Cross-Cultural Adaptation and Validation of the Hindi Version of the Fukushima Lumbar Spinal Stenosis Scale-25 in Indian Patients with Lumbar Spinal Stenosis: A Cross-Sectional Psychometric Study

Ethics statement

This study was approved by the Institutional Ethics Committee (Ethics Code: DRB/Ortho/2025/68). Approval was obtained before enrollment began, and every participant provided written informed consent prior to taking part. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Study design and setting

We carried out this cross-sectional psychometric validation study in the Department of Orthopaedics at a large tertiary referral institute in North India offering comprehensive spinal services between June 2025 and April 2026. This institute is among the largest tertiary referral institutions in the country and serves a catchment of approximately nine to ten Indian states, so that its patient population encompasses a wide spectrum of clinical presentations and of socioeconomic, educational, and regional backgrounds.

Participants

We enrolled 160 consecutive Hindi-speaking patients with a diagnosis of symptomatic LSS. To qualify, patients had to be older than 40 years; report neurogenic claudication and/or radiating lower-limb pain or numbness for at least 3 months; have LSS confirmed radiologically on lumbar spine magnetic resonance imaging and graded using the Schizas morphological classification ^[12]; and be able to read and understand Hindi well enough to complete a self-administered questionnaire on their own. This requirement reflects the design of the FLS-25 as a self-administered instrument so that its properties were examined in the format for which it was developed. Enrollment of consecutive patients, without selection on the basis of severity, comorbidity, or pattern of disease, was intended to limit selection bias, and patients with mild-to-moderate as well as severe stenosis, with central as well as foraminal disease, and with or without spondylolisthesis were included (Table 1). We excluded anyone with prior lumbar spine surgery, coexisting myelopathy or peripheral neuropathy, walking limitation driven mainly by hip or knee osteoarthritis, uncontrolled psychiatric or cognitive impairment that would compromise reliable self-report, or unwillingness to consent. Fifty-four patients in the cohort were evaluated while still awaiting surgery (the pre-operative subgroup), and 106 had already had surgery at least 3 months earlier (the post-operative subgroup); a further 48 patients, drawn from across both groups, completed assessments both before surgery and 3 months afterward and formed the subset used for the responsiveness analysis.

Translation and cross-cultural adaptation

We followed the cross-cultural adaptation framework set out by Beaton et al. ^[13] to translate the English-language FLS-25 into Hindi, the same general approach used in earlier adaptations of this scale. Two bilingual, native Hindi-speaking translators worked independently and without knowledge of the questionnaire’s clinical purpose to produce separate forward translations, which a panel that included an orthopedic spine surgeon then merged into one version. A translator with no prior exposure to the original English instrument back-translated this harmonized version into English. An expert committee made up of orthopedic surgeons and linguists compared the original, the forward translations, and the back translation, resolving disagreements over meaning, idiom, and cultural fit to arrive at a pre-final Hindi draft. Fifteen Hindi-speaking patients with LSS, drawn from outside the main study sample, then piloted this draft so we could gauge how comprehensible, clear, and culturally acceptable it was; interviewers noted any item patients found confusing along with their suggested rewording. Because only minor wording changes emerged from this step, the expert committee reached consensus on the final H-FLS-25 without further revision.

Instruments

The H-FLS-25 comprises 25 items scored on a five-point Likert scale running from 0 (strongly disagree) to 4 (strongly agree); summed scores range from 0 to 100, with higher totals reflecting more severe symptoms and greater functional impairment. Alongside the H-FLS-25, participants filled out the already-validated Hindi ODI ^[14], a 10-domain measure of low back pain-related disability expressed as a percentage (0–100%), and a 10-cm VAS anchored at 0 (no pain) and 10 (worst imaginable pain) for both back and leg pain. The Hindi ODI was chosen as the principal external comparator because it is a validated Hindi-language measure of back-related disability and, to our knowledge, no validated Hindi disease-specific instrument for LSS existed against which the H-FLS-25 could be benchmarked.

Procedure

Baseline data collection took place at the outpatient clinic, where participants completed the H-FLS-25, Hindi ODI, and VAS. We re-administered the H-FLS-25 by telephone to a subset of patients roughly 2 weeks later to gauge test-retest reliability; a 2-week gap is standard in test-retest work on functional status questionnaires because it is long enough for patients to forget their earlier answers but short enough that their clinical status is unlikely to have genuinely changed ^[15]. Because baseline completion was self-administered whereas the retest was conducted by telephone, the retest also examined, incidentally, the stability of scores across two modes of administration. The 48 patients in the matched subgroup used for the responsiveness analysis were assessed once before surgery and again 3 months afterward; the treating surgeon chose decompression alone or decompression with instrumented fusion based on whether instability or spondylolisthesis was present.

Statistical analysis

We summarized demographic and clinical characteristics, including operative details such as surgery type and number of levels treated in the post-operative subgroup, using means, standard deviations (SD), and frequencies as appropriate. Cronbach’s alpha served as our measure of internal consistency for the total scale and each domain, treating values above 0.70 as acceptable, whereas test-retest reliability was captured through the intraclass correlation coefficient (ICC; two-way mixed-effects model, absolute agreement) and item-level agreement through weighted kappa statistics. To evaluate construct (convergent) validity, we computed Pearson correlations between H-FLS-25 total and domain scores on one hand and ODI and VAS scores on the other, treating coefficients of 0.40 or above as acceptable and taking higher values as evidence of stronger convergence. For known-groups validity, we compared H-FLS-25 scores between patients with mild-to-moderate stenosis (Schizas grade A/B) and those with severe stenosis (Schizas grade C/D) using independent-samples t-tests. We tabulated the mean, SD, and range for each of the 25 items along with floor and ceiling effects, defined as the proportion of participants scoring at the lowest or highest possible value for that item, and flagged any floor or ceiling effect exceeding 15% as notable. We explored the scale’s underlying factor structure through exploratory factor analysis with principal component extraction and varimax rotation, checking sampling adequacy with the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity, retaining factors with eigenvalues above 1, and treating item loadings of 0.40 or higher as meaningful. Within the matched pre-/post-operative subgroup, we quantified responsiveness to surgery using effect sizes, calculated as the mean score change divided by the pre-operative SD, with an effect size of 0.8 or above considered large. Measurement error was quantified using the standard error of measurement (SEM = SD × √[1−ICC]) and the smallest detectable change at the 95% confidence level (SDC95 = 1.96 × √2 × SEM) ^[16]. In the absence of an external anchor, a distribution-based minimal important change (MIC) was estimated as one half of the baseline SD ^[17] and examined against the magnitude of one SEM ^[18]. The SD used for these calculations was the pre-operative value in the matched surgical subgroup (Table 2), combined with the ICCs reported in Table 3. We took a two-sided P-value under 0.05 as statistically significant and carried out all analyses in the Statistical Package for the Social Sciences (IBM Corp., Armonk, NY, USA).

Sample size considerations

The adequacy of the sample was judged against accepted recommendations for psychometric research. For exploratory factor analysis, a minimum of five respondents per item and not fewer than 100 participants is widely advised ^[19], which corresponds to 125 participants for a 25-item instrument; the present sample of 160 yields 6.4 respondents per item. Samples of this order are, moreover, adequate when communalities are moderate to high, and the factors are well determined ^[20], a condition supported here by a KMO value above 0.80 and item loadings of 0.42 to 0.79. For test-retest reliability, at least 50 participants are recommended ^[21]; 142 patients completed the retest, nearly three times this threshold. The sample also exceeds those of the Iranian (n = 131) ^[9] and Korean (n = 129) ^[10] adaptations of the same instrument, and the responsiveness subgroup of 48 patients exceeds the 27 patients on whom responsiveness was assessed in the Korean adaptation ^[10]. We therefore regard the sample as more than adequate for the principal objectives of this study.

Reliability

Internal consistency was high both for the total score and for each of the five domains that emerged from the factor analysis, with Cronbach’s alpha exceeding 0.80 for most domains and reaching 0.89 overall (Table 3). Test-retest reliability was likewise strong: the ICC for the total score topped 0.90, and domain-level ICCs fell in the good-to-excellent range. Weighted kappa values for individual items were mostly above 0.60, corresponding to substantial-to-almost-perfect agreement and indicating that responses to all 25 items held up well across the 2-week retest interval.

Construct and known-groups validity

The H-FLS-25 total score correlated strongly with both the ODI (r = 0.77) and VAS leg pain (r = 0.79), and moderately strongly with VAS back pain (r = 0.71), all significant at P < 0.001, supporting construct validity (Table 4). Domain-level correlations mirrored this pattern overall: the activity limitation and physical pain domains aligned most closely with disability and leg pain scores, while the mental/psychological health domain showed weaker, though still significant, associations. Patients with severe stenosis (Schizas grade C/D) scored significantly higher on the H-FLS-25 total than those with mild-to-moderate stenosis (Schizas grade A/B) (P < 0.001), confirming that the instrument can distinguish between these clinically distinct severity groups. The weaker association of the mental/psychological health domain with the ODI and VAS scores (r = 0.52–0.60) than that of the total score (r = 0.71 to 0.79) (Table 4) further indicates that the instrument is not merely a surrogate for pain or generic disability and offers preliminary indirect support for divergent construct behavior.

Item-level score distribution

Table 5 reports the mean, SD, range, and floor/ceiling effects for all 25 items. Because no patient scored at either the minimum (0) or maximum (100) possible total, the H-FLS-25 covered a wide range of symptom severity without bunching at either end. Item 7 (“I can’t lie on my back”) showed a floor effect, suggesting it discriminates poorly among patients with mild or no difficulty in that position. Items 19 (“I often have no feeling in my legs”) and 20 (“If I walk for a while, my symptoms worsen and I can’t walk anymore”) both showed ceiling effects, implying these items may be less able to register further gains once walking tolerance and sensory symptoms have largely recovered.

Factor structure

A KMO value above 0.80 together with a significant Bartlett’s test of sphericity (P < 0.001) confirmed that the data were suitable for factor analysis. Principal component analysis with varimax rotation yielded a five-factor solution, each factor with an eigenvalue above 1 (Fig. 1), which we labeled physical pain, activity limitation, gait instability, physical activity, and mental/psychological health. These five factors together explained roughly 66% of the total variance, and every item loaded at least 0.40 on its assigned factor, with loadings spanning about 0.42–0.79 (Table 6), which supports the structural validity of the H-FLS-25 in this population. As exploratory analysis is the accepted first step when an instrument is introduced into a new linguistic and cultural setting, as in the Korean adaptation ^[10], this structure should be regarded as provisional pending confirmatory factor analysis in an independent sample.

Responsiveness to surgical treatment

All 48 patients who completed both pre-operative and 3-month post-operative assessments showed significant gains across every H-FLS-25 domain as well as on the ODI and VAS (Table 2 and Fig. 2). The H-FLS-25 total score produced the largest effect size (2.65), ahead of the activity limitation (1.60) and physical pain (1.55) domains, with the physical activity domain trailing somewhat (1.03). Effect sizes for all measures are compared in Fig. 3. VAS leg pain showed a larger effect size (1.63) than either VAS back pain (1.10) or the ODI (1.05), hinting that the H-FLS-25 may be especially attuned to the leg-dominant symptom relief that typically follows surgical decompression.

Measurement error and interpretability

Table 7 presents the measurement error indices. The SEM of the H-FLS-25 total score was 3.7 points on the 0–100 scale, corresponding to a smallest detectable change of 10.3 points at the 95% confidence level. The mean improvement of 34.7 points in the surgical subgroup was more than 3 times this threshold. The distribution-based MIC, estimated at 6.6 points for the total score (one half of the baseline SD), lay well below the observed mean change, which was more than 5 times this value. Domain-level SEM values ranged from 0.8 to 1.7 points, and the observed mean changes exceeded the corresponding SDC95 for the physical pain, activity limitation, gait instability, and mental/psychological health domains, whereas that for the physical activity domain (3.3 points) approached but did not exceed its SDC95 (3.7 points), indicating that this domain is best interpreted at the group level. These MIC values should be regarded as provisional, because an anchor-based estimate was not available.

Conflict of Interest:

Nil

Source of Support:

Nil

Consent:

The authors confirm that informed consent was obtained from the patient for publication of this article

How to Cite this Article

Raj S, Kumar V, Teja AT, Uppal R, Basil F, Goni VG. Cross-Cultural Adaptation and Validation of the Hindi Version of the Fukushima Lumbar Spinal Stenosis Scale-25 in Indian Patients with Lumbar Spinal Stenosis: A Cross-Sectional Psychometric Study. J Orthop Case Rep. 2026 November;16(11):504-513. doi: 10.13107/jocr.2026.v16.i11.8538.