From Data to Clinical Decisions in Chronic Wound Care: Evaluation of the Interpretability, Minimal Important Difference, and Smallest Detectable Change of the Wound‐QoL‐14 in a Spanish Population

ABSTRACT

The Wound‐QoL‐14 questionnaire has proven to be a valid instrument for assessing patients with chronic wounds. However, cutoff scores are not established, and the smallest detectable change and the minimal important difference have not been defined. We aimed to assess the interpretability, smallest detectable change and the minimal important difference of the Wound‐QoL‐14 questionnaire in a Spanish population with chronic wounds. At baseline, patients responded to a 5‐Likert global anchor question about how much their wounds had been a problem, and bands of severity were calculated. At a re‐test phase they responded to a 7‐Likert anchor question about the severity of the change in wound status, and smallest detectable change and minimal important difference were calculated using a combination of anchor‐based and distribution‐based methods. At baseline, stratification of 323 patients according to simplified bands of severity (0–0.25/0.26–1.00/1.01–2.00/2.01–3.00/3.01–4) showed a moderate correlation with the global anchor question (polychoric correlation = 0.653), but the level of agreement (41.5%) was poor (Krippendorff's alfa coefficient = 0.609). For 197 re‐test patients, the Wound‐QoL‐14 individual smallest detectable change was lower than the weighted minimal important difference (0.39 vs. 0.53 points, respectively). These findings support the interpretability of the Spanish version of the Wound‐QoL‐14 questionnaire for clinical and research use.

Key Points

• The scores of the Spanish version of the Wound‐QoL‐14 lacked clinical interpretation.

• We provided cutoff points to interpret clinical data, as well as the smallest detectable change and the minimal important difference.

• Five bands of severity were identified (Not at all: 0.00–0.20/Slightly: 0.21–1.00/Moderately: 1.01–82.25/Strongly: 2.26–3.20/Very strongly: 3.21–4.00).

• The smallest detectable change (SDC) was 0.39 points, rounded to 0.40.

• The minimal important difference (MID) for improvement was 0.53 points, rounded to 0.50.

1. Introduction

Assessing health‐related quality of life (HRQoL) in patients with chronic wounds is an essential component of research and patient‐centered clinical decision‐making, since these wounds impose a significant burden to patients, not only physical but also emotional and social [1]. Patient‐reported outcome measures (PROM) are useful tools for quantifying HRQoL. Compared to generic instruments like EQ‐5D or Short Form, disease‐specific instruments offer more detailed insights into HRQoL and are generally more sensitive to the impact of interventions [2]. The Wound‐Quality of Life (Wound‐QoL) questionnaire has become a widely accepted PROM for patients with chronic wounds, and it is one of the most commonly used condition‐specific questionnaires [2]. The original version, the Wound‐QoL‐17, consists of 17 items covering three key domains of patients' everyday life, body, and psyche [3]. A shorter version, the Wound‐QoL‐14, was later developed in response to the need for shorter instruments more suitable for routine clinical practice and large‐scale research [4]. Three items with low discriminative power were eliminated to build this shorter version, which preserves the original structure while maintaining strong psychometric properties. However, the psychometric validity of any PROM does not imply its clinical interpretability, which is an aspect frequently underestimated in validation studies [5].

Interpretability refers to the ability to assign qualitative clinical meaning to the quantitative scores obtained on a questionnaire, or to the change in scores [6]. Although not considered a psychometric property as such, it is an important characteristic of measurement instruments. The first proposal to interpret the Wound‐QoL instrument has recently been reported, based on a mixed sample from several European countries [7].

Closely related to interpretability are two other important parameters, namely the minimal important difference (MID) and the smallest detectable change (SDC). MID is defined as the smallest change in a score that patients perceive as beneficial or detrimental, indicating genuine improvement or deterioration which, in turn, would potentially prompt a change in their clinical management [8]. SDC represents the smallest change that exceeds the measurement error of the instrument, thus indicating a ‘true’ change beyond random variability. While the SDC is a statistical property derived from the reliability of the instrument, the MID is anchored in patients' perception and clinical relevance [9].

Interpretability, MID and SDC have been scarcely evaluated in the field of chronic wounds in general, and for Wound‐QoL surveys in particular. Previous reports include only one interpretability study involving both versions [7], and two studies in which MID and SDC were estimated for the Wound‐QoL‐17 version [10, 11]. Therefore, MID and SDC for Wound‐QoL‐14 remain to be calculated.

2. Objective

The aim of this study was to evaluate the interpretability of the Spanish version of the Wound‐QoL‐14 questionnaire in patients with chronic wounds. Specifically, we aimed to: (1) calculate cutoff points to define severity bands for Wound‐QoL‐14 scores, and (2) estimate the SDC and the MID to facilitate the interpretation of clinical data.

3. Materials and Methods

3.1. Study Design and Ethical Considerations

This research was part of the AlbaMad study, which was a collaborative, multicenter study conducted during 2024 in Spain to validate the specific Wound‐QoL‐14 survey in Spain [12]. Healthcare professionals involved in wound care were invited to participate through several scientific forums. However, no specific recruitment campaign or targeted call for patient participation was conducted. Consecutive patients aged 18 years or older presenting with chronic wounds of any aetiology of at least 1 month's duration were invited to participate during routine clinical visits at the participating centres. Exclusion criteria included hospitalization, refusal to participate, or inability to complete the questionnaires due to limited comprehension or language barriers. Written informed consent was obtained from all participants, and no financial compensation was provided to patients or healthcare professionals.

The questionnaires were self‐administered under the supervision of healthcare professionals specializing in the treatment of chronic wounds. Caregivers were permitted to provide assistance only when necessary, limiting themselves to reading the items verbatim, without interpreting or paraphrasing the responses.

Interpretability was evaluated in the total sample of 323 patients from the AlbaMad study. The SDC and the MID were evaluated in a subsample of 197 patients who completed the retest phase.

Approval from our Institutional Review Board (internal code 2023‐140) was granted. The requirements of the Declaration of Helsinki and Spanish regulations were met.

3.2. Sample Size

Interpretability was evaluated on the total sample of 323 patients who participated in the AlbaMad study. SDC and MID were assessed on the sample of 197 patients who also participated in the retest phase [12].

3.3. Wound‐QoL‐14 Questionnaire

The Wound‐QoL‐14 questionnaire consists of 14 items which are answered on a 5‐point Likert scale (0 = not at all to 4 = very much), with a 7‐day recall period. Item scores are averaged to obtain a score for the total scale and the three subscales (Body, Psyche and Everyday Life) [12].

3.4. Statistical Methods

Items left blank were imputed manually by following the hybrid method of the k‐Nearest Neighbour Imputation within Strata, where k = 6.

Categorical variables were expressed as frequency and percentage. Quantitative variables were expressed as mean with standard deviation (SD) and also as median with percentiles 25 and 75.

Correlations between quantitative variables were calculated by using the Spearman's rank coefficient (rho). The association between ordinal categorical variables was explored using polychoric correlations, which are more appropriate than Spearman's rho coefficient when dealing with variables with ordinal categories [13].

The free program jamovi 2.4.10 (The jamovi project, Sydney, Australia) was used for analyses. A p value < 0.05 was considered significant.

3.5. Interpretation of Single Scores

3.5.1. Global Anchor Question

All 323 patients who completed the Wound‐QoL‐14 survey at baseline also responded to a global anchor question (GAQ) aimed to rate the clinical severity of their wounds. The GAQ is a single‐item Likert‐type scale that serves as an external reference standard. It captures patients' perceptions of their health status, allowing them to be classified in order to establish meaningful benchmarks that facilitate clinical interpretation. Specifically, patients rated on a 5‐point scale the extent to which their wounds had been a problem over the last month (1‐Not at all/2‐Slightly/3‐Moderately/4‐Strongly/5‐Very strongly) [14].

3.5.2. Bands of Clinical Severity Defined by Cutoff Points

Bands of clinical severity are empirically defined score intervals corresponding to specific qualitative levels of a condition's severity. They are useful for transforming numerical scores into distinct qualitative categories, thereby enabling the grouping of patients into clinically meaningful levels of severity.

The first requisite to look for cutoff points using the GAQ at T₀ was to find at least a moderate correlation (rho > 0.5) between the Wound‐QoL‐14 scores and the GAQ categories. If accomplished, Wound‐QoL‐14 scores with two decimals were ordered from lowest to highest. After counting the number of patients scoring each concrete value, they were distributed alongside the five categories of clinical severity (from ‘Not at all’ to ‘Very strongly’) according to their responses to the GAQ. Each category was assigned a value from 0 to 4 (0‐Not at all/1‐Slightly/2‐Moderately/3‐Strongly/4‐Very strongly). Then, the mean, the median, and the mode of the categories for each Wound‐QoL‐14 score were calculated. These results were tabulated, and the table was subsequently examined looking for Wound‐QoL‐14 scores which had associated an increment of 1 point for the mean, the median, and the mode. Those Wound‐QoL‐14 scores were considered as potential candidates to be cutoff points to distribute the patients into the five categories of clinical severity [14]. The association between these two ordinal variables (the patients' actual distribution, and the distribution based on chosen cutoff points) was examined through polychoric correlations, and their level of agreement through the ordinal Krippendorff's α coefficient for agreement (Poor agreement: α ≤ 0.67/Acceptable agreement: 0.67 < α < 0.79/High agreement: 0.80 < α < 0.99/Perfect agreement: α = 1) [15].

3.5.3. Top Box Responses

The total number of top box responses, defined as selections of either of the two highest response categories for each item—‘quite’ or ‘very’—, was calculated for each participant. A very strong association between the number of top box responses and the Wound‐QoL‐14 global score was expected (Spearman rank correlation coefficient higher than 0.8) [7].

3.6. Interpretability of Change Scores

3.6.1. Floor and Ceiling Effects

Floor and ceiling effects for the Wound‐QoL‐14 were calculated, defined as less than 15% of patients reaching the worst or the best score, respectively [5]. HRQoL surveys should lack floor and ceiling effects to detect changes.

3.6.2. Patient Global Rating of Change & Health Worker Global Rating of Change

At the follow‐up visit conducted 1 month after the baseline assessment (T1, n = 197), both patients and the attending healthcare professionals simultaneously completed their respective global rating of change anchor questions to assess perceived clinical evolution, in order to calculate the SDC and MID. The patient global rating of change (GRC‐P) and the health worker global rating of change (GRC‐HW) were the same 7‐point Likert scale asking patients and health workers how much wounds had changed since the baseline visit (1‐Much worse/2‐Moderately worse/3‐Slightly worse/4‐Unchanged/5‐Slightly better/6‐Moderately better/7‐Much better), and if they found the change to be important or not [14]. Polychoric correlation and ordinal Krippendorff's α were used to evaluate the correlation and the level of agreement between the GRC‐P and GRC‐HW. Then, both GRC‐P and GRC‐HW were used to calculate SDC and MID, which were termed SDC‐P and SDC‐HW, and MID‐P and MID‐HW, respectively.

3.6.3. Score Differences Between T₀ and T₁

Wound‐QoL‐14 scores range from 0 points (best) to 4 points (worst). T ₁ scores were subtracted from T ₀ scores (Score difference = T ₀ − T ₁) to get improvements as positive differences and deterioration as negative differences.

3.6.4. Smallest Detectable Change

The SDC is the minimal change a survey is able to detect beyond the measurement error [11]. Accordingly, SDC‐P and SDC‐HW were calculated only for patients whose wounds were considered to remain unchanged. These groups were formed by summing those patients rating ‘Unchanged’ with those rating either ‘Slightly better’ or ‘Slightly worse’ who besides considered the change as nonimportant.

SDC can be calculated either for single individuals or for an entire group. Individual SDCs were calculated according to the following equation:

$${\mathrm{SDC}}_{\text{individual}}=\left[{1.96}^{*}\surd 2^{*}\left({\mathrm{SD}}_{\text{unchanged}}\times \surd 1-{\mathrm{CCC}}_{\text{unchanged}}\right)\right]$$ (1)

where SD_unchanged stands for the SD of change scores, whereas the reliability coefficient CCC_unchanged was the concordance correlation coefficient for the patients who remained in the same health status. To note, the expression between parentheses represents the standard error of measurement (SEm).

Group SDCs were calculated according to the equation:

$${\mathrm{SDC}}_{\text{group}}={\mathrm{SDC}}_{\text{individual}}/\surd \mathit{n}$$ (2)

where n is the sample size of patients who remained unchanged.

3.6.5. Minimal Important Difference

The MID‐P and MID‐HW were estimated for the Woud‐QoL‐14 survey and also for its domains based on a combination of anchor‐based methods and distribution‐based ones, as recommended [9, 16].

3.6.5.1. Anchor‐Based Methods

The ‘Slightly improved’ or ‘Slightly deteriorated’ subgroups were formed by those patients who responded ‘Slightly better’ or ‘Slightly worse’ to the anchor question and they also considered the change as being clinically important. As a prerequisite to check if GRC‐P and GRC‐HW questions were valid anchors to evaluate the MID for change in these subgroups, Spearman correlation coefficients between the change in the scores of Wound‐QoL‐14 and also its domains with both the GRC‐P and GRC‐HW should be greater than 0.3 [14, 16, 17, 18].

As a second requisite, the distribution of the change scores was graphically depicted to examine the shape of the density curves. Whether the shape of MID curves for improvement and for deterioration were similar, a single MID for change could be calculated. On the contrary, if the curves were different in shape, a MID for improvement and a different MID for deterioration were calculated [8].

The first method to estimate the MID for change was the mean average change (AC). The mean AC in score for the ‘Slightly improved’ or ‘Slightly deteriorated’ subgroups of patients was assimilated to the MID for improvement or deterioration, respectively.

The second method was based on building receiver operating characteristic (ROC) curves. The score for which sensibility and specificity sum to the maximum (Youden method) was assimilated to the MID value for change [9, 19].

Third, the social comparison method determined the MID as the mean of two differences. On one hand, the difference of the mean score between patients who rate themselves as ‘Slightly improved’ and those who rate themselves as ‘Unchanged’. On the other hand, the difference in mean score between patients who rate themselves as “Slightly deteriorated” and those who rate themselves as ‘Unchanged’ [19].

3.6.5.2. Distribution Based Methods

The first MID was calculated as half the SD of the mean change of the 197 scores [9]:

$$\mathrm{MID}={0.5}^{*}{\mathrm{SD}}_{\text{change}}$$ (3)

Second, MID was assimilated to the upper limit of the confidence interval of the standard error of measurement (SEm), which was calculated as follows [9]:

$$\mathrm{MID}={1.96}^{*}{\mathrm{SD}}_{\text{unchanged}}\times \surd 1-{\mathrm{CCC}}_{\text{unchanged}}$$ (4)

Third, MID was assimilated to the Cohen's d effect size for paired samples, also known as standardized response mean (SRM). SRM was calculated as the ratio between the mean change scores divided by the SD of the change scores [9].

Finally, a weighted MID mean value was calculated from the six methods, giving the anchor‐based methods more weight [16], specifically twice. MIDs for the Wound‐QoL‐14 survey and related domains were graphically represented along with the SDC as a lower limit reference.

Additionally, weighted MID mean values were divided by their related SDCs to calculate the reliability ratio [5]. The value should be higher than 1, meaning that clinically important change is distinguishable from measurement error.

4. Results

Overall, 197 out of 323 patients who participated in the AlbaMad study completed the survey and related documents both at baseline (T ₀) and at follow‐up (T ₁). Patients' flowchart is shown in Figure 1, and characteristics of these two related populations are shown in Table 1.

FIGURE 1.

Flowchart of patients participating in the AlbaMad study. PAT and HW denote that the number in parenthesis is the number of patients considered to have experienced the corresponding level of change according to the answer to the 7‐point Likert anchor question by patients and health workers, respectively.

TABLE 1.

Characteristics of the study population and results of the Wound‐QoL‐14.

Characteristics			T 0 (n = 323)	T 1 (n = 197)
Age, years		Mean (SD)	71.1 (14.6)	72.2 (14.2)
		Median [P25‐P75]	74 [62–82]	75 [63–83]
Sex, males		n (%)	168 (52.0)	96 (48.7)
Education level	No formal education	n (%)	65 (20.1)	40 (20.3)
Primary		n (%)	147 (45.5)	91 (46.2)
Secondary		n (%)	84 (26.0)	48 (24.4)
University		n (%)	27 (8.4)	18 (9.1)
Wound aetiology	Venous	n (%)	131 (40.6)	93 (47.2)
PAD/ischemic a		n (%)	63 (19.5)	37 (18.8)
Diabetic		n (%)	37 (11.5)	15 (7.6)
Others		n (%)	92 (28.5)	52 (26.4)
Wound duration (months)		Mean (SD)	18.9 (44.7)	22.5 (49.2)
		Median [P25‐P75]	6 [3–14]	6 [3–18]
Wound size (cm2)		Mean (SD)	20.2 (45.8)	25.1 (50.9)
		Median [P25‐P75]	4.7 [1.2–15.6]	7.0 [1.6–21.8]
Wound pain (NRS points)		Mean (SD)	4.2 (3.1)	4.6 (3.2)
		Median [P25‐P75]	4 [2–7]	5 [2–7]
Wound‐QoL‐14	Global score	Mean (SD)	1.71 (1.06)	1.72 (1.13)
		Median [P25‐P75]	1.57 [0.79–2.57]	1.43 [0.71–2.79]
Body subscale		Mean (SD)	1.39 (1.13)	1.52 (1.19)
		Median [P25‐P75]	1.25 [0.50–2.00]	1.25 [0.50–2.50]
Psyche subscale		Mean (SD)	2.23 (1.20)	2.18 (1.22)
		Median [P25‐P75]	2.25 [1.25–3.25]	2.25 [1.25–3.25]
Everyday Life subscale		Mean (SD)	1.62 (1.35)	1.59 (1.39)
		Median [P25‐P75]	1.40 [0.40–2.80]	1.2 [0.20–2.80]

Abbreviation: PAD: peripheral artery disease.

^a The term ‘Ischemic’ referred to other ischemic conditions different to PAD, such as arteriolosclerosis.

4.1. Interpretability: Bands of Severity

All categories were represented in the GAQ (Figure S1).

The correlation between the Wound‐QoL‐14 scores at T ₀ and the GAQ was higher than 0.5 (rho = 0.627; p < 0.001) (Figure S2), although the categories showed a great level of overlap (Figure S3).

A total of 12 bands were examined in order to identify cutoff values for stratifying patients according to HRQoL impairment. Then, and for the sake of simplicity, the cutoff points defining the best band (band #8) were rounded to ease its clinical use and interpretation (band #11: Not at all: 0.00–0.20/Slightly: 0.21–1.00/Moderately: 1.01–2.25/Strongly: 2.26–3.20/Very strongly: 3.21–4.00) (Table S1). The frequency of distribution of GAQ scores according to Wound‐QoL‐14 bands is shown in Figure S4.

Regarding top box responses, patients answered approximately one third of the items of the Wound‐QoL‐14 with one of the two upper response options (mean ± SD: 5.10 ± 4.37/Median [P25‐P75]: 4 [1–8.5]). As expected, the number of top box responses strongly correlated with patients' Wound‐QoL‐14 score (rho = 0.948; p < 0.001) (Figure S5).

4.2. Smallest Detectable Change

As previously reported [12], no relevant floor or ceiling effects were found (0.3% and 1.5%, respectively). GRC‐P and GRC‐HW scores are shown in Table 2. As expected, the correlation between them was excellent (polychoric correlation = 0.932) as well as interrater reliability (ordinal Krippendorff's α = 0.885) with a good level of agreement (73.6%).

TABLE 2.

Global rating of change anchor question for patients and healthcare workers.

	Answer to the anchor question	Patients, n (%)	Wound‐QoL‐14 baseline, mean ± SD	Wound‐QoL‐14 follow‐up, mean ± SD	Wound‐QoL‐14 increment (T 0 − T 1), mean ± SD
GRC‐P	Much worse	1 (0.5)	2.64 (NA)	3.29 (NA)	−0.64 (NA)
	Moderately worse	7 (3.5)	1.44 (0.99)	1.89 (0.91)	−0.45 (0.46)
	Slightly worse	9 (4.6)	1.24 (1.02)	1.49 (0.97)	−0.25 (0.87)
	Unchanged	15 (7.6)	1.95 (1.31)	1.95 (1.02)	+0.005 (0.59)
	Slightly better	39 (19.9)	1.83 (1.28)	1.39 (1.02)	+0.44 (0.88)
	Moderately better	61 (30.9)	1.51 (0.98)	0.95 (0.82)	+0.56 (0.71)
	Much better	65 (33.0)	1.89 (1.14)	0.78 (0.76)	+1.11 (1.05)
GRC‐HW	Much worse	2 (1.0)	1.47 (1.66)	2.15 (1.62)	−0.68 (0.05)
	Moderately worse	5 (2.5)	1.54 (0.82)	1.64 (0.90)	−0.10 (1.03)
	Slightly worse	14 (7.1)	1.47 (1.24)	1.55 (1.07)	−0.08 (1.04)
	Unchanged	13 (6.6)	1.44 (1.30)	1.54 (1.14)	−0.09 (0.53)
	Slightly better	43 (21.8)	1.93 (1.18)	1.42 (0.96)	+0.52 (0.81)
	Moderately better	54 (27.5)	1.73 (1.07)	1.12 (0.95)	+0.61 (0.90)
	Much better	66 (33.5)	1.71 (1.11)	0.70 (0.68)	+1.00 (0.96)

Abbreviations: GRC‐HW: Healthcare worker global rating of change; GRC‐P: Patient global rating of change.

SDC‐P and SDC‐HW for Wound‐QoL‐14 and domains were calculated based on the analysis of the 30 and 29 patients who remained unchanged, respectively (Table S2). In general, SDC‐P values were lower than SDC‐HW values (Table 3).

TABLE 3.

Smallest detectable change of the Wound‐QoL‐14 and domains for patients and healthcare workers.

	SDC patients (n = 30)		SDC healthcare workers (n = 29)
	Individual	Group	Individual	Group
Global score	0.39	0.07	0.75	0.14
Body subscale	0.64	0.12	0.71	0.13
Psyche subscale	1.13	0.21	1.64	0.30
Everyday Life subscale	0.46	0.08	0.86	0.16

Abbreviation: SDC: smallest detectable change.

4.3. Minimal Important Difference

The 7‐points anchor question was valid to calculate anchor‐based MIDs for change since all Spearman correlation coefficients between both the GRC‐P and GRC‐HW and the change in the scores of the Wound‐QoL‐14 and domains were statistically significant (p < 0.001) and higher than the preestablished threshold of 0.3, except for the Everyday life domain (Table S3).

The ‘Slightly improved’ subgroup was constituted by 26 patients for GRC‐P and 32 for GRC‐HW, whereas the ‘Slightly deteriorated’ subgroup was only formed by 7 patients for GRC‐P and 9 for GRC‐HW. After examining the shape of the distribution of the Wound‐QoL‐14 global change scores for the three GRC‐P and GRC‐HW groups, MIDs for improvement and for deterioration were deemed different enough to deserve to be calculated separately (Figure S6). However, ‘Slightly deteriorated’ subgroups both for GRC‐P and GRC‐HW were deemed to be too small for a reliable calculation of the MIDs for deterioration. Consequently, only anchor‐based MIDs for improvement were calculated.

All MID values for improvement calculated according to six methods as well as the weighted mean for Wound‐QoL‐14 and its domains are shown in Table 4. The relationship of all these MID values together with the SDC is depicted both for patients and healthcare workers (Figure 2). Only two reliability ratios out of eight were higher than 1, specifically for Wound‐QoL‐14 global score and Everyday Life domain for patients.

TABLE 4.

Minimal important difference for improvement of Wound‐QoL‐14 and its domains for patients and healthcare workers.

		Anchor			Distribution			Weighted mean
		AC	SC	ROC	0.5*SD	1.96*SEm	SRM
Patient	Global score	0.64	0.59	0.33	0.48	0.55	0.62	0.53
	Body subscale	0.78	0.58	0.38	0.57	0.86	0.62	0.61
	Psyche subscale	0.53	0.67	0.38	0.59	1.11	0.51	0.60
	Everyday Life subscale	0.55	0.45	0.50	0.57	0.64	0.44	0.52
Healthcare worker	Global score	0.65	0.51	0.25	0.34	0.53	0.62	0.48
	Body subscale	0.79	0.51	0.38	0.31	0.50	0.62	0.53
	Psyche subscale	0.56	0.65	0.13	0.51	3.21	0.51	0.77
	Everyday Life subscale	0.61	0.43	0.50	0.40	1.69	0.44	0.62

Abbreviations: AC: average change; ROC: receiver operating curve; SEm: Standard error of measurement; SC: social comparison; SD: standard deviation; SRM: standardized response mean.

FIGURE 2.

Graphic representation of minimal important difference for improvement for the Wound‐QoL‐14 survey and its domains both for patients (left panel) and health workers (right panel). Weighted means are joined by the dashed line, and smallest detectable change by the dotted line. AC: average change; ROC: receiving operator characteristic (Youden method); SC: social comparison; SDC: smallest detectable change; SRM: standardized response mean. 0.5*SD: Half the standard deviation. 1.96*SEm: Upper limit of the confidence interval of the standard error of measurement.

5. Discussion

Assessing interpretability allows HRQoL instruments to be more connected to the realities of clinical decision‐making. Severity bands and MIDs enable the evaluation of the impact of care improvement programs, educational interventions, or new therapies from the patient's perspective [16]. Despite the above, there is little information available on the interpretability and estimation of the MID for HRQoL questionnaires in patients with chronic wounds. The reason could be that most studies to date have focused primarily on the psychometric validation of these instruments [3, 20, 21, 22] paying little attention to their interpretability or clinical utility in routine practice [10, 11]. Our results help fill this gap by providing empirically derived cutoff points and MID estimates for the Wound‐QoL‐14 in a Spanish population with chronic wounds.

5.1. Bands of Clinical Severity for Wound‐QoL

Regarding bands of clinical severity, Janke et al. [7] recently published the first study proposing cutoff points for the Wound‐QoL‐17 and Wound‐QoL‐14 versions. However, the Wound‐QoL‐14 scores were not measured directly but were derived from the 17‐item version. These authors did not intend to determine the optimal cutoff points, but rather they sought cutoff values that would be easy to apply in clinical practice. This was an essential methodological difference to us, as we were looking for the best thresholds. However, the bands proposed by Janke et al. [7] could serve as a starting point for future studies seeking to determine reference bands for other countries, given that population differences (e.g., in pain tolerance) make it advisable for each country to conduct its own studies to assess the variability of its thresholds. In line with this, the severity bands we proposed largely coincide with those of Janke et al. The key difference is that our bands are not equidistant, suggesting that there was a clinical or functional justification behind each cutoff point [16].

The best level of agreement between actual GAQ ordinal categories and ordinal categories determined by applying the best band cutoffs was found to be moderate by both Janke et al. and us, despite the methodological differences outlined below. These authors assessed the association using Spearman's rank correlation coefficients and the level of agreement using the weighted kappa coefficient of agreement [7]. Instead, we used polychoric correlations and Krippendorff's correlation coefficient, which are more suitable for handling ordinal variables [13, 23]. Regardless of the methodology, we found a moderate level of agreement because of the great extent of overlap in scores between GAQ categories (see Figure S5). This problem will be shared by any other researchers who would find a similar overlap in GAQ scores.

Clinically, categories related to impairment appear to be the most important since stratification by severity can aid in the more efficient allocation of resources such as time, materials, or personnel. Therefore, taking into account only top boxes could be a possible way to overcome the limitation caused by overlapping scores, as recently reported by Janke et al. [7]. In fact, both Janke et al. and we found a strong and significant correlation between Wound‐QoL‐14 scores and the number of top boxes scored by patients. That coincidence strongly suggests that taking these responses into account could be a useful, simple, and straightforward approach in clinical settings that would help overcome the limitation imposed by overlapping scores on the GAQ.

5.2. Smallest Detectable Change

Beyond patient stratification, we also determined the individual and group SDC. Establishing an individual SDC allows differentiation between changes in HRQoL scores due to random variability and real changes. Additionally, establishing a group SDC allows us to interpret whether observed changes in group mean values reflect a real change beyond the measurement error. This is particularly important in clinical trials and population studies, where conclusions are often drawn at the group level. Despite this, and to the best of our knowledge, there is only one study involving the Wound‐QoL‐17 which was performed on Danish patients [11], whereas the SDC for the Wound‐QoL‐14 questionnaire had not been evaluated until now. Compared to our findings, individual SDC values reported by Knudsen et al. were notable higher [11]. The huge discrepancy in individual SDC values can be partly explained by the reliability coefficients because, as shown in Equation (1), the SDC value is higher the lower the reliability coefficient, and vice versa. Knudsen et al. used the classical intraclass correlation coefficient for agreement (ICC) as the reliability coefficient, whereas we selected the CCC, which is more suitable for non‐normally distributed samples. As an example, their ICC of 0.88 for global score yielded and individual SDC of 0.77 points, while our higher CCC of 0.915 yielded a lower SDC of 0.39 points.

We expanded the calculation of SDC beyond patients to also include healthcare workers. Interestingly, individual and group SDC‐HWs for the global Wound‐QoL‐14 score were twice as high as SDC‐Ps, indicating that clinicians need to observe greater changes than patients to consider the change as relevant.

5.3. Minimal Important Difference

As for the MID, it would be useful to determine what minimal changes in the score of any HRQoL instrument―and, eventually, in its subscales―would constitute a significant improvement or deterioration. This, in turn, would help clinical decision‐making and also sample size calculation of clinical trials [14]. However, MIDs for HRQoL instruments focused on chronic wounds have barely been determined. To the best of the authors' knowledge, there are two reports on the Wound‐QoL‐17 questionnaire [10, 11]. The MID for the Wound‐QoL‐14 had not been determined until now. We explored several methods for calculating MIDs, which included three anchor‐based and three distribution‐based methods, and obtained six different MID values for patients and healthcare workers. Then, for the sake of simplicity, we proposed a single MID value of 0.5 points for the global scale and of 0.6 points for its domains.

The lack of similar studies prevents a direct comparison of our results, but an indirect comparison can be made with the two studies in which the longer version of the Wound‐QoL‐17 questionnaire was used [10, 11]. In short, our results were consistent with those found by Topp et al. [10], but largely different from the MID values obtained by Knudsen et al. [11].

Topp et al. calculated a MID estimate of ±0.47 points using a distribution‐based method (n = 221), whereas using anchor‐based methods they estimated a MID for improvement of 0.52 points (n = 89) and a MID for worsening of 0.47 points (n = 10). Overall, they recommended a change of 0.5 points in the Wound‐QoL‐17 as an appropriate MID for German patients with chronic wounds [10]. Our proposal is largely consistent with this recommendation.

Knudsen et al. calculated a MID estimate of 1.36 points using a distribution‐based method and of 1.29 points using an anchor‐based method in 155 Danish patients with chronic wounds [11]. These authors merely described the results, but they did not recommend any single MID value.

Since both papers were published almost simultaneously and the authors were unaware of each other's article, the reason for such a huge difference in MID values was not discussed. It is known that there can be great variability in MID depending on the method used to calculate it [19], and our findings are a good example of this. Additionally, the variability in MIDs could be explained, at least in part, by fundamental methodological differences in the grouping of patients according to the anchor question.

Topp et al. used a 5‐point Likert scale as an anchor question, whereas Knudsen et al. and we used a 7‐point Likert scale. Neither Topp et al. nor Knudsen et al. asked the patients if they found the difference in wound status clinically important, as we did. From here, the criteria for grouping patients were different. Topp et al. formed the group ‘No change’ simply with the patients who answered this option to the anchor question [10]. Knudsen et al. formed the group ‘No change’ by combining the three categories that contained the patients who responded ‘No change’ and ‘A little better/worse’ [11]. We further refined this method to form the group ‘Unchanged’ by taking into account if the patients who responded ‘Slightly better/worse’ had also found the difference not important, as previously reported [14]. The following recalculation serves as an example of how patient grouping can partly affect the MID. We calculated a MID for improvement of 0.64 points according to the described anchor‐based AC method, but this value would increase to 0.75 and 0.84 points after applying to our patients the grouping methodology used by Topp et al. and Knudsen et al., respectively. However, the difference between the MID for improvement calculated by Knudsen et al. (1.29 points) and the MID proposed by Topp et al. and us (0.50 points) is so large that there must be more explanatory factors we are currently unaware of. In any case, according to the MID value calculated by Knudsen et al., patients must score more than 1 point of change to consider such change as clinically relevant. Such a large change seems unrealistic, considering that the scale ranges from 0 to 4 points.

Overall, we recommend 0.50 points as MID for improvement of the global score of the Wound‐QoL‐14 based on four facts. First, 0.50 points is consistent with the proposal by Topp et al. for the Wound‐QoL‐17 using different methodology in a different population [10]. Second, we used a combination of distribution‐ and anchor‐based methods, as recommended [16]. Concerning anchor‐based methods, they included both patient‐ and clinician‐reported baseline questions regarding perceived change in wound status. This dual‐reference approach captures both subjective experience and clinical perspective, providing a more complete and robust interpretation of meaningful change [17, 24, 25]. Third, our data reasonably met nine out of the 10 core criteria for selecting an optimal MID, which were proposed by Wan et al. [8] after expanding the credibility assessment framework developed by Carrasco Labra et al. [25]. The exception was the precision of MID (criterium number four), which could probably be resolved by recruiting a larger sample of ‘Unchanged’ patients. Fourth, the reliability ratio for the MID for improvement of the global score of the Wound‐QoL‐14 was higher than 1.

Regarding domains, Topp et al. calculated MIDs ranging between 0.41 and 0.63, but they considered such variability a great discrepancy and, consequently, did not make a general recommendation for subscales [10]. In our case, MID‐Ps ranged from 0.52 to 0.61 and MID‐HWs ranged from 0.53 to 0.77. Despite the degree of variability being similar to that found by Topp et al., we think it is reasonable to recommend a single MID for improvement of 0.60 points for all domains. However, this is a weak recommendation that should be taken with caution until further research is conducted, as most reliability coefficients for the domains were lower than 1. Again, values for domains calculated by Knudsen et al., which ranged from 1.20 to 1.76 [11], were deemed too high to be taken into consideration.

Until now, anchor‐based MIDs were calculated based exclusively on patients' perspective [10, 11] despite the recommendation to include clinicians' perspective [16] as we did. We found similar values for MID‐Ps and MID‐HWs, which, in our opinion, make our findings more robust.

5.4. Limitations

Several limitations should be acknowledged when interpreting the results of this study. First, external validity could be compromised by the lack of information regarding certain aspects of the participation process, such as the number of eligible patients who declined to participate and the level of adherence to the study protocol across the collaborating centres.

Second, a possible double selection bias should be considered. On the one hand, it is possible that patients who had more favourable clinical conditions were more likely to participate. On the other hand, healthcare professionals who chose to participate may be those most committed to evidence‐based optimal wound care, which could positively influence the observed results. In line with this, an additional limitation is that the sensitivity of the Wound‐QoL‐14 to detect clinical deterioration was not evaluated due to the small number of patients in our sample who reported a worsening of their wound status during the follow‐up period. We cannot know if the lower number of patients who worsened was due to the high quality of care or because the patients who worsened were less likely to participate in the retest phase.

Third, the study design did not include domain‐specific anchor questions at baseline, which made it impossible to evaluate the interpretability of domain scores.

The last limitations were related to the ‘Unchanged’ subgroup. The sample size was relatively small and, besides, we could not rule out the possibility that some patients reported ‘No change’ because they did not accurately recall their previous condition after the one‐month period. For future research, we recommend adding a first short‐term re‐test interval in order to obtain a larger group of patients classified as unchanged and to avoid recall bias. As an example, Knudsen et al. re‐administered the questionnaire within a shorter interval (1 week), which resulted in a large number of participants being categorized as having ‘no change’ than in our case [11].

Despite these limitations, we believe that the present study provides valuable information on the interpretability of the Wound‐QoL‐14 in real‐world clinical settings and offers practical guidance for clinicians and researchers working with patients affected by chronic wounds.

6. Conclusion

We conducted a study to define the interpretability thresholds of the Wound‐QoL‐14 instrument, as well as MIDs and SDCs for patients and healthcare workers, using a robust, multi‐method approach. By integrating patient stratification with a rigorous estimation of MID values, we provided a comprehensive framework to improve both clinical decision‐making and outcome monitoring in the field of chronic wound management. Future research should aim to validate these thresholds in diverse populations and explore their applicability in longitudinal studies assessing treatment effects. At the same time, the harmonization of methodological criteria in the design of future studies is of utmost importance to allow for comparisons.

Akcnowledgements

The authors are especially grateful to the following Spanish scientific societies for disseminating and publicizing participation of health professionals dedicated to wounds in the AlbaMad study: Asociación Española de Enfermería Vascular y Heridas (AEEVH), Grupo Nacional para el Estudio y Asesoramiento en Úlceras por Presión y Heridas Crónicas (GNEAUPP), Sociedad Española de Heridas (SEHER), Sociedad Gallega de Heridas (SGH). Also, the authors are grateful to Diputación de Albacete and Universidad de Castilla‐La Mancha for funding the Article Processing Charges.

Funding

The authors received funds from Diputación de Albacete and from Universidad de Castilla‐La Mancha to pay the Article Processing Charges, but these organizations had no role during the preparation of this manuscript.

Disclosure

The authors have nothing to report.

Ethics Statement

This study was conducted after obtaining approval from our Institutional Review Board (Comité de Ética e Investigación con medicamentos de la Gerencia de Atención Integrada de Albacete; internal code 2023‐140).

Consent

Patients provided written informed consent before participating.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Distribution of responses to the anchor global question at baseline (0‐Not at all/1‐Slightly/2‐Moderately/3‐Strongly/4‐Very strongly). Categories represented in the GAQ: Not at all: 9% (n = 30)/Slightly: 25% (n = 80)/Moderately: 30% (n = 98)/Strongly: 21% (n = 68)/Very strongly: 15% (n = 47)/Mean ± SD: 2.07 ± 1.12/Median [P25‐P75]: 2 [1–3].

Figure S2: Distribution of the Wound‐QoL‐14 scores by the global question categories at baseline.

Figure S3: The overlap between the global question and Wound‐QoL‐14 scores explains why severity bands cannot achieve a high level of agreement.

Figure S4: Frequency distribution of global question scores according to selected Wound‐QoL‐14 cutoff points (band #11 in Table S1).

Figure S5: Scatterplot of Wound‐QoL‐14 scores with the number of top box responses (“quite” and “very”).

Figure S6: Distribution of the Wound‐QoL‐14 global change scores for the ‘Slightly deteriorated’ (upper), ‘Unchanged’ (middle), and ‘Sightly improved’ (bottom) subgroups, according to the anchor question for patients (left panel) or health workers (right panel). ‘Slightly deteriorated’ and ‘Sightly improved’ subgroups showed quite different density shapes.

Table S1: Agreement between bands evaluated with polychoric correlations and ordinal Krippendorff's α coefficient of agreement. Blue band #8 represents the category with the highest level of agreement, whose cutoff points were rounded to facilitate clinical application in orange band #11 (our proposed band). Yellow band #12 corresponds to the application to our sample of the cutoff points proposed by Janke et al. 7.

Table S2: Individual smallest detectable change for the Wound‐QoL‐14 instrument for both patients and healthcare workers.

Table S3: Spearman rank correlations between both the 7‐points anchor question for both patients and healthcare workers and scores of Wound‐QoL‐14 and subscales.

Horcajada‐Reales C., Gerónimo‐Pardo M., Augustin M., and Selva‐Sevilla C., “From Data to Clinical Decisions in Chronic Wound Care: Evaluation of the Interpretability, Minimal Important Difference, and Smallest Detectable Change of the Wound‐QoL‐14 in a Spanish Population,” International Wound Journal 23, no. 10 (2026): e71065, 10.1111/iwj.71065.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.