|Home | About | Journals | Submit | Contact Us | Français|
We aimed to compare our normative data for quantitative interference pattern (IP) analysis of the anal sphincter to previously published data. In 28 nulliparous women, we performed IP analysis during quantitative concentric needle electromyography (QEMG) of the anal sphincter. At each sampling site, a 500-ms epoch was analyzed. The data were log transformed. Linear regression lines (with 95% confidence intervals) were calculated from the log transformed variables “turns–second” and “amplitude–turn.” These confidence intervals were then transformed back into the original parameters to yield scatterplots with confidence curves. The mean turns–second were 203 (SD 174). The mean amplitude (mcv)–turn was 266 (SD 87). The regression coefficients for the log-transformed variables are constant=1.5, slope=0.3, and resultant cloud of raw data has a convex upper boundary. These appear slightly different than previously published reports, potentially influencing the determination of normal and abnormal studies.
Vaginal delivery is repeatedly implicated as contributing to the development of pelvic floor disorders such as urinary incontinence, fecal incontinence, and pelvic organ prolapse in women [1–3]. Injury to the nerves serving the pelvic muscles is thought to be one of the reasons for the development of these conditions . Because symptoms may not manifest for years, the influence of nerve injury at the time of childbirth has been questioned.
Unfortunately, the current and most commonly used tests to document nerve injury may be insensitive. One of the most commonly used neurophysiologic test for documenting presence or absence of neuropathy in the pelvic floor is a partial nerve conduction study of the distal pudendal nerve, known as pudendal nerve terminal motor latency. However, only when the largest, and most heavily myelinated nerves are lost does the pudendal nerve conduction test become abnormal. Recently, new electromyography (EMG) advances (multiple motor unit action potential analysis and interference pattern (IP) analysis) make it easier to detect subtle nerve injury in postpartum women even when pudendal nerve conduction velocity measures are normal . These, combined with other reports, have prompted the Committee on Clinical Neurophysiology at the International Consultation on Incontinence to suggest utilizing other electrodiagnostic tests, including sacral reflexes and concentric needle EMG . That these quantitative EMG parameters are changed even after uncomplicated, spontaneous (nonoperative) vaginal deliveries in asymptomatic women suggest that these abnormalities may be preclinical markers for future development of pelvic floor disorders.
During needle EMG, a small needle electrode located near muscle fibers detects summated electrical potentials and displays it as a waveform called a motor unit action potential, much like surface electrodes do during an electrocardiogram (Fig. 1). During increased contraction (e.g., to maintain continence), more motor units (a single nerve cell, its axon, and all the skeletal muscle fibers it serves) are called upon, and the resulting action potentials in the EMG signal “interfere” with one another, creating the “interference pattern” (Fig. 2). Although generally assessed in a subjective fashion, computer algorithms can aid in the measurement of objective data from the EMG waveform. The number of motor units, firing frequency of the motor unit, and recruitment of new motor units (as increased force of contraction is required) can be gleaned from the interference pattern.
Unlike most other skeletal muscles, the pelvic floor muscles contract tonically, and techniques for obtaining and analyzing data are challenging. One quantitative concentric needle electromyography (QEMG) technique, IP analysis, is very promising due to its potential ease of use and lack of operator-induced bias.  EMG signals recorded at varying levels of baseline and voluntary contraction can be analyzed. When nerve injury occurs, fewer motor units contribute to the creation of force during muscle contraction creating a “reduced” interference pattern (Fig. 2).
Prior to using this tool for pelvic floor muscles, reliable and discriminating normative data needs to be established. Previous investigators have utilized nulliparous  as well as a combination of nulliparous and parous women  to serve as comparative controls. In order to further study the acute neurophysiologic effects of a first pregnancy and delivery (whether by vaginal or cesarean delivery) on the pelvic floor; we aimed to establish normative data for quantitative interference pattern analysis of the external anal sphincter (EAS) in a group of healthy nulliparous women and compare it to previously published data.
Approval for this investigation was obtained from the Institutional Review Board at Oregon Health and Science University. Between July 2002 and August 2003, 28 healthy nulliparous women without complaints of pelvic floor disorders underwent a standardized evaluation as part of a larger study designed to determine the effect of vaginal delivery to the pelvic floor. These women were recruited by posted flyers from the student health center at Portland State University and from local Planned Parenthood clinics. Examinations were performed by one examiner (WTG) who was not masked to subject group. Each woman was modestly compensated for her participation. Each subject then underwent a battery of comprehensive pelvic floor neurophysiologic examinations which included pudendal and perineal nerve terminal motor latency determinations, sacral reflexes (clitoral–anal, urethral–anal, and bladder–anal), and quantitative concentric needle EMG of the external anal sphincter. Prior to needle EMG, a small amount of topical anesthesia (lidocaine–prilocaine cream) was placed at the anal verge.
One aspect of the neurophysiologic exam will be detailed here. We performed quantitative concentric needle EMG using a Medtronic Keypoint EMG machine (Medtronic Corporation, Minneapolis, MN, USA) equipped with interference pattern analysis software. Filter settings were 5 Hz to 10 kHz; the gain was 100 μV/div to 2 mv/div, and the sweep speed was 10 ms/div.
As described in detail in previously published reports [5, 10], we used a 37-mm concentric needle to sample four to five unique sites (in different motor unit territories) on each side of the EAS. This was accomplished by initially placing one finger in the anus (to gauge the direction of the anal canal) and then by advancing the needle in a direction parallel to the anal canal (perpendicular to the longitudinal direction of the EAS muscle fibers) approximately 0.5 cm for each sample acquisition (Fig. 1).
Subjects were asked to incrementally increase voluntary contraction at each site to maximum in order to recruit more motor units and create a so-called interference pattern. With each change in needle position, the interference pattern was sampled at baseline muscle activity and at the consciously increased muscle activity. Simultaneous force or pressure readings were not measured.
The EMG signal was recorded in a digital format on a Sony TCD-D7 digital audio tape (DAT) recorder (Sony Corporation, Tokyo, Japan; frequency response: 2 Hz to 22 kHz) using the optical digital output interface of the Medtronic Keypoint machine. The original uncompressed digital data was then played through the optical digital input (bypassing any digital–analog conversions) and analyzed using the interference pattern software on the Keypoint system after the completion of the experimental protocol. The time stamp on the DAT recorder was noted with each new site to ensure that the IP analysis, performed from the DAT recording, acquired motor units from different areas in the muscle and prevented the sampling of the same motor units.
During each site sampling, several crisp-sounding 500-ms epochs (coinciding with each incremental change in muscle contraction efforts) were analyzed for the following EMG parameters: number of turns–second (any peak or trough of the signal where amplitude changes by 100 μV); amplitude–turn (change in volts between two turns); number of short segments (parts of EMG signal that have sharp activity (very short rise times); percent activity (percent of time during the epoch that sharp activity occurs); envelope (that peak to trough amplitude exceeded only by 1% of peaks and troughs during epoch). Those rare data points that were recorded as a value of “zero” (i.e., an epoch of little or no crisp EMG activity) were discarded.
SPSS for windows v.11.0 (SPSS Inc, Chicago, IL, USA) was used for all statistical calculations. For the pooled data, means and standard deviations of continuous data were calculated. A Turns–Amplitude scatterplot was then created. In order to create “clouds” of data to be used for future individual subject comparisons, the data were first log (10) transformed. Univariate linear regression was used to calculate a slope and intercept from the transformed data. Linear regression lines (with 95% confidence intervals) were then created from the log-transformed variables “turns–second” and “amplitude–turn”. The variables from the upper and lower confidence limit lines were then transformed back into the original parameters to yield confidence curves to serve as upper and lower boundaries to which future individual scatterplots can be compared.
Because Weidner's  and Podnar's  previously published data exists in different forms, and a direct comparison to our data cannot be made without the raw data that created those reports, a qualitative comparison figure of data curves was created using the following method. We used the slope and intercept values from our log-transformed regression line (utilized to create the normative cloud as described above) as well as from Podnar's  manuscript and transformed them to create separate regression “curves” (the 95% confidence limits of these curves represent the boundaries of the respective “clouds”). Weidner did not separately report the log-transformed regression parameters but published a graph of a “cloud” similar to the one described above. The abscissa and ordinate data points for the data curve (but not the confidence limits) were estimated from the figure and replotted. All data curves were then plotted together on the same graph to reveal similarities and differences in shape and location.
The mean age was 27 years (range 18–38 years), the mean body mass index was 24.5 kg/m2 (range 17–44 kg/m2)and 96% were Caucasian. None of the women reported pelvic floor disorders nor had evidence of pelvic floor abnormalities on the standardized, structured pelvic exam. Further demographic and clinical exam findings are detailed elsewhere . Each study subject provided a mean of 21 (range 5–35) data points (from both the left and right side) to the pool of 490 data points. Due to discomfort (and early termination of the study), two subjects provided only eight and five data points to the pool, whereas four subjects provided more than 30 data points. The resultant quantitative IP parameters are shown in the Table 1. Log transformation for both variables yielded a regression line with the following variables, 0.3 (SE 0.019) for slope and 1.5 (SE.04) for intercept. To determine if the subjects with less than ten and more than 30 data points altered the results, the regression analysis was repeated with these subjects removed. No difference in the log-transformed regression variables was noted, 0.29 (SE 0.02) for slope and 1.5 (SE 0.05) for intercept. The predicted data along those 95% confidence bands were transformed back to the original values, and the resultant 95% confidence curves are shown overlying the original scatterplot in Fig. 3. The regression parameters were then compared to previously published reports in Table 2. The position and shape of the raw data curves (without the confidence limits that create the cloud boundaries) are shown in Fig. 4.
As previous investigators have shown [8, 9, 11], IP analysis is a feasible technique to study the neuromuscular integrity of the tonically contracting striated muscles of the pelvic floor. Using a similar technique as previous investigators, we demonstrate that the raw IP parameters we obtained for the external anal sphincter appear similar to the previous studies that separately reported them for nulliparous controls. However, slight differences in the location and shape of the scatterplots are noted.
Podnar  has suggested that IP analysis may be more sensitive to subtle reinnervation because these changes may be more prominent in “high-threshold motor units” (i.e., those activated at greater force requirements) which are more likely detected by IP analysis. One of the ways that this technique can be utilized to detect subtle changes may be in “clouds analysis” . Cloud analysis essentially relies on scatterplots of normative data, with boundary curves (incorporating 95% of the data points). Data points obtained during IP analysis from any subsequent individual patient or study subject are then compared to these “clouds” or boundary curves. In the particular method described here, the axes are number of turns (essentially representing the number (or density) of motor units contributing to the muscle contraction) and the amplitude of each turn (with larger amplitudes representing larger motor units). When (for example) more than 10% of an individual's IP data points fall outside the cloud, we might diagnose the muscle as neuropathic (abnormal due to injury sustained by the nerve) or myopathic (abnormal due to some intrinsic or acquired abnormality of the myocytes). In general, when the clumps of data are to the upper left side of a “cloud,” we consider it to be evidence of neuropathy. This occurs because with nerve injury, denervated muscle fibers rely on “collateral sprouting” from the adjacent axons of surviving motor units. The surviving motor unit becomes “larger” (generating more detectable electrical energy, i.e., a larger amplitude) by incorporating more muscle fibers served by the same nerve cell. Fewer motor units then contribute to muscle contraction (even if the number or volume of muscle fibers remain unchanged). As a result, the number of turns decreases, but the amplitude of each turn is greater. This pushes the scatterplot above and to the left of the appropriate normative data “cloud.”
In using this technique and in making comparisons to previously published work, there are several important principles to point out. Both amplitude and turns increase as the force of contraction increases as more (and larger) motor units are recruited (Henneman  principle). Generally, amplitude increases with the force of the contraction to maximum, but turns will plateau prior to maximum force contraction. However, it is difficult to reliably measure force of contraction at the anal sphincter because it does not act on a bony lever through a tendinous insertion, and normal individuals have varying abilities to voluntarily produce a strong contraction.
The utility of the clouds technique is that it does not measure nor require a standardized force of muscular contraction. Therefore, the extent to which the muscles were contracted in each of the three studies compared could explain the observed minimal differences of the individual IP parameter values (turns–second, amplitude–turn, etc.) as seen in Table 1. As suggested by Stalberg  though, “clouds analysis” theoretically can be used irrespective of the force of muscle contraction, and as described in the “Materials and methods” section, a cloud analysis in fact plots amplitude versus turns– second at varying levels of contraction.
However, by following the same method as others to generate the shape of the IP “cloud”, we find that our data curve is potentially dissimilar to both Weidner and Podnar. In order to attempt to explore that question statistically, we compared the slope of Podnar's regression equation (whose data included men and women who were both nulliparous and parous) to the slope from our regression equation using a two-sided t-test and found the p-value to be only 0.20 (Table 2). Weidner's curve falls in between these two extremes. So, although they appear visually different, they may in fact be statistically similar.
Despite the assertion that clouds analysis allows us to disregard force measurements in limb muscles , it has previously been shown that the shape of the cloud (concavity or convexity of the cloud boundaries) is dependent on the force of muscle contraction. At low to moderate force levels, the interference pattern becomes fuller due to the sequential recruitment of higher-threshold motor units, and the increased firing rate of motor units already activated. Although, by Henneman's principle, the amplitude of the newly recruited motor units is larger, the turns–amplitude ratio increases only slightly, and the cloud boundary is concave downward. At higher force levels, the mean amplitude of the interference pattern increases much more than the number of turns detectable, and the resultant cloud is concave upward.
It is difficult to determine where on the continuum of force generation a given subject lies simply by viewing the perineum during these studies. Some subjects have very high baseline tone, whereas others have much less activity. The coordination needed to activate the muscles is variably present. Finally, there is no easy way to determine percent maximum effort. Despite our efforts to test a wide range of contraction forces, our study subjects may have been skewed to one or the other end of the force spectrum. It is still possible that our subjects were analyzed at different points along the force curve than either Weidner's or Podnar's. Until more precise tools are developed, it might be prudent to simultaneously measure pressure with anal or vaginal manometers.
Using cloud analysis, the shape and location of the normative cloud affects the discrimination and cutoff between normal and abnormal. Because facial muscles do not move bones through a measurable range of motion and are not easily measured for force generation, they are similar to pelvic floor muscles. However, recently, the same cloud analysis technique was shown to be effective in discriminating between subjects with normal facial muscles, and those that had been treated with botulinum toxin for blepharospasm (inducing neuropathic changes) .
In summary, this study points out potential subtle differences in study subjects and protocol. As we continue to use this test in the evaluation of pelvic floor disorders and how childbirth affects the nerves and muscles of the pelvic floor, we need to ensure that we standardize as much as possible to improve the discrimination between normal and abnormal.
Financial support: NIH K12 HD-01243. Medical Research Foundation of Oregon.
Presented at: 26th Annual Scientific Meeting of the American Urogynecologic Society, September 15–17, 2005. Atlanta, GA, USA.
Conflicts of interest None.
W. Thomas Gregory, Division of Urogynecology and Reconstructive Pelvic Surgery, Department of Obstetrics and Gynecology, Oregon Health and Science University, 3181 SW Sam Jackson Park Road, Mail Code L466 Portland, OR 97239, USA.
Amanda L. Clark, Division of Urogynecology and Reconstructive Pelvic Surgery, Department of Obstetrics and Gynecology, Oregon Health and Science University, 3181 SW Sam Jackson Park Road, Mail Code L466 Portland, OR 97239, USA.
Kimberly Simmons, Division of Urogynecology and Reconstructive Pelvic Surgery, Department of Obstetrics and Gynecology, Oregon Health and Science University, 3181 SW Sam Jackson Park Road, Mail Code L466 Portland, OR 97239, USA.
Jau-Shin Lou, Department of Neurology, Oregon Health and Science University, Portland, OR, USA.