When a disease is studied only in patients in a hospital, the symptomatic and morphologic features seen in those hospitalized patients may not accurately reflect the wild type disease which is prevalent in the population as a whole. This may result in an inaccurate picture of the disease. This is called 'Berkson's fallacy' (he described it but did not commit it),[1, 2] and it may result in an increased chance of a positive or negative correlation being found where no or weak correlation actually exists. Since a description by Rokitansky in 1869, [3] endometriosis has usually required surgery in a hospital or other health care facility for diagnosis. Thus, the risk has always silently existed that endometriosis could be spuriously linked to symptoms or other disease states due to the effects of Berkson's fallacy acting unrecognized in hospitalized patients. Modern hindsight suggests that the decades-old effects of Berkson's fallacy have led to misinterpretation of the symptoms of infertility and pain associated with endometriosis. This in turn has led to confusion and ineffective treatment which are the unfortunate modern hallmarks of the disease.

Pregnancy as prophylaxis

The historical origin of the notion that endometriosis is positively associated with infertility began in an earlier, simpler time when it was thought that the pregnancy rate in married women should approach 100%. Early studies [4, 5, 6] of surgically diagnosed patients with obliteration of the cul-de-sac, large ovarian chocolate cysts (most of which were non-endometriotic), and pelvic adhesions found a pregnancy rate of only 40% in married women [5] without accounting for other important fertility factors, such as the male factor, age, and duration of conception attempts. Such patients today would be classified as Stage III or IV by the revised American Fertility Society (rAFS) classification system for endometriosis [7]. However, even in a modern large referral practice devoted almost exclusively to patients with endometriosis, rAFS Stage III and IV patients constitute only about 18% of surgically diagnosed patients [8]. Therefore, these early studies inaccurately characterized the morphological spectrum of endometriosis toward the severe end. Then, as now, the true prevalence of endometriosis and the symptom of pregnancy among the general population of women was unknown. Without these important prevalence figures, the observation of a 60% rate of infertility [5] among patients with chocolate cysts of the ovary has no relevance.

Observing this fertility rate of about 40%, Meigs [6] postulated that repeated pregnancies delayed the appearance of endometriomata of the ovaries, thus introducing the concept of pregnancy as prophylaxis or therapy which was later repeated by others [9]. He did not pause to wonder why parous women in his own study were not protected against the disease. So convinced was Meigs that early and frequent childbearing would help prevent endometriosis, that in later years he urged parents to support their newly married children so the young couples could reproduce early without economic pressures [10]. He was especially interested in educated couples having children and his views generated a storm of controversy at the time.[11]

In the modern era, we can easily identify the causes and effects of Berkson's fallacy on such early mischaracterization of disease. For example, the actual visual spectrum of disease is far greater than that found by Sampson and Meigs, and we now realize that most women do not have Stage III or IV disease. Also, the ovaries are not as commonly involved by endometriosis as suggested by these earlier studies. Unfortunately, the history of study of endometriosis is marked by authors simply repeating what has been published in the past, even if their own findings are contradictory.

Haydon also favored the concept that endometriosis is more common in infertile women.[12] Support for his conclusion came from his retrospective study of 569 cases. He found that 262 patients were parous, whereas 302 were not. He concluded that the incidence of relative infertility was therefore 53%. However, his numbers were not corrected for patients not attempting pregnancy. When 122 of his unmarried patients are eliminated from consideration, the birth rate in his population was actually 58%. However, follow-up was available in only 291 of his patients, so the actual eventual birth rate may have been higher. This important distinction regarding his work has been overlooked to this day.[13]

Counsellor and Crenshaw found in a retrospective study of patients with endometriosis that 56.2% of married couples were sterile.[14] They acknowledged the possibility that patients presenting for surgery for infertility constituted a biased population.

Pregnancy as 'treatment'

From observations of decreased fertility in patients with endometriosis, authors began to make unsupported leaps of faith. Beecham believed pregnancy to be a cure for endometriosis and dogmatically stated: 'Nature (since the beginning of time) has employed an efficient prophylactic and curative measure for endometriosis, i.e. pregnancy.' No reference or evidence was offered to support this speculation.[15]

Modern studies have found infertile patients to be more likely to have disease than fertile controls undergoing laparoscopic tubal ligation.16, 17, 18] This would seem to support the notion that pregnancy is a good treatment for endometriosis. However, fertile patients undergoing tubal ligation could be a 'super fertile' and therefore unrepresentative subset of the total population of women. The notion of pregnancy as treatment seems to be a charade resulting from misinterpreting the reduced fecundity of women with the disease as positive evidence of a beneficial effect of pregnancy in those without the disease, compounded by the observation of symptom reduction during pregnancy in women with the disease.

No one has ever done the simple study which would have proven whether pregnancy eradicates endometriosis. This would require making a surgical diagnosis of endometriosis in a woman and not treating the disease. After an intervening pregnancy, the pelvis could be checked by reoperation to see if the disease were gone

Does endometriosis cause fertility? Does pregnancy cause endometriosis?

Not all authors, however, have found a high rate of infertility among patients with endometriosis. At odds with his later published work,[14] Counseller [19] observed a 61.3% crude birth rate and a 71.5% crude pregnancy rate among 737 married patients with endometriosis. Bennet found a corrected birth rate of 88% in patients with only endometriosis.[20] Dougherty and Anderson found that 87% of their patients with endometriosis had previously been pregnant and wondered whether pregnancy caused endometriosis.[21] Andrews and Larsen found that 72% of their married patients with endometriosis had been pregnant prior to their diagnosis.[22] One modern longitudinal cross-section study of largely untreated patients with endometriosis found no clearly consistent protective effect of parity against extent of disease as measured by number of pelvic areas involved and found that most patients had previously conceived, even though this study population included women never attempting pregnancy.[23] Although it might be tempting, as Dougherty and Anderson did, to conclude from observations such as these that pregnancy causes endometriosis or that fertility is caused by endometriosis, this would be as outrageous as observing a higher rate of infertility associated with endometriosis and concluding, as did Meigs and several generations of gynecologists, either that lack of childbearing causes endometriosis or that infertility is caused by endometriosis.

Pregnancy effects on endometriosis

The notion that pregnancy has a cytotoxic effect on endometriosis has been refuted by several authors. MacArthur and Ulfelder reviewed the behavior of 24 cases of endometriosis during pregnancy in a literature review.[24] They were unable to verify that pregnancy had a beneficial effect on endometriosis. They concluded that its behavior in pregnancy is highly variable, that permanent regression after pregnancy is the exception and may never truly occur, and that regression of endometriosis in the rare instances in which it occurred during pregnancy appeared to be ascribable to diminished tissue responsive- ness to hormonal stimulation postpartum, rather than to necrosis and actual disappearance of endometriosis.

Further evidence that pregnancy is not toxic to endometriosis has been reported. Hanton and colleagues described two cases of symptomatic ovarian endometrioma, one during pregnancy, the other immediately postpartum.[25] Norenberg and colleagues reported a case of diaphragmatic pregnancy adjacent to endometriosis.[26]

It was clearly a leap of faith to observe a slightly higher percentage of nulligravidas than gravidas with endometriomas and then conclude that endometriosis is a cause of sterility or caused by delayed childbearing when parous women in the same study also suffered from it [5, 6, 27]. Still lacking is a simple prospective study with accurate antenatal and postpartum laparoscopic control proving histologically that untreated endometriosis is improved or absent after pregnancy.

Although the scientific evidence supporting pregnancy as prophylaxis or treatment of endometriosis seems unconvincing, this did not stop efforts to mimic its presumed beneficial state as suggested by Berkson's fallacy.

Pseudopregnancy treatment of endometriosis

Kistner initiated the concept of pseudopregnancy as treatment for endometriosis.[28] Pseudopregnancy was based on the unproven 'fact' that pregnancy improves existing endometriosis or prevents its development. There was no proof of this 'fact' at that time.[28] The anecdotal impetus for pseudopregnancy came from observations on a 25-year-old patient without a proven diagnosis of endometriosis. A dark brown, collapsed cystic structure on her right ovary was removed at Cesarean section and submitted for microscopy. Although no glandular element was found, it was claimed that microscopic evidence of endometrial stroma was found in a field of decidual change and occasional necrosis. The term 'necrobiosis' was coined to describe this histologic finding, and because endometriosis was not found, it was concluded that pregnancy had cured her disease. It was postulated that eventually all of the endometriotic stroma might have undergone complete necrosis and resorption if exposed to the hormonal effects of pregnancy long enough. Beyond the scientifically objectionable assumption of 'fact' and the obvious observation that if nine months of pregnancy had not resulted in complete necrosis of endometriosis then pregnancy is incomplete therapy for the disease, it is risky to base an entire ethos of treatment on a single patient who did not have proven endometriosis. This study was small (12 patients) and loose: it did not require biopsy proof of disease, used several dose regimens, assessed endometrial but not endometriotic response to therapy, did not attempt to quantify extent of disease, and did not mention length of follow-up.

A second loosely designed study of 58 patients, 30 of whom had previous conservative surgery, treated with several medicines in several dose regimens for varying lengths of time found a 'satisfactory' response in 93.3% of surgical-medical patients but in only 75% of medically treated patients.[29] Six of 24 medical patients had subsequent surgery and all had endometriosis. Two of 30 surgical-medical patients had subsequent surgery and one had endometriosis. Four patients with vaginal endometriosis were treated medically, but one had repeated biopsies to assess treatment. Disease was eventually absent in that patient, who was judged a medical cure despite the obvious possibility that the disease was surgically removed by repeated biopsies. This surgically treated patient has been referred to by other authors as an example of successful medical therapy.[18] Beyond the complexity of again using multiple dose regimens of multiple medications, with or without surgery, in patients whose extent of disease is not known, the apparent superior results obtained with surgery were not emphasized. It was concluded that medical therapy with norethynodrel was optimal.

Other reports on pseudopregnancy followed,[30, 31, 32] mentioning endometrial biopsy as a way to assess response of endometriosis to therapy since it was assumed that endometriosis was simply misplaced normal endometrial tissue. It is now realized that endometriosis has low and varying populations of hormone receptors,[33, 34, 35] as compared with native endometrium, so a difference in response would be expected. Therefore, the response of the endometrium is not pertinent to the treatment of endometriosis.

It is actually impossible to assess the response of endometriosis to medical treatment, since it is impossible to obtain biopsies before and after treatment from the same group of cells. An assumption must be made that any biopsy taken either before or after treatment will be an accurate representation of all other disease in the pelvis. Considering the protean visual, histologic, biochemical and hormonal characteristics of endometriosis, this assumption may not be valid.

Although some patients with endometriosis may have diminution of symptoms while on therapy or successful conception following therapy, these observations by themselves do not prove eradication of disease by pseudopregnancy therapy. Considering the historical context in which it was developed, it came as no surprise that rates of persistence of disease of 90-100% were reported after pseudopregnancy therapy.[22, 36]

The effect of the menopause on endometriosis

It has been observed for many years that endometriosis is primarily a disease of the menstrual years. Goldstein and colleagues [37] reported the youngest patient with the disease, 10.5 years old, while the oldest patient, mentioned by Haydon, was 78.[12] It has long been held that the hypoestrogenic state accompanying menopause improves or cures endometriosis, while estrogen stimulates or perhaps causes it.

Sampson observed the relative rarity of postmenopausal endometriosis and made the optimistic and highly qualified conclusions that: 'I hope and expect that the cessation of ovarian function will cause any adenomatous tissue which was left in the pelvis to atrophy',[5] and that: '...the implantations will usually, possibly always, atrophy after all ovarian tissue is removed ... All of them probably cease to grow and actually atrophy after the menopause'.[38]

Meigs reinforced the concept of the importance of the cyclic action of estrogen and progesterone on the maintenance of endometriosis.[6] Although two of his 16 patients with ovarian endometriomas were menopausal, he nonetheless believed that without the estrogenic stimulus provided by the ovaries, endometriosis shrinks and atrophies. In addition, he proposed bilateral oophorectomy as treatment which would make the catamenial activity of the ovarian endometrioma cysts stop. He reasoned that removal of the ovaries would make the cysts within the ovaries stop growing and slowly atrophy (sic). He referred to 'case number 8946' as an example of shrinkage of residual endometriosis after castration. There is no case number 8946 in the paper, but there is a case number 8496. This patient had a lime-sized mass which persisted on the left side of the pelvis for 25 months postoperatively, it then disappeared within 2 months. Although both ovaries had been removed, it was assumed that this was a remnant of the left ovary. Beyond the obvious observations that Meigs did not really know what the mass was and therefore made unjustified conclusions and that parts of his paper may have been poorly worded at best, if 25 months of castration did not make endometriosis go away, then castration is not an efficient treatment of the disease.

Although the notion that lack of cyclic estrogenic activity will result in physical eradication of endometriosis may not have been specifically stated by early authors, many authors up to the present have mentioned castration with retention of disease as satisfactory treatment of endometriosis.

In 1936, Cattell and Swinton [39] stated that castration 'will cause the lesion to recede and usually relieve symptoms'. No supporting references were offered.

Cattell, [40] later noting that 6 of 11 patients with significant bowel disease continued to have symptoms or abnormal x-rays after removal of the ovaries and retention of disease, stated: 'It seems safe to remove the ovaries without resecting the bowel if the diagnosis of endometriosis is confirmed by frozen section'. The rationale for favoring a treatment that usually left pathology behind was not explained.

Fallon and colleagues [41] stated that: '...all endometriosis ... regresses after removal of the ovaries ... Following ovarian deactivation, endometriosis gradually regresses, with insignificant residual'. Partial oophorectomy was mentioned as a means of slightly reducing estrogen production, possibly with diminution of symptoms. No references were offered in support of these opinions.

Counseller and Crenshaw [14] noted the difficulty sometimes encountered in attempting conservative surgery in extensive disease, and left ovarian tissue in only 38.5% of their patients. They stated that: 'Obviously, the quickest and most certain way for the relief of pain is the destruction of the ovarian function...'. Counseller [19] had stated this earlier. No references were offered in support of their opinions.

None of the above studies showed actual destruction of endometriosis resulting from decreased estrogen effect, but relied on symptomatology to gauge success of treatment. As with pregnancy, absence or diminution of symptoms does not prove elimination of disease.

Probably because endometriosis is less symptomatic after the menopause, and because fewer menopausal patients seek treatment for it, not as much is known about the disease in postmenopausal patients as following pregnancy. However, it is clear from published reports that older women can have problems due to endometriosis.

In two studies of large numbers of patients with endometriosis, 2.5% and 3.7% of the patients were postmenopausal [42, 43].

In another study of postmenopausal endometriosis by Kempers and colleagues [44] 138 patients 45 years of age or older who were menopausal for at least two years were compiled. Only one patient had been on estrogen replacement therapy. Forty-one had clinically significant disease including intestinal disease, widespread pelvic involvement or endometriomas over 3 centimeters in diameter (61% of these patients had been fertile, another indication of the baseline fertility potential among these patients). Among eight patients with intestinal endometriosis, the disease was classified histologically as atrophic in all, including two with small bowel obstruction. In 26 patients with histologic proof of disease, 20% had histologic evidence of cellular activity of their disease, but none of these was symptomatic. The conclusions were that: symptomatic postmenopausal endometriosis occurs, albeit less prevalent than symptomatic premenopausal disease, postmenopausal disease can be independent of estrogen support, and the histologic appearance of the disease is not helpful in predicting whether it is symptomatic.

Although some reports of postmenopausal endometriosis have detailed patients who have not been exposed to endogenous or exogenous estrogen, one patient with postmenopausal endometriosis associated with ovarian hyperthecosis has been reported.[45]

Although there is no question that the disease usually appears visually to be quiescent post menopause, and microscopically the glands may appear atrophic, scientific evidence of an actual toxic effect of the menopause on endometriosis is lacking. Nonetheless, menopause has frequently been proposed as treatment of endometriosis, and many physicians believe that this will cure a patient in the absolute sense of the word: that the disease will ultimately be gone as a result of anti-cellular toxicity of low estrogen levels. In a reprise of the developments preceding initiation of pseudopregnancy therapy, the fact that post-menopausal patients are less commonly afflicted with symptomatic disease has led to the conclusion that endometriosis is cured by the hypoestrogenic states of natural or surgical menopause.[46] Castration or medical suppression of ovarian function is a natural suggestion for treatment of endometriosis if it is believed that women do not need estrogen and that endometriosis is cured by some type of anti-cellular activity resulting from the lack of estrogen characterizing menopause. Ovarian suppression, therefore, is the goal of pseudo-menopausal treatments of endometriosis.

Pseudomenopausal treatments of endometriosis

Danazol was the first medicine introduced for the treatment of endometriosis by pseudo-menopause. Gonadotropin-releasing hormone (GnRH) agonists were the second.

One of the first studies on danazol was illustrated by many 'before and after' photographs of the female pelvis purporting to show resolution of endometriosis after danazol therapy.[47] This led immediately to great hopes that endometriosis could be cured by medical therapy which mimicked nature. However, not all the patients had biopsies taken after therapy, and the notion of consistent, profound anti-gonadotropin suppression by danazol was later found to be erroneous.[46, 48] It is now realized that danazol does not eradicate endometriosis of any stage or location,[49, 50, 51, 52, 53] and the hopes dashed by danazol seem to be a combined result of historically misplaced expectations, errors of visual identification and of not confirming clinical observations by biopsy [50, 54]. Recent reports on the use of danazol for endometriosis have shown an actual decrease in fertility in treated women with minimal or mild disease compared to operative laparoscopy [52] or observation only, [55, 56, 57] while the improved pregnancy rate following surgical therapy for disease associated with pelvic adhesions has long been recognized. Since there is a surprisingly high background pregnancy rate in untreated disease in infertile patients, [55, 56, 58] the true effect of medical or surgical therapy on fertility is difficult to determine, since it may be small. It seems most likely that much of the 'improved' fertility after treatment is due to the natural history of endometriosis, not to the treatment itself.

Urologists have questioned the use of danazol for endometriosis involving the urinary tract, noting it to be ineffective for relief of obstructive ureteral endometriosis [59]. Since danazol does not eradicate endometriosis, it is not surprising that it is ineffective in long-term relief of pain [60, 61, 62, 63]. Danazol has been found to reduce high density lipoprotein (HDL) cholesterol [64].

GnRH agonists seek to mimic the hypoestrogenic menopausal state by inducing a profound suppression of pituitary gonadotropins. However, since this approach shares danazol's historical basis of origin, one would naturally predict that GnRH agonists would replicate danazol's shortcomings, although with different side-effects.

An early report [65] on the efficacy of GnRH agonists assessed response of the endometrium rather than endometriosis, assessed the visual appearance of endometriosis at the conclusion of therapy when errors of visual identification are common, [50] and spoke of disappearance and resorption of lesions without biopsy proof. The definite impression was left that GnRH agonist therapy resulted in physical eradication of disease.

A large multicenter randomized, double-blind, study of 213 treated patients was based on the undefined and unreferenced statement that: 'endometriosis resolves after ovariectomy and menopause,' [66] and went on to find that treatment with a GnRH agonist, nasal nafarelin, was as effective as danazol in reduction of symptoms and severity of disease while on treatment. A post-treatment pregnancy rate of 39% was reported for all stages of disease. Treatment response was evaluated by laparoscopy done at the end of the treatment period, presence or absence of disease was apparently not confirmed by biopsy, and follow-up was done only for patients attempting pregnancy.

Misinterpretation of published studies in favor of conventional dogma continues to be a problem. For example, a study published in 1977 recommended conservative surgery for endometriosis in younger patients, but oophorectomy in older patients if a hysterectomy was performed, since a high percentage of these patients required estrogen replacement therapy within 5 years postoperatively [67]. If the disease was resected and the ovaries were left in, only 6% of patients required reoperation for endometriosis. A recent study of GnRH agonist therapy [68] stated that: 'bilateral oophorectomy is the most effective treatment of endometriosis' and cited the 1977 study as the basis of this belief, although the earlier study said nothing of the kind. Nonetheless, it was proposed that GnRH agonists might therefore be effective treatment for endometriosis. Since the earlier study was cited out of context, and since there is no historical support for the concept of hypoestrogenic toxicity against endometriosis, it came as no surprise when the more recent study found that GnRH agonists do not eradicate endometriosis [69]. Other investigators have confirmed that GnRH agonists do not eradicate endometriosis [69]. Reversible reduction in trabecular bone accompanying the hypoestrogenic state has been reported, [70] and pain recurrence follows therapy by gonadotropin-releasing hormone.[71]

Current attempts at pseudomenopausal therapy share another feature which makes it impossible ever to mimic the menopausal state exactly. Menopause is characterized by low estrogen and high gonadotropin levels, whereas danazol and the GnRH agonists seek to lower the ovarian production of estrogen in part by reducing pituitary gonadotropin output. Thus, only a 'pseudo' pseudomenopause is achieved. The effect of high levels of gonadotropins on endometriosis is unknown.

It is becoming popular to subject women with pelvic pain to a three to six month trial of GnRH agonist therapy, the thinking being that if pain relief occurs, then endometriosis is likely to be present so medical therapy can be continued indefinitely. Actually, the evidence is clear that symptomatic response to such a trial of medical therapy will only narrow the field of possible diagnoses to at least six estrogen-mediated conditions that can cause pain: uterine fibroids, adenomyosis, primary dysmenorrhea, ovulation pain, endometriosis, and unknown. Trying to discriminate between these six conditions with diagnostic imaging tests such as ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI) is difficult. Fibroids might be found if large enough, not all cases of adenomyosis or endometriosis would be evident, and none of the rest would be found. Thus, at the end of an expensive six months of treatment with its associated side-effects and after several types of scans, the clinician will still be unable to guarantee a diagnosis of endometriosis. Such a trial of therapy thus resolves down to whether or not to continue successful treatment of symptoms without a diagnosis.

Discussion

Berkson's fallacy seems to be affecting thought on endometriosis to this day. In the beginning, the weak pathoconsortive association between endometriosis and the symptom of infertility was observed in hospitalized patients, and it was inferred that endometriosis is causative of infertility, that it is promoted by voluntarily delayed childbearing, and that it can be cured or delayed by pregnancy. These unproven opinions are repeated to patients to the present time. Other authors, observing the rarity of symptomatic disease in hypoestrogenic women and perhaps not realizing the importance of estrogen to women, concluded that menopause, by its hypoestrogenic effect, cured or improved endometriosis and is good treatment, and this notion also persists to this day. These beliefs have indirect support from temporary symptom improvement, but for the most part are fueled by continuing errors in visual identification of disease as well as unreferenced hopes, beliefs, expectations and dogma all of which arise in part from the seductiveness of an unconfirmed belief [72]. Scientific studies based on these notions continue to be produced.

If either pregnancy or menopause resulted in the destruction of endometriosis, there should by now be abundant microscopic evidence of this effect in humans. Sadly, the simple studies which would have confirmed or denied the basis of all medical therapy have never been done. These studies would be simple to conduct: patients with surgically proven endometriosis would be surgically reinvestigated after pregnancy or menopause to see if their endometriosis was gone.

Since medical therapy cannot be used purposefully to eradicate endometriosis, there must be some other reason for its ethical use, such as relief of symptoms. However, hope of permanent relief of symptoms by medical therapy is misplaced because the cause of the symptoms is not eliminated, a permanent change in the disease is not induced, and it is unclear whether it can be suppressed forever. In the modern era of concern over medical costs and efficiency, consumer attention is toward a desire to get to the root cause of disease, not simply to mask symptoms. In this regard, current medical therapy is a cliché.

Emphasis on fertility as a cardinal symptom of the disease or endpoint of therapy has also contributed to continued disappointment and confusion, not only because infertility is a less common and less specific symptom than pain, but also because findings referable to patients with infertility may not be at all pertinent to the greater number with pain. It is poor medical practice to treat patients with one symptom based on study results of another symptom which may not be directly related to the disease. The presumed inherent 'goodness' of conservative medical management, the temporary improvement experienced by some patients during treatment, and the desire to avoid the morbidity of surgery with its ill-defined risk of postoperative adhesions are not necessarily convincing reasons to embrace current or future medical therapy of endometriosis.

References

1. Berkson J. Limitations of the application of fourfold table analysis to hospital data. Biometrics 1946;2:47-53.
2. Berkson J. The statistical study of association between smoking and lung cancer. Proc Mayo Clinic 1955;30:319-48.
3. Schenken RS. Pathogenesis. In: Schenken R (ed) Endometriosis: contemporary concepts in clinical management. Philadelphia: JB Lippincott, 1989:1-48.
4. Cullen TS. The distribution of adenomyomas containing uterine mucosa. Arch Surg 1920; 215-83.
5. Sampson JA. Perforating hemorrhagic (chocolate) cysts of the ovary. Arch Surg 1921;3: 245-323.
6. Meigs JV. Endometrial hematomas of the ovary. Bost Med Surg J 1922;187:1-13.
7. The American Fertility Society. Revised American Fertility classification system of endometriosis: 1985. Fertil Steril 1985;44:351-2.
8. Redwine DB. The visual appearance of endometriosis and its impact on our concepts of the disease. Prog Clin Biol Res 1990;323: 393-412.
9. Fallon J, Brosnan JT, Manning JJ, Moran WG, Meyers J, Fletcher EM. Endometriosis: a report of 400 cases. Rhode Island Med J 1950;33: 15-23.
10. Meigs JV. Etiologic role of marriage age and parity; conservative treatment. Obstet Gynecol 1953;2:46-53.
11. Meigs JV. An interest in endometriosis and its consequences. Am J Obstet Gynecol 1960;79: 625-35.
12. Haydon GB. A study of 569 cases of endometriosis. Am J Obstet Gynecol 1942;43:704-9.
13. Burns WN, Schenken RS. Pathophysiology. In: Schenken R (ed) Endometriosis: contemporary concepts in clinical management. Philadelphia: JB Lippincott, 1989:83-126.
14. Counsellor VS, Crenshaw JL. A clinical and surgical review of endometriosis. Am J Obstet Gynecol 1951; 62:930-42.
15. Beecham CT. Surgical treatment of endometrio- sis with special reference to conservative surgery in young women. JAMA 1949;139:971-2.
16. Hasson HM. Incidence of endometriosis in diagnostic laparoscopy. J Reprod Med 1976; 16:135-8.
17. Drake TS, Metz SA, Grunert GM, O'Brien WF. Peritoneal fluid volume in endometriosis. Fertil Steril 1980;34:280-1.
18. Strathy JH, Molgaard CA, Coulam CB, Melton LJ. Endometriosis and infertility: a laparoscopic study of endometriosis among fertile and infertile women. Fertil Steril 1982;38:667-72.
19. Counseller VS. Surgical procedures involved in the treatment of endometriosis. Surg Gynecol Obstet 1949;89:322-7.
20. Bennet ET. Endometriosis in the older age group. Am J Obstet Gynecol 1953;65:100-8.
21. Dougherty CM, Anderson MR. Endometriosis and adenomyosis. Am J Obstet Gynecol 1964;89:23-37.
22. Andrews WC, Larsen GD. Endometriosis: treatment with hormonal pseudopregnancy and/or operation. Am J Obstet Gynecol 1974;118:643-51.
23. Redwine DB. The distribution of endometriosis in the pelvis by age groups and fertility. Fertil Steril 1987;47:173-5.
24. McArthur JW, Ulfelder H. The effect of pregnancy upon endometriosis. Obstet Gynecol Surv 1965;20:709-33.
25. Hanton EM, Malkasian GD, Dockerty MB, Pratt JH. Endometriosis: symptomatic during pregnancy. Am J Obstet Gynecol 1966;95: 1165-6.
26. Norenberg DD, Gundersen JH, Janis JF, Gundersen AL. Early pregnancy on the diaphragm with endometriosis. Obstet Gynecol 1977;49:620-2.
27. Sampson JA. Heterotopic or misplaced endometrial tissue. Am J Obstet Gynecol 1925; 10:649-64.
28. Kistner RW. The use of newer progestins in the treatment of endometriosis. Am J Obstet Gynecol 1958;75:264-78.
29. Kistner RW. The treatment of endometriosis by inducing pseudopregnancy with ovarian hormones. A report of fifty-eight cases. Fertil Steril 1959;10:539-56.
30. Kistner RW. Infertility with endometriosis. A plan of therapy. Fertil Steril 1962;13:237-45.
31. Kistner RW. Current status of the hormonal treatment of endometriosis. Clin Obstet Gynecol 1966;9:271-92.
32. Kourides IA, Kistner RW. Three new synthetic progestins in the treatment of endometriosis. Obstet Gynecol 1968;31:821-8.
33. Bergqvist A, Rannevik G, Thorell J. Estrogen and progesterone cytosol receptor concentration in endometriotic tissue and intrauterine endometrium. Acta Obstet Gynecol Scand Suppl 1981;101:53-8.
34. Gould SF, Shannon JM, Cunha GR. Nuclear estrogen binding sites in human endometriosis. Fertil Steril 1983;39:520-4.
35. Janne O, Kauppila A, Kokko E, Lantto T, Ronnberg L, Vihko R. Estrogen and progestin receptors in endometriosis lesions: Comparison with endometrial tissue. Am J Obstet Gynecol 1981;141:562-6.
36. Andrews MC, Andrews WC, Strauss AF. Effects of progestin-induced pseudopregnancy on endometriosis: clinical and microscopic studies. Am J Obstet Gynecol 1959;78:776-83.
37. Goldstein DP, DeCholnoky C, Eman J. Adolescent endometriosis. J Adolesc Health Care 1980;1:37-41.
38. Sampson JA. Ovarian hematomas of endome- trial type (perforating hemorrhagic cysts of the ovary) and implantation adenomas of the endometrial type. Bost Med Surg J 1922; 186:445-56.
39. Cattell RB, Swinton NW. Endometriosis. With particular reference to conservative treatment. N Engl J Med 1936;214:341-6.
40. Cattell RB. Endometriosis of the colon and rectum with intestinal obstruction. N Engl J Med 1937;217:9-16.
41. Fallon J, Brosnan JT, Moran WG. Endo- metriosis. Two hundred cases considered from the viewpoint of the practitioner. N Engl J Med 1946:235:669-73.
42. Scott RB, TeLinde RW. External endometriosis - the scourge of the private patient. Ann Surg 1950;131:697-720.
43. Henriksen E. Endometriosis. Am J Surg 1955; 90:331-7.
44. Kempers RD, Dockerty MB, Hunt AB, Symmonds RE. Significant postmenopausal endometriosis. Surg Gynecol Obstet 1960;3:348-56.
45. Montes M, Beautyman W, Haidak G. Cholesteatomatous endometriosis. Am J Obstet Gynecol 1961;82:119-23.
46. Reyniak JV, Lauersen NH. Danazol - a versatile pharmacologic agent. Fertil Steril 1982;37:475-7.
47. Dmowski WP, Cohen MR. Treatment of endometriosis with an antigonadotropin, danazol. A laparoscopic and histologic evaluation. Obstet Gynecol 1975;46:147-54.
48. Fraser IS, Markham R, McIlveen J, Robinson M. Dynamic test of hypothalamic and pituitary function in women treated with danazol. Fertil Steril 1982;37:484-8.
49. Schweppe K-W, Wynn RM, Beller FK. Ultrastructural comparison of endometriotic implants and eutopic endometrium. Am J Obstet Gynecol 1984;148:1024-37.
50. Evers JLH. The second-look laparoscopy for evaluation of the result of medical treatment of endometriosis should not be performed during ovarian suppression. Fertil Steril 1987;47:502-4.
51. Dmowski WP. Visual assessment of peritoneal implants for staging endometriosis: do number and cumulative size of lesions reflect the severity of a systemic disease? Fertil Steril 1987;47:382-4.
52. Fayez JA, Collazo LM, Vernon C. Comparison of different modalities of treatment for minimal and mild endometriosis. Am J Obstet Gynecol 1988;159:927-32.
53. Brosens IA, Verleyen A, Cornillie F. The morphologic effect of short-term medical therapy of endometriosis. Am J Obstet Gynecol 1987;157:1215-21.
54. Greenblatt RB, Borenstein R, Hernandez-Ayup S. Experiences with danazol (an antigonadotropin) in the treatment of infertility. Am J Obstet Gynecol 1974;118:783-7.
55. Seibel MM, Berger MJ, Weinstein FG, Taymor ML. The effectiveness of danazol on subsequent fertility in minimal endometriosis. Fertil Steril 1982;38:534-7.
56. Bayer SR, Seibel MM, Saffan DS, Berger MJ, Taymor ML. Efficacy of danazol treatment for minimal endometriosis in infertile women. J Rep Med 1988;33:179-83.
57. Telimaa S. Danazol and medroxyprogesterone acetate inefficacious in the treatment of infertility in endometriosis. Fertil Steril 1988;50: 872-5.
58. Collins JA, Wrixon W, Janes LB, Wilson EH. Treatment-independent pregnancy among infertile couples. N Engl J Med 1983;309:1201-9.
59. Jepsen JM, Hansen KB. Danazol in the treatment of ureteral endometriosis. J Urol 1988;139: 1045-6.
60. Fedele L, Arcaini L, Bianchi S, Baglioni A, Vercellini P. Comparison of cyproterone acetate and danazol in the treatment of pelvic pain associated with endometriosis. Obstet Gynecol 1989; 73:1000-4.
61. Fedele L, Bianchi S, Viezzoli T, Arcaini L, Candiani GB. Gestrinone versus danazol in the treatment of endometriosis. Fertil Steril 1989;51: 781-5.
62. Dmowski WP. Pseudomenopause: a new approach in treating endometriosis. Contrib Obstet Gynecol 1976;8:107-14.
63. Barbieri RL, Evans S, Kistner RW. Danazol in the treatment of endometriosis: analysis of 100 cases with a 4-year follow-up. Fertil Steril 1982; 37:737-46.
64. Dmowski WP, Radwanska E, Binor Z, Tummon I, Pepping P. Ovarian suppression induced with Buserelin or danazol in the management of endometriosis: a randomized, comparative study. Fertil Steril 1989;51:395-400.
65. Lemay A, Maheux R, Faure N, Jean C, Fazekas ATA. Reversible hypogonadism induced by a luteinizing hormone-releasing hormone (LH- RH) agonist (Buserelin) as a new therapeutic approach for endometriosis. Fertil Steril 1984;41: 863-71.
66. Henzl MR, Corson SL, Moghissi K, Buttram VC, Berqvist C, Jacobson J et al. Administration of nasal nafarelin as compared with oral danazol for endometriosis. A multicenter double-blind comparative clinical trial. N Engl J Med 1988; 318:485-9.
67. Williams TJ, Pratt JH. Endometriosis in 1,000 consecutive celiotomies: incidence and management. Am J Obstet Gynecol 1977;129:245-50.
68. Steingold KA, Cedars M, Lu JKH, Randle D, Judd HL, Meldrum DR. Treatment of endometriosis with a long-acting gonadotropin-releasing hormone agonist. Obstet Gynecol 1987;69:403-11.
69. Nisolle-Pochet M, Casanas-Roux F, Donnez J. Histologic study of ovarian endometriosis after hormonal therapy. Fertil Steril 1988;49:423-6.
70. Dawood MY, Lewis V, Ramos J. Cortical and trabecular bone mineral content in women with endometriosis: effect of gonadotropin-releasing hormone agonist and danazol. Fertil Steril 1989;52:21-6.
71. Franssen AMHW, Kauer FM, Chada DR, Zijlstra JA, Rolland R. Endometriosis: treatment with gonadotropin-releasing hormone agonist Buserelin. Fertil Steril 1989;51:401-8.
72. Feinstein AR. Clinical epidemiology: the archi- tecture of clinical research. Philadelphia: WB Saunders, 1985:408.