Development, validity and reliability of the Italian version of the Copenhagen neck functional disability scale

The CNFDS consists of 15 items that evaluate the impact of NP on headache, ability to sleep, concentration, activities related to work, daily activities and leisure activities. It includes also questions of psychosocial nature, such as decreased social contact, influence on emotional relationships with family members and attitudes toward the future. Each item can be answered as ‘yes’ (0 points), ‘occasionally’ (1 point), and ‘no (2 points). To avoid repetitive answering, response options are reversed after the fifth question. The total score can range from 0 to 30, with higher scores indicating worse disability.

The cross-cultural adaptation process was carried out in accordance with the recommendations proposed by Beaton et al. [16].

Step 5: Pretesting. The prefinal CNFDS was administered to 30 participants responding to the study inclusion criteria and who were asked about any misunderstanding, conflicting, or ambiguous word or sentence. These participants did not have any questions about the scale and all the questions were well understood. There was no multiple answer question or missing answer. After the pretesting, the final version of the scale was obtained (i.e. CNFDS-I), Additional file 1: Table S1.

Consecutive outpatients seeking treatment or evaluation between March 2015 and February 2016 at the Physical Medicine and Rehabilitation Unit of “Organizzazione Sanitaria Meridionale Assistenza Inabili Recupero Minori” (OSMAIRM) were assessed for inclusion.

The inclusion criteria were: non-specific chronic NP (lasting > 12 wk) with or without arm pain, at least 18 years old, and ability to read and speak Italian fluently. The NP area was defined according to the IASP definition [17] as the area bounded by the nuchal fold on the top, an imaginary transverse line passing through the tip of the first chest spinous process on the bottom, and a sagittal suture tangential to the lateral edges of the neck on both sides.

Patients with neurological signs in the arms were excluded because there was a lack of instrumental examination for confirming the diagnosis (i.e. cervical radiculopathy) [18]. Other exclusion criteria were: lawsuits, cognitive impairment, fractures, cancer central neurological signs and severe psychiatric disorders.

A booklet asking information about demographic (e.g. age, sex) and clinical characteristics (e.g. pain duration) was administered to each patient. The same booklet contained the Italian versions of the Neck Disability Index (NDI-I) [14], of the Neck Bournemouth Questionnaire (NBQ-I) [15], and of the CNFDS-I; a Visual Analogue Scale (VAS) to assess pain intensity was also included [19].

The NDI is the most commonly tested and translated neck-related disability scale [20]. It consists of 10 items and each question is scored on a 6-point scale ranging from 0 (no disability) to 5 (full disability). Total score ranging from 0 to 50 which can also be expressed as a percentage. The NBQ is a short questionnaire consisting of seven items representing aspects of the biopsychosocial model relevant to patients with NP [21]. Each item is scored on a 0–10 numerical scale, where zero represents absence of limitation, for a total score ranging from 0 to 70 points. The NBQ-I has displayed acceptable construct validity and responsiveness in Italian patients with chronic NP [15].

The VAS is a horizontal line, 100 mm in length, asking patients to rate their pain intensity at the moment, “no pain” and “worst pain” are the extremes. The VAS has exhibited satisfactory test-retest reliability and construct validity in patients with chronic pain [22].

The COnsensus-based standards for the Selection of health Measurement INstruments(COSMIN) definitions of measurement properties were adopted in this study [23].

All the questions were well accepted, there were no problems in the instrument’s comprehensibility. The questionnaire was completed with a mean of 120 s (standard deviation = 40). No missing responses or multiple answers were found.

Exploratory factor analysis revealed that the first factor explained 83% of the total variability (eigenvalue = 4.122), while the second factor explained only 13% (eigenvalue = 0.648). The ratio between the first and the second eigenvalues was equal to 6.36. The scree plot (Fig. 1) also indicated a clear unidimensional pattern for the CNFDS-I. Factor loadings are presented in Table 2 and show as the item 9 appear deflected. We decided not to delete this item because we consider its content (Have you been bothered by headaches during the time that you have had neck pain?) very relevant to patients with NP, because it presented a sufficient item total correlation (Table 2), and because we would not like to create a CNFDS Italian version that differs from other versions in the number of items.

The Cronbach α for the scale was 0.83, indicating satisfactory results. Cronbach’s α when an item was excluded ranged from 0.12 to 0.69, while correlation of each item with the total score ranged from 0.24 to 0.76 (Table 2).

CNFDS-I correlations with the NDI-I and the NBQ-I were equal to 0.846 and 0.708, respectively. The correlation between the CNFDS-I and the VAS was 0.570. All these correlations met the expected hypotheses and the CNFDS-I construct validity was considered satisfactory.

Summary in Table 3.

The ICC_agreement was equal to 0.997 with a 95% confidence interval ranging from 0.995 to 0.999. This result represents excellent test-retest reliability.

The SEM_agreement equalled 3.0 scale points (10% of the scale range), while the SDC was 8.31 scale points (27% of the scale range).

The CNFDS-I had no floor or ceiling effects, in fact no patients with a minimum and maximum score were identified in the test nor in the retest.

This study describes the cross-cultural adaptation and evaluation of the CNFDS-I in patients with chronic NP. The CNFDS-I is a unidimensional and internally consistent tool with excellent test-retest reliability and sufficient construct validity; the SDC was equal to 8.31 scale points (27% of the scale range). These results indicate that the CNFDS-I is ready to be used as a measuring method of neck-related disability in Italian patients with chronic NP.

This is the first study to perform a factor analysis of the CNFDS as the original study [8], while the previous ones [33‐36] did not do that. The CNFDS-I resulted to be unidimensional and this represents a key finding, as it indicates that it is appropriate to use its total sum score [24]. Meanwhile, this finding does not support the use of the subscales proposed by the French developers [35]. In contrast with the CNFDS-I, other neck-related disability tools available in Italian (i.e. NDI-I and NPQ-I) were not shown to be unidimensional [14, 15, 37], questioning the suitability of using their total sum scores.

The unidimensionality of the NDI has been questioned also by other studies in other language versions [38, 39]. A future study can aim at comparing directly the unidimensionality of all these tools, to check if the CNFDS is indeed the best performing method from a psychometric point of view. Item 9 was the only CNFDS-I item with a low factor loading; future studies should assess if this result is repeated in other languages.

The Cronbach’s α of the entire questionnaire (α = 0.83) was major than the predefined threshold, and it was similar to the values obtained in the original version (α = 0.90) French (α = 0.83) Polish (α = 0.90) and Iranian (α = 0.84) [33‐36]. The comparison is simplified in Table 3. This indicates a high interrelation of the items.

Construct validity was analyzed by comparing the CNFDS-I to the NDI-I and NBQ-I. The very high correlation between CNFDS-I and NDI-I (ρ = 0.846) suggests that the theoretical construct of these two instruments may be very similar [32]. We chose the NDI as a comparative questionnaire because it is the most validated and internationally used [15, 40]. The correlation between the total scores of CNFDS-I and NBQ-I was high too (ρ = 0.708). A second disability comparator questionnaire was included to compare the biopsychosocial aspects examined by the NBQ-I [13, 21]; the correlation between the two questionnaires outlines that their content is not totally different as described by Ferreira et al. [13], possibly because they may differ across cultures. Foremost, to fully compare the content validity of these different instruments, it is necessary to conduct a head-to-head comparison study asking patients and clinicians if the most relevant neck-related disability aspects are included in each questionnaire [8]. Additionally, a consensus-based definition for neck-related disability should be established before further content assessment of the tools. These considerations are not totally new, as a recent systematic review clearly highlighted that the content validity of patient-reported outcome measures is understudied [41], and that a major effort should be made to fill this evidence gap.

The authors of French version [35], compared the CNFDS with the VAS scale, found a moderate correlation (0,45 Spearman’s r).

Similarly, we have identified a correlation of ρ = 0.570. This result reflects those by Fejer [42], highlighting the moderate correlation between NP intensity and disability, which are strongly associated. Therefore, the CNFDS and the VAS evaluate different aspects, though maintaining a certain degree of correlation.

Moreover, after validating the CNFDS in Iran, Azhari et al. [43] investigated the most important aspects covered by the CNFDS, emphasizing the fact that its primary use is to gather and measure the patients’ disability, not their perceived pain.

CNFDS-I test-retest reliability was excellent as highlighted in the original development study [8] and other studies [24, 34‐36]. The ICC agreement represents an absolute measure of reliability and it is suggested that it should be interpreted in combination with a relative measure such as the SDC [44]. The SDC of the CNFDS-I was found to be 27% of the scale range; this may suggest that repeated measurements with this tool do not give similar results, but there is not a standard cut-off value to determine whether a SDS is indeed small enough. Moreover, a measurement error larger than 20% of the scale range has been found for various broadly used questionnaires [45]. A comparison between SDC and minimal important change should be performed to have a better insight into the CNFDS measurement error. In light of these considerations, future studies on the CNFDS should calculate its minimal important change to be able to better interpret this instruments’ changes in scores.

This study did not assess responsiveness, defined as the ability of an instrument to detect changes in the construct to be measured [23]. It is an important measurement property to use an instrument as an outcome measurement instrument and it has not been evaluated in any language for the CNFDS. Overall, it remains unclear which neck-related disability instrument should be preferred in Italian patients with NP; thus, there is an urgent need of a head-to-head clinimetric study comparing content validity, structural validity and responsiveness of various instruments (including NDI-I and NBQ-I) in the same patients.