The physiological basis of pelvic floor exercises in the treatment of stress urinary incontinence


Fifty years after the introduction of pelvic floor exercises by Kegel1046, there is a renewed interest in their use to treat stress urinary incontinence in women. There are two important reasons for this: doubts about the apparently high cure rate with surgery2,3 and the need for an effective treatment which can be used commonly outside the hospital. Pelvic floor exercises may be this treatment. Complications are rare4, expensive urodynamic testing before pelvic floor exercises is probably not necessary5,6 and the woman herself undertakes the treatment in the privacy of her home. It may benefit women with additional urinary symptoms6,8. There is evidence that with pelvic floor exercises some women are cured, thus avoiding surgery9,11. For women who have had an operation, pelvic floor exercises could benefit residual urinary symptoms9,12. Some authors feel that pre-operative pelvic floor exercises may improve the outcome of surgery13.

However, there are problems with pelvic floor exercises. Improvement is more common than cure. Treatment in hospital with intensive physiotherapy and biofeedback14 is expensive. It is not known if pelvic floor exercises are useful only in genuine stress incontinence, or whether they are beneficial also in incontinence associated with a low-pressure urethra. Various investigators have different exercise programmes13–16 and so the International Continence Society could make no recommendations regarding the performance of pelvic floor exercises17. Similar biofeedback devices which are highly effective in one setting18 are ineffective in another7. In a direct comparison surgery for stress incontinence was better than pelvic floor exercises in the short term9. A summary of the long term results of pelvic floor exercises showed an average cure rate of 50%14. The results of later research showed various long term effects: a high relapse rate19, a low relapse rate20, or maintenance of the level of continence achieved by the pelvic floor exercises21.

Although pelvic floor exercises are effective in many women, our understanding of the mechanisms of this treatment is incomplete. This paper is an analytical review of the literature on the physiological mechanisms of pelvic floor exercises, including 220 papers, 85% of which were published in the last decade.

Clinical background

The aim of pelvic floor exercises in stress urinary incontinence is to achieve continence using the striated muscle of the pelvic floor. Women with stress incontinence are continent at rest but lose urine on straining. Normally, additional forces must be generated when straining to prevent leakage of urine from the urethra. The periurethral bundles of the levator ani muscles have been identified as the primary continence mechanism during straining22,23. X-ray video recordings have established that the movements of muscular elements in the pelvis during straining are the same in both continent and incontinent women24. This indicates that the muscular occlusive mechanism of the urethra during straining is present in women with stress incontinence, but is inefficient.

Skeletal muscle in adults contains two types of fibre: type I (slow twitch) fibres, capable of longer but quite weak contractions, and type II (fast twitch) fibres, among which the IIB subgroup are capable of very short but powerful contractions. The presence of both main types of fibre in the levator and muscles has been confirmed histochemically using different staining technique25–31. There has been no study so far to specifically investigate the subpopulations A, B and C of type II skeletal muscle of levator ani, seen in other muscles32–33, although two reports acknowledge the presence of these subpopulations30,34.

The anterior part of the levator ani is the puborectalis muscle, also referred to as the pubovaginalis muscle35. Medial bundles of the puborectalis (the fibres of Luschka) terminate at their upper end on the posterior surface of the pubic bones at the level of the urethra; yet they do not penetrate the urethral wall, an observation from cadaveric dissection25,36 confirmed in vivo by magnetic resonance imaging37–39. These fibres exert an indirect external occlusive effect on the urethral lumen. This can be achieved at three degrees of intensity:

  • 1Involuntary, at rest (which represents continuous tonic low-level activation via type I (fibres).
  • 2Reflex, on straining (type II fibres).
  • 3Voluntary (probably both type I and type II fibres).

Two distinct aspects of this classification should be emphasised: a specialised reflex function of the type II fibres in compressing the urethral wall during training27,29, when a quick contraction is needed; and the fact that the levator ani muscles are constantly subjected to contractions. Basal tonic activity was confirmed in a number of studies40–43, and was measured electromyographically; it is approximately 150 μV43. In one study perineal contraction during coughing produced a pressure increase within the urethra of 10 to 20 mm Hg, while voluntary contraction resulted in a similar increase of about 20 mm Hg42, in accordance with our classification. The urethral pressure on straining is closely associated with the electrical activity in the levator ani43.

The levator ani belongs to a group of few skeletal muscles which are subjected to prolonged work; other examples are the external anal sphincter, the diaphragm in the region of the cardia, and some laryngeal muscles. Other muscles respond to such conditions with a decrease in their type II fibre content44,45. Koelbl et al.29 estimated the puborectalis type II content to be 19%; however, in this study the sampling was done remotely from the urethra. It should be kept in mind that different regions of the pelvic floor25, and even parts of the same muscle26 can have different compositions. A quantitative estimation of periurethral fast-twitch fibre content in post-mortem specimens showed that they account for only 4% of the fibre population26, a very small proportion when compared with other muscles46, and a logical adaptation to low-level contractions. Since periurethral type II fibres have larger diameters (mean 60 pm) than type I (mean 45 μm)26, their contribution to the bulk of the normal muscle at the level of the urethra is slightly higher than 4%. Interestingly, a diminished number of fast-twitch fibres in the levator ani muscles was consistently reported in stress incontinence28,29. This prompted us to seek the origin of stress incontinence in muscular overuse47. More pronounced regressive changes and altered composition were observed in samples obtained from the right puborectalis muscle compared with the left in 10 parous women in the study of Fischer et al.30, but these authors did not mention the route of delivery of their women and the possible adverse effect of episiotomy cannot be assessed. Furthermore, the morphology of the medial bundles of the levator ani muscle was found to be reasonably30 or closely48 associated with the outcome of surgery for stress incontinence. There has been no study which associates the outcome of pelvic floor exercises with the morphology of the levator ani muscles. Ultrasound may be important as an indirect measure of the function of levator ani. Two reports came from one institution and referred to the overall thickness of the pelvic floor (mean 9.4 mm)49,50.

Surgical observations on the anterior paravaginal fascial defects seen in women with stress incontinence are also important. The initial report by Richardson et51 was confirmed by other authors52, and the high cure rates of stress incontinence with fascial repair underline the importance of the attachment of the upper puborectalis muscle. Indeed, if the vagina and the paravaginal tissue are torn from the stabilising arcus tendineus fasciae pelvis, the effect of the puborectalis on the urethra may be diminished. Altered configuration of musculofascial attachments at the level of the proximal urethra was shown in early reports of magnetic resonance imaging exploring the lower urinary tract in stress incontinence37,38. Vaginography has been proposed to investigate the status of the fascial structures related to the levator ani muscles53.

In some women with stress incontinence both histological studies30,31 and magnetic resonance imaging37,39 have shown differences in the concentration of skeletal muscle fibres between the left and right sides of the levator ani muscles. This finding may have clinical significance, for we do not know whether compression of the urethra should be symmetrical to assure continence.

The prerequisite for proper contraction in skeletal muscle is the integrity of its motor units (individual motor neurones and the muscle fibres they innervate), and for pelvic floor exercises to be effective significant denervation of the levator ani should not have occured. The available evidence suggests that in stress incontinence denervation is only partial54,58. As the puborectalis muscle is situated in the immediate proximity of the vagina and is thus stretched during childbirth, vaginal delivery may be a predisposing factor to denervation of the puborectalis. The average limit for elongation of the nerves of the pelvic floor muscles before direct damage occurs is thought to be 15% (range 6%-22%)59. Two reports from independent centres, comparing pudendal nerve terminal motor latencies before and after childbirth, showed that denervation occurs during delivery60,61 while throughout pregnancy the nerve's conduction is minimally affected61. There are several possible mechanisms of denervation58,59: denervation at the neuromuscular junction, overstretching of the nerve trunk, or damage to the lumbo-sacral plexus. Although in some women partial denervation is due to a temporary pressure effect62, in others it persists for years63. Compression of the pudendal nerve inside the pudendal canal can also be responsible for stress incontinence: complete relief of symptoms occurred in six women and improvement in another five out of twelve women with stress incontinence after surgical opening of the Of note, in the study of Sultan et al.60 neurological damage to the pudendal nerve after vaginal delivery was more pronounced on the left side. Electromyographic evaluation of the levator ani before therapy has been considered useful in urinary incontinence65 and other pelvic floor disorders66.

The consequences of partial muscular denervation include degeneration and atrophy of some fibres, re-innervation and compensatory hypertrophy of others. Local fibre degeneration and atrophy in the levator ani muscles have been reported 30,31,67. Allen et al.58 found that 80% of 66 women had re-innervation of the pelvic floor after vaginal delivery, as measured by the increased duration of individual motor unit potentials with concentric needle electromyography. Gilpin et al.28 found puborectalis fibre hypertrophy in stress incontinence, the diameter of type 1 fibres being greater than type II. Morley et al.34 reported hypertrophy of both type I and type II fibres in stress incontinence, but especially type II. These findings suggest that the pattern of denervation may vary, and that the re-innervated and hypertrophied type II fibres are not functionally equal to normal fibres. In another study magnetic resonance imaging demonstrated fatty degeneration and partial replacement of striated muscle by connective tissue in 16 of 24 women with stress incontinence39.

The pelvic floor muscles68–70 and intramuscular connective tissue70 contain intracellular oestrogen and progesterone receptors69 which are also present in the surrounding connective tissue68,71, indicating that the puborectalis muscles and their fascial neighbourghood are targets for steroid hormones. More pronounced degenerative changes in the levator ani were observed in postmenopausal than in premenopausal women with stress incontinence67. The number and diameter of both the type I and type II fibres were shown to decrease with increasing age29, as did also the puborectalis glycogen content72. A meta-analysis of the efficacy of oestrogen therapy for urinary incontinence in postmenopausal women found a significant beneficial effect for women with stress incontinence73. However, no data are available on oestrogen administration in per-and postmenopausal women attending programmes of pelvic floor exercises.

Action of pelvic floor exercises

Although there have been a number of reviews of pelvic floor exercises in the treatment of stress urinary incontinence13–16, none have examined the effect of these exercises on the pelvic muscles. The classical explanation was restoration of tone and muscular function1, a view still held13,14. Since there are no studies of the effects of pelvic floor exercises on the morphology of the pelvic muscles, the produced changes are uncertain. Some authors ascribe to pelvic floor exercises some effects of strength training, like increase in total number of activated and effective motor units and muscular hypertrophy74. A comment on the mechanism of action in a handbook on pelvic floor exercises states that these exercises may increase muscular strength and endurance of the pelvic floor, or may increase the reflex action of these muscles through fast-twitch fibres recruitment75. We believe that physiologically pelvic floor exercises should have two main goals: stimulation of existing fast-twitch fibres; and transition of the predominant type I fibre into type II.

Physiological aspects

Skeletal muscle shows an inherent phenotypic plasticity that provides the possibility for it to be significantly remodelled. In an overburdened or abnormal muscle, transition from type II to type I fibres takes place45,76, with remodelling of isoforms of contractile proteins77. This transition affects the metabolism of the muscle, type I fibres undergoing aerobic metabolism, type II B fibres anaerobic metabolism.

There are few biochemical studies on the pelvic muscles in women with urinary incontinence. We have estimated glycogenolysis in samples of puborectalis muscle obtained from women with stress incontinence undergoing posterior colporrhaphy72. In obtaining these specimens we followed anatomical descriptions very closely78. In 22 of 28 samples of puborectalis muscle we found significant glycogenolysis with the activation of hydrolytic decomposition of glycogen. This observation was suggestive of intracellular acidosis, being a possible biochemical background for fibre type transition72. In animals, elevated intracellular hydrogen ion concentrations have been shown to adversely affect actomyosin interaction79 and tension development80 during contraction to a greater degree in fast-twitch fibres than in slow-twitch fibres.

When some fibres of the puborectalis muscle become dysfunctional as a result of denervation or detachment from the endopelvic fascia the remainder are subjected to extra work47. If the demand exceeds the physiological adaptative capacities of human muscle, preferential damage to type II fibres occurs81. This will exacerbate the effects of the transformation of type II to type I fibres. In animals, the process of muscular regeneration after excessive work will also lead to the reconstitution of type I fibres with aerobic metabolism82.

In human beings different programmes of regular exercise induce different changes within skeletal muscle. Endurance training increases aerobic glycolysis in muscle, reduces fatiguability and increases the amount of type I fibres83, while sprint training triggers the slow-to-fast-twitch transition83. Whether strength training prompts a transition between fibre pools is not clear83. Yet, sprinters and throwers undertaking long term explosive anaerobic training may exhibit in their muscles a proportion of type I fibres as low as 20%24, and type I to II A transitions have been demonstrated after strenuous exercises in humans85. The length of training seems important. There is a suggestion that adrenergic β2-receptor stimulation is helpful in accomplishing such transition86.

This research into the basic science of stress incontinence and skeletal muscle physiology implies that pelvic floor exercises should not be of the endurance type. In women, there is an inverse relationship between the concentration of fast-twitch fibres and muscle power87, suggesting that in pelvic floor exercises the induced contractions should be short and repeated, but not necessarily powerful. At least some failures of pelvic floor exercises can be ascribed to unsuitable endurance training.

Skeletal muscle performs work in a similar pattern. There are two phases during muscular contraction: the isometric phase, where muscle tension increases without change in length; and the isotonic phase, where muscle tension is static with decreasing lenght. The state of basal low-level contraction in the levator ani at rest is not likely to change this pattern significantly. It is thought that slow-twitch fibres are capable of developing similar peak power of contraction as that of fast-twitch fibres (200–250 kN. m-2)88. An important difference is in the sensitivity to ionised calcium, which is seven times greater in fast-than in slow-twitch fibres88. This partly explains the velocity of contraction in type II fibres. Power of contraction has many contributing factors, including the initial length of the given muscle. Recent research has shown that the cross-sectional area of each type of fibre is an important determinant of the power of muscular contraction89.

Both the initial length and the cross-sectional area of fibres of the puborectalis muscle are altered in women with stress incontinence. Detachment of the anterior vaginal fascia from the arcus tendineus fasciae pelvis will shorten the anterior portion of the puborectalis muscle. The transition from type II to type I fibres will alter the diameter of the muscle fibres, type I fibres being 25% smaller than type II fibres. Both these mechanisms result in decrease in the external force compressing the urethra.

It should also be noted that voluntary contraction of the pelvic floor is an action distinct from Valsalva's manoeuvre and from abdominal straining with bearing down, the difference being demonstrated with dynamic magnetic resonance imaging90. Many researchers have emphasised the selection of the correct muscle groups in pelvic floor exercise6,10,91, that is, the need to contract the muscles of the pelvic floor and the avoidance of abdominal straining with bearing down. Pelvic floor exercises should concentrate on the anterior part of the pelvic floor, not the posterior6. Objective monitoring of the squeesing technique has been postulated10.


These considerations of the pathophysiology of the skeletal muscle of the pelvic floor in stress urinary incontinence indicate that the target for pelvic floor exercises should be the type II B fibres of the puborectalis muscles. These fibres are physiologically adapted for quick twitches and result in brief, strong contractions, required for both reflex occlusion of the urethra during straining and for voluntary retention of urine. Their deficiency is an important mechanism in the pathogenesis of stress incontinence. Since transition between the types of fibres occurs, the predominant population of type I fibres can be the source of type II fibres with pelvic floor exercises, which should promote such a transformation.

Future studies should evaluate subpopulations A, B and C of the type II fibres of the puborectalis muscle in health and disease. Our findings of acidosis in the medial bundles of the levator ani muscles in stress incontinence need confirmation by other authors, perhaps using different techniques. We need to develop methods of evaluating clinically the morphological and electromyographic abnormalities in the levator ani muscles, in order to select women for pelvic floor exercise14. Ultrasound may be a useful clinical investigation because it is generally available, whereas magnetic resonance imaging is likely to be confined to research, owing to its expense. Assessment of the levator ani should be on both sides of the body, and the anatomical site of the assessment should be annotated precisely. We should perform randomised trials to investigate the effects of different frequencies and intensities of contractions of the levator ani in programmes of pelvic floor exercises.


The authors would like to thank Dr M. Zendzian-Piotrowska, Department of Physiology, Białystok Medical University, for her help in the review of the literature.