Kreyenfeld_et_al_englisch.indd Order-Specifi c Fertility Rates for Germany Estimates from Perinatal Statistics for the Period 2001-2008 Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski Abstract: Until 2008, Germany’s vital statistics did not include information on the biological order of each birth. This resulted in a dearth of important demographic indicators, such as the mean age at fi rst birth and the level of childlessness. Re- searchers have tried to fi ll this gap by generating order-specifi c birth rates from survey data, and by combining survey data with vital statistics. This paper takes a different approach by using Perinatal Statistics to generate birth order-specifi c fertility rates for the period 2001 to 2008. Perinatal Statistics includes information on births that took place in German hospitals. Out-of-hospital births, which account for about 2 % of all births, are not included in the Perinatal Statistics. In a sensitivity analysis, we show how robust our estimates are to the inclusion of out-of-hospital births. Our general assessment is that the Perinatal Statistics is a valuable source for generating order-specifi c fertility rates, regardless of whether out-of-hospital births are included. Keywords: Birth order · Fertility · Germany · Eastern and Western Germany · Perinatal Statistics 1 Introduction Across Europe, we have witnessed a postponement of fi rst-time motherhood, and an increase in the share of women who remain childless throughout their lives (So- botka 2004). Despite the obvious importance of these social changes, it is remarka- ble to recall that, until very recently, some countries did not provide suffi cient infor- mation to document this change in behaviour. Important demographic indicators, such as the mean age at fi rst birth or the share of permanently childless women, were not available from vital statistics of countries such as Austria, Belgium, Lux- embourg, Switzerland, Germany, France and the UK, because order-specifi c fertil- ity information were not recorded in these countries. Most countries have since Comparative Population Studies – Zeitschrift für Bevölkerungswissenschaft Vol. 35, 2 (2010): 207-224 © Federal Institute for Population Research 2010 URL: www.comparativepopulationstudies.de DOI: 10.4232/10.CPoS-2010-06en URN: urn:nbn:de:bib-cpos-2010-06en8 • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski208 reformed their vital statistics and Statistical Offi ces have started collecting data on births by biological order. Austria did so in 1984; France in the year 1998, while England and Wales are expected to ratify a regulation stipulating that order-specifi c data can be collected from 2011 onwards. Germany’s Statistical Offi ce has collected this type of data since 2008, but has not published the data yet as the quality has not been fully checked. Order-specifi c birth information will therefore soon be available for almost all European countries. However, we still lack order-specifi c fertility information for the past. Researchers have tried to fi ll this gap by drawing on other sources. Order- specifi c fertility information has been generated from survey data (Toulemon 2001; Smallwood 2002). Others have tried to combine survey data with vital statistics (Birg et al. 1990; Kreyenfeld 2002; Handcock et al. 2000). This paper uses Perinatal Statistics to generate order-specifi c birth rates for the period 2001-2008, a period for which no birth order-specifi c information is available for Germany. The Perinatal Statistics has been monitored by the Bundesgeschäfts- stelle Qualitätssicherung (BQS) (Institute for Quality and Safety). It includes clinical records for all children who were delivered in German hospitals. To our knowledge, it is the only data source that contains information on the biological order of hospital births at a population level for Germany. This data has primarily been used in the fi eld of medicine, but some few demographic studies exist that used these data for analysing fertility behaviour (Hullen 2003; Voigt/Hullen 2005; Birg/Flöthmann 1996). These previous studies, however, were only able to draw on data from selected federal states, while this study uses data from the Perinatal Statistics for the whole of Germany for the period 2001 to 2008. We have structured the paper as follows. In Part 2, we give a detailed account of the Perinatal Statistics provided by the BQS. Part 3 explains the method used to generate age- and order-specifi c fertility rates with this data. In addition, we provide some initial results on order-specifi c fertility behaviour in Germany. Part 4 investi- gates how robust our estimates are, if we not only use data from the Perinatal Sta- tistics, which only includes hospital births, but also consider out-of-hospital births. In this section we additionally draw upon data on out-of-hospital births provided by the Gesellschaft für die Qualität in der außerklinischen Geburtshilfe e.V. (QUAG). Part 5 summarises the fi ndings and concludes. 2 Data and variables 2.1 General description of the Perinatal Statistics Data for this investigation comes from the BQS, which is an institution that moni- tors the quality of care in German hospitals. One of its responsibilities is to collect the clinical records of the maternity wards of all German hospitals. This initiative dates back to the Münchner Perinatalstudie, which was conducted in the period 1975-1977. The main goal of this study was to report on the quality of the maternity wards of all the clinics in the city of Munich. The Münchner Perinatalstudie was the Order-Specifi c Fertility Rates for Germany • 209 fi rst step towards introducing a systematic statistics that contained the character- istics of a clinical birth, i.e., whether it was a stillbirth, whether a Caesarean was conducted, and whether other complications occurred during delivery or during pregnancy. This procedure was gradually implemented throughout Bavaria and was adopted in the other federal states over the course of the 1980s (Bundesgeschäftss- telle Qualitätssicherung 2002: 190). After German unifi cation, the eastern German federal states joined the Perinatal Statistics. While it was initially voluntary, it be- came compulsory for the hospitals to collect this data in 1995 (Bundesgeschäftss- telle Qualitätssicherung 2008: 196). At that time, data collection was still organised on a federal level, and no central register had been implemented. Finally, in 2001, the BQS was assigned the task of establishing a nationwide registry. Since this time, it has been mandatory for all public and private hospitals to document all births (both still and live), and make these records available to the BQS, which stores this information in a data fi le called Datensatz Geburtshilfe.1 Data for the Perinatal Statistics are collected by the staff of the hospital where the woman delivers the child. The characteristics of a newborn, such as weight, sex and physical condition, are taken from the medical records of the child. Information on the medical background of the current and previous pregnancies is copied from the pregnancy record (Mutterpass) of the mother (Jahn/Berle 1996: 132; Reime et al. 2008). If this is not available, the mother is asked to provide this information. Ad- ditional data (such as smoking habits) are gathered by the staff upon admission to the hospital (Voigt et al. 2006; Schneider et al. 2008). The Perinatal Statistics provides a rich set of variables for micro-level analysis for medical and demographic research. However, we should also point out some shortcomings of the data. First, the Perinatal Statistics does not include deliver- ies that take place at home, in birth centres, or in the offi ces of midwives.2 These births make up for about 2 % of all births (see tab. 5). The fact that these births are not included in the Perinatal Statistics is a potential problem because women who opt for an out-of-hospital birth differ in several ways from women who choose the “standard” path of delivering their children in a hospital. Since women only rarely choose to have their fi rst child at home or in a birth centre, out-of-hospital births are often births to older women who already have at least one child (Loytved/Wenzlaff 2007, see also fi g. 4). We will turn to this issue again later on in a sensitivity analysis (part 4). 1 When we talk about Perinatal Statistics, we mean the Datensatz Geburtshilfe of the BQS. The Perinatal Statistics were assessed through remote execution by sending SPSS syntax scripts to the BQS by email. In 2010, responsibilities have shifted and the “Aqua-Institut” is now monitor- ing this data (http://www.aqua-institut.de/). 2 If delivery was planned at home or at a birth centre, but the woman was transferred to the hospital during the process of giving birth, the birth was entered into the Perinatal Statistics. Loytved and Wenzlaff (2007: 9) show that in roughly 12 percent of all cases in which a birth was intended to occur at home or in a birth centre, the mother was transferred to a hospital. From the perspective of the Perinatal Statistics, about one percent of all deliveries in hospitals were originally expected to occur at home or in a birth centre (BQS-Bundesauswertung 2007: 103). • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski210 Another shortcoming of the Perinatal Statistics is that data quality varies quite substantially between the calendar years. Although it is mandatory for all hospitals to provide information on all births since 2001, the Perinatal Statistics does not cover all hospital births and not all German federal states in the early years. For 2001, Hesse, Schleswig-Holstein and Saarland did not provide any data to the cen- tral register of the BQS. In the subsequent years, these federal states participated, but some hospitals still failed to provide information. These gaps have closed over the years. Since 2004, almost all clinical births have been covered in the Perinatal Statistics. 2.2 Selection of sample For this study, we use the Perinatal Statistics for the period 2001-2008. Since women can have multiple births, different observation units could be employed. Hence, it is possible to conduct investigations at the “mother level” or at the “child level”. There are also analyses which used singletons only (Krafczyk 2007; Voigt et al. 2006). For our investigation, the observation unit is the child. This corresponds to the vital statistics, which mainly counts live-born children (and not births). To improve the fl ow of the paper, we use the word “birth” referring to live-born children. The total number of live-born children in the Perinatal Statistics for the period 2001 to 2008 is 4,982,707. In a few cases, it was not possible to identify the year of birth, the order of the birth or the age of the mother. These cases were deleted from the analysis. Furthermore, births in which the mother was older than age 50 or younger than age 11 were also omitted. This leaves us with 4,978,381 live-born children in the fi nal data set. Table 1 displays the number of births in our sample and compares it with data from the vital statistics provided by the Statistical Offi ce. As can be seen in the ta- ble, differences between the vital statistics and the Perinatal Statistics are particu- larly large in 2001, but they decline over the years. The difference in coverage in the year 2001 can largely be attributed to the fact that Hesse, Schleswig-Holstein and Saarland failed to deliver their data. Coverage increases continuously, and in 2004 about 95 % of all births are covered in the Perinatal Statistics. Tab. 1: Coverage of births in the Perinatal Statistics 2001 2002 2003 2004 2005 2006 2007 2008 Total Births in Perinatal Statistics 452,826 608,122 590,293 671,228 665,718 655,951 667,016 667,227 4,978,381 Births in Vital Statistics 734,475 719,250 706,721 705,622 685,795 672,724 684,862 682,514 5,591,963 Perinatal Statistics/ Vital Statistics 0.62 0.85 0.84 0.95 0.97 0.98 0.97 0.98 0.89 Source: Statistisches Bundesamt (2009a), BQS Perinatal Statistics (own estimates) Order-Specifi c Fertility Rates for Germany • 211 2.3 Variables The variables used in our analysis are age, birth order and region. The age of the mother was generated by the BQS, which provided the data. It was generated by subtracting the birth year of the child from the year of birth of the mother. Age is used in single age categories. We also distinguish between western Germany and eastern Germany, the latter one including the city-state of Berlin.3 Our key variable of interest is the order of birth, which is broken down into the following categories: fi rst-, second-, third-, fourth- and higher-order births. The order of the birth was generated by adding a “one” to the number of previously live-born children. One aspect that required special attention concerns the multiple births. For singletons, the birth order can be generated by simply adding a one to the number of previously live-born children. This strategy needed to be modifi ed for multiple births, because the numbers of previously born children do not take into account the current birth. In other words, if a woman gives birth to twins and does not have any prior children, the number of previous live births is zero in the Perinatal Statistics. If we would ap- ply the same strategy as for singletons, we would assign erroneously both children the birth order one. We corrected this bias by assigning one of the twins the order “one”, the other one the order “two”.4 Table 2 tabulates the distribution of births by birth order and calendar year in our fi nal data set. The table shows that about 50 % were fi rst-; 35 % second-; 10 % third-; and 5 % fourth- and higher-order children.5 3 The German Statistical Offi ce usually excludes Berlin when it generates fertility rates by eastern and western Germany (Kreyenfeld/Pötzsch/Kubisch 2010). It is, however, a problem to exclude Berlin from the Perinatal Statistics. In contrast to the vital statistics, which uses the place of resi- dence of the mother, the Perinatal Statistics surveys the place of birth of the child. This means that we cannot generate the fertility rates by federal state, because we would have to use a different defi niton of region for the weights and for the base population than in the Perinatal Statistics. This is particularly true for Berlin, because some women who live in the federal state of Brandenburg choose a hospital in Berlin to deliver their child. 4 We fi rst generated a “master table” that contained the number of all children by order, age and region. Furthermore, we generated a “multiple birth table” that contained multiple births by order, age and region. From the master table, we subtracted half of all multiple births of a given order. These children were then added to the next birth order. For simplicity, we assumed that all multiple births are twin births. 5 Data by age and region have been made available as a MPIDR Technical Report (http://www. demogr.mpg.de/en/publications/technicalreports.htm). This report also contains a detailed description of the potential of the Perinatal Statistics. The Perinatal Statistics provides a rich source for demographic analyses as detailed information on past pregnancies, live and still births, abortions and miscarriages. Unfortunately, this data cannot be used to generate expo- sure rates as pregnancy dates of past pregnancies are not included. • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski212 3 Method and fi rst results 3.1 Estimating order-specifi c fertility rates from Perinatal Statistics (Method 1) Our main goal is to use the Perinatal Statistics to generate age- and order-specifi c fertility rates for Germany. Since fertility patterns in eastern and western Germany continue to differ (Konietzka/Kreyenfeld 2007), we will estimate separate fertility rates for eastern Germany (including Berlin) and western Germany (excluding Ber- lin). In order to generate these rates, the number of births and the base population of women are necessary. Data on the number of women by age and region (eastern and western Germany) comes from German vital statistics. A problem arises when calculating fertility rates because the Perinatal Statistics does not include all births. In order to calculate birth rates, we weighted the data of the Perinatal Statistics to match the total number of births in Germany. We have constructed a weighting factor that considers age, year and region (eastern and western Germany). Let BV be the number of births in the vital statistics, BP be the number of births in the Perinatal Statistics, be our estimated number of births, i be the order of the births, and w be the weighting factor. The weight ac- counts for age (a), year (t) and region (r). The weighting factor is generated by the ratio of the number of births in the vital statistics and the number of births in the Perinatal Statistics: Tab. 2: Number of live-born children by birth order, absolute numbers and column percent 2001 2002 2003 2004 2005 2006 2007 2008 Total Absolute numbers 1st child 220,525 294,635 285,152 324,693 321,622 315,772 324,234 325,759 2412,390 2nd child 156,934 211,359 206,091 233,625 231,540 226,136 228,948 228,698 1723,330 3rd child 51,166 69,566 67,159 77,217 76,703 77,599 77,577 76,900 573,885 4th+ child 24,202 32,563 31,893 35,693 35,854 36,444 36,258 35,870 268,777 Total 452,826 608,122 590,293 671,228 665,718 655,951 667,016 667,227 4,978,381 Column % 1st child 49% 48% 48% 48% 48% 48% 49% 49% 48% 2nd child 35% 35% 35% 35% 35% 34% 34% 34% 35% 3rd child 11% 11% 11% 12% 12% 12% 12% 12% 12% 4th+ child 5% 5% 5% 5% 5% 6% 5% 5% 5% Total 100% 100% 100% 100% 100% 100% 100% 100% 100% Source: BQS Perinatal Statistics (own estimates) B̂ )r,t,a(B )r,t,a(B )r,t,a(w P V (1) Order-Specifi c Fertility Rates for Germany • 213 In the next step, the numbers of births for each order i from the Perinatal Sta- tistics , differentiated into age (a), year (t) and region (r), are multiplied by the weights. This gives the estimated number of births: To generate age- and order-specifi c fertility rates, the estimated births are re- lated to the number of women by age, region and calendar year. 3.2 Preliminary results on order-specifi c fertility behaviour in Germany It is beyond the scope of this paper to provide a comprehensive overview of east- ern and western German fertility dynamics. However, we still seek to provide some basic demographic indicators based on our estimated values. Table 3 displays the order-specifi c TFR, while table 4 shows the mean ages at childbirth.6 Figures 1 and 2 graph these indicators. p iB )r,t,a(w)r,t,a(B)r,t,a(B̂ Pii (2) 6 In the tables and graphs, we have used ages 15-44, because we only had aggregated values on the number of births from the vital statistics for the ages 45 and older and 14 and younger. In order to generate order-specifi c fertility rates for these ages, we would have had to generate a separate weighting factor. Tab. 3: Order-specifi c TFR (ages 15-44) 2001 2002 2003 2004 2005 2006 2007 2008 Eastern Germany 1st child 0.66 0.66 0.67 0.69 0.67 0.67 0.69 0.71 2nd child 0.38 0.39 0.40 0.42 0.41 0.42 0.45 0.46 3rd child 0.11 0.11 0.11 0.12 0.12 0.13 0.13 0.14 4th+ child 0.06 0.05 0.06 0.06 0.06 0.06 0.07 0.07 Total 1.21 1.21 1.23 1.28 1.26 1.28 1.34 1.38 Western Germany 1st child 0.69 0.68 0.67 0.67 0.66 0.65 0.67 0.67 2nd child 0.48 0.47 0.48 0.48 0.47 0.46 0.47 0.47 3rd child 0.15 0.15 0.15 0.16 0.15 0.16 0.16 0.16 4th+ child 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.07 Total 1.38 1.37 1.36 1.37 1.35 1.34 1.37 1.38 Germany 1st child 0.68 0.67 0.67 0.68 0.66 0.65 0.67 0.68 2nd child 0.46 0.46 0.46 0.47 0.46 0.46 0.47 0.47 3rd child 0.14 0.14 0.14 0.15 0.15 0.15 0.15 0.16 4th+ child 0.07 0.06 0.07 0.07 0.07 0.07 0.07 0.07 Total 1.35 1.34 1.34 1.36 1.34 1.33 1.37 1.38 Source: BQS Perinatal Statistics (own estimates), Statistisches Bundesamt (2009a) • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski214 Regarding fi gure 2, changes in the age at fi rst birth are most remarkable. In 2001, western German women were, on average, 27.4 years of age at fi rst birth; by 2008, they were 28.7 years old. This fi nding suggests that the trend towards postponing the fi rst birth has not yet halted, despite the fact that West Germany was one of the countries where fertility postponement started rather early. Ages at second birth have increased to a lesser extent than ages at fi rst birth. For higher-order births, there is basically no change over time in the mean age at childbirth. The results also show that the order-specifi c TFRs did not change much between 2001 and 2008. In eastern Germany, there is a stronger increase in the age at fi rst birth than in western Germany. We observe an increase in the age at fi rst birth of about 1.5 years between 2001 and 2008. However, eastern German women are in 2008 still one year younger at fi rst birth than their counterparts in the West. East-West differences in the age at second birth are smaller, which suggests that eastern Germans space fi rst and second births farther apart than western Germans. The most surprising fi nding for eastern Germany is the strong increase in the TFR for second births (see fi g. 1). Given that second birth rates dropped radically in eastern Germany after unifi cation (Sackmann 1999; Huinink 2005; Kreyenfeld 2009), this result could represent a fi rst indication of a recuperation of second birth rates. Compared to the pattern in western Germany, one still needs to draw attention to the low progression ratios to the third child. During GDR-time, third and higher or- Tab. 4: Mean age at childbirth by birth order (ages 15-44) 2001 2002 2003 2004 2005 2006 2007 2008 Eastern Germany 1st child 26.12 26.35 26.60 26.85 26.97 27.07 27.29 27.47 2nd child 29.32 29.54 29.66 29.86 29.94 30.10 30.45 30.67 3rd child 31.42 31.64 31.62 31.62 31.63 31.83 32.07 32.21 4th+ child 33.15 33.23 33.09 33.00 33.10 33.21 33.08 33.34 All births 27.94 28.14 28.34 28.55 28.68 28.85 29.10 29.30 Western Germany 1st child 27.43 27.57 27.74 27.95 28.10 28.26 28.49 28.69 2nd child 29.88 30.04 30.15 30.30 30.43 30.57 30.78 30.98 3rd child 31.46 31.57 31.65 31.79 31.87 31.96 32.19 32.35 4th+ child 33.06 33.09 33.19 33.26 33.32 33.41 33.49 33.56 All births 28.99 29.14 29.28 29.46 29.60 29.76 29.97 30.15 Germany 1st child 27.14 27.32 27.50 27.73 27.87 28.02 28.25 28.45 2nd child 29.81 29.96 30.08 30.23 30.35 30.49 30.72 30.92 3rd child 31.51 31.58 31.66 31.76 31.83 31.94 32.17 32.33 4th+ child 33.12 33.12 33.19 33.21 33.28 33.37 33.41 33.52 All births 28.80 28.96 29.11 29.29 29.43 29.59 29.80 29.99 Notes: Mean ages were generated based on age-specifi c fertility rates. Source: BQS Perinatal Statistics (own estimates), Statistisches Bundesamt (2009a) Order-Specifi c Fertility Rates for Germany • 215 Fig. 1: Order-specifi c TFR (ages 15-44) 0,0 0,2 0,4 0,6 0,8 2001 2002 2003 2004 2005 2006 2007 2008 First (East) First (West) Second (East) Second (West) Third (East) Third (West) Year Source: BQS Perinatal Statistics (own estimates), Statistisches Bundesamt (2009a) Fig. 2: Mean age at childbirth by birth order (ages 15-44) 25 26 27 28 29 30 31 32 33 2001 2002 2003 2004 2005 2006 2007 2008 First (East) First (West) Second (East) Second (West) Third (East) Third (West) Year Notes: Mean ages were generated based on age-specifi c fertility rates. Source: BQS Perinatal Statistics (own estimates), Statistisches Bundesamt (2009a) • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski216 der birth rates were rather low (Kreyenfeld 2004) and our investigation based on the Perinatal Statistics suggests that low progression ratios to the third child are still a characteristic of contemporary eastern German fertility. 4 Sensitivity analysis 4.1 Statistics on out-of-hospital births (QUAG Statistics) A key question for our analyses is how robust the estimated number of births is given the fact that the Perinatal Statistics does not include all births. In particular, the question arises of whether the results are biased because home births are not included. Until recently, no information on out-of-hospital births has been available, and medical as well as epidemiological studies used only hospital births for their in- vestigations. In 1999, however, the midwives in Germany founded the Gesellschaft für Qualität in der außerklinischen Geburtshilfe e.V. (QUAG). The aim of QUAG, like the BQS, is to monitor the quality of medical care. A questionnaire similar to the one used in collecting the Perinatal Statistics is conducted for out-of-hospital births. Although participation is voluntary, about 80 % of all out-of-hospital births are covered in this data source, which we refer to as QUAG Statistics in the follow- ing (Loytved 2009: 6). Because the variables available in the QUAG Statistics are almost the same as the ones used in the Perinatal Statistics, the two sources can be combined in a straightforward manner. An essential measure of quality for the QUAG Statistics is how many of all out-of- hospital births are included in the data set. Table 5 compares the number of births in the QUAG Statistics with the calculated total number of out-of-hospital births in Germany.7 As can be seen from this table, coverage of the out-of-hospital births in the QUAG Statistics ranges from 68 to 91 %. This is substantially lower than the coverage of the Perinatal Statistics. However, the sample size is still high enough to give us a reasonable estimate of the age and parity distribution of out-of-hospital births. 4.2 Age structure of out-of-hospital births The share of out-of-hospital births is, at 1 to 2 % of all births, rather low (see tab. 5). However, they differ from hospital births systematically by order and age of the mother. This can be discerned from fi gure 3 and fi gure 4 exemplarily for the year 2008. Figure 3 plots the age pattern of births in the Perinatal Statistics and in the 7 Unfortunately, there are no offi cial records on the total number of out-of-hospital births in Ger- many. They can neither be generated from the Perinatal Statistics as this statistics is not com- plete during the fi rst years after the central register had been established. The data in table 5 therefore comes from the „Gesundheitsberichterstattung des Bundes“ (Statistisches Bundes- amt 2009b). We should note, however, that this is probably the upper benchmark for the actual amount of out-of-hospital births (Loytved 2009: 6). Order-Specifi c Fertility Rates for Germany • 217 QUAG Statistics. The fi gure supports previous fi ndings that out-of-hospital births are mostly births to older women (Loytved/Wenzlaff 2007). Figure 4 plots the dis- tribution of births by birth order in both the Perinatal and the QUAG Statistics. This fi gure shows that out-of-hospital births are more likely to be of higher-order births than hospital births. Only 17 % of births in the Perinatal Statistics of 2008 were of order three and higher, compared to 28 % in the QUAG Statistics. 4.3 Estimating order-specifi c fertility rates from Perinatal Statistics and QUAG Statistics (Method 2) The differences in the age structure between hospital and out-of-hospital births raise the question of to what extent our previous results (Method 1) are affected by the omission of out-of-hospital births. In the following, we address this issue by using additional information on out-of-hospital births. For this purpose, we have constructed a similar weighting factor as before (see part 3). However, instead of just using the births from the Perinatal Statistics, we also accounted for the births in the QUAG Statistics. Let BV be the number of births in the vital statistics, BQ be the number of births in the QUAG Statistics, BP be the number of births in the Perinatal Statistics, i be the order of the births, and w* be the new weighting factor. The weight again accounts for age (a), year (t) and region (r). It is generated by dividing the number of births from the vital statistics by the sum of the number of births in the Perinatal and the QUAG Statistics: Tab. 5: Coverage of births in QUAG Statistics 2001 2002 2003 2004 2005 2006 2007 2008 Number of births in QUAG Statistics 8,245 8,219 8,561 8,686 8,305 8,315 8,188 8,280 Out-of-hospital births in Germany* 9,841 11,236 9,415 12,130 12,260 10,859 10,997 9,799 Number of births in Germany 734,475 719,250 706,721 705,622 685,795 672,724 684,862 682,514 Coverage of out-of-hospital births in QUAG Statistics 84% 73% 91% 72% 68% 77% 74% 84% Share of out-of-hospital births out of all births in Germany 1% 2% 1% 2% 2% 2% 2% 1% * The number of out-of-hospital births was generated as the difference between the number of births in the vital statistics and the number of hospital births taken from the Gesundheitsberichterstattung des Bundes (Statistisches Bundesamt 2009b). Source: BQS Perinatal Statistics and QUAG Statistics (own estimates), Statistisches Bundesamt (2009a/b) )r,t,a(B)r,t,a(B )r,t,a(B )r,t,a(w QP V * (3) • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski218 Fig. 3: Age pattern of births in Perinatal and QUAG Statistics in 2008 (percent of births by single ages out of all births) 0 1 2 3 4 5 6 7 8 9 15 20 25 30 35 40 45+ Perinatal Statistics QUAG Statistics Percent Age of mothers in years Source: BQS Perinatal Statistics and QUAG Statistics (own estimates) Fig. 4: Distribution of births by order in Perinatal Statistics and QUAG Statistics in 2008 0 25 50 75 Perinatal Statistics QUAG Statistics 5 9 12 19 34 40 49 31 First birth Second birth Third birth Fourth and higher order birth Percent Note: If percent do not add up to 100, this needs to be attributed to rounding errors. Source: BQS Perinatal Statistics and QUAG Statistics (own estimates) Order-Specifi c Fertility Rates for Germany • 219 In the next step, the number of order-specifi c births a n d by age (a), year (t) and region (r) from the Perinatal and the QUAG Statistics are multiplied by the new weights to obtain the estimated number of births: 4.4 Comparison of estimates from Method 1 and Method 2 Table 6 compares the TFR from Methods 1 and 2. In the fi rst half of the table (under the heading “Method 1”), the estimates that we derived from the Perinatal Statistics are displayed. In the second part of the table (under the heading “Method 2”) are the estimates based on the Perinatal Statistics and the QUAG Statistics. The last part of the table gives the difference between both estimates. The comparison shows that both methods provide similar results. Differences in the TFR never exceed the value of 0.002. Table 7 displays estimates of the mean age at childbirth by birth order. Again, we observe hardly any differences between the two methods. Thus, we can conclude that measures of centrality, like the mean age at childbirth or the TFR, are not af- fected by the fact that the Perinatal Statistics does not include births that have not occurred in hospitals. Finally, fi gure 5 displays how the age pattern of births differs between the two methods in the year 2008. Here, we generated a ratio; i.e., we divided the number of births from Method 1 by the number of births from Method 2 for each age. As can be seen from the fi gure, differences between the two methods increase at higher ages. This is compatible with the idea that Method 2, unlike Method 1, accounts for out-of-hospital births, which are more prevalent at higher ages. However, the ratio is mostly between 0.99 and 1.01, which indicates that the differences in the estimated number of births between both methods do not exceed 1 %. It is only for the ages below 20 and above 40 that the differences can increase to up to 2 %. From this sen- sitivity analysis we conclude that it might be suffi cient to just draw on the Perinatal Statistics to estimate order-specifi c fertility rates in the future. p iB )r,t,a(w)r,t,a(B)r,t,a(B)r,t,a(B̂ *Qi P i * i (4) Q iB • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski220 Tab. 6: TFR by birth order (ages 15-44), comparison of Method 1 and Method 2 Source: QUAG Statistics and Perinatal Statistics (own estimates) 2001 2002 2003 2004 2005 2006 2007 2008 Method 1 1st child 0.683 0.675 0.669 0.676 0.663 0.652 0.675 0.680 2nd child 0.456 0.457 0.460 0.467 0.462 0.456 0.469 0.472 3rd child 0.142 0.144 0.144 0.148 0.147 0.152 0.155 0.155 4th+ child 0.065 0.065 0.065 0.065 0.066 0.068 0.070 0.070 Method 2 1st child 0.681 0.673 0.667 0.675 0.661 0.651 0.673 0.678 2nd child 0.458 0.458 0.461 0.468 0.462 0.457 0.470 0.473 3rd child 0.143 0.145 0.144 0.149 0.148 0.152 0.155 0.156 4th+ child 0.065 0.065 0.066 0.066 0.066 0.069 0.070 0.071 Method 2 - Method 1 1st child -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 2nd child 0.001 0.000 0.001 0.001 0.001 0.001 0.001 0.001 3rd child 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 4th+ child 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Tab. 7: Mean ages at childbirth by birth order (ages 15-44), comparison of Method 1 and Method 2 2001 2002 2003 2004 2005 2006 2007 2008 Method 1 1st child 27.144 27.325 27.502 27.726 27.875 28.017 28.251 28.446 2nd child 29.814 29.960 30.076 30.229 30.348 30.492 30.723 30.919 3rd child 31.505 31.582 31.662 31.764 31.830 31.940 32.169 32.331 4th+ child 33.120 33.115 33.189 33.213 33.280 33.369 33.415 33.519 Method 2 1st child 27.128 27.312 27.489 27.714 27.862 28.003 28.237 28.432 2nd child 29.813 29.958 30.074 30.227 30.346 30.492 30.722 30.916 3rd child 31.514 31.592 31.673 31.773 31.840 31.948 32.183 32.346 4th+ child 33.140 33.129 33.201 33.223 33.295 33.387 33.430 33.533 Method 2 - Method 1 1st child -0.016 -0.013 -0.013 -0.011 -0.013 -0.014 -0.014 -0.014 2nd child -0.001 -0.002 -0.002 -0.002 -0.002 0.000 -0.001 -0.003 3rd child 0.009 0.011 0.011 0.009 0.011 0.008 0.013 0.014 4th+ child 0.019 0.014 0.012 0.010 0.015 0.018 0.015 0.014 Notes: Mean ages were generated based on age-specifi c fertility rates. Source: QUAG Statistics and Perinatal Statistics (own estimates) Order-Specifi c Fertility Rates for Germany • 221 5 Summary and conclusions Until recently, German vital statistics has not included order-specifi c fertility data so that important demographic measures, such as the mean age at fi rst birth or the or- der-specifi c TFRs cannot be generated. The main goal of this paper was to delineate how this dearth of demographic data can be overcome by estimating order-specifi c fertility rates by using Perinatal Statistics. As Perinatal Statistics does not include all births in Germany, they had to be re-weighted in such a way that they matched the Fig. 5: Ratio of estimated births from Method 2 and Method 1 by single ages, Germany 2008 97,5 100,0 102,5 15 20 25 30 35 40 Method 2 Method 1 97,5 100,0 102,5 15 20 25 30 35 40 Method 2 Method 1 97,5 100,0 102,5 15 20 25 30 35 40 Method 2 Method 1 97,5 100,0 102,5 15 20 25 30 35 40 Method 2 Method 1 Age of Women Age of Women Age of Women Age of Women First birth Second birth Third birth Fourth and higher birth Notes: Method 1: Estimates based on Perinatal Statistics; Method 2: Estimates based on combination of Perinatal Statistics and QUAG Statistics (own estimates) • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski222 offi cial number of births from the vital statistics. Based on these estimates and addi- tional data on the population of females, we have generated age- and order-specifi c fertility rates for eastern and western Germany, as well as for Germany in total. We have also tried to assess the size of the bias created by the fact that out-of- hospital births, which account for about 2 % of all births in Germany, are not includ- ed in the Perinatal Statistics. Our sensitivity analysis has shown that the inclusion of out-of-hospital births (QUAG Statistics) does not have a noticeable impact on our estimates. In particular, measures of centrality, like the mean age at childbirth or the TFR by order of birth do not change in any substantial manner, if we additionally use data from the QUAG Statistics. Apparently, the share of out-of-hospital births is too small to affect our estimates to a signifi cant degree. We have also used this data to give a rough overview on the trends in order- specifi c fertility behaviour in Germany. Our estimates show that the age at fi rst birth is continuously increasing in western and eastern Germany during the period 2001-2008. If one compares ages at fi rst birth in both parts of the country, strong differences in the age at fi rst-time-parenthood prevail. Women in the eastern states are more than one year younger at fi rst birth than their counterparts in the west- ern states in 2008. Another remarkable aspect about eastern German fertility is the strong increase in second birth rates in recent years which might suggest that east- ern Germany has broken the “trend towards the one-child family” (Huinink 2005). Third birth rates have somewhat increased in eastern Germany too, but they are still much lower than in the western states of Germany. Overall, this fi rst investigation shows persisting differences in order-specifi c dif- ferences in behaviour between eastern and western Germany. It remains to be seen how these order-specifi c differences develop in the future and whether fertility be- haviour in eastern and western Germany converges any further. As Germany has recently reformed its vital statistics and provided order-specifi c fertility information, it will be possible to merge the time series from the Perinatal Statistics with the one from the vital statistics to investigate the trend in order-specifi c birth behaviour. We particularly want to thank Renate Meyer, Burkhard Fischer and Heike Schwarze from the Bundesgeschäftsstelle Qualitätssicherung (BQS) for their valuable com- ments on the Perinatal Statistics. Furthermore, we are very grateful to Anke Wie- mer, who provided us with data from the Gesellschaft für Qualität in der außerkli- nischen Geburtshilfe (QUAG). We would also like to thank our colleagues at the Max Planck Institute for Demographic Research, in particular Joshua Goldstein, Vladimir Shkolnikov and Aiva Jasilioniene for their valuable comments. Many thanks also to Tomáš Sobotka and Tomas Freijka for their comments on the paper. For language editing, we would like to thank Miriam Hils. Order-Specifi c Fertility Rates for Germany • 223 References Birg, Herwig; Filip, Daniel; Flöthmann, E.-Jürgen, 1990: Paritätsspezifi sche Kohorten- analyse des generativen Verhaltens in der Bundesrepublik Deutschland nach dem 2. Weltkrieg. IBS-Materialien 30. Universität Bielefeld. Institut für Bevölkerungsfor- schung und Sozialpolitik Birg, Herwig; Flöthmann, E.-Jürgen, 1996: Entwicklung der Familienstrukturen und ihre Auswirkungen auf die Belastungs- bzw. Transferquotienten zwischen den Generatio- nen. Studienbericht im Auftrag der Enquete-Kommission des Deutschen Bundestages „Demographischer Wandel“. In: Enquete-Kommission „Demographischer Wandel“ des Deutschen Bundestages (eds.): Herausforderungen unserer älter werdenden Ge- sellschaft an den einzelnen und die Politik. Studienprogramm 1 und 2. Heidelberg: RV Decker Bundesgeschäftsstelle Qualitätssicherung, 2002: BQS Outcome 2002. URL: [http://info. bqs-online.de/outcome/16n1/Buaw-2002-16n1-qr.pdf, 27-10-2009] Bundesgeschäftsstelle Qualitätssicherung, 2007: BQS-Bundesauswertung 2006, Ge- burtshilfe. URL: [http://www.bqs-qualitaetsreport.de/2006/ergebnisse/leistungsbe- reiche/geburtshilfe/geburtshilfe.pdf/download, 28-10-2009] Bundesgeschäftsstelle Qualitätssicherung, 2008: BQS-Qualitätsreport 2007. Kapitel 18 Geburtshilfe. URL: [http://www.bqs-qualitaetsreport.de/2007/ergebnisse/leistungs- bereiche/geburtshilfe/gebh.pdf/view, 26-02-2009] Handcock, Marc S.; Huovilainen, Sami M.; Rendall, Michael S., 2000: Combining regis- tration-system and survey data to estimate birth probabilities. In: Demography 37,2: 187-192 Huinink, Johannes, 2005: Ostdeutschland auf dem Weg zur Ein-Kind-Familie. In: Dienel, Christiane (eds.): Abwanderung, Geburtenrückgang und regionale Entwicklung: Ursachen und Folgen des Bevölkerungsrückgangs in Ostdeutschland. Wiesbaden: VS Verlag für Sozialwissenschaften: 231-246 Hullen, Gert, 2003: Demographische Analysen der Fertilität mit Hilfe der Perinatalstatis- tik. In: Der Gynäkologe 36: 62-68 Jahn, Albrecht; Berle, Peter, 1996: Zur Qualität antepartaler Daten in der Hessischen Perinatalerhebung: Ein Vergleich mit Angaben im Mutterpaß und Ergebnissen einer Schwangerenbefragung. In: Geburtshilfe und Frauenheilkunde 56: 132-138 Konietzka, Dirk; Kreyenfeld, Michaela (eds.), 2007: Ein Leben ohne Kinder. Kinderlosig- keit in Deutschland. Wiesbaden: VS Verlag für Sozialwissenschaften Krafczyk, Dietmar, 2007: Vergleichende Untersuchungen zum Einfl uss von vorausgegan- genen Totgeburten, Aborten und Schwangerschaftsabbrüchen bei den Müttern auf die somatische Klassifi kation der Neugeborenen nach Geburtsgewicht und Schwan- gerschaftsdauer: Analyse eines Schwangerenkollektivs mit 1.065.202 Einlingsgebur- ten der Jahre 1995-2000 der Bundesrepublik Deutschland. Medizinische Fakultät. Diss. Rostock: Universität Rostock Kreyenfeld, Michaela, 2002: Parity specifi c birth rates for West Germany: An attempt to combine survey data and vital statistics. In: Zeitschrift für Bevölkerungswissenschaft 27,3: 327-357 Kreyenfeld, Michaela, 2004: Fertility decisions in the FRG and GDR. In: Demographic Research Special Collection 3: 276-318 • Michaela Kreyenfeld, Rembrandt Scholz, Frederik Peters, Ines Wlosnewski224 Kreyenfeld, Michaela, 2009: Das zweite Kind in Ostdeutschland: Aufschub oder Ver- zicht? In: Cassens, Insa; Luy, Marc; Scholz, Rembrandt (eds.): Die Bevölkerung in Ost- und Westdeutschland: Demografi sche, gesellschaftliche und wirtschaftliche Entwick- lungen seit der Wende. Wiesbaden: VS Verlag für Sozialwissenschaften: 100-123 Kreyenfeld, Michaela; Pötzsch, Olga; Kubisch, Karolin, 2010: Human Fertility Database Documentation: Germany. Rostock: Max Planck Institute for Demographic Research Loytved, Christine, 2009: Qualitätsbericht 2006. Außerklinische Geburtshilfe in Deutsch- land. Im Auftrag der „Gesellschaft für Qualität in der außerklinischen Geburtshilfe e.V.“. URL: [http://www.quag.de/downloads/QUAG_bericht2006.pdf, 28-10-2010] Loytved, Christine; Wenzlaff, Paul, 2007: Außerklinische Geburt in Deutschland. German Out-Of-Hospital Birth Study 2000-2004. Bern: Huber Reime, Birgit; Schücking, Beate A.; Wenzlaff, Paul, 2008: Reproductive outcomes in ado- lescents who had a previous birth or an induced abortion compared to adolescents’ fi rst pregnancies. BMC Pregnancy and Childbirth 8. URL: [http://www.biomedcentral. com/bmcpregnancychildbirth, 23-12-2010] Sackmann, Reinhard, 1999: Ist ein Ende der Fertilitätskrise in Ostdeutschland absehbar? In: Zeitschrift für Bevölkerungswissenschaft 24,2: 187-211 Schneider, Sven; Maul, Holger; Freerksen, Nele; Pötschke-Langer, Martina, 2008: Who smokes during pregnancy? An analysis of the German Perinatal Quality Survey 2005. In: Public Health 122: 1210-1216 Smallwood, Steve, 2002: New estimates of trends in births by birth order in England and Wales. In: Population Trends 108: 32-48 Sobotka, Tomáš, 2004: Is lowest-low fertility in Europe explained by the postponement of childbearing? In: Population and Development Review 30,2: 195-220 Statistisches Bundesamt, 2009a: GENESIS-Online, die Internet-Datenbank des Statistischen Bundesamtes. URL: [https://www-genesis.destatis.de/genesis/online, 27-02-2009] Statistisches Bundesamt, 2009b: Entbindungen und Geburten im Krankenhaus nach Ländern. (Excel Sheets provided upon correspondence to Statistisches Bundesamt (Ute Bölt)), 30-10-2009 Toulemon, Laurent, 2001: How many children and how many siblings in France in the last century? In: Population & Sociétés 374: 1-4 Voigt, Manfred; Hullen, Gert, 2005: Die paritätsspezifi sche Fertilität. Demographische Analysen der Perinatalstatistik. In: Der Gynäkologe 38,7: 650-654 Voigt, Manfred; Hermanussen, Michael; Wittwer-Backofen, Ursula; Fusch, Christoph; Hesse, Volker, 2006: Sex-specifi c differences in birth weight due to maternal smoking during pregnancy. In: European Journal of Pediatrics 165: 757-761 A German translation of this authorised original article by the Authors is available under the title “Ordnungsspezifi sche Geburtenraten für Deutschland: Schätzungen auf Basis der Perinatalstatis- tik für die Jahre 2001-2008”, DOI 10.4232/10.CPoS-2010-06de or URN urn:nbn:de:bib-cpos-2010- 06de2, at http://www.comparativepopulationstudies.de. Michaela Kreyenfeld ( ), Rembrandt Scholz, Frederik Peters, Ines Wlosnewski. Max Planck Institute for Demographic Research, Rostock, Germany E-Mail: kreyenfeld@demogr.mpg.de, scholz@demogr.mpg.de, peters@demogr.mpg.de, ines.wlosnewski@uniklinik-freiburg.de URL: www.demogr.mpg.de Published by / Herausgegeben von Prof. Dr. Norbert F. Schneider Layout and print: Federal Institute for Population Research, Wiesbaden (Germany) Managing Editor / Redaktion Frank Swiaczny Copy Editor / Schlußredaktion Dr. Evelyn Grünheid Scientifi c Advisory Board / Wissenschaftlicher Beirat Jürgen Dorbritz (Wiesbaden) Paul Gans (Mannheim) Johannes Huinink (Bremen) Dirk J. van de Kaa (Den Haag) Marc Luy (Wien) Notburga Ott (Bochum) Peter Preisendörfer (Mainz) Comparative Population Studies – Zeitschrift für Bevölkerungswissenschaft www.comparativepopulationstudies.de ISSN: 1869-8980 (Print) – 1869-8999 (Internet) Board of Reviewers / Gutachterbeirat Martin Abraham (Erlangen) Laura Bernardi (Lausanne) Hansjörg Bucher (Bonn) Claudia Diehl (Göttingen) Andreas Diekmann (Zürich) Gabriele Doblhammer-Reiter (Rostock) Henriette Engelhardt-Wölfl er (Bamberg) E.-Jürgen Flöthmann (Bielefeld) Alexia Fürnkranz-Prskawetz (Wien) Beat Fux (Zürich) Joshua Goldstein (Rostock) Karsten Hank (Mannheim) Sonja Haug (Regensburg) Franz-Josef Kemper (Berlin) Hans-Peter Kohler (Philadelphia) Michaela Kreyenfeld (Rostock) Aart C. Liefbroer (Den Haag) Kurt Lüscher (Konstanz) Dimiter Philipov (Wien) Tomáš Sobotka (Wien) Heike Trappe (Rostock) © Federal Institute for Population Research 2010 – All rights reserved Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext