CoChat
Sign Up
Cited by 1Genetic Associations and EpidemiologyMental Health Treatment and AccessMental Health Research Topics

The development of an environmental risk score using Swedish National Registers and its impact on subsequent episodes of major depression

Open access
Kendler, Kenneth S., Lönn, Sara L., Sundquist, Jan, Sundquist, Kristina·2025·Psychological Medicine
DOI 10.1017/s0033291725000583PMC PMC12080639
Open this paper in CoChat — free accountOpen PDF

Abstract

Background: Stressful life events (SLEs) increase the risk for subsequent major depression (MD) episodes. Most prior studies of SLEs utilized questionnaires or interview-based assessments. We sought to develop and evaluate an environmental risk score (ERS) for MD from multiple classes of SLEs obtained from national Swedish registries. Methods: We assessed, in the entire adult population of Sweden (n = 7,105,712), the occurrence of 52 categories of SLEs derived from registry information for the 6 months prior to January 9, 2010 and the risk for MD registration over the subsequent 6 months. Weights for our ERS were obtained from a random half of our sample and ERS and its relationship to MD risk was evaluated in the second half. Results: The ERS was robustly related to risk for subsequent MD episodes. Women were more sensitive to the depressogenic effect of the ERS than men. Those with prior episodes of MD had larger absolute increases in MD risk from our ERS than those without previous episodes. Genetic risk for MD was associated with a greater sensitivity to the depressogenic effects of the ERS. A co-sibling control analysis suggested that most, but not all, of the association between ERS and subsequent risk for MD was causal. Conclusions: Valid measures of SLEs that predispose to risk for MD can be assessed from high-quality registry data. While not all event categories (e.g. interpersonal or romantic difficulties) can be assessed, this method avoids problems with accurate dating and recall bias and can be performed in very large samples.

AI snapshot

Written from the paper’s full text.

TL;DR

Using Swedish national registers, the authors built an environmental risk score (ERS) from 52 stressful-life-event categories and found it predicted subsequent major depression (MD) registration in the entire adult population (n = 7,105,712). Women, individuals with prior MD, and those with higher genetic risk were more sensitive to the ERS, and co-sibling analysis suggested most but not all of the association was causal.

Key findings

  • In the full sample, 11.3% had at least one SLE in the 6 months before baseline and 2.0% had an MD registration in the following 6 months; MD cases had a higher SLE proportion (16.9% vs. 11.3%) and higher FGRS_MD (0.33 vs. −0.01).
  • The ERS predicted MD with a whole-population beta of +0.97 (95% CI 0.94–1.00); the co-sibling beta was +0.82 (0.78–0.86), and +0.84 (0.80–0.89) after excluding family-member events, suggesting modest familial confounding and a largely causal association.
  • Women were more sensitive to the ERS than men (ERS × female interaction 0.4004, 95% CI 0.3405–0.4604); prior MD showed a larger absolute increase on the additive scale (interaction 0.9190, 95% CI 0.8408–0.9973), though logistic models indicated approximately 19% reduced sensitivity per prior episode (HR 0.81).
  • Higher genetic risk was associated with greater ERS sensitivity (ERS × FGRS_MD interaction 0.1104, 95% CI 0.0779–0.1429), and the weighted ERS had 80% greater predictive power than an unweighted sum of events.

Methods

  • Population-based cohort using linked Swedish nationwide registers; the sample was the entire adult population (n = 7,105,712), with 52 SLE categories derived from registry information and MD outcome defined by ICD-10 codes F32/F33 from inpatient, specialist, and primary care registries over the 6 months after baseline.
  • The sample was randomly split in half: weights for the ERS were estimated in a training half using a linear probability model, and the ERS was evaluated in the testing half; analyses used linear and quadratic models compared by adjusted R² and AIC, additive-scale interactions with sex, prior MD, and FGRS_MD, and a co-sibling linear mixed model with a random effect for full-sibling pairs.
  • FGRS_MD was based on 1st- to 5th-degree relatives (mean 40.1 relatives per proband), standardized by year of birth and county into a z-score; FGRS analyses were run on the subset with valid FGRS_MD measures.

Limitations

  • Stated: Swedish registry data cannot assess all relevant stressors (e.g., serious relationship or workplace difficulties, loss of confidants, extended social network crises); the 6-month exposure/outcome timing likely underestimates ERS effects; and not modeling conditionality for spouse/child events introduces conservative bias.
  • Stated: the ERS is “tuned” to predict MD, some SLE categories overlap (13 of 271 correlations >0.50, 4.8%), and MD is based on first registration rather than self-reported onset, with variable delay from onset to care likely attenuating temporal associations.
  • Stated: the co-sibling design only partly controls familial confounding because full siblings share about 50% of their genetic risk; the kindling interpretation is consistent with the data but does not provide direct evidence of the specific mechanism.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1017/s0033291725000583

Cites

44 of 44 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

3 of 35 cited sources checked · 1 supported · 1 flagged · accuracy not yet measured

  • Premisecitation

    The population prevalence of treated MD in Sweden (20.0% in females and 10.8% in males for those born 1970-1990) suggests that the study is not substantially under-ascertaining MD.

    “the population prevalence of treated MD nationally from those born from 1970 to 1990, equals 20.0 in females and 10.8% in males which suggests that we are not substantially under-ascertaining this disorder”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Kendler, Ohlsson, Lichtenstein, Sundquist, & Sundquist, 2018 source

      The Limitations section (page unavailable) reports treated-major-depression rates of 12.2% in women and 7.1% in men, not the quoted 20.0% and 10.8%. It also estimates that only about 50–60% of people with lifetime depression seek treatment, contrary to the suggestion that studying treated cases does not substantially under-ascertain major depression. (LLM, conf=0.82)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured
  • Premisecitation

    Several prior twin studies show evidence of increased sensitivity to the depressogenic effects of environmental exposures in those with high genetic risk.

    “several prior twin studies, assessing a range of environmental adversities, show evidence of increased sensitivity to the depressogenic effects of these environmental exposures in those with high genetic risk”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites Kendler et al., 1995 source
    • cites Strachan, Duncan, Horn, & Turkheimer, 2017 source

      The Results (G × E interaction; page unavailable) report that genetic variance in depressive symptoms increased with neighborhood deprivation, supporting the claim's genetic-sensitivity point for one exposure. This paper is one twin study, not evidence that several prior twin studies assessed a range of adversities. The Discussion also says it found no direct effect of neighborhood deprivation on depressive symptoms after accounting for familial factors, so a depressogenic effect of the exposure is not established. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
    • cites Wichers et al., 2009 source
  • Findingown data

    The entire sample included 7,105,712 individuals.

    “Our entire sample included 7,105,712 individuals of whom 11.3% had at least one SLE in the 6 months prior to January 9, 2010 and 2.0% had an MD registration over the 6 months after January 9, 2010.”
    Quote in the paperNot a substantial claim
  • Findingown data

    In the preliminary stages, a 6-month time frame for the ERS had a moderately stronger predictive impact on MD risk than a 3-month time frame.

    “we explored 3 versus 6-month time frames for the assessment of our ERS finding the latter to have a moderately stronger predictive impact on MD risk”
    Quote in the paperNot a substantial claim
  • Findingown data

    Using logistic regression, those with a prior MD are less sensitive to the effects of environmental stressors with an HR of 0.81 per prior depressive episode, meaning an approximately 19% reduced sensitivity to stressors for each previous episode.

    “Here we show, using logistic regression, that those with a prior MD are less sensitive to the effects of environmental stressors with an HR of 0.81 per prior depressive episode, meaning an approximately 19% reduced sensitivity to stressors for each previous episode.”
    Quote in the paperSubstantial claim
  • May combine two claims. 12
  • Classified as probably not a substantial claim. 9
  • The quote only partly states this claim. 1

  • May combine two claims.

    Premisecitation

    SLEs are modestly heritable and/or run in families.

    “SLEs are modestly heritable (Bolinskey, Neale, Jacobson, Prescott, & Kendler, 2004) and/or run in families (Middeldorp, Cath, Vink, & Boomsma, 2005)”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Bolinskey, Neale, Jacobson, Prescott, & Kendler, 2004

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Middeldorp, Cath, Vink, & Boomsma, 2005 source

  • May combine two claims.

    Findingown data

    In sensitivity analyses, the effects of the ERS on MD risk were highest within 2 months of the January 9, 2010 cut-off but declined only slightly out to 6 months, suggesting only a modest loss of signal.

    “the effects of the ERS on MD risk were highest within 2 months of the January 9, 2010 cut-off but declined only slightly out to 6 months suggesting that we experienced only a modest loss of “signal” here”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    Rates of major depression in the 6 months after January 9, 2010 were approximately twice as high in women as in men.

    “As expected, rates of MD in the 6 months after January 9, 2010 were approximately twice as high in women as in men.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    2.0% of the sample had a major depression registration over the 6 months after January 9, 2010.

    “Our entire sample included 7,105,712 individuals of whom 11.3% had at least one SLE in the 6 months prior to January 9, 2010 and 2.0% had an MD registration over the 6 months after January 9, 2010.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    A linear model fitted the data well but was modestly improved by adding a quadratic function, with the fit dropping 88 AIC units and a modest improvement in predicted variance.

    “A linear model fitted the data well but the model was modestly improved by the addition of a quadratic function with the fit dropping 88 AIC units and a modest improvement in predicted variance.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    A number of specific events have considerably stronger depressogenic effects in women (e.g., assault with a weapon, criminal conviction, and spousal cancer) and far fewer events have stronger depressogenic effects in men (e.g., physical assault and spousal heart attack).

    “we show a number of specific events with considerably stronger depressogenic effects in women (e.g. assault with a weapon, criminal conviction, and spousal cancer) and far fewer events with stronger depressogenic effects in men (e.g. physical assault and spousal heart attack)”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Across all tables, 13 of 271 tetrachoric correlations between individual SLEs (4.8%) were >0.50, suggesting that content overlap among individual SLEs is not a major problem in interpreting the findings.

    “we had 271 correlations of which 13 (4.8%) met that standard”
    • Also stated in: § Limitations · hedged
      “These results suggest that content overlap among our individual SLEs is not a major problem in the interpretation of our findings”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The data on changes in the association between SLEs and MD episodes over time are consistent with the kindling hypothesis but do not present any direct evidence of the specific mechanism involved.

    “while this data – changes in the association between SLEs and MD episodes over time – is consistent with the kindling hypothesis – but does not present any direct evidence of the specific mechanism involved”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The environmental risk score was developed using a split-half design: weights were obtained from one random half of the sample and the score was evaluated in the second half.

    “We split our entire sample into two random halves and used one half as a training set to develop our MD ERS and then we applied the risk model on the testing half to evaluate the derived MD ERS.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The 52 categories of registry-derived events are valid indicators of stressful life events.

  • May combine two claims.

    Premisecitation

    Death of a close relative (parents, siblings, spouse, and children) is associated with a substantial and sustained increased risk for incident and recurrent episodes of major depression, as well as for stress reaction and alcohol use disorder.

    “the death of a close relatives (parents, siblings, spouse, and children) was associated with a substantial and sustained increased risk for incident and recurrent episodes of MD as well as for stress reaction and alcohol use disorder (Kendler, Lonn, Sundquist, & Sundquist, 2023).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Kendler, Lonn, Sundquist, & Sundquist, 2023

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationnone given

    The registry-based method avoids problems with accurate dating and recall bias.

    “While not all event categories (e.g. interpersonal or romantic difficulties) can be assessed, this method avoids problems with accurate dating and recall bias and can be performed in very large samples.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The 52 categories of registry-derived events are valid indicators of stressful life events.

  • May combine two claims.

    Premisecitation

    Severe events and/or multiple stressful events occurring close together in time have an increased aggregate effect on depressive risk.

    “consistent with prior findings that severe events and/or multiple stressful events occurring close together in time have an increased aggregate effect of depressive risk”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Frank et al., 1996 source
    • cites Kendler et al., 1999 source
    • cites Miller & Ingham, 1985 source

  • Classified as probably not a substantial claim.

    Premisecitation

    Genetic risk for MD increases exposures to certain classes of SLEs.

    “prior evidence has shown that genetic risk for MD increases exposures to certain classes of SLEs”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Kendler & Karkowski-Shuman, 1997 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Interview-based measures of stressful life events likely produce more valid results than questionnaire-based measures.

    “Nearly all such studies have relied on a questionnaire or interview-based retrospective assessments of SLEs completed by respondents with the interview-based measures likely producing more valid results (Brown, Sklair, Harris, & Birley, 1973; Dohrenwend, Link, Kern, Shrout, & Markowitz, 1987; Dohrenwend, Raphael, Schwartz, & Skodol, 1993; Dohrenwend, Dohrenwend, Dodson, & Shrout, 1984; Paykel, 198”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Brown, Sklair, Harris, & Birley, 1973 source
    • cites Dohrenwend, Link, Kern, Shrout, & Markowitz, 1987

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Paykel, 1983 source

  • Classified as probably not a substantial claim.

    Premisenone given

    A range of confounders could both influence the risk of stressful life events and depressive episodes.

    “Given that a range of confounders could both influence the risk of SLEs and depressive episodes, it would be inappropriate to assume that the predictive associations that we describe between SLE exposure and subsequent risk for MD are entirely causal.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationnone given

    Not all event categories, such as interpersonal or romantic difficulties, can be assessed from registry data.

    “While not all event categories (e.g. interpersonal or romantic difficulties) can be assessed, this method avoids problems with accurate dating and recall bias and can be performed in very large samples.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The 52 categories of registry-derived events are valid indicators of stressful life events.

  • May combine two claims.

    Study design and conductreasoning

    The Swedish registry data used did not permit assessment of all relevant environmental stressors for MD, so the ERS covers a more restricted range of adversities than interview or questionnaire-based assessments.

    “official Swedish registry data did not permit the assessment of all relevant environmental stressors for MD”
    • Also stated in: § Limitations · plain
      “our method of SLE assessment covers a more restricted range of adversities than is possible with interview or questionnaire-based assessments”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Kendler et al., 1998 source

  • Classified as probably not a substantial claim. May combine two claims. The quote only partly states this claim.

    Interpretationnone given

    The ERS method has several important advantages in SLE assessment compared to questionnaire or interview-based measures, including objectivity, lack of dependence on recall biases and inaccuracies in dating, and scalability to large samples.

    “This method has several important advantages in SLE assessment compared to questionnaire or interview-based measures, including objectivity in the assessment, lack of dependence on recall biases and inaccuracies in dating, and scalability to large samples.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    By design, the current ERS is tuned to predict MD.

    “by design, our current ERS is “tuned” to predict MD”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductnone given

    The ERS method is not without important limitations, especially the inability to assess certain important areas of adversity and the lack of precise timing of the onsets of depressive episodes.

    “It is, however, not without important limitations, especially the inability to assess certain important areas of adversity and the lack of precise timing of the onsets of depressive episodes.”
    Quote in the paperSubstantial claim
20 of 20

  • The paper relies on this without support.

    Questionnaire or interview-based assessments of stressful life events are time-consuming and typically available on only limited samples.

    “However, such assessments are time-consuming and typically available on only limited samples.”

  • The paper relies on this without support.

    The 52 categories of registry-derived events are valid indicators of stressful life events.

    “We assessed, in the entire adult population of Sweden (n = 7,105,712), the occurrence of 52 categories of SLEs derived from registry information for the 6 months prior to January 9, 2010 and the risk for MD registration over the subsequent 6 months.”

  • The paper relies on this without support.

    The Swedish national registries provide high-quality data suitable for assessing stressful life events.

    “Valid measures of SLEs that predispose to risk for MD can be assessed from high-quality registry data.”

  • The paper relies on this without support.

    The Swedish national registry data accurately record the environmental stressors and MD diagnoses used to construct the ERS.

    “we do not have measures of self-report onsets for MD but rather need to rely on first registrations in the Swedish medical registries”
4 of 4

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Figure 1 is assigned two different subjects

    Minor

    The analysis calls Figure 1 the overall ERS-MD relationship, but the Figure 1 caption identifies it as the contributions of individual stressful events. The overall relationship is captioned as Figure 2.

  • Results use a different exposure cutoff

    Minor

    The Results place the exposure and outcome windows around January 9, 2010, whereas the stated calendar windows place their boundary at September 1, 2010.

  • Conflicting baseline dates

    Minor

    The Methods set the baseline at October 1, 2010, but the Limitations describe exposures ending August 30 and outcomes beginning September 1, 2010. Those windows use September 1, not October 1, as the baseline.

  • Model D is described as both excluding and including an interaction

    Minor

    The text assigns both the two-predictor model and the model with an interaction to Model D. Table 2 instead places the two-predictor model in Model C and the interaction in Model D.

4 of 4

Cites

Flagged first, then by citations

2 of the 46 references couldn’t be matched to a known paper.

Research works printed but not linked: 1

  1. 10.1375/twin.8.3.224 · DOI

Cited by

Sorted by citations

Related

Authors
Kendler, K. S., Lönn, S. L., Sundquist, J., & Sundquist, K.
Affiliations
  • Virginia Commonwealth University (US)
  • Lund University (SE)
Corresponding author
Kendler, Kenneth S.
ORCID
Venue
Psychological Medicine
Publisher
Cambridge University Press (CUP)
Published
2025
Volume
55
Pages
e74-e74
ISSN
  • 0033-2917 (ISSN-L)
  • 1469-8978
DOI
10.1017/s0033291725000583
PMC
PMC12080639
OpenAlex
W4408236775
Semantic Scholar
c89e36198bc9622817129b5da769f81b84808912
Cited by
1
References
46
Citations by year
2025: 1
FWCI
0.54
Type
journal article
Language
en
Open access
Licence
Locations
  • Psychological Medicine (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Lund University Publications (Lund University) (submittedVersion)
  • PubMed Central (submittedVersion, cc-by)
Funders
Vetenskapsrådet (2021-06467, 2020-01175)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
  • Social Sciences › Psychology › Social Psychology › Mental Health Treatment and Access
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Mental Health Research Topics
Keywords
environmental risk score, life events, major depressive disorder, depression recurrence, national registry, registry data, sex differences, genetic risk, causal inference
MeSH terms
Adolescent, Adult, Aged, Major Depressive Disorder, Depressive Disorder, Major, Female, Humans, Life Change Events, Male, Middle Aged, Registries, Risk Factors, Sex Factors, Stress, Psychological, Sweden, Young Adult

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-14.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-14.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review