Portal de Periódicos da CAPES

Advanced Maternal Age and Offspring Outcomes: Reproductive Aging and Counterbalancing Period Trends

2016; Wiley; Volume: 42; Issue: 1 Linguagem: Inglês

10.1111/j.1728-4457.2016.00105.x

1728-4457

Kieron Barclay, Mikko Myrskylä,

Intergenerational Family Dynamics and Caregiving

Population and Development ReviewVolume 42, Issue 1 p. 69-94 ARTICLEOpen Access Advanced Maternal Age and Offspring Outcomes: Reproductive Aging and Counterbalancing Period Trends Kieron Barclay, Kieron BarclaySearch for more papers by this authorMikko Myrskylä, Mikko MyrskyläSearch for more papers by this author Kieron Barclay, Kieron BarclaySearch for more papers by this authorMikko Myrskylä, Mikko MyrskyläSearch for more papers by this author First published: 08 April 2016 https://doi.org/10.1111/j.1728-4457.2016.00105.xCitations: 63AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Women and men in the developed world are having children at later ages. Since 1970 the mean age at first birth has increased in each of the 23 OECD countries for which data are available, at a rate of 0.08 years per calendar year, and now averages 28 years. Over the period 1995–2011, postponement of childbearing has been increasing faster, at 0.10 years per calendar year. In Germany and the UK, the mean age at first birth exceeds 30 years (OECD 2014). Advanced-age fertility has also been increasing: in Sweden in 2013, a quarter of all births were to mothers aged 35 or older. The potential consequences of postponement are numerous, including decreasing period fertility (Bongaarts and Feeney 1998) and negative health outcomes for children as a result of reproductive aging (Jacobsson, Ladfors, and Milson 2004). Although parental socioeconomic resources typically increase with age (Powell, Steelman, and Carini 2006), advanced maternal age is associated with increased risks of Down syndrome, childhood cancer, and autism (Durkin et al. 2008; Johnson et al. 2009; Yip, Pawitan, and Czene 2006). The research documenting these negative child outcomes, however, neglects the potential benefits of being born at a later date. For many important outcomes such as health and educational attainment, secular trends across the OECD countries are positive, so being born into a later birth cohort would appear to be beneficial. We illustrate this proposition using data from Sweden and a sibling–comparison design. We show that the macro-level trends outweigh the individual-level risk factors. In the process, we find that fertility postponement even up to maternal ages above 40 is associated with positive long-term outcomes for children. These results are likely to extend to other countries where health has been improving and educational access has been expanding, such as the United States and much of Europe. Figure 1 illustrates the changing patterns in fertility timing in Sweden. In 1968, approximately 75 percent of all births were to mothers aged less than 30, and fewer than 10 percent of births were to mothers aged 35 or above. Over a 45-year period childbearing at later ages at all parities has become more common; by 2013 approximately 60 percent of births were to mothers aged 30 or older, and 5 percent to mothers aged 40 or older. There are many reasons for the increase in the mean of maternal age at birth over these years. Much of the fertility postponement has been attributed to the use of the contraceptive pill, the expansion of career opportunities for women, and increasing economic uncertainty (Kohler, Billari, and Ortega 2002; Sobotka 2004). In the United States attitudinal and structural changes in the 1960s and 1970s increased opportunities for women in education and the labor market, as did the introduction of oral contraception (Goldin and Katz 2002), leading to improvements in gender equality. Figure 1Open in figure viewerPowerPoint Percent of births in Sweden each year by age of mother at the time of birth, 1968–2013 SOURCE: Swedish Register data, compiled by the authors. While Sweden did not have a strong feminist movement (Gelb 1989), the country's high level of gender equality relative to other countries can largely be attributed to the fact that achieving equality has been a goal of successive governments since the 1960s (Hoem 1995). Women in Sweden today have greater educational attainment than men (OECD 2013), and the tendency to delay childbearing until completing one's education is likely to be part of the explanation for the increase in maternal age over time, as are increased career opportunities, particularly in the large public sector, and financial resources (Blossfeld and Huinink 1991). Since the 1960s in Sweden there have been a number of notable shifts in fertility and family formation behavior, collectively labeled the second demographic transition (Van de Kaa 1987; Lesthaeghe 2010). One of these has been an increase in the prevalence of less-committed relationships, which is likely to help explain why more women delay childbearing to older ages. Furthermore, Sweden's political and social system has been described as one of "statist individualism" (Berggren and Trägårdh 2006; Eklund, Trägårdh, and Berggren 2011). The country's tax and welfare systems are designed to minimize dependence on the family and enable individuals to pursue their own goals (Trägårdh 1990), making the timing of childbearing a choice that is likely to be relatively independent of familial pressures. Delayed childbearing may be desirable from a woman's own life-course perspective and beneficial for the child, especially among younger mothers for whom socioeconomic position and resources may be rising rapidly. However, fertility postponement increasingly means a rise in advanced-age motherhood rather than a decline in young-age motherhood. This is potentially alarming as there are known risks associated with childbearing at older ages, and it has been suggested that mothers may not be fully aware of these risks (Benzies 2008). Advanced maternal age is associated with a gradual deterioration of the intrauterine environment and decreased viability of embryos due to an age-dependent decrease in oocyte quality (Abdalla et al. 1993).1 These changes mean that older mothers are at higher risk of pregnancy complications. The risks of miscarriage, preterm birth, low birth weight, stillbirth, and Down syndrome increase exponentially with age (Jacobsson, Ladfors, and Milson 2004; Yoon et al. 1996; Andersen et al. 2000). Danish register data for the period 1978 to 1992, for instance, showed that 9 percent of pregnancies intended to be carried to full term to mothers aged 20–24 ended in spontaneous abortion, while the corresponding figure was 20 percent for ages 35–39 and 41 percent for ages 40–44 (Andersen et al. 2000). Research has also shown that the disadvantages for the offspring of older mothers can extend throughout adulthood. Children born to older mothers are at greater risk of Alzheimer's disease (Rocca et al. 1991), hypertension (Brion et al. 2008), diabetes (Gale 2010), cancer (Hemminki and Kyyrönen 1999), and mortality (Kemkes-Grottenthaler 2004), and those born to the oldest mothers also have lower self-rated health and are more likely to be obese (Myrskylä and Fenelon 2012). It is possible that these negative long-term outcomes are a consequence of low birth weight or pre-term birth, since not all studies have been able to adjust for those mediating factors. Research suggests that lower birth weight has a negative causal impact on height, as well as on cognitive ability in adulthood, educational attainment, and earnings (Conley and Bennett 2000; Black, Devereux, and Salvanes 2007). Research has also shown that below-average birth weight is associated with increased mortality risk in adulthood (Osler et al. 2003). While well-defined physiological mechanisms account for the relationship between advanced maternal age and poor perinatal and infant outcomes, it is unclear whether the long-term negative effects on children of being born to an older mother are causal. Recent studies suggest that the increased mortality of the offspring of older mothers in adulthood is at least partially explained by the death of parents when the offspring are younger (Myrskylä and Fenelon 2012; Myrskylä et al. 2014). A fact that has yet to receive much attention is that the age at which a woman chooses to have a child is related to period conditions. A woman born in 1960 who had a child at age 20 would have given birth in 1980. If the same woman had chosen to have a child at age 40, that child would be born in 2000. This makes a significant difference to the expected health and education of the average child, since the second half of the twentieth century has witnessed a number of secular improvements. These include improving medical and public health conditions, indicated by lower age-specific mortality and an increasing life expectancy (Oeppen and Vaupel 2002); and by increases in average height, a useful indicator of improvements in early-life conditions, of populations across the developed world (Komlos and Lauderdale 2007). The second half of the twentieth century and the beginning of the twenty-first have also been characterized by a steady expansion of educational systems across Western Europe and the United States (Breen and Jonsson 2007; Breen et al. 2009; Breen 2010). Today more people than ever continue their formal education beyond the legally defined minimal age. The expansion of access to tertiary education has been particularly striking: in the 1940s and 1950s only a small fraction of the population obtained a bachelor's degree. While we should not ignore other long-term trends such as increasing socioeconomic inequality or rising rates of obesity, the positive progress in public health conditions and educational access has extended the opportunity for longer lives and advanced learning to more people than ever before. In this study we examine the extent to which these secular improvements outweigh the disadvantages that have been shown to be associated with being born to an older mother. One previous study used a comparable research design. Myrskylä et al. (2013b) analyzed IQ at age 18 by maternal age and found that secular positive trends outweigh any potential individual aging-related outcomes, so that IQ increased monotonically with maternal age. This study, however, analyzed only men and focused on a measure of cognitive ability that has been claimed to be increasing over successive cohorts without any real gains in intelligence (Flynn 1987; Emanuelsson and Svensson 1990). It is unclear whether the same pattern would be observed for women and for outcomes for which measurement is reliably consistent over birth cohorts. For Swedish men and women born between 1960 and 1991 we show that individuals born to older mothers, including those at the oldest ages, are taller, remain longer in the educational system, are more likely to attend university, and perform better on standardized tests than their siblings who were born when their mothers were younger. Analyzing these multiple outcomes requires us to use several data sets, each of which is based on high-quality Swedish administrative register data. Our results show that in a regime characterized by improving social conditions, postponing parenthood is beneficial for children even when the individual maternal aging-related effects might be negative. These results are also likely to apply to other countries where health is improving and education is expanding. Before we present our data and results, we summarize the changes that have been taking place in Sweden with respect to education, height, and physical fitness. Education Education in Sweden is state funded at all levels, and tertiary education is free for Swedish and European Union citizens (Högskoleverket 2012). Students in tertiary education are eligible for financial support from the Swedish state for living costs in the form of study grants and low-interest student loans. The cohorts on whom we focus in this study were born between 1960 and 1982. This means that they were in secondary school between around 1976 and 1998, a period of substantial change in the Swedish educational system (Halldén 2008). Between the 1960s and 2000s, tertiary education enrollment increased substantially (Breen et al. 2009). In 2012 approximately 33 percent of the Swedish population had attained post-secondary education, slightly higher than the OECD average. This educational expansion has clearly benefited individuals born into later birth cohorts, which has implications for patterns of educational attainment by maternal age at the time of birth. Height Research suggests that taller individuals have lower mortality (Davey Smith et al. 2000a), greater health-related quality of life (Christensen et al. 2007), and superior cognitive ability (Case and Paxson 2008). Height in adulthood is strongly related to both length at birth (Sørensen et al. 1999) and height in childhood, with a correlation of approximately 0.7 (Power, Lake, and Cole 1997). Mothers of infants with greater birth weight also have lower all-cause and cause-specific mortality (Davey Smith et al. 2000b). The overall pattern suggests that healthier mothers give birth to longer infants, who retain a height advantage into adulthood and also have greater relative health themselves. Swedes have been growing taller since at least the early nineteenth century and gained approximately 10 cm between 1900 and 2000 (Gustafsson et al. 2007). A similar historical increase in height, attributable to improvements in nutrition and public health (Hatton 2013), is found in a wide range of other countries (Komlos and Lauderdale 2007), and greater stature in historical populations is also associated with lower premature mortality (Gunnell, Rogers, and Dieppe 2001). Physical fitness Physical fitness is a component of overall health. By physical fitness we mean aerobic fitness, the ability of the body to deliver oxygen to the muscles and use it to generate energy for physical activity; the most common measure of that capacity is maximal oxygen uptake (Armstrong and Welsman 2007). We use a closely correlated measure called maximal working capacity, explained below. Greater physical fitness is associated with lower mortality risk at all ages (Blair et al. 1996) and with greater self-rated health (Shirom et al. 2008). Unlike height, it is far less clear whether the physical fitness of the Swedish population has improved in recent decades. One study found that the aerobic fitness of adolescents in Sweden decreased between 1974 and 1995 (Westerstahl et al. 2003), but it is not known whether this was due to an increase in body mass index or to less daily physical activity. Other research has found that while the functional fitness of the healthiest group of adolescents was approximately the same in 2001 as it was in 1987, the fitness of the least-healthy group of adolescents has fallen substantially (Ekblom, Oddsson, and Ekblom 2004). Taking a global perspective, Tomkinson and Olds (2007) present data which indicate that aerobic fitness among 6–19-year-olds was improving from 1958 to the 1970s, but since the 1970s has been in steady decline worldwide. Data This study uses Swedish administrative register data. Because of different data availability for various outcome variables, we will study several cohort groups, which we describe in more detail below. Details on selecting the final sample for each set of analyses are given in Table S1.* The range of birth cohorts that we study is 1960–1991. In Sweden each individual has a unique personal identification number (PIN), which enables us to link the records of an individual across the various administrative registers. We draw heavily upon the Swedish multi-generational register, which contains information on each individual as well as that individual's parents. The main family members of interest are the mother, father, and siblings. We use information on the biological mother and father to identify the sibling group and use information on the biological mother to calculate maternal age at the time of birth. Our main analyses use fixed effects specified at the level of the sibling group, so the regressions identify the parameters of interest from between-siblings comparison. Given our use of sibling fixed effects, we omit only children. We also drop sibling groups with twins and other multiple birth individuals since those individuals exhibit no variation in maternal age at the time of birth. The term "cohort cut" in Table S1 refers to individuals who are lost when restricting the sample to specific birth cohorts. All of the descriptive statistics and results presented below are based on the final sample that is detailed in Table S1. Because children born to older parents may benefit from the accumulation of parental socioeconomic resources, we also adjusted for the time-varying occupational status of parents and for time-varying household income. Data on occupational status are available only from censuses prior to the 1990s, so we draw data from the 1960, 1970, 1975, 1980, 1985, and 1990 censuses. Using data on the mother and father, we categorize household socioeconomic status according to the higher of the two parents' occupational statuses. A reliable measure for parental income is available only from 1970 because of changes in how individuals and households were taxed. Our measure for parental income combines the earnings of the mother and father in the year before the index child was born, to account for the fact that parental income typically decreases immediately after the birth of a child owing to lower levels of labor market participation. The 1970s was a period of high inflation in Sweden (Edvinsson and Söderberg 2011), hence we adjust our measure of combined parental income for a measure of inflation based on the consumer price index. Educational attainment To examine educational attainment, we use data on cohorts born 1960–1982. We examine educational attainment in the year in which individuals turn 30, using two different measures. The first is the number of years of educational attainment achieved by that age. This measure is based on the number of years that correspond to the specific level of education achieved by age 30, and may not in all cases reflect the actual number of years that an individual spent in the educational system. The second measure is a binary variable indicating whether individuals had entered tertiary education by age 30. The reason for using this second measure is that not all individuals in Sweden have finished their education by age 30, but the vast majority of people who earn a bachelor's degree would have started that degree before age 30 (Högskoleverket 2012). The Swedish education system today is divided into three sections: nine years of compulsory schooling, three additional years of upper secondary education, and tertiary education (Halldén 2008). Tertiary education consists of two parts: a traditional university education and vocational tertiary education. The variable for highest educational level and the corresponding years of education required to reach that level come from the Swedish education registers and Statistics Sweden (Halldén 2008; Statistics Sweden 2000). In the analyses of educational attainment, we also adjust for birth order, since research consistently demonstrates that later-born children have lower educational attainment than first-borns (Black, Devereux, and Salvanes 2005; Barclay 2015). Grade point average at age 16 The data on grade point average (GPA) are taken from grades earned during the final year of compulsory education, at which time students are typically 16 years old. The system for assigning grades in the Swedish high school system has changed several times over the past decades, so we limit our analyses to the period 1998–2007, during which the grade system stayed constant. This means studying cohorts born 1982–1991, who were aged 16 between 1998 and 2007. During this period in the Swedish compulsory school system, grades ranged from pass with special distinction to pass with distinction, pass, or fail. To construct an overall score, each of these grades was assigned a numerical score (Skolverket 2010). The overall GPA score for each child was calculated by summing his or her grades based on the best grades in 16 subjects, and the scores ranged from 0 to 320 (Skolverket 2010; Turunen 2014). A score between 0 and 159 represents a mean mark of fail, and a score between 160 and 239 is equivalent to a mean mark of pass. In these analyses of GPA we again also adjust for birth order, as research has shown that later-born children have lower educational performance than first-borns even in high school (Kantarevic and Mechoulan 2006; Härkönen 2014). Physical fitness and height To examine physical fitness and height we use data from the Swedish military conscription register on cohorts born between 1965 and 1977. Only men were required to attend conscription tests. Our outcome measure for examining physical fitness is maximal working capacity (MWC). MWC, the maximum resistance attained in watts when riding on a stationary bicycle for a period of 5 to 10 minutes, is an important predictor of mortality among men (Sandvik et al. 1993). Height is measured in centimeters. In our analyses of physical fitness and height, we also adjust for birth order, as research demonstrates that, compared to first-borns, later-born children have lower physical fitness (Barclay and Myrskylä 2014) and are shorter (Myrskylä et al. 2013a). We also include a covariate for the age at which individuals took the conscription test, ranging from 17 to 20, to adjust for any potential differences in physical fitness or height by age. Statistical analyses We present results based upon several cohort groups and outcomes. While there are some small variations, described below, we pursue the following general strategy. Model 1 is a standard regression model (OLS or logistic) estimating a between-family comparison for the bivariate relationship between maternal age at the time of birth and the outcome variable in question. Model 2 is a fixed-effects regression model (OLS or logistic) comparing siblings within the same family to one another to estimate the relationship between maternal age at the time of birth and the outcome variable. In these models we also adjust for birth order. In the analyses using the military conscription register, we also adjust for age at the time of the conscription tests; for the vast majority (99.9 percent) of individuals eligible under our criteria (see Table S1), this is between 17 and 20. Model 3 is a fixed-effects regression model (OLS or logistic) that is the same as Model 2 except for the inclusion of a categorical variable for year of birth, in individual years. Model 2 captures the total effect of maternal age on child outcomes. This total effect includes not only the potential individual-level factors such as reproductive aging and accumulation of social resources, but also the impact of changing period conditions. For an individual, the period conditions systematically change with maternal age; thus Model 2 describes how child outcomes change with changing maternal age for an individual mother. Model 3 removes the influence of changing period conditions and estimates the net effect of maternal age. The use of fixed effects in Models 2 and 3 means that we perform a within-family comparison, comparing siblings within the same family to one another. This estimator minimizes residual confounding by inherently adjusting for all factors that are shared by the siblings and remain constant, such as parental height, parental cognitive skills, and the size of the sibling group. We demonstrate the hierarchy of our models based upon the approach for studying educational attainment measured in years by age 30: (Model 1) (Model 2) (Model 3)where yij is the measure on each of the outcome variables for individual i in sibling group j. Model 1 does not use fixed effects and is a standard OLS model performing a between-family comparison. In Models 2 and 3 αj is introduced as the sibling fixed effect. MABij is age of the mother at the time of birth for individual i in sibling group j in five-year categories; BIRTHORDERij is the birth order of individual i in sibling group j; and BIRTHYEARij is the year of birth of individual i in sibling group j. In the analyses of educational outcomes, we adjust for SEXij, the sex of individual i in sibling group j, although the sex ratio at birth does not meaningfully vary by maternal age (James 1987). The key coefficient of interest is β1, the estimate for maternal age at the time of birth. Results Summary statistics Table 1 provides summary statistics for the analytical sample for each of the cohort groups. For each of the outcomes the descriptive statistics suggest an inverse U-shaped association by maternal age so that those born to the youngest and oldest mothers score lowest. In addition, the summary statistics suggest an improvement in the outcomes over time. Table 1. Descriptive statistics for years of educational attainment by age 30, grade point average (GPA) at age 16, and physical fitness and height at ages 17 to 20, by maternal age at the time of birth, Sweden Maternal age 15–19 20–24 25–29 30–34 35–39 40–44 45+ All Educational attainment at age 30 N 87,160 489,445 597,801 306,903 93,687 15,823 794 1,591,613 % 5.5 30.8 37.6 19.3 5.9 1.0 0.0 100.0 Female % 48.9 48.6 48.5 48.6 48.3 49.4 46.9 48.5 Birth order Mean 1.1 1.5 1.8 2.3 2.8 3.5 4.3 1.9 Birth year Mean 1967.6 1969.5 1971.1 1972.2 1972.0 1971.1 1971.1 1970.7 Education by birth year All years 11.5 12.2 12.9 13.1 13.0 12.6 12.3 12.6 1960–1964 11.2 11.7 12.2 12.2 12.0 11.7 11.4 11.9 1965–1969 11.4 11.8 12.4 12.4 12.2 12.0 11.6 12.1 1970–1974 11.8 12.4 13.1 13.1 12.9 12.7 12.3 12.8 1975–1979 12.2 12.9 13.6 13.8 13.7 13.5 13.5 13.4 1980–1982 11.9 12.6 13.4 13.8 13.9 13.8 13.3 13.4 GPA at age 16 N 8,558 127,390 236,545 154,746 48,460 6,709 183 582,591 % 1.5 21.9 40.6 26.6 8.3 1.2 0.0 100.0 Female % 48.5 49.1 48.6 48.6 48.4 48.7 48.1 48.7 Birth order Mean 1.1 1.4 1.8 2.1 2.5 2.8 3.4 1.8 Birth year Mean 1985.2 1986.1 1986.7 1987.3 1987.8 1988.2 1988.3 1986.8 Education by birth year All years 165.9 189.9 207.9 217.2 218.7 214.8 206.0 206.8 1982–1984 166.4 191.0 211.7 218.6 216.3 207.5 185.9 205.7 1985–1989 166.3 190.6 208.7 218.5 218.9 215.8 206.8 207.7 1990–1991 155.3 182.2 200.3 213.2 219.1 215.0 208.0 205.1 Physical fitness and height N 11,991 71,340 85,522 37,103 9,233 1,379 65 216,633 % 5.5 32.9 39.5 17.1 4.3 0.6 0.0 100.0 Conscription age Mean 17.7 17.8 17.8 17.8 17.8 17.8 17.8 17.8 Birth order Mean 1.1 1.5 1.9 2.4 2.9 3.6 4.1 1.8 Birth year Mean 1968.5 1969.7 1971.0 1971.9 1971.9 1971.7 1972.6 1970.7 Physical fitness (MWC) by birth year All years 289.5 298.1 304.2 303.3 299.2 293.7 297.8 300.9 1965–1969 285.9 294.0 300.3 299.2 293.3 291.7 286.0 296.0 1970–1974 297.2 303.4 307.7 306.4 302.0 293.3 298.9 305.4 1975–1977 289.0 295.0 301.5 301.3 299.8 296.8 300.9 300.0 Height by birth year All years 178.5 179.1 179.7 180.0 179.9 179.5 179.4 179.5 1965–1969 178.4 179.0 179.6 179.8 179.4 178.9 179.6 179.2 1970–1974 178.6 179.2 179.9 180.1 180.2 179.6 179.7 179.7 1975–1977 178.4 179.0 179.7 180.1 180.1 180.0 178.5 179.7 NOTE: See text for description of samples and units of measurement. For the number of years of education the highest mean was among individuals born to mothers aged 30–34, with 13.1 years of schooling. As the mother's age increases or decreases, the mean years of education decrease. Compared to an individual born to a mother aged 30–34, an individual born to a mother aged 15–19 had spent 70 percent of a standard deviation less time in education by the time he or she had reached age 30, and the equivalent figure for an individual born to a mother aged 45 and older was 35 percent of a standard deviation. For GPA at age 16 (range 0–320), children born to mothers aged 35–39 had the highest mean scores, at 218.7. Those born to mothers 45 and older had a mean score of 206.0, which is 20 percent of standard deviation lower. The lowest mean scores were for those born to mothers aged 15–19, at 165.9, which is only slightly above a mean mark of failure (159 points). For children born to teenage mothers in cohorts 1990–1991, the mean GPA is actually below the failure threshold. The summary statistics for physical fitness show that the highest mean MWC, at 304W, was among male children born to mothers aged 25–29. The mean MWC decreases for both younger and older maternal ages. Men born to teenage mothers had an MWC of 290W, or 29 percent of a standard deviation lower than those born to mothers aged 25–29, while men born to mothers aged 40–44 had an MWC of 294W, which is 21 percent of a standard deviation lower than those born to mothers aged 25–29. Previous research examining how MWC varies by age in Sweden has shown that the mean value for men aged 20–29 is 303W (Wohlfart and Farazdaghi 2003), slightly below the mean score for men born to mothers aged 25–29. The same study found that the mean score for men aged 30–39 was 288W, which implies that men born to teenage mothers have a level of physical fitness approximately equivalent to being at least ten years older than they were when taking these conscription tests. The pattern by birth year shows that the mean M