Programme with abstracts
08:00 - 09:00
09:00 - 09:10
Sven Oskarsson, Department of Government, Uppsala University and Chair of the Swedish Network for Register-Based Research (SWE-REG)
9:10 – 10:00
Catarina Almqvist Malmros, Professor at the Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
Title: “Challenges and opportunities for register-based research”
10:00 - 10:30
10:30 – 12:00
Theme: Research on child-and early adulthood
Chair: Annika Rosengren, Department of Molecular and Clinical Medicine, University of Gothenburg
Oral session 1.1
Title: Religion and mental health in young adulthood – a register based study on differences by religious affiliation in sickness absence due to mental disorders in Finland
Speaker: Jan Saarela, Åbo Akademi University
Background: Religiosity and spirituality are known to be positively correlated with health. This is the first study to analyse the interrelation between religious denomination and sickness absence due to mental disorders using population register data with detailed ICD-codes.
Methods: A follow-up study was based on the entire population born in Finland in 1984–1996 (N=794,476). Each person was observed from age 20 in the period 2004–2018. Cox proportional hazards models were applied to analyse the association between religious denomination and first-time sickness allowance receipt from any cause and mental disorder. Mental disorders were further divided into severe mental illness (F20-F31), depression (F32-F33), anxiety (F40-F48), and any other mental disorder (all other F codes). Men and women were analysed separately.
Results: Differences in sickness absence due to mental disorders by religious affiliation were substantial. As compared to members of the Evangelical Lutheran state church, the hazard for non-affiliated women was 1.34 (95% CI: 1.30-1.39), while that for women with any other religion was 1.27 (1.19-1.35), after adjusting for own and parental characteristics. Corresponding numbers for men were 1.45 (1.39-1.50) and 1.42 (1.30-1.54), respectively. The gradient was larger for severe mental illness and depression than for anxiety and other mental disorders. For sickness absence due to any cause there was no difference between Lutherans, non-affiliated, and those with any other religion.
Conclusions: The strong association between mental disorders and church membership as measured from administrative registers should be further examined by epidemiologists and public health practitioners.
Oral session 1.2
Title: Effect of paediatric obesity treatment on morbidity and mortality in young adulthood: Real-world data from the Swedish childhood obesity treatment register
Speaker: Resthie Putri, Department of Clinical Science, Intervention and Technology (CLINTEC), Karolinska Institutet
Introduction: The long-term impact of paediatric obesity treatment efficacy on adult outcomes remains unclear. We evaluated its effect on young adult morbidity and mortality using real-world data.
Methods: A nationwide cohort data of children (6–17 years) undergoing ≥1 year obesity treatment from the Swedish childhood obesity treatment register (BORIS)(1996–2019) were linked with several national registers. Individuals were followed from the date they turned 18 years to 2020. Outcomes were type 2 diabetes (T2D), hypertension, fractures, dyslipidaemia, depression/anxiety, and mortality. Treatment response was categorised into: “remission” (no longer obesity), “good response” (body mass index standard deviation score [BMI SDS] reduction ≥0.25 units), “poor” (BMI-SDS increase ≥0.25), otherwise “intermediate”. Cox regression (attained age as timescale), adjusted for age and degree of obesity at baseline, sex, socio-economic position, and migrant background was applied.
Results: Of 6,713 individuals, 56% were males. Median age at baseline was 12.1 (Q1–Q3: 10.1-14.3) years, at end of follow-up 21.9 (19.7–25.3) years. Compared to poor response, good response had lower risk for T2D (35 vs 73/1000 PY, HR=0.42 [95% CI 0.23–0.77]), dyslipidaemia (15 vs 37/1000 PY, HR=0.31 [0.13–0.75]), and mortality (2.5 vs 18/1000 PY, HR=0.42 [0.30–0.58]). Obesity remission showed similar reduced risk, and lower risk for hypertension (HR=0.40 [0.24-0.65]). Treatment response did not affect the risk of anxiety/depression or fractures.
Conclusion: Good treatment response and obesity remission in paediatric obesity treatment are associated with reduced risk for metabolic morbidity and overall mortality in young adulthood.
Oral session 1.3
Title: Intergenerational transmission of psychiatric conditions and psychiatric, behavioral, and psychosocial outcomes in offspring: A hierarchical factor model approach
Speaker: Mengping Zhou, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
Importance: Psychiatric conditions in parents are associated with many psychiatric and non- psychiatric outcomes in offspring. ?However, it remains unknown whether this intergenerational transmission was attributable to broader psychopathology comorbidity or to specific conditions.
Objective: To estimate associations between general and specific factors of psychopathology in parents, and a wide range of register-based outcomes in their offspring.
Design, Setting and Participants: This Swedish national registers-based cohort study included 2 947 703 individuals born between 1970 and 2000, and followed up through December 31, 2013.
Exposures: Hierarchical factor model consisting of one general and three specific psychopathology factors fit to 9 parental psychiatric diagnoses and violent criminal court convictions.
Outcomes: 31 outcomes sorted into 6 broad clusters: psychotic-like outcomes, neurodevelopmental outcomes, internalizing outcomes, externalizing outcomes, behavior and accidents, and psychosocial outcomes.
Results: Of 2 947 703 individuals, 1 518 252 (51.5%) are males and the mean (SD) age at the end of follow-up is 28.7 (8.92). The general psychopathology factor in parents was significantly associated with all 31 offspring outcomes ?(Odds Ratio (OR) range: 1.08–1.40). The specific psychotic factor in parents was primarily associated with psychotic-like outcomes (OR range: 1.05–1.25), and the specific internalizing factor in parents was primarily associated with offspring internalizing (OR range: 1.11–1.13) and neurodevelopmental outcomes (OR range: 1.02–1.10). The specific externalizing factor in parents was associated with externalizing (OR range: 1.21–1.32) and internalizing outcomes (OR range: 1.01–1.13).
Oral session 1.4
Title: The legacy of exile for children of refugees: examining adaptation across multiple areas of life
Speaker: Lisa Harber-Aschan, Demography Unit, Department of Sociology, Stockholm University
While much is known about the lives of refugees, less is known about the lives of their children, particularly with respect to inequalities they may face in adulthood. Using longitudinal register data for the population of Sweden, we holistically study a range of life outcomes at age 30 using six measures: tertiary education, earnings, unemployment, housing support, childbearing, and health (any unplanned hospital visits). We estimate differences between the Swedish-born children of refugees (G2 refugees) and (a) children of Swedish-born parents, (b) children of non-refugee immigrants (G2 non-refugees), and (c) foreign-born children of refugees (G1.5 refugees). We use parametric Generalized Linear Models, varying the link function according to the distribution of the outcome, stratifying the analyses by sex and country of origin. We find the outcomes for G2 refugees are extremely heterogeneous. Overall, compared to children of two Swedish-born parents, G2 refugees at age 30 have lower fertility (IRR=0.85; 95% CI: 0.83,0.87), poorer health (OR=1.13; 95% CI: 1.02,1.26) and experience inequalities in annual earnings (β=-17,029, 95% CI: -19374,-14,321) and unemployment (OR=1.72; 95% CI: 1.60,1.85), but do not have lower education (OR=0.99; 95% CI: 0.95,1.03). However, there is considerable variation by parental country of birth; for example, children of refugees from Lebanon are those most likely to experience socioeconomic disadvantage, while children of refugees from Iran have similar levels of unemployment despite having substantially higher education and earnings. Our results indicate that the adaptation of children of refugees is not a uniform process, with strong support for segmented adaptation.
Theme: Use of registers in clinical research
Chair: Jesper Lagergren, Department of Molecular medicine and Surgery, Karolinska Institutet
Oral session 2.1
Title: Prescription antibiotics use prior to colorectal cancer surgery and risk of postoperative complications: A Swedish national population-based study
Speaker: Sai San Moon Lu, Umeå University, Department of Radiation Sciences, Oncology Unit
Background: Gut microbiome composition is believed to impact the risk of surgical site complications after colorectal cancer surgery. Antibiotics affect the gut microbiome, but evidence for a role in surgical site complications is inconclusive. We investigated prescription antibiotics use up to 4.5 years before surgery in relation to the risk of surgical site infections, including anastomotic leakage, within 30 days after colorectal cancer surgery.
Methods: We conducted a cohort study using complete-population data from Swedish national registers between 2005 and 2020.
Results: The final study population comprised 26 527 colon cancer and 12 312 rectal cancer cases. A dose-response relationship was observed between antibiotics use and risk of surgical site infections in colon cancer (adjusted odds ratio (aOR) for low use = 1.19, 95% confidence interval (CI) 1.03-1.37, aOR for very high use = 1.78, 95% CI 1.32-2.42, versus no use, Ptrend <0.001). For anastomotic leakage in specific, the association persisted (Ptrend = 0.047). The urinary tract antiseptic methenamine hippurate, which has no known effect on gut microbiota, was not associated with surgical site infections. In rectal cancer patients, antibiotics use was not associated with surgical site infections or anastomotic leakage. For cardiovascular and/or neurological complications, which we included as a negative control due to the lack of a clear microbiome hypothesis, associations with surgical site infections were null in both colon and rectal cancer.
Conclusion: Prescription antibiotics use is associated with a higher risk of surgical site infections, including anastomotic leakage, after colon cancer but not rectal cancer surgery.
Oral session 2.2
Title: Non-erosive gastro-oesophageal reflux disease and incidence of oesophageal adenocarcinoma: population-based cohort study in three Nordic countries
Speaker: Dag Holmberg, Department of Molecular medicine and Surgery, Karolinska Institutet
Background: Gastroesophageal reflux disease (GERD) is associated with an increased risk of esophageal adenocarcinoma, but whether patients with non-erosive GERD, i.e., GERD symptoms without esophageal mucosal abnormalities, are at an increased risk is unknown.
Methods: This population-based cohort study included all 486,556 patients who underwent endoscopy for GERD in Finland, Denmark, or Sweden from 1987-2019. Non-erosive GERD was defined by an absence of esophagitis and any other esophageal diagnosis within 12 months of the endoscopy, while erosive GERD was defined by the presence of esophagitis at endoscopy. Standardized incidence ratios (SIR) with 95% confidence intervals (95%CI) were calculated by dividing the observed number of esophageal adenocarcinomas in the non-erosive and erosive GERD cohorts, analyzed separately, by the expected number, derived from the general populations in Finland, Denmark, and Sweden of the corresponding age, sex, and calendar period.
Results: Among 285,811 patients with non-erosive GERD, 228 developed esophageal adenocarcinomas during 2,081,051 person-years of follow-up. The overall risk of esophageal adenocarcinoma was similar to that of the background population (SIR 1.04, 95%CI 0.91-1.18), and did not increase with longer follow-up. Patients with erosive esophagitis at endoscopy (200,745 patients, 1,750,249 person-years, and 542 esophageal adenocarcinomas) had a notable increased overall risk of esophageal adenocarcinoma (SIR 2.36, 95%CI 2.17-2.57), which became more pronounced with longer follow-up.
Conclusion: Patients with non-erosive GERD seem to have a similar risk of esophageal adenocarcinoma compared to the corresponding background population. This finding is relieving for the majority of GERD patients with normal endoscopy and may guide follow-up strategies.
Oral session 2.3
Title: Circulating proteins and incident myocardial infarction: appraising sex-specific differences and causality
Speaker: Olga Titova, Department of Surgical Sciences, Uppsala University
Background: The molecular pathways of the development of myocardial infarction (MI) are not fully understood. We investigated the relationships between circulating cardiometabolic proteins and MI risk using observational and Mendelian randomization (MR) approaches and explored potential sex-specific differences.
Methods: The cohort study included 11,751 participants (55-93 years). Data on 256 proteins assessed with Olink proximity extension assays, biochemical and questionnaire-based information was used. Participants were followed up for incident MI and death over 8 years through linkage to the Swedish National Patient Register and the Cause of Death Register. MR analyses were conducted using index cis-genetic variants strongly related to the proteins as instrumental variables.
Results: Using a discovery and validation approach, we identified 16 proteins associated with incident MI in the entire cohort following adjustment for potential confounders and multiple testing. Among these proteins, strongest magnitude of association was for growth differentiation factor 15, followed by osteoprotegerin, chitinase-3-like protein 1, retinoic acid receptor responder protein 2, urokinase plasminogen activator surface receptor, tissue factor pathway inhibitor (TFPI), cathepsin D, C-X-C motif chemokine 16, matrix metalloproteinase 12 (MMP-12), C-C motif chemokine 15, transferrin receptor protein 1, P-selectin, low-density lipoprotein receptor, and T-cell surface glycoprotein CD1c. Multiple sex-specific relationships were observed with more associations identified among women. TFPI and MMP-12 had an opposite direction of association in MR analyses.
Conclusions: This study identified several novel and previously established associations between circulating proteins and incident MI; and, for the first time, suggested sex-specific patterns in multiple protein-MI associations.
Oral session 2.4
Title: Decreased risk of esophageal adenocarcinoma after gastric bypass surgery in a cohort study from three Nordic countries
Speaker: Jesper Lagergren, Department of Molecular medicine and Surgery, Karolinska Institutet
Background: Obesity is strongly associated with esophageal adenocarcinoma and moderately with cardia adenocarcinoma, but whether weight loss prevents these tumors is unknown.
Methods: This population-based cohort study included patients with an obesity diagnosis in Sweden, Finland, or Denmark. Participants were divided into a bariatric surgery group and a non-operated group. The incidence of ECA was first compared with the corresponding background population by calculating standard incidence ratios (SIR) with 95% confidence intervals (CI). Second, the bariatric surgery group and the non-operated group were compared using multivariable Cox regression, providing hazard ratios (HR) with 95% CI, adjusted for sex, age, comorbidity, calendar year, and country.
Results: Among 748,932 participants with an obesity diagnosis, 91,731 underwent bariatric surgery, predominantly gastric bypass (n=70,176; 76.5%). The SIRs of ECA decreased over time following gastric bypass, from SIR=2.2 (95% CI 0.9-4.3) after 2-5 years to SIR=0.6 (95% CI <0.1-3.6) after 10-40 years. Gastric bypass patients were also at a decreased risk of ECA compared to non-operated patients with obesity (adjusted HR=0.6, 95% CI 0.4-1.0 [0.98]), with decreasing point estimates over time. Gastric bypass was followed by a strongly decreased adjusted risk of esophageal adenocarcinoma (HR=0.3, 95% CI 0.1-0.8), but not of cardia adenocarcinoma (HR=0.9, 95% CI 0.5-1.6), when analyzed separately. There were no consistent associations between other bariatric procedures (mainly gastroplasty, gastric banding, sleeve gastrectomy, and biliopancreatic diversion) and ECA.
Conclusion: Gastric bypass surgery may counteract the development of esophageal adenocarcinoma in morbidly obese individuals.
Theme: Political participation and social cohesion
Chair: Maria Brandén, Institute for Analytical Sociology, Linköping University
Oral session 3.1
Title: Propinquity as a mediator of ethnic endogamy - A discrete choice analysis of the Swedish partner market
Speaker: Jesper Lindmarker, Institute for Analytical Sociology, Linköping University
Romantic partnerships between groups is considered a social thermometer of intergroup boundaries. To date, however, it has been challenging to account for the types of partners who are locally available and to understand the effect of spatial mechanisms. In this paper, we use updated, multivariate tools and data to formally examine whether and to what degree residential propinquity—i.e., the spatial proximity of singles (with different ethnic backgrounds)—accounts for patterns of ethnic homogamy. Specifically, we employ discrete choice, conditional logistic regression models to longitudinal micro-data from the Swedish registers to investigate whether segregation mediates ethnic homogamy. We identify all different-sex couples who form cohabiting unions between 1991 and 2017 and eventually either married or had a child together. For each union we sample counterfactual unions and estimate parameters that maximise the likelihood that the actual partner was chosen over the counterfactuals. Because these are micro-models, we can estimate the relative importance of multiple partner dimensions net of differences in group sizes along these dimensions. We find that propinquity has a substantial effect for the likelihood of union formation even within smaller regions. Further, we find substantial differences in endogamous tendencies between ethnic groups, net of partner market characteristics, suggesting preferential and norm differences. Lastly, we find that accounting for segregation through propinquity mediates endogamy estimates. Not accounting for the geographical distribution of groups risks overstating the degree of endogamy for segregated groups while understating it for lesser segregated groups.
Oral session 3.2
Title: Mandatory Civil Service and Its Impact on Social Cohesion
Speaker: Maria Brandén; Institute for Analytical Sociology, Linköping University
Bringing members of different social groups into contact (intergroup contact) has been theorized to be an important mechanism toward increasing social cohesion and improving intergroup relations. This study examines the role of intergroup contact on intergroup relations by analyzing a unique social context in Sweden—mandatory military service. During military service, men were exposed to peers that they likely would not have met otherwise, and engaged in training that required a high degree of team work. We test whether this impacts social cohesion by examining how exposure to individuals with different immigrant backgrounds is associated with the likelihood to partner with someone with immigrant background. All analyses build on Swedish population registers. Our findings suggest that exposure to individuals with a background in other countries in military service is associated with a higher likelihood to partner with someone from such groups, supporting the hypothesis that the exposure to mixed environments in the military service can increase social cohesion.
Oral session 3.3
Title: The Causal Effects of Wealth on Political Participation and Political Preferences
Speaker: Erik Lindqvist, Swedish Institute for Social Research, Stockholm University
We test whether random, positive wealth shocks in the form of lottery prizes induce changes in political participation and political preferences. Compared to suitably matched controls, large-prize winners are no more likely to cast votes in national elections or run for political office. We also find no winnings during childhood significantly affect future political participation. But political preferences change as a result of winning the lottery. Winners of larger lottery prizes become more negative toward taxes on wealth, real estate and inheritances. Consistent with material self-interest influencing preferences, these effects diminish over time as lottery wealth dissipates. Effects on other political preferences are statistically insignificant, though they often go in the direction of a more right-wing political orientation. We find no evidence that lottery wealth changes moral values or strengthen beliefs in the importance of hard work for success in life.
Oral session 3.4
Title: Does Adolescence Neighbourhood Civic Norms and SES Moderate the Influence of Individual-level Genetic Resources on Political Participation?
Speaker: Oskar Pettersson, Department of Government, Uppsala University
The distal, yet causal relationship between individual-level genetic factors and adult-life political participation has been shown in recent twin studies and genomic studies. Something that has not yet received proper attention is the question of whether the effects of genetic factors—here referred to as "individual-level genetic resources," and captured using a polygenic index for educational attainment (EA PGI)—depend on social context during the formative years. Accordingly, this paper tests whether adolescence neighborhood-level civic norms and SES moderate genetic effects on adult-life political participation. The results provide lower-bound estimates of a negative interaction between genetic propensity for education and neighborhood civic norms and SES when it comes to voting in the EU elections: the genetic effects are lower for individuals whose adolescence neighborhood was high in political participation and in SES. This suggests a type of substitutability mechanism. No clear evidence of interaction is found with regards to voting in national elections. There is also limited evidence that genetic effects on non-electoral participation are higher for the very same individuals—a fact that may suggest different gene-environment interaction mechanisms depending on the form of participation.
Theme: Pandemic research
Chair: Jonas Björk, Department of Laboratory Medicine, Lund University
Oral session 4.1
Title: Prevalence and Risk Factors of Post-Covid Conditions among people with intellectual disability in Sweden: what we should know
Speaker: Ailiana Santosa, School of Public Health and Community Medicine, University of Gothenburg
Background: People with intellectual disability (PwID) have been disproportionately affected by the COVID-19 pandemic, thus are at greater risk of post-COVID condition (PCC). Studies of PCC on PwID, however, are lacking. This study aims to estimate the prevalence and risk factors associated with PCC among PwID in Sweden.
Methods: Individuals with intellectual disabilities (ICD-10 F70–73, F78–79, F84, F88–89, Q90) living in the two largest Swedish regions (Västra Götaland and Stockholm) on 1 Jan 2020 were identified from national registers and primary healthcare databases. PCC was defined as a primary or secondary diagnosis ICD-10 U09.9, with the first event 28 days after infection. We described sociodemographics, comorbidities, and prescription medications among PwID with and without PCC.
Results: Among 83290 PwID (mean age 37.2±12.9), 247 (0.3%) had been diagnosed with PCC. PwID with PCC, compared to PwID without PCC, were more female-dominated (59.1%vs.39.4%), received at least two doses of COVID-19 vaccine (71%,61%), and had higher prevalence of thyroid disease (10.9%,4.8%), obesity (15.8%,8.4%), hearing loss problems (8.5%,5.8%), hypertension (13.4%,4.8%), generalized anxiety disorder (8.9%,3.9%), depression (23.5%,9.6%), PTSD (9.7%,2.6%), bipolar disease (6.5%,1.9%) and eating disorder (4.9%,1.9%). Furthermore, PwID with PCC were more likely to take prescription medicines, including antidepressants, neuroleptics, tranquilizers, and antieileptics.
Conclusion: The findings provide insight into how COVID-19 affects PwID, including comorbidities, vaccinations, and treatment, despite the low PCC occurrence among PwID.
Oral session 4.2
Title: Population-wide Long-Term Effects of Covid-19 Infections Asserted with Serology Testing
Speaker: Nicholas Baltzer, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
The COVID-19 pandemic caused significant upheaval in healthcare consumption and preparedness, with many questions regarding how long it would tax the healthcare systems. We aimed to investigate if there was an increased long-term healthcare usage in individuals with a SARS-CoV-2 infection during 2020.
We followed 365,354 individuals from Stockholm county, serology-tested in 2020, using the SARS-CoV-2 Quality Registry linked to the Stockholm regional healthcare data warehouse. We extracted their medical history up until October 2022. The sero-positive and sero-negative individuals were matched 1:1 on age, sex, 2019 medical history, and date of last serology, and we compared their differences in healthcare usage over 2021/22. The differences were estimated as Incidence rate ratios (IRR), computed as a quasi-Poisson logistic regression with robust standard errors, while the differences over time were estimated as a difference-in-differences model with matched baseline healthcare consumption from 2019.
Of the 365,354 individuals, 272,918 were sero-negative in 2020 and 73,814 were sero-positive. In preliminary analyses, the IRRs were 1.0 for healthcare usage in primary care, 0.96 for outpatient specialist care, and 0.98 for inpatient care during the follow-up period of 2021/22. Difference-in-differences plotting showed no substantial differences in healthcare use during the follow-up time, ranging from 0.01 to -0.01. We observed an increased healthcare usage in patients hospitalized for COVID-19 in 2020 unlike non-hospitalized.
No significant differences in healthcare consumption were found between serology-negative and serology-positive individuals on the population level. The COVID-19 pandemic has had limited impact on the long-term healthcare usage in individuals infected with SARS-CoV-2.
Oral session 4.3
Title: Smoking status and COVID-19 related risk in four Swedish non-communicable disease registries, a non-intuitive association?
Speaker: Brian Kirui, School of Public Health and Community Medicine, University of Gothenburg
Rationale: The impact of smoking on COVID-19 has been controversial due to conflicting evidence on both incidence and severity.
Objectives: To investigate the effect of smoking on COVID-19 infection, hospitalization, intensive care unit (ICU) admission, and death.
Methods: We included subjects in the Swedish population with registered smoking information during 2019 from four Swedish national chronic disease quality registers: Airways, Diabetes, Heart failure and Cardiac disease. Four COVID-19 outcomes were captured from 1 Jan 2020 until 31 Dec 2021. Using logistic regression, we estimated adjusted odds ratios (aOR) with 95% confidence intervals (95%CI) between smoking status and COVID-19 outcomes, comprehensively controlling for relevant confounders, for each register and pooled across registers.
Results: The combined cohort of the registries comprised 506755 subjects, 13.0% current, 32.2% former and 54.7% never smokers. During follow-up, there were 10.1% COVID-19 infections, 2.5% hospitalizations, 0.3% ICU admissions, and 0.5% COVID-19 deaths. Compared to never-smokers, current smokers had a lower risk of COVID-19 infection (aOR 0.66, 95%CI 0.64-0.68), hospitalization (0.61, 0.57-0.66), ICU admission (0.41, 0.33-0.50), and death (0.82, 0.69-0.97), while former smokers had a significantly higher risk of COVID-19 death (1.21, 1.11-1.33). Similar results were seen in separate analysis within each register.
Conclusion: Explaining the findings presents complexities. Theories like nicotine's anti-inflammatory effects have been debunked, and nitric oxide's antiviral role remains unsubstantiated. Despite our rigorous efforts to control for confounding, the results for current smokers persist. Selection bias or other methodological issues are further potential explanations that will be presented and discussed.
Oral session 4.4
Title: Fast-paced register-based research during a health crisis
Speaker: Jonas Björk, Department of Laboratory Medicine, Lund University
Register-based research during the recent COVID-19 pandemic contributed with answers to urgent societal questions. This research could promptly establish main risk factors for severe disease and identify especially vulnerable population groups. It highlighted sociodemographic differences in disease but also in health seek behavior, testing, and subsequently in vaccination uptake. Register-based research allowed for longer, population-wide follow up of vaccine effectiveness and safety than the restricted and short-term randomized clinical trials that preceded approval of the different vaccines. Waning effects and varying effectiveness across virus variants of concern were also monitored, and effects of public health interventions evaluated. In this review, I will give examples of successful population studies linked with register data that were conducted during the pandemic. I will discuss key elements of these studies that made them possible to conduct rapidly and with high quality. I will also discuss some lessons learnt from research during a pandemic, and structural infrastructure changes that should be considered in order to improve disease surveillance and preparedness for future health crises that are bound to occur again, sooner or later.
Oral session 4.5
Title: The National Quality Registry for SARS-CoV-2
Speaker: Sara Nordqvist-Kleppe, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
The COVID-19 pandemic led to an unprecedented global collaboration towards identification, disease tracing, containment, protection, and eradication, of the SARS-CoV-2 pathogen. Collation of data from different sources is important for the healthcare sector to adequately respond to a pandemic. To quality assure SARS-CoV-2 testing and treatment in Sweden, the National Quality Registry for SARS-CoV-2 (COVID-19) (NKCOV) was founded.
NKCOV currently collects PCR-, serology-, treatment-, and antigen-, data from laboratories, with the intent to cover every lab and region that handles SARS-CoV-2 testing. Data is currently available from Stockholm region and Dalarna, with more pending.
NKCOV holds 3,337,171 PCR tests, of which 82.6% are negative, 16.6% are positive, and 0.8% are inconclusive. 98.3% of tests were drawn from the upper respiratory tract, and 99.3% of test results are qualitative. The sample dates are evenly distributed over the wild-type, alpha, delta, and omicron waves. NKCOV further holds 654,241 serology tests, of which 70.2% are negative, 25.3% are positive, and 4.5% are inconclusive. 96.1% of results are qualitative, with 86.4% of tests targeting the nucleocapsid, 7.7% targeting both nucleocapsid and spike protein, and 5.9% targeting only the spike protein. 80.2% of the test samples were drawn during the wild-type wave, and 15.8% during the alpha wave. Further results will be presented during the conference.
The NKCOV registry is operational, covering Stockholm and Dalarna to date, with data available upon request for research. We are working to expand to other regions and to collect data on treatments for patients with COVID-19.
12:00 – 13:00
13:00 – 13:50
Petter Lundborg, Professor at the Department of Economics and the Centre for Economic Demography, Lund University
Title: “Using Quality Registers for Social Science Research: Examples from the Danish IVF register”
13:50 – 14:20
14:20 – 15:50
Theme: Long-term and life-course studies
Chair: Helena Backman, Public Health and Clinical Medicine, Umeå University
Oral session 5.1
Title: The interrelation between unemployment and overqualification among second generation and ancestral population in Sweden
Speaker: Wooseong Kim, Department of Sociology, Stockholm University
Previous studies found that the second generations––the children of immigrants born in the host country––show higher risks of both overqualification and unemployment compared to natives with two native parents. Our study contributes to labor market and immigrant integration literatures by investigating whether there is an interrelation between overqualification and unemployment such that it generates a vicious circle of accumulating labor market disadvantages over career, and if this is stronger among the second generations. Specifically, we address two following questions: how does past overqualification/unemployment affect current unemployment/overqualification, and how do they vary between ancestral Swedes and 9 second generation origin groups over career? We use dynamic correlated random-effects multinomial models to estimate probabilities of transiting from past to current employment statuses. We use Swedish population register data to follow individuals over their early working careers. Our results show that past overqualification and unemployment increase the risk of current unemployment and overqualification among second generations and ancestral Swedes. However, ancestral Swedes are more likely to experience transient disadvantages due to higher probabilities of transitioning to matched employment, while the second generations are more likely to fall into or remain in unemployment. Higher transition risks from overqualification to unemployment, together with a higher unemployment persistence, suggest that second generations experience more severe entrapment into unemployment compared to ancestral Swedes. Furthermore, contrary to the career mobility hypothesis, the probability of transiting to a matched job is higher among the unemployed than among the overqualified for all ancestry groups.
Oral session 5.2
Title: Level of education modifies asthma mortality in Norway and Sweden - The Nordic EpiLung Study
Speaker: Helena Backman, Public Health and Clinical Medicine, Umeå University
Background: The relationship between socioeconomic status (SES), asthma and mortality is complex and multifaceted, and it is not established if educational level modifies the association between asthma and mortality.
Aim: To study that association between asthma and mortality in Sweden and Norway, and to what extent educational level modifies this association.
Study Design and Methods: Within the Nordic EpiLung Study, >56,000 individuals aged 30-69 years participated in population-based surveys on asthma and associated risk factors in Sweden and Norway during 2005-2007. Data on educational level and 10-year all-cause mortality were linked by national authorities. The fraction of mortality risk attributable to asthma was calculated, and Cox regression was used to estimate hazard ratios (HR) and 95% confidence intervals (95%CI) for mortality related to asthma, stratified by educational level.
Results: In total, 5.5% of all deaths was attributed to asthma. When adjusted for potential confounders, the HR for mortality related to asthma was 1.71 (95%CI 1.52-1.93). Those with primary level of education had higher hazard of all-cause death related to asthma than those with tertiary level (HR 1.80, 95%CI 1.48-2.18, vs. HR 1.39, 95%CI 0.99-1.95).
Conclusion: Asthma was associated with an overall 71% increased all-cause mortality and 5.5% of deaths can be attributed to asthma. Educational levels modified the risk of mortality associated with asthma, with the highest risk among those with primary education.
Oral session 5.3
Title: Sequence analysis of sickness absence and disability pension states among privately employed white-collar workers in Sweden; a prospective cohort study
Speaker: Kristin Farrants, Division of Insurance Medicine, Karolinska Institutet
Background: There is very little scientific knowledge about sickness absence (SA) and disability pension (DP) among privately employed white-collar workers.
Aim: We aimed to gain knowledge about future patterns of SA/DP over time in a cohort of the privately employed white-collar workers in Sweden.
Methods: A 7-year prospective cohort study, using microdata linked from held by Statistics Sweden, the Social Insurance Agency, and the Board of Health and Welfare. We analysed information for all the 1,283,516 privately employed white-collar workers in 2012 in Sweden, using sequence analysis to describe the future annual states of SA (in spells >14 days) and DP, and multinomial logistic regression to analyse the odds of sociodemographic variables belonging to each observed cluster.
Results: We identified five clusters of future sickness absence and disability pension patterns during follow-up: (1) ‘low or no SA/DP’ (88.7% of all), (2) ‘SA due to somatic diagnosis’ (5.2%), (3) ‘SA due to mental diagnosis’ (3.4%), (4) ‘not eligible for SA/DP’ (1.4%), and (5) ‘DP’ (1.2%). Men, highly educated, high-income earners, and those working in industry or service sector were more likely to belong to the first and the fourth cluster. The second, third, and fifth clusters consisted mainly of women, low educated, low-income earners, those working in education, care, nursing, and social services sectors.
Conclusion: In general, the absolute majority of the privately employed white-collar workers had sequences without SA/DP during the 7-year follow-up. The risk of belonging to a cluster characterised by having SA/DP varied by sociodemographic factors.
Oral session 5.4
Title: Life expectancy of gastric cancer survivors compared to the background population
Speaker: Wille Leijonmarck, Department of Molecular medicine and Surgery, Karolinska Institutet
Background: Life expectancy among 5-year gastric cancer survivors compared to the background population is unknown and needs to be clarified for effective monitoring strategies and evidence-based information to patients.
Methods: Nationwide population-based cohort study including all gastric adenocarcinoma patients who had survived at least 5 years following gastrectomy from 2006-2015 in Sweden, with follow-up during 6-10 postoperative years. The observed survival in this cohort was divided by the expected survival, retrieved from the entire Swedish population of the corresponding age, sex, and calendar year, yielding relative survival rates with 95% confidence intervals (CIs) using the life time table method. Stratified analyses were performed by categories of age, sex, calendar year, education level, comorbidity, neoadjuvant therapy, tumour sub-location, and pathological tumour stage.
Results: Among 767 patients who met the inclusion criteria, the relative survival was reduced among gastric cancer survivors compared to the corresponding population, and the difference increased in magnitude for each later year (from 97.3% [95% CI: 95.4%-99.1%] year 6 to 86.6% [95% CI: 82.3%-90.9%] year 10). The decline in relative survival tended to be more pronounced among patients who underwent gastrectomy in earlier calendar years, had fewer years of formal education, had more comorbidities, did not receive neoadjuvant therapy, and had more advanced pathological tumour stage.
Conclusion: Gastric adenocarcinoma survivors have a reduced life expectancy compared to the corresponding background population, and more so in certain subgroups. This finding underscores the relevance of closer monitoring of these patients.
Theme: Fertility research
Chair: Paul Dickman, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
Oral session 6.1
Title: The Power of the Pill: Evidence from Oral Contraceptive Sales
Speaker: Kelly Ragan, Stockholm School of Economics
The introduction of oral contraception (’the pill’) was a watershed in fertility control. Yet, the pill’s impact on childbearing has been hard to establish due to limited data. I exploit rich data on pill sales across markets and time to characterize how Swedish teens’ access to pharmacies shaped pill sales and fertility after the pill’s introduction. Pill sales are highly sensitive to pharmacy access, and teen fertility responds strongly to pill sales. The unique data establish an important link in the causal chain from pill access to women’s fertility. Estimated fertility responses to the pill could explain the halving of teen births after the pill’s introduction, a finding which is robust to accounting for trends in female education, abortion provision, and other factors.
Oral session 6.2
Title: Do depression and loneliness affect fertility intentions and their realization? Evidence from Norway and Sweden
Speaker: Erik Carlsson, Demography Unit, Stockholm University
Depression and loneliness are widespread problems in modern European and North American societies. Although there are several reasons to expect mental health to affect fertility, there is so far little research on the topic. Earlier research has primarily explored the opposite direction of causality, i.e. how fertility affects mental health. This study examines how depression and loneliness are associated with the propensity to report a positive short-term fertility intention, the propensity to realize a positive fertility intention, and the propensity to abandon rather than postpone a positive fertility intention. The study data come from the Norwegian and Swedish Generations and Gender Surveys from 2007/2008 and 2012/2013, together with their population register-based follow-ups. The follow-up data to the Norwegian GGS stretches until 2011, while the follow-up data to the Swedish GGS stretches until 2021. We measure depression by using the shortened version of the Center for Epidemiologic Studies Depression Scale, and measure loneliness by using the De Jong Gierveld short scale for emotional and social loneliness. Results show that depression is negatively associated with the propensity to report a positive fertility intention among Norwegian women and that both depression and loneliness are negatively associated with the propensity to realize a positive fertility intention among Swedish women. To our knowledge, this study is the first to show that depression and loneliness may affect fertility intentions and their realization.
Oral session 6.3
Title: Do Hormonal Contraceptive Sales Influence Mortality and Suicide Among Teen Girls? Evidence from the Introduction of the Pill in Sweden
Speaker: Kelly Ragan, Stockholm School of Economics
The introduction of oral contraception (‘the pill’) was a watershed in fertility control which was quickly adopted by Swedish women. Yet, recent medical studies have suggested that the use of combined hormonal contraception substantially increases suicide risk among teenage girls. I exploit the pill’s introduction as a type of ‘natural experiment’ and couple this with rich date on pill sales across markets and time to quantify the pill’s spread and investigate the relationship between pill sales and mortality among 800,000 Swedish teens. The Swedish setting is well suited for study since young women’s access to contraceptive services were largely unrestricted and pill use heavily skewed toward teenagers and women in their early twenties. Quarterly data on hormonal contraceptive sales across 70 local markets provide a rich source of variation in exposure to the pill from which to link individual case of death data. Although the hormonal contraceptives sold during this period were many times stronger than the low dose contraceptives studied in recent medical literature, the data do not support the view that the pill’s diffusion led to a significant increase in suicide among young women in Sweden.
Oral session 6.4
Title: Ethnic fertility differentials among Yugoslavian-born immigrants in Sweden
Speaker: Erik Carlsson, Demography Unit, Stockholm University
Many immigrants to North America and Western Europe originate from countries where fertility patterns differ considerably between ethnic groups. Yet, earlier research on immigrant fertility in Western destination countries typically does not distinguish among immigrants’ origin at a finer level than country of birth, an approach that risks hiding variation between sub- and transnational ethnic groups. This study uses Swedish population register data on the so-called information language of newly arrived immigrants to distinguish between BCMS-speaking (Bosnian/Croatian/Montenegrin/Serbian) and Albanian-speaking women immigrating to Sweden from the former Yugoslavia during 1992-2001. To my knowledge, this study is the first to use data on immigrants’ information language to identify subnational ethnic groups. The study uses event-history analysis to compare parity-specific transition rates of about 29,000 Yugoslavian-born women and 2.4 million native-born women from 1992 to 2017. Results show that Albanian-speakers have higher first-, second-, and third-birth transition rates than both BCMS-speakers and native women. The transition rates of BCMS-speakers are considerably closer to those of natives. Results point to partial adaptation towards native fertility patterns by duration of stay within the G1 and between the G1 and G1.5 among both BCMS-speakers and Albanian-speakers. The study contributes to the understanding of fertility patterns among Yugoslavian migrants, which is one of the largest immigrant origin groups in several European destination countries. The study also makes a general contribution to research on immigrant fertility in demonstrating that distinguishing among immigrants from the same origin country by subnational ethnicity can uncover considerable within-group heterogeneity.
Theme: Miscellaneous topics in register-based research
Chair: Anita Berglund, Institute of Environmental Medicine, Karolinska Institutet
Oral session 7.1
Title: Cholinesterase inhibitors and non-steroidal anti-inflammatory drugs and the risk of peptic ulcers: a self-controlled study
Speaker: Máté Szilcz, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
BACKGROUND: Non-steroidal anti-inflammatory drugs (NSAIDs) should be used with caution in older adults. Their gastrointestinal adverse event risk might be further reinforced when using concomitant cholinesterase inhibitors (ChEIs). We aimed to investigate the association between NSAIDs and ChEI use and the risk of peptic ulcers in older adults.
METHODS: Register-based self-controlled case series including adults ≥65 years with new prescriptions of ChEIs and NSAIDs, diagnosed with incident peptic ulcer in Sweden, 2007-2020. We identified persons from the Total Population Register individually linked to several nationwide registers. We estimated the incidence rate ratio (IRR) of peptic ulcer with a conditional Poisson regression model for four mutually exclusive risk periods: use of ChEIs, NSAIDs and the combination of ChEIs and NSAIDs, compared with the non-treatment in the same individual. Risk periods were identified based on the prescribed daily dose, extracted via a text-parsing algorithm, and a 30-day grace period.
RESULTS: Of 70,060 individuals initiating both ChEIs and NSAIDs, we identified 1500 persons with peptic ulcer (median age at peptic ulcer 80 years). Compared with the non-treatment periods, the risk of peptic ulcer substantially increased for the combination of ChEIs and NSAIDs (IRR: 9.0, [6.8-11.8]), more than for NSAIDs alone (5.2, [4.4-6.0]). No increased risk were found for the use of ChEIs alone (1.0, [0.9-1.2]).
DISCUSSION: The risk of peptic ulcer was associated with the concomitant use of NSAIDs and ChEIs, over and beyond the risk of NSAIDs, underlining the importance of carefully considering the risk of peptic ulcers when co-prescribing these drugs.
Oral session 7.2
Title: Using fine-grained population-based grid data to assess genetic self-selection into residential neighborhoods across the life-span
Speaker: Rafael Ahlskog, Department of Government, Uppsala University
Gene-environment correlation, or rGE, is known to cause confounding in conventional genetic discovery studies. One way that rGE can come about is through self-selection into environments. In this study, we leverage a large sample of genotyped Swedish dizygotic twin pairs, coupled with population-wide register data, to investigate genetically driven self-selection into residential neighborhoods. We use grids as small as 250x250 meters to construct highly fine-grained neighborhood characteristics. The high level of geographic granularity makes it possible to distinguish not just regions or districts, but e.g. higher or lower SES neighborhoods in the same area. Furthermore, access to high-quality population-wide data for a large selection of variables means that we can construct neighborhood characteristics not just based on education or income, but on things like occupational status, political participation and ethnic diversity as well as anthropometric traits like height or BMI.
Using these fine-grained neighborhood characteristics, we can investigate how within-pair differences in polygenic indices for a range of traits are related to self-selection into residential neighborhoods during different periods of the life cycle. This allows us to look at not just the existence of, but also the specific character of, self-selection as a driver of gene-environment correlation. Preliminary results show that the twin with e.g. a higher polygenic index for educational attainment is causally more likely to self-select into a high-SES neighborhood when moving out of their childhood home, and that this social stratification persists both over the life-cycle and across generations.
Oral session 7.3
Title: Trajectories of physical activity before and after a cancer diagnosis in persons with type 2 diabetes
Speaker: Emerald Heiland, Department of Surgical Sciences, Uppsala University
Background: We identified physical activity (PA) trajectories pre- and post-cancer diagnosis in persons with type 2 diabetes, explored trajectory determinants, and probability of post-trajectory membership depending on pre-cancer PA. Methods We included 26,258 persons with diabetes, who had a cancer diagnosis (from January 1, 2004). Data were from the Swedish National Diabetes Register and National Cancer Register. Annual (2000-2021) self-reported PA (frequency of walking 30min or equivalent) was used, excluding the year immediately before and after cancer diagnosis. Group-based trajectory modelling and multivariable logistic regression were used to identify the trajectories and assess potential determinants, respectively.
Results: Seven trajectories emerged pre-diagnosis: average maintainers (50.0%), high maintainers (12.3%), increasers (11.1%), very high maintainers (11.0%), decreasers (6.3%), low maintainers (6.0%), and low decreasers (3.3%). Six similar trajectories emerged post-diagnosis. Pre-diagnosis decreasers, compared with average maintainers, were more likely to be older, women, single, with lower education, higher HbA1c and body mass index (BMI), and smokers. Increasers were more likely to be older, men, married/cohabiting, low HbA1c and BMI, and non-smokers. Those who were in the low decreases PA pre-cancer diagnosis trajectory were more likely to be in the high average maintainers after diagnosis. Discussion Most patients were in the average or high PA groups before and after a cancer diagnosis. PA behaviour may depend on sociodemographic factors, glycemic control, and body composition. PA may improve after a cancer diagnosis even if low prior to the diagnosis. Conclusion Register data can provide high resolution, long-term information to answer research questions in innovative ways.
Oral session 7.4
Title: Antireflux surgery versus antireflux medication and risk of esophageal adenocarcinoma in patients with Barrett’s esophagus
Speaker: Dag Holmberg, Department of Molecular medicine and Surgery, Karolinska Institutet
Background and aims: Antireflux treatment is recommended to reduce esophageal adenocarcinoma in patients with Barrett’s esophagus. Antireflux surgery (fundoplication) counteracts gastroesophageal reflux of all types of carcinogenic gastric content, and reduces esophageal acid exposure to a greater extent than antireflux medication (proton pump inhibitors). We examined the hypothesis that antireflux surgery prevents esophageal adenocarcinoma to a larger degree than antireflux medication in patients with Barrett’s esophagus.
Methods: This multi-national and population-based cohort study included all patients with a diagnosis of Barrett’s esophagus in any of the national patient registries in Denmark (2012-2020), Finland (1987-1996 and 2010-2020), Norway (2008-2020), or Sweden (2006-2020). Patients who underwent antireflux surgery were compared with non-operated patients using antireflux medication. The risk of esophageal adenocarcinoma was calculated using multivariable Cox regression, providing hazard ratios (HR) with 95% confidence intervals (CI) adjusted for age, sex, country, calendar year, and comorbidity.
Results: The cohort consisted of 33,939 patients with Barrett’s esophagus. Of these, 542 (1.6%) had undergone antireflux surgery. During up to 32 years of follow-up, the overall HR was not decreased in patients having undergone antireflux surgery compared to non-operated patients using antireflux medication but rather increased (adjusted HR 1.9, 95% CI 1.1-3.5). Additionally, HRs did not decrease with longer follow-up, but instead increased for each follow-up category, from 1.8 (95% CI 0.6-5.0) within 1-4 years of follow-up to 4.4 (95% CI 1.4-13.5) after 10-32 years of follow-up.
Conclusion: Patients with Barrett’s esophagus who undergo antireflux surgery do not seem to have a decreased risk of esophageal adenocarcinoma compared to those using antireflux medication.
16:00 – 17:00
Topic: “How to make Sweden a paradise for register-based research”
Panel: Catarina Almqvist Malmros (Karolinska Institutet), Per Bergstrand (the Swedish Research Council), Thomas Frisell (Karolinska Institutet), Petra Hasselqvist (Swedish Association of Local Authorities and Regions)
Moderators: Sven Oskarsson and Anita Berglund
Title: Weight trajectories through adulthood and prostate cancer incidence, aggressiveness, and death: The Obesity and Disease Development Sweden (ODDS) pooled cohort study
Speaker: Marisa Da Silva, Department of Translational Medicine, Lund University
Introduction: There is inconclusive evidence whether body fatness is associated with prostate cancer (PCa) incidence but convincing evidence for a positive association with PCa death. Body fatness is often measured as body mass index at one point in time. Instead, we investigated weight trajectories throughout adulthood and in age spans.
Methods: The Obesity and Disease Development Sweden (ODDS) study is a pooled cohort with ~4 million participants from 17 cohorts linked to national registries. We sampled 258,848 men, with minimum three weight observations between age 17-60, enrolled in 1963-2014, and followed up through 2019. We used a linear mixed effects model, with and without linear age splines, to estimate individual weight trajectories (intercept and slope) for inclusion in a multivariable-adjusted Cox model.
Results: During median 43 years of follow-up, 23,348 participants were diagnosed with PCa at mean age of 70 (SD=8) years. The hazard ratio of weight trajectory (quintile 1 vs. quintile 4) and aggressive PCa was 1.09 (95% CI: 1.00-1.19) and 1.31 (95% CI: 1.18-1.45) for PCa death. The association between weight trajectory (1kg/year) and PCa (aggressive and death) was strongest in the youngest age span (17 to <30) and strong in the oldest (45-60) for PCa death. In the case-only analysis the association with PCa death was strongest in the oldest age span.
Conclusion: Our findings suggest that preventing rapid weight increase that commonly occurs in young adulthood may reduce the risk of PCa incidence and death, while preventing rapid weight increase in old adulthood may improve PCa survival.
Title: Collider stratification bias as a potential explanation for the inverse obesity paradox in prostate cancer - a simulation analysis
Speaker: Josef Fritz, Department of Translational Medicine, Lund University
Collider (stratification) bias is often brought up as a potential explanation when the association of obesity with disease diagnosis is of different magnitude than with disease outcome. It has been well studied in instances where obesity, despite increasing disease risk, is associated with improved disease survival (“obesity paradox”). For prostate cancer (PCa), an “inverse” form of the obesity paradox has consistently been reported, including analyses of large Swedish cohort data linked to the National Prostate Cancer Register; body mass index (BMI) is negatively associated with localized PCa diagnosis (hazard ratio [HR] ~0.9 per 5-kg/m2 increase), but positively associated with PCa-specific death (HR~1.2) in localized PCa cases. Collider bias as a potential explanation in this setting is unexplored. Using simulations, we demonstrate that collider bias is unlikely to distort the PCa death HR by more than ±0.02, which is mainly due to the modest association of BMI with localized PCa risk. A substantially stronger association with PCa diagnosis, i.e. HR≤~0.6-0.7, would be needed to introduce relevant collider bias, in addition to the existence of strong unmeasured risk factors for both PCa diagnosis and death (which could be genetic risk, for example). Collider bias is an insufficient explanation for the inverse obesity paradox in localized PCa. Even in large observational studies, random variability in effect estimates outweighs the maximum magnitude of collider bias. Classical confounding and other sources of biases bear a higher potential of distorting observed associations between BMI and PCa death than collider bias.
Title: Assessing RETTS Accuracy within the Emergency Care System using the Lund Outcome Set for Evaluation of Triage Standard: Establishing a Benchmark for Triage
Speaker: André Johansson, Faculty of Medicine, Lund University
Background: Most emergency departments [ED] in Sweden use the Rapid Emergency Triage and Treatment System [RETTS] to prioritize patient care. The validation of RETTS remains an unsolved problem beyond studies using singular outcome measures with limited ability to capture the complexities of triage. Lund Outcome Set for Evaluation of Triage [LOSET] is a newly developed multi-modal standard for triage validation. It encompasses a range of clinically important outcomes including interventions, lab results, specific types of admissions, mortality, and diagnoses individually for red (highest) and orange (second highest) priority.
Methods: We will source data from Region Halland, encompassing nearly 400,000 adult patient visits to two EDs from 2017 to 2023, and a secondary dataset from Region Skåne covering 500 ED visits from 2017 and 2018. Patient outcomes will be labelled using the LOSET standard, relying on interventions, lab results and admission in the ED and final diagnosis in either the ED or the corresponding ward post-admission. We will manually review a subset of the data to ensure alignment between final ward diagnoses and initial ED presentations. Balanced accuracy [(sensitivity + specificity) / 2] will be our primary metric to evaluate RETTS' proficiency in identifying red and orange priorities. Sub-group analysis will be made on age groups, gender, versions of RETTS and periods marked by the Covid-19 pandemic.
Expected results: We anticipate highlighting areas where RETTS is both accurate and might need improvement. Additionally, we aim to offer methodological insights to assist future triage system validation efforts both nationally and globally.
Title: Medium-to-long term effects of maternal COVID-19 infection during pregnancy on the child: a register-based study
Speaker: Huiqi Li, University of Gothenburg
Maternal COVID-19 infection during pregnancy (MCIP) is considered a major event and may lead to adverse effects of the development of the child, prenatally and postpartum. Previous studies found MCIP associated with preterm birth and low birth weight. Due to the short follow-up available after the pandemic, it is unknown if MCIP influences medium-to-long term childhood development, for example height and weight development, and language development. We aim to evaluate such effects and assess the mediating factors, including preterm birth, low birth weight, difficulty in breastfeeding, and postpartum depression.
We will obtain data on children’s development from the Child Health Services Quality Register (Barnhälsovårdsregistret, BHVQ) and link it to the established SCIFI-PEARL (Swedish COVID-19 Investigation for Future Insights – a Population Epidemiology Approach using Register Linkage) project, which includes data from various national and regional registers. The rich data on maternal status from National Medical Birth Register (Medicinska födelseregistret) and the Swedish Pregnancy Register (Graviditetsregistret), data on SARS-CoV-2 test-positive results from SmiNet, and data from National Patient Register, would enable us to identify children exposed to MCIP of varied severity and a wide range of covariates of interest. We will also consider vaccination status of the mother and SARS-CoV-2 variants of concern during pregnancy.
To date, usage of the data from BHVQ have been limited. For this study, most data are already in-house, and linked BHVQ data will be delivered by mid-October 2023. Analyses have been initiated, and preliminary results will be shared on the Swedish Register-Based Research Summit in November.
Title: Andas Sverige and Breathe Sweden: current status and plans ahead
Speaker: Sanna Kjellberg, School of Public Health and Community medicine, University of Gothenburg
Background: “Andas Sverige” (andassverige.se) and “Breathe Sweden” (breathesweden.com) are recently launched web platforms aimed at: (1) facilitating interactive communication between respiratory researchers and the public, and (2) enhancing collaboration both within and outside the Swedish respiratory research community.
Results: Initially based on a collaboration between the West Sweden Asthma Study and the Obstructive Lung Disease in Northern Sweden study, eight other studies across Sweden have now joined the platforms, including both pediatric and adult studies. More studies are in the processes of joining or being invited. A steering committee has been established to lead the activities of the platforms, with one representative from each study. An easily accessible database comprising metadata for each study is being assembled and will be published on the “Breathe Sweden” platform in the autumn of 2023 to enhance cohort-discovery and collaboration between study groups. Plans are in place to prepare joint grant applications that will, where possible, include cross-cohort data pooling or analyses. The first annual in-person research meeting is scheduled for Spring 2024, where outputs from each study and the consortium will be presented by junior and senior members of each cohort, with opportunities created for more collaborations.
Future perspectives: The platforms are constantly evolving to enhance utility. Our future plans include increasing engagement on “Andas Sverige” by communicating ongoing research to the broader public through animations and storytelling presentations. Additionally, we are planning future implementations of the platform concept in other Nordic countries, such as creating a “Breathe Norway” and “Breathe Finland” platform.
Title: Exposure to air pollution during pregnancy and risk of premenopausal breast cancer - a Swedish nation-wide cohort study including more than 1 million women
Speaker: Jessica Edlund, Department of Radiation Sciences, Umeå University
Background: Air pollution has been linked to breast cancer risk, but previous studies have seldom considered specific exposure windows, such as pregnancy. During pregnancy the breast undergoes substantial changes and exposures may have a stronger impact than if they occurred at other time-periods.
Aim: To identify associations between air pollution exposure during pregnancy and risk of premenopausal breast cancer. Method: Using nation-wide data from Swedish registers we constructed a cohort consisting of all women in Sweden giving birth to their first child between 1991 and 2015 (excluding women with a previous cancer diagnosis). Data on air pollution (NO2, PM10 and PM2.5), was based on the women’s home residences, both during and between pregnancies. Associations between air pollution and premenopausal breast cancer was evaluated using Cox proportional hazards regression to estimate adjusted hazard ratios (HR), using age as the time scale.
Results: The full cohort included 1 016 980 women and 2 102 778 pregnancies. During the study period, 12 176 (1.2 %) were diagnosed with breast cancer by age 50. A 10 unit increase in NO2, PM10 and PM2.5, (during a woman’s first pregnancy), was associated with premenopausal breast cancer, HR = 1.04 (95% CI = 1.01, 1.07), HR = 1.08 (95% CI = 1.04, 1.13) and HR = 1.11 (95% CI = 1.05, 1.18), respectively.
Conclusion: Analyses of air pollution levels (NO2, PM10 and PM2.5) during the first pregnancy suggests an association with premenopausal breast cancer. Additional analyses considering time between pregnancies are currently underway.
Title: A genetically informed study of residential mobility behaviour in relation to neighbourhood ethnic diversity
Speaker: Qinya Feng, Department of Government, Uppsala University
Do native individuals relocate in response to the presence of ethnic minorities and immigrants in their local area? Despite extensive research on the impact of ethnic diversity on socio-political attitudes, there is a lack of systematic investigation into the actual residential mobility behaviour of native people concerning social diversity and its underlying factors. Recent studies also highlight the potential genetic influence on both residential environments and mobility behaviour, raising concerns about genetic confounding in the relationship between neighbourhood characteristics and mobility behaviour. To address these gaps, this project employs population-wide Swedish register data spanning from 1990 to 2019 and Swedish twin data to explore how micro-contextual ethnic diversity influences the residential mobility behaviour of native individuals. The first part of the project focuses on identifying yearly geolocation changes and constructing personalised micro-contexts, often referred to as "bespoke neighbourhoods." It measures various contextual characteristics for each individual. The second part of the project examines whether changes in micro-context ethnic diversity prompt individuals to relocate to neighbourhoods with lower ethnic diversity. In addition to the population-wide analysis, this section leverages a unique Swedish twin sample linked to molecular genetic data and employs polygenic indices to investigate the role of genetics in the studied relationship. The project will shed new light on the behaviour that shapes individuals’ living environments and can also be politically consequential in contemporary society.
Title: Risks and Biases in Data Linkage Consent: Evidence from Survey Experiments
Speaker: Elis Carlberg Larsson, Institute for Analytical Sociology, Linköping University
In this article, we investigate the risks and biases arising from requesting data linkage consent. Specifically, we examine how asking respondents to consent to linking their survey responses to data from administrative records affects their future survey participation. We also investigate the characteristics of individuals who consent to data linkage. To examine these aspects, we employ an iterative experimental survey design. More precisely, data linkage consent was requested from respondents in randomized sequential batches, enabling the estimation of causal effects of a data linkage consent request on future survey participation. Our analysis of the experimental results reveals that asking for data linkage consent does not influence participation rates in future surveys, and individuals being asked for their data linkage consent exhibit a lower likelihood of deregistering from the survey panel. Additionally, we identify heterogeneous effects on the likelihood of deregistering, where individuals who refused to report educational information are more likely to deregister. Furthermore, we find substantive differences between respondents who provided data linkage consent and those who did not, with regard to various sociodemographic characteristics and political attitudes. We found that education and gender are important predictors of data linkage consent. In terms of political attitudes, supporters of conservative and nationalistic parties are less likely to consent to data linkage compared to their left-wing and liberal counterparts. These findings contribute to our understanding of the risks associated with requesting data linkage consent and shed light on potential selection biases that may impact data quality in data linkage studies.
Title: Religious Pluralism and Religious Decline: Baptist rates
Speaker: Elis Carlberg Larsson, Institute for Analytical Sociology, Linköping University
In religious studies, a long-standing debate revolves around the role of religious pluralism in religious participation. In his seminal book, ”The Sacred Canopy”, Berger argued that as individuals are exposed to more religious world-views, each of those views becomes less plausible, leading to a negative impact on individuals’ religiosity. This argument has been controversial and has been challenged by rational-choice scholars who argue that religious pluralism is likely to enhance the quality of available religious options, thereby increasing individuals’ incentives to participate. However, empirical patterns have shown mixed results, providing inconclusive insight into the role of religious pluralism. Scholars have emphasized the importance of studying local exposure, preferably through social network connections. Nevertheless, due to the rarity of data containing both religious and social network information, such studies are scarce. As a contribution, I leverage panel data covering the entire Swedish population from 2001 to 2017 to this line of inquiry. I employ an inverse probability weighting design to statistically estimate how parents’ decisions to baptize their children are influenced by religious pluralism channeled through their social ties. These weights are estimated using a machine learning algorithm called Bayesian additive regression trees (BART). The results indicate that exposure to religious pluralism leads to a decreased likelihood of baptizing one’s child. Furthermore, I find that parents’ religious composition moderates the extent to which religious pluralism influences their baptism behavior. Additionally, there are substantial heterogeneous effects depending on what social tie exposes one to religious pluralism. In conclusion, I discuss the theoretical implications of these findings and the importance of adopting a social network approach to the study of religion.