Introduction
Prostate cancer is the second largest cancer-related cause of death in men and accounts for 16% of all cancer diagnoses, with a rising incidence during the last years [
1,
2]. Active surveillance is a relatively new treatment strategy for men with early prostate cancer. It consists of initially withholding radical treatment, but instead monitoring the disease and switching to active therapy only when progression occurs. Active surveillance may reduce overtreatment, but may cause anxiety and distress while living with ‘untreated’ cancer.
It is important to have an instrument that adequately measures anxiety specifically related to prostate cancer. To assess prostate cancer-specific anxiety, Roth et al. developed the Memorial Anxiety Scale for Prostate Cancer (MAX-PC) [
3]. They found that the MAX-PC captured anxiety among men with prostate cancer that might be missed using other more general anxiety measures, while, for example, it is more strongly associated with changes in PSA (prostate-specific antigen) level [
4].
We assessed the validity of the Dutch adaptation of the MAX-PC in Dutch men on active surveillance.
Results
Out of the 150 questionnaires sent, 129 were completed at a mean of 2.67 months (SD 1.74) after diagnosis (response rate 86%). Patient characteristics are shown in Table
1, and questionnaire scores and distributions are shown in Table
2. All 129 men completed all 18 MAX-PC items. In 8 men, the DCS, CES-D, STAI-6, or SF-12 score were discarded, as one or more items of one of these measures were missing.
Table 1
General, medical, and demographic patient characteristics (N = 129)
General |
Total number of patients | 129 |
Mean age (year) (SD) | 64.9 (6.9) |
Mean time (months) between questionnaire completion and diagnosis (SD) | 2.7 (1.7) |
Medical characteristics |
Mean prostate-specific antigen level (ng/ml) (SD) | 5.7 (1.9) |
Clinical stage
|
T1C (%) | 91 (71) |
T2 (%) | 38 (29) |
Demographics |
Education |
Low (primary, secondary) (%) | 86 (67) |
High (college, university) (%) | 42 (33) |
Missing | 1 |
Employed |
Yes (%) | 50 (60) |
No (%) | 76 (40) |
Missing | 3 |
Hospital |
Academic/referral centre (%) | 61 (47) |
Other (%) | 68 (53) |
Marital status |
Married/living together (%) | 119 | (92) |
Other (%) | 10 | (8) |
Table 2
Questionnaire scores and distributions (N = 129)
MAX-PC total | 13.9 | 8.8 | 14.0 | (6–20) | 0–54 | 0–39 | 2 | 0 |
Prostate cancer anxiety | 9.3 | 6.8 | 9.0 | (3–14) | 0–33 | 0–29 | 5 | 0 |
Prostate-specific antigen anxiety | 0.3 | 1.0 | 0.0 | (0–0) | 0–9 | 0–6 | 85 | 0 |
Fear of recurrence | 4.3 | 2.5 | 4.0 | (2–6) | 0–12 | 0–12 | 6 | 1 |
DCS | 27.5 | 13.7 | 28.1 | (18.8–36.3) | 0–100 | 0–67.2 | 1 | 0 |
CES-D | 5.7 | 6.1 | 4.0 | (0.5–9.2) | 0–60 | 0–24 | 25 | 0 |
STAI-6 | 35.9 | 9.0 | 35.0 | (30–40) | 20–80 | 20–66.7 | 5 | 0 |
SF-12 MCS | 54.1 | 8.5 | 55.6 | (52.2–60.1) | Mean 50, SD 10 | 25.5–67.1 | 0 | 0 |
Only CES-D and the ‘PSA anxiety’ subscale showed floor effects, with 25 and 85% of subjects exhibiting the most favorable low score, respectively. No ceiling effects were observed.
The Cronbach’s alpha coefficients for the ‘prostate cancer anxiety’ subscale, the ‘PSA anxiety’ subscale, the ‘fear of recurrence’ subscale, and the total MAX-PC were 0.91, 0.64, 0.85, and 0.77, respectively.
The initially fitted CFA model in which the items were assigned to the same factors as in the original publication did not fit very well (χ
2 271.81 with 132
df,
P < 0.001, RMSEA = 0.081, CFI 0.95). However, the modification indices in the sequence of subsequently fitted models indicated that model fit could be substantially improved by freeing parameters in the error covariance matrix only, while leaving the factor structure unchanged. Adding 10 extra covariance parameters among a total of 153 of these parameters resulted in a just adequately fitting model (χ
2 148.61 with 122
df,
P = 0.051, RMSEA = 0.037, CFI 0.99) that had the same factor structure as the original. These freed parameters indicated the presence of small neglected factors. Table
3 presents the factor loadings and standard deviations of the final adequately fitting CFA model.
Table 3
Factor loadings (and standard deviations) in the final adequately fitting confirmatory factor analysis model
1 | 0.59 (0.07) | – | – |
2 | 0.58 (0.07) | – | – |
3 | 0.62 (0.07) | – | – |
4 | 0.77 (0.07) | – | – |
5 | 0.64 (0.07) | – | – |
6 | 0.52 (0.06) | – | – |
7 | 0.71 (0.07) | – | – |
8 | 0.39 (0.06) | – | – |
9 | 0.38 (0.06) | – | – |
10 | 0.64 (0.08) | – | – |
11 | 0.66 (0.06) | – | – |
12 | – | 0.36 (0.06) | – |
13 | – | 0.14 (0.04) | – |
14 | – | 0.21 (0.05) | – |
15 | – | – | 0.48 (0.07) |
16 | – | – | 0.48 (0.06) |
17 | – | – | 0.60 (0.06) |
18 | – | – | 0.67 (0.06) |
The correlation coefficients of the MAX-PC scores with DCS, CES-D, STAI-6, and SF-12 MCS are shown in Table
4. The ‘PSA anxiety’ subscale did not show any relevant correlations (
r < 0.3), while all other correlations were >0.3. In line with prior hypotheses, the strongest correlations of the MAX-PC and the three subscales were seen with STAI-6. The
P-value for the difference between the correlation MAX-PC total—DCS (
r = 0.41) and MAX-PC total—CESD (
r = 0.48) was 0.49, for MAX-PC total—DCS (
r = 0.41) versus MAX-PC total—STAI-6 (
r = 0.66)
P was 0.008. All other possible differences between correlations were significant at the 0.001 level. Correlations were in line with hypotheses in >75%.
Table 4
Construct validity
DCS | 0.36** | 0.08 | 0.45** | 0.41** |
STAI-6 | 0.59** | 0.27** | 0.59** | 0.66** |
CES-D | 0.46** | 0.18* | 0.40** | 0.48** |
SF-12 MCS | −0.36** | 0.04 | −0.38** | −0.39** |
Discussion
We largely reproduced the structure and the validity of the MAX-PC as a measure for prostate cancer-specific anxiety in a sample of Dutch prostate cancer patients on active surveillance. To our knowledge, no other questionnaires for assessing prostate cancer-specific anxiety are available.
The ‘PSA anxiety’ subscale performed relatively poor with a Cronbach’s alpha of 0.64 and with no relevant correlations with other scores. These problems with the ‘PSA anxiety’ subscale were also observed in the original version of the MAX-PC [
3,
4]. The abnormal score distribution (85% of men in our population exhibited the lowest possible score) limits the value of the ‘PSA anxiety’ subscale in our study.
Compared to the internal consistency scores reported in other studies (alpha of the total MAX-PC 0.89-0.90; subscales ‘prostate cancer anxiety’, ‘PSA anxiety’, and ‘fear of recurrence’: 0.90–0.91, 0.54–0.64, and 0.82–0.85, respectively [
3‐
5]), Cronbach’s alpha for the total MAX-PC was somewhat lower in our cohort but similar for the subscales.
CFA largely confirmed the three-factor structure as used in the original publication. Correlation analysis provided evidence for the construct validity of the total score and the ‘prostate cancer anxiety’ and ‘fear of recurrence’ subscales but not of the ‘PSA anxiety’ subscale. These findings are also in line with results of the original version [
3].
Our study has limitations. Future validation studies should incorporate test–retest reliability, because this is an important quality measure for questionnaires that have a discriminative purpose such as the MAX-PC and longitudinal validity. Second, our data lack any psychiatric assessment or clinical diagnosis, so cut-off points for clinical prostate cancer-specific anxiety could not be established. Finally, we evaluated only a specific subgroup of patients with prostate cancer, i.e. men who are on active surveillance and who received the diagnosis no longer than 6 months earlier. As clinimetric properties may vary between different study populations, it is recommended to further validate the MAX-PC in other prostate cancer patient cohorts, e.g. before and after surgery or radiation therapy. Only with a multiple-group model or a direct comparison with the original version of the MAX-PC, the above-mentioned assertions on the validity of the Dutch version of the MAX-PC can be confirmed.
In conclusion, we found positive evidence for the appropriateness of the MAX-PC to identify and quantify prostate cancer-specific anxiety. It may allow for comparisons between Dutch patients and other international observations and for comparisons of the effect of treatments and/or supportive measures. However, some weaknesses in the original version, especially regarding the ‘PSA anxiety’ subscale, were also replicated in the adapted Dutch version. The ‘PSA anxiety’ subscale of MAX-PC may need to be revised.
Acknowledgments
The authors thank Laraine Visser-Isles for her backward-translation of the questionnaire and all 129 respondents for participation.