Menstrual patterns and fecundity among non-lactating and lactating cycling women in rural highland Bolivia: implications for contraceptive choice

Contraception 62 (2000) 181–187

Original research article

Menstrual patterns and fecundity among non-lactating and lactating cycling women in rural highland Bolivia: implications for contraceptive choice Virginia J. Vitzthuma,b,*, Hilde Spielvogelc, Esperanza Caceresc, Julia Gainesa a

Department of Anthropology, Binghamton University, State University of New York, Binghamton, NY 13902-6000, USA Institute for Primary and Preventative Health Care, Binghamton University, State University of New York, Binghamton, NY 13902-6000, USA c Bolivian Institute of High Altitude Biology, La Paz, Bolivia

b

Received 13 July 2000; accepted 7 September 2000

Abstract Choosing an appropriate contraceptive method, particularly one based on fertility awareness, depends in part upon the degree of a woman’s cycling regularity. However, while the suppressive effect of lactation on ovarian function is well established, the potential influence of continued breastfeeding on menstrual patterns once post-partum cycling has resumed is largely unexamined. This longitudinal study in a population of non-contracepting Aymara women (n ⫽ 191 providing 665 non-truncated menstrual segments) identified conceptions and fetal loss via urine tests for hCG and classified segments accordingly to test the hypotheses that (1) cycles in lactating women are significantly different in length and regularity from those of non-lactating women, and (2) cycles in women living at high altitude are significantly different in length and regularity from those at lower altitudes. Analyses found that segments are significantly longer and regularity tends to be less common among lactating than non-lactating women; however, the rate of conception is twice as great among the former than the latter, and the distribution of conception outcomes does not differ between the two groups. Menstrual regularity is not the norm in this population, even among those who are not currently breastfeeding. High altitude per se does not appear to influence menstrual cycling as both regularity and segment length are comparable to other populations, hence women indigenous to high altitude are suitable candidates for a wide variety of contraceptive choices. In addition, these findings suggest that studies of fecundability limited to non-lactating women may be biased towards those of relatively lower fecundity. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Menstruation; Contraception; Fecundity; Cycle regularity; Breastfeeding

1. Introduction Bolivia has the second highest national total fertility rate (TFR) in South America (4.2 births per woman, Paraguay ⫽ 4.4 births per woman) and a much higher TFR (6.4) in its rural regions [1]. To reduce the infant and maternal mortality rates associated with short birthspacing, in recent years, there has been a concerted effort to improve family planning services. For cultural reasons as well as economic and infrastructural limitations, “natural” methods based upon fertility awareness are very attractive to Bolivian couples. As of 1998 [1], of those reporting current use of contraception (31.4%) the greatest percentage rely on the “rhythm” * Corresponding author. Tel.: ⫹1-607-777-4450; fax: ⫹1-607-7772477. E-mail address: [email protected] (V.J. Vitzthum).

method (41.4%). In rural settlements (⬍2000 inhabitants), where 32% of the national population of 7.9 million reside, more than half (51.8%) of current contraceptors use the rhythm method. The second most popular method, the IUD, is used by only 22.3% of all contraceptors and by only 16.6% of rural contraceptors. Successful use of the rhythm and other fertility awareness methods depends in part upon the regularity of a woman’s cycle. As in many cultures, breastfeeding among Bolivian women typically continues for many months while menstruating. The median duration of post-partum amenorrhea is reported to be 10.2 months whereas that of breastfeeding is 17.6 months [1]. Although lactation is well documented to suppress ovulation [2], there has been little study of the possible effect of sustained breastfeeding on menstrual patterns once post-partum cycling has resumed. Current WHO guidelines [3] recommend that breastfeeding be continued

0010-7824/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 0 1 0 - 7 8 2 4 ( 0 0 ) 0 0 1 6 4 - 5

182

V.J. Vitzthum et al. / Contraception 62 (2000) 181–187

through at least the first, if not second, year post-partum irrespective of menstrual resumption. Thus, whether or not menstrual patterns differ among breastfeeding women, who may comprise half or more of all menstruating women in some populations, is central to providing appropriate health care and contraceptive advice. In addition, data from women in industrialized countries indicate that intrapopulational variation in menstrual patterns is related to a host of environmental factors including diet and activity levels [4]. It also may be the case that regional differences in these factors have resulted in interpopulational variation in menstrual cycling. For example, women in a Papua New Guinea tribe are estimated to have a median cycle duration of 36 days, the longest thus far reported [5]. The hypoxic conditions that typify high altitude settings have been hypothesized by several authors [6] to negatively impact reproductive functioning. The altiplano (a mountainous plain at altitudes of 4000 meters and greater) is home to 43% of Bolivians; an additional 30% reside in the valle regions where 48.4% of the population are rural residents and many rural communities are located as high as 3500 meters [1]. Peru, Chile, Ecuador, and Argentina also have significant proportions of their populations living at higher elevations as do Ethiopia, Nepal, and Tibet. Until the current study, no data on menstrual cycles in women indigenous to high altitude had been collected to either confirm or refute the hypothesized effects of hypoxia on cycle characteristics. Significant ecologically influenced variation in menstrual patterns would have implications for contraceptive choice, particularly for those methods dependent upon cycling regularity. Here we report on menstrual patterns in a sample of non-contracepting Aymara women (n ⫽ 191 providing 665 non-truncated segments) residing at 4000 m. As the first prospective longitudinal study of menstrual patterns in a population indigenous to high altitude, these data test the hypotheses that [1] cycles in lactating women are significantly different in length and regularity from those of nonlactating women, and [2] cycles in women living at high altitude are significantly different in length and regularity from those at lower altitudes.

2. Methods and materials Project REPA (Reproduction and Ecology in Provincı´a Aroma) is a multidisciplinary longitudinal study of reproductive functioning and health among rural Aymara families indigenous to the Bolivian altiplano. Preliminary studies began in 1989, with more than two years of continuous fieldwork from 1995 to 1997 and additional data collection through 1999. Representing more than 80% of the eligible participants, 316 women, 19 – 40 years of age and currently in stable sexual unions, were recruited during 12 months beginning in November 1995 from 30 communities, principally dependent upon agropastoralism for subsistence, scat-

Table 1 Segment types based on hCG tests (n ⫽ 697) No.

Description

Code

N

1 2 3 4 5 6 7 8

Non-conception/non-post-fetal-loss Likely post-fetal-loss Post-fetal-loss Likely conception ⫹ likely fetal loss Conception ⫹ fetal loss Conception ⫹ term birth Conception ⫹ induced abortion Conception ⫹ unknown outcome

NC/NPFL LPFL PFL LC CFL CT CA CU

612 13 5 20 10 23 1 13

tered over approximately 200 square km situated about half-way between La Paz and Oruro. All study protocols were approved by the IRB at the University of California, Riverside. Of the 316 women, 125 were pregnant and/or lactating and non-cycling throughout the study’s duration; the remaining 191 contributed data on their patterns of menstruation. Of these, 93 were not breastfeeding and 98 were lactating at the time of the first observed menstrual segment. Of the latter, the subsample sizes for categories of time postpartum were 5 (⬍6 months), 27 (6 to ⬍12 months), 35 (12 to ⬍18 months), 20 (18 to ⬍24 months), 7 (⬎24 months), and 4 (unverifiable). Adopting WHO guidelines [7], menstrual intervals are referred to as segments rather than cycles to avoid the presumption of normal cycling, and segment length is defined as the first day of menses up to and including the day before the subsequent menses. A menses episode is the period in days from the first through the final appearance of blood during waking hours. Throughout participation, cycling women were visited every other day by a bilingual (Aymara-Spanish) female promotora to record menstrual status and collect a saliva sample for later assays of progesterone and estradiol [8]. Beginning at cycle day 24/25, a urine sample was collected to detect hCG, evidence of a conception, using StanBio QuPID kit (sensitive at 25 mIU/mL hCG, Stanbio Laboratory, San Antonio, TX, USA). Sample collection continued until the next menses or, if positive, until evidence of a fetal loss (two sequential negative tests) or the sixth month of gestation. From 1 to 8 segment initiations were observed for each participant. Women who were lactating but non-cycling at the time of initial recruitment were visited weekly until the first postpartum menses, then followed as described above. The sample of 191 women reported 853 segment initiations of which 837 completed menses episodes were recorded. Of these, 665 completed (non-truncated) segments were observed, the remainder having been lost to: conceptions to term (23), to an unknown outcome (13), or to induced abortion (1); termination of observation (112); and participant withdrawal (23). Completed segments were classified according to the observed outcome of the urine test for hCG (Table 1; six segments could not be classified). Two or more sequential positive tests, separated by two days, followed by negative tests was classified as a clear

V.J. Vitzthum et al. / Contraception 62 (2000) 181–187 Table 2 Selected sample characteristics (n ⫽ 191)

Age (years) Education (years) Menarche (years) Age at first birth (years) Parity

183

3. Results

Median

Mean

SD

Range

30 5 14 20 4

29.4 5.0 14.2 20.0 4.0

4.9 2.9 1.5 3.0 2.3

20–38 0–13 10–19 15–31 0–11

conception and fetal loss (Type 5). A single and/or faint positive test followed by a negative test was classified as a likely conception with a presumed fetal loss (Type 4). Segments following a clear (Type 3) or presumed fetal loss (Type 2) were distinguished from those neither following a fetal loss nor in which a detectable conception occurred (Type 1). Statistical analyses were carried out with SPSS for Windows (Version 10.0). For comparison with previous studies, descriptive statistics are given for a sample of all segments. However, the inclusion of multiple segments from women who did not conceive during the study may bias the sample towards women of lower fecundity and, perhaps, aberrant segments. Furthermore, within such a sample not all data points are independent, confounding statistical analyses. Therefore, units of analyses for statistical comparisons are the mean or median of each woman’s segments, and a sample comprising a single non-truncated segment from each woman (either the first observed Type 2 or 3 segment, or if none, the first observed Type 1 segment).

3.1. Sample characteristics Sample characteristics are shown in Table 2. Mean age was 29.3 years, and all but nine women in the sample were between 22 to 36 years of age, a period during which the probability of an ovulatory cycle is relatively high and constant [9]. Mean recalled age at menarche was 14.1 years, a value roughly midway in the ranges reported for other samples from high altitude [10]. Mean age at first birth was 20 years and mean parity was 4; thus, women average four live births in the first 9 years of childbearing. While age at menarche is likely to be subject to recall bias and should be viewed cautiously, age at first birth and parity were crossed check by multiple methods and are likely to be accurate. 3.2. Segment type and length Segment types (Table 3) vary in mean length whether the sample comprises all segments or a single segment from each woman (for all non-conception [Types 1–3] segments: median test, ␹2 ⫽ 6.501, p ⫽ 0.039). Segments in which conception and fetal loss are likely to have occurred (Type 4) are predictably the longest (36.4 days; vs Type 1 ⫽ 28.5 days, n ⫽ 142: median test, ␹2 ⫽ 7.429, p ⫽ 0.006). In addition, segment variability is far greater among segment Types 2– 4, with a far longer right tail, than in Type 1 segments. Type 1 segments average 28.5, 28.5, and 29.1

Table 3 Segment length in non-truncated segments Segment typea and sample composition Types 1–4 All segments One segment per woman (types 1–3)b Individual means Individual means: non-lactatingc Individual means: lactating 1 (NC/NPFL) All segments One segment per woman Individual means Individual means: non-lactatingd Individual means: lactating 2 (LPFL) All segments One segment per woman 3 (PFL) All segments/1 per woman 4 (LC) All segments One per womane a

N

Mean

649 162 162 83 74

28.9 28.9 29.4 28.6 30.4

612 159 159 82 72

Range

Median

5.6 5.3 4.2 3.4 4.8

13–76 17–61 20.3–53.3 23.4–43.0 20.3–53.3

28 28 28.3 27.7 30.0

28.5 28.6 29.1 28.4 29.9

4.8 4.4 3.4 3.2 3.6

13–76 17–49 20.3–43.5 23.4–43.0 20.3–43.5

28 28 28.3 27.7 29.8

13 11

34.5 32.5

12.2 9.9

25–62 25–61

30 30

5

33.8

10.1

25–51

31

19 18

36.1 36.4

12.5 12.8

22–71 22–71

31 31

Segment types defined in Table 1. Types 1 vs. 2 vs. 3 (one segment per woman): median test, ␹2 ⫽ 6.501, p ⫽ 0.039. c Non-lactating vs. lactating (types 1– 4): 2 sided t-test, t ⫽ 2.697, p ⫽ 0.008. d Non-lactating vs. lactating (type 1): 2 sided t-test, t ⫽ 2.586, p ⫽ 0.011. e Types 1 vs. 4 (one segment per woman): median, ␹2 ⫽ 7.429, p ⫽ 0.006. b

SD

184

V.J. Vitzthum et al. / Contraception 62 (2000) 181–187

Table 4 Regularity of segments Minimum # of observed segments

All non-conception segment typesa n

% Predictable b

Total 3 4 5 Non-lactatersd 3 4 5 Lactatersd 3 4 5

regular

semi-regular

122 91 69

35 32 35

64 47 35 57 44 30

Type 1 segments only

% Non-regular c

n

of total

of predictable

39 34 23

26 34 42

53 52 40

39 40 42

39 29 22

22 31 36

32 23 27

39 39 20

29 38 53

% Predictable

% Non-regular

regular

semi-regular

of total

of predictable

109 78 61

38 35 36

40 33 23

22 32 41

51 49 39

50 42 34

62 44 34

40 41 41

39 30 24

21 29 35

49 42 37

55 63 43

46 34 24

35 27 29

41 38 17

24 35 54

54 58 37

a

Segment types defined in Table 1. Regular: all segments are 26 –32 days long. c Semi-regular: all but one segment are 26 –32 days long. d Includes only those maintaining lactation status throughout observation. b

days for the all-segments, single-segment, and individualmeans samples. The variance (sd2) is greatest for the allsegments sample (23.04) and approximately halved for the sample of individual means (12.25), reflecting the substantial between-woman as well as within-woman variation in segment length. Segment length was not found to be associated with either current age or age at menarche. 3.3. Non-lactating vs. lactating cycling women Based on analyses of the sample of individual means, whether considering the all-segments sample (Types 1– 4) or a sample of only Type 1 segments, segment length was significantly longer in lactating than in non-lactating women (28.6 vs. 30.4, p ⫽ 0.008, and 28.4 vs. 29.9, p ⫽ 0.011, respectively; Table 3). In the Types 1– 4 sample, the greater segment length of breastfeeding women reflects their relatively greater number of conceptions and fetal losses compared to the non-lactating women. As would be expected, variance of segment length was about twice as great among breastfeeding than non-breastfeeding women (23.04 vs. 11.56) in the Types 1– 4 sample and, though less, still greater in the Type 1 sample (13.69 vs. 10.24). 3.4. Regularity Regularity is shown in Table 4. For comparability, regular is defined as having all of a woman’s observed segments of length 26 –32 days inclusive [10]. Of 191 women, 122 were observed for at least three non-conception nontruncated segments, 91 for at least four, and 69 for at least five. In each of these samples, only about a third (32–35%) of the women could be considered regular. Defining predictable as those having no more than a single segment

outside the range of 26 –32 days, 58 –74% of the women met this criteria (depending upon the number of segments observed), the sample of five observed segments having the lowest predictability. Among breastfeeding women, regularity and predictability were less common than among non-lactating women; however, the differences are not significant. If only Type 1 segments are considered, virtually identical patterns and proportions are observed. Similarly, if the range of days defining regularity is 3 days ⫾ 28 days (observed mean for the non-lactating sample) or 30 days (mean for lactating sample), there was no appreciable change in the frequencies of either regularity or predictability. 3.5. Fecundity Fecundity is shown in Table 5. The rate of conceptions was nearly twice as high among lactating cycling women than non-lactaters (46 of 328 segments vs. 21 of 369 segments, ␹2 ⫽ 13.88, p ⬍0.0001). However, there were no differences between the two groups in the distribution of conception outcomes, that is, fetal loss is no more likely among lactating than non-lactating women, nor was the time to fetal loss significantly different (55.0 vs. 56.7 days). These findings are contrary to the opinion that lactation increases the risk of loss of the new conceptus, a belief that may lead some women to be more “casual” about contraception while nursing than is advisable to avoid pregnancy. The mean and median post-partum timing of conceptions among lactating cycling women was 472.5 days (SD ⫽ 160.0) and 474 days, respectively, ranging from 151 to 940 days. Of these conceptions, as might be expected, a significantly (p ⬍0.03) greater proportion occurred in those beyond compared to those within one year postpartum (34

V.J. Vitzthum et al. / Contraception 62 (2000) 181–187

185

Table 5 Distribution of conceptions and outcomes in non-lactating and lactating cycling women Number of segmentsa

Lactation status

Non-lactating cycling Lactating cycling Total

Conception outcomesb

Total

Non-Conception

Conception

Fetal lossc

Term

Unknown

369 328 697

348 282 630

21 46 67

8 22 30

8 15 23

5 8 13

Distribution of conceptions in non-lactating vs. lactating women: ␹2 ⫽ 13.88, p ⬍0.0001. Distribution of conception outcomes: Non-significant differences (␹2 ⫽ 0.725). Analysis does not include one abortion. c Non-significant difference in days to fetal loss (non-lactating vs. lactating: 43.0/56.7/39.8 vs. 32.5/55.0/48.3 for median/mean/SD, respectively). a

b

conceptions among 62 women vs. 11 among 32, respectively). Conception outcomes, however, were not associated with the post-partum timing of conception nor is the average time to fetal loss significantly different between lactating women conceiving before compared to after one year postpartum (43.0 vs. 58.5 days).

4. Discussion Regardless of the unit of analysis, mean and median segment length in these Aymara women is about 29 and 28 days, respectively. Eliminating those segments in which a putative fetal loss occurred, or which followed a fetal loss, does not appreciably change these estimates, reflecting in part the relatively low proportion of these segments in the total sample (34 of 162 women, 37 of 649 segments) and the fact that their mean duration, though significantly longer than that of Type 1 segments, is well within the normal range of the latter. What does vary markedly among the

samples is the estimate of variance, ranging from 12.25 in the sample of individual means of Type 1 segments to 31.36 in the sample of all Types 1– 4 segments. This demonstrates the sensitivity of estimates of population variance of segment length to the unit of analysis and the inclusion of other than Type 1 segments, by necessity the practice in most inquiries. With these caveats, Table 6 lists those populations for which roughly comparable data on segment length is available in the literature. Despite the differences of ecology, lifestyle and nutritional status, this Aymara sample falls neither particularly low nor high in the range of observed segment lengths. Only about one-third of the Aymara women could be considered as having regular cycles based on the criterion proposed for a study of Mayan women [11], whether 3, 4, or 5 segments have been observed or whether assessment of regularity is restricted to Type 1 segments. That irregularity does not appear to increase with a greater number of observations suggests that there is a core set of women for whom regularity is the norm and for whom contraceptive methods

Table 6 Segment lengths in days in several populations # Cyclesa

Sample description Canada/US [12]

Denmark [13] Denmark [14] Guatemala [11] India (Bengalee) [15] South India [16]

Mali [17] New Guinea [5] Switzerland [18] UK [19]

US [20]

a b

20–24 years 25–29 years 30–34 years 20–35 years 25–34 years 18–39 years Adult 20–24 years 25–29 years 30–34 years 15–53 years 18–44 years 20–40 years 21–25 years 26–30 years 31–35 years 20–24 years 25–29 years 30–34 years

Units of analysis are italicized. Mean of medians for all cycles at each year of age.

3,320 3,412 6,691 1,061 880 782

400 37 31,644 2,606 4,387 5,224 1,494 3,078 3,762

# Womena 257 266 505 295 434 301 110 583 699 413 54 36

Median

Mean

SD

30.5 29.4 28.8

3.99 6.34 6.53

28.4 30.4 30.1 32.3 31.2 31.1

2.6 3.97

29.3 28.8 28.4 29.2 28.7 27.9

4.15 3.94 3.92 3.69 3.57 3.55

29

28.5 36 27.8b 29 28 28

4.9 3.9 4.0

186

V.J. Vitzthum et al. / Contraception 62 (2000) 181–187

requiring cycle regularity would be an appropriate choice. Nonetheless, the majority of women cannot be considered to have regular cycles, a finding that emphasizes the importance of carefully assessing a woman’s regularity in the course of deciding upon a contraceptive and offering viable alternatives should the criterion of regularity not be met. In addition, given that only one-third of this sample could be considered regular, it is not unlikely that some proportion of the Bolivian couples reporting use of the rhythm method (41.4% of current contraceptors) are not ideal candidates for this method. In the study of Mayan women, 54.5% of those reporting regularity proved to be irregular during three observed cycles and were possibly at increased risk for conception. A similar lack of recognition of irregularity may be the case among Bolivian women, putting some women at undue risk of pregnancy and also undermining confidence in a method that works well for couples meeting the criteria for use. There is little comparative data on regularity available in the literature. Among Mayan women, regularity was estimated for those who had reported having regular cycles. To approximate such a sample here, predictability for the Aymara women was defined as having no more than one observed segment outside the range of 26 –32 days; women who are predictable might be expected to report themselves as regular. For the sample of three observed segments, as was observed in the Mayan study, 74% might be expected to have reported themselves as regular. Of these, 53% would have proved to be irregular, very comparable to the 54.5% irregularity observed among those Mayan women who had reported themselves as regular. This suggests that while regularity is not the norm among Aymara women, neither does it characterize an unusually low proportion of the women, at least compared to Mayan women. Among lactating women, segment length is significantly longer by about two days than among non-lactaters, whether considering all or only Type 1 segments. Menstrual regularity also tends to be non-significantly lower among lactaters. However, the greater length and possibly lower regularity is not associated with a reduction in fecundity. Rather, the rate of conceptions is approximately twice as high among the lactating women with no differences from non-lactating women in proportion of fetal loss or time to fetal loss. This finding does not contradict the well-established fact that women who breastfeed on demand are at a lower risk for the return of post-partum menses. Rather, the finding here that a greater proportion of the conceptions occurring in lactating women is beyond rather than before one year post-partum argues that even upon menstrual resumption, with continued breastfeeding fecundity is relatively lower, increasing over time post-partum. The findings presented do suggest, however, that there is a higher proportion of more fecund women among the sample of lactating than non-lactating cycling women. This is to be expected in a non-contraception-using population, because those with relatively higher natural fecundity are

more likely to become pregnant and subsequently breastfeed. Two important implications derive from this observation. First, while it remains important to promote continued breastfeeding for the health of infant and mother (assuming adequate nutritional resources for both), it is essential that the contraceptive needs of these women not be less than fully recognized because they are lactating. The somewhat lower proportion of regularity (23%–32%) coupled with the possibility of higher fecundity compared to non-lactating counterparts argues that contraceptive choices other than the rhythm method must be available, along with adequate counseling. Because fertility awareness methods are the most easily shared of contraceptives, adequate dissemination of information regarding the criteria for use is critical. Second, investigations of fecundity, fecundability, and contraceptive efficacy that limit their samples to non-lactating women may be inadvertently biased towards women of lower fecundity (i.e., they are not breastfeeding because they were less able to conceive). Hence, such samples are not representative of the total population of women of reproductive age and estimates of demographic and reproductive variables may be biased. Turning to the question of whether hypoxic conditions negatively impact characteristics of the menstrual cycle in women indigenous to high altitude, the data here do not support this hypothesis. Segment length is neither unusually long nor short in this Aymara sample, and cycling regularity appears to be no less than that observed among Mayan women at far lower altitudes. It is worth noting that the assumption that altitude is the most significant ecological difference between the Aymara, or any other high altitude population, compared to other groups is almost certainly false. The Bolivian altiplano is characterized by cold and harsh winds, high solar radiation, seasonal aridity and limited agricultural output, poverty and poor nutritional intake, and a general absence of electricity, fuel, clean water, proper sanitation, paved roads, and adequate health care. At least with regard to the characteristics of the menstrual cycle examined here, Aymara women appear to be adapted to these conditions and are no more or less suitable candidates than others for high quality counseling and a wide variety of contraceptive choices. Acknowledgments Our appreciation to the study participants, to Dr. I. Sinai for her helpful commentary, and to Kate Loran for assistance in manuscript preparation. Supported by NSF Grant SBR-9506107 and Binghamton University, SUNY. References [1] Instituto Nacional de Estadı´stica, Ministerio de Hacienda, Bolivia. Encuesta Nacional de Demografı´a y Salud 1998. Bolivia, December 1998.

V.J. Vitzthum et al. / Contraception 62 (2000) 181–187 [2] Contraceptive efficacy of lactational amenorrhoea. Lancet 1992;339: 227. [3] World Health Organization (WHO). The World Health Organization’s infant feeding recommendation. WHO Weekly Epidemiological Record 1995;17:117–220. [4] Cumming DC, Wheeler GD, Harber VJ. Physical activity, nutrition, and reproduction. Ann NY Acad Sci 1994;709:55–76. [5] Johnson PL, Wood JW, Campbell KL, Maslar IA. Long ovarian cycles in women of highland New Guinea. Hum Biol 1987;59:837– 45. [6] Wiley AS. The ecology of low natural fertility in Ladakh. J Biosoc Sci 1998;30:457– 80. [7] Snowden R. Patterns and Perceptions of Menstruation. St. Martin’s Press, NY, 1983. [8] Vitzthum VJ, Bentley GR, Caceres E, et al. Correlates of conception in high altitude Aymara women. FASEB 1998;12(5):A726. [9] Lipson SF. Ellison PT. Normative study of age variation in salivary progesterone profiles. J. Biosoc Sci 1992;24:233– 44. [10] Greksa LP. Age of menarche in Bolivian girls of European and Aymara ancestry. Ann Hum Biol 1990;17:49 –53. [11] Burkhart MC, de Mazariegos L, Salazar S, Hess T. Incidence of irregular cycles among Mayan women who reported having regular cycles: implications for fertility awareness methods. Contraception 1999;59:271–5.

187

[12] Chiazze L Jr, Brayer FT, Macisco JJ, Parker MP, Duffy BJ. The length and variability of the human menstrual cycle. JAMA 1968; 203:89 –92. [13] Kolstad HA, Bonde JP, Hjøllund NH, Jensen TK, Henriksen TB, Ernst E, Giwercman A, Skakkebæk NE, Olsen J. Menstrual cycle pattern and fertility: a prospective follow-up study of pregnancy and early embryonal loss in 295 couples who were planning their first pregnancy. Fertil Steril 1999;71:490 – 6. [14] Mu¨nster K, Schmidt L, Helm P. Length and variation in the menstrual cycle—a cross-sectional study from a Danish county. Br J Obstet Gynaecol 1992;99:422–9. [15] Datta N. Influence of seasonal variations on the reproductive cycles in women. Popul Rev 1960;4:46 –55. [16] Jeyaseelan L, Rao PSS. Correlates of menstrual cycle length in South Indian women: a prospective study. Hum Biol 1993;65:627–34. [17] Strassmann BI. The biology of menstruation in Homo sapiens: total lifetime menses, fecundity, and nonsynchrony in a natural-fertility population. Curr Anthropol 1997;38:123–9. [18] Vollman RF. The menstrual cycle. Philadelphia, PA: WB Suanders, 1977. [19] Monari P, Montanari A. Length of menstrual cycles and their variability. Genus 1998;54:95–188. [20] Bailey J, Marshall J. The relationship of the post-ovulatory phase of the menstrual cycle to total cycle length. Biosoc Sci 1970;2: 123–32.