Translation, Validation, and Diagnostic Performance of the Michigan Neuropathy Screening Instrument (MNSI) in Bangla

ABSTRACT

Backgrounds and Aims

Diabetic peripheral neuropathy (DPN) can result in foot ulcers and ultimately limb amputation. The Michigan Neuropathy Screening Instrument (MNSI) is a recommended screening tool for DPN. However, it must be appropriately translated and validated before use in new linguistic and cultural contexts. The study aimed to translate the MNSI into Bangla and assess its validity and applicability among Bangladeshi Diabetic patients.

Methods

A cross‐sectional study was conducted to develop Bangla version of the MNSI questionnaire (MNSIq‐Bl) following ISPOR Task Force recommendations. The translation process included reconciliation, expert committee review, and pilot testing among 20 purposively selected patients with diagnosed diabetes mellitus. Validation was performed among purposively selected 50 diabetic patients attending outpatient Department of Endocrinology, Dhaka Medical College, adhering to the inclusion and exclusion criteria between July 2018 and June 2019. Cut‐off values for the MNSIq‐Bl and MNSIe (questionnaire and physical examination) scores were recommended compared with the result of the Nerve Conduction Study (NCS). Ethical clearance (MIU‐DMC/ECC/2019/42, dated February 6, 2019) from the due authority and informed written consent were obtained from the respondents before data collection.

Results

Strong inter‐interviewer reliability was observed for both MNSIq‐Bl (r = 0.977) and MNSIe (r = 0.923), with significance at p < 0.001. Cronbach's α for MNSIq‐Bl (excluding Items 4 and 10) ranged from 0.7 to 0.8, indicating acceptable internal consistency. The area under the curve (AUC) for MNSIq‐Bl was 0.850 at a cut‐off score of ≥ 2.5 with 68% sensitivity and 75% specificity. For MNSIe, the AUC was 0.918 at a cut‐off score of ≥ 0.5, with 90% sensitivity and 75% specificity.

Conclusion

The Bangla version of MNSI (MNSI‐Bl) is a valid and reliable tool for screening diabetic peripheral neuropathy among Bangla‐speaking patients, demonstrating good diagnostic performance.

Summary

• To screen for diabetic peripheral neuropathy, the Michigan Neuropathy Screening Instrument (MNSI) is widely used; however, a validated Bangla version of the tool is not available.

• This study successfully translated and validated the MNSI into Bangla, demonstrating strong reliability and validity.

• The Bangla MNSI (MNSI‐Bl) can be a practical tool for screening peripheral neuropathy among Bangla‐speaking diabetic populations, aiding timely clinical intervention.

1. Introduction

Globally, diabetes mellitus (DM) is a rapidly expanding health concern, affecting millions and leading to significant morbidity and mortality [1]. The International Diabetes Federation (IDF) has estimated that by 2045, the global burden of diabetes patients will rise to 783 million, especially in low‐ and middle‐income countries [2]. One of the most common and debilitating complications of diabetes is diabetic peripheral neuropathy (DPN), which affects nearly 50% of diabetic patients over their lifetime [3]. DPN can lead to severe consequences, including loss of protective sensation, chronic infections, foot ulcers, and eventual limb amputation, significantly impacting the quality of life and increasing healthcare costs [4].

Bangladesh has witnessed a sharp rise in diabetes prevalence, increasing from 2.29% in 1995 to 12.01% in 2019, with an estimated 13 million adults currently living with diabetes [5]. Given this high disease burden, early detection and management of DPN are crucial in preventing severe complications. The reported prevalence of DPN among diabetic patients in Bangladesh ranged from 19.7% to 35% [6]. Moreover, nerve conduction study (NCS) is considered the gold standard for DPN diagnosis, but its high cost and limited availability make it impractical for routine use in resource‐constrained settings [7]. The American Diabetes Association (ADA) advocates annual screening for DPN in all diabetic patients using validated tools [8]. In case of suspected patients with DPN, using a screening tool before suggesting NCS can decrease the out‐of‐pocket (OOP) expenditure of the diabetic patients, which is currently 13.03% of their total OOP [9].

However, clinical diagnosis remains challenging due to the subjective nature of symptoms and variability in presentation. The Michigan Neuropathy Screening Instrument (MNSI) is a widely used tool for DPN assessment due to its simplicity, cost‐effectiveness, and noninvasive nature [10]. It consists of two parts: a self‐administered questionnaire assessing neuropathic symptoms and a clinician‐administered physical examination evaluating foot appearance, vibration sensation, ankle reflexes, and monofilament testing [11].

Despite its effectiveness, the MNSI has primarily been available in English, limiting its applicability in non‐English‐speaking populations. Cross‐cultural adaptation and validation of such screening tools are necessary to ensure their reliability and validity across diverse linguistic and cultural contexts [12].

MNSI has been translated and validated into multiple languages. However, a Bangla version has not been added to this list despite the high prevalence of diabetes in Bangladesh. Validation of the Bangla version of the MNSI screening instrument can facilitate early detection, timely referral, appropriate intervention, and improved clinical outcomes for diabetic patients in Bangladesh. Thus, this study aims to translate and validate the Bangla version (MNSI‐Bl) of the MNSI scale, and to recommend a cut‐off value for screening DPN for general practitioners to be applied in low‐resource settings.

2. Methodology

This was a cross‐sectional study conducted at the department of physiology of Dhaka Medical College (DMC) between July 2018 and June 2019. In this study, firstly, the MNSI instrument was translated and validated through piloting among 20 diagnosed cases of diabetes mellitus. Secondly, for screening, both the MNSI questionnaire and the MNSI physical examination score, cut‐off value was recommended based on study among 50 diabetic patients aged 18–60 years. In both sections of the study (translation validation process and recommendation for cut‐off value), respondents were purposively included. Registered cases of DM in the outpatient Department of Endocrinology, DMC, were selected for the study. Respondents with chronic diseases such as chronic renal failure, hypothyroidism, arthritis, cancer, hepatic dysfunction, Guillain–Barré syndrome, cerebral vascular disease, and hereditary sensory motor neuropathy were excluded. In addition, chronic alcohol abusers and respondents with psychological problems were excluded. For translation validation ISPOR task force recommendation regarding principles of good practice was considered. For translation and validation we adopted the recommended stages (Figure 1) to ensure the validity of the process consisting of (1) forword translation, (2) reconciliation, (3) back translation, (4) reconciliation and review of back translation, (5) expert committee review and finally, (6) piloting, and (7) debriefing and finalization (to be used among larger population to evaluate the applicability of the scale by the general practitioner in resource poor settings as an appropriate facility based assessment and referral tool as well) [12, 13]. To measure the validity and reliability, the investigator's and a neurologist's MNSI‐Bl scores were compared. On the other hand, receiver operating characteristic (ROC) curve analysis was applied to find out the cut‐off value of MNSIq‐Bl and MNSIe score for screening DPN. After validating MNSIq‐Bl and MNSIe, to find out the cut‐off value of the tool, we used the diagnostic accuracy sample size calculator, estimating sensitivity 75% and specificity 75%, prevalence of DPN among diabetic patients 34%, with a margin of error 0.2 and 95% confidence interval. Considering the expected dropout of 5%, the final sample size was calculated to be 56. Finally, we completed our study among 50 purposively selected diabetic patients.

Figure 1.

Flowchart for translation and validation process of MNSI.

2.1. Concept for Translation and Face Validity

To fulfill the objective of the validation of the MNSI scale, we approached to original author for the use of the questionnaire (https://eprovide.mapi-trust.org/my-eprovide/myrequests/update/e3fb010ebe6f197ed1212, request ID no: 213037, dated July 4, 2019). In this study, the MNSI scale was intended to be used among a new population (different from the population where the scale was designed and prescribed) in Bangladesh. As the MNSI questionnaire is a self‐administered questionnaire, the language and cultural differences by nation are considered while using the MNSI scale. But a Bangla version of the MNSI scale is still unavailable. So, before applying the scale in any original study, researchers conducted validation process of MNSI instruments in accordance with Beaton et al.'s recommendations for adaptation of self‐report measures [12].

3. The Procedure of Translation and Validation of the Questionnaire Part

3.1. Stage 1: Forward Translation

MNSI questionnaire was translated into two Bangla formats (B1 and B2) by one physician (Translator 1) and one nonphysician (Translator 2) education expert, while both were native Bangla speakers and had fluency in English. Translator 1 could understand the concept of the scale and its context of application in Bangla. Additionally, he was expected to provide more reliable wording of the text, considering the link with the clinical perspective. On the other hand, Translator 2 was neither informed nor aware of the concept of the scale or its application. Moreover, as an education specialist, it was expected that she would have conceptualized differently than Translator 1, as she is a nonphysician, and her language would be more likely to represent the general people.

3.2. Stage 2: Reconciliation

Both the Bangla formats (B1 and B2) were checked by an interpreter (Interpreter 1) who compiled the two Bangla formats into one pre‐final Bangla format through a reconciliation process. Interpreter 1 had efficiency in the Bangla language, and her fluency in English was also considered. She was expected to understand the perspective of both translators. Again, the synthesis was done in the presence of both translators, where a written report was compiled about the agreement, disagreement and the process of how the issues were resolved. During the whole process in all the issues of disagreement, none of the experts were allowed to compromise his or her inner feelings; rather, the utmost effort was to resolve the issue according to the consensus of all of them.

3.3. Stage 3: Back Translation

Then, two English experts independently translated this Bangla format (B3) back into two English formats individually (E1 and E2). Both the back translators were fluent in English, while the back Translator 1 (for E1) was an English version teacher with 20 years of experience. On the other hand, the back Translator 2 (for E2) was a specialist physician. Before they do the back translation, their verbal assurance of no previous knowledge about the questionnaire was confirmed. None of them were allowed to see the original English version before or during their back translation process.

3.4. Stage 4: Reconciliation of Back Translation

Another English expert checked and synthesized these two English formats into a back‐translated one compiled English format (E3). The person who reviewed the English translated versions (E1 and E2) and reconciled was an English native speaker. He checked both versions in the presence of both the back translators to reconcile the variation between the back Translators 1 and 2. The final version of the back translated copy was created according to the consensus of all three members, where the discussion continued for a common consensus rather than any compromise of one's own view.

3.5. Stage 5: Expert Committee's Review (Back Translation Review and Harmonization by Expert Committee)

An expert committee was formed, consisting of a methodologist, a neurologist, and two translators, one who translated Bangla B3 and another one who back‐translated the English version E3. The committee compared the back‐translated English format (E3) with the original English questionnaire. They discussed the discrepancies between the original version of the MNSI scale and the back‐translated, reconciled version (E3). Finally, recommended the pre‐final Bangla version (B3) for the next stage of pilot testing.

3.6. Stage 6: Pilot Testing

The pre‐final Bangla (B3) version was applied to 20 individuals who were not included in the study subjects. The instrument was applied by one neurologist and the researcher herself to find out the internal consistency of the MNSI Bangla version scoring (questionnaire part and examination part). After evaluation, both the results were compared statistically to identify the inter‐interviewer differences in results.

3.7. Final Approval (Proofreading)

After the completion of pretesting, debriefing was done. Based on the agreement of the expert committee, the pre‐final Bangla version was accepted as a final version of the valid Bangla questionnaire (B3).

3.8. Scoring of MNSI

This screening tool is divided into two sections: the first section is a self‐evaluation process where a self‐administered questionnaire with 15 items is organized, and the second section comprises a set of five physical examinations that must be performed by medical professionals in a medical setting.

3.9. The Scoring of MNSI Questionnaire Part

For section one, there are 15 items with dichotomous “Yes” and “NO” answer options for the respondents. This questionnaire attempts to gather information about patients' positive and negative sensory symptoms like pain, temperature sensation, tingling and numbness according to their perception. Among the 15 items, 11 items (Items 1–3, 5, 6, 8, 9, 12, 14, 15) are scored positively (“Yes” response counts as 1 point), and two items (Items 7 and 13) are scored negatively (“No” response counts as 1 point). The original scale excluded the rest of two items (Items 4 and 10) from the calculation of final scoring, where the former measures circulation and the latter measures general status. So, the score of a respondent can vary from 0 to 13.

3.10. The Scoring of MNSI Examination Part

There are five things measured by the physical examination section of the MNSI scale, including the appearance of the feet, ulceration, ankle reflex, and touch sensation of monofilament at the great toe.

3.11. Foot Inspection

On inspection, if any abnormality like dry skin, callous formation, fissures, frank ulceration or any other deformity was observed, it was scored 1 for each side.

3.12. Presence or Absence of Ulcer

Each ulcerated foot was scored as 1.

3.13. Ankle Reflex

A percussion hammer was used to examine the ankle reflex in the relaxed sitting position with the foot dependent and dorsiflexed slightly; the Achilles tendon was percussed directly. If the reflex is obtained, it is graded as present. If the reflex is absent, the patient is asked to perform the Jendrassik maneuver. Reflexes elicited with the Jendrassic maneuver are designated “present with reinforcement.” If the reflex is not elicited even with the Jendrassic maneuver, the reflex is considered absent. Scoring was based on Present—0, Present with reinforcement—0.5, and Absent—1.

3.14. Vibration Sensation

The test was performed on each side with the great toe unsupported. Vibration sensation was done bilaterally using a 128 Hz tuning fork, which was placed over the dorsum of the great toe on the bony prominence of the distal interphalangeal (DIP) joint. Patients kept their eyes closed and responded when they no longer sensed the vibration. If the examiner felt vibration for < 10 s longer, the respondent's vibration sensation in the great toe was mentioned as “present” with a score of 0. But if the examiner felt vibration for ≥ 10 s longer, it was recorded as “reduced” and was scored as 0.5. Finally, if no vibration was detected by the patient, it was scored as 1.

3.15. Monofilament Testing

The filament was to be initially prestressed (4–6 perpendicular applications to the dorsum of the examiner's first finger). The filament was applied to the dorsum of the great toe midway between the nail fold and the DIP joint. The filament was applied perpendicularly and briefly (< 1 s) with even pressure. Respondents kept their eyes closed and were asked to respond when they felt the filament. The response was considered normal and scored as 0 if the respondent could have eight correct responses out of 10. In case of seven correct responses, he or she was identified as having reduced response and was scored as 0.5. Lastly, if the respondents failed to have even a single correct answer, they were scored as 1.

3.16. Interpretation of Score

In both the questionnaire and examination sections, a higher score indicates more neuropathic symptoms.

3.17. NCS

Neurological examinations included measurement of nerve conduction velocity of the posterior tibial, deep peroneal, and sural nerves. Nerve conduction velocity was measured by a standard NCV‐EMG machine (NIHON KOHDEN), model: MEB9400, software: NeuropackS1 in a room temperature. Nerve conduction parameter followed the recommendations by the Antonio Conference on Diabetic neuropathy. All measurements were conducted with standard surface electrodes and were recorded. The active recording electrode G1 was placed on the center of the muscle belly, and the reference electrode G2 was placed distally, over the tendon to the muscle (Table 1). A qualified neurologist obtained the results for NCS. The test results were expressed in either “neuropathy” or “no neuropathy” based on any abnormality in any evaluated parameter (latency, amplitude, conduction velocity) in individual nerves during the test.

Table 1.

Description of sensory nerve and motor nerve conduction study.

Name of the nerve	Recording site	Placing of G1	Placing of G2	Stimulation site	Distal distance
Sural sensory nerve conduction	Posterior ankle	Posterior to the lateral malleolus	3–4 cm distally	Lateral calf	14 cm
Tibial nerve conduction	Abductor Hallucis Brevis (AHB) muscle	1 cm proximal and 1 cm inferior to the navicular prominence	Over the metatarsal‐phalangeal joint of the great toe	Medial ankle: slightly proximal and posterior to the medial malleolus Popliteal fossa: mid‐posterior knee over the popliteal pulse	9 cm
Common peroneal nerve conduction	Extensor Digitorum Brevis (EDB)	Over the muscle belly of the dorsal lateral foot	Distally over the metatarsal‐phalangeal joint of the little toe	Ankle: anterior ankle, slightly lateral to the tibialis anterior tendon Below the fibular head: lateral calf, one to two fingerbreadths inferior to the fibular head Lateral popliteal fossa: lateral knee, adjacent to external hamstring tendons.	9 cm

3.18. Statistical Analysis

All the parameters were expressed as mean ± SD (standard deviation). Pearson's correlation coefficient (r) analysis was done to evaluate concurrent validity between the MNSIq‐Bl and MNSIe scores, conducted by the neurologist and the investigator. The internal consistency (reliability) of the MNSIq‐Bl was evaluated through inter‐rater reliability, calculating the intraclass correlation coefficient (ICC) at a 95% confidence interval and Cronbach's α. The cut‐off value for Cronbach's α was determined based on [14] recommendation including excellent (0.90 and above), good (0.90> α ≥ 0.80), acceptable (0.80> α ≥ 0.70), questionable (0.70> α ≥ 0.60), poor (0.60> α ≥ 0.50) and unacceptable reliability (0.5 > α) [14]. The MNSI‐Bl scale's (questionnaire and physical examination) diagnostic capacity was compared with the result of NCS based on sensitivity, specificity, positive predictive value, and negative predictive value analysis. ROC curve analysis was applied to find out a cut‐off value of the MNSI score for screening DPN. Analysis was performed by using SPSS (Statistical Package for Social Science) version 21.

4. Result

The MNSI‐Bl was initially pilot tested on 20 respondents to assess its preliminary validity. Diagnostic performance was evaluated in an additional cohort of 50 respondents.

4.1. Section A: MNSI instrument Translation Validation for MNSI‐Bl (N = 20)

4.1.1. Distribution of Score

Distribution of both MNSIq‐Bl score and MNSIe score followed the Gaussian distribution irrespective of the measurement done by either neurologist or investigator (Table 2).

Table 2.

Distribution of statistics of the MNSIq‐Bl score and MNSIe score measured by the neurologist and the investigator (n = 20).

Distribution of statistics	MNSIq score		MNSIe score
	Neurologist's score	Investigator's score	Neurologist's score	Investigator's score
Mean	2.15	2.25	0.63	0.65
Median	2.0	2.0	0.25	0.25
Variance	1.92	2.09	0.602	0.58
Std. deviation	1.38	1.44	0.77	0.76
Minimum	0.00	0.00	0.00	0.00
Maximum	5.00	5.00	2.50	2.00
Range	5.00	5.00	2.50	2.00
Interquartile range	1.75	1.75	1.00	1.00
Skewness	0.361	0.210	1.03	0.72
Kurtosis	0.407	−0.120	0.19	−0.90

4.1.2. Reliability Assessment for MNSIq‐Bl and MNSIe Score

To evaluate the inter‐interviewer variability, Pearson's correlation was used. The scores between the investigator and the neurologist show a strongly positive correlation coefficient in both questionnaires (MNSIq‐Bl; r = 0.945, N = 20, p < 0.00) and examination section (MNSIe r = 0.923, N = 20, p < 0.001) (Figures 2 and 3). Intraclass correlation coefficient for MNSIq‐Bl was found to be 0.973 with 95% confidence interval (CI) (range: 0.93–0.89), and for MNSIe it was 0.923 with 95% CI (range: 0.90–0.98).

Figure 2.

Relationship of MNSIq‐Bl score measured by neurologist and investigator.

Figure 3.

Relationship of MNSIe score measured by neurologist and investigator.

Internal consistency of the MNSIq‐Bl was measured based on item‐to‐item correlations. The item‐to‐total score correlations and Cronbach's α were measured for 13 items of the main questionnaire, excluding Items 4 and 10. The Cronbach's α (of the instrument) was within an acceptable range (minimum 0.703 to maximum 0.762; therefore, the score was within ≥ 0.8 to ≥ 0.7) (Table 3).

Table 3.

Subscale Item to total correlations and Cronbach's α (n = 20).

Question item	Scale mean if item deleted	Scale variance if item deleted	Corrected item‐total correlation	Cronbach's α if item deleted	Satisfaction
Item 1	2.6667	6.023	0.482	0.722	Acceptable
Item 2	2.5333	5.637	0.607	0.704	Acceptable
Item 3	2.8	6.786	0.21	0.752	Acceptable
Item 5	2.7667	6.461	0.347	0.738	Acceptable
Item 6	2.7333	6.823	0.149	0.76	Acceptable
Item 7	2.8	6.855	0.174	0.755	Acceptable
Item 8	2.9333	7.375	−0.065	0.762	Acceptable
Item 9	2.8	6.579	0.319	0.741	Acceptable
Item 11	2.7	6.01	0.513	0.719	Acceptable
Item 12	2.7333	6.409	0.344	0.739	Acceptable
Item 13	2.6333	5.689	0.623	0.703	Acceptable
Item 14	2.7667	6.185	0.491	0.723	Acceptable
Item 15	2.7333	6.202	0.446	0.727	Acceptable

4.2. Section B: Diagnostic Performance of the MNSI‐Bl in Comparison With NCS Evaluation (N = 50)

Sociodemographic characteristics of the respondents are shown in Table 4.

Table 4.

Sociodemographic characteristics of the respondents (N = 50).

Parameters	Subjects
Age (years)	48.44 ± 9.44
	(22–60)
Gender (%)
Male (%)	25 (50%)
Female (%)	25 (50%)
Body mass index (BMI) (kg/m2)	28.72 ± 2.76
	(23–35.86)
Systolic blood pressure (mmHg)	123.90 ± 15.85
	(100–160)
Diastolic blood pressure (mmHg)	79.00 ± 10.20
	(60–95)
Fasting blood glucose (mmol/L)	12.09 ± 3.72
	(7–22.78)
HbA1c (%)	8.80 ± 1.71
	(6.9–13.2)
Duration of diabetes (years)	6.35 ± 4.75
	(1–20)

4.3. Evaluation of the Respondents for DPN on NCS

The NCS diagnosed 22 (44%) subjects with DPN among 50 respondents. Simultaneously, among the respondents (N = 50), the mean MNSIq‐Bl and MNSIe scores were 2.48 (± 1.64) and 1.47(± 0.36), respectively. Respondents with DPN had higher MNSIq‐Bl scores (t(48) = 5.24, 95% CI: 1.17– 2.63, p < 0.001) and MNSIe scores (t(48) = 6.67, 95% CI: 3.07–5.79, p < 0.001) compared with those without DPN (Table 5).

Table 5.

Comparison of the NCS and MNSI‐Bl assessment (N = 50).

DPN (NCS)	MNSIq‐Bl score Mean ± SD (Range)	p value	MNSIe Mean ± SD (Range)	p value
Yes (22, 44%)	3.55 ± 1.50 (2–7)	< 0.001*	2.89 ± 1.60 (0–6)	< 0.001*
No (28, 56%)	1.64 ± 1.06 (0–4)		0.36 ± 0.67 (0–2)

4.4. Validity Assessment for MNSIq‐Bl

The ROC analysis shows the AUC was 0.850 (Figure 4). Consequently, the ROC analysis identified a cut‐off value of ≥ 2.5, with 68.2% sensitivity and 75% specificity for MNSIq‐Bl (Table 6).

Figure 4.

ROC curve for MNSIq‐Bl score of study subjects (N = 50).

Table 6.

Diagnostic performance of MNSI‐Bl based on proposed cut‐off value (N = 50).

Variable	True positive	False positive	False negative	True negative	Sensitivity	Specificity	Positive predictive value	Negative predictive value
MNSIq‐Bl ≥ 2.5	15 (30%)	7 (31.8%)	7 (31.8%)	21 (42%)	68.2%	75%	68.2%	75%
MNSIe ≥ 0.5	20 (40%)	7 (31.8%)	2 (4%)	21 (42%)	90.9%	75%	74.1%	91.3%

4.5. Validity Assessment for MNSIe

For MNSIe, according to the ROC analysis, the AUC was 0.918 (Figure 5). A cut‐off value of ≥ 0.5 was found corresponding with the 90.9% sensitivity and 75% specificity (Table 6).

Figure 5.

ROC curve for MNSIe score of study subjects (N = 50).

4.6. Performance of Each Item of the MNSIq‐Bl in Predicting Confirmed Neuropathy

Among 15 items of the MNSIq‐Bl, Items 1 and 2 showed significant discriminatory capability to diagnose DPN as an independent question. Sensitivity and specificity of questionnaire Item 1 were 64% and 93%, respectively. Questionnaire Item 2 showed the sensitivity of 59% and specificity of 39% (Table 7).

Table 7.

Performance of each item of the MNSIq‐Bl in predicting confirmed neuropathy of the study subjects (N = 50).

MNSIq‐Bl	With neuropathy (22)	Without neuropathy (28)	Total	Sensitivity	Specificity	x 2 test
						p value
Item 1	14 (63.6%)	2 (7.1%)	16	64.0%	93.0%	0.001
Item 2	13 (59.0%)	17 (60.7%)	30	59%	39%	0.002
Item 3	8 (36.4%)	1 (3.6%)	9	36.4%	96.4%	0.009
Item 4	18 (81.8%)	18 (64.3%)	36	81%	35.7%	0.292
Item 5	5 (22.7%)	6 (21.4%)	11	22.7%	56.4%	1
Item 6	1 (4.5%)	0	1	4.5%	100.0%	0.903
Item 7	22 (100%)	26 (93%)	48	100.0%	7%	0.581
Item 8	1 (4.5%)	2 (7.1%)	3	4.5%	92.9%	1
Item 9	2 (9.1%)	0	2	9.1%	100.0%	0.367
Item 10	13 (59.1%)	14 (50%)	27	59.0%	50.0%	0.723
Item 11	11 (50%)	8 (28.6%)	19	50.0%	71.4%	0.209
Item 12	11 (50%)	11 (39.3%)	22	50%	60.7%	0.229
Item 13	20 (91%)	25 (89.1%)	45	91%	10.7%	0.198
Item 14	5 (22.7%)	4 (14.1%)	31	22.7%	85.7%	0.689
Item 15	2 (9.1%)	1 (3.6%)	3	9.1%	96.4%	0.829

4.7. Performance of Each Item of MNSIe in Predicting Confirmed Neuropathy

Sensitivity and specificity of each item of the MNSIe part were analyzed, and the results showed that loss of ankle reflex (χ² (1, N = 50) = 17.56, p < 0.001, φ = 0.63), loss of vibration (χ² (1, N = 50) = 18.40, p < 0.001, φ = 0.65), and monofilament test (χ² (1, N = 50) = 32.33, p < 0.001, φ = 0.85) had significant capability for predicting neuropathy. Monofilament test had the highest sensitivity (81.8%) and specificity (100%), followed by loss of ankle reflex with 84.2% sensitivity, and 86.6% specificity. Finally, loss of vibration had 63.6% sensitivity and 96.4% specificity (Table 8).

Table 8.

Performance of individual components of MNSIe in predicting confirmed clinical neuropathy.

MNSIe score		With neuropathy (22)	Without neuropathy (28)	Total	Sensitivity	Specificity	χ 2 test p value
Presence of abnormality on inspection	No	17 (77.3%)	25 (89.1%)	42	62.5%	59.5%	0.446
	Yes	5 (22.7%)	3 (10.7%)	8
Loss of ankle reflex	No	6 (27.3%)	25 (89.3%)	31	84.2%	86.6%	< 0.001*
	Yes	16 (72.7%)	3 (10.7%)	19
Loss of vibration	No	8 (36.4%)	27 (96.4%)	35	63.6%	96.4%	< 0.001*
	Yes	14 (63.6%)	1 (3.6%	15
Response in monofilament test	No	4 (18.2%)	28 (100%)	32	81.8%	100%	< 0.001*
	Yes	18 (81.8%)	0	18

5. Discussion

The current study aimed to translate and validate the Bangla version of the MNSI for DPN. At the same time, the study also explored the diagnostic performance of the tool in comparison with NCS for Bangladeshi type 2 diabetic patients attending a tertiary‐level hospital. The study demonstrated that the translated instrument maintained good reliability and diagnostic validity. This shows that it is an effective screening tool for DPN, especially in resource‐limited clinical settings.

The MNSI has been translated and validated into several languages. It can even support in design of treatment modalities for DPN patients with common psychological problems like depression [15].

A comparison of methodology among the studies shows a variation in their followed steps. For example, the translation and validation process of the Turkish, Brazilian, Polish, and Arabic versions of the tool followed the Beaton et al. methodology [12]. While the Portuguese and the present study of the Bangla version were based ISPOR recommendation [16]. On the other hand, the authors of the Filipino version followed a mixed methodology [17].

The MNSI‐Bl demonstrated good consistency, with Cronbach's alpha values indicating acceptable reliability. These findings align with validation studies conducted in other languages, including Brazilian, Portuguese, Filipino, and Thai, which reported similar reliability coefficients [16, 17, 18, 19].

The Turkish version shows a moderate positive correlation between the questionnaire and the physical examination score [20]. On the other hand, the Polish version shows high stability in test‐retest inter‐item correlation for both the questionnaire (0.73) and the physical examination section (0.97) [21]. Echoing the other studies in the present study, the MNSI‐Bl version also showed a strong inter‐item correlation for both the questionnaire and the physical examination. All the existing evidence further supported the tool's consistency and applicability in clinical assessments.

Linguistic validation for different languages of the MNSI tool was carried out based on ROC curve analysis. The result varied among studies (Table 9). The cut‐off value of the questionnaire, considering AUC, ranged from 0.68 for the Filipino version to 0.91 for the Portuguese version [17, 22]. On the other hand, for examination, the AUC ranged from 0.79 for the Filipino version to 0.94 for the Turkish version [17, 20]. For instance, the MNSIq Filipino version had the highest cut‐off value of 4, where the sensitivity was 73.6% and the specificity was 52.4% [17]. Portuguese and Arabic versions had a similar cut‐off point of 3 with different sensitivity (100% and 93.9%, respectively) and almost similar specificity (64% and 65%, respectively) [16, 22]. On the other hand, the Turkish version found a cut‐off point of 3.5 along with 75.5% sensitivity and 68.1% specificity of and the Polish version found a cut‐off value of 3 with a sensitivity of 40% and a specificity of 100% [20, 21]. The present study for MNSIq‐Bl showed agreement regarding AUC (0.850), along with sensitivity (68%) and specificity (75%) with other studies. But at the same time, the study identified a lower cut‐off value (≥ 2.5).

Table 9.

Comparison of proposed cut‐off value and their diagnostic performance among different versions of MNSI scale.

Reference paper	Questionnaire				Examination
	Cut‐off value	Sensitivity	Specificity	AUC	Cut‐off value	Sensitivity	Specificity	AUC
Sutkowska et al. [21] (polish)	3	33%–40%	90%–100%	0.749	2	81%–84%	60%–70%	0.796
Reyhanioğlu et al. [20] (Turkish)	3.5	75.5%	68.1%	0.783	2.75	87.5%	93.6%	0.939
Barbosa et al. [22] (Portuguese)	≥ 3	100%	64%	0.913	≥ 2	86%	61%	0.798
Dagang et al. [17] (Filipino)	≥ 4	73.60%	52.4%	0.677	≥ 2.5	52.9%	84.1%	0.787
Abuzinadah et al. [16] (Arabic)	≥ 3	93.90%	65%	—	≥ 4	95.9%	62.5%	0.931
MNSI‐Bl	≥ 2.5	68%	75%	0.85	≥ 0.5	90%	75%	0.918

In previous studies, the cut‐off value for MNSIe was found to be two in Polish and Portuguese [16, 22] and four in the Arabic version [21]. On the other hand, the AUC also varied from the highest 0.939 in the Turkish version [17] to the lowest 0.787 Filipino version [20]. In the present study, the MNSIe found a lower cut‐off (≥ 0.5) value compared with other studies. But the AUC for MNSIe was found to be 0.918, which is similar to other studies.

In this paper, individual item analysis of the questionnaire revealed that questions related to numbness (Item 1) and burning pain (Item 2) exhibited high discriminatory capabilities. On the other hand, Dagang et al. [17] showed that stand‐alone Items 1, 2, 4, and 9 demonstrated considerable capacity to diagnose DPN [17]. It may be due to our small sample size in the present study (50) to response against all the items of questionnaire in compared to their sample size (150).

In the physical examination section, the monofilament test emerged as the most sensitive and specific predictor of neuropathy, followed by loss of ankle reflex and vibration sensation. On the contrary, Dagang et al. [17] said that loss of ankle reflex had the most significant predictive capability for peripheral neuropathy [17]. The reason for this difference may be due to the exclusion of ulceration response in our study and the exclusion of the monofilament test in Dagang et al. [17] study.

6. Limitation

Despite the study's significance in providing a validated screening tool for DPN in Bangladesh, it has several limitations. Firstly, the translation process did not involve professional linguists; however, it adhered to established guidelines for cross‐cultural adaptation to ensure accuracy. Secondly, the sample size was relatively small, which may limit the generalizability of findings to the broader diabetic population. Additionally, the study was conducted in a single tertiary care hospital, which may not represent the diverse socioeconomic and geographic distribution of diabetic patients across Bangladesh. Future research with a larger, more diverse sample and additional diagnostic comparisons is necessary to strengthen the findings and further validate the Bangla version of MNSI (MNSI‐Bl).

7. Conclusion

The MNSI tool's Bangla (MNSI‐Bl) version has been verified in accordance with the language validation procedure. Furthermore, MNSI‐Bl is a reliable and valid screening tool. It can be used with cut‐off values of ≥ 2.5 for MNSIq‐Bl and ≥ 0.5 for MNSIe. Peripheral neuropathy can be screened with high sensitivity and specificity using MNSIq‐Bl as well as MNSIe. The study found the examination section, especially the monofilament test, stronger in detecting neuropathy for being more objective. In resource‐constrained settings, MNSI‐Bl can be a better alternative and more applicable tool to screen suspected DPN patients before recommending NCS. We recommend using simple and noninvasive tool like MNSI‐Bl and monofilament test by doctors to screen DPN among suspected patients in low‐resource settings, particularly in primary health care. Additionally, using the tool can aid in early DPN diagnosis and prompt referral for improved care.

Author Contributions

Nawshin Islam: conceptualization, data curation, investigation, validation, formal analysis, writing – original draft, writing – review and editing, methodology, funding acquisition. Qazi Shamima Akhter: supervision, resources. Md. Arifuzzaman: data curation, investigation, supervision, resources. Manasi Saha: data curation, investigation, resources. Nishat Rahman: data curation, investigation, resources. Mahbuba Akter: investigation, data curation. Khadijatul Busra: investigation, data curation. Keya Sarker: investigation, data curation. Mostafizur Rahman: data curation, investigation, project administration. Irfan Nowroze Noor: software, methodology, formal analysis, writing – review and editing, visualization, resources, conceptualization, validation, writing – original draft.

Ethics Statement

The study was committed to ensuring the human rights of the participants. The Research Review Committee of the Department of Physiology and the Ethical Review Committee of DMC approved the study protocol (MIU‐DMC/ECC/2019/42, dated‐06/02/2019). Institutional permission to interview the respondents was obtained from the director's office of the Dhaka Medical College and Hospital.

Consent

Although all the patients were invited, participation was voluntary, and respondents were informed about their right to withdraw his or her participation at any stage of research. They were informed about the benefits and risks of the study. The data was collected, stored and analyzed, maintaining the confidentiality of the data. Before data collection, informed written consent was obtained from each respondent in accordance with the Declaration of Helsinki.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

The lead author, Irfan Nowroze Noor, affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Islam N., Akhter Q. S., Arifuzzaman M., et al., “Translation, Validation, and Diagnostic Performance of the Michigan Neuropathy Screening Instrument (MNSI) in Bangla,” Health Science Reports 8 (2025): e71574, 10.1002/hsr2.71574.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.