G Model
ALCR-119; No. of Pages 13 Advances in Life Course Research xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Advances in Life Course Research journal homepage: www.elsevier.com/locate/alcr
Unintended pregnancy in the life-course perspective Cornelia Helfferich a,*, Angelika Hessling b, Heike Klindworth c, Ines Wlosnewski d a
Institute for Social Science Research on Women (SoFFI F.)/Ev. Hochschule Freiburg, Bugginger Str. 38, D 79114 Freiburg, Germany Federal Centre of Health Education, BZgA, Ostmerheimer Str. 220, D 51109 Ko¨ln/Cologne, Germany c Institute of Sociology, University of Freiburg, c/o Institute for Social Science Research on Women (SoFFI F.), Bugginger Str. 38, D 79114 Freiburg, Germany d Institute for Social Science Research on Women (SoFFI F.), Bugginger Str. 38, D 79114 Freiburg, Germany b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 23 May 2013 Received in revised form 31 March 2014 Accepted 1 April 2014
Objective: In this contribution unintended pregnancies are studied as a multidimensional concept from a life-course perspective. Standardized data on the prevalence of unwanted pregnancies in different stages of women’s life course are combined with a qualitative analysis of the subjective meaning of ‘‘unwanted’’ and of subjective explanations of getting pregnant unintentionally. Methods: The study ‘‘frauen leben 3’’ on family planning in the life course of 20–44 year old women was conducted on behalf of the Federal Centre for Health Education (BZgA) from 2011 until 2014 in four federal states in Germany. A standardized questionnaire was used to collect retrospective information on 4794 pregnancies (including induced abortions), and biographical in-depth interviews provide qualitative information on 103 unwanted pregnancies. The standardized data were analyzed with bivariate methods and multivariate logistic regression models. The qualitative procedure to construct typologies of subjective meanings consisted of contrasting cases according to the generative approach of Grounded Theory. Main results: In contrast to unwanted pregnancies, mistimed pregnancies are characterized to a greater extent by negligence in the use of contraceptives, by a positive reaction to the pregnancy and by a more general desire to have a child. Four different subjective meanings of ‘‘unwanted’’ are constructed in qualitative analysis. The logistic regressions show that the selected factors that increase the likelihood of an unwanted pregnancy vary according to age and stage in the life course. The quantitative analysis reveals furthermore that relationship with a partner had a significant effect in all stages of the life course. The qualitative interviews specify the age- and life course-related aspects of these effects. ß 2014 Published by Elsevier Ltd.
Keywords: Unintended pregnancy Unwanted pregnancy Mistimed pregnancy Life course Age at birth/abortion Family planning
1. Introduction Unintended pregnancies account for an appreciable proportion of total fertility. They are also considered an * Corresponding author. Tel.: +49 761 47812 690. E-mail addresses: [email protected] (C. Helfferich), [email protected] (A. Hessling), soffi@eh-freiburg.de (H. Klindworth), [email protected] (I. Wlosnewski).
important indicator of reproductive health (Mosher, Jones, & Abma, 2012). In an international overview, 40% of all pregnancies in West Europe and 48% of those in the USA were classified as unintended (Singh, Wulf, Hussain, Bankole, & Sedgh, 2009: 52f), whereas national Studies in France and the Netherlands report a proportion of approximately one third (Dreesen & Matthijs, 2010) The majority of studies on unintended pregnancies collect information on the prevalence of unintended pregnancies
http://dx.doi.org/10.1016/j.alcr.2014.04.002 1040-2608/ß 2014 Published by Elsevier Ltd.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 2
C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
by means of retrospective surveys on previous pregnancies (e.g. Font-Ribera, Pe´rez, Salvador, & Borrell, 2007; Santelli, Duberstein Lindberg, Orr, Finer, & Speizer, 2009; Sihvo, Bajos, Ducot, Kaminski, & The COCON Group, 2003). A cross-national comparison of existing surveys is problematic because of variations in the frame of reference (all pregnancies including induced abortions, or just live births), in the retrospective timeframe and even in the definition of ‘‘unintended’’. Pregnancy and fertility intentions1 are the subject of both demographic studies and research on reproductive health. Fertility intentions, defined as future intentions, expectations and the desire to have a certain number of children (or no children), are a key indicator for predictions of population development in demography, based on microeconomic models of sociological action theories (Rational Choice Theory: Huinink, 1995) or social-psychological action theories (Theory of Planned Behavior, Theory of Conjunctural Action; overview: Philipov, 2011). In contrast, the purpose of studies in reproductive health is to find the best preventative methods, such as sex education and access to contraceptives, to reduce the number of unwanted pregnancies and induced abortions. These studies focus on social and cultural framing conditions, on the procedural nature of events, and differentiate according to stage of the life course and target groups. Consequently, they emphasize the multidimensionality and complexity of pregnancy intentions (Bajos et al., 2003; BZgA, 2013; Santelli et al., 2009). In this contribution we continue the tradition of research on retrospectively reported unintended pregnancy as an aspect of reproductive health and follow FontRibera et al. (2007) and Sihvo et al. (2003) in adopting a life-course approach. We report on both standardized and qualitative findings on unintended pregnancy. The contribution forges a link between empirical studies of pregnancy intentions in the field of reproductive health and studies of fertility intentions in the field of demography. We can add to discussions in the field of demography on how behavioral models of fertility could be improved in four ways. First, inconsistencies – there is no desire for a child but a pregnancy occurs – are an explicit issue of our research. In demographic research, Philipov and Bernardi (2011, 553ff) propose extending the central term ‘‘fertility intention’’ to include not only intentions that are clearly conscious, can be explicitly formulated and (potentially) implemented without ambivalence. Neutral intentions (neither intending to have a child nor not to have a child) and ambivalent intentions (alternating between the desire for and the fear of a child, conflicting goals) should be regarded, too. Second, the behavioral elements of sexual activity and (non-)contraception assume a prominent position in research on unintended pregnancy. This supports Miller and Pasta (1995) who propose these
1 We adopt the terminology employed by Santelli et al. (2009) who use the term ‘‘pregnancy intentions’’ in the context of retrospective reporting of experienced pregnancies and the term ‘‘fertility intentions’’ for demographic forecasts of population development.
behaviors instead of births as the predictable measures in the Theory of Planned Behavior. They criticize that conception or birth is not a behavior in the proper sense, but rather an event. Third, a differentiation according to social position and life-course stage is central to research with a preventative orientation. This is in accordance with Morgan’s (2001) critical call for specifying fertility intentions for different social groups and for considering changes in fertility intentions over the life course in demographic studies. Fourth, our findings provide a better insight in the importance of the relationship as a context of pregnancy or fertility intentions, which is also increasingly sought after by demographers. After the outline of the theoretical background and the research questions (Section 2), data and methods are described (Section 3). The presentation of results (Section 4) starts with the comparison of the two most important subtypes of unintended pregnancies: mistimed and unwanted pregnancies (Section 4.1). In the next step (Section 4.2), logistic regressions are used to model constellations of explanatory factors for conceiving an unwanted pregnancy in different stages of the life course. Reconstructing the meaning of unwanted pregnancies from qualitative, biographical interviews (Section 4.3) contributes to the explanation of statistical correlations and clarifies subjective meanings of unwantedness. In the closing section (Section 5) the results and conclusions for demography are discussed. 2. Background and research questions The definition of unintended pregnancy and its operationalization varies between different studies. However, a standard measure comprising three categories has become established: intended, mistimed and unwanted (Mosher et al., 2012: 3f for the National Survey of Family Growth (NSFG) in the USA, with a further differentiation between ‘‘moderately’’ and ‘‘seriously mistimed’’ as more than or less than two years too early). The frequently used term ‘‘unintended’’ encompasses ‘‘unwanted’’ and ‘‘mistimed’’ (and sometimes additionally ‘‘undecided’’). The key dimensions for operationalizing the NSFG definition are – in addition to the desire (wanting) to have a child – the dimensions contraception and timing of the pregnancy (Santelli et al., 2003). Pregnancies that occur in spite of contraception are termed unwanted as are pregnancies with ‘‘no contraception and no desire for a child’’. Furthermore, it is suggested to include the scaled indicator ‘‘Happiness at being pregnant’’ as an emotional response to conceiving (Santelli et al., 2003). In the ‘‘London Measure of Unplanned Pregnancies’’, Barrett, Smith, and Wellings (2004) allocate scores to the six dimensions of contraceptive behavior timing, intention and feelings about having a baby, partner communication and preparation for pregnancy -, all of which are related to intendedness. The overall score is a summed value between 0 (highly unwanted) and 12 (highly wanted) that can be considered a quantified degree of planning. But Dreesen and Matthijs (2010) criticize that the dimensions are independent of each other and should therefore be separated analytically. Furthermore, in case of inconsistencies between dimensions an overall score
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
cannot provide information on which dimensions indicate that a pregnancy was intended and which dimensions imply it was not. Such inconsistencies have consistently been found in studies in different Western countries. They relate to the appreciable proportion of unintended pregnancies in which no contraceptives were used and/ or in which the reaction to conception was positive (for the USA: Edin, England, Fizgibbons Shafer, & Reed, 2007, 26; Sable & Libbus, 2000; Trussell, Vaughan, & Stanford, 1999 for a small, selective sample; for European countries: Dreesen & Matthijs, 2010 for the Netherlands, Barrett et al., 2004 for England, Sihvo et al., 2003 for France, FontRibera et al., 2007 for Spain, Helfferich, Karmaus, Starke, & Weller, 2001; Helfferich, Klindworth, & Kruse, 2005 for Germany). On the whole, ‘‘unintendedness’’ is considered a multidimensional concept. Consequently, ‘‘unintendedness’’ should be viewed as more of a continuum than a strict polar opposite to ‘‘intendedness’’ (Santelli et al., 2009, 88). However, comparisons of different kinds of unintendedness are rarely the subject of research. Differences concerning the socioeconomic characteristics of women with mistimed and unwanted pregnancies were found by Mosher et al. (2012), D’Angelo, Gilbert, Rochat, Santelli, and Herold (2004) and Sihvo et al. (2003, 605). Mosher et al. (2012, 11) show that on all three scales ‘‘Happiness when found out pregnant’’, ‘‘Trying hard to get/ to avoid pregnancy right before’’, ‘‘Wanted to get/to avoid pregnancy right before’’ the scores for mistimed pregnancies – and especially mistimed by more than two years – were between those for intended and for unwanted pregnancies. Sable and Libbus (2000, USA) come to a similar finding for women and Helfferich et al. (2005, 204f, Germany) for men. There is little research on the changes in the proportion of mistimed and unwanted pregnancies over the life course. Mosher et al. (2012) and D’Angelo et al. (2004) find a decline in mistimed pregnancies with increasing age or parity. Santelli et al. (2009, 88) differentiate the prevalence of mistimed and unwanted pregnancies in different stages of the life course: women who are still in education or training and not yet married reported more mistimed pregnancies, whereas women at the end of their childbearing period reported more unwanted pregnancies. All studies consistently find that pregnancies of under 24year-olds or under 20-year-olds are more frequently unwanted or unintended than pregnancies of older women (Dreesen & Matthijs, 2010; Font-Ribera et al., 2007; Helfferich et al., 2001). Our first research question reads: Is there a difference between mistimed and unwanted pregnancies leading to babies in terms of prior contraceptive behavior and emotional response to conceiving? Is the prevalence of mistimed and unwanted pregnancy related to age? Research on the factors influencing an unintended or unwanted pregnancy only partially differentiates according to life stage. Several studies confirm the role of age, relationship status, birth order or number of children already born, education and socioeconomic status. Relationship status, which can be operationalized as marital status (married/unmarried, cohabiting/non-cohabiting) or
3
relationship duration, is described in different studies as ‘‘the strongest determinant of unintended pregnancy’’ (Dreesen & Matthijs, 2010, 14; Font-Ribera et al., 2007, Spain, Netherlands). The specific relationship form is found to have an age-independent influence in Germany (Helfferich et al., 2001, 223). According to the American NSFG, the influence of the relationship is independent of life-course stage (Mosher et al., 2012, 8). Together with age, the birth order or number of children already born has a considerable effect. Mosher et al. (2012, 8) report that the proportion of unwanted pregnancies increases with the birth order. Studies on Germany (Feldhaus & Boehnke, 2008; Helfferich et al., 2001, 246) show a higher proportion of unwanted or mistimed pregnancies for the first child and more unwanted pregnancies for the third child. Several studies consider low income and a low level of formal education to be risk factors (Font-Ribera et al., 2007, 128; Mosher et al., 2012, 8f und 14; Rossier, Michelot, Cocon Group, & Bajos, 2006, 21). However, in Germany there is no significant difference in intendedness attributable to education level (Helfferich et al., 2001, 245). The duration of further education and training did however play a role, in so far as the pregnancies of young mothers still in education or training were more frequently unwanted or unintended (Helfferich, Hendel-Kramer, & Wehner, 2007). At the macrostructural level in Germany, specific characteristics of the federal states are associated with different prevalence of unintendedness. The city states (Berlin, Hamburg, Bremen), for example, have highest rate of induced abortions followed by the states in the East (with a historical tradition as a socialist country) (Statistisches Bundesamt, 2011, 37f). The rates of induced abortion correspond to higher levels of indicators for economic problems such as poverty or the unemployment rate. All of these factors interact and the effect of age and relationship status can disappear after controlling for other related variables (e.g. relationship duration; Dreesen & Matthijs, 2010; Rossier et al., 2006). In contrast, other factors, such as parity or participation in education or training, depend on age and life-course stage. A low educational level and young age at first birth (both risk factors for an unwanted pregnancy) are also interrelated. Several studies adopt the life-course perspective by looking for age-dependent and life-stage dependent conditions that contribute to an unintended pregnancy. Font-Ribera et al. (2007), for example, compare unintended pregnancies in the age groups 15–24, 25–39 and 40–49, as do Sihvo et al. (2003) with slightly different age boundaries. Rossier et al. (2006) compute the risk of unintended pregnancy as the product of the frequency of sexual activity, the use and the pearl index of contraception and the probability of conceiving. They show that sexual activity, access to contraception, desire for children, and the resources for raising a child are affected by separate age-dependent social influences and that their characteristics and their impact vary across age groups. Consequently, there are different explanations for unintended pregnancy among young (childless) women, for
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 4
C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
women in the middle age group who are in a stable longterm relationship and for women over 34 who already have children (Rossier et al., 2006; Sihvo et al., 2003). The situation in Germany is characterized by a high average age at first birth (2010: 29.2 years in West Germany and 27.4 in East Germany; Statistisches Bundesamt, 2012). It is mainly highly-qualified West German women who are worried about getting pregnant too early (for differences between East and West Germany see Goldstein, Kreyenfeld, Huinink, Konietzka, & Trappe, 2010). In Germany the is a widespread belief that the father or the mother should have completed their education or training and have a secure job before they have their first child (Institut fu¨r Demoskopie Allensbach, 2010). The prevalence rates for motherhood during vocational training (3%) or university studies (6%) are correspondingly low (BMFSFJ, 2010) and soon after having completed training or education the transition rate into parenthood increases sharply (‘‘institutional effect’’: Billari & Philipov, 2006). The second research question reads: Which factors affect the prevalence of unwanted pregnancies in younger, middle and higher aged women in Germany and how do these factors interact in each age group? Qualitative data can help understand causal mechanisms behind correlations uncovered by standardized surveys. In research on reproductive health, qualitative studies on the subjective orientations of target groups are considered important in order to develop targeted prevention programs. Such programs ‘‘must build upon cultural understanding of the problem to be prevented. Research should focus on the meaning of pregnancy intentions to women and the processes women and their partners use in making fertility decisions. (. . .). Both qualitative and quantitative research, have contributed to our understanding of fertility decision-making; both will be essential to the creation of more effective prevention programs’’ (Santelli et al., 2003: 99). Qualitative studies can describe the ‘‘perceived cultural appropriateness of behavior according to age, demographic position and personal biography’’ (Sihvo et al., 2003: 601). They are also being called for in demography (e.g. Von der Lippe & Fuhrer, 2004). Trussell et al. (1999) conclude that not all women believe equally that a child is something to be planned and proposed qualitative studies to explore subjective pregnancy intentions. Several qualitative studies highlight the diversity and ambiguity of subjective meanings of pregnancy intentions. Bernardi and Mynarska (2010) construct a typology of future fertility intentions: in addition to ‘‘surely yes’’ and ‘‘surely not’’ they describe types with different planning horizons (‘‘as soon as’’, ‘‘surely one day’’) as well as openness and instability (‘‘maybe’’, ‘‘at times’’; cf. Sobotka, 2011, 80ff). In addition to ‘‘planned pregnancy’’, Earle (2004) distinguishes between ‘‘laissez faire’’, ‘‘recalcitrant (ambiguous)’’ and ‘‘accidental’’ pregnancies in retrospective interviews with primagravidae. Helfferich and Kandt (1996, 60f) analyze the statements made in qualitative interviews by interviewees who had described a pregnancy as unwanted in a prior standardized survey. Further qualitative studies employ in depth interviews to investigated family planning embedded in biography
(e.g. Rijken & Knijn, 2009). Helfferich et al. (2001, 2005) study specific biographical contexts of family planning and unwanted pregnancies from the women’s and the men’s perspective. The potential of qualitative research, in particular qualitative biographical research in combination with standardized life-course research (Kelle & Erzberger, 1999), has by no means been exhausted. The third research question includes the subjective perspective and complements the second research question: What are the typical subjective interpretations of the situation of unwanted pregnancy in the narrative passages of interviews and how do these interpretations vary for pregnancies in young, middle and older age? Furthermore, we ask whether certain subjectively perceived contexts of unwanted pregnancy are associated with specific concepts of intendedness and, as part of this, with a subjective theory that explains how the unintended pregnancy came about. 3. Data and methods The three research questions were investigated with both standardized data (research question 1 and 2) and qualitative data (research question 3). Different analytical methods that consider the role of specific independent variables in subsamples of unintended pregnancies are employed in answering the research questions 1 and 2. 3.1. Data source The data are drawn from the study ‘‘Family planning in the life course of women’’ (‘‘frauen leben 3’’, 2012–2014), carried out by the Institute for Social Science Research on Women (SoFFI F.), Freiburg (Germany) on behalf of the Federal Centre for Health Education, BZgA, Cologne, and in cooperation with the Institute of Sociology of the University of Freiburg (fieldwork February to July 2012). For this study, retrospective standardized and qualitative data on reproductive behavior and reproductive events in the lives of 20 to 44 year old women were collected. 3.1.1. Standardized telephone interviews An age-stratified random sample of 4002 women was drawn from the telephone directory of four federal states in Germany (Berlin, Baden-Wu¨rttemberg, Saxony and Lower Saxony), which differ according to patterns of fertility (for example, abortion rate, age at birth of the first child) and social indicators (for example, poverty and unemployment rates). In total 4794 pregnancies (4427 live births, 367 induced abortions) were reported by women who at least once in their life experienced a pregnancy (n = 2584). The standardized questionnaire included questions on intended and non-intended pregnancies ending in a live birth as well as in abortion.2 3.1.2. Qualitative semi-structured interviews At the end of the telephone interview, women who said that they had experienced at least one unintended
2 No cases of stillbirth or miscarriage were considered. The term ‘‘abortion’’ applies to induced abortion only.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
pregnancy were invited to recount their biographies at a later date. A sample of ninety-seven women was recruited, who spoke of their lives from childhood until the date of the interview in semi-structured, in-depth interviews. The sample was constructed on the basis of socio-demographic data derived from standardized surveys in accordance with quality criteria to maximize its diversity and contrast in terms of age, education and number of children. The guidelines for the interview encouraged the interviewee to give a free narrative about the life-course stages of childhood, youth and adulthood with specific questions on sexual behavior, relationships and reproductive events. The face-to-face interviews lasted 45 min and the complete recordings were transcribed. The current analysis focuses on the reports of 103 unwanted pregnancies (including 45 abortions). In addition, there are also narratives on 34 mistimed pregnancies, 14 undecided and 63 wanted pregnancies. The standardized and qualitative data can be matched at the respondent level. 3.2. Data analysis 3.2.1. Bivariate analysis of standardized data In the analyses of Sections 4.1 and 4.2, the unit investigated is pregnancies. The dependent variable is pregnancy intention at the time of conception, determined by the answer to the question ‘‘Did you want to get pregnant at that time?’’ The response options were ‘‘wanted then or sooner’’,3 ‘‘wanted, but it should have been later’’ (=mistimed), ‘‘unwanted’’ and ‘‘ambivalent, undecided’’. According to a commonly used procedure (Font-Ribera et al., 2007; Santelli et al., 2009, 88; Sihvo et al., 2003), all terminated pregnancies were considered to be unwanted. The life-course perspective is incorporated into the analyses of the three research questions by means of the differentiation according to mother’s age at birth of her child or termination of the pregnancy (‘‘under 25’’, ‘‘25–34’’, ‘‘35– 44’’). The boundary for the youngest age group is drawn at 25 years because it is rare to have a stable job (considered one of the major prerequisites for having children in Germany) before reaching this age. The boundary for the oldest age group was drawn at 35 years because this age is a criterion for ‘‘late’’ motherhood (Engstler & Lu¨scher, 1991, 438). There were not enough pregnancies in women over 34 in the data set studied here to allow a further differentiation of this age group. Bivariate analyses using the chi-squared tests were computed to discover the differences between mistimed and unwanted pregnancies according to age and with respect to contraceptive behavior, the emotional reaction to pregnancy and general attitudes to children. The emotional reaction to conception was recorded with a six-point happiness scale, with 1 indicating ‘‘very happy’’ and 6 indicating ‘‘very unhappy’’. Furthermore, we asked whether contraceptives were in use at the time of conception (for all pregnancies). General attitudes toward children were
3 In the following the terms ‘‘wanted then or sooner’’, ‘‘wanted (then)’’ and ‘‘at the right time’’ are treated as equivalent.
5
operationalized with the variable ‘‘I always wanted to have children’’ (agree fully, disagree fully, agree partially). It was assumed that there are negative feelings associated with having an abortion and therefore it did not seem appropriate to ask interviewees who reported an abortion whether they had been happy with their pregnancy. Consequently, no data are available on the emotional response to pregnancies that were terminated. At this stage of the analysis only pregnancies ending in a live birth were included (n = 4402). Chi-squared tests with p-values below 0.05 indicate significant differences. 3.2.2. Multivariate analysis of standardized data To investigate the research question which factors affect pregnancy intentions in young, middle or older age, only pregnancies (births and induced abortions) described as ‘‘wanted then or sooner’’ or as ‘‘unwanted’’ are compared. The exclusion of mistimed pregnancies is justified by the results of investigating the first research question which show that in several dimensions the mistimed pregnancies resemble the wanted pregnancies more closely than the unwanted ones. Undecided pregnancies are excluded because of the small number of cases. For multivariate analyses, logistic regression was used to estimate the effects of selected independent variables on unwanted pregnancies. These multivariate models were estimated separately for the three predefined groups of age at conception. Results are presented as odds ratios that indicate the relative likelihood of experiencing an unwanted pregnancy. The independent variables selected on the basis of theoretical considerations (see Section 2) were collected or operationalized as follows: For all pregnancies, including those that were terminated, information on the respondent’s personal situation at the time of conception was recorded by means of a list of questions relating to a difficult relationship,4 financial or job insecurity and whether they were in education or training at the time. The response options were operationalized dichotomously (yes/no). For the age group over 34 at conception the variable ‘‘currently in education or training’’ could not be included in the analyses because of the small number of cases. Education level was categorized in four groups: low, medium, high and higher.5 It should be noted that this variable records the level of education at the time of the
4 The survey posed separate questions on whether, at the time of conception, the respondent was in a relationship or not and if so whether the relationship was instable or troubled, for example, because the partner did not want a child at that time. Since the overwhelming majority of pregnancies (85%) were conceived within a stable relationship, the separate categories for problematic relationships were merged in the further analyses. 5 Education level was operationalized as a combination of schooling, vocational training qualifications and tertiary education: low level = Hauptschule (9 years) and vocational training or Realschule/POS (10 years) without vocational training; medium level = Hauptschule and college of higher education/master school/technical school or Realschule/ POS and vocational training; high level = Realschule/(Fach-)abitur (12/13 years) and college of higher education/master school/technical school or (Fach-)Abitur with or without vocational training; higher level: (Fach)abitur and university degree (university or university of applied sciences) or currently a university (of applied sciences) student.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
6
4. Results
survey, not at the time of the pregnancy. In determining regional effects, the two West German federal states Baden-Wu¨rttemberg and Lower Saxony were merged into one variable (‘‘west’’) in the regression model since there were hardly any differences in the distributions of the relevant factors and the analysis was primarily concerned with East-West and urban-rural differences. Since third and further births were relatively rare, the number of children born prior to the pregnancy of interest was operationalized in three groups: no children, one child and two or more children. Marital status at the time of conception distinguishes between never married and previously married women. The seven variables indicating possible influences on the pregnancy intention were included in three blocks into the logistic regression. In the first step, the effect of relationship quality, financial and job security, number of children already born, education and region in the three age groups were estimated. In the second and third steps, the additional effects of being in education or training and marital status were included, and improvements in the model’s goodness of fit were tested. After excluding cases with missing data on the independent variables, information on 3872 pregnancies from 1933 women could be included in the logistic regression. Of these pregnancies, 80% were wanted then or sooner and 20% were unwanted. Independent variables are considered significant if their effect on unintended pregnancy is statistically significant at the 95% level.
Of all the pregnancies (including induced abortions) reported by the 20–44 year old women, 66.3% were wanted then or sooner, 13.3% were mistimed and 17.7% were unwanted. Contraceptives were in use before 12% of all pregnancies (and 35% of unintended pregnancies). 4.1. Mistimed, unwanted and intended pregnancies that led to babies Mistimed and unwanted pregnancies differ significantly in several analytic dimensions. As to mistimed pregnancies the use of contraceptives is significantly less likely than as to unwanted pregnancies (28.4–47.2%). The frequency of a positive reaction to the news of a pregnancy was significantly higher when the pregnancy was mistimed: the reaction to 55.9% of the mistimed pregnancies was (very) positive (scores of 1 or 2 on a six-point scale) and (very) negative (scores of 5 or 6) to 8.4%. In comparison, the proportion of positive reactions is lower among unwanted pregnancies (22.4%) and the proportion of negative reactions higher (35.4%). The pattern ‘‘no contraception and positive response’’ dominates mistimed pregnancies with 44.9%, but remains marginal at 15.3% for unwanted pregnancies. This shows that mistimed pregnancies, compared to unwanted pregnancies, are more similar to intended pregnancies, for which generally can be assumed that contraception was omitted and the reaction was positive. A (very) negative response is more likely if the pregnancy was unwanted and contraception was used (42.9%), a (very) positive response is more likely if the pregnancy was mistimed and no contraception was used (62.9%). Only 10.5% of all unintended (mistimed or unwanted or undecided) pregnancies that were carried to term were consistent in the sense that contraception was used and the reaction to conception was negative (score of 5 or 6 on the six-point scale). Of all pregnancies that ended with a live birth, the proportions of mistimed and unwanted pregnancies depend on the mother’s age. Mistimed pregnancies become rarer with increasing age. However, the share of unwanted pregnancy is smallest in the middle age group (25–34 years, see Table 1). The results underline the fear of getting pregnant ‘‘too early’’. But mistimed pregnancies still account for 12% of all pregnancies in the age group 25–34 years. Obviously ‘‘too
3.2.3. Analysis of qualitative data The qualitative analysis covered 58 unwanted pregnancy carried to term and 45 abortions. The dimensions along which the narratives differed were determined with a contrasting comparison of the minimal and maximal differences in accordance with the generative approach of Grounded Theory. The women’s accounts were then grouped into four categories according to the main focus of concern which was expressed while talking about the pregnancy, e.g. conflict with partner or going back to work. Within the four categories narrations from women of different age were compared. In a further step, subjective concepts of whether and how a child can be ‘‘planned’’ as well as the use of the term ‘‘unwanted’’ with respect to a pregnancy were extracted and grouped. The relationship between specific ideas about planning and the conditions within which an unwanted pregnancy was conceived was discussed.
Table 1 Intendedness of pregnancies ending in a live birth by mother’s age at the birth (in %). Age at birth*
Under 25 years
25–34 years
35 years and older
Total
Intendedness
n = 1171
n = 2824
n = 394
n = 4389
Wanted then or sooner Mistimed Undecided Unwanted Total
54.7 23.4 3.5 18.4 100.1
77.9 12.0 2.6 7.5 100
78.9 5.1 3.8 12.2 100
71.8 14.5 2.9 10.8 100
Source: BZgA, data set ‘‘frauen leben 3’’, 2012, 20–44 year old women in four federal states. * p < 0.001.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
7
15.6 3.1
"I always wanted to have children": agree
24.2 57.1
unwanted undecided mismed
47.5 7.9
"I always wanted to have children": do not agree/agree parally
wanted then or sooner
14.9 29.7 0.0
20.0
40.0
60.0
Fig. 1. Pregnancy intention and item ‘‘I always wanted to have children’’ (in %, age at birth younger than 25 years).
early for a child’’ does not only refer to biological age but also to other timelines (like the time since a previous birth) or it extends well beyond the age of 25 years. Mistimed and unwanted pregnancies differ along the mother’s long-term intention to have children. This intention was recorded with the item ‘‘I always wanted to have children’’. Fig. 1 shows for an age at birth under 25 years: If women agree with this item, a pregnancy was significantly more frequently declared to be mistimed and less often to be unwanted. This effect is pronounced in the youngest age groups (under 25) and disappears in the oldest age group (35 years and older). In the age group 25– 34 years (age at birth) the proportion of mistimed pregnancies is not influenced by the long-term intention to have children, but those who did not or partially agree with the item had a significantly higher proportion of unwanted pregnancies (17% vs. 6.3%). 4.2. Factors affecting conception of an unwanted pregnancy in young, middle and older age The comparison of mistimed and unwanted pregnancies has shown that along certain dimensions mistimed pregnancies are more similar to intended pregnancies. Therefore, the following multivariate analysis concentrates on the factors affecting conception of an unwanted pregnancy in comparison with a pregnancy that was wanted and the time was right or it should occur sooner (=intended). Table 2 shows the results of the logistic regressions used to estimate the likelihood (odds ratio) that a pregnancy was unwanted vs. wanted in dependency on selected factors. The logistic regressions were estimated separately for pregnancies in young women (
ALCR-119; No. of Pages 13 Advances in Life Course Research xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Advances in Life Course Research journal homepage: www.elsevier.com/locate/alcr
Unintended pregnancy in the life-course perspective Cornelia Helfferich a,*, Angelika Hessling b, Heike Klindworth c, Ines Wlosnewski d a
Institute for Social Science Research on Women (SoFFI F.)/Ev. Hochschule Freiburg, Bugginger Str. 38, D 79114 Freiburg, Germany Federal Centre of Health Education, BZgA, Ostmerheimer Str. 220, D 51109 Ko¨ln/Cologne, Germany c Institute of Sociology, University of Freiburg, c/o Institute for Social Science Research on Women (SoFFI F.), Bugginger Str. 38, D 79114 Freiburg, Germany d Institute for Social Science Research on Women (SoFFI F.), Bugginger Str. 38, D 79114 Freiburg, Germany b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 23 May 2013 Received in revised form 31 March 2014 Accepted 1 April 2014
Objective: In this contribution unintended pregnancies are studied as a multidimensional concept from a life-course perspective. Standardized data on the prevalence of unwanted pregnancies in different stages of women’s life course are combined with a qualitative analysis of the subjective meaning of ‘‘unwanted’’ and of subjective explanations of getting pregnant unintentionally. Methods: The study ‘‘frauen leben 3’’ on family planning in the life course of 20–44 year old women was conducted on behalf of the Federal Centre for Health Education (BZgA) from 2011 until 2014 in four federal states in Germany. A standardized questionnaire was used to collect retrospective information on 4794 pregnancies (including induced abortions), and biographical in-depth interviews provide qualitative information on 103 unwanted pregnancies. The standardized data were analyzed with bivariate methods and multivariate logistic regression models. The qualitative procedure to construct typologies of subjective meanings consisted of contrasting cases according to the generative approach of Grounded Theory. Main results: In contrast to unwanted pregnancies, mistimed pregnancies are characterized to a greater extent by negligence in the use of contraceptives, by a positive reaction to the pregnancy and by a more general desire to have a child. Four different subjective meanings of ‘‘unwanted’’ are constructed in qualitative analysis. The logistic regressions show that the selected factors that increase the likelihood of an unwanted pregnancy vary according to age and stage in the life course. The quantitative analysis reveals furthermore that relationship with a partner had a significant effect in all stages of the life course. The qualitative interviews specify the age- and life course-related aspects of these effects. ß 2014 Published by Elsevier Ltd.
Keywords: Unintended pregnancy Unwanted pregnancy Mistimed pregnancy Life course Age at birth/abortion Family planning
1. Introduction Unintended pregnancies account for an appreciable proportion of total fertility. They are also considered an * Corresponding author. Tel.: +49 761 47812 690. E-mail addresses: [email protected] (C. Helfferich), [email protected] (A. Hessling), soffi@eh-freiburg.de (H. Klindworth), [email protected] (I. Wlosnewski).
important indicator of reproductive health (Mosher, Jones, & Abma, 2012). In an international overview, 40% of all pregnancies in West Europe and 48% of those in the USA were classified as unintended (Singh, Wulf, Hussain, Bankole, & Sedgh, 2009: 52f), whereas national Studies in France and the Netherlands report a proportion of approximately one third (Dreesen & Matthijs, 2010) The majority of studies on unintended pregnancies collect information on the prevalence of unintended pregnancies
http://dx.doi.org/10.1016/j.alcr.2014.04.002 1040-2608/ß 2014 Published by Elsevier Ltd.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 2
C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
by means of retrospective surveys on previous pregnancies (e.g. Font-Ribera, Pe´rez, Salvador, & Borrell, 2007; Santelli, Duberstein Lindberg, Orr, Finer, & Speizer, 2009; Sihvo, Bajos, Ducot, Kaminski, & The COCON Group, 2003). A cross-national comparison of existing surveys is problematic because of variations in the frame of reference (all pregnancies including induced abortions, or just live births), in the retrospective timeframe and even in the definition of ‘‘unintended’’. Pregnancy and fertility intentions1 are the subject of both demographic studies and research on reproductive health. Fertility intentions, defined as future intentions, expectations and the desire to have a certain number of children (or no children), are a key indicator for predictions of population development in demography, based on microeconomic models of sociological action theories (Rational Choice Theory: Huinink, 1995) or social-psychological action theories (Theory of Planned Behavior, Theory of Conjunctural Action; overview: Philipov, 2011). In contrast, the purpose of studies in reproductive health is to find the best preventative methods, such as sex education and access to contraceptives, to reduce the number of unwanted pregnancies and induced abortions. These studies focus on social and cultural framing conditions, on the procedural nature of events, and differentiate according to stage of the life course and target groups. Consequently, they emphasize the multidimensionality and complexity of pregnancy intentions (Bajos et al., 2003; BZgA, 2013; Santelli et al., 2009). In this contribution we continue the tradition of research on retrospectively reported unintended pregnancy as an aspect of reproductive health and follow FontRibera et al. (2007) and Sihvo et al. (2003) in adopting a life-course approach. We report on both standardized and qualitative findings on unintended pregnancy. The contribution forges a link between empirical studies of pregnancy intentions in the field of reproductive health and studies of fertility intentions in the field of demography. We can add to discussions in the field of demography on how behavioral models of fertility could be improved in four ways. First, inconsistencies – there is no desire for a child but a pregnancy occurs – are an explicit issue of our research. In demographic research, Philipov and Bernardi (2011, 553ff) propose extending the central term ‘‘fertility intention’’ to include not only intentions that are clearly conscious, can be explicitly formulated and (potentially) implemented without ambivalence. Neutral intentions (neither intending to have a child nor not to have a child) and ambivalent intentions (alternating between the desire for and the fear of a child, conflicting goals) should be regarded, too. Second, the behavioral elements of sexual activity and (non-)contraception assume a prominent position in research on unintended pregnancy. This supports Miller and Pasta (1995) who propose these
1 We adopt the terminology employed by Santelli et al. (2009) who use the term ‘‘pregnancy intentions’’ in the context of retrospective reporting of experienced pregnancies and the term ‘‘fertility intentions’’ for demographic forecasts of population development.
behaviors instead of births as the predictable measures in the Theory of Planned Behavior. They criticize that conception or birth is not a behavior in the proper sense, but rather an event. Third, a differentiation according to social position and life-course stage is central to research with a preventative orientation. This is in accordance with Morgan’s (2001) critical call for specifying fertility intentions for different social groups and for considering changes in fertility intentions over the life course in demographic studies. Fourth, our findings provide a better insight in the importance of the relationship as a context of pregnancy or fertility intentions, which is also increasingly sought after by demographers. After the outline of the theoretical background and the research questions (Section 2), data and methods are described (Section 3). The presentation of results (Section 4) starts with the comparison of the two most important subtypes of unintended pregnancies: mistimed and unwanted pregnancies (Section 4.1). In the next step (Section 4.2), logistic regressions are used to model constellations of explanatory factors for conceiving an unwanted pregnancy in different stages of the life course. Reconstructing the meaning of unwanted pregnancies from qualitative, biographical interviews (Section 4.3) contributes to the explanation of statistical correlations and clarifies subjective meanings of unwantedness. In the closing section (Section 5) the results and conclusions for demography are discussed. 2. Background and research questions The definition of unintended pregnancy and its operationalization varies between different studies. However, a standard measure comprising three categories has become established: intended, mistimed and unwanted (Mosher et al., 2012: 3f for the National Survey of Family Growth (NSFG) in the USA, with a further differentiation between ‘‘moderately’’ and ‘‘seriously mistimed’’ as more than or less than two years too early). The frequently used term ‘‘unintended’’ encompasses ‘‘unwanted’’ and ‘‘mistimed’’ (and sometimes additionally ‘‘undecided’’). The key dimensions for operationalizing the NSFG definition are – in addition to the desire (wanting) to have a child – the dimensions contraception and timing of the pregnancy (Santelli et al., 2003). Pregnancies that occur in spite of contraception are termed unwanted as are pregnancies with ‘‘no contraception and no desire for a child’’. Furthermore, it is suggested to include the scaled indicator ‘‘Happiness at being pregnant’’ as an emotional response to conceiving (Santelli et al., 2003). In the ‘‘London Measure of Unplanned Pregnancies’’, Barrett, Smith, and Wellings (2004) allocate scores to the six dimensions of contraceptive behavior timing, intention and feelings about having a baby, partner communication and preparation for pregnancy -, all of which are related to intendedness. The overall score is a summed value between 0 (highly unwanted) and 12 (highly wanted) that can be considered a quantified degree of planning. But Dreesen and Matthijs (2010) criticize that the dimensions are independent of each other and should therefore be separated analytically. Furthermore, in case of inconsistencies between dimensions an overall score
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
cannot provide information on which dimensions indicate that a pregnancy was intended and which dimensions imply it was not. Such inconsistencies have consistently been found in studies in different Western countries. They relate to the appreciable proportion of unintended pregnancies in which no contraceptives were used and/ or in which the reaction to conception was positive (for the USA: Edin, England, Fizgibbons Shafer, & Reed, 2007, 26; Sable & Libbus, 2000; Trussell, Vaughan, & Stanford, 1999 for a small, selective sample; for European countries: Dreesen & Matthijs, 2010 for the Netherlands, Barrett et al., 2004 for England, Sihvo et al., 2003 for France, FontRibera et al., 2007 for Spain, Helfferich, Karmaus, Starke, & Weller, 2001; Helfferich, Klindworth, & Kruse, 2005 for Germany). On the whole, ‘‘unintendedness’’ is considered a multidimensional concept. Consequently, ‘‘unintendedness’’ should be viewed as more of a continuum than a strict polar opposite to ‘‘intendedness’’ (Santelli et al., 2009, 88). However, comparisons of different kinds of unintendedness are rarely the subject of research. Differences concerning the socioeconomic characteristics of women with mistimed and unwanted pregnancies were found by Mosher et al. (2012), D’Angelo, Gilbert, Rochat, Santelli, and Herold (2004) and Sihvo et al. (2003, 605). Mosher et al. (2012, 11) show that on all three scales ‘‘Happiness when found out pregnant’’, ‘‘Trying hard to get/ to avoid pregnancy right before’’, ‘‘Wanted to get/to avoid pregnancy right before’’ the scores for mistimed pregnancies – and especially mistimed by more than two years – were between those for intended and for unwanted pregnancies. Sable and Libbus (2000, USA) come to a similar finding for women and Helfferich et al. (2005, 204f, Germany) for men. There is little research on the changes in the proportion of mistimed and unwanted pregnancies over the life course. Mosher et al. (2012) and D’Angelo et al. (2004) find a decline in mistimed pregnancies with increasing age or parity. Santelli et al. (2009, 88) differentiate the prevalence of mistimed and unwanted pregnancies in different stages of the life course: women who are still in education or training and not yet married reported more mistimed pregnancies, whereas women at the end of their childbearing period reported more unwanted pregnancies. All studies consistently find that pregnancies of under 24year-olds or under 20-year-olds are more frequently unwanted or unintended than pregnancies of older women (Dreesen & Matthijs, 2010; Font-Ribera et al., 2007; Helfferich et al., 2001). Our first research question reads: Is there a difference between mistimed and unwanted pregnancies leading to babies in terms of prior contraceptive behavior and emotional response to conceiving? Is the prevalence of mistimed and unwanted pregnancy related to age? Research on the factors influencing an unintended or unwanted pregnancy only partially differentiates according to life stage. Several studies confirm the role of age, relationship status, birth order or number of children already born, education and socioeconomic status. Relationship status, which can be operationalized as marital status (married/unmarried, cohabiting/non-cohabiting) or
3
relationship duration, is described in different studies as ‘‘the strongest determinant of unintended pregnancy’’ (Dreesen & Matthijs, 2010, 14; Font-Ribera et al., 2007, Spain, Netherlands). The specific relationship form is found to have an age-independent influence in Germany (Helfferich et al., 2001, 223). According to the American NSFG, the influence of the relationship is independent of life-course stage (Mosher et al., 2012, 8). Together with age, the birth order or number of children already born has a considerable effect. Mosher et al. (2012, 8) report that the proportion of unwanted pregnancies increases with the birth order. Studies on Germany (Feldhaus & Boehnke, 2008; Helfferich et al., 2001, 246) show a higher proportion of unwanted or mistimed pregnancies for the first child and more unwanted pregnancies for the third child. Several studies consider low income and a low level of formal education to be risk factors (Font-Ribera et al., 2007, 128; Mosher et al., 2012, 8f und 14; Rossier, Michelot, Cocon Group, & Bajos, 2006, 21). However, in Germany there is no significant difference in intendedness attributable to education level (Helfferich et al., 2001, 245). The duration of further education and training did however play a role, in so far as the pregnancies of young mothers still in education or training were more frequently unwanted or unintended (Helfferich, Hendel-Kramer, & Wehner, 2007). At the macrostructural level in Germany, specific characteristics of the federal states are associated with different prevalence of unintendedness. The city states (Berlin, Hamburg, Bremen), for example, have highest rate of induced abortions followed by the states in the East (with a historical tradition as a socialist country) (Statistisches Bundesamt, 2011, 37f). The rates of induced abortion correspond to higher levels of indicators for economic problems such as poverty or the unemployment rate. All of these factors interact and the effect of age and relationship status can disappear after controlling for other related variables (e.g. relationship duration; Dreesen & Matthijs, 2010; Rossier et al., 2006). In contrast, other factors, such as parity or participation in education or training, depend on age and life-course stage. A low educational level and young age at first birth (both risk factors for an unwanted pregnancy) are also interrelated. Several studies adopt the life-course perspective by looking for age-dependent and life-stage dependent conditions that contribute to an unintended pregnancy. Font-Ribera et al. (2007), for example, compare unintended pregnancies in the age groups 15–24, 25–39 and 40–49, as do Sihvo et al. (2003) with slightly different age boundaries. Rossier et al. (2006) compute the risk of unintended pregnancy as the product of the frequency of sexual activity, the use and the pearl index of contraception and the probability of conceiving. They show that sexual activity, access to contraception, desire for children, and the resources for raising a child are affected by separate age-dependent social influences and that their characteristics and their impact vary across age groups. Consequently, there are different explanations for unintended pregnancy among young (childless) women, for
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 4
C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
women in the middle age group who are in a stable longterm relationship and for women over 34 who already have children (Rossier et al., 2006; Sihvo et al., 2003). The situation in Germany is characterized by a high average age at first birth (2010: 29.2 years in West Germany and 27.4 in East Germany; Statistisches Bundesamt, 2012). It is mainly highly-qualified West German women who are worried about getting pregnant too early (for differences between East and West Germany see Goldstein, Kreyenfeld, Huinink, Konietzka, & Trappe, 2010). In Germany the is a widespread belief that the father or the mother should have completed their education or training and have a secure job before they have their first child (Institut fu¨r Demoskopie Allensbach, 2010). The prevalence rates for motherhood during vocational training (3%) or university studies (6%) are correspondingly low (BMFSFJ, 2010) and soon after having completed training or education the transition rate into parenthood increases sharply (‘‘institutional effect’’: Billari & Philipov, 2006). The second research question reads: Which factors affect the prevalence of unwanted pregnancies in younger, middle and higher aged women in Germany and how do these factors interact in each age group? Qualitative data can help understand causal mechanisms behind correlations uncovered by standardized surveys. In research on reproductive health, qualitative studies on the subjective orientations of target groups are considered important in order to develop targeted prevention programs. Such programs ‘‘must build upon cultural understanding of the problem to be prevented. Research should focus on the meaning of pregnancy intentions to women and the processes women and their partners use in making fertility decisions. (. . .). Both qualitative and quantitative research, have contributed to our understanding of fertility decision-making; both will be essential to the creation of more effective prevention programs’’ (Santelli et al., 2003: 99). Qualitative studies can describe the ‘‘perceived cultural appropriateness of behavior according to age, demographic position and personal biography’’ (Sihvo et al., 2003: 601). They are also being called for in demography (e.g. Von der Lippe & Fuhrer, 2004). Trussell et al. (1999) conclude that not all women believe equally that a child is something to be planned and proposed qualitative studies to explore subjective pregnancy intentions. Several qualitative studies highlight the diversity and ambiguity of subjective meanings of pregnancy intentions. Bernardi and Mynarska (2010) construct a typology of future fertility intentions: in addition to ‘‘surely yes’’ and ‘‘surely not’’ they describe types with different planning horizons (‘‘as soon as’’, ‘‘surely one day’’) as well as openness and instability (‘‘maybe’’, ‘‘at times’’; cf. Sobotka, 2011, 80ff). In addition to ‘‘planned pregnancy’’, Earle (2004) distinguishes between ‘‘laissez faire’’, ‘‘recalcitrant (ambiguous)’’ and ‘‘accidental’’ pregnancies in retrospective interviews with primagravidae. Helfferich and Kandt (1996, 60f) analyze the statements made in qualitative interviews by interviewees who had described a pregnancy as unwanted in a prior standardized survey. Further qualitative studies employ in depth interviews to investigated family planning embedded in biography
(e.g. Rijken & Knijn, 2009). Helfferich et al. (2001, 2005) study specific biographical contexts of family planning and unwanted pregnancies from the women’s and the men’s perspective. The potential of qualitative research, in particular qualitative biographical research in combination with standardized life-course research (Kelle & Erzberger, 1999), has by no means been exhausted. The third research question includes the subjective perspective and complements the second research question: What are the typical subjective interpretations of the situation of unwanted pregnancy in the narrative passages of interviews and how do these interpretations vary for pregnancies in young, middle and older age? Furthermore, we ask whether certain subjectively perceived contexts of unwanted pregnancy are associated with specific concepts of intendedness and, as part of this, with a subjective theory that explains how the unintended pregnancy came about. 3. Data and methods The three research questions were investigated with both standardized data (research question 1 and 2) and qualitative data (research question 3). Different analytical methods that consider the role of specific independent variables in subsamples of unintended pregnancies are employed in answering the research questions 1 and 2. 3.1. Data source The data are drawn from the study ‘‘Family planning in the life course of women’’ (‘‘frauen leben 3’’, 2012–2014), carried out by the Institute for Social Science Research on Women (SoFFI F.), Freiburg (Germany) on behalf of the Federal Centre for Health Education, BZgA, Cologne, and in cooperation with the Institute of Sociology of the University of Freiburg (fieldwork February to July 2012). For this study, retrospective standardized and qualitative data on reproductive behavior and reproductive events in the lives of 20 to 44 year old women were collected. 3.1.1. Standardized telephone interviews An age-stratified random sample of 4002 women was drawn from the telephone directory of four federal states in Germany (Berlin, Baden-Wu¨rttemberg, Saxony and Lower Saxony), which differ according to patterns of fertility (for example, abortion rate, age at birth of the first child) and social indicators (for example, poverty and unemployment rates). In total 4794 pregnancies (4427 live births, 367 induced abortions) were reported by women who at least once in their life experienced a pregnancy (n = 2584). The standardized questionnaire included questions on intended and non-intended pregnancies ending in a live birth as well as in abortion.2 3.1.2. Qualitative semi-structured interviews At the end of the telephone interview, women who said that they had experienced at least one unintended
2 No cases of stillbirth or miscarriage were considered. The term ‘‘abortion’’ applies to induced abortion only.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
pregnancy were invited to recount their biographies at a later date. A sample of ninety-seven women was recruited, who spoke of their lives from childhood until the date of the interview in semi-structured, in-depth interviews. The sample was constructed on the basis of socio-demographic data derived from standardized surveys in accordance with quality criteria to maximize its diversity and contrast in terms of age, education and number of children. The guidelines for the interview encouraged the interviewee to give a free narrative about the life-course stages of childhood, youth and adulthood with specific questions on sexual behavior, relationships and reproductive events. The face-to-face interviews lasted 45 min and the complete recordings were transcribed. The current analysis focuses on the reports of 103 unwanted pregnancies (including 45 abortions). In addition, there are also narratives on 34 mistimed pregnancies, 14 undecided and 63 wanted pregnancies. The standardized and qualitative data can be matched at the respondent level. 3.2. Data analysis 3.2.1. Bivariate analysis of standardized data In the analyses of Sections 4.1 and 4.2, the unit investigated is pregnancies. The dependent variable is pregnancy intention at the time of conception, determined by the answer to the question ‘‘Did you want to get pregnant at that time?’’ The response options were ‘‘wanted then or sooner’’,3 ‘‘wanted, but it should have been later’’ (=mistimed), ‘‘unwanted’’ and ‘‘ambivalent, undecided’’. According to a commonly used procedure (Font-Ribera et al., 2007; Santelli et al., 2009, 88; Sihvo et al., 2003), all terminated pregnancies were considered to be unwanted. The life-course perspective is incorporated into the analyses of the three research questions by means of the differentiation according to mother’s age at birth of her child or termination of the pregnancy (‘‘under 25’’, ‘‘25–34’’, ‘‘35– 44’’). The boundary for the youngest age group is drawn at 25 years because it is rare to have a stable job (considered one of the major prerequisites for having children in Germany) before reaching this age. The boundary for the oldest age group was drawn at 35 years because this age is a criterion for ‘‘late’’ motherhood (Engstler & Lu¨scher, 1991, 438). There were not enough pregnancies in women over 34 in the data set studied here to allow a further differentiation of this age group. Bivariate analyses using the chi-squared tests were computed to discover the differences between mistimed and unwanted pregnancies according to age and with respect to contraceptive behavior, the emotional reaction to pregnancy and general attitudes to children. The emotional reaction to conception was recorded with a six-point happiness scale, with 1 indicating ‘‘very happy’’ and 6 indicating ‘‘very unhappy’’. Furthermore, we asked whether contraceptives were in use at the time of conception (for all pregnancies). General attitudes toward children were
3 In the following the terms ‘‘wanted then or sooner’’, ‘‘wanted (then)’’ and ‘‘at the right time’’ are treated as equivalent.
5
operationalized with the variable ‘‘I always wanted to have children’’ (agree fully, disagree fully, agree partially). It was assumed that there are negative feelings associated with having an abortion and therefore it did not seem appropriate to ask interviewees who reported an abortion whether they had been happy with their pregnancy. Consequently, no data are available on the emotional response to pregnancies that were terminated. At this stage of the analysis only pregnancies ending in a live birth were included (n = 4402). Chi-squared tests with p-values below 0.05 indicate significant differences. 3.2.2. Multivariate analysis of standardized data To investigate the research question which factors affect pregnancy intentions in young, middle or older age, only pregnancies (births and induced abortions) described as ‘‘wanted then or sooner’’ or as ‘‘unwanted’’ are compared. The exclusion of mistimed pregnancies is justified by the results of investigating the first research question which show that in several dimensions the mistimed pregnancies resemble the wanted pregnancies more closely than the unwanted ones. Undecided pregnancies are excluded because of the small number of cases. For multivariate analyses, logistic regression was used to estimate the effects of selected independent variables on unwanted pregnancies. These multivariate models were estimated separately for the three predefined groups of age at conception. Results are presented as odds ratios that indicate the relative likelihood of experiencing an unwanted pregnancy. The independent variables selected on the basis of theoretical considerations (see Section 2) were collected or operationalized as follows: For all pregnancies, including those that were terminated, information on the respondent’s personal situation at the time of conception was recorded by means of a list of questions relating to a difficult relationship,4 financial or job insecurity and whether they were in education or training at the time. The response options were operationalized dichotomously (yes/no). For the age group over 34 at conception the variable ‘‘currently in education or training’’ could not be included in the analyses because of the small number of cases. Education level was categorized in four groups: low, medium, high and higher.5 It should be noted that this variable records the level of education at the time of the
4 The survey posed separate questions on whether, at the time of conception, the respondent was in a relationship or not and if so whether the relationship was instable or troubled, for example, because the partner did not want a child at that time. Since the overwhelming majority of pregnancies (85%) were conceived within a stable relationship, the separate categories for problematic relationships were merged in the further analyses. 5 Education level was operationalized as a combination of schooling, vocational training qualifications and tertiary education: low level = Hauptschule (9 years) and vocational training or Realschule/POS (10 years) without vocational training; medium level = Hauptschule and college of higher education/master school/technical school or Realschule/ POS and vocational training; high level = Realschule/(Fach-)abitur (12/13 years) and college of higher education/master school/technical school or (Fach-)Abitur with or without vocational training; higher level: (Fach)abitur and university degree (university or university of applied sciences) or currently a university (of applied sciences) student.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
6
4. Results
survey, not at the time of the pregnancy. In determining regional effects, the two West German federal states Baden-Wu¨rttemberg and Lower Saxony were merged into one variable (‘‘west’’) in the regression model since there were hardly any differences in the distributions of the relevant factors and the analysis was primarily concerned with East-West and urban-rural differences. Since third and further births were relatively rare, the number of children born prior to the pregnancy of interest was operationalized in three groups: no children, one child and two or more children. Marital status at the time of conception distinguishes between never married and previously married women. The seven variables indicating possible influences on the pregnancy intention were included in three blocks into the logistic regression. In the first step, the effect of relationship quality, financial and job security, number of children already born, education and region in the three age groups were estimated. In the second and third steps, the additional effects of being in education or training and marital status were included, and improvements in the model’s goodness of fit were tested. After excluding cases with missing data on the independent variables, information on 3872 pregnancies from 1933 women could be included in the logistic regression. Of these pregnancies, 80% were wanted then or sooner and 20% were unwanted. Independent variables are considered significant if their effect on unintended pregnancy is statistically significant at the 95% level.
Of all the pregnancies (including induced abortions) reported by the 20–44 year old women, 66.3% were wanted then or sooner, 13.3% were mistimed and 17.7% were unwanted. Contraceptives were in use before 12% of all pregnancies (and 35% of unintended pregnancies). 4.1. Mistimed, unwanted and intended pregnancies that led to babies Mistimed and unwanted pregnancies differ significantly in several analytic dimensions. As to mistimed pregnancies the use of contraceptives is significantly less likely than as to unwanted pregnancies (28.4–47.2%). The frequency of a positive reaction to the news of a pregnancy was significantly higher when the pregnancy was mistimed: the reaction to 55.9% of the mistimed pregnancies was (very) positive (scores of 1 or 2 on a six-point scale) and (very) negative (scores of 5 or 6) to 8.4%. In comparison, the proportion of positive reactions is lower among unwanted pregnancies (22.4%) and the proportion of negative reactions higher (35.4%). The pattern ‘‘no contraception and positive response’’ dominates mistimed pregnancies with 44.9%, but remains marginal at 15.3% for unwanted pregnancies. This shows that mistimed pregnancies, compared to unwanted pregnancies, are more similar to intended pregnancies, for which generally can be assumed that contraception was omitted and the reaction was positive. A (very) negative response is more likely if the pregnancy was unwanted and contraception was used (42.9%), a (very) positive response is more likely if the pregnancy was mistimed and no contraception was used (62.9%). Only 10.5% of all unintended (mistimed or unwanted or undecided) pregnancies that were carried to term were consistent in the sense that contraception was used and the reaction to conception was negative (score of 5 or 6 on the six-point scale). Of all pregnancies that ended with a live birth, the proportions of mistimed and unwanted pregnancies depend on the mother’s age. Mistimed pregnancies become rarer with increasing age. However, the share of unwanted pregnancy is smallest in the middle age group (25–34 years, see Table 1). The results underline the fear of getting pregnant ‘‘too early’’. But mistimed pregnancies still account for 12% of all pregnancies in the age group 25–34 years. Obviously ‘‘too
3.2.3. Analysis of qualitative data The qualitative analysis covered 58 unwanted pregnancy carried to term and 45 abortions. The dimensions along which the narratives differed were determined with a contrasting comparison of the minimal and maximal differences in accordance with the generative approach of Grounded Theory. The women’s accounts were then grouped into four categories according to the main focus of concern which was expressed while talking about the pregnancy, e.g. conflict with partner or going back to work. Within the four categories narrations from women of different age were compared. In a further step, subjective concepts of whether and how a child can be ‘‘planned’’ as well as the use of the term ‘‘unwanted’’ with respect to a pregnancy were extracted and grouped. The relationship between specific ideas about planning and the conditions within which an unwanted pregnancy was conceived was discussed.
Table 1 Intendedness of pregnancies ending in a live birth by mother’s age at the birth (in %). Age at birth*
Under 25 years
25–34 years
35 years and older
Total
Intendedness
n = 1171
n = 2824
n = 394
n = 4389
Wanted then or sooner Mistimed Undecided Unwanted Total
54.7 23.4 3.5 18.4 100.1
77.9 12.0 2.6 7.5 100
78.9 5.1 3.8 12.2 100
71.8 14.5 2.9 10.8 100
Source: BZgA, data set ‘‘frauen leben 3’’, 2012, 20–44 year old women in four federal states. * p < 0.001.
Please cite this article in press as: Helfferich, C., et al. Unintended pregnancy in the life-course perspective. Advances in Life Course Research (2014), http://dx.doi.org/10.1016/j.alcr.2014.04.002
G Model
ALCR-119; No. of Pages 13 C. Helfferich et al. / Advances in Life Course Research xxx (2014) xxx–xxx
7
15.6 3.1
"I always wanted to have children": agree
24.2 57.1
unwanted undecided mismed
47.5 7.9
"I always wanted to have children": do not agree/agree parally
wanted then or sooner
14.9 29.7 0.0
20.0
40.0
60.0
Fig. 1. Pregnancy intention and item ‘‘I always wanted to have children’’ (in %, age at birth younger than 25 years).
early for a child’’ does not only refer to biological age but also to other timelines (like the time since a previous birth) or it extends well beyond the age of 25 years. Mistimed and unwanted pregnancies differ along the mother’s long-term intention to have children. This intention was recorded with the item ‘‘I always wanted to have children’’. Fig. 1 shows for an age at birth under 25 years: If women agree with this item, a pregnancy was significantly more frequently declared to be mistimed and less often to be unwanted. This effect is pronounced in the youngest age groups (under 25) and disappears in the oldest age group (35 years and older). In the age group 25– 34 years (age at birth) the proportion of mistimed pregnancies is not influenced by the long-term intention to have children, but those who did not or partially agree with the item had a significantly higher proportion of unwanted pregnancies (17% vs. 6.3%). 4.2. Factors affecting conception of an unwanted pregnancy in young, middle and older age The comparison of mistimed and unwanted pregnancies has shown that along certain dimensions mistimed pregnancies are more similar to intended pregnancies. Therefore, the following multivariate analysis concentrates on the factors affecting conception of an unwanted pregnancy in comparison with a pregnancy that was wanted and the time was right or it should occur sooner (=intended). Table 2 shows the results of the logistic regressions used to estimate the likelihood (odds ratio) that a pregnancy was unwanted vs. wanted in dependency on selected factors. The logistic regressions were estimated separately for pregnancies in young women (
