Health anxiety and cognitive interference: Evidence from the application of a modified Stroop task in two studies

Abstract

The number of studies using a modified Stroop task in the examination of health anxiety-related cognitive biases is limited, and their results are divergent. The use of research methods that preclude conscious processes is, however, essential. The purpose of the present two studies was to examine whether health anxiety is associated with information processing biases towards illness-related stimuli, as well as whether health anxiety interacts with perceived physical symptoms regarding such biases. In both studies a modified Stroop task was implemented. Fifty-one healthy university students in the first study, and 69 in the second study completed a modified Stroop task and filled out questionnaires regarding mood, state anxiety, health anxiety and perceived symptoms. According to the results of both studies, after controlling for positive and negative mood and state anxiety, individuals with higher levels of health anxiety displayed greater information biases towards illness-related words. Moreover, health anxiety did not interact with perceived physical symptoms, as shown in the second study. The implications of these findings are significant, not only as far as health anxiety is concerned, but also for everyday health-related behaviour.

Introduction

Cognitive-affective representations about health and illness are known to be strong predictors of health-related behaviour (Cioffi, 1991, Martin and Leventhal, 2004, Smith and Ruiz, 2004). One such cognitive-emotional factor associated with everyday health-related decision-making and behaviour is anxiety about personal health and the possibility of being ill, or health anxiety. Herein, we present two studies that focus on health anxiety and its relation to cognitive processing of health-related information.

The fear of having an illness, that is health anxiety, is a dimensional construct ranging from mild fear to clinical hypochondriasis (Salkovskis & Warwick, 1986). According to the cognitive-behavioural model, variables such as the cognitive processing of information and cognitive appraisals are significantly involved in the development and maintenance of health anxiety (Warwick & Salkovskis, 1990). Internal information (e.g., bodily sensations, physical symptoms) or external illness-related stimuli (e.g., hearing about an ill friend) have the potential to activate dysfunctional cognitions about health and illness. These in turn result in higher anxiety about personal health and affect health behaviours and decisions (e.g., they provoke an even more catastrophic interpretation of illness-related information; Salkovskis and Bass, 1997, Warwick and Salkovskis, 1990). Several studies, both correlational and experimental (e.g., Hadjistavropoulos et al., 2000, Marcus and Church, 2003, Marcus et al., 2007, Rief et al., 1998, Smeets et al., 2000) have offered considerable support to this model.

As the cognitive-behavioural model suggests, biased cognitive processing of internal or external illness-related information may be the start point for the elevation of health anxiety (Kellner, 1986, Warwick and Salkovskis, 1990). In fact, several studies have shown that, when people are threatened, they display significant information biases towards information related to the source of their concern (e.g., Block, 2005, Williams et al., 1996).

For the examination of informational processing, the use of specific research methods (such as visual probe tasks or perception of difficult to read words), which will preclude conscious processes that may interfere with the results (like self-report methods), is needed. Still, the number of studies on health anxiety that use such methods is quite small (Williams, 2004) and their findings are divergent. For instance, Durso, Reardon, and Shore (1991) showed that health anxiety is related to memory biases. Hitchock and Mathews (1992) found that there is an enhanced attentional sensitivity to illness-related information, while Pauli, Schwenzer, Brody, Rau, and Birbaumer (1993) also found attentional bias towards bodily symptoms in high health anxious individuals. More recently, Pauli and Alpers (2002) found memory and attentional biases in patients with somatoform disorders. On the other hand, Brown, Kosslyn, Delamater, Fama, and Barsky (1999), in two samples of hypochondriac patients and healthy or non-hypochondriac controls, did not find strong evidence for perceptual and memory bias. Further, Lees, Mogg, and Bradley (2005) found that high health anxious participants did not show greater attentional bias for health-related cues than low anxious participants.

A commonly used method to assess the extent to which individuals exhibit a non-conscious cognitive processing bias to specific threatening stimuli is the emotional Stroop task (MacLeod, 1991). The emotional Stroop task has been developed as a modification of the original Stroop (1935) procedure. During this task, individuals are asked to name the color of a word presented to them as quickly and as accurately as possible, while ignoring the meaning of the word. Participants are expected to show delays or make more errors in color-naming for words related to their concerns compared to neutral words (referred to as the “interference effect”). Interference is used as an indicator of processing biases towards a particular set of stimuli (words).

Studies using the emotional Stroop task have identified biases, for example, in emotional disorders (Coles and Heimberg, 2002, Williams et al., 1996) in post traumatic stress disorder (Buckley, Blanchard, & Neill, 2000), in women with a family history of breast cancer (Erblich, Montgomery, Cloitre, Valdimarsdottir, & Bovbjerg, 2003), in individuals with asthma (Jessop, Rutter, Sharma, & Albery, 2004), or in individuals with poorer self-assessed health (Williams, Wasserman, & Lotto, 2003). In all these studies, participants showed greater interference effects towards stimuli related to their source of concern. In any case, the emotional Stroop task has proved to be a simple but valid method for assessing cognitive interference (Dalgleish, 2005). Thus, it seems to form a convenient tool to examine whether biased processing of illness-related information is associated with health anxiety (Kellner, 1986, Warwick and Salkovskis, 1990).

The use of Stroop tasks in health anxiety studies is limited: Owens, Asmundson, Hadjistavropoulos, and Owens (2004) found an information processing bias towards illness-related cues in high health anxious persons, whereas such biases were not identified regarding other categories of emotionally laden stimuli (i.e., negative or positive emotion words). Lecci and Cohen’s (2002) study, however, resulted in different findings. They found that hypochondriacal tendencies were associated with slower reaction times to illness-related words in a modified Stroop task, but only when illness concerns were experimentally activated. Finally, Lim and Kim (2005) found that somatoform patients showed greater attentional bias for physical threat words.

Herein, we report the results of the two studies, in both of which a modified Stroop task was used. The purpose of our first study was to examine whether health anxiety is associated with non-conscious cognitive processing biases towards illness-related stimuli, as suggested by the cognitive model for health anxiety (Kellner, 1986, Warwick and Salkovskis, 1990), as well as with general threat-related stimuli. The inclusion of threat stimuli in this study was aimed in examining whether health anxiety is related only to stimuli with a specific content (i.e., illness) or to stimuli with undifferentiated threat value as well. The purpose of the second study was to replicate the findings of the first study. A second aim was to examine a possible interaction between health anxiety and current physical symptoms. Perceived symptoms are considered to be a significant source of health-related information, while they are involved in the development and the maintenance of health anxiety (Warwick & Salkovskis, 1990). Thus, their interaction with health anxiety might interfere with the information processing of health and illness-related stimuli. Our hypotheses were: (a) high health anxious participants will show significantly greater interference effects for illness-related stimuli in comparison with threat-related stimuli and neutral stimuli; (b) health anxiety interacts with current perceived symptoms regarding the attentional resources allocated to illness-related stimuli. Greater biases towards illness-related stimuli are expected when high health anxiety is combined with increased levels of physical symptoms.