European Journal of Obstetrics & Gynecology and Reproductive Biology 166 (2013) 190–195

Contents lists available at SciVerse ScienceDirect

European Journal of Obstetrics & Gynecology and

Reproductive Biology
journal homepage: www.elsevier.com/locate/ejogrb

Association between childhood psychosocial factors and induced abortion

Venla Lehti a,*, Andre Sourander b, Päivi Polo-Kantola c, Lauri Sillanmäki a, Tuula Tamminen d,
Kirsti Kumpulainen e
a
Department of Child Psychiatry, University of Turku, Turku, Finland
b
Department of Child Psychiatry, University of Turku and Turku University Hospital, Turku, Finland
c
Department of Obstetrics and Gynecology, Turku University Hospital, Turku, Finland
d
Department of Child Psychiatry, University of Tampere, Tampere, Finland
e
Department of Child Psychiatry, University of Eastern Finland, Kuopio, Finland

A R T I C L E I N F O A B S T R A C T

Article history: Objective: To examine the predictive associations between psychosocial risk factors in childhood and
Received 6 June 2012 having an abortion in adolescence or young adulthood.
Received in revised form 7 September 2012 Study design: This study is based on a nationwide cohort consisting of 2867 girls born in Finland in 1981.
Accepted 8 October 2012
The baseline assessment was conducted at age eight by three informants, and it included information on
psychiatric symptoms, school performance and family related risk factors. Register-based follow-up data
Keywords: on abortions were collected until the end of the year when the participants turned 28 years. They were
Abortion
available for 2694 participants. Cox proportional hazards model and logistic regression model were used
Childhood and adolescence
Psychosocial factors
for statistical analysis.
Family factors Results: Altogether 357 women (13.3%) had had an abortion for other than medical reasons during the
follow-up. Of the childhood factors, a high level of conduct problems, poor school performance, family
structure other than two biological parents, and mother with a low level of education were
independently associated with having an abortion. Comparison of the strength of associations between
childhood risk factors and first abortion under the age of 20 versus first abortion at a later age, showed no
significant differences. Neither did the comparison between one and more abortions.
Conclusions: At age eight there are already psychosocial factors which predict later abortion. This finding
needs to be considered when targeting preventive interventions and developing sexual health services.
ß 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Cross-sectional studies have shown that women with low

socioeconomic status [3] or with a history of intimate partner
There is wide variation in abortion rates across and within violence [4] are more likely to have an abortion. A longitudinal
different countries. While the average abortion rate in developed study showed that women with a single-parent family background
countries has been estimated to be 26 abortions in a year per 1000 also have an increased risk [5]. The only study which allows
women aged 15–44 [1], the rate is 8.6 abortions per 1000 women evaluation of predictive associations between psychiatric symp-
aged 15–49 in Finland [2]. Among Finnish women under the age of toms and having an abortion is a Norwegian cohort study, which
20 the rate is 12.0 abortions per 1000 women aged 15–19 [2]. In showed an association between conduct problems at age 15 and
Finland, an evaluation of the indication and permission to later abortion [6]. The study, however, was based on a relatively
terminate the pregnancy is needed from two different physicians. small sample, no independent effects of symptoms other than
The gestational age limit is generally 12 weeks, but in exceptional conduct problems were studied, and the risk factors were assessed
social conditions it can be extended until the gestational week only by the youths themselves [6]. Other studies have shown that
19 + 6 and for a severe fetal defect it can be extended until the unconventionality in adolescence and the use of psychotropic
gestational week 23 + 6. Most abortions are for social reasons, and drugs prior to pregnancy increase the risk of abortion [7,8]. In most
less than 4% of all abortions are due to a fetal defect or a medical studies information on abortions has been based on self-reports,
problem of the pregnant woman [2]. which is an unreliable way of data collection [9].
The aim of this study was to examine associations of family
related factors, psychiatric symptoms and school performance at
* Corresponding author at: Department of Child Psychiatry, University of Turku,
age eight with having an abortion under the age of 29. Our
Itäinen Pitkäkatu 1/Varia, 20014 Turku, Finland. Tel.: +358 50 306 1767. hypothesis was that family background and childhood behavioral
E-mail address: venla.lehti@utu.fi (V. Lehti). profile predict an abortion. Further, we hypothesized that

0301-2115/$ – see front matter ß 2012 Elsevier Ireland Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ejogrb.2012.10.017
 V. Lehti et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 166 (2013) 190–195 191

childhood factors are more strongly associated both with having an zero and the standard deviation as one. For the pooled variables,
abortion already in adolescence and with having more than one the mean score was taken of the standardized parent subscale and
abortion. the corresponding teacher subscale to give equal weight to the
informants. All the variables concerning psychiatric symptoms
2. Materials and methods were used as three-categorical: low level indicated no symptoms
or symptoms below the cutoff score based on the 50th percentile,
The study is a part of the ‘‘Finnish 1981 Birth Cohort Study’’. The intermediate level indicated symptoms between the level of the
baseline assessment was conducted in 1989 when the participants 50th and 90th percentile, and high level indicated symptoms above
were eight years old. Register data for the follow-up were collected the level of the 90th percentile.
until the end of the year 2009, during which the participants
turned 28. Informed consent was obtained from the children’s 2.4. School performance
parents at baseline. The method of combining information from
questionnaires and register data did not allow the identification of Teachers were asked about the child’s school performance by
the subjects. The Ethics Committee of the Intermunicipal Hospital giving them three options: the student is 1 = above average,
District of Southwest Finland approved the research plan. 2 = average or 3 = below average. In this study, the alternatives 1
and 2 were combined and those at the ‘‘below average’’ level were
2.1. Participants compared with those at the level of ‘‘average or above’’.

The original study sample was drawn from the total population 2.5. Family background
of Finnish children born in 1981 (n = 60,007). It consisted of 6017
children, which was 10% of the basic population. Of the selected The families were categorized into either those with two
children, 5813 (96.6%) took part in the study in 1989. The number biological parents or others. The mother’s educational level was
of female participants was 2867. Almqvist et al. [10] have measured by her completion of upper secondary school. Not
previously presented the design and subjects of this study. completing it was defined as a low educational level. The mother’s
Follow-up data were available for 2694 girls (94.0% of the age at the participating child’s birth was calculated through her
participants). The 6.0% loss was due to missing personal own birth year. Mothers born in 1961 or later were referred to as
identification numbers. young mothers because they had been 20 years old at most when
they had given birth in 1981. Those born in 1960 or earlier were
2.2. Data collection referred to as old mothers.

The baseline assessment was conducted with the help of 2.6. The outcome
schoolteachers. The researchers visited participating schools and
met the teachers to instruct them on data collection. The teacher At follow-up, the women in the cohort who had obtained an
who knew the child best completed a teacher’s questionnaire, abortion were identified by their personal identification number
which included questions on children’s psychiatric symptoms and from the Register on Induced Abortions and Sterilizations, which
school performance. The children completed questionnaires is maintained by the National Institute for Health and Welfare.
concerning depressive symptoms in the classroom. The teachers All public and private hospitals report abortions to this register.
sent parent questionnaires via the child to the parents and the It includes women’s personal identification number, and
parents returned them in a sealed envelope to the teachers. Parents information on sociodemographic characteristics, previous
gave information on family factors and children’s psychiatric pregnancies, contraception, indication for abortion, and details
symptoms. Only one parent questionnaire was collected from each on the procedure. The coverage and validity of the register have
child’s home. been shown to be very good [16]. Spontaneous abortions are not
included in the register. Furthermore, in this study, abortions
2.3. Explanatory variables measured at age eight which were due to a fetal birth defect or malformation or
medical problem of the pregnant woman were excluded,
2.3.1. Psychiatric symptoms because they are more likely to follow an intended pregnancy.
The parents completed the Rutter scale, parent version, which This study focused on abortions following an unwanted
has 31 items [11]. The teachers completed the Rutter scale, teacher pregnancy.
version, with 26 items [12]. The answers are rated on a scale of 0–2.
Both include three subscales: conduct problems, hyperactive 2.7. Statistical analysis
problems, and emotional problems. Parent and teacher reports
were pooled together. The children completed the Children’s Associations between childhood psychosocial risk factors and
Depression Inventory (CDI), a 27-item self-report [13]. The range of time to first abortion were analyzed using Cox proportional
scores for each item is 0–2. The question concerning suicide was hazards models. The first abortion before the end of the year 2009
excluded for ethical reasons, so the version of the CDI used in this was considered as the endpoint event. Those who did not have
study consisted of 26 questions. Previous studies have shown that abortions during the follow-up, or who died or moved abroad
the reliability of the Rutter parent scale [14] and teacher scale [12] during it without a preceding abortion, had censoring time to the
is good. The reliability of the CDI has varied in different studies end of follow-up, death, or moving date. The strength of the
[15]. In this study, the internal consistency of the scales evaluated associations was quantified using hazard ratios (HR) with 95%
by Cronbach’s alpha ranged from 0.46 (parent’s report of emotional confidence intervals (CI). Single predictor analyses were conducted
problems) to 0.83 (teacher’s report of hyperactivity and child’s first. The variables which were significant at level p < 0.1 were
report of depressive symptoms). selected for the two multipredictor models. Model 1 included
A standardization procedure was conducted to facilitate the variables related to family background. Model 2 included variables
comparison of different psychopathology subscales with different related to psychiatric symptoms and school performance. The
numbers of items. The three subscales of the parent and teacher variables which were significant at level p < 0.1 in model 1 or 2
reports and the CDI were standardized using the mean value as were selected for the final multipredictor model.
192 V. Lehti et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 166 (2013) 190–195

Logistic regression analysis was used to compare the register-

based information on the characteristics of women under and over
the age of 20 having their first abortion. It was also used for
estimating the strength of the association between the childhood
psychosocial risk factors and having the first abortion under or
over the age of 20. First, a multinomial logistic regression analysis
was conducted with a three-category response: no abortions, first
abortion under the age of 20, and first abortion over the age of 20. A
further analysis was then conducted with a binary response: first
abortion under the age of 20, and first abortion over the age of 20.
Similarly, a multinomial logistic regression analysis was conducted
with a three-category response: no abortions, one abortion, and
more than one abortion, and a further analysis was conducted with
a binary response: one abortion and more than one abortion. The
strength of associations was quantified using odds ratios (OR) with
95% confidence intervals. Statistical analyses were conducted
using SAS for Windows version 9.2 (SAS Institute Inc., Cary, NC,
USA, 2002-2008).

3. Results

Altogether 357 women (13.3% of the sample) had had an

abortion for other than medical reasons. Most of them, 279 women
(78.2%), had obtained one abortion, and 64 women (17.9%) had
obtained two. Only 14 (3.9%) had had more than two abortions. The
distribution of abortions in different age groups is illustrated in
Fig. 1. The age range of those who had had an abortion was 14.4–
28.9 years at first abortion. Median age at first abortion was 20.6
Fig. 2. Estimated survival curves for time to the first abortion. Data represent results
years (lower quartile 18.6 years and upper quartile 23.7 years).
from the single predictor Cox regression analysis. Low level of parent- and teacher-
Those who had had their first abortion under the age of 20 lived reported conduct problems indicates no symptoms or symptoms below the cutoff
less often in a relationship, had had fewer previous pregnancies, score based on the 50th percentile, intermediate level indicates symptoms between
and had terminated their pregnancy more often after the ninth the level of the 50th and 90th percentile, and high level above the level of the 90th
gestational week, compared with women who had had their first percentile.

abortion at an older age (Table 1).

The results of the Cox regression analysis are shown in Table 2.
high level of conduct problems were independently associated
In the single predictor analysis all explanatory variables apart from
with having an abortion during the follow-up.
parent- and teacher-reported emotional problems were associated
To compare the associations between childhood predictors and
with having an abortion during the follow-up. The findings
having the first abortion under or over the age of 20, multinomial
concerning the association between conduct problems and
logistic regression analyses with all the previously mentioned
abortion are illustrated in Fig. 2. The variables which were
explanatory variables and a three-category response was con-
significant at level p < 0.1 in the multipredictor model 1 were
ducted. There was no clear trend of the childhood predictors being
family structure, and mother’s educational level, while those in the
more strongly associated with either age group. Furthermore,
multipredictor model 2 were school performance, and parent and
when the two groups were compared by conducting a logistic
teacher report of conduct problems. They remained significant
regression analysis with a binary response, no significant
(p < 0.05) in the final multipredictor model. In other words, having
differences were found. The frequencies are shown in Table 1.
a family structure other than two biological parents, a mother with
When women with only one and those with repeated abortions
a low educational level, school performance below average, and a
were compared using a multinomial logistic regression analysis, all
odds ratios were higher for the outcome more than one abortion. In
a binary logistic regression analysis, however, there were no
significant findings. The closest to significance was the parent- and
teacher-reported hyperactive problems, which was close to being
more strongly associated with having more than one abortion (OR
2.1 for intermediate level of problems and 1.6 for high level of
problems, p = 0.051).

4. Comment

The findings of this study support our hypothesis that there is

an association between certain psychosocial factors measured at
age eight and having an abortion in later life. Women who have had
a high level of conduct problems, poor school performance, family
structure other than two biological parents or a mother with a low
Fig. 1. Proportion of girls in the sample who have had an abortion in each age group. educational level are at increased risk of having an abortion.
Only the first abortion of each participant is included. Numbers of abortions are In line with our results, an association has also been found
included at the top of bars. between adolescent conduct problems and abortions in later life
 V. Lehti et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 166 (2013) 190–195 193

Table 1
Characteristics of women having their first abortion, n = 357.

Women < 20 years, n = 152 Women  20 years, n = 205 pa

% (n) % (n)

Register information on abortions

***
Lives in a relationship
Yes 6.6 (10) 25.6 (52)
No 93.4 (142) 74.4 (151)
***
Number of previous pregnancies
0 94.7 (144) 71.2 (146)
1 4.0 (6) 18.1 (37)
2 1.3 (2) 9.3 (19)
3 – 1.5 (3)
**
Gestational age at the time of abortion
<9 weeks 41.5 (63) 58.1 (119)
9–12 weeks 52.0 (79) 37.1 (76)
>12 weeks 6.6 (10) 4.9 (10)
Contraception prior to pregnancy
Yes 63.1 (82) 68.7 (123)
No 36.9 (48) 31.3 (56)

Baseline information on participants

Family structure
Two biological parents 72.7 (104) 77.8 (154)
Other than two biological parents 27.3 (39) 22.2 (44)
Age of the girl’s mother
Old (born 1927–1960) 88.6 (124) 93.3 (181)
Young (born 1961–1965) 11.4 (16) 6.7 (13)
Educational level of the girl’s mother
High 23.0 (32) 20.0 (39)
Low 77.0 (107) 80.0 (156)
School performance
Average or above 85.5 (130) 82.7 (167)
Below average 14.5 (22) 17.3 (35)
Parent- and teacher-reported level of conduct problems
Low 44.8 (64) 46.7 (93)
Intermediate 33.6 (48) 35.2 (70)
High 21.7 (31) 18.1 (36)
Parent- and teacher-reported level of hyperactive problems
Low 53.1 (75) 62.3 (124)
Intermediate 29.1 (41) 19.6 (39)
High 17.7 (25) 18.1 (36)
Parent- and teacher-reported level of emotional problems
Low 57.5 (81) 51.8 (102)
Intermediate 30.5 (43) 36.0 (71)
High 12.1 (17) 12.2 (24)
Self-reported level of depressive symptoms
Low 44.1 (67) 48.5 (98)
Intermediate 42.8 (65) 39.1 (79)
High 13.2 (20) 12.4 (25)
a
Logistic regression analysis.
**
p  .01.
***
p  .001.

[6]. Conduct problems are characterized by disruptive and norm- This study showed that family structure other than two
breaking behavior, aggressiveness and risk-taking, which may biological parents increases the likelihood of a daughter having
increase the likelihood of unintended pregnancies. It is possible an abortion. Previously it has been shown that living in a single-
that girls with conduct problems become involved in risky parent family is associated with having an abortion [5], while
behavior voluntarily and perhaps impulsively [17], but they also others have found that disrupted family structure is associated
have an increased risk of being sexually abused [18]. with early sexual debut and teenage pregnancy [20,21]. There are
Teacher’s evaluation of the child’s school performance as several possible explanations for these associations. For example,
below average already in primary school was an independent high-conflict marriages increase the risk of violence, child abuse,
predictor of an abortion. Poor school performance in adolescence and problems in the parent-child relationship, while single
is associated with various health-compromising behaviors, parenthood decreases parental monitoring [22]. All of these
including high-risk sexual activity [19]. Furthermore, in later outcomes are known to increase the likelihood of sexual risk-
life, women with a low level of education have a remarkably taking [20,23,24].
higher risk of having an unintended pregnancy compared with In this study, girls whose mother had a low level of education
other women, especially those with a university education [3]. were at increased risk of having an abortion. It has been suggested
The finding of this study, however, suggests that educational that low parental socioeconomic status increases the risk of
achievement already in primary school may predict health adverse health and social outcomes through family level factors
disparities in later life. such as stressful life situations, poor access to services, and
194 V. Lehti et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 166 (2013) 190–195

Table 2
Associations between psychosocial factors in childhood and abortion in later life.

n Those who had Single predictor p Multipredictor p Final multipredictor p

an abortion (%) models, HR models 1 and 2, model, HR (95% CI)
(95% CI) HR (95% CI)

Model 1
Family structure
Two biological parents 2161 11.9
*** *** ***
Other than two biological parents 408 20.3 1.8 (1.4–2.3) 1.7 (1.3–2.2) 1.5 (1.2–2.0)
Age of the girl’s mother
Old (born 1927–1960) 2402 12.7
**
Young (born 1961–1965) 141 20.6 1.8 (1.2–2.6) 1.4 (0.9–2.1)
Educational level of the girl’s mother
High 743 9.6
*** ** *
Low 1771 14.9 1.6 (1.2–2.1) 1.5 (1.2–2.0) 1.4 (1.1–1.8)

Model 2
School performance
Average or above 2343 12.7
*** ** **
Below average 250 22.8 1.9 (1.4–2.5) 1.6 (1.2–2.2) 1.6 (1.2–2.2)
*** ** ***
Parent- and teacher-reported level of conduct problems
Low 1431 11.0
*
Intermediate 823 14.3 1.3 (1.05–1.7) 1.2 (0.9–1.5) 1.3 (0.98–1.6)
*** *** ***
High 297 22.6 2.2 (1.7–3.0) 1.8 (1.3–2.5) 1.8 (1.4-2.5)
***
Parent- and teacher-reported level of hyperactive problems
Low 1741 11.4
*
Intermediate 519 15.4 1.4 (1.1–1.8) 1.2 (0.9–1.6)
*** *
High 268 22.8 2.2 (1.6–2.9) 1.4 (1.02–2.1)
Parent- and teacher-reported level of emotional problems
Low 1487 12.3
Intermediate 790 14.4 1.2 (0.9–1.5)
High 254 16.1 1.3 (0.9–1.9)
*
Self-reported level of depressive symptoms
Low 1372 12.0
*
Intermediate 927 15.5 1.3 (1.1–1.7) 1.2 (0.9–1.5)
High 278 16.2 1.4 (0.996–1.9) 1.1 (0.8–1.5)

Note: Only the first abortion of each participant was included. Multipredictor model 1 included family related factors. Multipredictor model 2 included school performance
and psychiatric symptoms.
The overall p value for multicategorical variables is reported on the same row as the name of the variable.
HR: hazard ratio; CI: confidence interval.
*
p  .05.
**
p  .01.
***
p  .001.

health-compromising life style as well as through community- 4.1. Recommendation

level factors [25]. For example, in a US study, neighborhood
characteristics such as low level of social organization and In conclusion, this study shows that the pathways which lead to
concentrated poverty were associated with adolescents’ early or having an abortion may have their origins early in life. Although
multiple sexual relationships even when controlled for various the psychosocial factors identified in this study may have
individual-level factors [26]. independent effects, it is also possible that there are shared risk
Our hypotheses of the stronger associations between childhood factors which increase the likelihood of both adverse psychosocial
psychosocial risk factors and the first abortion obtained under the conditions in childhood and having an abortion in later life. The
age of 20 or repeated abortions were only partially supported. findings of this study have implications in the prevention of
There was a statistically non-significant trend for more powerful abortions. For example, tailoring sexual health education and
associations between the risk factors and repeated abortions. No improving access to sexual health services for girls with identified
support was found for the stronger associations between risk risk factors might help in reducing the abortion rates. On the other
factors and the first abortion obtained under the age of 20. One hand, women who undergo abortion may need particular support,
explanation might be that adolescents with more childhood since all the risk factors for abortions identified in this study are
psychosocial problems are more willing to give birth because of also known to increase the risk of later mental health problems
their lower educational expectations [27]. Thus their abortion rate [22,25,28,29].
in adolescence may be lower than expected on the grounds of their
presumably higher level of risky sexual behavior and unintended Acknowledgments
pregnancies.
This study was limited by having only one measurement of Professor Irma Moilanen (Department of Child Psychiatry,
psychosocial risk factors and the lack of information on possible University of Oulu, Oulu, Finland) and Professor Jorma Piha
mediators. No diagnostic interview was used for measuring (Department of Child Psychiatry, University of Turku, Turku,
psychiatric symptoms and there was no information on other Finland) are thanked for their comments to the manuscript.
childhood risk factors such as adverse life events. The sample size Statistician Hans Helenius (Department of Biostatistics, University
did not allow the use of other than binary variables for family of Turku, Finland) is thanked for advice in statistical analyses.
related factors and school performance. Venla Lehti’s work was financially supported by the National
 V. Lehti et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 166 (2013) 190–195 195

Graduate School of Clinical Investigation and by the grant from the [16] Gissler M, Ulander VM, Hemminki E, Rasimus A. Declining induced abortion
rate in Finland: data quality of the Finnish abortion register. International
Finnish Brain Foundation. Journal of Epidemiology 1996;25:376–80.
[17] Donohew L, Zimmerman R, Cupp PS, Novak S, Colon S, Abell R. Sensation
References seeking, impulsive decision-making, and risky sex: implications for risk-
taking and design of interventions. Personality and Individual Differences
[1] Sedgh G, Henshaw S, Singh S, Ahman E, Shah IH. Induced abortion: estimated 2000;28:1079–91.
rates and trends worldwide. Lancet 2007;370:1338–45. [18] Fergusson DM, Woodward LJ. Educational, psychosocial, and sexual outcomes
[2] Heino A, Gissler M, Soimula A. Induced abortions 2010—preliminary data. of girls with conduct problems in early adolescence. Journal of Child Psychol-
Statistical report. Helsinki: National Institute for Health and Welfare; 2011. ogy and Psychiatry and Allied Disciplines 2000;41:779–92.
[3] Font-Ribera L, Pérez G, Salvador J, Borrell C. Socioeconomic inequalities in [19] Neumark-Sztainer D, Story M, French SA, Resnick MD. Psychosocial correlates
unintended pregnancy and abortion decision. Journal of Urban Health of health compromising behaviors among adolescents. Health Education
2007;85:125–35. Research 1997;12:37–52.
[4] Coker AL. Does physical intimate partner violence affect sexual health? A [20] Lenciauskiene I, Zaborskis A. The effects of family structure, parent–child
systematic review. Trauma Violence Abuse 2007;8:149–77. relationship and parental monitoring on early sexual behaviour among ado-
[5] Sauvola A, Mäkikyro T, Jokelainen J, Joukamaa M, Järvelin MR, Isohanni M. lescents in nine European countries. Scandinavian Journal of Public Health
Single-parent family background and physical illness in adulthood: a follow- 2008;36:607–18.
up study of the Northern Finland 1966 birth cohort. Scandinavian Journal of [21] Vikat A, Rimpelä A, Kosunen E, Rimpelä M. Sociodemographic differences in
Public Health 2000;28:95–101. the occurrence of teenage pregnancies in Finland in 1987–1998: a follow up
[6] Pedersen W, Mastekaasa A. Conduct disorder symptoms and subsequent study. Journal of Epidemiology and Community Health 2002;56:659–68.
pregnancy, child-birth and abortion: a population-based longitudinal study [22] Kelly JB. Children’s adjustment in conflicted marriage and divorce: a decade
of adolescents. Journal of Adolescence 2011;34:1025–33. review of research. Journal of the American Academy of Child and Adolescent
[7] Gissler M, Artama M, Ritvanen A, Wahlbeck K. Use of psychotropic drugs Psychiatry 2000;39:963–73.
before pregnancy and the risk for induced abortion: population-based regis- [23] Markham CM, Tortolero SR, Escobar-Chaves SL, Parcel GS, Harrist R, Addy RC.
ter-data from Finland 1996–2006. BMC Public Health 2010;10:383. Family connectedness and sexual risk-taking among urban youth attending
[8] Martino SC, Collins RL, Ellickson PL, Klein DJ. Exploring the link between alternative high schools. Perspectives on Sexual and Reproductive Health
substance abuse and abortion: the roles of unconventionality and unplanned 2003;35:174–9.
pregnancy. Perspectives on Sexual and Reproductive Health 2006;38:66–75. [24] Wilson HW, Widom CS. An examination of risky sexual behavior and HIV in
[9] Jagannathan R. Relying on surveys to understand abortion behavior: some victims of child abuse and neglect: a 30-year follow-up. Health Psychology
cautionary evidence. American Journal of Public Health 2001;91:1825–31. 2008;27:149–58.
[10] Almqvist F, Ikäheimo K, Kumpulainen K, et al. Design and subjects of a Finnish [25] Bradley RH, Corwyn RF. Socioeconomic status and child development. Annual
epidemiological study on psychiatric disorders in childhood. European Child Review of Psychology 2002;53:371–99.
and Adolescent Psychiatry 1999;8(Suppl 4):3–6. [26] Browning CR, Burrington LA, Leventhal T, Brooks-Gunn J. Neighborhood
[11] Rutter M, Tizard J, Whitmore K, editors. Education, health and behaviour. structural inequality, collective efficacy, and sexual risk behavior among urban
London: Longman; 1970. youth. Journal of Health and Social Behavior 2008;49:269–85.
[12] Rutter M. A children’s behaviour questionnaire for completion by teachers: [27] Sihvo S, Bajos N, Ducot B, Kaminski M. Women’s life cycle and abortion
preliminary findings. Journal of Child Psychology and Psychiatry and Allied decision in unintended pregnancies. Journal of Epidemiology and Community
Disciplines 1967;8:1–11. Health 2003;57:601–5.
[13] Kovacs M. Children’s depression inventory, CDI, manual. Toronto: Multi- [28] Fergusson DM, Horwood LJ, Ridder EM. Show me the child at seven: the
Health Systems, Inc.; 1992. consequences of conduct problems in childhood for psychosocial functioning
[14] Rutter M, Graham P. Psychiatric disorder in 10- and 11-year-old children. in adulthood. Journal of Child Psychology and Psychiatry and Allied Disciplines
Proceedings of the Royal Society of Medicine 1966;59:382–7. 2005;46:837–49.
[15] Myers K, Winters NC. Ten-year review of rating scales. II: Scales for internal- [29] Sourander A, Multimäki P, Santalahti P, et al. Mental health service use among
izing disorders. Journal of the American Academy of Child and Adolescent 18-year-old adolescent boys: a prospective 10-year follow-up study. Journal of
Psychiatry 2002;41:634–59. the American Academy of Child and Adolescent Psychiatry 2004;43:1250–8.