Adaption and psychometric evaluation of the Hindi version of Neck Disability Index in the rural population of Northern India: A cross cultural study

Ethical approval

The translation, adaptation, and psychometric testing of the Hindi version of NDI was performed for the tool requirement of a subsequent registered clinical trial among Hindi speaking rural North Indian population. The present study was phase 1 of the clinical trial and includes a different study population. This study was approved by the Institutional Review Board of the Madhav University, Rajasthan (Ref. No: MU/IEC/21/14; dated 24/07/2021), and written informed consent for participation was obtained before patient recruitment. All procedures were conducted in accordance with the declaration of Helsinki. The purposes and the importance of the study were clarified to each participant. Participants were free to refuse to participate or answer any of the questions throughout the course of the study to ensure data confidentiality at all levels of the study, names of participants and any personal identifiers were not included. This study was part of larger trial that was registered prospectively with CTRI/2021/08/036040.

Participants

This study was conducted at the Physiotherapy outpatient department (OPD) of Sanskriti University Mathura, India and Jeevan Jyoti Hospital Mathura in the Northern State of India (Mathura city, Uttar Pradesh) from August 2021 to June 2022. Hence, the study protocol was presented to the Ethical member at the Madhav University and approval obtained. Further, the Ethical approval letter and the study proposal was submitted to the Clinical Trial Registry of India (Government of India) CTRI, which verified the application and granted permission to conduct the definitive study. The study settings are the associate centers of Madhav University, and proper permission was obtained from the two centers to conduct the study. A total of 220 participants with CNSNP were recruited across two centers, of which 211 participants (114 females and 97 males) completed all participant-reported socio-demographic data, clinical information using face to face interviews as the participants in general visit the study center for consultations and physical rehabilitation, hence we decided to use interview method to improve the quality of data collected. by trained assessors. Neck Disability Index – Hindi version (NDI-Hi) and Neck Pain and Disability Scale (NPAD) (Agarwal & Allison, 2006). This study followed recommendations of best practices for developing and validating scales (R. Article, 2017), which suggests at least 5-10 participants per item in the tool as recommended by the COSMIN checklist (Mokkink et al., 2010).

Considering the lack of specific guidelines and different authors used different formula to estimate sample size like; factor analysis formula, Kaiser-Meyer-Olkin sampling adequacy (values ranging from 0.50 to 0.60), regression formula and item-sample ratio estimation (Arafat & Chowdhury, 2016). We intended to include all eligible patients with CNSNP attending both the study center’s during the study period to improve the statistical estimations and consistency of the results.

Inclusion and exclusion criteria

Individuals aged 18 years and above, living with CNSNP for > 3 months duration, residing in rural area, who understand and speak Hindi language, and those willing to participate were included in this study. Individuals with acute neck pain, diseases causing neck disability, previous cervical spine injury, prolapsed intervertebral disc disease, traumatic vertebral fractures, neck surgery, clinically recognizable cognitive impairment, pregnancy, and those who could not read and speak Hind were excluded from the study. Individuals with chronic non-specific neck pain visiting the study area primarily for other problems were also excluded.

Instruments

Neck Disability Index (NDI)

The NDI, originally developed by Vernon and Mior (Vernon & Mior, 1991) is a 10-item or section patient-reported outcome measure designed to assess neck pain and pain-related disability. The tool consist of 10 sections designed to self-report on how neck pain impacts the activities of daily life, the domains are; intensity of pain, personal care (dressing, washing), lifting heavy and light weights, reading, headaches, concentration, work, driving, sleeping, and recreation. Each section consists of six activity-based exclusive responses expressing progressive levels of functional disability. Item scores range from 0 (no disability) to 5 (total disability) and Vernon also established categorization of scores as; 0 to 4 no disability, 5 to 14 mild disability, 15 to 24 moderate disability, 25 to 34 severe disability, and ≥ 35 complete disability. The total score is out of 50 (Vernon & Mior, 1991) and is also frequently normalized to 100 and reported as a percentage. The NDI (one-dimensional questionnaire) measures a single construct, and the questionnaire has demonstrated moderate differences in validity and reliability with a different patient population (Pietrobon et al., 2002).

Neck Pain and Disability Scale (NPAD)

The NPAD scale is a 20-item multi-dimensional questionnaire designed to assess neck pain and disability. The items are scored along a 10 cm visual analog scale (VAS) with the scores ranging from 0 to 5 in VAS. A higher score indicates greater disability. The Hindi version of NPAD has been cross-culturally translated and validated among 64 Indian individuals with cervical radiculopathy. The Hindi version of NPAD has been psychometrically tested and shown to have acceptable internal consistency and validity (Agarwal & Allison, 2006).

Translation and cross-cultural adaptation process

Tool adaptation and translation was conducted as per the guidelines suggested by Beaton et al. (2000). The translation and adaptation process for NDI consists of the following steps; forward translation, synthesis, backward translation, expert committee review, and synthesis, tool pretesting, and submission and appraisal of the written reports by the translation-adaptation coordinating committee (Beaton et al., 2000). In step one, forward translation was performed from English to Hindi language by two Hindi native translators (a professional translator and a post-graduate employee working in education department) who were fluent in the English language. One independent translator was blinded to process of forward translation of NDI. This was done to make sure the equivalency from a therapeutic point of view rather than literal equivalence. The other translator was informed about the purpose of the study and the concepts being studied. This was for contemplating the language used by the population and spotlight terms in the original questionnaire, the translation of which might have been obscure. Two forward-translated documents (T1 and T2) were produced. In step two, the independently translated T1 & T2 documents were shared among the two translators to synthesize NDI-Hi (version 1). Any inconsistencies, differences in the concepts and/or meaning, elusive wordings were sorted by discussion and agreement. In step three, the first draft NDI-Hi was back-translated to English language by two bilingual independent translators (native Hindi language speaking teachers with master’s degree in English) and not familiar with the construct being assessed and blinded to the process of forward translation. After discussion and agreement between back translators and the principal investigator (MS), a third translation was performed. In step four, all three documents were presented to the expert panel for review and discussion. The expert committee was comprised of four translators, two review committee experts from the university and two senior physiotherapy academicians who reviewed the Hindi version of the questionnaire. The panel discussed clarity, relevance, modifications, comprehension and synthesized the pre-final version for field testing Figure 1.

Figure 1. Study flowchart.

NDI, Neck disability index.

Pilot study and modifications

In step five, upon approval of the panel, NDI-Hi was piloted on 30 individuals with CNSNP (19 females and 11 males). A cognitive (qualitative) debriefing and the pretest pilot of the pre-final NDI-Hi was conducted to assess; 1. The time taken to complete the tool and clarity of the instructions in the tool (administration). 2. Internal consistency, if the items are logically ordered, necessity of the items in the tool (organization). 3. If the questions are direct, whether the questions are free from technical jargon, cultural appropriateness. As a result, the majority of participants reported a good understanding of the tool. However, 16 participants reported section eight ‘driving’ not applicable to them. Based on cognitive debriefing comments made by the participants, the committee agreed to reword section eight (driving) to fit the cultural context in India where most people do not drive car, rather travel by public and/or private transport and motorbike. For the participants who did not answer/drive, the total score was adjusted from 45 instead of 50 (multiplying 50/45). The final Hindi version of the NDI was then compared with the original one to achieve representational, fact-finding and contextual equivalence.

Psychometric testing of the NDI-Hi

Psychometric properties were assessed using the final version of the NDI-Hindi (Sidiq, 2023) on 211 patients with CNSNP for > 3 months duration across three centers. The test-retest reliability of the NDI-Hi version was determined by calculating Cronbach’s alpha with an alpha value of > 0.7 (deemed acceptable), > 0.8 (considered good), and > 0.9 (considered excellent) to grade the internal consistency (Taber, 2018). The minimum sample size required for two-tailed Cronbach’s alpha test was calculated using the Bonet formula based on assumptions; item in the scale (k) 10, power 0.90 (1 - β), type I error (α) 0.05, value of Cronbach’s alpha at null hypothesis (CA0) and expected value of Cronbach’s alpha (CA1) were identified at 0.0 and 0.8, respectively (Bonett, 2002). The required sample size was 70. NDI-Hi was re-administered on about 50% of the participants (n = 110) at the interval of 48 hours to assess its test-retest reliability. With a minimum of 2 days between the completions of NDI-Hi, sufficient time should have elapsed to minimize the bias associated with recall of previous responses (Marx et al., 2003). This is more than the power calculated sample.

To determine the dimensionality of the NDI-Hi, a stepwise exploratory factor analysis (EFA) was performed with Kaiser Meyer Olkin and Bartlett’s criteria with a retention rule of an Eigenvalue > 1 (items with loadings ≥ 0.4 were considered satisfactory). Independent factors were obtained with maximum likelihood using the Varimax rotation model. The floor and ceiling effects, distribution and completeness of items responses and missing values were examined for content validity. The concurrent validity method was used for the criterion-related validity; Pearson’s correlation analysis was done to establish relation between NDI-Hi and Visual Analogue Scale (VAS). The convergent construct validity was examined by comparing the NDI-Hi and Hindi version of NPAD using Spearman rank correlation coefficient (r). The r value of > 0.80 were considered excellent, and 0.61 to 0.80 very good. It is expected that the Hindi versions of NDI and NPAD would have moderate correlations since both tools measures the impact of neck pain on the functional status of the patients (Agarwal et al., 2006). Standard error of measurement (SEM) was calculated from the square root (SEM = SD √ (1-R)) of the within-subject variance from ANOVA for random-effect to evaluate the measurement error. MDC_95% was derived using the SEM to allow expression of the smallest magnitude of change that reveals the true change rather than measurement error by using the formula MDC_95% = 1.96 √2x SEM or 2.7*SEM. The Bland-Altman plot was also used to determine agreement (Bland & Giavarina, 2015) (Figure 2).

Figure 2. Bland-Altman plot of agreement between test and re-test scores of NDI-Hi.

Bold blue line representing the mean of difference, the red lines representing 95% LOA and the grey lines representing the 95% CI of the mean of the difference. NDI-Hi, Hindi-version of the Neck disability index; LOA, limits of agreement.

Statistical analysis

IBM SPSS Statistics (RRID: SCR_016479) version 20 was used to analyze the data (SPSS 20, IBM, Armonk, NY, USA). Characteristics and observations of the quantitative and qualitative variables were summed using the mean, standard deviation (SD) and frequencies or relative frequencies. Normality of scores from different instruments was examined using one-sample Kolmogorov-Smirnov test. Internal consistency, test-retest reliability, and convergent validity of the two scales used in the analysis were examined. The raw score of the NDI was transformed to 100 prior to psychometric testing of the tool. Authors in the present study were able to determine the internal consistency of the NDI-Hindi using Cronbach’s alpha index with values between 0.70 and 0.90 considered satisfactory. For test-retest reliability, interclass correlation coefficients (ICCs) and 95 percent confidence intervals (CIs) were determined. ICCs below 0.40 were considered low, those in the range of 0.4-0.70 were considered moderate, 0.70-0.90 was considered significant and values above 0.9 were considered exceptional respectively. The degree of correlation between NDI-Hindi and the NPAD at baseline was determined using Spearman’s rho correlation. The significance level was set at P < 0.05. Using the scree test plot, retention rule (item value) of Eigenvalue > 1, and Kaiser Meyer Olkin (KMO) and Bartlett’s test of sphericity, the dimensionality of the NDI-Hi was ascertained. The Maximum Likelihood with a Varimax rotation model was used for the EFA of the Hindi version of NDI, and the 0.4-factor loading principle was used for extraction (Zhang, 2015).

Participants

Demographic variables of the participants have been tabulated in Table 1 (Sidiq et al., 2023). All participants were assessed for their body mass index, duration of neck pain, pain medication, associated medical condition, marital status, employment status, level of education, income and mean and SD scores of NDI and NPAD. The majority of participants were female (n = 149, 70.6%), married (n = 152, 72.0%) and unemployed (n = 120, 56.9%). Near about 80% of the participants reported living with neck pain for 6 months and more.

Participant’s response to NDI-Hi

Table 2 displays the mean scores extracted from the participants in 10 sections of the NDI-Hindi. Participants were moderately affected due to neck disorder and the mean rating on NDI was 32.02 (6.9) out of 50 in total. Principle component analysis revealed two-factor structures (physical dysfunction and neurological dysfunction) of NDI-Hindi. The physical dysfunction included pain intensity personal care, lifting, work, driving, reading and recreation were moderately to severely affected with a mean score ranging from 3.36-3.92, while neurological dysfunction that included headaches, concentration and sleeping were mildly affected with mean scores 2.3 and standard deviation of 0.6.

Internal consistency of the NDI-Hi

The Cronbach’s alpha reliability was 0.962, which indicated that respondents were able to read and comprehend the 10 NDI-Hindi items well. However, all 10 indicators demonstrated a meaningful adjusted total association with their total scores. When these indicators were removed from the total, all of the items exhibited a corrected item-total correlation greater than 0.75, showing reasonably substantial shared covariance (Table 3).

Test-retest reliability of NDI Hi

Test-retest reliability was assessed seven days after the initial assessment and was found to be excellent. The test-retest reliability was excellent for the overall NDI-Hindi and average Interclass correlation coefficient (ICC) was 0.893, with a 95% Confidence interval (CI) (0.795, 0.937) (p < 0.001) (Table 2). The mean of item variances was 0.648, with inter-item covariance’s mean of 0.465 (Table 3).

Construct validity of NDI-Hi

Construct validity of the NDI-Hindi was estimated using principal component analysis (factor analysis) with varimax rotation (Table 4). Data met the assumptions (Kaiser–Meyer Olkin = 0.826) of sample adequacy and Bartlett’s test of sphericity. The total variance explained by two factors was 87.5%. The analysis found a two factor structure of scale; Factor 1 included the first seven items of the questionnaire (77.3% of variance), Factor 2 included three items (10.2% of variance) (Table 5). Factor loading was determined using a scree plot (Figure 3).

Figure 3. Scree plot indicating factor loading for NDI-Hi.

NDI-Hi, Hindi-version of the Neck disability index.

Convergent validity of NDI-Hi

The scores of NDI-Hindi showed excellent correlation with the total scores on Hindi-validated version of NPAD scale. Scores on neck pain and disability scale showed a significant positive correlation with all 10 items of NDI, with the highest correlation with lifting item (0.917, p = 001) and a least but significant correlation with the concentration item and sleeping (0.845, p = 0.001) (Table 6).

Study limitations

This study recruited samples beyond the recommended guidelines for cross-cultural adaptations of outcome measures and the required power calculated sample for test-retest reliability, which improved the statistical power of the psychometric evaluation of the NDI-HI version. However, there are few limitations to mention; this study is a cross-sectional design and hence any correlations should not be interpreted as causal effects. This study used only a self-reported questionnaire, the relationship between the activity limitations of neck pain (NDI) and physical tests were not considered and the study was done among rural northern Indian population only.