Critical illness polyneuropathy (CIP) affects 30% to 50% of the most severely critically ill patients and is the most frequent acute polyneuropathy in the intensive care unit (ICU) ¹ . CIP is often associated with an acute, primary myopathy called critical illness myopathy (CIM), and both conditions occur in patients with multiple organ dysfunctions and failure (MOF). Indeed, CIP and CIM represent the failure of the neuromuscular system in patients with MOF ¹ . CIP classically presents as a sensory-motor axonal polyneuropathy causing difficulty in weaning patients from a ventilator, flaccid limbs, and a possible reduction in deep tendon reflexes. Amplitude reduction of both the compound muscle action potential (CMAP) and the sensory nerve action potential (SNAP) is the predominant electrophysiological finding in CIM and CIP; latency and nerve conduction velocity remain normal or are only slightly decreased. CIM is a primary myopathy with distinctive electrophysiological and morphological findings ¹ . Other features of CIM include increased CMAP duration, normal SNAPs, reduced muscle excitability on direct stimulation and myopathic motor unit potentials on needle electromyography. The clinical features are often the same as for CIP, but sensation, if testable, is normal.

Traditional methods to diagnose CIP and CIM include manual testing of muscle strength using the UK Medical Research Council (MRC) score ² or dynamometry ³ to demonstrate severe weakness, electrophysiological tests to explore the function of peripheral nerves and muscles, and muscle biopsy ¹ . Conventional nerve conduction studies (NCS) with measurement of conduction velocity, CMAP and SNAP amplitudes, electromyography (EMG) and other specialised techniques such as direct muscle stimulation or axonal excitability testing may reveal nerve or muscle dysfunction with a high degree of specificity ⁴ . However, these techniques require specialised personnel, they are time-consuming and they do not allow diagnosis of small intra-epidermal nerve fiber pathology ⁵ . Conventional NCS-EMG may require up to 90 minutes to be completed ⁴ . Considering the high prevalence of ICU-acquired neuromuscular disorders, it is unrealistic for conventional NCS-EMG to be used as a large-scale screening tool.

Electrophysiological investigations of peripheral nerves and muscles offer several advantages. They are minimally invasive and easily repeatable, they can be performed at the bedside and the results are immediately available ⁶ . Electrophysiological alterations indicating CIP or CIM can also be demonstrated in non-collaborative patients. In comatose patients or in those with persisting sedation or septic encephalopathy who develop severe muscle weakness or paralysis after ICU admission, electrophysiological investigations can be performed to avoid unreasonably pessimistic prognosis by identification of CIP or CIM as the cause ⁷ . Electrophysiological alterations are not only essential to establish the diagnosis of CIP and CIM, but they also have an earlier onset than clinical signs or they can be documented at an earlier stage ^{7– 11} , thus offering the advantage of a timely diagnosis and the opportunity to conduct potentially valuable interventions before structural muscle-nerve alterations become established ⁴ . CMAP reduction is an early event which precedes clinical signs: its onset can be abrupt within 24 hours of normal findings ¹⁰ , and it can be observed as early as 48 hours before clinical signs in patients with sepsis ¹² . Moreover, NCS may also be useful to predict hospital mortality ⁹ , and short ¹³ and long-term morbidity ^{8, 14, 15} . CMAP after direct muscle stimulation (dmCMAP) may precede the development of ICU-acquired weakness by several days ¹¹ . Therefore there is a need for a rapid, simple, accurate and valid electrophysiological test to identify CIP and CIM early in ICU.

In a previous study called CRIMYNE ¹⁰ , we found that a simplified electrophysiological investigation, the peroneal nerve electrophysiological test (PENT), had high sensitivity (100%) and moderate specificity (67%) in identifying patients with a diagnosis of CIP or CIM using complete NCS as the reference diagnostic standard. However, the CRIMYNE study had methodological limitations that precluded the acceptance of PENT as a valid screening test in the ICU. Specifically, patients underwent complete NCS only if the PENT was abnormal, thus precluding the possibility of accurately detecting false negative results. Moreover, NCS assessors were not blind to the results of PENT. According to the STARD Statement for reporting studies of diagnostics accuracy, knowledge of the results of the index test can influence the reading of the reference standard, and vice versa. Such knowledge is likely to increase the agreement between results of the index test and those of the reference standard, leading to inflated measures of diagnostic accuracy ¹⁶ .

The accuracy of new diagnostic tests should be evaluated before their introduction into clinical practice, because invalid tests may yield exaggerated and biased results, which may trigger their premature adoption in clinical practice and lead to test overuse, increasing patient risks and costs, or to test underuse, missing opportunities to improve health ^{17, 18} . Test validation involves comparing the new test with a reference standard, defined as the best available method for establishing the presence or absence of the condition of interest ¹⁷ . Therefore, we designed the multicentre CRIMYNE-2 study according to rigorous methodology to provide an unbiased evaluation of the accuracy of PENT in diagnosing electrophysiologically-proven CIP and CIM diagnosed using complete NCS-EMG as the reference standard.

We conducted a prospective observational study between April 2010 and June 2012 in nine medical-surgical Italian ICUs joining the Margherita-Prosafe Project, an international research campaign that collects and analyses clinical data on all patients admitted to the participating ICUs ¹⁹ . The Project, promoted by the GiViTI ( Gruppo Italiano per la Valutazione degli Interventi in Terapia Intensiva, Italian Group for the Evaluation of Interventions in Intensive Care Medicine) is based on an electronic Case Report Form (eCRF), which was extended to collect the data of interest for the CRIMYNE-2 study. Two databases were used: Margherita-Due, supported until December 2010, and Prosafe, released on January 2011. Both databases were organized with a core data set and a supplemental module designed for the CRIMYNE-2 research project.

The core data of the Margherita-Prosafe Project included information related to the patients' condition on admission (demographics, admission diagnoses, comorbidities, surgical status), severity-of-illness scoring systems (Simplified Acute Physiology Score [SAPS II], Glasgow Coma Scale [GCS]); procedures, organ failures and complications arising during the ICU; ICU and hospital outcome. The CRIMYNE-2 module included the electrophysiological variables needed to diagnose CIP and CIM, as well as the final diagnosis concluded by the clinical neurophysiologist (see “Possible diagnoses”).

Inclusion criteria were age ≥ 18 years and ICU stay ≥ 3 days, because this patient population is at increased risk of developing CIP and CIM ^{1, 20} .

Exclusion criteria were a previous history of neuromuscular disorders or chronic conditions associated with neuromuscular disorders ( i.e., diabetes, renal failure, chronic alcohol abuse), use of neurotoxic or myotoxic drugs, evidence of altered neuromuscular transmission at repetitive stimulation test either caused by neuromuscular blocking agents or disease, lower limb disorders precluding NCS and EMG (for example edema, fractures, amputation, plaster casts), and terminal conditions. Sedation was not an exclusion criterion, and, if present, was not discontinued nor reduced prior to testing.

In each ICU, the inclusion and exclusion criteria were verified daily by the intensivist in charge for the CRIMYNE-2 study. Patients enrolled were subjected to the index test (PENT) and the reference diagnostic standard (NCS-EMG) on the same day by two independent clinicians who were blind to the results of the other test. The research team was allowed to exclude patients if another patient in the same ICU was being concomitantly studied.

The study was approved by the Ethics Committee of each participating centre (protocol number 554/2010). Detailed written information was provided to the patients and family members about the study protocol, the scope of research, and the safety of electrophysiological investigations. Written informed consent was obtained from the patient where appropriate. In case of altered consciousness, the Ethics Committees waived the requirement for consent, as in Italy relatives are not regarded as legal representatives of the patient in the absence of a formal designation ²¹ . Written informed consent was requested from all surviving patients as soon as they regained their mental competency. All investigations were conducted according to the principles expressed in the Declaration of Helsinki.

All data were centrally reviewed by four of the research team (NL, BG, GN, GB), and diagnoses were evaluated for consistency with predefined electrophysiological diagnostic criteria of CIP and CIM. In the case of discordant results, the centres were contacted and diagnoses were discussed with local personnel until a consensus was reached. If needed, supplemental electrophysiological material was requested from the participating centre and analysed.

During the study period, 342 patients were evaluated and 124 (36.3%) were enrolled ( Figure 2).

Discordant results after central revision were documented in five cases. Two cases, classified by the local clinical neurophysiologist as normal despite minimally reduced ulnar CMAP amplitudes, were confirmed as normal at central revision after upgrading the outdated limits of normality used centrally. One case classified locally as abnormal because of minimally reduced proximal peroneal CMAP amplitude (recorded above the fibular head) and normal distal peroneal CMAP (below the fibular head) was classified as normal after central revision, according to the study protocol. One case initially misclassified as having normal PENT because of erroneous transcription of the data was reclassified to abnormal PENT after re-examination of the original electrophysiological data. One case was eventually classified as showing inconclusive results. This patient had normal findings at PENT and abnormal findings at complete NCS-EMG performed 24 hours later. Because this was a protocol violation and CMAP reduction may develop rapidly, within 24 hours of normal findings ¹⁰ , we excluded the patient from the analysis.

Of the 124 patients enrolled, 121 (98%) were investigated with both the index test (PENT) and the reference standard (complete NCS-EMG) with no protocol violation ( Figure 2). The general characteristics of the study population are presented in Table 1. Acute neurologic patients were the main group of patients, with head trauma (35 patients), spontaneous intracranial haemorrhage (25) and aneurysmal subarachnoid haemorrhage (21) being the most frequent causes of admission. Patients had a severe clinical condition, with high prevalence of sepsis, single organ failure (25%) or MOF (62%), and prolonged duration of mechanical ventilation and ICU stay, particularly in patients with abnormal NCS-EMG. Hospital mortality was higher in patients with abnormal NCS-EMG results than in patients with a normal NCS-EMG ( Table 1).

Of 23 patients with normal NCS-EMG, all also had normal findings with PENT with no false negative results ( Figure 2). Of 97 patients with abnormal ENG-EMG, four had peroneal nerve mononeuropathy at complete NCS-EMG and abnormal findings at PENT, and were classified as false positives. The sensitivity of the PENT was 100% (95% CI 96.1-100.0) and the specificity 85.2% (95% CI 66.3-95.8). Sensitivity was comparable to that found in the CRIMYNE study (100%). Specificity was higher in this study compared to the CRIMYNE study, but the difference was not statistically significant (CRIMYNE: 67%, CRIMYNE-2: 85%, p=0.08).

The electrophysiological diagnosis in the 93 patients with abnormal findings at complete NCS-EMG was CIP in 35 patients (37.6%), combined CIP and CIM in 16 (17.2%) patients and undetermined in 42 (45.2%) patients.

Electrophysiological investigations were performed on median ICU day 9 (IQR 5-16). The median time needed to perform PENT and the complete NCS-EMG was 10 minutes (IQR 8.0-10.5) and 50 minutes (40–60). In 85 cases (70.2%), the peroneal nerve examined first was abnormal, thus eliminating the need for contralateral testing.

In this multicentre diagnostic accuracy study, we found that the PENT had 100% sensitivity and high specificity in diagnosing CIP or CIM, confirming the preliminary results of the previous CRIMYNE study ¹⁰ . The time needed to complete the test was 10 minutes, which was considerably shorter than the time needed to complete NCS-EMG (50 minutes).

Conventional NCS and EMG have never gained popularity in the ICU because they are time-consuming, not readily available, expensive, and technically challenging as they require specialised personnel. Therefore, our results may be important in promoting a wider use of electrophysiological investigations in the ICU. As a validated, high-sensitivity, minimally invasive, non-volitional and quick diagnostic test, PENT can accurately exclude CIP or CIM if the result is normal, and can be proposed for the assessment and monitoring of the neuromuscular function in the early stages of critical illnesses.

Abnormal PENT cannot discriminate between CIP, CIM or combined CIP and CIM; in fact, in our study, diagnosis was undetermined in almost half of patients. Because precise definition of pathology can be relevant for predicting the recovery of muscle strength ^{13, 14} , PENT can be used as a screening tool to select patients deserving further investigations. Moreover, “false positive patients” are those with peroneal nerve mononeuropathy who are worthy of medical attention. Involvement of the peroneal nerve is invariably detected in patients with CIP or CIM, possibly because it is the longest nerve of the body and, hence, it is mostly vulnerable to the energy deficit caused by tissue ischemia or dysoxia ^{7, 10} . In fact, the biosynthesis takes place in the neuronal cell body and then the structural components are moved into the axons and transported to their final destination to generate the action potential and to maintain axonal integrity, a process that requires considerable amount of energy ^{7, 10} . This might explain the reason why reduced peroneal nerve CMAP is highly sensitive compared to other tests. Pragmatically, measurements of peroneal CMAP could be implemented to diagnose all patients with ICU-acquired neuromuscular disorders, either CIP or CIM, while other tests, such as measurement of dmCMAP, could be used to differentiate CIM from CIP ²⁹ . Recent studies suggest that the duration of the CMAP can be prolonged in patients with CIM ³⁰ .

Five single-centre studies have used simplified electrophysiological tests of neuromuscular function in the ICU ^{10, 11, 31– 33} . In two randomized controlled trials on intensive insulin treatment ^{31, 32} , the diagnosis of CIP was suggested by the presence of abundant spontaneous activity in the form of positive sharp waves and fibrillation potentials on EMG ³⁴ . No reference diagnostic standard was used, thus the sensitivity and specificity of the test are unknown. In the CRIMYNE study ¹⁰ , we used the same index test and reference diagnostic standard of the CRIMYNE-2 study, but comparison was unblinded and was limited to patients with abnormal findings at the index test, which may have inflated the test sensitivity. In a prospective observational study in surgical ICU patients ¹¹ , the sensitivity and specificity of dmCMAP in predicting the development of ICU-acquired weakness were 83.3% and 88.8%, respectively. In a more recent study ³³ , sensitivity and specificity of combined peroneal nerve CMAP and sural nerve SNAP evaluations were 100% and 81%, respectively. However, in both studies all electrophysiological measurements were performed by the same investigators, and no blind assessment of the reference diagnostic test was implemented ³⁵ .

Among non-volitional non-electrophysiological tests, ultrasound-guided assessment of muscle mass can reliably detect early muscle changes. In a recent prospective cohort study in 63 critically ill patients who were developing acute muscle wasting, the rectus femoris cross-sectional area decreased significantly from days 1 to 7 by 12.5% ³⁶ . Simultaneous measurement of muscle strength and muscle mass has been suggested as a critical step in all future studies investigating ICU-acquired weakness ³⁷ . However, assessing muscle strength may be difficult in the early stage of acute disease, and the use of PENT could be of value in assessing the generation of the action potential, which is an essential pre-requisite for muscle contraction.

The prevalence of electrophysiological abnormalities was 76%, which is higher than reported in a systematic review [46% (95% C.I. 43–49%)] ²⁰ . We enrolled patients with high incidence of sepsis and MOF, and prolonged mechanical ventilation ⁵ . It is therefore not surprising that this patient group also had a high incidence of ICU-acquired neuromuscular disorders that are currently viewed not as isolated events, but rather as an integral part of the process leading to MOF ¹ . In two recent studies, severe muscle weakness was documented in 74% of patients ³⁸ and electrophysiological abnormalities in 87% ¹¹ . In an unselected population of critically ill children, the occurrence of generalised muscle weakness was substantially lower (1.7%) ³⁹ , reflecting the need for future studies to include much larger samples of unselected ICU patients to evaluate the true incidence of ICU-acquired neuromuscular disorders.

Several limitations of the current study are worth discussing. First, we did not test muscle strength clinically, and therefore, we could not evaluate the proportion of patients with pure electrophysiological alterations that developed definite CIP or CIM. Current recommendations suggest that patients with a MRC sum-score of less than 48 or reduced handgrip dynamometry should undergo physical rehabilitation without any further testing ^{1, 40} . NCS and EMG testing should be reserved for those patients not improving despite receiving such treatment. MRC and dynamometry can be easily implemented in the ICU, and inter-observer reliability is good. However, volitional tests require the patient’s collaboration, hence an inability to perform the test or the recording of low values may occur as a result of coma, delirium, sedation, injury, or simply poor patient motivation or attention. Recent studies suggest that manual muscle testing can be insufficient for the early detection of ICU-acquired neuromuscular disorders in most patients ⁴¹ . In fact, ascertainment of ICU-acquired weakness at ICU day 7 post-awakening, as originally defined ² , can be of limited value because many patients have been already discharged from the ICU at that time, reflecting the change in clinical ICU practice towards earlier discharge ³⁸ . Moreover, inter-observer agreement on MRC sum-score assessment is only moderate in patients with MRC scores that are lower than 48, probably because the ability to perform the volitional MRC sum-score among the most severe patients is variable ^{38, 42} . PENT is not dependent upon patients’ collaboration and could be performed at an early stage to detect initial functional derangement and to use the results to start investigational treatments with the aim of interrupting pathological mechanisms at their onset ⁴³ . Several studies have demonstrated that electrophysiological alterations are followed by muscle weakness, but they included a small number of selected ICU patients ^{7– 11} . PENT is a quick test and could be used in large ICU populations.

Second, the large majority of patients were critically ill neurologic patients, which may limit the generalisability of the results to other ICU populations. However, development of CIP and CIM is independent from the cause of ICU admission, and it rather depends on MOF, which is also a common complication in acute neurologic patients ⁴⁴ , as it is confirmed in this study.

Lastly, we excluded patients with diabetes, who may represent up to 16% percent of patients admitted to the ICU ⁴⁵ . Peripheral neuropathy is a common complication in diabetic patients ⁴⁶ . However, generalised symmetric polyneuropathy typically involves sensory nerves with mild reduction in distal sensory response amplitudes (e.g., those of the sural nerves) ⁴⁷ . In contrast, motor response amplitudes, as those tested with peroneal nerve evaluation, are generally preserved and decrease only in more advanced disease.

To the authors’ knowledge, this is the first time that a simplified electrophysiological test to assess the peripheral nerve and muscle function has been evaluated in an ICU population using a rigorous methodology based on independent and blinded comparison with a reference standard.

Measurement of CMAP of the peroneal nerve showed 100% sensitivity and high specificity in diagnosing probable CIP or CIM, and did not require patients’ collaboration. Potential useful applications of the test can be at the early ICU stage, when volitional tests can be rarely performed. At a later stage, before discharge from ICU or acute-care hospital, a normal test excludes CIP or CIM and the need for further electrophysiological investigations. If patients present abnormal values, they might have CIP, CIM or focal peripheral nerve complications such as peroneal entrapment neuropathy that warrant neurological consultation.

The application of this test to wider populations of ICU patients might allow a more precise estimation of the true incidence of ICU-acquired neuromuscular disorders. With adequately powered observational cohort studies, future studies might be able to evaluate the relationship between electrophysiologically-proven CIP or CIM and reduced muscle strength and mass in the ICU ^{36, 37} , as well the relationship between these acute changes and persistent muscle weakness or persistent physical dysfunction at long-term follow-up ⁴² .

The data referenced by this article are under copyright with the following copyright statement: Copyright: � 2014 Latronico N et al.

Data associated with the article are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).

figshare: Data set for the CRIMYNE-2 study on the validation of peroneal nerve test to diagnose polyneuropathy and myopathy in 121 patients, doi: http://dx.doi.org/10.6084/m9.figshare.1021506 ⁴⁸

Detailed written information was provided to the patients and family members about the study protocol, the scope of research, and the safety of electrophysiological investigations. Written informed consent was obtained from the patient where appropriate. In case of altered consciousness, the Ethics Committees waived the requirement for consent, as in Italy relatives are not regarded as legal representatives of the patient in the absence of a formal designation ²¹ . Written informed consent was requested from all surviving patients as soon as they regained their mental competency.