In order to make determinations as to whether an individual's score on an anxiety scale represents a significant deviation from what would normally be expected in the population, clinicians compare that score to normative values. These values are usually based on groups of psychologically healthy patients who are demographically comparable to the client being assessed. When these normative values are derived from persons who are different from the individual (e.g. with respect to age), incorrect assessment conclusions can be reached. In fact, research has established that age affects responses on psychological tests. There is evidence that mental disorders and their symptoms may manifest differently in the elderly. It is possible that some anxiety symptoms may be reduced while others may become more prominent as one ages. Wisocki et al. (1986), for example, have shown that seniors do not report many worries apart from health-related concerns. Other evidence suggests that anxiety may manifest as somatic symptoms in older adults. Therefore, basing assessment conclusions about an elderly individual on norms for younger persons may be problematic. Despite these issues, attempts to tailor existing measures for use with the elderly are rarely observed and normative values for many widely used scales are not available for seniors. Investigating the need for such norms is critical for the correct diagnosis of or the ability to rule out psychological conditions such as anxiety disorders. Development of age-appropriate norms could lead to better and more immediate care and treatment of seniors. The purpose of this project was to investigate whether some of the commonly used measures of anxiety should be normed separately for this population.
Little research has been carried out regarding the use of anxiety measures with elders. Stanley et al. (1996) cited the need to establish psychometrically sound anxiety measures for older adults and studied four anxiety measures (i.e. State-Trait Anxiety Inventory; the Fear Questionnaire; the Worry Scale; and the Padua Inventory). They concluded that young persons without anxiety complaints tend to obtain higher scores than older persons. A limitation of the study was that comparison means for younger adults were obtained from other studies that had utilized college undergraduates. The scores of college undergraduates are demographically homogeneous and may differ from those of young adults (e.g. ages 25 to 54) working in the community. Osman et al. (1993) found significant negative associations between scores on some Beck Anxiety Inventory items and age but most of their sample were relatively young community volunteers (age of males: M = 36.2 years, SD = 11.9; age of females: M = 37.1 years, SD = 12.0).
Steer et al. (1994) found the BAI levels of elderly medical outpatients to be lower than those of college undergraduates. As medical outpatients may score differently than healthy students regardless of age, their findings do not facilitate the identification of appropriate normative values for normal elders. Nonetheless, Steer et al. (1994) found that elderly medical outpatients obtain significantly lower scores than elderly psychiatric patients, supporting the utility of this measure with seniors.
Another normative study is that of Knight et al. (1983), who studied the STAI. Their findings suggested no score differences over the life span. Note, however, that the findings may not be generalizable to North America and Europe as they were generated in New Zealand. Furthermore, these investigators did not examine the possibility of sex by age interaction effects.
In addition to the BAI, the present study focused on the Anxiety Sensitivity Index. The construct of anxiety sensitivity (AS) has received considerable attention in recent years, but work on its application and utility with elders is lacking. Indeed, a recent state-of-the-art book on AS includes consideration of many issues but, notably, does not incorporate a chapter regarding the elderly. AS is specific, distinct from trait anxiety, and refers to a fear of anxiety and anxiety-related bodily sensations based on the belief that they will have harmful consequences. It has been shown to be elevated in individuals with anxiety disorders and is predictive of the occurrence of spontaneous panic attacks. Despite the widespread clinical use of the ASI, there are no normative data for this scale concerning elders.
Finally, we were interested in examining the Fear Survey Schedule. The FSS comprises of a list of a large number of stimulus situations to which fear is maladaptive (e.g. being ignored, being teased, the noise of vacuum cleaners, the sight of people with deformities). Although it is not appropriate to diagnose specific phobias based on the FSS alone, it is a widely used measure of fear and anxiety. Normative values on the FSS are lacking and their development is important as fear of specific situations captures an aspect of anxiety that is not tapped by many other instruments.
It was hypothesized that, as anxiety disorders manifest differently in older adults, scores on the FSS, BAI and ASI would vary as a function of age. Based on preliminary indicators in the pre-existing literature, it was expected that older persons would obtain lower scores on all measures.
Methods Participants The participants were 101 adults (37 male), with an age range from 40 to 93 years. They were recruited with the aid of a large community organization (i.e. an umbrella organization that brings together seniors'/ retired persons' community groups throughout our geographic region) and a public advertisement. Some participants contacted the investigators after reading the advertisement, while others responded to an announcement made (on our behalf) by a representative of the community organization.
Participants varied widely in terms of age, educational attainment and occupation. Participants were asked to indicate whether they were receiving treatment for a mental disorder. Five persons who responded affirmatively were excluded from the analyses.
Beck Anxiety Inventory. The BAI is a 21-item measure of anxiety symptom severity (e.g. nervousness, difficulty breathing, fear of dying). Participants rate how much each symptom distressed them over the past week using a four-point scale ranging from `not at all' to `severely'. The possible range of scores is 0-63. The validity and reliability of the instrument have been successfully established.
Anxiety Sensitivity Index. The ASI is a 16-item self-report inventory. The items are rated on a five-point scale and reflect the person's concern about the negative consequences of anxiety. The possible range of scores is 0-64. The ASI is also believed to measure susceptibility to panic disorder where the preoccupation is frequently with fear of dying as a result of anxiety symptoms (e.g. palpitations seen as a heart attack or hyperventilation as suffocation). The reliability and validity of the measure have been supported in studies of younger adults.
Fear Survey Schedule. The FSS is a measure of fear. It lists 108 stimulus situations (e.g. being teased, dentists, fish) and respondents are asked to indicate along a five-point scale how disturbing each stimulus situation is. Total scores range from 0-432. The reliability and validity of the instrument are well established.
Participants completed the BAI, the ASI and the FSS, along with a brief questionnaire requesting demographic information pertaining to age, sex, education and medical history. Subsequently, they anonymously returned them by mail (postage was pre-paid). Results
Of a total of 150 questionnaires, 101 were returned (return rate 67%). For the purpose of the analyses, the sample was divided into two groups: persons 60 years of age and over and persons younger than 60 years old.
We first examined the scores on the three dependent measures of interest (BAI, ASI and FSS) in order to determine whether they significantly deviated from the normal distribution. Our examination revealed that the distributions of scores on all three measures showed significant and moderate skewness. Tabachnick and Fidell (1989) recommend the square root transformation for normalizing such data. This was applied. Following the data transformation, we conducted a preliminary analysis in order to determine whether participants' level of education and retirement status were related to any of the three transformed variables. While education was not significantly related to any of the variables of interest, retirement status was correlated with the ASI scores. It was, therefore, concluded that retirement status would be used as a covariate in the analysis involving the ASI.
The main comparisons (based on age and sex) were conducted using 2 x 2 analyses. An analysis of variance (ANOVA) of the BAI transformed scores did not reveal significant main effects but identified a significant age by sex interaction, F(1,92) = 5.43, p < 0.05. Tukey's Honestly Significant Difference method was employed to assess the age by sex interaction apparent for the BAI. These post-hoc comparisons showed that older men obtained lower BAI scores than both younger men, Q = 4.13, p < 0.05, and women over 60 years of age, Q = 3.96, p < 0.05.
The transformed ASI scores were analyzed using a 2 x 2 analysis of covariance (ANCOVA), with retirement status as a covariate. These scores varied as a function of age, F(1,91) = 6.04, p < 0.05, and sex, F(1,91) = 4.52, p < 0.05, with older and male participants obtaining lower scores. The age group x sex interaction and the effect of the covariate were not significant.
The ANOVA involving the transformed total FAST score only revealed one significant effect. Specifically, scores varied as a function of participant sex, with female participants obtaining higher scores, F(1,92) = 13.31, p < 0.001. However, since FAST norms are often presented for individual items, individual items were also assessed (using square root transformations to correct for apparent deviations from normality) for age effects using 2 x 2 Cage group x sex) analyses of variance. Given the high number of analyses (108), the significance level was set at 0.001 in order to reduce the probability of Type I error rates. Age effects were identified with respect to five FAST items.
These findings suggest that separate ASI norms may be appropriate for older adults. The older participants obtained lower scores than the younger ones. In fact, the scores obtained by the younger adults appear to be comparable to those found in previous research. The ASI was initially developed as a measure of susceptibility to panic disorder. Reported prevalence rates of anxiety disorders among seniors vary in the literature.
Nonetheless, our findings are consistent with the view that if anxiety disorders are less prevalent among the elderly, discovering lower ASI scores among the older participants may simply reflect that elders are less susceptible to panic.
Consistent with the ASI findings, general anxiety (as measured by the BAI) was found to be lower among men over 60 than among younger men. This supports the need for separate norms for older men. Although investigations of the State Trait Anxiety scales did not reveal differences in anxiety scores between younger and older persons, consideration was not given to age by sex interactions as in the present study. Our findings suggest that it would be important for future normative investigations of anxiety in the elderly to assess for such interactions. Moreover, given that phobias and panic disorder are more common in women, it is not surprising that we found females to obtain higher FSS and ASI scores than their male counterparts.
We did not find any evidence to suggest that separate age-based norms would be necessary for the total FSS score. Nonetheless, our examination of specific differences on individual items was instructive. These differences suggest that anxiety about social embarrassment and success may be reduced among older persons. Furthermore, the findings suggest that separate FSS norms for males and females may be appropriate.
In summary, from a clinical standpoint, the present findings suggest that separate norms for older adults would be appropriate for use with the ASI. They also suggest that norms broken down by age and sex would be appropriate when the BAI is used with seniors. While separate norms for seniors do not appear to be necessary when assessing the FSS total score, sex differences exist. It would be valuable for investigators to direct future attention towards obtaining normative information across an array of anxiety and other psychopathology measures appropriate for application in seniors. This, indeed, remains an understudied yet important avenue of inquiry.
In terms of limitations, it is possible that additional differences would have been detected with a larger sample size. Furthermore, our recruitment strategy involved a community advertisement and seniors' organizations. This may limit the generalizability of our conclusions in that the findings may not necessarily be applicable to the entire population of community-dwelling elders. This possibility could be clarified through future investigation.
- Owens, K. M., Hadjistavropoulos, T., & Asmundson, G. J. (2000). Addressing the need for appropriate norms when measuring anxiety in seniors. Aging & Mental Health,4(4), 309-314. doi:10.1080/713649964.
The box directly below contains references for the above article.
Reflection Exercise #4
The preceding section contained information
about anxiety in seniors. Write
three case study examples regarding how you might use the content of this section
in your practice.
18 What suggested the need for separate anxiety measurement norms for older men? To select and enter your answer go to Test.