Psychosomatics 47:340-347, August 2006
doi: 10.1176/appi.psy.47.4.340
© 2006 Academy of Psychosomatic Medicine
Assessing Anxiety in Men With Prostate Cancer: Further Data on the Reliability and Validity of the Memorial Anxiety Scale for Prostate Cancer (MAX–PC)
Andrew Roth, M.D.,
Christian J. Nelson, Ph.D.,
Barry Rosenfeld, Ph.D.,
Adam Warshowski, M.A.,
Noelle O’Shea, R.N.,
Howard Scher, M.D.,
Jimmie C. Holland, M.D.,
Susan Slovin, M.D., Ph.D.,
Tracy Curley-Smart, R.N.,
Thomas Reynolds, M.A., and
William Breitbart, M.D.
Received February 2, 2005; revised September 2, 2005; accepted September 29, 2005. From the Dept. of Psychiatry and Behavioral Sciences, Dept. of Nursing, and Dept. of Medicine, Memorial Sloan-Kettering Cancer Center; and the Dept. of Psychology, Fordham University, Bronx, NY. Address correspondence and reprint requests to Dr. Nelson, Dept. of Psychiatry and Behavioral Sciences, Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021. e-mail: nelsonc{at}mskcc.org
|
ABSTRACT
|
Identifying which men with prostate cancer might benefit from mental health treatment has proven to be a challenging task. The authors developed the Memorial Anxiety Scale for Prostate Cancer (MAX–PC) in order to facilitate the identification of prostate cancer-related anxiety. A revised version of this scale was tested in a more clinically varied population. Ambulatory men with prostate cancer (N=367) completed a baseline assessment packet that included the MAX–PC and other psychosocial questionnaires. The MAX–PC showed high internal consistency and concurrent and discriminant validity. Factor analysis identified three distinct factors for the MAX–PC that corresponded to the intended subscales (General Prostate Cancer Anxiety, PSA (prostate-specific antigen) Anxiety, and Fear of Recurrence). PSA levels were not correlated with anxiety overall; however, anxiety was significantly higher among patients whose PSA levels were changing (i.e., rising, falling, and unstable), versus those with stable PSA levels. Also, in a multivariate analysis, the change in PSA levels was a significant predictor of MAX–PC scores, but not Hospital Anxiety and Depression Scale (HADS) scores. These results indicate that the MAX–PC is a valid and reliable measure of anxiety that assesses aspects of anxiety unique to men with prostate cancer, and it may provide a more sensitive measure of anxiety than the HADS for this population.
|
INTRODUCTION
|
Anxiety is frequently reported as the most common psychological reaction after a diagnosis of prostate cancer.1,2 Given the lack of clear consensus as to which treatments are optimal for any given prostate cancer patient, as well as the potential for unwanted side effects and possible disease progression, it is not surprising that many men experience intense anxiety.3 However, identifying which men suffer from clinically significant anxiety has proven to be challenging in this population. Studies of oncologists have typically revealed very low rates of recognition of patient distress, leaving many anxious patients untreated.4 Moreover, clinicians working with this population have frequently noted the potential for men to deny feeling anxious, while simultaneously displaying anxiety by delaying necessary interventions or demanding additional procedures such as repeated prostate-specific antigen (PSA) testing.5,6
Estimates of the prevalence of "clinically significant" anxiety in men with prostate cancer are complicated by the fact that methods for assessing anxiety have not been specifically tailored to this population. A number of studies have focused on either general psychological distress or traditional conceptualizations of depressive and anxiety disorders, but these studies may not accurately capture patients with significant but highly focused anxiety (i.e., anxiety centered on the illness or treatment). Nevertheless, a number of studies have documented high rates of psychological distress in men with prostate cancer. For example, Zabora and colleagues7 found that 30.5% of men with prostate cancer acknowledged significant psychological distress on the Brief Symptom Inventory (BSI).8 Roth and colleagues,1 using a single-item visual-analog scale, the Distress Thermometer, estimated that 13% of men with prostate cancer had severe psychological distress. Moreover, from the findings of the Hospital Anxiety and Depression Scale (HADS),9 they found that 33% of the sample had clinically significant anxiety. Despite these preliminary results, measures of general anxiety or distress may fail to identify those individuals who experience distress specifically focused on the illness or treatment.
In order to improve the recognition and measurement of anxiety related to prostate cancer, we developed a self-report measure, the Memorial Anxiety Scale for Prostate Cancer (MAX–PC).10 The original questions were generated with input from oncologists and psychosocial care specialists who had experience with prostate cancer patients and through review of other questionnaires that measured similar constructs (e.g., the Impact of Events Scale11). This 18-item scale was designed to tap three aspects of prostate-related anxiety: general anxiety related to prostate cancer and treatment, anxiety specifically related to PSA testing, and fear of recurrence/disease progression. In order to assess the usefulness of an earlier version of this instrument, we studied patient responses in a large sample of men with early-stage prostate cancer. This study found the MAX–PC to be a valid and reliable measure of anxiety, with strong concurrent and discriminant validity with other self-report measures. However, our first analysis of the MAX–PC was limited in a number of regards: it included relatively few men with advanced disease; it lacked a strong measure of general anxiety (e.g., a diagnosis of Generalized Anxiety Disorder); and it omitted or rephrased a number of original items based on our initial analyses. Hence, we have conducted the present investigation in order to assess the reliability and validity of this revised measure in a more severely ill sample as well as to analyze the extent to which disease-specific anxiety may be fueled by more general anxiety symptoms, versus situational factors such as disease state.
|
METHOD
|
Subjects
Subjects were recruited from the Urologic and Genito-Urinary Medical Oncology clinics at Memorial Sloan-Kettering Cancer Center from July 2001 to September 2002, while they were awaiting routine oncology appointments. Eligibility requirements included English fluency, diagnosis of prostate cancer, and a scheduled PSA test on the day of their visit to the clinic. Men who met these requirements were informed of the risks and benefits of study participation and provided written informed consent. This study was approved by the Institutional Review Board of Memorial Sloan-Kettering Cancer Center.
Procedure
Patients who provided informed consent were administered a series of self-report questionnaires to be completed while waiting for their scheduled appointment (the "baseline" assessment). Questionnaires included the Memorial Anxiety Scale for Prostate Cancer (MAX–PC10), the Hospital Anxiety and Depression Scale (HADS9), the Distress Thermometer,1 the Functional Assessment of Cancer Therapy Scale–Prostate Cancer Version (FACT–P13), the Generalized Anxiety Disorder Questionnaire (GAD–Q14), the Urinary Function Subscale for Prostate Cancer Patients,12 and a sociodemographic and medical questionnaire. Additional medical data, including PSA test results, stage of disease,15 and Gleason scores,16 were elicited from the hospital’s computerized medical database.
In order to assess the relationship between MAX–PC scores and changes in PSA test results, we categorized participants into one of four PSA change categories: rising, falling, steady, and unstable. These change categories were derived from the three PSA test results obtained before the baseline assessment. Rising and falling PSAs were defined on the basis of three consecutive PSA values that were either increasing or decreasing (respectively), regardless of the magnitude of the change. When PSA values fluctuated in an inconsistent manner (rising and then falling or the reverse), patients were classified as "steady" if the magnitude of the change did not exceed a 10% difference, whereas "unstable" PSA change was listed when the magnitude of the change exceeded 10% (based on the average of the three PSA test results).
Statistical Analyses
Reliability analyses included coefficient  calculations for both the MAX–PC total score and each of the three subscales, as well as calculating item/total correlations for each of the individual MAX–PC items. Also, the factor structure of the MAX–PC was analyzed by principal-components factor analysis to assess the extent to which the current data fit the anticipated three-factor model upon which the MAX–PC was originally developed.10 Validity of the MAX–PC was established by use of a series of correlation coefficients quantifying the association between MAX–PC total scores and individual subscales and the various measures of quality of life and psychological distress that had been administered concurrently.
|
RESULTS
|
Subject Characteristics
During the 15-month study period, a total of 367 men with prostate cancer consented to participate in the study. Of these 367 study participants, 164 (44.8%) were classified as having organ-confined disease (Tumor Stage 1 or 2), and 203 (55.2%) were classified with advanced disease (locally advanced or metastatic disease; Tumor Stage 3 or 4). The average length of time since initial prostate cancer diagnosis was 62.8 months (standard deviation [SD]: 54.4; range: 2.4–347.1). Not surprisingly, the average length of time since diagnosis was significantly longer for patients with advanced disease than for those with organ-confined disease (77.9 months versus 34.5 months; t[333]=7.53; p<0.0001). The average age of the sample was 65.8 years (SD: 9.3; range: 41–91). The vast majority of the sample was Caucasian (N=326; 88.8%), and 24 (6.0%) were African American, 8 were Hispanic (2.2%), and 11 (3%) were identified as other ethnic backgrounds; 305 men (83.8%) were currently married, and 271 (73.8%) had earned a college or advanced degree. Additional demographic and medical characteristics for this sample are presented in Table 1.
The sample demonstrated a moderate level of anxiety overall, although a number of individuals obtained high scores on the MAX–PC. We applied a cutoff of 27 to identify individuals with clinically significant anxiety, which represents an average rating of 1.5 on the 0–3 rating scale. Using this cutoff, 10.6% of the sample (N=39) were identified as having a high degree of anxiety (Figure 1). Using the GAD–Q to identify patients who met DSM-IV criteria for Generalized Anxiety Disorder (GAD), we found 12.5% of the sample (N=46) who met this threshold. However, the overlap between these two classifications was quite modest; only 4.6% of patients (N=17) met criteria for both a diagnosis of GAD and a clinically significant anxiety specifically focused on their disease.
Reliability Analysis
The MAX–PC total scores demonstrated strong internal consistency, with a coefficient of 0.90 and a median item/total correlation of 0.62 (range: 0.19–0.72; see Table 2). Likewise, the Prostate Cancer Anxiety (PCA) and Fear of Recurrence (Fear) subscales also demonstrated strong internal consistency (PCA: =0.90, Fear =0.85). The median item/total correlation of these two scales was 0.63 (range: 0.58–0.70) and 0.67 (range: 0.66–0.73), respectively. The internal consistency of the PSA Anxiety subscale was somewhat weaker, with a coefficient alpha of 0.54 and a mean item/total correlation for the PSA subscale of 0.44 (range: 0.19–0.52).
Factor Analysis
Principal-components factor analysis using a varimax rotation supported the three-factor model that had been identified in previous research (Table 3).10 All 11 items that comprised the PCA subscale had a factor-loading >0.50 on the first factor, and none of these items significantly loaded on the other two factors. The Fear subscale items also loaded clearly onto a second factor, with each item-loading >0.60 and no items loading >0.50 on the other two factors. The PSA subscale items also comprised a single factor, although one of the three item-loadings did not exceed 0.50, this third factor having a loading of 0.41. However, the other two PSA subscale items loaded heavily (>0.80) onto this third factor, and none of the three items loaded on the other factors.
Concurrent and Discriminant Validity Analysis
The MAX–PC total scores were significantly correlated with other measures of anxiety and psychological distress (see Table 4), including the HADS Anxiety subscale (r=0.51; p<0.0001), the Distress Thermometer (r=0.60; p<0.0001), HADS total scores (r=0.44; p<0.0001), and the FACT Emotional Well-Being subscale (r = –0.64; p<0.0001). Patients who met DSM–IV criteria for GAD on the basis of responses to the GAD–Q obtained significantly higher mean scores on the MAX–PC than did patients who did not meet criteria for GAD: 1.08 versus 0.65 (t[362]=5.27; p=0001).
As expected, the MAX–PC total scores demonstrated a somewhat weaker, although still statistically significant, correlation with the Depression subscale of the HADS (r=0.28; p<0.0001) as well as with other measures of overall quality of life, such as the FACT subscales (see Table 4). For example, the correlation between MAX–PC total scores and overall quality of life (FACT total score) was –0.34 (p<0.0001). The MAX–PC total scores were also negatively correlated with age (r = –0.18; p<0.0001), but were not associated with education. There was no correlation between baseline PSA and total MAX–PC scores. However, patients with advanced disease obtained significantly higher mean scores on the MAX–PC than did patients with localized disease (early-stage): 0.79 versus 0.56 (t[362]=3.96; p<0.0001).
The PCA subscale showed a comparable pattern of results as compared with the MAX–PC total scores, although with slightly weaker correlations (see Table 4). The Fear of Recurrence subscale, on the other hand, appears to be more sensitive to depression and physical distress. The Fear subscale, as compared with the MAX–PC total score or the PCA subscale, demonstrated stronger correlations with the Depression subscale of the HADS (r=0.41; p=0.0001) and FACT subscales. To illustrate, the Fear subscale demonstrated higher correlations on the FACT–G (r = –0.43; p=0001), the FACT–Physical subscale (r = –0.36; p=0.0001), the FACT–P Functional (r = –0.41; p=0.0001), and the FACT–P Prostate Concern subscale (r = –0.23; p=0.0001) as compared with the MAX–PC total scores. The Fear subscale was also the only scale to correlate significantly with Baseline PSA (r=0.12; p=0001). The PSA subscale showed the same pattern of significant correlations as the MAX–PC total scores; however, these correlations were weaker.
Differences in Mean MAX–PC Scores by PSA Progression
The MAX–PC total and subscale scores were then calculated for each of the PSA change groups, as described above (Table 5). There was a significant difference among the four PSA change groups (steady, rising, falling, and unstable) on MAX–PC Total scores (p=0.003). Also, there was a significant group difference in scores on the PCA and Fear of Recurrence subscales (p=0.045 and p=0.0001, respectively). There was no significant difference on the PSA subscale (p=0.32). Post-hoc analyses revealed that those patients who were classified as having a steady PSA reported significantly less anxiety than patients who had PSA levels that were rising, falling, or otherwise unstable.
Does the MAX–PC Measure a Unique Form of Anxiety?
To understand the determinants of anxiety related to prostate cancer, we used a multiple-regression analysis to identify predictors of MAX–PC scores. We considered both psychosocial and medical variables as possible predictors, on the basis of our hypothesis that both long-standing anxiety symptoms (as measured by the GAD–Q) and situational/medical factors (e.g., disease status as measured by current PSA level, the change in PSA scores or prostate cancer symptoms) would influence disease-related anxiety. In this analysis, MAX–PC total scores were used as the dependent variable, and the GAD–Q, FACT Prostate Cancer subscale, FACT Physical subscale, FACT Functioning subscale, Urinary Function subscale, use of hormone therapy (dichotomized variable contrasting patients on hormones versus those not on hormones), age, PSA level, and PSA progression (using a dichotomized variable contrasting patients with steady PSA levels versus those with rising, falling, or unstable PSA levels) were considered predictor variables. These variables generated a significant prediction model (p<0.0001), accounting for 18% of the variance in MAX–PC Total scores. As expected, GAD–Q, FACT–Functional, age, and PSA progression were significant predictors, whereas baseline PSA level, the FACT Prostate Cancer, and the FACT Physical subscales were not significant predictors (PSA level was retained in the prediction model because of the strong association between baseline PSA and PSA change).
We were also interested in determining whether pain predicted MAX–PC total scores. We did not directly assess pain in this study; however, FACT Physical Functioning and FACT Prostate subscales do contain questions that ask about pain. Since these FACT subscales were not significant predictors in the multivariate analysis, we pooled the four questions that pertain specifically to pain from these subscales and developed an ad-hoc pain measure. These four questions showed good reliability (Cronbach : 0.88) and were correlated with total MAX–PC scores (r=0.11; p<0.05). However, when added to the multivariate analysis, pain was not a significant predictor, and, as a result, it was not included in the final prediction model.
When this same set of variables was used in a regression model to predict scores on the HADS Anxiety subscale, largely similar results emerged, with a significant overall model (R2=0.21; p=0.0001) and a strong association with the presence of a generalized anxiety disorder (i.e., GAD–Q scores). However, PSA change was not significantly associated with HADS Anxiety scores in this model, indicating less association between this measure of anxiety and disease-specific factors (Table 6).
|
DISCUSSION
|
This study demonstrates strong reliability and validity for the revised 18-item MAX–PC. In a sample comprising roughly equal numbers of men with early and advanced disease, the MAX–PC demonstrated a high degree of reliability, as well as convergent and discriminant validity. Not surprisingly, patients who met DSM–IV criteria for GAD scored significantly higher on the MAX–PC than patients who did not meet criteria for GAD. However, consistent with our expectations, we found that MAX–PC scores were not simply a function of trait-like anxiety. Multiple-regression models predicting MAX–PC scores indicated that anxiety related to prostate cancer was a function of both generalized anxiety (GAD) as well as instability of PSA levels and diminished functional abilities. Indeed, the MAX–PC appeared to be more sensitive to "state" variables, such as changes in PSA levels, than were other, more general, measures of anxiety such as the HADS or Distress Thermometer. Thus, the MAX–PC appears to capture a unique type of anxiety among men with prostate cancer that might be missed using other, more general, anxiety measures.
The finding that men with decreasing PSA levels had anxiety levels comparable to patients with rising or unstable PSA levels was somewhat surprising. The impetus behind the development of the MAX–PC arose from clinical observations that men with prostate cancer become very anxious in relationship to their PSA tests. However, our expectation was that high or rising PSA levels would be associated with the highest level of anxiety, whereas men with stable or falling PSA levels would be less anxious. Contrary to this expectation, we found that men with falling PSA levels experienced anxiety comparable to those with rising or unstable PSA levels. Moreover, there was no association between PSA level and MAX–PC scores at the time of the baseline assessment. These findings suggest that prostate-cancer–related anxiety is heavily influenced by changes in PSA test results, regardless of the direction of change, but are not necessarily related to the actual PSA level. The relatively high levels of anxiety among men with falling PSA levels may indicate that patients who are still in the active phases of treatment are concerned that their improvement will continue.
The observation of significantly lower anxiety among men with stable PSA levels was evident across the total MAX–PC scores, as well as each of the three subscales. However, the PSA Anxiety subscale, which arguably should have been most sensitive to PSA-related concerns, did not reach statistical significance in this analysis. This "null" finding may be the result of limitations in the PSA Anxiety subscale. Despite having been changed from the initial iteration of the MAX–PC in hopes of increasing its applicability, this scale still had relatively weaker reliability and was handicapped by the low number of questions in the scale and by low rates of item endorsement (i.e., relatively few men reported substantial anxiety specifically focused on PSA testing). Despite the marginal psychometric properties of this subscale, we continue to believe that the occurrence of anxiety specifically focused on PSA testing is sufficiently important from a clinical perspective to warrant continued inclusion in the MAX–PC.
These data suggest that the MAX–PC may be an important clinical tool for oncologists treating men with prostate cancer. By providing a rapid and specific mechanism for identifying men with high levels of anxiety, appropriate referrals (e.g., for further mental health consultation) can be greatly facilitated. The cutoff score utilized in this study identified roughly 10% of the sample as having significant anxiety related to prostate cancer and its treatment. This proportion, although lower than some studies using more ambiguous criteria for assessing distress, is likely a more realistic appraisal of the frequency with which mental health intervention for anxiety is actually necessary or can feasibly be applied in this population. However, without stronger evidence to support an "optimal" cutoff score (as discussed below), the threshold applied in this analysis should be considered preliminary. Moreover, there are numerous clinical and empirical justifications for applying alternative cutoff scores, such as the need to identify individuals with any evidence of anxiety (which would justify using a lower cutoff), or only those individuals with extreme levels of anxiety (which would justify a higher cutoff), both of which might be appropriate in different settings.
Despite the strong psychometric findings, this study does have several limitations. First, despite the variation in disease-stage in this sample, the demographic characteristics are skewed toward the well-educated, married, white man. This may have implications for the generalizability of this instrument when assessing other ethnic groups, because these groups may experience anxiety in a variety of different ways. Second, there was no clinician-rated assessment of anxiety in this study, which limited our ability to develop an optimal cutoff score. Despite the use of a validated self-report measure of Generalized Anxiety Disorder, this diagnosis does not represent an appropriate "gold standard" because of the possibility that clinically significant anxiety will not take the form of GAD in this population. Third, this instrument does not overtly ask about the possible negative impact prostate cancer may have on these men’s lives (e.g., becoming a burden to their families or financial stressors related to medical bills and the potential loss of income), and whether this negative impact has led to anxiety. However, we believe that if these issues were the cause of anxiety, although not overtly queried on the measure, these men would endorse more anxious thoughts when thinking about prostate cancer, and, as a result, have a higher score on the instrument. Despite these limitations, this study provides important additional evidence for the validation of the MAX–PC as a brief instrument to assess anxiety in men with prostate cancer, and it has some significant advantages over other existing measurement tools.
|
REFERENCES
|
- Roth AJ, Kornblith AB, Batel-Copel L, et al: Rapid screening for psychologic distress in men with prostate cancer: a pilot study. Cancer 1998; 82:1904–1908[CrossRef][Medline]
- Slovin S: Biochemical relapse in prostate cancer: is PSA promoting stress and anxiety? (electronic version). Medscape Hematology-Oncology 2002; 5
- Roth A, Scher H: Genitourinary malignancies, in Psycho-Oncology. Edited by Holland JC. New York, Oxford University Press, 1998, pp 339-358
- Fallowfield L, Ratcliffe D, Jenkins V, et al: Psychiatric morbidity and its recognition by doctors in patients with cancer. Br J Cancer 2001; 84:1011–1015[CrossRef][Medline]
- Lofters A, Juffs HG, Pond GR, et al: "PSA-itis": knowledge of serum prostate antigen and other causes of anxiety in men with metastatic prostate cancer. J Urol 2002; 168:2516–2520[CrossRef][Medline]
- Roth AJ, Passik SD: Anxiety in men with prostate cancer may interfere with effective management of the disease. Primary Care and Cancer 1996; 16:30–37
- Zabora J, Brintzenhofe-Szoc K, Curbow B, et al: The prevalence of psychological distress by cancer site. Psycho-Oncology 2001; 10:19–28[CrossRef][Medline]
- Derogatis LR, Melisaraatos N: The Brief Symptom Inventory (BSI): an introductory report. Psychol Med 1983; 13:595–606[Medline]
- Zigmund AS, Snaith RP: The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 1983; 67:361–370[Medline]
- Roth AJ, Rosenfeld B, Kornblith AB, et al: The Memorial Anxiety Scale for Prostate Cancer: validation of a new scale to measure anxiety in men with prostate cancer. Cancer 2003; 97:2910–2918[CrossRef][Medline]
- Horowitz M, Wilner N, Alvarez W: Impact of Events Scale: a measure of subjective stress. Pychosom Med 1979; 41:209–218
- Litwin MS, Hays RD, Fink A, et al: Quality-of-life outcomes in men treated for localized prostate cancer. JAMA 1995; 273:129–135[Abstract]
- Esper P, Mo F, Chodak G, et al: Measuring quality of life in men with prostate cancer using the Functional Assessment of Cancer Therapy-Prostate (FACT-P) instrument. Urology 1997; 50:920–928[CrossRef][Medline]
- Roemer L, Borkovec M, Sharon P, et al: A self-report diagnostic measure of generalized anxiety disorder. J Behav Ther Psychiatry 1995; 26:345–350[CrossRef]
- American Joint Committee on Cancer: Manual for Staging of Cancer, 4th Edition. Edited by Beahrs OH, Henson DE, Hutter RVP, et al. Philadelphia, PA, Lippincott, 1992
- Carroll PR, Lee KL, Fuks ZY, et al: Cancer of the prostate, in Cancer: Principles and Practice of Oncology, 6th Edition. Edited by DeVita VT Jr, Hellman S, Rosenberg SA. Philadelphia, PA, Lippincott, Williams & Wilkins, 2001
This article has been cited by other articles:
|
|
|
|
 
N. S. Consedine, B. A. Adjei, P. M. Ramirez, and J. M. McKiernan
An Object Lesson: Source Determines the Relations That Trait Anxiety, Prostate Cancer Worry, and Screening Fear Hold with Prostate Screening Frequency
Cancer Epidemiol. Biomarkers Prev.,
July 1, 2008;
17(7):
1631 - 1639.
[Abstract]
[Full Text]
[PDF]
|
|
|
Get information about faster international access.
Privacy Policy
Copyright © 2006
Academy of Psychosomatic Medicine.
All rights reserved.
Home
| Search
| Current Issue
| Past Issues
| Subscribe
| All APPI Journals
| Help
| Contact Us
|
TOP
ABSTRACT
INTRODUCTION
