Clinical Neurophysiology 115 (2004) 16831688 www.elsevier.

com/locate/clinph

Usefulness of an increase in size of motor unit potential sample

Simon Podnar*
Institute of Clinical Neurophysiology, Division of Neurology, University Medical Center Ljubljana, SI-1525 Ljubljana, Slovenia Accepted 21 February 2004 Available online 25 March 2004

Abstract Objective: It is known that the sensitivity of quantitative electromyographic (EMG) analysis of motor unit potentials (MUPs) improves with an increase in MUP sample size to more than 20. However, no normative data and estimate of sensitivity have been published. Methods: In the present study sample sizes of 5, 10, 15, 20, 30 and 40 MUPs were obtained from the external anal sphincter (EAS) muscles of 81 controls and 70 patients with cauda equina lesions. For each sample size normative limits and sensitivities for mean values and outliers were calculated for 8 MUP parameters. Results: As the size of the MUP samples increased, normative limits narrowed and sensitivities increased for both statistics of all MUP parameters (sensitivities were 26% at 10, 44% at 20, and 67% at 40 MUPs with mean values and outliers of MUP area, duration and number of turns). Conclusions: Our results conrmed a substantial increase in the sensitivity of MUP analysis by enlargement of the MUP sample size to more than 20 MUPs. The gain in sensitivity seem to be greater than the increase obtained by examination of contralateral EAS muscle. Signicance: Findings might be useful to clinical neurophysiologists planning strategies for electrodiagnostic evaluation of lower sacral segments. q 2004 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
Keywords: External anal sphincter muscle; Motor unit potentials; Needle electromyography; Sample size; Sensitivity

1. Introduction A basic statistical theorem states that the variance of samples is reduced by increasing their size. A reduction in variance narrows the reference limits in both controls and patients, thus improving their separation (increasing sensitivity). This was previously demonstrated in a small group of myopathic muscles for motor unit potential (MUP) sample sizes of 5 20 MUPs (Engstrom and Olney, 1992). For neuropathic muscles a further increase in the sensitivity of MUP analysis was demonstrated for MUP sample sizes of 20 100 MUPs (sensitivity for individual statistics of different MUP parameters was 30 62% at 20, 49 82% at 40, and 80 100% at 100 MUPs) (Podnar and Mrkaic, 2003). That study, however, used random computer
Abbreviations: EAS, external anal sphincter; EMG, electromyography; MUPs, motor unit potentials. * Tel.: 1-522-1500; fax: 1-522-1533. E-mail address: simon.podnar@kclj.si (S. Podnar).

sampling of MUPs from common control and neuropathic MUP pools (Podnar and Mrkaic, 2003). During diagnostic quantitative electromyography (EMG) MUPs are sampled directly from individual muscles, which form distinct MUP pools. Such sampling is thus not random. Results of our previous study (Podnar and Mrkaic, 2003) can thus only be taken as a rough (theoretical) estimate of the effect of an increase in MUP sample sizes on reference limits and sensitivities for the diagnosis of neuropathy, useful only for comparison between samples of different sizes. That study did thus not provide valid normative data and realistic sensitivities of quantitative EMG analysis for MUP samples of different sizes. In the present study we followed a similar protocol to that in our previous study (Podnar and Mrkaic, 2003), but instead of sampling from common control and neuropathic MUP pools, MUP samples of different sizes were obtained from the external anal sphincter (EAS) muscles of individual controls and patients with neuropathic lower sacral lesions.

1388-2457/$30.00 q 2004 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2004.02.016

1684

S. Podnar / Clinical Neurophysiology 115 (2004) 16831688

2. Materials and methods From the database of subjects with quantitative MUP analysis all subjects with at least 35 MUPs sampled from the EAS muscle were identied. A control group was recruited from subjects who participated in our normative studies (Podnar et al., 2000, 2002), and from subjects referred to our EMG laboratory for electrodiagnostic evaluation of minor sacral dysfunction (i.e. slight urinary stress incontinence, premature ejaculation, etc.), or ano-genital pain. From the control group all examinees with a medical history or ndings on neurological examination pointing to a possible disorder of the sacral nervous system were excluded. [In these patients sacral dysfunction was most probably due to local causes (e.g. prostatic hypertrophy, disrupted bladder support, etc.).] In contrast, subjects in the patient group all had a history pointing to an adequate causative event, perianal sensory loss on clinical examination or spontaneous denervation activity in the lower sacral myotomes on concentric needle EMG (performed 3 10 weeks after the onset of the condition), and radiological ndings supportive of a cauda equina or conus medullaris lesion. The National Ethics Committee of Slovenia approved the study, and all subjects gave informed consent. MUPs were sampled from the subcutaneous and deeper EAS muscles by standard concentric needle EMG electrode, following previously described insertions (Podnar et al., 1999). Needle electrodes were inserted into the middle of the anterior and posterior halves (two insertion sites on each side for each the subcutaneous and the deeper EAS muscle), and then angled backwards and forwards in a systematic manner, so that at least 4 5 sites, separated at least 10 mm, were sampled in each of 4 different EAS muscles. An advanced EMG system (Keypoint, Medtronic Functional Diagnostics, Skovlunde, Denmark), with standard lter settings (5 Hz 10 kHz), and the template-operated multiMUP analysis program were used (Podnar and Vodusek, 1999; Podnar et al., 2002; Stalberg et al., 1995). In multiMUP analysis the operator initiates the computerized acquisition of the previous 4.8 s of EMG activity, from which MUPs are then automatically extracted, quantitatively described and sorted into up to 6 classes, each representing consecutive discharges of an individual MUP (Podnar et al., 2002; Stalberg et al., 1995). Eight MUP parameters [amplitude, duration, area, number of phases, number of turns, thickness ( area/amplitude) (Nandedkar et al., 1988), size index ( 2 log(amplitude) area/ amplitude) (Sonoo and Stalberg, 1993), and spike duration] were evaluated. From each included subject the rst 5, 10, 15, 20, 30, and 40 MUPs were included, starting with the left (when present), followed by the right subcutaneous EAS muscle, and, when necessary, continued with the left, followed by the right deeper EAS muscle. In every subject the mean value for all 8 evaluated MUP parameters was determined for each sample size. From means obtained in individual

subjects the group mean and SD was calculated. For sample sizes of 10, 20, 30, and 40 MUPs, in addition, from individual controls data the 10th and the 90th percentile values of all evaluated MUP parameters were determined. Using non-parametric statistics from 10th percentile values obtained in individual subjects lower reference limits were set at the group 5th percentile (95% specicity for values below normative limits), and from 90th percentile values upper reference limits were set at the group 95th percentile (95% specicity for values above normative limits) (Podnar and Mrkaic, 2003; Podnar et al., 2002; Stalberg et al., 1994). Thus, using control MUPs for mean values normative limits (mean ^ 2SD) for MUP samples of 6 sizes, and for outliers normative limits for 4 MUP sample sizes indicated were calculated (95% specicity). The values obtained in patients were then compared to normative limits for MUP samples of the same size, and the sensitivities were calculated by dividing the number of patient samples of that particular size outside (above or below) the normative limits by the number of all patients samples of that size. The sensitivity obtained by separate and simultaneous use of mean values and outliers was calculated separately for each of 8 evaluated MUP parameters. In addition, the cumulative sensitivities for 3 MUP parameters (MUP area, number of turns, and duration) (Podnar and Mrkaic, 2002), and for all 8 evaluated MUP parameters were determined.

3. Results Eighty-one controls with 4225 MUPs, and 70 patients with 4261 MUPs fullled the inclusion criteria. Of the included subjects, 5 controls and 3 patients had less than 40 (but more than 35) MUPs. 3.1. Normative limits As the size of the MUP samples increased, normative limits for mean values and outlier limits of all evaluated MUP parameters progressively narrowed. This trend also continued for MUP sample sizes of more than 20. A similar narrowing of reference limits was observed in patients. For most MUP parameters patients had similar lower reference limits and much larger upper reference limits compared to controls (Table 1, Fig. 1). 3.2. Sensitivity The sensitivity for mean values and outliers progressively increased as the number of MUPs within the sample increased (Table 1). Such relations were also obtained for MUP sample sizes of more than 20 MUPs and were common to all evaluated MUP parameters (Fig. 2). For few MUP parameters at some sample sizes, sensitivities were higher for mean values (at sample size of 40 MUPs only for

Table 1 Effect of size of the MUP sample on reference limits and sensitivity for mean values and outliers for 8 evaluated MUP parameters MUP parameter Statistics, sample size (N) Mean Control (mean ^ SD) Area (mV ms) 10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40 353 ^ 209 354 ^ 154 351 ^ 123 344 ^ 116 5.50 ^ 1.37 5.48 ^ 1.09 5.45 ^ 0.88 5.42 ^ 0.88 2.93 ^ 0.70 2.95 ^ 0.51 2.95 ^ 0.42 2.93 ^ 0.40 2 0.21 ^ 0.39 2 0.21 ^ 0.30 2 0.20 ^ 0.24 2 0.22 ^ 0.23 0.83 ^ 0.16 0.83 ^ 0.14 0.83 ^ 0.11 0.82 ^ 0.10 409 ^ 177 414 ^ 135 414 ^ 113 407 ^ 107 2.87 ^ 0.68 2.85 ^ 0.52 2.86 ^ 0.44 2.84 ^ 0.43 3.05 ^ 0.55 3.09 ^ 0.40 3.09 ^ 0.33 3.07 ^ 0.30 Patient (mean ^ SD) 497 ^ 425 498 ^ 323 484 ^ 299 492 ^ 256 6.36 ^ 2.26 6.47 ^ 1.75 6.40 ^ 1.60 6.51 ^ 1.46 3.24 ^ 0.82 3.37 ^ 0.60 3.34 ^ 0.54 3.39 ^ 0.48 2 0.01 ^ 0.46 2 0.01 ^ 0.39 2 0.03 ^ 0.35 2 0.00 ^ 0.33 0.92 ^ 0.22 0.92 ^ 0.17 0.91 ^ 0.16 0.93 ^ 0.15 495 ^ 251 501 ^ 218 489 ^ 187 496 ^ 159 3.30 ^ 1.16 3.38 ^ 0.78 3.34 ^ 0.70 3.39 ^ 0.64 3.20 ^ 0.52 3.35 ^ 0.39 3.33 ^ 0.35 3.39 ^ 0.48 Sensitivity (%) 9 19 19 30 11 17 23 24 10 16 20 23 9 13 16 16 13 13 14 23 10 13 13 16 19 21 23 23 4 9 16 20 Outliers Control limits (595%) 461416 511293 541189 551189 1.6012.56 1.8012.7 1.7811.76 1.8011.26 0.97.1 1.07.1 1.07 1.07 2 1.61 to 1.54 2 1.54 to 1.50 2 1.43 to 1.20 2 1.43 to 1.28 0.411.80 0.411.77 0.421.73 0.421.66 781219 621217 871289 871225 1.07.6 1.187.12 1.187.04 1.186.82 1.06.3 1.96.1 1.96.1 2.06.0 Patient limits (595%) 57 2475 53 2178 52 2149 50 1806 1.7816.56 1.7814.86 1.7814.63 1.7814.75 0.9 8.4 1.0 8.3 1.0 8.3 1.0 8.1 2 1.47 to 2.06 2 1.51 to 1.84 2 1.50 to 1.74 2 1.49 to 1.74 0.402.16 0.412.00 0.411.97 0.402.11 89 1788 85 1982 84 1836 88 1603 1.179.53 1.009.08 1.188.97 1.088.68 1.9 7.2 2.0 7.0 2.0 7.0 2.0 6.2 Sensitivity (%) 11 24 30 39 13 14 34 43 14 17 24 20 9 17 36 40 19 20 23 33 14 21 21 27 16 26 26 31 10 13 6 4 14 29 30 43 16 20 40 43 16 23 27 30 11 21 37 44 21 24 29 49 16 23 23 29 21 31 34 36 10 19 17 20 Combined, sensitivity (%)

Duration (ms)

S. Podnar / Clinical Neurophysiology 115 (2004) 16831688

Turns (N)

Size index

Thickness (ms)

Amplitude (mV)

Spike duration (ms)

Phases (N)

MUP, motor unit potential. MUP samples were obtained from the external anal sphincter (EAS) muscles of 81 controls and 70 patients with cauda equina or conus medullaris lesions. Outlier limits were obtained as the group 5th percentiles of 10th percentile values obtained in individual subjects and the group 95th percentiles of 90th percentile values obtained in individual subjects. Fractional numbers of outlier reference limits for number of phases and number of turns is due to statistical methodology applied, and is aimed to reduce error introduced by integer numbers.

1685

1686

S. Podnar / Clinical Neurophysiology 115 (2004) 16831688

parameters (MUP area, number of turns and duration) (Podnar and Mrkaic, 2002) were 26, 44, 54 and 67%, and for all 8 MUP parameters, 44, 51, 69 and 77%, respectively.

4. Discussion Utility of a diagnostic test depends critically on its diagnostic powerability to separate normal and pathological subjects. To increase diagnostic power of MUP analysis a number of approaches are possible: (1) improvement in computer software for better sampling of most pathologic MUPs [e.g. very polyphasic, unstable, with late components-satellites (Podnar and Fowler, 2004)]; (2) development of new MUP parameters [e.g. thickness (Nandedkar et al., 1988), size index (Sonoo and Stalberg, 1993)] and employment of advanced statistical techniques (e.g. discriminant analysis, multivariate analysis) with better discrimination abilities; (3) increase in reproducibility of studies by standardization of MUP sampling [i.e. positions and depths of needle insertions (Podnar et al., 1999) level of muscle activity (Podnar and Vodusek, 1999)], and of MUP editing (i.e. duration measurement, MUP deletion); (4) improvement in quality of normative data by using standardized techniques on larger and better selected groups of controls. One possibility to increase diagnostic power of MUP analysis is also by increase in size of MUP sample obtained from individual muscle (Engstrom and Olney, 1992; Podnar and Mrkaic, 2003). In the present study, with an increase in MUP sample size from 20 to 40, we demonstrated a signicant narrowing of mean values and outlier reference limits in controls, and in patients with neuropathic lesions for all evaluated MUP parameters (Table 1, Fig. 1). This, most importantly, led to a signicant rise in the sensitivity of MUP analysis as MUP sample size increased (Table 1, Fig. 2). As previously discussed, these results were not surprising, and only reected a basic statistical theorem stating that variance of samples is reduced by increasing their size (Podnar and Mrkaic, 2003). The observed narrowing of reference limits, and the increase in sensitivity in the present study were, however, much smaller than reported in our previous study (compare respectively Figs. 1 and 2 from this study with Figs. 1 and 3 from our previous study) (Podnar and Mrkaic, 2003). Again, as previously explained, this result was also expected, it being due to a difference in MUP sampling between the two studies. In the present study sampling from individual EAS muscles was employed, while control and patient MUP pools were sampled in our previous study (Podnar and Mrkaic, 2003). Although sampling of larger number of MUPs from individual muscles (examinees) reduces intrasubject variability, it does not reduce inter-subject variability. Such sampling is, as a consequence, not random. In contrast, both intra-subject and inter-subject variability are reduced by increasing the size of MUP samples from

Fig. 1. Effect of the size of the motor unit potential (MUP) sample (X-axis) on the 5th (circles) and 95th percentile limits (squares) for mean values (Yaxis) of controls (open symbols) and patients with cauda equina or conus medullaris lesions (solid symbols). Results for MUP area are presented, but similar results were also obtained for other MUP parameters, and for outliers (see Table 1). Lower 95th percentile limit for sample size of 5 MUPs compared to sample sizes of 10 and 15 MUPs observed in patients is coincidental, and was not observed with other MUP parameters. At sample sizes of more than 30 MUPs note saturation in narrowing of limits. [Compare with Fig. 1 in our previous study (Podnar and Mrkaic, 2003).]

number of turns and number of phases), while for all other MUP parameters the sensitivities for outliers were higher. The relations between MUP parameters changed somewhat over the range of sample sizes (Table 1, Fig. 2). When, at a sample size of 40, only typical neuropathic changes (MUP parameters above normative limits) were assessed, the highest sensitivity was found for MUP duration (41%), followed by MUP area (36%), size index (33%), spike duration (31%), number of turns (27%), thickness (26%), amplitude (24%), and the number of phases (20%). However, when MUP parameters below normative limits were added, the ranking of MUP parameters changed to MUP thickness (49%), followed by size index (44%), area and duration (43%), spike duration (36%), number of turns (30%), amplitude (29%), and number of phases (20%) (Table 1, Fig. 2). For sample sizes of 10, 20, 30 and 40 MUPs the cumulative sensitivities of the combined mean values and outlier criteria for 3 MUP

Fig. 2. Effect of the size of the motor unit potential (MUP) sample (X-axis) on the sensitivity of mean values criterion (Y-axis) to distinguish between MUP samples from controls and patients for 8 evaluated MUP parameters. Similar results were also obtained for outliers. [Compare with Fig. 3 in our previous study (Podnar and Mrkaic, 2003).]

S. Podnar / Clinical Neurophysiology 115 (2004) 16831688

1687

common MUP pools, which makes such sampling completely random (Podnar and Mrkaic, 2003). The present study, in addition to valid normative data for mean values and outliers of MUP sample sizes of 10, 20, 30, and 40 MUPs, thus provided a valid estimate of the sensitivity of MUP analysis at different sizes of MUP samples (Table 1). In spite of the limitations discussed above, a signicant increase in sensitivity was demonstrated, often even with quite small changes in reference limits. This can probably be explained by contribution of both control and patient reference limits to increase in sensitivity, and by progressively smaller changes required to obtain a certain raise in sensitivity. When both mean values and outlier criteria of 3 MUP parameters (MUP area, number of turns and duration) that proved necessary in quantitative MUP analysis (Podnar and Mrkaic, 2002) were applied, their cumulative sensitivity increased from 44 to 67% for 20 and 40 MUPs, respectively. This increase was somewhat larger when all 8 MUP parameters evaluated in the present study were applied (from 51 to 77% for 20 and 40 MUPs, respectively). On the other side, with sample sizes below 20 MUPs sensitivity of MUP analysis also rapidly declined (for 3 MUP parameters from 44 to 26% for 20 and 10 MUPs, respectively) as demonstrated also previously (Engstrom and Olney, 1992). Based on the present and formerly published data (Engstrom and Olney, 1992; Podnar and Mrkaic, 2003), we believe that sampling 20 MUPs is a minimum requirement for exclusion of muscle pathology with some certainty. In contrast, using outlier criteria only few abnormal MUPs need to be obtained to demonstrate pathology (Stalberg et al., 1994). For a sample size of 20 MUPs normative limits obtained in the present study were similar to our previously published normative data (Podnar et al., 2000, 2002). However, the sensitivity of MUP analysis for both statistics (mean values and outliers) of individual MUP parameters and of their combinations was slightly lower in the present study compared to our previous results (Podnar and Vodusek, 2001; Podnar et al., 2002), which was probably due to a bias towards less affected patients (muscles) in the present study. It is more difcult and often even impossible to sample a large number of MUPs (at least 35 in the present study) in severely affected patients. However, excluding these patients from the present study seems appropriate, because such extended MUP sampling might be necessary only in patients with slight and equivocal affection (Stalberg et al., 1994). In the present, and our previous studies (Podnar and Mrkaic, 2003; Podnar et al., 2000, 2002), in spite of a well recognized non-Gaussian distribution of MUP amplitudes in individual muscles this MUP parameter was presented by parametric statistics (mean, SD). This was done because whatever the distribution of the original samples, distribution of their mean values is Gaussian. In clinical practice, continuation of MUP analysis after 20 MUPs have been sampled would be needed only in patients with a suspected neuropathic or myopathic disorder, and normal results at that point. The question

then is, whether to sample another 20 MUPs from the same muscle (and thus increase the sample size from 20 40 MUPs), or to sample 20 MUPs from another muscle affected by the same pathologic condition. With a sensitivity of 44% at 20 MUPs from the EAS muscle found in the present study, 56% (100 2 44 56%) of patients with a cauda equina or conus medullaris lesion would have normal quantitative EMG ndings after sampling of 20 MUPs from the rst EAS muscle examined. In our previous study, the sensitivity in the rst subcutaneous EAS muscle was 66%, but dropped to 46% for the subcutaneous EAS muscle contralateral to the normal subcutaneous EAS muscle (Podnar, 2003). Assuming a similar drop in sensitivity here, the sensitivity in the second EAS muscle would be 31% instead of 44% (44 46/66 31%). Examination of the second EAS muscle would thus result in an increase in sensitivity of 17% (56 31 17%), from 44 to 61%. This is 6% less than gain in sensitivity obtained by increasing the MUP sample size from 20 to 40 from an individual EAS muscle in the present study. Although based on extrapolation our results seem to suggest a somewhat greater gain in sensitivity of MUP analysis by increasing the size of the MUP sample from 20 to 40 in the original EAS muscle, than by examination of 20 MUPs from the contralateral subcutaneous EAS muscle. For the EAS muscle our results thus demonstrated a slightly greater increase in the sensitivity of MUP analysis by narrowing of the normative limits due to an increase in the MUP sample size, than by variation in the severity of neuropathic changes between the two sides caused by a cauda equina or conus medullaris lesion. However, other possible information not provided by increase in MUP sample size, but obtained by examination of the contralateral subcutaneous EAS muscle (i.e. spontaneous denervation activity in acute situation) should also be considered when decision about further strategy of EMG examination is made. In EAS muscles of patients with chronic cauda equina or conus medullaris lesions spontaneous denervation activity is usually not detected, which favors increase in MUP samples. Although test retest evaluations were not performed in the present study, we expect that increase in sample size also increases reproducibility of MUP sampling. In this situation, inter-individual variability does not affect results, which should reect in even more pronounced decrease of variance, similar to decrease observed in our previous study (Podnar and Mrkaic, 2003). This is particularly important for sensitivity of follow-up studies. In the present study several larger MUP samples were obtained from the EAS muscles of both sides. This might reduce validity of our results, because in patients with cauda equina lesions different severity of neuropathic lesions in the contralateral EAS muscles is common. However, as we always included MUPs from the left EAS muscles rst, and because in a group as a whole EAS muscles from both sides were similarly affected, we believe that adding MUP samples from the right EAS muscles did not introduce

1688

S. Podnar / Clinical Neurophysiology 115 (2004) 16831688

a systematic error, which would importantly distort our results. However, because of high frequency of asymmetric and unilateral cauda equina lesions we would recommend to sample 40 MUPs from one side (20 MUPs from the subcutaneous, and 20 MUPs from the deeper EAS muscles) in this patient population. MUPs from the subcutaneous and the deeper EAS muscles can be evaluated together, as we found only minor difference in MUP duration (Podnar and Vodusek, 2000), which, however, did not require calculation of separate normative data (Podnar et al., 2000). The relation between the gain in sensitivity obtained by an increase in sample size and that obtained by examination of additional muscles is, however, expected to vary in different muscles. The EAS is a special muscle, which might cause differences between our results, and results obtained in a typical limb skeletal muscle. Furthermore, ndings in different clinical situations are also expected to vary. More uniform neuropathic lesions are expected to favor an increase in MUP sample size, while more heterogeneous neuropathic lesions are expected to favor assessment of additional muscles. In cases where only one muscle (with quantitative EMG normative data) affected by a particular pathology is available, the sensitivity of MUP analysis can be improved solely by increasing the MUP sample size in that muscle. However, pathological ndings (e.g. spontaneous denervation activity, clear neurogenic changes) in other relevant muscles may also add valuable diagnostic information. In conclusion, our present study demonstrated a signicant increase in the sensitivity of MUP analysis in the EAS muscles of patients with cauda equina lesions by enlargement of the MUP sample size from 20 to 40 MUPs, which might be larger than obtained by sampling 20 MUPs from the contralateral EAS muscle.

Jones for language review. The study was supported by the Ministry of Education, Science and Technology of the Republic of Slovenia, Grant No. J3 7899.

References
Engstrom JW, Olney RK. Quantitative motor unit analysis: the effect of sample size. Muscle Nerve 1992;15:27781. Nandedkar SD, Barkhaus PE, Sanders DB, Stalberg EV. Analysis of amplitude and area of concentric needle EMG motor unit action potentials. Electroencephalogr Clin Neurophysiol 1988;69:561 7. Podnar S. Electromyography of the anal sphincter: which muscle to examine? Muscle Nerve 2003;28:3779. Podnar S, Fowler CJ. Sphincter electromyography in diagnosis of multiple system atrophy: technical issues. Muscle Nerve 2004;29:1516. Podnar S, Mrkaic M. Predictive power of motor unit potential parameters in anal sphincter electromyography. Muscle Nerve 2002;26:389 94. Podnar S, Mrkaic M. Size of motor unit potential sample. Muscle Nerve 2003;27:196201. Podnar S, Vodusek DB. Standardisation of anal sphincter EMG: high and low threshold motor units. Clin Neurophysiol 1999;110:1488 91. Podnar S, Vodusek DB. Standardization of anal sphincter electromyography: uniformity of the muscle. Muscle Nerve 2000;23:122 5. Podnar S, Vodusek DB. Standardization of anal sphincter electromyography: utility of motor unit potential parameters. Muscle Nerve 2001; 24:94651. Podnar S, Rodi Z, Lukanovic A, Trsinar B, Vodusek DB. Standardization of anal sphincter EMG: technique of needle examination. Muscle Nerve 1999;22:4003. Podnar S, Vodusek DB, Stalberg E. Standardization of anal sphincter electromyography: normative data. Clin Neurophysiol 2000;111: 22007. Podnar S, Vodusek DB, Stalberg E. Comparison of quantitative techniques in anal sphincter electromyography. Muscle Nerve 2002;25:8392. Sonoo M, Stalberg E. The ability of MUP parameters to discriminate between normal and neurogenic MUPs in concentric EMG: analysis of the MUP thickness and the proposal of size index. Electroencephalogr Clin Neurophysiol 1993;89:291303. Stalberg E, Bischoff C, Falck B. Outliers, a way to detect abnormality in quantitative EMG. Muscle Nerve 1994;17:3929. Stalberg E, Falck B, Sonoo M, Stalberg S, Astrom M. Multi-MUP EMG analysisa two year experience in daily clinical work. Electroencephalogr Clin Neurophysiol 1995;97:14554.

Acknowledgements The author thanks Prof. David B. Vodusek, and Prof. Janez Zidar for review of the manuscript and Dr Dianne