• No results found

Descriptive analyses were performed to examine sample characteristics and the distribution of scores for the scales included in this study. A total of 2.156 participants identified as female, whereas 1.489 identified as male. The remaining 94 patients did not report their gender;

however, the frequencies undoubtedly confirm an overrepresentation of female participants.

All participants were in the ages of 16-97 years, with an average duration of pain of 8.06 years. Of all the 3.572 participants who answered the question regarding occupational status, 34.2 % reported that they were currently working, studying, or in military service, whereas 61.4 % reported that they were currently unemployed. Sample characteristics and descriptive statistics for the ODI, NRS, GSE and PCS are presented in Table 2 and Table 3. Note that the recoded variable of GSE (reversed scores) was included instead of the original GSE variable.

27 Table 2

Sample characteristics for ODI, NRS, GSE and PCS

ODI = Oswestry Disability Index; NRS = 0-10 NRS Usual Pain Intensity; GSE = General self-efficacy scale; PCS= Pain Catastrophizing Scale

Characteristics N (%)

Ntotal = 3.739

Mean (range) SD

Age 3.739 (100%) 49.73 (16-97) 15.62

Duration of pain (years) 3.426 (91.6%) 8.06 (0-73) 9.08

Sex

Female Male

3.645 (97.5%) 2.156 (56.6 %) 1.489 (39.8 %)

Education

Comprehensive school (1-10 years) Secondary/vocational school (11-13 years) College (14-17 years)

Higher university degree (>17 years)

3.637 (97.3%) 554 (14.8%) 1621 (43.4%) 1135 (30.4%) 327 (8.7%)

Employment/student

Employed/Student/Military Service Unemployed

3572 (95.5%) 1277 (34.2%) 2295 (61.4%)

Social benefits

Retirement pension Sick leave

Work assessment allowance (AAP) Disability pension

Unemployment benefits Social benefit

Others

3099 (82.9%) 484 (12.9%) 363 (9.7 %) 1000 (26.7%) 1048 (28.1%) 18 (0.5%) 64 (1.7%) 122 (3.3%)

28 Table 3

Descriptive statistics for ODI, NRS, GSE and PCS

Questionnaire Mean SD

ODI 42.75 17.54

NRS 7.19 1.79

GSE 21.35 5.53

PCS 22.75 13.10

ODI = Oswestry Disability Index; NRS = 0-10 NRS Usual Pain Intensity; GSE = General self-efficacy scale; PCS= Pain Catastrophizing Scale

Results: correlation analysis

The results from the Pearson correlation analyses are presented in Table 4. The results showed a significant correlation between pain intensity and pain-related disability, r = .42, p < .001.

Significant correlations between the independent variables were found; self-efficacy and pain intensity, r = .145, p < .001, self-efficacy and pain catastrophizing, r = .395, p < .001 and pain intensity and pain catastrophizing, r = .349, p <. 001. Note that the recoded GSE scale (reversed) was included and not the original GSE variable in the correlation analysis, thus the positive correlational relationship signifies that lower self-efficacy is associated with higher pain catastrophizing and pain intensity.

Table 4

Pearson correlation coefficients, all variables

ODI GSE NRS PCS

ODI 1 .307** .421** .379**

GSE .307** 1 .145** .395**

PI .421** .145** 1 .349**

PCS .379** .395** .349** 1

ODI = Oswestry Disability Index; NRS = 0-10 NRS Usual Pain Intensity; GSE = General self-efficacy scale; PCS= Pain Catastrophizing Scale

** p < .001

29 Results: mediation analysis

The results from Baron & Kenny’s (1986) method was further corroborated with Hayes Process in SPSS as well as the Sobel test. The results showed a significant direct effect of Pain Intensity (NRS) on the Oswestry Disability Index (ODI), b = .35, 95% CI [0.24 – 0.45].

The relationship between pain intensity and pain-related disability is depicted in Figure 2. A total effect of b = 4.00, p < .001 was derived from the analysis.

Figure 2

Path diagram illustrating the total effect between the independent and dependent variable

Figure 3 represents the proposed mediation analysis, wherein our results showed that pain intensity score is related to General Self Efficacy (GSE) score, b = .45, p < .001. Furthermore, the results from our analysis suggests that our proposed mediator, self-efficacy, is related to pain-related disability, b = .77, p < .001. A comparison of the total effect and the direct effect reveals that the association between pain intensity and pain-related disability is reduced when self-efficacy is included as a mediator, indicating that a significant partial mediation effect is present, albeit small-moderate. This is further corroborated by observed increase in the R2 value when including self-efficacy as mediator between pain intensity and pain-related disability, R2 = .021 without mediator variable versus R2 = .24 with the mediator variable. In this specific case, the R2-values reveal that reported pain intensity alone might explain 2.1%

of the variance in related disability, demonstrating that 97.9% of the variability in pain-related disability must be accounted for by other variables. Including self-efficacy in the model increased R2 to 24%, signifying that the combined influence of pain intensity and self-efficacy better explains pain-related disability than pain intensity does in isolation, however with 76% of the variability still unaccounted for.

Total effect: b = 4.00, p < .001 R2 = .021

Pain-related disability (ODI) Pain Intensity (NRS)

30

The Sobel test was conducted in order to verify that the results showed a consistent mediation effect. The Sobel test established a significant indirect effect of z = 2.18, p = .03. Running the analysis with Hayes PROCESS model 4 gave identical results.

Figure 3

Path diagram of the mediation model, including the results from the mediation analysis

General Self-Efficacy (GSE)

Pain Intensity (NRS) Pain-related disability (ODI)

Results: moderation analyses

Two simple independent moderation analyses were performed using hierarchical regression with each dependent variable. Neither analysis resulted in a significant interaction effect, hence there was no real indication to assume a moderated mediation effect either. Running the moderation analyses with PROCESS model 1 generated identical results. Note that the

insignificant interaction term (pain intensity x pain-related disability) in both moderation analyses showed high multicollinearity (highest VIF-value: 24.804 and lowest tolerance-value: 0.040). The results from the hierarchical regression analyses are presented in Table 5 and Table 6.

b = .45, p < .001 b = .77, p < .001

Direct effect: b = 3.65, p < .001 Indirect effect: b = .35, 95% CI [0.24 – 0.45]

R2 = .24

31 Table 5

Results from moderation analysis with PCS moderating the relationship between NRS and ODI

Dependent variable = Oswestry Disability Index (ODI)

NRS = 0-10 NRS Usual pain intensity; PCS = Pain Catastrophizing Scale; NRS x PCS = 0-10 NRS Usual Pain intensity and Pain Catastrophizing Scale

Table 6

Results from moderation analysis with PCS moderating the relationship between NRS and GSE

Variables B SE B t P R2 R2 change

Step 1

Constant 17.217 .382 45.075 p < .001 .156 .156

NRS .025 .055 .464 p = .643

PCS .171 .008 22.023 p < .001

Step 2

Constant 16.848 .700 24.068 p < .001 .156 .000

NRS .077 .099 .781 p = .435

PCS NRS x PCS

.189 -.002

.031 .004

6.162 -.629

p < .001 p =.529

Dependent variable: General Self-efficacy (GSE)

NRS = 0-10 Usual pain intensity; PCS = Pain Catastrophizing Scale; NRS x PCS = 0-10 NRS Usual Pain intensity and Pain Catastrophizing Scale

Variables b SE B t P R2 R2 change

Step 1

Constant 10.931 1.086 10.066 p < .001 .226 .226

NRS 3.325 .154 21.562 p < .001

PCS .303 .021 14.208 p < .001

Step 2

Constant 11.183 1.959 5.708 p < .001 .226 .000

NRS 3.290 .273 12.038 p < .001

PCS .290 .086 3.376 p = .001

NRS x PCS .002 .011 .155 p = .887

32 Results: moderated mediation analysis

A moderated mediation analysis was performed with Hayes PROCESS using model 8, despite of the lack of indications of a possible moderated mediation effect. As predicted from the simple moderation analyses, no significant moderator effects were found. The interaction variable, NRS x PCS, was found to be insignificant, b = .0137, p = .2273. Additionally, R2 change was .0004, further demonstrating an almost nonexistent accountability of the moderator on the dependent variable.

Based on the results it is inferred a partial mediation effect in the relationship, with GSE partly mediating the relationship between NRS and ODI. Furthermore, the results showed no significant moderator effects, neither for the simple regression analyses nor the moderated mediation analysis.

33