Fear, Anxiety, Worry, and Breast Cancer Screening Behavior: A Critical Review

Despite some recent controversy (1), the bulk of the current literature confirms that breast cancer screening is an effective means for reducing breast cancer mortality (2–7). While screening rates are improving, a significant number of women are not screening at nationally recommended rates (8); the American Cancer Society recommends yearly mammograms for women over the age of 40 (9).

Screening rates have been linked to a wide range of social factors including background variables such as age (10–13), socioeconomic status (SES), and education (14–17) as well as other proximal indices of SES such as medical insurance (13, 18). Likewise, marital status (10), physician recommendation (4, 16, 19–21), and minority status (2, 22–25) may also play important roles.

Despite the volume of this research, there are good reasons to explore the impact of other variables. First, structural and demographic factors such as age, income, marital status, and ethnicity cannot be directly or easily modified. Hence, although the study of these variables can help identify those at risk for a poor screening profile, such research offers little direction in terms of viable interventions. Second, while physician recommendation remains critical, and women claim that they are more likely to get a mammogram if their physician recommends it (4, 19, 21, 26), most women still do not screen frequently enough although rates of physician recommendation are high (27), even among low income, minority groups (16). It seems increasingly clear that if we are to improve screening rates, we must develop interventions that target variables that are both amenable to change and for which there is room for improvement.

In the current review, we focus on research examining how the construct variously called “fear,” “anxiety,” or “worry” relates to breast cancer screening behavior. There are four reasons for this focus. First, emotions are well-documented motivators (28, 29) and are central to models of both self-regulation and health behavior (30–33) as well as models regarding the “uptake” of health-promoting messages (34–37).

Second, emotions and emotion regulatory styles differ systematically across ethnic groups; these differences may have important implications for explaining differences in screening behavior (16). Compared to European Americans, African Americans report lower negative emotion and stress (38–41), lower depression (42–44), and less grief following interpersonal losses (45). They are less likely to disclose in therapy (46, 47), may be less emotionally expressive during conflict than European groups (38), and manifest global differences in coping with stress (48).

Third, recent views of ethnicity have suggested that viewing it exclusively as a proxy for SES, access to services, or education is not useful (38, 49–51). Ethnicity is most usefully viewed as a global factor that corresponds to a pattern of social and individual factors (52); recent research suggests that differences in emotions and emotion regulation may explain substantial portions of the variance otherwise “explained” by ethnic group membership (53, 54).

Lastly, and perhaps most importantly, the literature linking emotion variables to breast cancer screening behavior is growing larger and is more fragmented. Researchers are investigating a range of divergent and atheoretical “emotion” constructs, and few connections between the various research programs are apparent. In the current review, we focus on the most commonly researched emotion variable, variously called cancer fear, anxiety, or worry, as well as the manner in which fear is regulated. We outline the extant literature before identifying trends and discrepancies and offer some preliminary interpretations and directions for future research. As will become clear, the state of the available literature is such that a more sophisticated meta-analytic approach is not possible.

In the context of screening behavior, the construct variously called anxiety, fear, or worry remains the most extensively studied emotion variable. Currently, however, it is unclear whether fear acts as a barrier or facilitator of screening (see Table 1). On the one hand, a fear of cancer and the medical establishment has been linked to poorer screening (55, 56), particularly among African American women (57). A fear of “finding something wrong” has been cited as a key obstacle to screening among both Hispanic (58) and Black (59) groups and a feeling that “it is better not to know” is a reported barrier among several European samples (60–62). Comparisons of screeners and nonscreeners have suggested that trait anxiety was significantly more pronounced among women who did not attend a screening in response to an invitation than among those who did (60), although other research has indicated no relation between screening behavior and trait anxiety (63).

Other studies, however, have suggested that greater fear(s) is associated with a higher likelihood of screening (10, 16, 63–65), perhaps for both intentions and actual behavior (63). One Swedish study of 434 nonattenders and 515 attenders showed that the women who worried most about breast cancer were more likely to screen (65). Another study of 1384 U.S.-born European and African American women, together with samples of English-speaking Caribbean, Haitian, Dominican, and eastern European women in New York City, showed that worry about cancer was positively associated with mammography and clinical breast examination frequency, even when background variables such as education and SES were controlled (16). Women who report recent symptoms of anxiety have been shown to be more likely to accept an invitation for a screening (10, 66) and those with a lifetime history of phobic disorders tend to present earlier with breast symptoms (67), again, even when demographics are controlled. The two prospective studies that are available (68, 69) are equally conflicted with one suggesting improved screening at intermediate levels of worry (68) and the other suggesting a negative relation between worry and mammography (69).

Emotions like fear, anxiety, and worry rarely occur in an unregulated state (Table 2). Because of their strong motivational qualities, people are impelled to regulate most emotional experiences (28). Several studies have linked a form of emotion regulation termed “denial” with delays in responding to possible breast cancer symptoms (64, 70, 71). One study, in which the coping styles of 100 women were assessed before biopsy, found that women who did not view a lump as a cause for concern (“nonidentifiers”) all used denial as a coping mechanism; they were thrice more likely to use avoidant coping mechanisms than identifiers (72). Other research has shown that beliefs indicative of realism and a willingness to directly face breast problems—attributes that appear antithetical to denial—were more prominent in women who screened compared with those who did not (73). In contrast, however, two studies have reported findings showing that the emotion regulatory style of repression may be associated with improved screening. One study compared 210 women who self-referred for screenings with 210 nonattenders and found that those who self-referred evinced higher repressiveness than nonattenders (74). Most recently, a study of 1364 women has also found that greater repression was associated with increased mammography and clinician breast examinations even when demographics and background characteristics were controlled (16).

The literature on cancer-related fear, worry, or anxiety, emotion regulation, and screening behavior is growing. As the above review makes clear, however, data are frequently contradictory, making it difficult to coherently interpret them and thus develop practical solutions and interventions. In offering some interpretations and preliminary explanations for discrepancies, we focus on four interrelated issues: (a) the consideration of the target (and extent) of women's anxiety or fear, (b) the need for consensually accepted definitions and measurements of emotions, (c) the lack of explicit theoretical models relating emotions to screening behavior, and (d) the issues of generalizability, particularly across age and ethnic groups as well as across different stages of the cancer screening process.

The most salient problem confronting researchers interested in the role of emotions in screening behavior is that they cannot determine exactly what women are afraid of or how these diverse fear components relate to one another or to screening behavior. Much research leaves the object of women's fear unspecified. Women's fears surrounding breast cancer seem to encompass nearly “everything” but certainly include fear of a breast cancer diagnosis (58–62, 75), fear of pain/discomfort (60), and, more complicating, fear of embarrassment. To this list, we can add fear of the medical establishment (55, 75, 76), radiation (60), nonspecific “cancer worry” (10, 16, 64, 65), general anxiety (10, 66), or phobia (67).

Given this state of affairs, clearly identifying the elements of cancer and the screening process that women are most afraid of and how these anxieties relate to screening behavior are important research mandates. The model below offers a preliminary framework for organizing how diverse fears and anxieties may relate to screening behavior.

In the model, we identify three discriminable sources of anxiety: (a) fear of screening components, (b) fear of screening outcomes, and (c) undifferentiated cancer fear. However, although we think these components empirically separable, they are likely heavily interactive as indicated by the double-headed arrows. Because of this likelihood, the model suggests that fear has diverse effects on screening behavior, although we note a generally inverse relation between fear specificity and likelihood of screening.

In terms of screening outcome, the available data seem to suggest that undifferentiated fear or anxiety regarding “getting cancer” may generally be facilitative of screening, at least where the fear occurs within manageable limits (see below), and the target population has access to, and some faith in, the efficacy of the available screening procedures. As has been noted by many authors, treatment efficacy perceptions are important factors in screening behavior (24, 77–79) although their relations with emotional responding and coping remain poorly understood. Studies based on idiosyncratically developed instruments have tended to return a positive relation between nonspecified cancer worry and screening behavior (16, 63, 65, 66, 80, 81). Conversely, however, fear of the medical establishment and of the pain associated with mammography, fear of radiation, and fear that the procedure will prove embarrassing seem likely to deter screening, as does a fear of a breast cancer diagnosis. Operationally, these latter studies have tended to target more specific worries such as “pain at mammography,” “fear of radiation,” “fear of disease,” or “fear of hospitalization and/or surgery” (59, 60, 64). Fear of negative outcomes appears to occupy a middle ground in terms of its impact on screening behavior. As such, its role in generating or inhibiting screening behavior may depend on the relations between this type of fear and other fears (see Fig. 1).

Our suspicion is that more specific cancer-related worries enable and encourage coping strategies that do not involve screening and thus do not generate the same impetus to act that more generalized cancer worries do. In many cases, specific fears are often about components of the screening process itself. For component-specific fears, avoiding screening situations will reduce anxiety and the individual may be less likely to screen. In contrast, where the anxiety is about cancer itself or is more diffuse, the feelings are more difficult to “pin down” and avoid. Acting to reduce generalized cancer worry may, in many cases, involve engaging in a screening behavior as the individual seeks to reduce their anxiety. However, these interpretations must be taken as preliminary, and researchers must begin to clearly define the aspects of cancer and the cancer screening process. Without knowing the sources of women's fears, much less their relation to outcome behavior, providers and health professionals interested in the development of interventions will be inevitably ill informed.

In many ways, the issue evident above stems inextricably from a more general issue involving the need to develop theories regarding why the anxieties and fears that revolve around breast cancer and breast cancer screening processes relate to screening. Researchers need to more systematically develop, test, and operationalize emotion variables in this domain (69), preferably by engaging with established models of emotions and health (30, 31, 53, 82, 83). Well-developed models of emotion have the potential to clarify complex literatures by offering frameworks within which to consider when anxiety will (and will not) facilitate screening. Here, we focus on two aspects of this overarching issue.

First, research tends to assume that the relation between fear/anxiety/worry and screening behavior is linear when, in fact, there are some grounds to suspect that the relation may be more complex. It may be, for example, that moderate levels of anxiety induce preventive care behaviors (76) while too little promotes inactivity and too much promotes avoidance of both the anxiety and the settings that elicit it. There are few data that bear on this possibility, although one recent study of 6512 predominantly Caucasian (97%) women demonstrated that women who described moderate levels of worry were more likely to have an annual mammogram than those who reported mild or severe levels (84). Operationally, this possibility may be reflected in differential relations occurring when “fear” (55, 58, 59, 61, 62, 75, 76) versus “worry” (10, 16, 65) are used. “Fear” measurements seem likely to index ratings within a high range that may be dysfunctional in the upper extreme, turning women to repressive regulatory styles or avoidance, while “worry” may access a lower, more manageable range of the affect, with lower levels indicating either apathy or avoidance and greater levels of “worry” indicating a level of anxiety conducive to appropriate vigilance and self-care (69).

Second, there is a pressing need to systematically operationalize and report such operationalizations in the measurement of fear, anxiety, and worry in this important public health domain. The studies summarized in Table 1 reveal a wide range of fear and anxiety measurement methods; they diverge in terms of both the precise source of anxiety they target and in how psychometrically robust the measurements used are. Studies targeting “breast cancer worry” (e.g., Refs. 16, 65, 68, 76) do not always identify how many items from a large questionnaire are used in a worry subscale (65, 76) and sometimes equate negative cancer attitudes with anxiety (76). Other studies use only a single item to measure breast cancer worry (68)—Pap smear, mammography, or breast self-examination worry (85)—and most fail to report reliability coefficients for multi-item scales. Single-item measures are notoriously unreliable and only four studies report reliability coefficients (16, 63, 65, 80): 0.73, 0.85, 0.79, and 0.71 in these studies, respectively. Interwoven with these reliability issues are other psychometric issues regarding scale validity and the precision of terms. Some studies measure general anxiety (72, 74, 86) and some do not make it clear what is being measured (10). For example, without describing actual scale items regarding “breast cancer worry,” it is difficult to determine if the measure is telling us more about the frequency or severity of cancer worries. Studies frequently employ concepts such as “risk” and “worry” interchangeably when models of health behavior suggest they may have distinct predictive and theoretical value.

Overall, psychosocial approaches to breast cancer screening are being hampered by the lack of adequate instrumentation. In many cases, the assessment of anxiety, either about cancer or the screening process or a more general affective trait, does not appear to have been a focus of the study with a consequence that psychometric properties are poor. Perhaps as a result, authors have continued to develop their own sets of questions, inadvertently perpetuating an increasingly fragmented field. Because the interests of different research groups diverge, their measurements assess different facets of anxiety, making links between the different research programs difficult to discern. The need for multiple-item assessment and the description of items is likewise pressing as reliability and validity concerns remain key issues.

Lastly, the above literature pays little attention to, and infrequently tests, the generalizability of models relating fear, fear regulation, and screening behavior across ethnic groups, despite the presence of both theory (28) and data (53, 54), suggesting interethnic generalizability to be unlikely. Cancer researchers clearly need to begin systematically testing their models for generalizability, particularly across age and ethnic groups. Cancer research (16, 66, 87, 88) as well as research in several other areas show that emotions and emotion regulation both vary as a function of ethnicity (38–47).

Furthermore, barriers to mammography in asymptomatic Black and White women differ, with Black women more likely to report worry and fear as barriers and Whites more likely to report inconvenience, procrastination, and being too busy (59). Black women report higher scores of cancer anxiety than the majority population and are less likely to seek mammography (76). Such data are compatible with two interpretations. One likely possibility is that African American women are more afraid of cancer than Caucasian women. Complementarily, the screening behaviors of these minority women may be more strongly inhibited by fear. It has been suggested that contradictions in the relations between emotion regulation and screening may reflect changes in the way denial operates at different stages of cancer and the testing process (27). Our suggestion is that the same is likely to prove true of how cancer-related fears affect the screening behavior of women from different ethnic groups. Researchers need to use established theory, measurement tools, and analytic strategies more systematically to examine the possibility that these variables impact screening behavior differentially among distinct groups of women.

It has recently been suggested that differences in health between African American women and Caucasian women are caused by African American women tending to put their health at risk at the service of regulating negative emotions (88). The eating of “comfort foods” (typically of high calorie, sugar, and fat contents) as well as smoking and drinking are short-term solutions to stress and negative emotions but have a deleterious effect on long term-health. Although this claim has not yet been empirically investigated, differences in emotional experience and emotional regulation across ethnicities remain worthy of future investigation, especially concerning their influence on care-seeking behavior (38, 53).

In addition to further empirical work, much scholarly and conceptual analysis remains necessary. If the contradictions in the data describing fear, anxiety, worry, and screening are to be resolved, the conceptual ambiguities responsible for these contradictions must be clarified. The similarities and differences between the various emotion and emotion regulation constructs must be more clearly spelled out so that the findings accumulated by researchers studying these constructs can be more coherently related. Such conceptual distinctions will pave the way for more rigorously differentiated research studies. As is implied above, the study of denial and repression as distinct variables may allow us to develop a clearer and more coherent understanding of the relationship between emotion regulation and screening behavior. Empirically differentiating the impact of variables such as fear, worry, and anxiety might be similarly helpful. In short, breast cancer screening researchers should begin to articulate a finer-grained body of data, which may then offer us clearer and more specific ideas about how health care practitioners can intervene.