|Home | About | Journals | Submit | Contact Us | Français|
Approximately 11% of U.S. women undergo surgery for pelvic floor dysfunction, including genital organ prolapse and urinary and fecal incontinence. The major risk factor for developing these conditions is giving vaginal birth. Vaginal birth is a remarkable event about which little is known from a biomechanical perspective. We first review the functional anatomy of the female pelvic floor, the normal loads acting on the pelvic floor in activities of daily living, and the functional capacity of the pelvic floor muscles. Computer models show that the stretch ratio in the pelvic floor muscles can reach an extraordinary 3.26 by the end of the second stage of labor. Magnetic resonance images provide evidence that show that the pelvic floor regions experiencing the most stretch are at the greatest risk for injury, especially in forceps deliveries. A conceptual model suggests how these injuries may lead to the most common form of pelvic organ prolapse, a cystocele.
Women have been expected to pay a lifelong price for a baby’s birth. Pelvic floor dysfunction, in the form of a type of hernia termed pelvic organ prolapse (including cystocele, rectocele, and/or uterine prolapse) and urinary and fecal incontinence are considered inevitable sequelae for some women who experience injuries during birth. Of the risk factors that have been examined, parity has the strongest association with risk of requiring surgery for pelvic organ prolapse. Compared with nulliparous women, women with one child were 4 times more likely and those with two children were 8.4 times more likely to develop pelvic organ prolapse requiring hospital admission (1). Vaginal delivery is also the single most important risk factor for developing stress urinary incontinence, and the need for treatment increases inexorably with advancing age.
With the burgeoning elderly population, the latent injuries caused in childbirth will affect more and more women. Although surgery for pelvic organ prolapse is effective in restoring anatomy, functional outcomes have not been as satisfactory. Despite the obvious fact that these problems are structural failures, it is only recently that researchers have begun biomechanical analyses to evaluate the mechanisms of normal support and failure.
The term pelvic floor refers to all of the muscles, the connective tissue, and the organs that fill the cavity of the pelvic canal. The pelvic floor muscles form a diaphragm that spans the pelvic cavity. Its fibers have a U shape surrounding the levator hiatus, an arrangement that allows their constant activity to close the hiatus, providing support for the superincumbent pelvic and abdominal organs (Figure 1). The levator ani muscle consists of five parts: the pubovaginal, puboperineal, puboanal portions, which form the pubovisceral complex, and the puborectalis and iliococcygeus muscles. These parts of the levator ani muscle form three different regions of the pelvic floor. Region 1 is the iliococcygeal portion, which forms a relatively flat, horizontal shelf spanning the potential gap from one pelvic sidewall to the other near the sacrum. Region 2 includes the pubovisceral muscle, consisting of fatigue-resistant (type 1) muscle fibers that arise from the pubic bone on either side of the symphysis and attach to the walls of the pelvic organs and the perineal body; these help to close the urogenital hiatus (the opening in the levator ani through which the urethra and vagina pass). Region 3, the puborectal muscle, forms a sling around and behind the rectum, just cephalad to the external anal sphincter.
The normal baseline activity of the levator ani muscle keeps the urogenital hiatus closed against the opening action of intrabdominal pressure. The muscle exerts a resultant force in a ventro-cephalic direction, thereby helping to compress the rectum, vagina and urethra, from back to front, as it equilibrates intraabdominal pressure. To counteract increased hydrostatic pressure and the effect of greater abdominal wall muscle tone, this postural activity exerts a 92% larger vaginal closure force in the upright posture than in the supine posture, in an analogous manner to the erector spinae muscle activity required to maintain upright trunk posture. Indeed, in the upright posture, the pelvic floor muscles are preactivated before rapid arm movements, just as the erector spinae muscles are (2).
A maximum voluntary contraction of the levator ani muscles causes the pubovisceral muscles and the puborectalis muscles to additionally increase the vaginal closure force by 46%, further compressing the rectum, distal vagina and urethra behind the pubic bone distally and against intraabdominal pressure more proximally. It is this compressive force and pressure that one feels when palpating a pelvic floor muscle contraction vaginally. Finally, maximal contraction of the mid and dorsal iliococcygeus muscles (Region 1) elevates the central region of the posterior pelvic floor but likely contributes little to a vaginal measurement of levator strength because these muscles do not act about the vagina.
A popular method of testing pelvic floor muscle strength has involved the use of vaginally placed balloon-type sensors. However, their radial compliance precludes the isometric measurements favored by muscle physiologists. We therefore invented a two-billed instrumented speculum to make the isometric measurements of vaginal closure force outlined in the two preceding paragraphs. Knowing that the physiological cross-sectional area of the Region 2 muscles is approximately 2.8 cm2 (3) and that striated muscle can maximally develop 2.8 Newtons (N) cm−2 parallel with its fibers, we can predict that the right and the left sides of this muscle each develop approximately 5.6 N, or a total force acting to close the vagina of 11 N. This is consistent with the mean (± SD) maximum value of 10 (± 2) N of vaginal closure force we have measured in upright stance (4).
The pelvic floor, viewed cranially, is a hexagonally shaped structure connected to the bony pelvis at the sacrum, ischial spines, and pubic bones. In the upright posture it is loaded by the hydrostatic load due to the height of the fluid abdominal contents beneath the diaphragm, by part of the weight of the genital organs, and by any abdominal pressurization due to abdominal muscle and diaphragm muscle postural tone and/or respiratory activity. In published studies, we have measured hydrostatic pressures of 40 cmH2O (the traditional pressure units in urology and gynecology) in the bladder, just above the pelvic floor, so this is approximately the hydrostatic load carried by the pelvic floor in the upright posture. The area of the average female pelvic floor is approximately 94 cm2, distributed with approximately 53 cm2 located anterior to and 41 cm2 located posterior to the ischial spines (5).
Knowing pelvic floor area and the pressure acting on it, one can estimate the load acting normal to the pelvic floor (calculated as the product of the hydrostatic pressure times the floor area) as approximately 37 N in quiet standing. This load decreases to approximately 19 N in the supine posture. However, during a maximum cough, the strong diaphragmatic contraction causes intravesical pressure to reach approximately 140 cmH2O, thereby placing a peak load of 129 N on the pelvic floor. A small percentage of this load is inertial as downward momentum of the abdominal contents is arrested by the pelvic floor. During another common activity of daily living, straining at stool, co-contractions of the diaphragm and the abdominal wall cause the peak intravesical pressure to typically reach 100 cmH2O over several seconds, thereby exerting a peak load on the pelvic floor of 92 N. Inhibition of the levator ani muscle allows the urogenital hiatus to open for bowel and/or bladder evacuation.
During labor, the pelvic floor undergoes remarkable changes to allow a 3.5 kg baby to be passed from inside to outside of the body. The first stage of labor consists of progressively stronger uterine contractions that help to drive the fetal head through a ripened cervix causing it to dilate and thin. The second stage of labor starts after the cervix is completely dilated and the fetal head begins to touch the posterior pelvic floor. During this approximately 90 min stage, spontaneous uterine contractions occur every 3 min for a duration of 1 min. These contractions increase the uterine pressure near the fetal head from 2.6 to 8.5 kPa (6), engaging the head in the pelvis and helping to push it through the birth canal along what is known as the Curve of Carus. When, during the second stage of labor, the woman is instructed to push during uterine contractions by bearing down via a maximal contraction of her diaphragm, this increases the intrauterine pressure to approximately 19 kPa.
If the average diameter of the molded fetal head is 9 cm, then its cross-sectional area is approximately 63 cm2. The above intrauterine pressures would then apply expulsive forces on the fetal head: 16 N at rest, 54 N during a uterine contraction, and 120 N during a volitional push. In a recently published computer simulation, which takes into account the viscoelastic nature of the pelvic floor muscles, we show that a single volitional push timed to coincide with the height of each uterine contraction is a more efficient pushing style than pushing at other times. If the woman becomes exhausted and can no longer make progress, then an instrumented delivery is indicated. A vacuum device is applied to the fetal head to apply an additional traction force of up to 113 N, typically with up to four pulls, each coinciding with a uterine contraction and push (7). In the cases when obstetric forceps have to be used, the additional traction force can reach 200 N (8).
All women sustain stretching of their pelvic floor during birth, but only some experience injury (Figure 2). Specific features of injury during vaginal birth influence whether a woman develops prolapse later in life. Several factors that can be grouped together as descriptors of difficult vaginal delivery are associated with increased occurrence of prolapse: forceps delivery, a prolonged second stage of labor, and large infant birth weight have all been associated with pelvic organ prolapse. Unfortunately, because of the overlapping nature of these different factors, it is difficult to determine which of them is causal and which of them is associated. Forceps delivery is used when there has been a prolonged second stage of labor, and both of these factors are increased in infants of large size. The role of childbirth in causing damage to the levator ani muscle, which is associated with both vaginal delivery and with pelvic organ prolapse, is probably the mediating mechanism in these injuries.
In one study of 160 primiparous women we found that 32 (20%) exhibited damage to the levator ani muscles on magnetic resonance (MR) scans. None of the 80 nulliparous women serving as controls had injuries, thereby identifying birth as a cause of the type of levator ani muscle injury seen in women with pelvic floor dysfunction. Twenty-nine of these injuries occurred in the pubovisceral muscle; only three injuries occurred in the iliococcygeal portion of the muscle. This was a study originally designed to study stress urinary incontinence: equal numbers of women that developed de novo stress incontinence and women that remained continent after their first birth were recruited. Because the stress-incontinent group were twice as likely to have defects as the continent group, the occurrence of these defects in a group of primiparous women not oversampled for stress incontinence would be somewhat lower than this estimate. However, even if this estimate were halved, it would still indicate that 1 in 10 women delivering their first infant will sustain levator damage. An association has been demonstrated between MR-visible injury and de novo stress urinary incontinence as well as pelvic organ prolapse. In the case of the de novo stress urinary incontinence, use of forceps, anal sphincter laceration, and episiotomy increased the odds ratio for levator muscle injury by 14.7-, 8.1-, and 3.1-fold, respectively (9). Women with levator injuries were 3.5 years older and had a 78-min-longer second stage. These morphological studies identify the role of vaginal birth in the injury and indicate where some of the damage is occurring but do not yield information about the mechanism(s) of injury. In a more recent study, three-dimensional (3D) ultrasound scans of women before and after vaginal birth have confirmed that these types of injury occur during vaginal delivery. There is great variation in the amount of injury, from a few fibers on one side of the levator to the entire pubovisceral muscle, both with and without distortion of surrounding tissues.
There have been several suggestions for why the levator ani muscles might be injured. Information from electrodiagnostic studies has demonstrated that birth causes changes in mean motor unit duration after vaginal birth as well as changes in pudendal terminal motor latency. Abnormal tests have also been found in women with prolapse or stress incontinence. Although the pudendal nerve innervates the voluntary urethral and anal sphincters, it does not innervate the levator ani muscles, which receive their own nerve supply from the sacral plexus. Currently, it is not clear whether the visible levator defects are from neurological or stretch injury.
A geometric model (Figure 3) has suggested that some muscle damage during the second stage of labor may come from overstretching because those parts of the muscle that are stretched the most are those parts that are seen to be injured (10). The pubovisceral (pubococcygeal) muscle, the shortest and most medial levator ani muscle, had the largest tissue strain with a stretch ratio of 3.26. Regions of the iliococcygeus, pubococcygeus, and puborectalis muscles reached maximal stretch ratios of 2.73, 2.50, and 2.28, respectively. These values considerably exceed the maximum stretch ratio of 1.5 tolerated by striated muscle in nonpregnant animal preparations. Tissue stretch ratios were found to be proportional to fetal head size. For example, increasing fetal head diameter by 9% increased medial pubovisceral stretch by the same amount. This analysis revealed that the region of muscle injured most often, the pubovisceral (pubococcygeal) portion, was the portion of the muscle that underwent the greatest degree of stretch, and the second area of observed injury, the iliococcygeal muscle, was the muscle stretched second most. Furthermore, when the portion of the muscle at risk was identified in cross sections cut in the same orientation as axial MR scans, the pattern of predicted injury matched the injury seen in MR scans (Figure 4). Finally, the maximum stretch ratio was found at the end of the second stage of labor when the fetal head crowned (Figure 3). These theoretical findings suggesting stretch-induced injury are supported by studies showing increased insulin-like growth factor-1 splice variants that indicate stretch and overload in women after first vaginal birth.
We do not yet know whether, during a birth, injury occurs to a pubovisceral muscle in its active or its passive state or both. The best predictors of stretch injury in active muscle are the strain or the product of force and strain (work), with strain and work essentially equivalent in their predictive value (11). The best predictor of injury in passive muscle, as measured by the force deficit after one min, is the negative work done on the muscle. When a passive muscle is stretched, its force depends on two factors: the strain rate and the product of the strain times the strain rate. An order-of-magnitude increase in the strain rate can increase the peak force by 25%. Hence, a physician performing an instrumented delivery (vacuum or forceps instrumentation) is probably wise to keep the rate of pelvic muscle stretch as low as possible by delivering the fetal head as slowly as is reasonable. We recognize that injury can also occur when a shoulder is delivered or when the head is delivered occiput-posterior, but the mechanics of these injury mechanisms remain to be investigated.
One might speculate how 85–90% of first time mothers can undergo vaginal birth without the fetal head overstretching and rupturing the U-shaped loop of pubovisceral muscle tissue. One possibility is that the structure that lies in series with this muscle could protect it by stretching more than the muscle itself (12), just as a fusible link protects the wiring harness of an automobile against an electrical short. That structure, the perineal body, located between the vaginal and rectum and comprised of relatively soft connective tissue, has material properties that do appear to change in late pregnancy, but these remain to be quantified.
During pregnancy, hormones affect the biochemical composition of the solid matrix and hydration phases constituting each pelvic floor tissue. Remodeling mechanisms lead to changes in the organization, orientation, and diameter of the collagen fibers as well as the crimp structure of the collagen fibrils, reinforcing each tissue. Such effects can significantly affect the short- and longer-term viscoelastic properties of the vaginal wall, the pubovisceral muscles, and the perineal body, for example. They will largely determine (a) the extent and rate at which these structures can be stretched by an expulsive force acting cyclically on the fetal head, and (b) the resistance to stretch provided by those structures. The more a tissue exhibits creep behavior, the further it will stretch under a constant load. And the more it exhibits relaxation behavior, the more the stress in a tissue will decrease over time when held at a constant length, thereby helping to lower the risk of rupture in the next loading cycle. Were a tissue to exhibit viscoplasticity, it would behave like a solid below a critical level of stress, but above that level, it would flow like a viscous liquid. There is evidence that tensile failure in some soft tissues can be predicted by the product of the stress times the strain extant in the tissue, so mechanisms that lower one or both these variables will reduce the risk of rupture. Pregnancy is known to significantly affect the instantaneous stiffness and relaxation behavior of vaginal tissues in rat (14; Figure 5). However, accurate constitutive laws are lacking for pregnant human pelvic floor tissues, and the effects of pregnancy on injury at any tissue level, and on structural failure, are as yet largely unknown.
One finite-element model of the second stage of vaginal birth has estimated the maximum levator muscle stretch to be 1.6 (15). This estimate is artificially low because the geometric data for the model’s levator muscles were taken from a cadaver. Because a cadaver naturally lacks pelvic muscle tone and its pelvic floor is often stretched during embalming, it normally exhibits an enlarged levator hiatus and this will reduce the calculated stretch ratio. A more recent finite-element model (16), based on the MR data of the levator muscles from a living woman having levator muscle tone, demonstrated a maximum stretch ratio of 3.5, thereby corroborating the maximum stretch ratio predicted by Lien (10, 12; Figure 6). The maximum stretch ratio is expected to depend on a number of factors, including the size of the fetal head relative to that of the maternal pelvis, the degree of molding of the fetal head, perineal descent, and the subpubic arch angle. Hence, the maximum stretch ratio is expected to range widely among individuals.
A geometric model has also been used to predict the stretch ratios in the nerves innervating the levator ani, urethra, and anal sphincter during the second stage of vaginal labor (17). The results showed that the inferior rectal branch exhibited the maximum strain, 35%, and this strain varied by 15% from the scenario with the least perineal descent to that with the most perineal descent. The strain in the perineal nerve branch innervating the anal sphincter reached 33%, whereas the branches innervating the posterior labia and urethral sphincter reached values of 15% and 13%, respectively. It was concluded that during the second stage (a) nerves innervating the anal sphincter are stretched beyond the 15% strain threshold known to cause permanent damage in nonpregnant appendicular nerve, and (b) the degree of perineal descent is shown to influence pudendal nerve strain.
We developed a simplified biomechanical model (Figure 7) of anterior vaginal wall prolapse (cystocele) (18). The anterior vaginal wall is fixed at its lower margin to a structure called the perineal membrane and suspended at its upper end by two mesenteric like structures: the cardinal and uterosacral ligaments. It rests against the rectum and posterior vaginal wall, which are supported by the pubovisceral portion of the levator ani muscles. Intra-abdominal pressure acts perpendicular to the anterior vaginal wall, which pushes against the rectum and levator ani muscles, tending to displace them backward. If muscle force is sufficient to resist this pressure, then there is not net downward displacement of the anterior vaginal wall. If, however, the muscle force is insufficient to resist this pressure, then the hiatus opens and the suspending cardinal and uterosacral ligaments are not stiff enough to prevent downward displacement of the vaginal wall, then the distal vaginal wall will protrude through the genital hiatus. We then examined the effect on model behavior of an impairment of the levator ani muscle, such as that sustained as a birth-related injury. Simulations revealed that the larger the pubovisceral muscle impairment, the larger the anterior wall prolapse became. A 90% impairment of apical support led to an increase in anterior wall prolapse from 0.3 to 1.9 cm at 60% pubovisceral muscle impairment and from 0.7 to 2.4 cm at 80% pubovisceral muscle impairment. These results suggest that a prolapse can develop as a result of impairment of the muscular and apical supports of the anterior vaginal wall. Among women with prolapse, 55% have major injuries to the pubovisceral muscles compared with normal women, who exhibit 16% injury. These findings have functional consequences in that, on average, we found women with prolapse generate 40% less vaginal closure force during maximal voluntary muscle contraction. It is this impairment that may lead to conditions fostering the onset of a cystocele.
We apologize in advance to authors whose work could not be cited due to space limitations. We gratefully acknowledge the contributions of our many collaborators and the financial support of PHS grants P50 HD044406, P30 AG024824, and R01 HD 38665.
Dr. De Lancey has consulted for American Medical Systems, Inc., regarding the anatomy of pelvic organ prolapse, and both authors have received grant support from Johnson & Johnson, Inc.