We included 120 patients. Sixty-four of them (53.3%) completed the questionnaire by themselves. To 55 of them (45.8%), nurses read out the questionnaire. For one patient (0.8%), there is no information about the mode of questionnaire completion. The gender distribution was almost equal (n=63, 52.5% women). The patients had an average age of 73 ± 14 years (min=18, max=98). On average, the wound size decreased from 9.88 cm² at T0 to 6.84 cm² at T1 (Table 1). The most common diagnoses (Table 2) were diabetic foot ulcer (n=37, 30.8%), venous ulcer (n=20, 16.7%) and ulcer caused by trauma (n=19, 15.8%). The majority of wounds were located on the lower legs (n=61, 50.8%) and feet (n=48, 40.0%). The most frequent comorbidities (Table 2) were cardiovascular diseases (n=81, 67.5%), diabetes (n=58, 48.3%) and peripheral vascular diseases (n=53, 44.2%).
Table 1: Sociodemographic and wound-specific descriptive statistics for the total sample and subsamples
|
Total sample (N=120)
|
Self-completion (N=64)
|
Read-out (N=55)
|
Self-completion vs. read-out
|
|
n
|
%
|
n
|
%
|
n
|
%
|
p-value1
|
Male (vs. female)
|
57
|
47.5
|
34
|
53.1
|
22
|
40.0
|
0.153
|
Smoker (yes)
|
21
|
17.5
|
10
|
15.6
|
10
|
18.5
|
0.676
|
Living alone (yes)
|
48
|
40.0
|
21
|
32.8
|
27
|
49.1
|
0.071
|
Compression therapy (yes)
|
69
|
57.5
|
34
|
53.1
|
34
|
61.8
|
0.339
|
|
M
|
SD
|
M
|
SD
|
M
|
SD
|
p-value2
|
Age [years]
|
73.23
|
14.24
|
68.89
|
14.71
|
78.42
|
11.99
|
<0.001
|
Length [m]
|
1.72
|
0.10
|
1.73
|
0.10
|
1.70
|
0.11
|
0.183
|
Weight [kg]
|
82.52
|
19.85
|
84.36
|
17.03
|
80.00
|
22.79
|
0.238
|
BMI [kg/m²]
|
27.75
|
5.48
|
28.11
|
4.74
|
27.30
|
6.33
|
0.427
|
Wound duration [weeks]
|
34.55
|
65.02
|
22.28
|
39.61
|
49.50
|
84.43
|
0.023
|
Wound size at T0 [cm²]
|
9.88
|
21.33
|
9.39
|
13.69
|
10.57
|
27.89
|
0.766
|
Wound size at T1 [cm2]
|
6.84
|
18.67
|
7.33
|
15.05
|
6.36
|
22.42
|
0.786
|
Number of medications
|
8.6
|
4.39
|
8.56
|
4.72
|
8.49
|
3.89
|
0.929
|
Number of comorbidities
|
2.87
|
1.37
|
2.83
|
1.42
|
2.87
|
1.31
|
0.860
|
1 p-value according to Chi-square test, 2 p-value according to unpaired t-test
Table 2: Wound diagnoses and co-morbidities at baseline
|
|
Frequency
|
Percentage
|
|
Wound diagnoses
|
|
Diabetic foot ulcer
|
37
|
30.8
|
|
|
Venous ulcer
|
20
|
16.7
|
|
|
Traumatic ulcer
|
19
|
15.8
|
|
|
Arterial ulcer
|
14
|
11.7
|
|
|
Malignant ulcer
|
8
|
6.7
|
|
|
Mixed ulcer
|
6
|
5.0
|
|
|
Pressure ulcer
|
5
|
4.2
|
|
|
Vasculitis
|
1
|
0.8
|
|
|
Other
|
10
|
8.3
|
|
Comorbidities1
|
|
Cardiovascular diseases
|
81
|
67.5
|
|
|
Diabetes
|
58
|
48.3
|
|
|
Peripheral vascualar diseases
|
53
|
44.2
|
|
|
Chronic obstructive pulmonary disease
|
26
|
21.7
|
|
|
Rheumatism
|
25
|
20.8
|
|
|
Renal failure
|
21
|
17.5
|
|
|
Oncologic disease
|
17
|
14.2
|
|
|
Thrombosis/pulmonary embolism
|
14
|
11.7
|
|
|
Surgical intervention
|
13
|
10.8
|
|
|
Neurological disorders
|
11
|
9.2
|
|
|
Thyroid disorders
|
10
|
8.3
|
|
|
Psychiatric disorders
|
8
|
6.7
|
|
|
Other
|
12
|
10.0
|
|
1Percentages do not sum up to 100.0% because multiple answers were possible.
Descriptive statistics of the Wound-QoL items, the global scale, and the subscale can be seen in Table 3.
Insert Table 3
Number of missing values
Of the seventeen items, nine items at T0 and eight items at T1 showed no missing values. One item (‘climbing stairs has been difficult because of the wound’) showed a large number of missing values (T0: 29.2%, T1: 25.0%). All patients with missing values for this item filled in or stated during the interview that this item was not applicable to their situation, but this was not a response option in the Wound-QoL. For the remaining items, the number of missing values ranged from 0.8% to 2.5% at both time points equalling to one to three patients per item. Regarding the global and subscale scores, only for one patient at T0 and three patients at T1 the subscale ‘everyday life’ could not be calculated because of too many missing values.
Floor and ceiling effects
The global score showed no ceiling effect at either time point and a minor floor effect only at T1 (0.8%). Although the ‘body’ and ‘psyche’ subscales did not show ceiling effects at T0, the ‘psyche’ subscale showed a minor ceiling effect at T1 (0.8%). The ‘everyday life’ subscale showed minor ceiling effects at both time points (T0: 1.7%, T1: 3.4%). All subscales showed floor effects at both T0 (body: 12.5%, psyche: 9.2%, everyday life: 6.7%) and T1 (21.7%, 7.5%, 11.1% respectively). Floor effects in ‘body’ and ‘everyday life’ subscales were less pronounced in the patients who completed the questionnaire by themselves (T0: 6.3%, 4.7%, T1: 18.8%, 8.1% respectively) compared to the patients in the read-out group (T0: 20.0%, 9.3%, T1: 25.5%, 14.8% respectively).
Changes in the mean scores
The mean values decreased over time for both Wound-QoL global score (T0: 1.29, T1: 1.12) and Wound-QoL subscale scores (body: T0: 0.98, T1: 0.81, psyche: T0: 1.38, T1: 1.27, everyday life: T0: 1.59, T1: 1.36). The T-test results revealed significant improvements for the global score (t(119)=2.566, p=0.012), the ‘body’ subscale (t(119)=2.221, p=0.028) and the ‘everyday life’ subscale (t(119)=2.500, p=0.014), but not for the ‘psyche’ subscale (t(119)=1.136, p=0.258).
When we consider each of the subgroups, a significant change was observed in the self-completion group (global score: p=0.005, body: p=0.038, everyday life: p=0.002, psyche: p=0.076), but not in the read-out group (global score: p=0.820, body: p=0.371, everyday life: p=0.783, psyche: p=0.404).
Internal consistency
The internal consistency of the Wound-QoL global score was high at both times points (T0: α=0.889, T1: α=0.918). With regard to the subscales, the internal consistency was highest for the ‘everyday life’ subscale (T0: α=0.895, T1: α=0.925), followed by the ‘psyche’ subscale (T0: α=0.794, T1: α=0.811) and the ‘body’ subscale (T0: α=0.673, T1: α=0.687). The self-completed and read-out questionnaires showed the same patterns.
Item selectivity
The item selectivity of the items of the global score ranged from r=0.251 to r=0.768 at T0 and from r=0.395 to r=0.793 at T1. The items with the highest correlation coefficients were: ‘I have had trouble with everyday activities because of the wound’ (T0: r=0.768, T1: r=0.793), ‘the wound has limited my recreational activities’ (T0: r=0.760, T1: r=0.723), ‘the wound has forced me to limit my contact with other people’ (T0: r=0.754, T1: r=0.727) and ‘I have had trouble moving around because of the wound’ (T0: r=0.712, T1: r=0.728). It should also be noted that these four items showed the highest correlation coefficients in both the self-completion and the read-out group.
The item selectivity for the ‘body’ subscale ranged from r=0.369 to r=0.769 at T0 and from r=0.515 to r=0.775 at T1; for the ‘psyche’ subscale, it ranged from r=0.677 to r=778 at T0 and from r=0.593 to r=807 at T1 and for the ‘everyday life’ subscale, it ranged from r=0.703 to r=0.890 at T0 and from r=0.707 to r=0.870 at T1. The correlation between the global scale and subscales was highest for the ‘everyday life’ subscale (T0: r=0.867, T1: r=0.874), followed by the ‘psyche’ subscale (T0: r=0.801, T1: r=0.801) and the ‘body’ subscale (T0: r=0.632, T1: r=0.689).
Item selectivity generally showed minor effects and was similar for both the self-completion and the read-out subgroup.
Convergent validity
The correlation between EQ-5D-3L and Wound-QoL was significant (T0: r=-0.451, p<0.001, T1: r=-0.501, p<0.001). The same accounts for the correlation between pain VAS and Wound-QoL (T0: r=0.232, p=0.012, T1: r=0.372, p=0.001). Although the correlation between the wound size and Wound-QoL was significant at T1 (r=0.228, p=0.015), it was not significant at T0 (r=0.124, p=0.178). These correlations with the EQ-5D-3L represent moderate to large effect sizes, whereas the other correlations represent small to moderate effect sizes21.
For the self-completion subgroup, the correlation between EQ-5D-3L and Wound-QoL was significant at both time points (T0: r=-0.611, p<0.001, T1: r=-0.501, p<0.001). For the read-out subgroup, the correlation between EQ-5D-3L and Wound-QoL was significant at both time points as well (T0: r=-0.306, p=0.023, T1: r=-0.556, p<0.001). Additionally, for the read-out subgroup, the correlation between pain VAS and Wound-QoL was significant at both time points (T0: r=0.357, p=0.008, T1: r=0.486, p=0.003). The correlations with the EQ-5D-3L again represent moderate to large effect sizes, whereas the correlations with the pain VAS represent moderate effect sizes21. Table 4 shows the results regarding convergent validity for the total group and the subgroups.
Table 4: Convergent validity of the Wound-QoL global score
|
|
EQ-5D-3L
|
Pain VAS
|
Wound size
|
|
|
T0
|
T1
|
T0
|
T1
|
T0
|
T1
|
Total sample
|
r
|
-0.451
|
-0.501
|
0.232
|
0.372
|
0.124
|
0.228
|
|
p-value
|
<0.001
|
<0.001
|
0.012
|
0.001
|
0.178
|
0.015
|
|
n
|
119
|
119
|
116
|
75
|
119
|
113
|
Subgroup: Self-completion
|
r
|
-0.611
|
-0.501
|
0.180
|
0.306
|
0.143
|
0.242
|
p-value
|
<0.001
|
<0.001
|
0.165
|
0.055
|
0.263
|
0.063
|
|
n
|
63
|
64
|
61
|
40
|
63
|
60
|
Subgroup: Read-out
|
r
|
-0.306
|
-0.556
|
0.357
|
0.486
|
0.032
|
0.212
|
p-value
|
0.023
|
<0.001
|
0.008
|
0.003
|
0.816
|
0.132
|
|
n
|
55
|
54
|
54
|
35
|
55
|
52
|
Significant results are marked bold; r: Spearman correlation coefficient; n: number of patients; VAS: visual analogue scale
Responsiveness
Significant correlations were found between changes in Wound-QoL and changes in EQ-5D-3L (r=-0.373, p<0.001), changes in pain VAS (r=0.239, p=0.044) and changes in wound size (r=0.235, p=0.013). Although the effect sizes were moderate for correlations between changes in Wound-QoL and changes in EQ-5D-3L, the effect sizes were small for the correlations between changes in Wound-QoL and changes in pain VAS and wound size.
For the self-completion subgroup, only the correlation between changes in Wound-QoL and changes in EQ-5D-3L was significant (r=-0.408, p=0.001). The effect size was moderate. For the read-out subgroup, the correlation between changes in Wound-QoL and changes in EQ-5D-3L (r=-0.285, p=0.037), as well as the correlation between changes in Wound-QoL and changes in wound size (r=0.290, r=0.037), were significant, each representing small effect sizes. Table 5 shows the results regarding responsiveness for the total sample and the subgroups.
Table 5: Responsiveness of the Wound-QoL global score
|
|
EQ-5D-3L
|
pain VAS
|
Wound size
|
Total sample
|
r
|
-0.373
|
0.239
|
0.235
|
p-value
|
<0.001
|
0.044
|
0.013
|
N
|
118
|
71
|
112
|
Subgroup: Self-completing
|
r
|
-0.408
|
0.293
|
0.184
|
p-value
|
0.001
|
0.078
|
0.162
|
N
|
63
|
37
|
59
|
Subgroup: Read-out
|
r
|
-0.285
|
0.154
|
0.290
|
p-value
|
0.037
|
0.383
|
0.037
|
n
|
54
|
34
|
52
|
Significant results are marked bold; r: Spearman correlation coefficient; n: number of patients; VAS: visual analogue scale
Time of completion
For nine patients, the time needed to complete the Wound-QoL questionnaire was recorded. The time needed ranged from 0:57 minutes (self-completion) to 3:53 minutes (read out) at T0.