The Sources of Parent-Child Transmission of Risk for Suicide Attempt and Suicide Death in Swedish National Samples

Abstract

Objective –

To clarify the sources of parent-offspring transmission for suicide attempt (SA) and suicide death (SD).

Methods –

Three sources of parent-child resemblance (genes+rearing, genes only and rearing only) are examined in parents and offspring from 4 family types from Swedish national samples: intact, not-lived-with biological father, lived with stepfather, and adoptive. Parent-offspring resemblance was assessed primarily by tetrachoric correlation.

Results –

For SA→SA transmission, best estimate tetrachoric correlations (± 95% CIs) for genes+rearing, genes only and rearing only, were, respectively, +0.23 (0.23–0.24), +0.13 (0.11–0.15) and +0.14 (0.11–0.16). SA was more strongly transmitted to sons than daughters. Parental psychiatric disorders accounted for 40% of the genetic transmission but had no impact on rearing effects. For SD→SD transmission, best estimates of tetrachoric correlations for genes+rearing, genes only and rearing only cross-generational transmission were, respectively: + 0.16 (0.15–0.18), +0.07 (0.02–0.12) and −0.05 (−0.17–0.07). Although the SA-SD genetic correlation was high (+0.84), we could reject the hypothesis that they reflected behaviors only differing in severity on the same continuum of genetic liability.

Conclusions –

The transmission across generations of SA is moderately strong and arises equally from genetic and rearing effects. Parental psychiatric illness explains almost half of the genetic transmission of SA but none of the rearing effect. SD is modestly transmitted across generations, probably via genetic effects although rearing may play a role. While SA and SD share a substantial proportion of their hereditary risk, they do not, from a genetic perspective, simply reflect milder and more severe forms of the same diathesis.

Suicide is in a very marked and striking manner hereditary; and this is a strong ground for regarding it as constitutional …Dr. [Benjamin] Rush, M. Esquirol, and others have recorded instances of the hereditary transmission of this propensity. M. Falret has collected a variety of observations on the subject, and has concluded that, of all the kinds of insanity, the form distinguished by this tendency is probably that which the most frequently becomes hereditary.(1) Prichard 1837 p. 284

A long catalogue could be given of all sorts of inherited … predisposition[s] to various diseases… [one example being that] several members of the same family, during three or four successive generations, have committed suicide. (2) Darwin 1868 p. 16

Psychiatry has long been interested in the familial transmission of suicidal behavior. Despite a substantial number of family and twin studies of both suicide attempts (SA) and suicide death (SD) which have confirmed important familial/genetic contributions (3–5), we are aware of only two prior adoption studies specifically focused on SD (6, 7) and one on SA (8). The study of Schulsinger (6) examined only aggregated biological and adoptive relatives of adoptees with SD while those of Petersen et al (7) (8) examined biological and adoptive siblings of adoptees with, respectively, SD and SA. To our knowledge, no prior study has examined rates of suicidal behaviors in the offspring of biological and adoptive parents with SD and/or SA.

We conducted such a study that explored the sources of cross-generational transmission of SA and SD. We utilized an expanded adoption design involving four family types: intact nuclear families, families with a non-lived-with biological father (who sired the child but never lived with or near his offspring), families containing a stepfather, and adoptees and their biological and adoptive parents. From these four family types, we explored the resemblance for SA and SD in parents and offspring from parents who provided to their children both genes and rearing, genes only, or rearing only.

We began by asking three questions about SA:

• 1What are the relative roles of genetic versus rearing effects in the cross-generational transmission of SA?
• 2Does parent-offspring transmission of SA differ as a function of the sex of the parent or the child?
• 3What proportion of the cross-generational transmission of SA results from parental psychiatric or substance use disorders and do these proportions differ across the kind of parent-child relationship? We hypothesize that the contributions of these disorders to parent-offspring resemblance for SA will be stronger for genetic than for rearing cross-generational transmission.

Then we examine two questions about SD:

• 4What are the relative roles of genetic versus rearing effects in the cross-generational transmission of SD?
• 5What is the nature of the relationship between the transmitted genetic liability to SA and to SD? What is their genetic correlation and do SA and SD lie on a single dimension of genetic liability with SA representing a milder and SD a more severe form?

METHODS

We collected information on individuals from Swedish population-based registers with national coverage linking each person’s unique personal identification number which, to preserve confidentiality, was replaced with a serial number by Statistics Sweden. We utilized the following sources: Multi-Generation Register; Population and Housing Censuses; Swedish Hospital Discharge Register (national coverage 1987–2015 and partial coverage 1969–1986); Outpatient Care Register (national coverage 2001–2015) and Mortality Register (1969–2015). We secured ethical approval for this study from the Regional Ethical Review Board of Lund University (No. 2008/409, 2012/795, and 2016/679).

Suicide Attempt (SA) was identified in the Hospital Discharge, Outpatient (Specialist) Care Registry and Suicide Death (SD) in the Mortality Registry by the following ICD codes: ICD-8/9: E95, E98; ICD-10: X60-X89, Y10-Y34, Y87, Y90, Y91. In the analysis of SA, individuals with SD were excluded. We evaluated the hypothesis that SA and SC were disorders of differing severity on the same liability distribution in our largest sample of genetic informative parent-offspring correlations: not-lived-with-fathers and their offspring.

The database was created by entering all individuals in the Swedish population born in Sweden 1960–1990. The database also included the number of years, during ages 0–15, that individuals resided in the same household and geographical area as their biological mother, biological father and possible stepfather. From 1960 to 1985, we used household ID from the Population and Housing Census to define family types available every 5^th year and so some errors in family assignment might be possible. The household ID includes all individuals living in the same dwelling. From 1986 onwards, we defined family type using the Family ID from the Total Population Register available yearly. The Family ID is defined by related or married individuals registered at the same property. Furthermore, adults registered at the same property who have common children, but are not married, are registered in the same family. We created family types by investigating with whom the offspring shared the same household ID/Family ID when they were 0–15. In the detection of stepparents during the period 1986 and onwards for an offspring living with his/her mother, we only capture the stepfather if he is married to the mother of the offspring and/or has a common child together with her. For the years without this information, we approximated the household with the information from the closest year.

We thereby defined four kinds of families: i) intact families (offspring resided from 0–15 with their biological mother and father), ii) “not-lived-with” father families (offspring never resided in the same household or near their biological father), iii) stepfather families (offspring did not reside from ages 0–15 with their biological father and, from ages 0–15, resided ≥ 10 years with a non-biologically related male, 18–50 years older, iv) adoptive families (offspring adopted < age 5, with information available on both adoptive parents and ≥ 1 biological parent. Individuals adopted by biological relatives or an adoptive parent living with a biological parent were excluded). To maximize the number of adoptees, we included offspring born 1955–1990. The not-lived-with and stepfathers were defined so that their relationship with their offspring maximally resembled that seen between an adoptee and, respectively, his/her biological and adoptive parent.

We examined the tetrachoric correlation between all possible combinations of SA and SD in our parent-offspring pairs because this measure of association is easy to interpret in genetic-epidemiologic terms (i.e., equals the “correlation of liability”) and is insensitive to changes in base rates. We also present odds ratios for key results from logistic regression. We also investigated sex-specific transmission for the SA to SA transmission.

Finally, we used a linear probability model to examine the degree to which the transmission of SA from parents to children resulted from the transmission of psychiatric illness. As the base rates of SA were different in the different family types, we opted for the linear probability model, where the result are presented on the additive scale and more comparable between family types. In the first model, we included SA in parent while the second model added occurrence in parents of major depression, anxiety disorders, drug abuse, alcohol use disorder, bipolar illness and non-affective psychosis (for definitions see appendix).

To combine results from the different samples, we used the Olkin-Pratt (OP) meta-analytical approach. We calculate the combined correlations and the p-values for the heterogeneity tests that evaluate the null hypothesis that effects are similar across samples. We calculated the genetic correlation between SA and SD (see appendix for formula). Statistical analyses were performed using SAS 9.4 (9) and R 3.6.1 (10).

RESULTS

Sample and Results Presentation

Descriptive features of our four family samples are provided in table 1. Intact families are the most common followed by, respectively, not-lived-with father families, stepfather families and adoptive families. SAs are much more common than SDs in parents and offspring of all family types. Both forms of suicidal behaviors were rarer in the offspring and parents of intact families than in the other family types and less common in the step- and adoptive parents than in the biological parents and not-lived-with fathers.

Table 1 –

Sample Sizes and Rates of Suicide Attempts and Suicide Death in our Four Family Types

	Intact Families	Not-lived-with Father Families	Stepfather Families	Adoptive Families
Sample Size of Offspring	2,175,259	152,436	73,785	15,624
Year of Birth (offspring) (25–50–75 percentile)	1966–1974–1982	1969–1976–1983	1969–1975–1981	1959–1963–1967
Sample size of mothers	1,177,498	131,812	64,539	13,923
Suicide Attempt
Offspring Total	2.8%	6.3%	5.5%	6.7%
Offspring # of attempts (25–50–75 p-tile)*	1–1–2	1–1–2	1–1–2	1–1–2
Offspring Male/Female	2.7%/2.9%	5.9%/6.8%	5.2%/5.7%	6.1%/7.4%
Biological Mother	2.2%	7.0%	5.2%	8.5%
Biological Father	1.9%	7.9%		7.4%
Step/Adoptive-Mother				2.5%
Step/Adoptive-Father			3.4%	1.5%
Suicide Death
Offspring	0.3%	0.6%	0.5%	1.0%
Offspring Male/Female	0.4%/0.2%	0.8%/0.3%	0.8%	1.2%/0.8%
Biological Mother	0.2%	0.5%	0.3%	2.1%
Biological Father	0.5%	2.7%		2.4%
Step/Adoptive-Mother				0.5%
Step/Adoptive-Father			0.8%	0.9%

^*Among individuals with at least 1 attempt

To aid in following our subsequent results, we outline here their presentation. Table 2 provides separate results for all informative maternal-child and paternal-child correlations for both SA, SD and between SA and SD across our four family types. Table 3 compares results from the maternal-child and paternal-child relationships and obtains from them a weighted estimate (i.e., a parental-child correlation) which are depicted in figure 1. Table 4 presents correlations across the four kinds of parent-child relationships: mother-daughter, mother-son, father-daughter and father-son.

Table 2 –

Cross-Generational Transmission of Suicide Attempt and Suicide Death as Assessed by Tetrachoric Correlation (95% CIs) in Four Different Family Types

Relationship	Phenotypes: Parent to Offspring	Sources of Resemblance	Intact Families		Not-lived-with Father Families		Stepfather Families		Adoptive Families		Weighted Estimate Across Families		P-value for test of heterogeneity
			TC*	95% CIs	TC*	95% CIs	TC*	95% CIs	TC*	95% CIs	TC*	95% CIs
Mother – Offspring	Attempt to Attempt	Genes+ Rearing	0.25	0.24–0.25	0.20	0.19–0.22	0.19	0.16–0.22			0.22	0.21–0.23	<0.0001
	Death to Death		0.18	0.15–0.22	0.20	0.13–0.28	0.14	0.00–0.28			0.18	0.16–0.21	0.72
	Attempt to Death		0.12	0.11–0.14	0.12	0.10–0.14	0.14	0.08–0.20			0.12	0.11–0.13	0.82
	Death to Attempt		0.10	0.08–0.12	0.10	0.06–0.15	0.08	0.00–0.15			0.10	0.09–0.11	0.89
	Attempt to Attempt	Genes only							0.14	0.09–0.19	0.14	0.09–0.19	-
	Death to Death								−0.02	−0.19–0.15	−0.02	−0.19–0.15	-
	Attempt to Death								0.11	0.02–0.21	0.11	0.02–0.21	-
	Death to Attempt								0.12	0.05–0.20	0.12	0.05–0.20	-
	Attempt to Attempt	Rearing only							0.10	0.03–0.17	0.10	0.03–0.17	-
	Death to Death								0.04	−0.23–0.31	0.04	−0.23–0.31	-
	Attempt to Death								0.00	−0.15–0.16	0.00	−0.15–0.16	-
	Death to Attempt								0.07	−0.01–0.19	0.07	−0.01–0.19	-
Father-Offspring	Attempt to Attempt	Genes+ Rearing	0.24	0.24–0.25							0.24	0.24–0.25	-
	Death to Death		0.14	0.11–0.16							0.14	0.11–0.16	-
	Attempt to Death		0.08	0.06–0.10							0.08	0.06–0.10	-
	Death to Attempt		0.07	0.06–0.09							0.07	0.06–0.09	-
	Attempt to Attempt	Genes only			0.14	0.12–0.16			0.08	0.01–0.15	0.13	0.11–0.15	0.08
	Death to Death				0.08	0.03–0.13			0.07	−0.12–0.26	0.08	0.03–0.13	0.92
	Attempt to Death				0.07	0.03–0.11			0.03	−0.12–0.17	0.07	0.02–0.11	0.61
	Death to Attempt				0.08	0.06–0.11			0.09	−0.01–0.19	0.08	0.06–0.10	0.85
	Attempt to Attempt	Rearing only					0.14	0.11–0.17	0.15	0.06–0.24	0.14	0.11–0.17	0.83
	Death to Death						−0.12	−0.32–0.07	−0.04	−0.29–0.21	−0.08	−0.22–0.06	0.59
	Attempt to Death						0.06)	−0.02–0.13	−0.02	−0.22–0.18	0.05	−0.03–0.12	0.49
	Death to Attempt						0.00	−0.06–0.06	0.20	0.10–0.29	0.06	0.01–0.11	0.0004

^*TC- tetrachoric correlation

Table 3 –

Cross-Parent Weighted Estimates from Weighted Estimates for Mothers and Fathers and from Suicide Attempt to Suicide Death and Suicide Death to Suicide Attempt as Assessed by Tetrachoric Correlation (95% CIs) in Four Different Family Types

Disorders and Nature of Parent to Child Analyses		Mothers		Fathers		Weighted Estimate		Uncorrected P-value for test of heterogeneity
		TC*	95% CIs	TC*	95% CIs	TC*	95% CIs
Attempt to Attempt	Genes and Rearing	0.22	0.21–0.23	0.24	0.24–0.25	0.23	0.23–0.24	0.007
	Genes Only	0.14	0.09–0.19	0.13	0.11–0.15	0.13	0.11–0.15	0.71
	Rearing Only	0.10	0.03–0.17	0.14	0.11–0.17	0.14	0.11–0.16	0.32
Death to Death	Genes and Rearing	0.18	0.16–0.21	0.14	0.11–0.16	0.16	0.15–0.18	0.02
	Genes Only	−0.02	−0.19–0.15	0.08	0.03–0.13	0.07	0.02–0.12	0.30
	Rearing Only	0.04	−0.23–0.31	−0.08	−0.22–0.06	−0.05	−0.17–0.07	0.42
Attempt to Death	Genes and Rearing	0.12	0.11–0.13	0.08	0.06–0.10	0.11	0.10–0.12	0.004
	Genes Only	0.11	0.02–0.21	0.07	0.02–0.11	0.08	0.04–0.12	0.44
	Rearing Only	0.00	−0.15–0.16	0.05	−0.03–0.12	0.04	−0.04–0.11	0.59
Death to Attempt	Genes and Rearing	0.10	0.09–0.11	0.07	0.06–0.09	0.09	0.08–0.10	0.001
	Genes Only	0.12	0.05–0.20	0.08	0.06–0.10	0.08	0.07–0.10	0.22
	Rearing Only	0.07	−0.01–0.19	0.06	0.01–0.11	0.06	0.02–0.11	0.84
Attempt to Death and Death to Attempt	Genes and Rearing	0.11	0.10–0.12	0.09	0.08–0.10	0.10	0.09–0.11	0.024
	Genes Only	0.08	0.04–0.12	0.08	0.07–0.10	0.08	0.07–0.09	1
	Rearing Only	0.04	−0.04–0.11	0.06	0.02–0.11	0.06	0.02–0.09	0.65

^*Tetrachoric Correlation

Figure 1.

Best Weighted Estimates of Tetrachoric Correlations Reflecting Genes+Rearing, Genes Only and Rearing Only Parent-Child Relationships for Suicide Attempt and Suicide Death

Table 4 –

Cross-Generational Transmission of Suicide Attempt as a Function of the Sex of the Parent and the Sex of the Child as Assessed by Tetrachoric Correlation (95% CIs) in Four Different Family Types

	Phenotypes: Parent to Offspring	Sources of Resemblance	Intact Families		Not-lived-with Father Families		Stepfather Families		Adoptive Families		Weighted Estimate Across Families		P-value for test of heterogeneity Across Relationship Types
			TC*	95% CIs	TC*	95% CIs	TC*	95% CIs	TC*	95% CIs	TC*	95% CIs
	Mother to Daughter	Genes+ Rearing	0.23	0.22–0.24	0.21	0.19–0.23	0.16	0.12–0.20			0.22	0.21–0.23	<0.0001
	Mother to Son		0.26	0.25–0.27	0.20	0.18–0.22	0.23	0.19–0.26			0.24	0.23–0.25
	Father to Daughter		0.21	0.20–0.22							0.21	0.20–0.22
	Father to Son		0.28	0.27–0.29							0.28	0.27–0.29
	Mother to Daughter	Genes only							0.12	0.04–0.19	0.12	0.04–0.19	0.007
	Mother to Son								0.16	0.09–0.23	0.16	0.09–0.23
	Father to Daughter				0.14	0.11–0.16			0.10	0.01–0.20	0.14	0.11–0.16
	Father to Son				0.20	0.18–0.22			0.05	−0.05–0.15	0.19	0.17–0.21
	Mother to Daughter	Rearing only							0.04	−0.08–0.15	0.04	−0.08–0.15	0.083
	Mother to Son								0.15	0.05–0.25	0.15	0.05–0.25
	Father to Daughter						0.12	0.08–0.17	0.13	0.01–0.25	0.12	0.09–0.15
	Father to Son						0.16	0.12–0.21	0.16	0.04–0.29	0.16	0.13–0.20
	To Sons	Genes + Rearing									0.26	0.25–0.27	<0.0001
	To Daughters										0.22	0.21–0.22
	From Mothers										0.23	0.22–0.34	0.034
	From Fathers										0.25	0.24–0.25
	To Sons	Genes only									0.18	0.17–0.21	0.016
	To Daughters										0.14	0.11–0.16
	From Mothers										0.14	0.09–0.19	0.275
	From Fathers										0.17	0.15–0.19

^*Tetrachoric Correlation

Transmission of Suicide Attempt

As noted, table 2 contains the tetrachoric correlation for SA→SA mother-offspring and father-offspring transmission across our four family types as well as the weighted estimate across all informative families for parent-offspring pairs with a genes+rearing, genes only and rearing only relationship. The final column provides a heterogeneity test for those estimates. The estimates for mothers and fathers, their weighted estimate and the heterogeneity analyses are seen in table 3.

For SA→SA parent-offspring transmission, the overall estimates for genes+rearing, genes only and rearing only relationships were (± 95% CI), respectively: +0.23 (0.23–0.24), +0.13 (0.11–0.15) and +0.14 (0.11–0.16) (figure 1). No heterogeneity of estimation was seen for the genes only and rearing only relationships. For the genes+rearing analyses, where larger samples were available, inter-sample statistical heterogeneity was seen although the absolute difference in estimates was modest. The parallel ORs (± 95% CI) for these three associations equaled, respectively: 3.28 (3.21–3.35), 1.76 (1.66–1.86) and 1.89 (1.67–2.13).

Effect of Sex on Suicide Attempt Transmission

We tested, across the weighted estimates from our family types, the heterogeneity of parent-offspring transmission of SA across mother-daughter, mother-son, father-daughter and father-son pairs (table 4). Significant heterogeneity was seen in genes+rearing and genes only but not rearing-only relationships. To determine the origin of this heterogeneity, for both genes+rearing and genes only relationships, we compared the results for transmission to sons versus daughters and transmission from mothers versus fathers. In the genes+rearing sample, both comparisons differed significantly. However, the difference was more significant and larger for transmission to sons versus daughters compared to from fathers versus from mothers. For genes only relationships, the only significant difference was stronger transmission to sons than to daughters.

Impact of Psychiatric Illness on Cross-Generational Transmission

To examine the degree to which the transmission of SA from parents to children resulted from the transmission of psychiatric illness, we examined transmission of SA using a linear probability model across all of our family types and repeated those analyses controlling for the occurrence in parents of major depression, anxiety disorders, drug abuse, alcohol use disorder, bipolar illness and non-affective psychosis. We explored the degree to which the association for SA was attenuated by the addition of parental psychiatric disorders to the model. For the genes+rearing relationship, the weighted raw and controlled betas were, respectively, 6.14 (6.04–6.25) and 5.48 (5.38–5.59), producing a modest attenuation of 11% (figure 1). For the genes-only parent-offspring relationships, the parallel figures were 4.10 (3.68–4.52) and 2.48 (2.02–2.94), demonstrating a 40% attenuation. By contrast, for the rearing-only parent-offspring relationships, the parallel figures were 4.15 (3.31–4.98) and 4.18 (3.31– 5.05), demonstrating no attenuation.

Transmission of Suicide Completion

The tetrachoric correlation for SD→SD transmission in mother-offspring and father-offspring pairs are seen in table 2. For SD→SD parent-offspring transmission, the best estimates of the tetrachoric correlations across family types and mothers and fathers for genes+rearing, genes only and rearing only parent-child relationships were (± 95% CI), respectively: + 0.16 (0.15–0.18), +0.07 (0.02–0.12) and −0.05 (−0.17–0.07) (figure 1). The latter two of these estimates were imprecisely known because of the rarity of SD. Modest statistical heterogeneity was found only for genes+rearing relationships, which were stronger for mother-child than father-child pairs (table 2).

Transmission of Suicide Attempt to Suicide Death and Suicide Death to Suicide Attempt

Table 2 contains the tetrachoric correlation for SA→SD and SD→SA transmission from mothers and fathers to offspring across our four family types. No statistical heterogeneity is detected across family types. When we compare the weighted estimates in table 3, in genes+rearing analyses, results are significantly stronger for mothers than fathers, but no differences are seen for the genes only and rearing only relationships. Finally, at the bottom of table 3, we compare the aggregate estimates for SA→SD and SD→SA transmission. For genes+rearing relationships, the estimate of the former was statistically greater than the latter. Across family types, mothers and fathers, and SA→SD and SD→SA transmission, for genes+rearing, genes only and rearing only parent-child relationships were (± 95% CI), respectively: + 0.10 (0.09–0.11), +0.08 (0.07–0.09) and +0.06 (0.02–0.09).

From these results, we calculated the cross-generation genetic correlation of SA and SD to equal +0.84 (0.74–1.00), suggesting substantial sharing in the genetic risk factors for SA and SD. To test whether SA and SD were, from a genetic perspective, behaviors of differing severity on the same continuum of liability, we fit a three-category multiple threshold model to results from not-lived-with fathers and their children that include i) no SA or SD, ii) SA only and iii) SD only. The polychoric correlation (SE) was estimated at +0.13 (0.01), close to that obtained for SA alone but, the model fit poorly (x²= 40.26, df = 3, p <0.001).

DISCUSSION

We had five goals, the results for which we review in turn. First, we sought to clarify the magnitude and nature of the cross-generational transmission of SA. We found substantial parent-offspring resemblance for SA estimated in intact families by a tetrachoric correlation and OR as, respectively +0.23 and 3.28. Supporting the representativeness of our sample, this OR is within the range from a meta-analysis of family studies of SA: OR=2.77 (2.22–3.48) (4). The cross-generational transmission of SA resulted, in nearly equally degrees, from genetic and rearing effects. Calculating heritability from an adoption study is straight-forward as it equals twice the tetrachoric correlations for genes-only parent-child relationships, here equaling 26% (0.22–0.30).

Our findings were internally consistent across multiple family samples with one exception. Correlations from the mother-offspring genes+rearing relationship in the not-lived-with and stepfather families were moderately lower than that seen in the intact families. This introduced modest heterogeneity in the weighted average for genes+rearing relationships for mothers compared to fathers that was detectable due to our large sample sizes. The general similarity of our estimates for genes+rearing, genes only and rearing only across different family types supports the assumptions underlying our extended adoption design.

While we are unaware of prior adoption studies of parent-child transmission of SA, Peterson et al (8) found marginally significant evidence for excess risk for SA in biological siblings of Danish adoptees with SA. Our findings can also be usefully compared with those of three twin studies which examined self-report SA (11–13). Estimates of the heritability of SA were variable and known imprecisely, ranging from 17 to 55%. Shared environment was estimated in two studies at 19% (13) and 8% (11).

Our second goal was to examine sex effects in the parent-offspring transmission of SA. We found such effects of genes+rearing and genes only relationships, and in both of them, the most prominent finding was that transmission was stronger to sons than to daughters. While there is a large literature on sex effects in suicidal behavior (14), we have been unable to find prior comparable results.

Third, we sought to clarify the degree to which the parent-offspring transmission of SA resulted from the transmission of the liability to psychiatric and substance use disorders. Most importantly, in genes only parent-offspring relationships, adding six major psychiatric and substance disorders to the model resulted in a 40% reduction in transmission of SA. This suggests that approximately two-fifths of the parent-offspring genetic resemblance for liability to SA can be accounted for by transmission of risk to psychiatric and substance use disorders. These findings are not congruent with those from the Danish sibling adoption study noted above which found no reduction in resemblance for SA between adoptee and their biological siblings after controlling for psychiatric admissions (8). However, a summary of twin studies of SA concluded that “the presence of genetic vulnerability to nonfatal suicidal behaviors cannot be fully explained by genetically-influenced psychopathology ((5) p. 266)” and a similar conclusion has been reached from family studies of SA (3).

However, we found quite different results in our rearing-only parent-child pairs where the addition of parental psychiatric and substance use disorders to the model produced no attenuation in the magnitude of transmission of SA. These results suggest qualitive differences in the nature of the risk to SA from parents who provide genes versus only rear their offspring. For rearing only parents, our results would be consistent with social learning as a prime mechanism for the cross-generational transmission of risk to SA (15).

Our fourth goal was to examine the sources of parent-offspring resemblance for SD, a goal compromised by the rarity of SD in our sample and the resulting low power of our analyses. Consistent with the two prior adoption studies of SD which examined risk for SD in either all biological relatives of SD adoptees (6) or just their siblings (8), we found evidence for genetic transmission for SD. However, the magnitude of the transmission, although imprecisely known, was smaller than that for SA. Our results are also consistent with the partial adoption study (16) which examined resemblance for SD only between biological parents and their adopted-away offspring in an early Swedish adoptive cohort (born from 1946–68 and thereby overlapping for fourteen years with our 1955–1990 cohort) and reported a hazard ratio for SD slightly higher than the ORs obtained in our study.

Our direct estimates of rearing effects for SD were not significant and very imprecisely known. However, rearing effects can also be indirectly assessed from the difference in the correlations estimated from genes+rearing and genes only relationships. For SD, this would be estimated at +0.09. We also have weak evidence for rearing effects in our SA→SD analyses with positive estimates that do not differ from zero. Overall, SD is too rare in these Swedish families to be confident about the sources of parent-offspring transmission. Genetic effects for SD clearly exist but whether cross-generational environmental effects exist remains uncertain.

Our final goal was to understand the relationship between the cross-generational genetic transmission of SA and SD. Adding to our information about the parent-offspring transmission for both SA and SD, we examined their cross-transmission – that is from SA to SD and SD to SA. From the genetic correlations obtained from these four analyses, we could calculate a cross-generational genetic correlation between SA and SD. It was high (+0.84) but short of unity. One way to see why we found such a high correlation is to note that the genetic transmission from SA in parents to SD in children (+0.08) was nearly the same as the genetic transmission from SD to SD (+0.07). However, we were well powered in not-lived-with fathers and their offspring to test, from a genetic perspective, whether SA and SD could be assumed to reflect behaviors resulting from the same genetic liability differing only in severity. We could reject this hypothesis, indicating important qualitative differences in the nature of the transmitted liability to SA and to SD.

Limitations

These results should be interpreted in the context of four potentially important limitations. First, SA was ascertained from medical registries and are likely to represent, on average, more severe attempts than those obtained by self-report or interview-based methods. Approximately 40% of nonfatal suicide attempts do not obtain medical attention (17). Second, what is the quality of the diagnosis of SD in Sweden? In Sweden, unexpected deaths occurring outside of hospitals are all investigated through a forensic autopsy (18). Furthermore, the accuracy of diagnoses in the Cause of Death Register is evaluated every second year in an independent study of random samples (19). A recent systematic review of the reliability of suicide statistics found three studies from Sweden among the 31 meeting inclusion criteria all three of which supported the accuracy of the Swedish data (20). Third, in our definition of suicide, undetermined deaths were included, to minimize the possible impact of spatial and secular trends in the assignment of cases of SD (21). Fourth, we studied subjects born across three decades, which also required our definition of stepfathers to differ slightly across cohorts, and our findings could mask large cohort differences. We investigated this by analyzing separately the first and second chronological halves of our cohort. As seen in appendix table 1, differences were quite small between the older and younger cohort.

Conclusion

SA is strongly transmitted from parents to their children and this transmission arises nearly equally from genetic and rearing effects. While parental psychiatric and substance use disorders can explain almost half of the genetic transmission of SA across generations, these disorders had no effect on the impact of being reared by a non-biological parent with a SA. SD is modestly transmitted across generations, with genetic effects likely being more important than rearing effects. While SA and SD are substantially genetically correlated, a model that proposes that they reflect quantitatively different degrees of severity on the same continuum of liability fails badly.