A randomized, double-blind study of six combined oral contraceptives

CONTRACEPTION

A RANDOMIZED, DOUBLE-BLIND STUDY OF SIX COMBINED ORAL CONTRACEPTIVES

Task Force on Oral Contraceptives WHO Special Programme of Research, Development and Research Training in Human Reproduction

Dr S. Koetsawang and Dr 0. Kiriwat (Siriraj Hospital, Mahidol University, Bangkok, Thailand); Dr A. Sheth, Dr V. Choudhary, Dr K. Hazari, Dr J. Joshi, Dr S. Mehta and Dr K. Virkar (Institute for Research in Reproduction, Bombay, India); Dr A.N. Gupta (Postgraduate Institute of Medical Education and Research, Chandigarh, India); Dr V. Aimakhu, Mrs G. Delano, Dr 0. Ladipo and Professor 0. Ojo (University of Ibadan, Ibadan, Nigeria); Dr L. Andolsek and Dr Majda Kustrin (Family Planning Institute, University Clinical Hospital, Ljubljana, Yugoslavia); Professor T.K. Chatterjee, Mrs L. Niles and Dr D.S.O. Wacha (University of Zambia, Lusaka, Zambia); Dr R. Apelo (ReproductiveBiology Centre, University of the PhilIppines, Manila, Philippines); Dr J. Lopez (Hospital "Jose Joaquin Aguirre", University of Chile, Santiago, Chile); Dr S.C. Chew and Professor S.S. Ratnam (Kandang Kerbau Hospital for Women, University of Singapore, Singapore, Republic of Singapore); Dr L. Kovacs, Dr S. Zalanyi Jr. and Dr Andras Tekulics (University Medical School, Szeged, Hungary); M.A. Belsey, P.E. Hall, R.A. Parker, 0. Ayeni, A. Pinol and C. Li Hoi Foo (World Health Organization, Geneva, Switzerland)

Submitted for publication October 22, 1981 Accepted for publication February 3, 1982

Requests for reprints: Dr Mark A. Belsey, Medical Officer Special Programme of Research, Development and Research Training in Human Reproduction World Health Organization 1211 Geneva 27, Switzerland

MARCH 1982VOL.25NO.3

231

CONTRACEPTION

ABSTRACT A randomized controlled clinical trial comparing six combined oral contraceptives with 5Opg or less of ethinyl estradiol was undertaken in 10 WHO Collaborating Centres for Clinical Research in Human Reproduction. A total of 2430 women entered the trial and were observed for 28,077 woman-cycles. Ali low-dose combined oral contraceptives demonstrated equivalent efficacy with one-year pregnancy rates of one to six percent. However, discontinuation rates for medical reasons differed significantly between the treatment groups, with the preparation containing ZOpg ethinyl estradiol and that containing 4OOpg norethisterone acetate being associated with higher discontinuation rates due to bleeding disturbances. Even among the preparations which did not differ in discontinuation rates, the reasons for discontinuation did differ. Women receiving norethisterone preparations tended to discontinue because of bleeding disturbances while those receiving the levonorgestrel-containing pfeparations tended to discontinbe because of complai,nts of nausea and vomiting. INTRODUCTION During the past ten years, more than 40 different combined oral contraceptive preparations have been commercially available or provided through family planning programmes in developed and developing countries. It is estimated that more than 50 million women annually now use oral hormonal contraceptives(l), for the most part combined oral contraceptive preparations containing 5Opg or less of ethinyl estradiol or mestranol in combination with a progestogen. The rationale for choosing between the many low-dose combined oral contraceptives available is not based on scientific study. Despite a plethora of clinical and laboratory reports on individual preparations, reports of clinical trials comparing the effectiveness, incidence of side-effects and rates of discontinuation for different low-dose preparations are few(l,2). As part of its ongoing investigations into the safety and efficacy of fertility control methods, especially those issued in developing countries, the WHO Special Programme of Research, Development and Research Training in Human Reproduction has undertaken a multi-centred randomized double-blind study of six low-dose combined oral contraceptives. The preparations were selected on the basis of either their existing or expected widespread use In In order to determine the difference between the developing countries. effects of a 3Opg and 5Opg dose of ethinyl estradiol (EE) with a constant level of levonorgestrel (LNC), one preparation not yet commercially availabLe was included. Progestogens ln the preparations included in the study were limlted to norgestrel and norethisterone since both are known to act directly on progesterone receptors and nearly all other progestogens of the lY-nor-steroid type are metabolized to norethisterone before being active in this way.

232

MARCH

1982 VOL. 25 NO. 3

CONTRACEPTION METHODS AND MATERIALS A total of 2762 women attending family planning clinics between November 1975 and January 1’179 in ten WHOCollaborating Centres for Clinical Research in Human Reproduction (CCCR’s)* volunteered to participate in the study. Of these, 2430 women started in the study after satisfying the criteria for inclusion. To be included in the trial, a woman had to be healthy; to have had no contraindications for oral contraceptive use such as hypertension, heart disease, diabetes, etc.; to have been between the ages of 18 and 38, inclusive; if post-partum, to be at least 28 days post-partum and to have re-established menstruation; if lactating, to have been lactating at least 165 days and re-established menstruation; not to have used oral contraceptives within 28 days or long-acting injectable hormonal contraceptives within 90 days of starting the treatment groups; to have had regular menstrual cycles In addition, the woman with the usual length within 21 to 35 days, inclusive. must have agreed not to use any other contraceptive methods during her participation in the trial. Women were able to withdraw from the trial at any time at their own request, and were discontinued by the clinical trial staff when pregnancy was diagnosed or when severe medical conditions developed such as cardiovascular disease, hypertension or liver disease. Preparations The six (1) (21 (3) (41 (5) (6)

preparations

compared were:

ethinyl estradiol 50pg + levonorgestrel 15O)rg (EE 50 + LNG 150) ethinyl estradiol 3Opg + levonorgestrel 15Opg (EE 30 + LNG 150) mestranol 5Opg + norethisterone lmg (MES 50 + NET lmg) ethinyl estradiol 5Opg + norethisterone acetate lmg (EE 50 + NAc 1) ethinyl estradiol 20pg + norethisterone acetate lmg (EE 20 + NAc 11 ethinyi estradiol 35pg + norethisterone acetate 4OOpg (EE 35 + NAc 400)

The pills were produced by the manufacturers, but repackaged in twenty four 28-day cycle packets and identified by random numbers in such a way that neither the clinical trial staff nor the women knew which preparation was used by a particular woman+. At each of the centres women volunteering ana eligible to participate were randomly assigned to receive one of the six oral contraceptive preparations. In all circumstances, the pill regimen was begun on the fifth day of the menstrual cycle. The women were scheduled to return to the clinics at intervals of three cycles (84 days) for up to eight regular visits (24 cycles per woman). A woman was discontinued from the trial if the first 14 or more pills were missed in a cycle. Pregnancies occurring after discontinuation were not included in the analysis. A totai of 28,077 woman-cycles were observed. Data collection was standardized and included information on age, obstetric history, previous contraceptive use, menstrual data, past medical history, current complaints and a physical examination, including a pelvic examination. On follow-up visits, complaints, significant findings on physical examination, blood pressure and weight were recorded. Menstrual diary cards were maintained by the women. 4

for Research in Reproduction in Siriraj Hospital (Bangkok), Institute Postgraduate Institute of Medical Education and Research (Bombay), (Chandigarh), University of Ibadan (Ibadan), Family Planning Institute University of the Philippines (Ljubljana), University of Zambia (Lusaka), Kandang Kerbau Hospital for (Manila), University of Chile (Santiago), Women (Singapore), University Medical School, (Szeged). + One woman in one centre examined her pills with a magnifying was able to identify the company imprint on her pills.

MARCH 1982VOL.25N0.3

glass

and

233

CONTRACEPTION

At the end of the trial, the steroid content was measured in both opened and sealed unused pill samples from each of the centres. For the temperate countries, the hormone levels rn ~111s from unopened packs ranged from 92% to 104% (mean 99.4%) of the stated dosage level; for the similar samples from the troprcal countries, the range was 91.1% to 101.5% (mean Y6.4%). Levels In opened packets were slightly lower. Analysis The statrstical procedures used rncluded standard t-tests, analysrs of variance or X2 test on contingency tables. Extensive use was maoe of the test for heterogeneity of counts which ignores slight differences in sample size. Net life table discontanuation rates were calculated according to the log-rank test(3,4). Life table analysis was calculated from the frrst day of the menstrual period in which prlls were started until the patrent was terminated or completed the study after 24 cycles of use. Loss to follow-up was considered as a termination of use and was included in the "all reason discontinuation rate". Although in theory these women might be obtarnrng oral contraceptives from other sources, in practice the few studies that exist suggest that the majority of such women have drscontinued the method(5). Since net cumulative discontinuation rate estimates involve adjustment for competrng risks, the sum of the rates for specific reasons is greater than the overall drscontinuation rate for "all reasons". Definitions were standardized. A segment was defined as from the start of one menstrual-like bleeding episode to the start of the next menstrual-like bleeding episode. Bleeding disturbances derived from the menstrual diary cards were defined as follows: amenorrhea:

longest bleeding-free

segment greater than 60 days*

infrequent bleeding:

longest bleeding-free and not over 60 days

segment greater than 35 days

frequent bleeding:

shortest complete segment less than 24 days

Irregular bleedrng:

shortest complete segment less than 24 days and longest menstrual segment greater than 35 days

prolonged bleedrng:

bleeding/spotting

episode

longer than 7 days.

In addition to these categorres derived from the menstrual diary cards, the women's complaints of bleeding disturbance were categorrzed and included amenorrhea, irregular bleeding, prolonged bleeding, light bleeding and spotting.

* Although amenorrhea 1s commonly defined rn terms ot 90 days, because follow-up visits were in segments of 84 days f 14 days, the use of 60 days for the definition was found to be more convenient in the analysis.

234

MARCH

1982 VOL. 25 NO. 3

CONTRACEPTION

RESULTS Random allocation of women participating in the trial resulted in relatively equal numbers of subjects in each of the six treatment groups (Table I). As would be expected among the 10 participating centres, there were wide differences in the characteristics of the women starting in the trial. However, as a consequence of the random allocation of subjects within each centre, of 22 variables recorded on admissron, only 3 differed These variables were the statistically (p (0.05) between the six groups. outcome of the last pregnancy (Table I), the length of the open birth interval We considered these differences and the length of the usual menstrual cycle. not to be of practical importance or affecting the results of the study since the magnitudes of the differences were small, and, with the number of statistical tests performed, the probability of finding three tests Furthermore, neither these three variables significant was about p = 0.095. nor any of the other admisslon characteristics correlated with the probabrlity of continuation in the trial, pregnancy rates, side-effects or medical or personal reasons for discontinuation. By two years, discontinuation rates for all reasons differed signirrcantly between treatment groups, being mainly attributable to discontinuations because of side-effects or medical complications. Discontinuations because of loss to follow-up or personal reasons did not ditfer between groups (Table II). Neither the one- nor two-year net life-table pregnancy rates differed significantly between treatment groups which ranged from one to six percent (Table II). Although all ten centres contributed similar numbers of subjects to the study, of the 72 pregnancies, 38 were recorded in the two Indian centres, none were recorded in one centre, and three other centres accounted for 2 to 3 pregnancies each. One-year discontinuation rates for ail reasons and for medical reasons ranged from 47.8% to 56.0% and from 23.9% to 34.7%, respectively (Table II). Discontinuations for increased blood pressure (55 cases), other cardiovascular disease (2 cases), liver dysfunction (15 cases), non-malignant breast lesions (3 cases), post-coital bleeding (1 case) or other major medical" reasons did not differ among the treatment groups. Discontinuations for bleeding problems (200 cases) were generally more common in association with the norethisterone or norethisterone acetate preparations. The preparations containing 20"g ethinyl estradiol and lmg norethisterone acetate had a much higher rate of discontinuations for bleeding problems (63 cases.1 than any of the other preparations. Discontinuations for symptoms of nausea (36 cases) and vomiting (26 cases) were more common with the two levonorgestrel+ontaining preparations (Table II). Discontinuation for other minor medical complaints did not differ among groups. These included other gynaecological complaints (18 cases), other gastrointestinal complaints or perceived weight change (57 cases), skin complaints (1Y cases), headache, and dizziness and other central nervous system or psychological symptoms (162 cases).

* Other major medical reasons for discontinuation included breast malignancy (1 case), other neoplasias (1 case), pulmonary tuberculosis (5 cases), cholelithiasis (7 cases), Papinaucolou smear grade III (5 cases), pyelonephritis (3 cases), typhoid fever (5 cases), hyperthyroidism (2 cases) and one case each of migraine and seizure disorder.

MARCH

1982 VOL. 25 NO. 3

235

3 w

%

8

g c g EJ

Percent live birth Percent abortions Percent none

Outcome last pregnancy

Mean age (sd)

59.9 31.6 8.3

25.4 (4.74)

59.1 35.7 5.2

26.0 (4.73)

209

214

Total number observed at 12th cycle

66.3 27.8 5.8

26.3 (5.03)

203

61.8 31.4 6.8

26.2 (4.70)

192

398

400

406

399

Total starting trial

33

36

24

36

Never started oral contraceptives

29

22

460

458

22

452

NAc 1.0

EE .05

31

466

NET 1.0

MES .05

Excluded for obstetric or gynaecological criteria, medical reasons, age or incorrect treatment group allocation

Number volunteering to participate

EE .03 LNG .15

EE .05 LNG .15

EE .035

70.0 23.8 6.3

64.0 29.9 5.6

26.0 (4.69)

197

181

26.1 (4.69)

411

23

29

463

NAc 0.4

416

32

15

463

NAc 1.0

EE .02

RECRUITMENT AND SELECTED PATIENT CHARACTERISTICS OF WOMEN VOLUNTEERING TO PARTICIPATE

Table I

63.5 30.0 6.3

26.0 (4.77)

1196

2430

184

148

2762

Total

vz

3

P

3

rates

rates

* PC

4.7 33.8 34.3 9.4 60.6

Nausea and vomiting* All medical reasons*** Personal reasons Loss to follow-up All reasons*

4.2 6.2

3.6 27.9 48.9

0.05

.05 .15

0.5 2.6 0.7 1.2 2.9

at

at

EE LNG

Amenorrhoea*** Irregular bleeding** Prolonged bleeding Light bleeding Spotting**

Accidental pregnancy All menstrual***

Discontinuation 676 days

Accidental pregnancy All medical reasons All reasons

Discontinuation 360 days

II

.03 .15

4.6 35.0 34.3 10.8 62.0

1.5 2.8 0.3 0.0 2.8

5.1 5.5

3.5 26.1 47.8

EE LNG

.05 1.0

2.2 40.3 36.2 7.7 64.9

0.8 30.9 37.3 8.2 60.4

** P ( 0.01

3.3 6.2 1.2 2.0 6.8

4.2 13.4

3.5 32.4 53.5

EE .05 NAc 1.0

3.3 2.8 0.3 2.2 2.8

1.0 9.2

1.0 23.9 49.8

MES NET

.02 1.0

***

1.7 46.4 34.5 10.9 68.8

4.0 39.4 31.9 11.2 63.5

2.6 8.5 1.5 1.6 9.0

6.0 15.6

6.0 31.8 52.3

EE ,035 NAc 0.4

P < 0.001

13.7 4.2 1.8 2.4 5.3

23.2

5.0

4.5 34.7 56.0

EE NAc

CUMIJLATIVE NET LIFE-TABLE DISCONTINUATION RATES FOR ALL REASONS AND ALL MEDICAL REASONS

Table

0.024 0.001 0.374 0.244 0.039

0.002 0.255 0.130 0.001

0.0001

0.067 0.0001

0.068 0.003 0.056

P value log rank test

CONTRACEPTION

The pattern of bleeding abnormalities by treatment group as recorded on the menstrual diary cards paralleled the rates of discontinuation for bleeding complaints and the overall rate of bleeding complaints. By all categories of bleeding abnormalities, there were significant differences among the women in the different treatment groups, with the two levonorgestrel preparations and the mestranol 5Opg + norethisterone lmg having, in that order, the least disruption of the bleeding patterns (Table III). During the first three cycles, significantly more women receiving the levonorgestrel-containing preparations had normal bleeding patterns. The preparations containing 5Opg of mestranol or ethinyl estradiol and lmg of norethisterone or norethisterone acetate had a somewhat less frequent incidence of normal cycles. By the 10th to 12th cycle, the differences between these latter four preparations completely drsappears, whereas the differences persrsted between these four and the preparations containing 2Opg ethinyl estradiol + lmg norethisterone acetate and the 35pg ethinyl estradiol + 4OOpg norethisterone acetate (Table III). DISCUSSION With few exceptions, most published reports of clinical trials of steroidal contraceptives have been limited to the description of the effectiveness, continuation rates, clinical side-effects and bleeding patterns associated with either a single preparation or the same steroidal combinations in different doses or ratios (6, 7). At the same time, there has been in the developing world the impression that the published reports contrast with the rarely published experiences within the family planning programmes of developing countries in terms of effectiveness, continuation rates, disturbances of bleeding patterns and other side-effects. Although not comparable, the one-year life table pregnancy rates in the current studies is higher than might be expected on the basis of the published In published reports of different ratios and doses of estrogen Pearl rates. and progestogen in combined oral contraceptives, there appears as much as a ten-fold difference in pregnancy rates, though the hrghest rates reported are experience. Obviously, no greater than a 1.5 pregnancies/100 women-years many factors, including statistical chance, contribute to differences in In our own studies, the pregnancy rates vary widely in pregnancy rates. different centres, being as low as none among 208 women in one centre, to as The nature of the participating high as 20 among 233 women in another. centre, the selectivity in case recruitment and the degree of compliance with pill-taking regimens appear to have contributed to the variation between centres In pregnancy rates. Despite the higher pregnancy rates in our own studies, we can conclude that of the preparations studied, none was signrficantly different from any of the others, in terms of efficacy In preventing undesired pregnancies. However, the family planning administrator is also concerned with undesired pregnancies occurring after a woman discontinues oral contraceptive In this instance, use due to the contraceptive side-effects or complications. the preparation with 2Opg of ethinyl estradiol and that with 4OOpg norethisterone acetate can be considered as unsatisfactory because of the higher discontinuation rates. The former preparation was shown to be SignificantLy less acceptable and effective than ethinyl estradiol 3O$rg + levonorgestrel 15Opg in a controlled clinical trial reported from the United Kingdom(Z). Mestranol is an effective estrogen only after conversion to ethinyl estradiol. As measured by various end points such as changes in vaginal

MARCH 1982 VOL. 25 NO. 3

l-3*** lo-12*** 19-21***

NOTE:

80.4 83.6 87.5

72.6 86.6 87.1

7.6 19.8 2.2 10.1

315 167 85

288 160 68

312 150 64

11.1 27.1 4.5 11.7

(cycles l-3)

64.9 78.1 92.7

314 172 70

20.0 42.6 10.6 14.8

46.2 59.3 68.8

EE 0.02 NAc 1.0

*** = Pd

0.001

Longest menstrual segment greater than 35 days and not Over 60 days Shortest complete menstrual segment less than 24 days Shortest complete menstrual segment less than 24 days and longest segment greater than 35 days Bleeding or spotting episode longer than 7 days

Bleeding:

317 165 80

8.6 20.3 1.3 5.7

PATTERNS

71.1 83.2 88.2

PATTERNS

EE 0.05 NAc 1.0

CALENDARS

BY CYCLE SEGMENT+

BLEEDING

NUMBER OF WOMEN OBSERVED

6.9 12.3 1.3 4.7

Conditions are not mutually exclusive + The number of wxnen observed includes only those who completed bleeding calendars for the entire 3rycle segment

Prolonged bleeding:

Infrequent bleeding: Frequent bleeding: Irregular bleeding:

Mu 0.05 NET 1.0

BASED ON BLEEDING

EE 0.03 LNG 0.15

PERCENT OF WOMEN WITH ABNORMAL

of Abnormal

l-3 10-12 19-21

Definitions

Cycles:

Infrequent bleeding*** Frequent bleeding** Irregular bleeding*** Prolonged bleeding**

Cycles:

EE 0.05 LNG 0.15

IRREGULARITIES

PERCENT OF WOPIEN WITH NORMAL BLEEDING

BLEEDING

Table III

302 155 67

11.6 46.0 5.6 15.8

45.4 65.2 65.8

EE 0.035 NAc 0.4

CONTRACEPTION

smear,

suppression of ovulation and follicle stimulating hormone, and effects on lipid metabolism, mestranol compared in equal doses to ethinyl estradiol is considered to vary in equivalence from 1 to 0.6(8). In having either less or equivalent activity than ethinyl estradiol, the lower pregnancy rate and discontinuation rates for MES/NET compared with EE/NAc was unexpected and difficult to explain. They may just represent statistical variation, although in one recent report bleeding patterns have been shown to differ when MES and EE preparations have been compared(9). Even in instances where the discontinuation rates for medical reasons were similar, the symptom complaint associated with discontinuation differed between the combined oral contraceptives containing levonorgestrel and those containing norethisterone or norethisterone acetate. If in a population, women do not tolerate bleeding disturbances or headaches, citing these symptoms as the major reason for discontinuation of the combined oral contraceptives, then the family planning administrator might expect a levonorgestrel-containing oral contraceptive to be more satisfactory, whereas if gastro-intestinal complaints are more often the cause of discontinuation, then norethisterone preparations might be preferred. The higher incidence of normal bleeding cycles associated with duration of oral contraceptive use is only partially accounted for on the basis of women with bleeding disturbances discontinuing in the trial as evident from the persistently lower rate of normal cycles after 19 to 21 cycles among those receiving ethinyl estradiol 2Oug + norethisterone acetate lmg and ethinyl estradiol 35pg + norethisterone acetate 4OOug. The persistently higher rate of bleeding disturbances of these latter two preparations throughout the trial, in contrast to the other preparations, suggests that there is an optimal dose-ratio of estrogen and progestogen to obtain the least disruption of bleeding cycles(6,7).

240

MARCH 1982 VOL. 25 NO. 3

CONTRACEPTION REFERENCES

1.

George Washington University Medical Population Reports: million users. 1 (1974)

Center. Oral contraceptives Oral contraceptives, Series

2.

trial of two Bounds W, Vessey M 6 Wiggins P. A randomized double-blind low-dose combined oral contraceptives. British J of Obstet Gynaecol, 86: 325-329 (1979)

3.

Cox DR. Regression models and life Society B, 34: 187-220 (1972)

4.

Asen SP, Roy S, Pike MC, Casagrande J&Mishell for the statistical evaluation of intrauterine J. of Obstet and Gyn, 128: 329 (1977)

5.

Slocumb JC, Kunitz oral contraceptives 243-247 (1979)

6.

Arnt IC, Ferrari A, Natal Sartoretto J & Woutersz TB . Low-dose combination oral contraceptives: A controlled clinical study of three different norgestrel-ethinyl estradiol ratios. Fertility and Sterility, 28: 549-553 (1977)

7.

Lawson JS, Yuliano SE, Pasquale SA 6 Oterman JJ. Optrmum dosage of an oral contraceptive: A report from the study of seven combinations of norgestimate and ethinyl estradiol. American J of Obstet and Gyn, 134: 315-320 (1979)

a.

The relative potency of mestranol to ethinyl estradiol Dorfman RI. evaluated by various estrogen end points in volunteer women. An unpublished review prepared for the World Health Organization (1975)

9.

Goldzieher JW, Brandon Chenault C, Woutersz TB, Schneider BE 6 effects of ethinyl estrogens, used with and without Hansen PR. Clinical progestational agents, on bleeding patterns, weight and blood pressure. Contraception, 22: 369-381 (1980)

tables.

Journal

Royal

- 50 A, No

Statistical

DR Jr. A new procedure contraception. American

SJ 6 Odoroff CL. Complications with use of IUD and among Navajo women. Public Health Reports 94 (3):

MARCH 1982 VOL. 25 NO. 3

241