleicher, U. M., & Ammon, J. (1998). Mode of breast cancer detection: A study from the German part of the Maas-Rhine-
3149
EUREGIO. European Journal of Cancer Prevention, 1, S41–S46. Sloman, S. A. (2002). Two systems of reasoning. In T. Glovitch, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 379–396). Cambridge, UK: Cambridge University Press. Smith, L. K., Pope, C., & Botha, J. L. (2005). Patients’ helpseeking experiences and delay in cancer presentation: A qualitative synthesis. Lancet, 366(9488), 825–831. Svenson, O. (1989). Eliciting and analyzing verbal protocols in process studies of judgment and decision-making. In H. Montgomery, & O. Svenson (Eds.), Process and structure in human decision-making (pp. 65–81). Chichester, UK: Wiley. Toulmin, S. (1969). The uses of argument. Cambridge, UK: Cambridge University Press. Wool, M. S. (1986). Extreme denial in breast cancer patients and capacity for object relations. Psychotherapy and Psychosomatics, 46, 196–204. Zsambok, C. E. (1997). Naturalistic decision-making: where are we now? In C. E. Zsambok, & G. Klein (Eds.), Naturalistic decision making (pp. 3–16). Mahwah, NJ: Lawrence Erlbaum Associates.