Screening Value of the I‐Douleur Neuropathique 4 Questionnaire for Small Fibre Neuropathy in Patients With Painful Syndromes: Insights From 872 Skin Biopsies

ABSTRACT

Background

Small‐fibre neuropathy (SFN), which is defined by the sole involvement of small sensory fibres, often leads to neuropathic pain. The diagnosis of SFN relies on clinical, neurophysiologic, and morphologic abnormalities. A decrease in intraepidermal nerve fibre density (IENFD) on a skin biopsy specimen is considered useful in the diagnosis of SFN in the appropriate clinical context. The self‐administered Douleur Neuropathique 4 (I‐DN4) questionnaire, designed to differentiate neuropathic from somatic pain, shows high sensitivity and specificity but has not been specifically investigated with regard to SFN.

Methods

This retrospective study examined the diagnostic accuracy of I‐DN4 for SFN in 872 patients who were systematically assessed at the time of skin biopsy according to local standards of routine care, after excluding central nervous system diseases and other peripheral nerve disorders by careful clinical examination and medical history.

Results

An I‐DN4 score of ≥ 3 had a sensitivity of 93% and a specificity of 18.13% for diagnosing SFN. ROC analysis revealed an AUC of 0.53, indicating no discriminative ability for this condition. There were no correlations between the I‐DN4 scores and the adjusted IENFD or proximo‐distal IENFD ratio. Only the sensation of burning was a moderate predictor of SFN risk (odds ratio [OR] = 1.65; 95% confidence interval [CI], 1.04–2.68; p = 0.038). Hypoesthesia to pinprick was also associated with a moderate risk of a reduced IENFD at the distal leg (OR = 1.52; 95% CI, 1.03–2.25; p = 0.040).

Conclusions

I‐DN4 is not an effective standalone tool for screening SFN in patients with painful syndromes.

In this retrospective study of 872 patients with pain undergoing skin biopsy, after excluding central nervous system diseases and other peripheral nerve disorders through clinical examination and medical history, the self‐administered DN4 questionnaire (I‐DN4) demonstrated high sensitivity (93%) but very low specificity (18.13%) for small fibre neuropathy, with an overall diagnostic performance close to chance (AUC = 0.53). I‐DN4 scores did not correlate with intraepidermal nerve fibre density (IENFD) or the proximo‐distal IENFD ratio. These findings indicate that the I‐DN4 is not a reliable standalone screening tool for small fibre neuropathy in patients with painful syndromes.

1. Introduction

Neuropathic pain is defined by the International Association for the Study of Pain (IASP) as pain caused by a lesion or disease of the somatosensory nervous system. Small fibre neuropathy (SFN) has recently emerged as a common cause of neuropathic pain. This condition encompasses a broad spectrum of neuropathies characterised by the predominant or selective involvement of thinly myelinated Aδ fibres and unmyelinated C fibres [1]. As nerve conduction studies (NCS) are unremarkable since they explore only large myelinated fibres, diagnosing SFN can be difficult to establish. A skin biopsy quantifying intraepidermal nerve fibre density (IENFD) after PGP 9.5 staining is a key technique for detecting small fibre impairment [2]. However, this diagnostic method is both time consuming and not widely accessible.

The Douleur Neuropathique 4 (DN4) is a 10‐item clinician‐administered questionnaire developed to discriminate neuropathic pain from somatic pain [3]. The first seven items assess the characteristics of the pain and can also be self‐administered (I‐DN4). The last three items involve an examination of the patient to evaluate hypoesthesia to touch and prick, as well as whether the pain can be induced or exacerbated by brushing. A recent meta‐analysis of 27 studies reported a pooled sensitivity and specificity of 0.89 and 0.88 for the DN4, respectively, while a meta‐analysis of nine studies reported respective values of 0.83 and 0.81 for the I‐DN4 [4]. The Joint European Academy of Neurology–European Pain Federation–Neuropathic Pain Special Interest Group of the IASP guidelines on neuropathic pain assessment strongly recommend using both the DN4 and I‐DN4 questionnaires for the assessment of painful syndromes and skin biopsy in the diagnostic workup of neuropathic pain [4]. However, since the DN4 and I‐DN4 questionnaires were not designed with patients with SFN in mind, their diagnostic performance for this specific condition has never been investigated.

This study aimed to determine whether the characteristics of the pain—as described by the first seven items of the DN4 questionnaire (i.e., the I‐DN4)—are an effective screening tool for SFN in individuals with unexplained painful syndromes, without clinical signs or known alternative neurological diagnoses, such as a central nervous system disorder, post‐herpetic neuralgia, radiculopathy, or related conditions.

2. Methods

2.1. Study Design and Participants

This single‐centre retrospective study was conducted at the Neurology Department of Limoges University Hospital in France. Since May 2010, as part of routine care, patients with pain referred for a skin biopsy have been systematically assessed using the DN4 questionnaire. We included all patients who had both skin biopsy results and a completed DN4 questionnaire in their medical records. The 10 items of the French version of the DN4 questionnaire were assessed by the same examiner who performed the skin biopsy immediately before its completion. The examining physician, always a neurologist, conducted an interview and non‐standardised clinical examination to rule out any other apparent neurological causes of pain, particularly those originating from central nervous system lesions, which precluded the biopsy. In addition, patients were not included if, after reviewing the medical records available at our institution (largest hospital in our rural region), another neurological cause was established following the biopsy, such as a central nervous system disorder, postherpetic neuralgia, radiculopathy, or related conditions. However, patients affected by or who ultimately developed a large fibre neuropathy could be included. Other medical data included age, height, weight, signs of small sensory fibre impairment, comorbidities, and NCS results.

2.2. Skin Biopsies and SFN Definition

Skin biopsies were performed using a 3‐mm disposable circular punch after administering local anaesthetic at two sites: the distal leg (10 cm above the lateral malleolus) and proximal thigh (20 cm below the anterior superior iliac spine). Samples were processed with PGP9.5 staining as previously described [5]. IENFD was manually counted according to the 2010 European Federation of Neurological Societies/Peripheral Nerve Society (EFNS/PNS) guidelines using a Nikon DxM1200 light microscope across at least three non‐consecutive sections. Quantifications were performed by two individuals (LM and LR) who were required to reach a consensus through a double‐blind process.

IENFD values were considered indicative of SFN diagnosis if they fell within the lowest 0.05 quantile adjusted for sex and decade, based on established normative values [6], which are routinely used for patient care in our centre. A table summarising the age‐specific cut‐offs employed in this study is provided in the Supporting Information (Table S1). Patients were considered affected by SFN if they had both a reduced IENFD at the distal leg and clinical signs of small fibre impairment (pinprick sensory loss and/or allodynia) in a distribution consistent with peripheral neuropathy [2], regardless of whether nerve conduction studies indicated concomitant large fibre involvement (mixed fibre neuropathy). For a subsequent analysis, pure SFN was defined as meeting the criteria for SFN without any evidence of large fibre impairment on NCS.

2.3. Statistical Analysis

The study population characteristics were described using descriptive statistics, with quantitative values presented as means (standard deviations). We used the Student's t‐test for quantitative variables and the chi‐squared test for qualitative variables to compare patients with and without SFN. The diagnostic accuracy of the first seven items of the DN4 questionnaire (i.e., I‐DN4 questionnaire) was evaluated using receiver operator characteristic (ROC) and precision‐recall (PR) curves, comparing patients with and without a confirmed diagnosis of SFN, regardless of the presence of concomitant large fibre impairment on NCS (mixed fibre neuropathy). The area under the ROC curve (AUC) with a 95% confidence interval (CI) was calculated for the entire cohort. The same analysis was performed for the subgroup of patients without diabetes and without any evidence of large fibre impairment on NCS, thereby including only pure SFN and excluding those with mixed fibre neuropathy.

Given that the normative values of IENFD may be debated, we performed an additional set of analyses to assess whether our results were biased. First, we evaluated the diagnostic accuracy of the I‐DN4 questionnaire using an alternative set of published normative values [7]. Second, we calculated age‐ and sex‐adjusted IENFD values at the distal leg using software available at http://www.skinbiopsylab.it [8], and we also determined the proximo‐distal ratio of IENFD, defined as the ratio of the distal leg to thigh IENFD. Pearson correlations were used to assess the correlation between I‐DN4 questionnaire scores and (1) adjusted IENFD values at the distal leg and (2) the proximo‐distal IENFD ratio.

Univariate logistic regression analyses were conducted to evaluate the associations between each of the seven items of the I‐DN4 questionnaire and the presence of SFN. As the last three items of the DN4 may be redundant with our definition of SFN and could introduce bias, their association was analysed using univariate logistic regression only with the risk of an IENFD in the lowest 0.05 quantile, adjusted for sex and decade.

All statistical analyses were two‐tailed with a significance level of 0.05, performed using Rstudio (version 1.4.1717) and the precrec package [9].

2.4. Ethics

This study was conducted following the principles outlined in the Declaration of Helsinki. Owing to the retrospective design of this study, individual written informed consent was not required. The study protocol was approved by the Ethics Committee of the University Hospital of Limoges (Approval Number 10‐2023‐02).

3. Results

This study included 872 patients with results from both skin biopsies and DN4 questionnaires from May 2010 to March 2024. Of these, 100 (11.5%) were considered to be affected by SFN, including those with mixed fibre neuropathy. Among the 83 NCS available in the SFN group, 49 (59%) were normal, reflecting pure SFN. The mean age was 54.7 years, and 63.6% of the patients were female. Diabetes mellitus was more frequent in the SFN group (20% vs. 7.9%; p = 8.8e‐5). Detailed patient characteristics are presented in Table 1.

TABLE 1.

Patient characteristics.

Variables	Total (n = 872)	No SFN (n = 772)	SFN (n = 100)	p
Age (years)	54.7 (14.2)	54.6 (14.2)	55.5 (14.6)	0.29
Female sex	555 (63.6%)	497 (64.4%)	58 (58%)	0.21
BMI (kg/m2)	25.5 (5.8)	25.4 (5.6)	26.7 (7)	0.19
Diabetes mellitus	81 (9.3%)	61 (7.9%)	20 (20%)	8.8e‐5
I‐DN4 ≥ 3	725 (83.1%)	632 (81.9%)	93 (93%)	0.005
I‐DN4 score	4.2 (1.7)	4.2 (1.8)	4.5 (1.4)	0.23
IENFD (fibres/mm)	6.58 (3.60)	7.14 (3.38)	2.27 (1.87)	2.2e‐16
Normal NCS	544/691 (78.7%)	495/608 (81.4%)	49/83 (59%)	3.0e‐6

Note: Quantitative variables are expressed as mean (standard deviation), and qualitative variables as n (%).

Abbreviations: BMI, body mass index; I‐DN4, self‐administered Douleur Neuropathique 4 questionnaire; IENFD, intraepidermal nerve fibre density; NCS, nerve conduction study; SFN, small fibre neuropathy.

An I‐DN4 score of at least 3 had a diagnostic sensitivity of 93% (95% CI: 86.11%–97.14%) and specificity of 18.13% (95% CI: 15.48%–21.04%), with positive and negative predictive values of 12.83% (95% CI: 12.14%–13.55%) and 95.24% (95% CI: 90.6%–97.65%), respectively. The positive likelihood ratio was 1.14 (95% CI: 1.07%–1.21%), and the negative likelihood ratio was 0.39 (95% CI: 0.19%–0.80%). The accuracy was 26.72% (95% CI: 23.81%–29.79%). ROC curve analysis showed an AUC of 0.53 (95% CI: 0.48–0.58) when comparing patients with and without a diagnosis of SFN (Figure 1A). Applying another set of published normative IENFD values [7] to the same ROC analysis yielded nearly identical results, with an AUC of 0.53 (95% CI: 0.44–0.61) (Figure S1).

FIGURE 1.

Diagnostic utility of the I‐DN4 questionnaire in detecting SFN (n = 872). (A) Receiver operating characteristic (ROC) and precision‐recall (PR) curves for the I‐DN4 questionnaire comparing patients with and without SFN, including mixed fibre neuropathy showing an area under the curve (AUC) of 0.53 (95% CI: 0.48–0.58). (B) ROC‐PR curve analysis in the subgroup of patients without diabetes with normal nerve conduction studies, thus excluding mixed fibre neuropathy, showing an AUC of 0.53 (95% CI: 0.44–0.61). (C) Pearson correlation between I‐DN4 scores and adjusted IENFD (r = −0.014; p = 0.68). (D) Pearson correlation between I‐DN4 scores and the proximo‐distal IENFD ratio (r = −0.0048; p = 0.89). CI, confidence interval; I‐DN4, Self‐administered Douleur Neuropathique 4 questionnaire; IENFD, Intraepidermal nerve fibre density; SFN, Small‐fibre neuropathy.

Among patients without diabetes with an unremarkable NCS, the AUC was 0.53 (95% CI: 0.44–0.61) (Figure 1B), reflecting similar discriminability of the I‐DN4 in pure SFN.

Pearson correlation analysis revealed no significant correlation between the I‐DN4 score and adjusted IENFD at the distal leg (r = −0.014; p = 0.68) (Figure 1C) or the proximo‐distal IENFD ratio (r = −0.0048; p = 0.89) (Figure 1D).

Univariate logistic regressions applied to all seven items of the I‐DN4 questionnaire revealed that no item was significant except for burning sensation, with an OR of 1.65 (95% CI, 1.04–2.68; p = 0.038). Hypoesthesia to pinprick was associated with the risk of a reduced IENFD at the distal leg, with an OR of 1.52 (95% CI, 1.03–2.25; p = 0.040) (Table 2).

TABLE 2.

Univariate logistic regression analysis of the items in the DN4 questionnaire.

Items	OR	95% CI	p
A
1 Burning	1.65	1.04–2.68	0.038
2 Painful cold	0.99	0.63–1.50	0.93
3 Electric shocks	1.28	0.83–1.99	0.27
4 Tingling	1.05	0.64–1.81	0.84
5 Pins and needles	1.45	0.88–2.48	0.16
6 Numbness	1.03	0.64–1.69	0.91
7 Itching	1.08	0.69–1.65	0.74
B
8 Hypoesthesia to touch	1.41	0.99–1.99	0.054
9 Hypoesthesia to pinprick	1.52	1.03–2.25	0.040
10 Brushing	1.03	0.74–1.43	0.85

Note: The items in the DN4 questionnaire that were (A) associated with the risk of small fibre neuropathy or (B) above a 0.05 quantile value adjusted on age and sex of ankle intraepidermal nerve fibre (IENF) density according to EFNS/PNS guideline. Bold values signifies < 0.05.

Abbreviations: CI, confidence interval; DN4, Douleur Neuropathique 4; EFNS/PNS, European Federation of Neurological Societies/Peripheral Nerve Society; OR, odds ratio.

4. Discussion

This study presents a 14‐year investigation at a single centre involving 872 individuals with pain who underwent skin biopsy. We found that while the I‐DN4 score is sensitive, it lacks specificity in identifying SFN in patients with unexplained painful syndromes and no clinical evidence of an alternative central or peripheral nervous system disorder, when using the standard cut‐off of 3/7. The overall diagnostic performance of the I‐DN4 was low and near random chance. The metrics remained nearly identical when applied only to pure SFN, excluding patients with mixed fibre neuropathy.

Our results underscore the challenges in diagnosing SFN. While some I‐DN4 items are undoubtedly present in SFN, most of these symptoms are poor discriminators compared with those for other painful syndromes. Only the sensation of burning appeared to be a moderate predictor of the risk of SFN. The DN4 questionnaire was originally designed to distinguish between neuropathic and non‐neuropathic pain. By extension, one might consider diagnosing SFN when unexplained pain is accompanied by a positive DN4 or I‐DN4 score, normal NCS, and the absence of other evident neurological causes. We observed this trend in patients referred for skin biopsies at our centre and now demonstrate that this approach is ineffective. It is important to note that the neuropathic pain group in the study by Bouhassira et al. [3] was highly heterogeneous, with only 12 (13.5%) cases classified as unspecified “polyneuropathies”. Currently, over 300 studies are referenced on clinicaltrials.gov using the search term “DN4”. While the I‐DN4 and DN4 scores are clinically valuable for the initial categorisation of painful syndromes in daily medical practice, clinicians and researchers should exercise caution in extrapolating their application to populations beyond those originally studied.

The strengths of this study include the large sample size, the systematic collection of DN4 questionnaire data in the medical records, and the concurrent administration of the DN4 with the skin biopsy procedure. The quantification of IENFD was well standardised, as all assessments were conducted by the same two investigators (L.M. and L.R.) for 14 years, using a double‐blind reading process. Given the potential controversy surrounding the normative values of IENFD, we also demonstrate the absence of a correlation between the I‐DN4 score and the age‐ and sex‐adjusted IENFD at the distal leg. Moreover, since most cases of SFN are of toxic or metabolic origin with a length‐dependent pattern, the absence of correlation between the proximo‐distal IENFD ratio and the I‐DN4 score strengthens our conclusions.

Diagnosing SFN is challenging because there is no gold standard. It must be based on consistent evidence and the exclusion of differential diagnoses. The primary limitation of our study is that despite all patients being examined by a neurologist, our definition of SFN for the ROC analysis relied solely on a reduced IENFD above the 0.05 quantile values for age and sex combined with clinical signs of small fibre impairment. This association is enough to diagnose SFN according to the Besta criteria [2]. However, even if the IENFD quantification is a useful examination for investigating SFN, a normal IENFD does not exclude the diagnosis. PGP 9.5 staining detects nerve fibres that are intact enough to be quantified but do not provide information about their function [10]. The use of Quantitative Sensory Testing (QST), which is part of the Besta criteria (but not used in our centre), would likely have improved the detection of SFN in some of our patients. However, as reported in the original study, since the diagnostic performance of skin biopsy is substantially higher than that of QST, we believe that the addition of QST would only marginally affect our results. This is further supported by the fact that, as described above, we found no association between the proximo‐distal IENFD ratio and the I‐DN4, despite the strong inverse correlation between IENFD and all thermal thresholds of the QST reported in the Besta study.

The unclear cause of pain in our patients without SFN remains a significant limitation. We did not specifically investigate these patients, as most were outpatients referred from the pain and rheumatology departments. Moreover, the concept of nociplastic pain has recently emerged as a cause or mechanism of neuropathic‐like pain without overt anatomical involvement of the nervous system [11]. Although this concept has no firm physiological basis to date, it is considered that patients with fibromyalgia and related disorders experience this particular type of pain. Indeed, the descriptors used by patients with fibromyalgia frequently fulfill the first seven items of the DN4 questionnaire (excluding the last three relying on clinical examinations) [12, 13]. An alternative explanation of our findings would be that at least some patients who did not have a reduced IENFD suffered from an early stage of SFN or had a ‘pure functional’ form of SFN. To our knowledge, this last concept has not been firmly demonstrated except in the rare patient who suffers from genetic painful sodium channelopathy [14].

Another limitation is that some patients may have exhibited genuine neuropathic pain, that is, pain linked to damage to the nervous system injury, but not specifically due to small fibre involvement as evaluated by skin biopsy. However, as part of routine care, all patients were interviewed and examined by the neurologist pre‐biopsy. The neurologist, who could cancel the biopsy, aimed to avoid unnecessary, time‐consuming biopsy analyses by excluding patients with clinical signs or a history suggestive of alternative neurological conditions associated with pain, such as multiple sclerosis, post‐herpetic pain, or radiculopathy with a nerve root distribution, among others. This potential bias could therefore be partly limited, enabling us to estimate the performance of the I‐DN4 in distinguishing small nerve fibre damage, particularly in patients referred by pain units, in the absence of any other obvious neurological condition.

Our research highlights the challenging diagnostic process in suspected SFN cases and the limited discriminatory ability of the symptoms included in the I‐DN4 questionnaire in this context, except for a sensation of burning. Although IENFD quantification through skin biopsy is valuable, it is time‐consuming and not widely accessible. Therefore, there is a critical need for the development of new bedside tools to assist clinicians in the diagnosis of SFN. These tools may benefit from standardised quantification of pinprick hypoesthesia, which was significantly reduced when IENFD was altered and provides a robust clinicopathological rationale for SFN.

Author Contributions

Simon Frachet: conceptualisation, methodology, software, validation, formal analysis, investigation, data curation, writing – original draft, visualisation. Emilie Soust: conceptualisation, methodology, investigation, writing – original draft. Laurence Richard: investigation, resources, writing – review and editing. Aurore Danigo: investigation, resources, writing – review and editing. Claire Demiot: investigation, writing – review and editing. Laurent Magy: conceptualisation, methodology, writing – original draft, supervision.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Cut‐off values of intraepidermal nerve fibre densities used to define small fibre impairment. The 0.05 quantile was derived from Bakkers et al. (Neurology, 2009).

Figure S1: Receiver operating characteristic and precision‐recall curves for the I‐DN4 questionnaire comparing patients with and without SFN, using Lauria et al. (J Peripher Nerv Syst, 2010) normative values for intraepidermal nerve fibre densities. AUC = 0.54 (95% CI: 0.49–0.59).

Frachet S., Soust E., Richard L., Danigo A., Demiot C., and Magy L., “Screening Value of the I‐Douleur Neuropathique 4 Questionnaire for Small Fibre Neuropathy in Patients With Painful Syndromes: Insights From 872 Skin Biopsies,” European Journal of Neurology 33, no. 2 (2026): e70499, 10.1111/ene.70499.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.