Objectives Neuromuscular dysfunction in critically ill patients is attributed to either critical illness myopathy (CIM) or critical illness polyneuropathy (CIP) or a combination of both. However, it is unknown whether differential diagnosis has an impact on prognosis. This study investigates whether there is an association between the early differentiation of CIM versus CIP and clinical prognosis.
Methods The authors included mechanically ventilated patients who featured a Simplified Acute Physiology Score II (SAPS-II) ≥20 on three consecutive days within the first week after intensive care unit (ICU) admission. Fifty-three critically ill patients were enrolled and examined by conventional nerve-conduction studies and direct muscle stimulation (184 examinations in total). The first examination was conducted within the first week after admission to the ICU.
Results In this cohort of critically ill patients, CIM was more frequent (68%) than CIP (38%). Electrophysiological signs of CIM preceded electrophysiological signs of CIP (median at day 7 in CIM patients vs day 10 in CIP patients, p<0.001). Most patients with CIP featured concomitant CIM. At discharge from ICU, 25% of patients with isolated CIM showed electrophysiological signs of recovery and significantly lower degrees of weakness. Recovery could not be observed in patients with combined CIM/CIP, even though the ICU length of stay was significantly longer (mean 35 days in CIM/CIP vs mean 19 days in CIM, p<0.001).
Conclusion Prognoses of patients differ depending on electrophysiological findings during early critical illness: early electrophysiological differentiation of ICU acquired neuromuscular disorder enhances the evaluation of clinical prognosis during critical illness.
- Critical illness myopathy
- critical illness polyneuropathy
- direct muscle stimulation
- intensive care unit
- clinical neurology
- intensive care
- motor neuron disease
Statistics from Altmetric.com
- Critical illness myopathy
- critical illness polyneuropathy
- direct muscle stimulation
- intensive care unit
- clinical neurology
- intensive care
- motor neuron disease
Since the early description of critical illness myopathy (CIM)1 and critical illness polyneuropathy (CIP),2 3 the primary cause of these Intensive Care Unit (ICU) acquired weaknesses remains unresolved.4 ICU acquired weakness complicates recovery in critically ill patients and prolongs the duration of mechanical ventilation and length of stay in ICU.5 6 Some authors advise against differentiation of CIM and CIP, as differentiation is complicated and may not result in consequences.7 8
Electrophysiological abnormalities such as low compound muscle action potentials or pathological spontaneous activity can be detected within 1 week of ICU admission but do not allow for distinguishing CIM and CIP.9 10 In sedated patients without voluntary muscle contraction, electrophysiological differentiation between CIM and CIP is difficult. CIM diagnosis relies on direct demonstration of muscle membrane dysfunction.11 12 Authors using this technique of direct muscle stimulation found myopathy to be more frequent in critically ill patients.7 11–15 CIP with associated sensory nerve involvement is assessed by a reduction in sensory nerve amplitudes, whereas assessment of motor CIP remains difficult. Rich and colleagues introduced a ratio that divides nerve-evoked compound muscle action potentials (neCMAP) by direct muscle-evoked compound muscle action potentials (dmCMAP) in order to differentiate between myopathy (this ratio is expected to be around 1) and motor CIP axonopathy (this ratio is expected to be small and around zero).13
In this longitudinal study of 53 critically ill patients, conventional nerve-conduction studies and direct muscle stimulation were combined to determine whether there is any association between the early differentiation of CIM versus CIP and clinical prognosis.
The study was approved by our local ethics committee. Written informed consent of legal proxies was obtained according to the Declaration of Helsinki.
We performed this observational study in a 14-bed surgical ICU over a period of 18 months. Patients who required mechanical ventilation and featured a Simplified Acute Physiology Score II ≥ 20 (SAPS-II)16 on three consecutive days within the first 7 days after ICU admission were eligible for inclusion. Pre-existing neuromuscular disorders, severe head trauma or bleeding diathesis were previously defined as exclusion criteria. All patients were treated according to our standard operating procedures of intensive care medicine, adopting evidence-based bundles for severe sepsis.17 18 Severity of illness was monitored daily by repeated ratings of SAPS-II and Sequential Organ Failure Assessment score (SOFA).19
Every 3 days after admission to the ICU, electrophysiological bedside studies were recorded by portable two-channel Keypoint Medtronic equipment (Skovlunde, Denmark). Once either CIM or CIP was detected or once patients showed adequate awareness, diagnostic testing was repeated once a week. Given that patients featured sufficient awareness at ICU discharge (Ramsay score ≤2),20 the muscle strength of upper and lower limbs was evaluated and graded according to the Medical Research Council score.21 Whenever possible, we examined two proximal/distal muscles in each extremity and divided the total by the number of muscles examined.
Motor-nerve-conduction velocity and compound muscle action potential amplitude after nerve stimulation (neCMAP) were unilaterally performed in median, peroneal as well as in tibial nerves and recorded from abductor pollicis brevis, extensor digitorum brevis and abductor hallucis muscles. Sensory nerve-conduction studies were unilaterally conducted in sural and median nerves. Surface electrodes were used for stimulation and recording. In case of missing recordings or in the presence of oedema, subdermal electrodes were used in all patients. Nerve-conduction measurements were compared with normal values from age-matched individuals that were provided by the neurophysiological laboratory of the Charité.
During sedation, electromyography was performed in deltoid, biceps brachii, extensor digitorum longus, abductor pollicis brevis, rectus femoris and tibialis anterior muscles using concentric needle electrodes to assess pathological spontaneous activity. As soon as patients showed sufficient awareness and voluntary muscle contraction was possible, quantitative electromyography was applied whenever possible in extensor digitorum longus and tibialis anterior muscles. A total of 20 different motor unit action potentials (MUAP) were sampled by random insertion of a concentric needle electrode into four different regions of an examined muscle, each recorded at 10 ms, 50 μV and filter settings of 500 Hz and 10 kHz.22 23 The mean duration of collected non-polyphasic MUAPs was compared with normal values from healthy age-matched volunteers.22 24
Assessment of compound muscle action potential amplitudes following direct muscle stimulation (dmCMAP) was performed by longitudinal placement of either conventional stimulating surface electrodes or by subdermal electrodes along muscle fibres just proximal of the distal tendon insertion in case of oedema. Muscles were stimulated by gradually increasing strength (from 10 to 100 mA) at 1 Hz and a pulse duration of 0.1 ms. For recordings, disposable concentric needle electrodes (length 25 mm or 37 mm; diameter 0.46 mm) and/or disposable gel surface electrodes were used and placed 15–50 mm proximal to the stimulating electrode, guided by muscle twitch. Whenever no twitch was visible, the recording concentric needle electrode was pointed at four different directions so as not to miss small amplitudes. Muscles were assumed to be inexcitable if responses could still not be obtained. dmCMAP amplitudes were measured peak to peak. Filter settings were 500 Hz and 10 kHz. Limb temperature was kept at >32°C. The examination included tibialis anterior and abductor pollicis brevis. According to Trojaborg and colleagues, dmCMAP amplitudes recorded by concentric needle electrodes <3 mV were considered to be pathological and consistent with myopathy.11
Electrophysiological measurements of tibialis anterior and abductor pollicis brevis muscles of healthy volunteers (age 22–74 years) in our laboratory provided reference values for dmCMAP amplitudes using surface electrodes (n=17) or concentric needle electrodes (n=8).
To assess motor CIP, we calculated neCMAP/dmCMAP ratios (recorded in tibialis anterior muscle with peroneal nerve stimulation at the knee AND in abductor pollicis brevis muscle with median nerve stimulation at the wrist) as introduced by Rich and colleagues: ratios <0.5 indicate motor neuropathy; ratios >0.5 in combination with reduced dmCMAP amplitudes indicate myopathy, while ratios >0.5 in presence of normal dmCMAP amplitudes indicate normal findings.13
To determine muscle-fibre conduction velocity (MFCV), latencies of muscle-fibre action potentials were determined and calculated for the measured distance between electrodes. The perpendicular position of recording needle electrodes was ensured. Responses earlier than 8 ms were likely to be conducted via intramuscular nerve twigs and were not included.11 12 14
The diagnostic criteria for electrophysiological examination were as follows: (1) ICU control: patients presenting no pathology; (2) ICU unspecific: patients presenting unspecific pathology (pathological spontaneous activity and reduced neCMAP) not verifying myopathy or neuropathy; (3) CIM patients: patients presenting reduced dmCMAP in at least one muscle examined in addition to unspecific findings and normal sensory/motor nerve conduction velocity (isolated CIM); (4) CIP patients: patients presenting reduced SNAP and/or ne/dmCMAP ratio <0.5 in addition to unspecific findings (isolated CIP); (5) CIM/CIP patients: patients presenting characteristics of combined CIM and CIP—reduced dmCMAP AND reduced SNAP and/or ne/dmCMAP ratio <0.5.
Patients were classified according to their most severe electrophysiological findings during their ICU stay. To compare electrophysiological data between patient groups, findings from the first examination presenting the most severe electrophysiological classification were chosen for each patient.
Results are expressed as arithmetic mean±SD for electrophysiological data, median and (25/75) percentiles (if number of patients less than four, only the median is expressed) for categorical or non-normally distributed data, or frequencies (%) for qualitative data, respectively. Statistical tests were conducted with non-parametric tests by Mann–Whitney U test for two independent samples, Kruskal–Wallis test for three or more independent samples, and the Fisher exact test for qualitative data. In case of small samples, greater differences in sample sizes, large but unbalanced groups, data sets containing ties or sparse data, tests were carried out in an exact version. A diagnostic test performance was evaluated by receiver operating characteristics analysis using MUAP duration in tibialis anterior muscle <11.1 ms as the electrophysiological gold standard for diagnosing myopathy in patients capable of voluntary muscle contraction or, alternatively, using amplitudes of dmCMAP in tibialis anterior muscles of sedated patients not being capable of voluntary muscle contraction.
Kaplan–Meier curves were estimated to show the cumulative incidence of different electrophysiological disorders developing over time and to estimate probabilities for ICU discharge after the first day of awareness in CIM and CIM/CIP patients. Differences between groups considering cumulative incidences were tested using a univariate logrank test.
In univariate Cox proportional hazard regressions, we tested the impact of CIM and CIM/CIP as well as the illness severity on the duration between the first day of adequate awareness and ICU discharge (as dependent variable). In Cox regressions with time-dependent covariates, dmCMAP and SNAP amplitudes were included as indicators of myopathy and neuropathy, respectively, while repeated recordings of SAPS-II and SOFA score during adequate awareness were included as indicators of illness severity. These variables were also analysed by a stepwise (backward) procedure of multivariate Cox regression accounting for time-dependent covariates. HRs with 95% CIs (95% CI (HR)) and corresponding p values were calculated for each risk factor. A p value of <0.05 (two-sided) was considered statistically significant. Statistical analysis was performed using SPSS, Version 14 (SPSS, Chicago, Illinois) and SAS, Version 9.1 (SAS Institute, Cary, North Carolina).
Two hundred and twelve patients required mechanical ventilation and featured SAPS II≥20 on three consecutive days within the first 7 days after ICU admission and were therefore eligible for inclusion. Patients with pre-existing neuromuscular disorder (n=24), severe head trauma (n=14) or bleeding diathesis (n=33) (thrombocytopaenia <20 000/μl) were excluded. Sixteen patients could not be included due to logistical reasons. Written informed consent by legal proxy could not be obtained in 72 patients. Finally, 53 patients were included, and a total of 184 electrophysiological examinations were conducted. One or two examinations were conducted in nine patients each, three exams in 12 patients and four or more examinations in 23 patients.
Patient classification is shown in figure 1. Only one patient with pre-existing Wilson's disease and two previous liver transplantations showed reduced SNAP amplitudes without any evidence of motor CIP or CIM at both exams. However, pre-existing sensory polyneuropathy could not be ruled out.25 To prevent confusion, all three patients classified as ICU unspecific as well as the only patient with isolated sensory nerve involvement were not considered in the tables and figures.
Patients' characteristics are shown in table 1.
The amplitudes of nerve-conduction studies are shown in table 2. Predominant involvement of the lower limbs was observed in all patients. Parameters such as motor/sensory nerve conduction velocity, distal motor latency or F-Wave did not deviate from normal values in any of the examined patients. Pathological spontaneous activity such as fibrillation potentials or positive sharp waves were mostly of moderate activity and could be observed in different muscles from patients classified as ‘ICU unspecific,’ ‘CIM’ and ‘CIM/CIP.’ Tibialis anterior and extensor digitorum longus muscles were most frequently affected.
Muscle-specific electrophysiological data are shown in table 3. In quantitative electromyography, we observed an increased incidence of polyphasic potentials, and the recruitment pattern at maximum effort was fully or only mildly reduced despite severe weakness in patients classified as ‘CIM/CIP’ and respectively ‘CIM’, respectively, indicating a myopathy. We did not find any signs of denervation such as a reduced recruitment pattern or elevated MUAP amplitudes.
In healthy subjects (age 22–74 years), dmCMAP amplitudes were not below 0.6 mV when recorded by surface electrodes and not below 3 mV when recorded by concentric needle electrodes. For surface electrodes, 95% CIs were between 0.9 mV and 6 mV for tibialis anterior muscle and between 0.6 mV and 20 mV for abductor pollicis brevis muscle. Considering concentric needle electrodes, the 95% CIs were between 4 mV and 19 mV and between 6.8 mV and 13 mV, respectively.
MFCV was positively correlated with dmCMAP amplitude, with reduced amplitudes indicating a slower MFCV (ρ=0.55 and R Quadrate=0.401).
We did not observe any isolated motor CIP in any patient, as ne/dmCMAP ratios were consistently >0.5, which indicates either myopathy in the presence of reduced dmCMAP amplitudes (CIM- and CIM/CIP patients) or normal findings in the presence of normal dmCMAP amplitudes (healthy subjects and ICU-controls). Receiver operating characteristics analysis verified dmCMAP in tibialis anterior muscle during sedation as a predictor of myopathy, as later diagnosed by MUAP duration in the same muscle once voluntary contraction was applicable. The best relationship of sensitivity (70%) to specificity (83.3%) was observed at the cut-off value of 3.2 mV for dmCMAP, which is compatible with standard values from Trojaborg and colleagues.11
The onset of electrophysiological pathology is shown in figure 2. Six patients could not be included in analysis due to treatment in external hospitals prior to ICU admission. Abnormal dmCMAP amplitudes occurred significantly earlier than abnormal SNAP (p<0.001), indicating that CIM occurs prior to CIP with associated sensory nerve involvement during early critical illness. CIM patients showed a reduced dmCMAP amplitude median at day 7 (5/11), while CIM/CIP patients showed a reduced SNAP amplitude median at day 10 (4/13).
Confounders prolonging ICU length of stay showed that classification as CIM or CIM/CIP independently influenced ICU length of stay, whereas illness severity was comparable between both groups (p=0.005) (table 4).
Once sedation was ended, patients classified as CIM/CIP stayed significantly longer than patients classified as CIM (p=0.05) (figure 3).
On discharge from ICU, some patients classified as CIM featured amplitude recovery of dmCMAP in tibialis anterior muscles (figure 4A) and neCMAP in tibialis and peroneal nerves, while patients classified as CIM/CIP consistently showed reduced amplitudes of dmCMAP and neCMAP (figure 4B).
On ICU discharge, muscle strength according to the MRC score was significantly lower in patients classified as CIM/CIP (n=14; examination was precluded in four patients due to death and in two patients due to logisitical reasons; mean MRC score in upper limbs 3.5; mean MRC score in lower limbs 3.25) than in patients classified as CIM (n=8; examination was precluded in six patients due to death and in two patients due to logistical reasons; mean MRC score in upper limbs 4.5, p=0.002; mean MRC score in lower limbs 4.0, p=0.004).
Direct muscle stimulation facilitates diagnosis of CIM in the early course of critical illness. During analgesia and sedation, other methods of clinical assessment are not applicable. Electrophysiological signs of CIM precede electrophysiological signs of CIP. Isolated CIP was not observed in any patient, it occurred only in combination with myopathy. Clinical courses of patients classified as CIM and CIM/CIP differ. Both CIM and CIM/CIP independently influence ICU length of stay after the end of sedation. However, patients classified as CIM/CIP feature significantly higher degrees of weakness at ICU discharge and longer ICU lengths of stay than patients classified as CIM. Electrophysiological recordings showed that some patients classified as CIM showed signs of recovery at discharge from ICU, while all patients classified as CIM/CIP consistently featured electrophysiological pathology at ICU discharge.
By comparing dmCMAP amplitudes with MUAP duration in quantitative electromyography—the gold standard of proving myopathy22 23—we were able to show that assessment of dmCMAP amplitudes represents a valuable tool to differentiate between CIM and CIP during the early course of critical illness, when voluntary muscle contraction is not applicable due to sedation (sensitivity 70%, specificity 83.3%).
The technique of direct muscle stimulation has been evaluated in healthy subjects26 27 and patients suffering from weakness and/or weaning failure caused by critical illness.7 11–13 28 Published reference data for dmCMAP depend on recording characteristics of electrodes: concentric needle electrodes (Trojaborg et al11: 8.0±0.9 mV, lower limit ≥3 mV, n=18; Lefaucheur et al12: 9.61±2.36 mV, lower limit ≥ 4.88 mV, n=12 AND our data: 8.0±2.1 mV, lower limit ≥3 mV, n=8), subdermal electrodes (Trojaborg et al11: 4.5±1.7 mV, lower limit ≥1 mV, n=18) or surface electrodes (our data: 2.8±0.4, lower limit ≥0.6 mV, n=17). Assessing dmCMAP with surface electrodes may be of advantage in patients with bleeding diathesis. However, measurements with concentric needle electrodes have the advantage of also recording smaller activity from within the deeper muscle.
MFCV values of healthy subjects (Troni et al26: 3.53–4.24 m/s male and 2.96–3.74 m/s female; Trojaborg et al11: 6.4±0.3 m/s; Allen et al14: 3.0–5.5 m/s) and CIM patients (Trojaborg et al11: 4.5±0.2 m/s; Allen et al14: 2.32±1.12 m/s, our data: 5.2±1,1 m/s) are various. Interestingly, patients classified as ICU control showed a reduction in MFCV (5.9±1.6 m/s, n=3) and dmCMAP amplitude (5.6±1.8 mV, n=12) compared with healthy volunteers (our data and Trojaborg et al11), possibly indicating early impairment of muscle membrane excitability on a subclinical level that is not accompanied by distinct levels of weakness after the end of sedation.15 This indicates that critical illness in general causes impairment of muscle membrane excitability; however, in order to cause muscle organ failure, an additional pathomechanism is essential.
Ne/dmCMAP ratios <0.5 are supposed to indicate motor axonopathy.13 Since this ratio was >0.5 in all of our patients, the presence of motor axonopathy is questionable. Z'Graggen and colleagues assessed the existence of membrane depolarisation in motor nerves by applying nerve excitability testing,29 proving a nerve membrane affection but not finally proving the existence of a motor axonopathy.
It should be emphasised that SNAP abnormalities indicate sensory neuropathy and cannot serve as definite evidence towards neuropathic involvement in clinical weakness. Early reports attributed weakness in critically ill patients mostly to distal motor axonopathy on the basis of non-specific electrophysiological abnormalities and neglected the possibility of primary muscle fibre disorder.3 30 By applying the technique of direct muscle stimulation, we and others were able to show that CIM is frequent in critically ill patients.7 11–15 However, ne/dm CMAP ratios did not add any more information to the diagnosis of myopathy than dmCMAP amplitudes.
Onset and incidence of neuromuscular disorder
To the best of our knowledge, this is the first study reporting that CIM is verified significantly earlier than CIP. Unspecific findings such as pathological spontaneous activity or reduced neCMAP amplitudes do not differentiate between CIM and CIP and were observed within the first week after ICU admission. This is in line with three earlier studies describing early onset of neuromuscular dysfunction in the ICU without differentiating between myopathy and neuropathy.9 10 31 We presume that early pathological spontaneous activity is related to muscle membrane depolarisation leading to elevated excitability.
Coexistence of pathological spontaneous activity and reduced dmCMAP amplitudes was surprising to us, as we expected a reverse relationship due to contrary pathology. However, this is in line with findings from a rat model of CIM describing concomitant membrane depolarisation and reduced excitability, which was attributed to voltage-gated sodium-channel dysfunction.32 33 It was furthermore reported that endotoxin of Gram-negative bacteria causes a hyperpolarised shift in the gating of voltage-gated sodium channels in human skeletal muscle,34 which, in the presence of muscle membrane depolarisation in critically ill patients,35 will finally cause muscle membrane inexcitability.
We observed both systemic inflammation and illness severity during early critical illness to present significant risk factors for development of CIM or CIM/CIP. It nevertheless remains unresolved as to why some patients show more severe or combined affection of muscles and nerves than others.
Recovery from neuromuscular disorder
Some authors advise against adoption of electrophysiological differential diagnosis7 8 since distinguishing between CIM and CIP would not be associated with clinical prognosis. We recently reported that CIM constitutes the primary reason for ICU acquired weakness presenting in critically ill patients suffering from sepsis, systemic inflammatory response syndrome or multiple organ failure once sedation is ended.15 For the first time, we report that ICU length of stay is markedly prolonged in patients classified as CIM/CIP compared with patients with isolated CIM and that this does not result from illness severity after sedation was ended. Expecting subsequent prolongation of recovery time, it was interesting to observe that patients classified as CIM/CIP still featured severe weakness at ICU discharge in contrast to CIM patients. This is consistent with observations by Guarneri and colleagues describing a better long-term prognosis (1 year after hospital discharge) in patients diagnosed as having isolated CIM.36 It should nevertheless be mentioned that patients classified as CIM/CIP showed more pronounced dmCMAP amplitude reduction and shorter MUAP duration, both characteristics of pronounced myopathy, than patients with pure CIM.
As some CIM patients show recovery of dmCMAP amplitude reduction at ICU discharge, we presume inactivation of voltage-gated sodium channels in structural intact muscle fibres to be reversible after successful treatment/elimination of potentially involved factors, which would explain a return to normal function within days.7 11 13 37 Recovery was not observed in patients classified as CIM/CIP, which may be either due to pronounced muscle dysfunction with selective myosin-filament loss in fast twitch muscle fibres OR due to muscle denervation causing muscle membrane depolarisation potentially counteracting recovery.11 35 37 38
In conclusion, we were able to show that clinical prognosis differs according to electrophysiological differential diagnosis during early critical illness. CIM in combination with CIP was associated with more severe weakness at ICU discharge and longer ICU length of stay than isolated CIM. During the early course of critical illness, when voluntary muscle contraction is not applicable due to sedation, we recommend conventional electrophysiological recordings in combination with direct muscle stimulation (adds another 5–15 min) to maintain precise differential diagnosis. This supports better prediction of weaning difficulties which occur in both CIM and CIM/CIP patients, and furthermore assists clinicians in estimating motor function recovery at ICU discharge.
The authors want to thank W Trojaborg for advice and assistance in direct muscle stimulation techniques and quantitative electromyography.
Funding Deutsche Forschungsgemeinschaft.
Competing interests None.
Patient consent Obtained.
Ethics approval Ethics approval was provided by the Local Ethics Committee Charité, Berlin.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.