The joint effects of genetic liability and the death of close relatives on risk for major depression and alcohol use disorder in a Swedish national sample

BACKGROUND

To determine whether genetic risk factors for major depression (MD) and alcohol use disorder (AUD) interact with a potent stressor - death of spouse, parent, and sibling - in predicting episodes of, respectively, MD and AUD.

METHODS

MD and AUD registrations were assessed from national Swedish registries. In individuals born in Sweden 1960-1970, we identified 7586, 388 459, and 34 370 with the loss of, respectively, a spouse, parent, and sibling. We started following subjects at age 18 or the year 2002 with end of follow-up in 2018. We examined time to event - a registration for MD within 6 months or AUD within a year - on an additive scale, using the Nelson-Aalen estimator. Genetic risk was assessed by the Family Genetic Risk Score (FGRS).

RESULTS

In separate models controlling for the main effects of death of spouse, parent, and sibling, FGRS, and sex, significant interactions were seen in all analyses between genetic risk for MD and death of relative in prediction of subsequent MD registration. A similar pattern of results, albeit with weaker interaction effects, was seen for genetic risk for AUD and risk for AUD registration. Genetic risk for bipolar disorder (BD) and anxiety disorders (AD) also interacted with event exposure in predicting MD.

CONCLUSIONS

Genetic risk for both MD and AUD act in part by increasing the sensitivity of individuals to the pathogenic effects of environmental stressors. For prediction of MD, similar effects are also seen for genetic risk for AD and BD.

The Impact of the Good Behavior Game on Risk for Drug Use Disorder in an Agent-Based Model of Southern Sweden

OBJECTIVE

Drug use disorder (DUD) is a worldwide problem, and strategies to reduce its incidence are central to decreasing its burden. This investigation seeks to provide a proof of concept for the ability of agent-based modeling to predict the impact of the introduction of an effective school-based intervention, the Good Behavior Game (GBG), on reducing DUD in Scania, Sweden, primarily through increasing school achievement.

METHOD

We modified an existing agent-based simulation model of opioid use disorder to represent DUD in Scania County, southern Sweden. The model represents every individual in the population and is calibrated with the linked individual data from multiple sources including demographics, education, medical care, and criminal history. Risks for developing DUD were estimated from the population in Scania. Scenarios estimated the impact of introducing the GBG in schools located in disadvantaged areas.

RESULTS

The model accurately reflected the growth of DUD in Scania over a multiyear period and reproduced the levels of affected individuals in various socioeconomic strata over time. The GBG was estimated to improve school achievement and lower DUD registrations over time in males residing in disadvantaged areas by 10%, reflecting a decrease of 540 cases of DUD. Effects were considerably smaller in females.

CONCLUSIONS

This work provides support for the impact of improving school achievement on long-term risks of developing DUD. It also demonstrated the value of using simulation modeling calibrated with data from a real population to estimate the impact of an intervention applied at a population level.

Death of parent, sibling, spouse, and child in a Swedish national sample and risk of subsequent stress reaction, major depression, alcohol-use disorder, and drug-use disorder

BACKGROUND

To determine, in a general population, how much rates of stress reactions (SR), major depression (MD), alcohol-use disorder (AUD) and drug-use disorder (DUD) increase after the death of close relatives.

METHODS

SR, MD, AUD, and DUD registrations were assessed from national Swedish registries. From the population followed from 2000 to 2018, those exposed to death of a close relative in 2002-2016 were matched to unexposed controls and analyzed in males and females by a controlled pre-post design using a difference-in-difference method.

RESULTS

Substantial, brief increases in risk for SR and more modest prolonged increases in MD were observed after death of relatives in both men and women greatest with children, followed by spouses, parents, and siblings. Relatively long-lasting modest increases in AUD but not DUD were also observed following death of relatives. The absolute increases for SR and MD were greater in females than males and for AUD greater in males than females. However, logistic regression analyses showed most effects did not differ significantly by sex. Consistently larger increases in disorder risk were seen with the death of younger v. older parents, siblings, and spouses and with accidental v. non-accidental death in children.

CONCLUSIONS

Applying a matched cohort design to Swedish population registries, death of close relatives was associated with, and likely caused, substantial increases in rates of SR, MD, and AUD, consistent with smaller prior clinical investigations. Through such registries, we can, in large representative samples, integrate the impact of exposures to selected environmental adversities into disorder risk pathways.

Risk factors for the development of opioid use disorder after first opioid prescription: a Swedish national study

BACKGROUND

We need to better understand the frequency and predictors of opioid use disorder (OUD) after first opioid prescription (OP).

METHODS

We followed 1 516 392 individuals from the Swedish population born 1980-2000, from 1 July 2007, until 31 Dec 2017. We examined putative risk predictors with univariable and multivariable Cox Models and the potential causal effects of predictors by propensity score and co-sibling analyses.

RESULT

Of the individuals in our cohort, 24.8% (375 404) received a first OP, of whom 3034 (0.90%) developed a subsequent first OUD. The hazard ratio (HR) (± 95% CIs) for OUD after OP equaled 7.10 (6.75-7.46), with a mean time to onset of 3.41 (2.39) years. The strongest putative risk factors for development of OUD after OP were prior psychiatric and substance use disorders, criminal behavior, parental divorce/death, poor school performance, current community deprivation, divorce, and male sex. Few predictors differed across sexes. OP renewal was associated with a HR of 3.66 (3.41-3.93) for OUD. Co-sibling and propensity score analyses suggested that at least a moderate proportion of the risk factor-OUD association was likely causal. A risk score to predict OUD after OP had an AUC of 0.85, where nearly 60% of cases scoring in the top decile.

CONCLUSIONS

In a general population sample, an OP represents a substantial risk factor for subsequent OUD. Many of the risk factors for OUD after OP can be readily assessed at the time of potential OP, permitting clinicians to evaluate the risk of iatrogenic OUD.

Testing Phenotypic Models of Posttraumatic Stress Disorder and Alcohol Use Disorder Comorbidity Using Longitudinal Registry Data

OBJECTIVE

Two predominant phenotypic models of causality exist to explain the high co-occurrence of posttraumatic stress disorder (PTSD) and alcohol use disorder (AUD): the self-medication and susceptibility models. Population-based longitudinal studies that simultaneously examine both models are needed. Thus, the goal of the present study is to test these models using the Swedish National Registries.

METHOD

Registries were used to conduct longitudinal Cox proportional hazard models (n ≈ 1.5 million) and cross-lagged panel models (N ≈ 3.8 million) with follow-up periods of ~23 years.

RESULTS

Covarying for cohort and socioeconomic status, Cox proportional hazards model results found strong support for the self-medication model. Results showed that PTSD predicted increased risk for AUD among both men (HR = 4.58 [4.42, 4.74]) and women (HR = 4.14 [3.99, 4.30]), significantly more so for men (interaction HR = 1.11 [1.05, 1.16]). Support was also found for the susceptibility model, although the effects were lower in magnitude than those for the self-medication model. AUD increased risk for PTSD among men (HR = 2.53 [2.47, 2.60]) and women (HR = 2.06 [2.01, 2.12]), and significantly more so for men (interaction term HR = 1.23 [1.18, 1.28]). Cross-lagged model results of simultaneously testing both models found support for bidirectionality. The PTSD-to-AUD paths and the AUD-to-PTSD paths were of modest effect for men and women.

CONCLUSIONS

The results from both complementary statistical approaches demonstrate that the models of comorbidity are not mutually exclusive. Although the Cox model results evidenced more support for the self-medication pathway, the cross-lagged model results suggest that the prospective relationships between these disorders are nuanced across development.

The actions and interactions of family genetic risk scores for alcohol use disorder and major depression on the risk for these two disorders

We know little about how genetic risk factors for two disorders jointly act and interact in predisposing to illness. Therefore, in the Swedish population, born 1970-1990 (n = 2,116,082) and followed through 2015, we examine, using additive Cox models, the impact of the family genetic risk scores (FGRS) for alcohol use disorder (AUD) and major depression (MD), their interaction with each other and with the relevant comorbid disorder on risk for AUD and MD. FGRS scores are constructed using rates of illness in first-fourth degree relatives. FGRS for AUD and MD interacted in predicting of both disorders and one FRGS (e.g., for AUD) interacted with the phenotype of MD to predict that disorder (e.g., AUD). These FGRS interactions were not substantially attenuated by adding interactions with the disorders. These results replicated across sexes. In predicting risk for a given disorder, we rarely consider genetic liabilities for other disorders. But such effects were here significant and interactive. Furthermore, the primary disorder genetic risk interacts with comorbid disorders. The pathways to risk for disorders from their and other disorders' genetic liability may be more complex than commonly considered.

Predicting the Onset of Opioid Use Disorder in the Swedish General Population

OBJECTIVE

Given the public health importance of opioid use disorder (OUD), we sought to understand better its risk predictors in the Swedish general population.

METHOD

We examined the Swedish population, born 1950-1970 (n = 2,092,359), and followed through 2018. Using Cox, logistic, and co-sibling models, we explored associations between a wide range of putative risk factors and a first onset of OUD--assessed through medical, criminal, and pharmacy registers--in the entire cohort and in the cohort wherein prior cases of drug use disorder (DUD) were censored.

RESULTS

OUD was predicted by the following four risk factor domains: (a) externalizing syndromes, especially prior non-opioid DUD; (b) psychopathology; (c) psychosocial factors, including social class and immigrant and marital status; and (d) serious injuries and pain diagnoses. When predicting OUD as the first form of DUD, the importance of pain diagnoses as a predictor increased. Co-sibling analyses suggested that the association of some of these risk factors with OUD onset was likely largely causal, whereas others were a mixture of causal effects and familial confounding. An aggregate risk score from these individual risk factors had reasonable receiver operating characteristic (ROC) curve performance.

CONCLUSIONS

OUD is a multifactorial syndrome for which risk can be meaningfully predicted by prior externalizing syndromes, internalizing and psychotic psychopathology, indicators of psychosocial status, and predictors of pain diagnoses. Some important differences were seen in the prediction of any OUD onset versus OUD onset as the first form of DUD. Much of the effect of these predictors appear, in co-sibling analyses, to likely reflect causal influences.

The causal nature of the association between resting pulse in late adolescence and risk for internalizing and externalizing disorders: a co-relative analysis in a national male Swedish sample

BACKGROUND

Resting pulse is robustly and inversely associated with the risk for externalizing disorders and may be positively associated with internalizing disorders. We know little about the causal nature of these associations.

METHODS

We examined resting pulse at conscription examination in 369 301 males born 1960-80 with a mean (s.d.) follow-up of 29.1 (7.7) years. From pulse rates, we predicted, using Cox models, the risk for criminal behavior (CB), drug abuse (DA), alcohol use disorder (AUD), major depression (MD), and anxiety disorders (AD), assessed from medical, criminal, and pharmacy registries. Co-relative analyses were conducted on the general population, cousin, half-sibling, full-sibling, and monozygotic pairs discordant for the outcome. Twin/sibling modeling for pulse was performed using OpenMX.

RESULTS

Familial resemblance for pulse resulted entirely from genetic factors. In the general population, the risk for externalizing disorders (CB, DA, and AUD) and internalizing disorders (MD and AD) were, respectively, significantly associated with low and high resting pulse rate. For CB, DA, and AUD, co-relative analyses showed that the inverse association with pulse resulted entirely from familial common causes (aka 'confounders'). By contrast, co-relative analyses found that the association between higher pulse and MD and AD resulted from direct causal effects.

CONCLUSIONS

Resting pulse has a negative and positive association with, respectively, the risk for externalizing and for internalizing disorders. Co-relative analyses indicate that the nature of these associations differ, suggesting that elevated pulse appears to directly increase the risk for internalizing disorders while the reduced pulse is a risk index for underlying traits that predispose to externalizing disorders.

Chains of risk for alcohol use disorder: Mediators of exposure to neighborhood deprivation in early and middle childhood

Our goal was to test a cascade model to identify developmental pathways, or chains of risk, from neighborhood deprivation in childhood to alcohol use disorder (AUD) in young adulthood. Using Swedish general population data, we examined whether exposure to neighborhood deprivation during early and middle childhood was associated with indicators of social functioning in adolescence and emerging adulthood, and whether these were predictive of AUD. Structural equation models showed exposure to neighborhood deprivation was associated with lower school achievement during adolescence, poor social functioning during emerging adulthood, and the development of AUD for both males and females. Understanding longitudinal pathways from early exposure to adverse environments to later AUD can inform prevention and intervention efforts.

Time-specific and cumulative effects of exposure to parental externalizing behavior on risk for young adult alcohol use disorder

BACKGROUND

Previous studies indicate that parental externalizing behavior (EB) is a robust risk factor for alcohol use disorder (AUD) in their children, and that this is due to both inherited genetic liability and environmental exposure. However, it remains unclear whether the effects of exposure to parental EB vary as a function of timing and/or chronicity.

METHODS

We identified biological parents with an alcohol use disorder, drug abuse, or criminal behavior, during different periods of their child's upbringing, using Swedish national registries. Logistic regression was used to determine whether the effect of parental EB exposure during different developmental periods differentially impacted children's risk for young adult AUD (ages 19-24). In addition, we tested how multiply affected parents and/or sustained exposure to affected parents impacted risk.

RESULTS

While parental EB increased risk for young adult AUD, timing of exposure did not differentially impact risk. Having a second affected parent increased the risk of AUD additionally, and sustained exposure to parental EB across multiple periods resulted in a higher risk of young adult AUD than exposure in only one period.

CONCLUSIONS

In this well-powered population study, there was no evidence of "sensitive periods" of exposure to national registry-ascertained parental EB with respect to impact on young adult AUD, but sustained exposure was more pathogenic than limited exposure. These findings suggest developmental timing does not meaningfully vary the impact, but rather there is a pervasive risk for development of young adult AUD for children and adolescents exposed to parental EB.

Resilience and Risk for Alcohol Use Disorders: A Swedish Twin Study

BACKGROUND

Resilience has been shown to be protective against alcohol use disorders (AUDs), but the magnitude and nature of the relationship between these 2 phenotypes are not clear. The aim of this study was to examine the strength of this relationship and the degree to which it results from common genetic or common environmental influences.

METHODS

Resilience was assessed on a 9-point scale during a personal interview in 1,653,721 Swedish men aged 17 to 25 years. AUD was identified based on Swedish medical, legal, and pharmacy registries. The magnitude of the relationship between resilience and AUD was examined using logistic regression. The extent to which the relationship arises from common genetic or common environmental factors was examined using a bivariate Cholesky decomposition model.

RESULTS

The 5 single items that comprised the resilience assessment (social maturity, interest, psychological energy, home environment, and emotional control) all reduced risk for subsequent AUD, with social maturity showing the strongest effect. The linear effect by logistic regression showed that a 1-point increase on the resilience scale was associated with a 29% decrease in odds of AUD. The Cholesky decomposition model demonstrated that the resilience-AUD relationship was largely attributable to overlapping genetic and shared environmental factors (57 and 36%, respectively).

CONCLUSIONS

Resilience is strongly associated with a reduction in risk for AUD. This relationship appears to be the result of overlapping genetic and shared environmental influences that impact resilience and risk of AUD, rather than a directly causal relationship.

Psychological strength assessed in late adolescence and risk for criminal behavior: a Swedish prospective cohort and twin analysis

BACKGROUND

Certain personality traits predispose to criminal behavior (CB). We further clarify this relationship in a Swedish national sample.

METHOD

Psychological strength (PS) was assessed on a nine-point scale at personal interview in 1 653 721 Swedish men aged 18-20 years. We examined the association between PS and total, violent and recurrent CB over the lifetime (logistic regression), prospectively (Cox regression) and by bivariate Cholesky decomposition in 2507 monozygotic and 2244 dizygotic twin pairs (OpenMx).

RESULTS

Examining linear effects by logistic regression, PS was robustly associated with lifetime risk of total CB (per point, odds ratio = 0.74) and even more strongly associated with risk for violent (0.69) and recurrent CB (0.52). Prospective predictions of these three forms of CB by PS were similar, with hazard ratios of 0.80, 0.73 and 0.54, respectively. Twin modeling demonstrated that, for all three CB types, the association with PS arose almost entirely from familial effects. Common shared environment accounted for 72, 56 and 43% of the phenotypic correlation between PS and, respectively, total, violent and recurrent CB. Parallel figures for common genetic effects were for 24, 37 and 54%, respectively.

CONCLUSIONS

PS is strongly related to risk for total CB, and even more strongly for violent and, especially, recurrent CB. This association is probably not causal but rather results from shared familial risk factors that make an impact both on PS and risk for CB. PS has a stronger overall correlation with more severe criminal outcomes and a higher proportion of that correlation results from common genetic factors.

A Swedish national twin study of criminal behavior and its violent, white-collar and property subtypes

BACKGROUND

We sought to clarify the etiological contribution of genetic and environmental factors to total criminal behavior (CB) measured as criminal convictions in men and women, and to violent (VCB), white-collar (WCCB) and property criminal behavior (PCB) in men only.

METHOD

In 21 603 twin pairs from the Swedish Twin Registry, we obtained information on all criminal convictions from 1973 to 2011 from the Swedish Crime Register. Twin modeling was performed using the OpenMx package.

RESULTS

For all criminal convictions, heritability was estimated at around 45% in both sexes, with the shared environment accounting for 18% of the variance in liability in females and 27% in males. The correlation of these risk factors across sexes was estimated at +0.63. In men, the magnitudes of genetic and environmental influence were similar in the three criminal conviction subtypes. However, for violent and white-collar convictions, nearly half and one-third of the genetic effects were respectively unique to that criminal subtype. About half of the familial environmental effects were unique to property convictions.

CONCLUSIONS

The familial aggregation of officially recorded CB is substantial and results from both genetic and familial environmental factors. These factors are moderately correlated across the sexes suggesting that some genetic and environmental influences on criminal convictions are unique to men and to women. Violent criminal behavior and property crime are substantially influenced respectively by genetic and shared environmental risk factors unique to that criminal subtype.

The etiologic role of genetic and environmental factors in criminal behavior as determined from full- and half-sibling pairs: an evaluation of the validity of the twin method

BACKGROUND

Twin studies have shown that criminal behavior (CB) is influenced by both genetic and shared environmental factors. Could these results be replicated using full-siblings and half-siblings?

METHOD

In 911 009 full-siblings reared together (FSRT), 41 872 half-siblings reared together (HSRT) and 52 590 half-siblings reared apart (HSRA), CB was assessed from the Swedish Crime Register. Modeling, including testing for age differences and rearing status, was performed using the OpenMx package.

RESULTS

Five sibling models were fitted examining FSRT and HSRT 0-2 years different in age, and both FSRT and HSRT, and FSRT, HSRT and HSRA 0-10 years different in age with and without a specified shared environment indexing age differences. Heritability estimates for CB ranged from 33 to 55% in females and 39 to 56% in males, similar to those found in our prior twin study on the same population. Estimates for the shared environment varied from 1 to 14% in females and 10 to 23% in males, lower than those estimated in the twin study. The specified shared environment indexed by sibling age differences was significant in all models tested.

CONCLUSIONS

Heritability estimates for CB from full- and half-siblings closely approximated those found from twins in the same population, validating the twin method. Shared environmental estimates were lower, suggesting the presence of shared environmental factors for CB specific to twins. When rearing status can be assessed, full- and half-siblings offer an additional method for assessing the role of genetic and environmental factors in complex disorders. However, age differences in siblings may need to be included in the models.

Environmental transmission of violent criminal behavior in siblings: a Swedish national study

BACKGROUND. Violent criminal behaviour (VCB) runs strongly in families partly because of shared environmental factors. Can we clarify the environmental processes that contribute to similarity of risk for VCB in siblings?

METHOD

We assessed VCB from the Swedish National Crime Register for the years 1973-2011 in siblings born 1950-1991. We examined by conditional logistic and Cox proportional hazard regression, respectively, whether resemblance for VCB in sibling pairs was influenced by their age difference and whether VCB was more strongly 'transmitted' from older→younger versus younger→older siblings.

RESULTS

In our best-fit logistic model, for each year of age difference in full sibling pairs, the risk for VCB in the sibling of a case versus control proband declined by 2.6% [95% confidence interval (CI) 2.2-3.0]. In our best-fit Cox model, the hazard rate for VCB in a sibling when the affected proband was older versus younger was 1.4, 2.1 and 2.9 respectively for a 1-, 5- and 10-year difference in siblings.

CONCLUSIONS

Controlling for genetic effects by examining only full siblings, sibling resemblance for risk for VCB was significantly greater in pairs closer versus more distant in age. Older siblings more strongly transmitted risk for VCB to their younger siblings than vice versa. These results strongly support the importance of familial-environmental influences on VCB and provide some insight into the possible mechanisms at work.

Relapse prevention and residual symptoms: a closer analysis of placebo-controlled continuation studies with escitalopram in major depressive disorder, generalized anxiety disorder, social anxiety disorder, and obsessive-compulsive disorder

OBJECTIVE

Analyses of data from 4 relapse-prevention studies with escitalopram were conducted in order to compare patients with and without residual symptoms with regard to relapse rates and global illness during double-blind, 24-week continuation periods.

METHOD

Clinical Global Impressions-Severity of Illness scores and relapse status in 4 studies published from 2005 to 2007, 1 each in major depressive disorder (MDD), generalized anxiety disorder, social anxiety disorder, and obsessive-compulsive disorder (OCD), were analyzed using mixed-effects model repeated measures as a function of Montgomery-Asberg Depression Rating Scale (MADRS) scores on items 1, 3, and 7 at randomization.

RESULTS

All studies showed a statistically significant (P < .0001) standardized effect size of about 0.7 for escitalopram versus placebo, with a number needed to treat approximately 4. Patients with residual symptoms (MADRS score > 0) and without residual symptoms (MADRS score = 0) at the start of continuation treatment were defined by how patients scored on 3 core items of the MADRS: depressed mood (observed), inner or psychic tension, and lassitude. At randomization, patients with a residual symptom were globally more ill than patients without such a symptom. Patients who did not continue active treatment worsened, even if they were initially free of a residual symptom. In contrast, patients who continued receiving escitalopram remained stable or further improved, regardless of residual symptoms or diagnosis. No clear picture emerged regarding whether patients with residual symptoms had a higher relapse rate.

CONCLUSIONS

The presence of residual symptoms is associated with significantly worse overall illness severity in all 4 diagnostic groups and with a higher (although not significantly) risk of relapse for patients with MDD or OCD. The greatest difference in all of the studies was between patients treated with escitalopram (relapse rates ~ 20%) and placebo (relapse rates of about 50%).