Refine
Year of publication
Document Type
- Article (34)
Language
- English (34) (remove)
Has Fulltext
- yes (34)
Is part of the Bibliography
- no (34)
Keywords
- - (9)
- Epidemiology (4)
- Iodine (2)
- Thyroid (2)
- Thyroid-stimulating hormone (2)
- all-cause mortality (2)
- cardiopulmonary exercise testing (2)
- comorbidities (2)
- lung function (2)
- prognosis (2)
Institute
- Institut für Community Medicine (11)
- Kliniken und Polikliniken für Innere Medizin (11)
- Klinik und Poliklinik für Kinder- und Jugendmedizin (3)
- Klinik und Poliklinik für Mund-, Kiefer- und Gesichtschirurgie/Plastische Operationen (3)
- Institut für Diagnostische Radiologie (2)
- Klinik und Poliklinik für Augenheilkunde (2)
- Institut für Biometrie und Medizinische Informatik (1)
- Institut für Klinische Chemie und Laboratoriumsmedizin (1)
- Institut für Med. Biochemie u. Molekularbiologie (1)
- Klinik für Psychiatrie und Psychotherapie (1)
Publisher
- MDPI (6)
- Nature Publishing Group (5)
- S. Karger AG (5)
- Public Library of Science (PLoS) (4)
- Springer Nature (4)
- Wiley (4)
- BioMed Central (BMC) (3)
- BioScientifica (1)
- Upsala Medical Society (1)
Objective
Maternal pre-pregnancy underweight, overweight and obesity might increase the risk for worse short- and long-term outcome in the offspring. There is a need for further study into the relationship between maternal pre-pregnancy body mass index (BMI) and the combined outcome of physical development, state of health and social behavior in children. Question: Is maternal pre-pregnancy BMI associated with the child outcome in terms of physical development, state of health and social behavior (school and leisure time behavior) at the age of 9 to 15 years?
Methods
In the population-based birth cohort study Survey of Neonates in Pomerania (SNIP) children at the age 9–15 years and their families were re-examined by questionnaire-based follow-up. 5725 mother–child pairs were invited to SNiP-follow-up. This analysis is based on the recall fraction of 24.1% (n = 1379). Based on the maternal pre-pregnancy BMI (ppBMI), 4 groups were formed: underweight (ppBMI < 19 kg/m2, n = 117), normal weight (ppBMI 19–24.99 kg/m2, n = 913, reference), overweight (ppBMI 25–30 kg). /m2, n = 237) and obesity (ppBMI > 30 kg/m2, n = 109).
Results
In the multiple regression model, the BMI-z-score for children of mothers in the underweight group was −0.50 lower, and 0.50/1.07 higher in the overweight/obese group (p < 0.001) compared to reference at median age of 12 years. No differences were found in children of underweight mothers with regard to social behavior (interaction with friends and family), school and sports performance (coded from “very good” to “poor”), other leisure activities (watching television, using mobile phones, gaming), and health (occurrence of illnesses) compared to children of normal weight mothers. In contrast, maternal pre-pregnancy overweight and obesity were associated with lower school and sports performance, and higher screen time (smart phone, gaming, television) compared to children of normal weight mothers.
Conclusion
Maternal pre-pregnancy overweight and obesity but not underweight was negatively associated with school performance and leisure time behavior in the offspring at 9–15 years of age.
Aims
Sphingosine-1-phosphate (S1P) is a signaling lipid, which is involved in several cellular processes including cell growth, proliferation, migration and apoptosis. The associations of serum S1P levels with cardiac geometry and function are still not clear. We investigated the associations of S1P with cardiac structure and systolic function in a population-based sample.
Methods and results
We performed cross-sectional analyses of 858 subjects (467 men; 54.4%), aged 22 to 81 years, from a sub-sample of the population-based Study of Health in Pomerania (SHIP-TREND-0). We analyzed the associations of serum S1P with structural and systolic function left ventricular (LV) and left atrial (LA) parameters as determined by magnetic resonance imaging (MRI) using sex-stratified multivariable-adjusted linear regression models. In men, MRI data showed that a 1 µmol/L lower S1P concentration was associated with an 18.1 mL (95% confidence interval [CI] 3.66–32.6; p = 0.014) larger LV end-diastolic volume (LVEDV), a 0.46 mm (95% CI 0.04–0.89; p = 0.034) greater LV wall thickness (LVWT) and a 16.3 g (95% CI 6.55–26.1; p = 0.001) higher LV mass (LVM). S1P was also associated with a 13.3 mL/beat (95% CI 4.49–22.1; p = 0.003) greater LV stroke volume (LVSV), an 18.7 cJ (95% CI 6.43–30.9; p = 0.003) greater LV stroke work (LVSW) and a 12.6 mL (95% CI 1.03–24.3; p = 0.033) larger LA end-diastolic volume (LAEDV). We did not find any significant associations in women.
Conclusions
In this population-based sample, lower levels of S1P were associated with higher LV wall thickness and mass, larger LV and LA chamber sizes and greater stroke volume and work of the LV in men, but not in women. Our results indicate that lower levels of S1P were associated with parameters related with cardiac geometry and systolic function in men, but not in women.
Objective: Menopause is associated with multiple health risks. In several studies, a higher incidence or a higher risk for obstructive sleep apnea (OSA) in post-menopausal than pre-menopausal women is reported. This study was designed to verify such a connection between menopause and OSA in a population-based sample.
Methods: For a subsample (N = 1209) of the Study of Health in Pomerania (N = 4420), complete polysomnography data was available. Of these, 559 females completed a structured interview about their menstrual cycle. Splines and ordinal regression analysis were used to analyze the resulting data.
Results: In the ordinal regression analysis, a significant association between the apnea–hypopnea index (AHI) and menopause indicated that post-menopausal women had a substantially higher risk of OSA. In accordance with previous studies, risk indicators such as body mass index (BMI), age, and the influence of hysterectomies or total oophorectomies were included in the model.
Conclusions: Our results clearly confirmed the assumed connection between menopause and OSA. This is important because OSA is most often associated with male patients, and it warrants further research into the underlying mechanisms.
Purpose
The aim is to investigate the associations of the mother’s socioeconomic and lifestyle factors and life satisfaction with the delivery of a small for gestational age (SGA) infant.
Methods
Data from 4598 participants of the population-based birth cohort study Survey of Neonates in Pomerania (SniP) including comprehensive information on pregnancies, mothers, and their offspring in Western Pomerania, Germany were used in this study. The associations were analyzed using linear and logistic regression models.
Results
After logistic regression analysis adjusted for height of the mother, women who delivered SGA infants, had lower education (p < 0.01) and smoked more frequently during pregnancy (p < 0.01) compared with mothers of adequate for gestational age (AGA) neonates. A mother with less than 10 years of education and one who continued smoking during pregnancy had an odds ratio (OR) of 2.23 [95% confidence interval (CI) = 1.44 to 3.46] and 2.68 (95% CI = 2.06–3.49) of having an SGA infant, respectively. There was no association between the employment of the mother (p = 0.28), the monthly income (p = 0.09), the family status (p = 0.80), the number of friendships outside the household that the mother would not wish to relinquish (p = 0.47), the number of people that she could rely on in case of an emergency (p = 0.75), or alcohol consumption prior to (p = 0.14) or during the pregnancy (p = 0.99) with SGA. Finally, women who delivered SGA infants were more frequently dissatisfied with their employment (p = 0.03) and financial status (p < 0.01).
Conclusions
Women who delivered SGA infants had more associated socioeconomic and lifestyle risk factors and were more frequently dissatisfied with their life conditions than mothers of AGA neonates.
Background and aims
Prevention measures for cardiovascular diseases (CVD) have shifted their focus from lipoproteins to the immune system. However, low-grade inflammation and dyslipidemia are tightly entangled. The objective of this study was to assess the relations between a broad panel of inflammatory biomarkers and lipoprotein subclass parameters.
Methods
We utilized data from the population-based Study of Health in Pomerania (SHIP-TREND, n = 403). Plasma concentrations of 37 inflammatory markers were measured by a bead-based assay. Furthermore, we employed nuclear magnetic resonance spectroscopy to measure total cholesterol, total triglycerides, total phospholipids as well as the fractional concentrations of cholesterol, triglycerides, phospholipids, ApoA1, ApoA2 and ApoB in all major lipoprotein subclasses. Associations between inflammatory biomarkers and lipoprotein subclasses were analyzed by adjusted linear regression models.
Results
APRIL, BAFF, TWEAK, sCD30, Pentraxin-3, sTNFR1, sTNFR2, Osteocalcin, Chitinase 3-like 1, IFN-alpha2, IFN-gamma, IL-11, IL-12p40, IL-29, IL-32, IL-35, TSLP, MMP1 and MMP2 were related with lipoprotein subclass components, forming two distinct clusters. APRIL had inverse relations to HDL-C (total and subclasses) and HDL Apo-A1 and Apo-A2 content. MMP-2 was inversely related to VLDL-C (total and subclasses), IDL-C as well as LDL5/6-C and VLDL-TG, IDL-TG, total triglycerides as well as LDL5/5-TG and HDL4-TG. Additionally, we identified a cluster of cytokines linked to the Th1-immune response, which were associated with an atherogenic lipoprotein profile.
Conclusion
Our findings expand the existing knowledge of inflammation-lipoprotein interactions, many of which are suggested to be involved in the pathogeneses of chronic non-communicable diseases. The results of our study support the use of immunomodulatory substances for the treatment and possibly prevention of CVD.
Background
We investigated the association between low cardiorespiratory fitness and liver fat content (LFC) in the general population.
Materials and Methods
We evaluated data from 2151 adults (51.1% women) from two population-based cohorts of the Study of Health in Pomerania (SHIP-2 and SHIP-TREND-0). We analysed the cross-sectional associations of peak oxygen uptake (VO2peak) with LFC, assessed by magnetic resonance imaging proton density fat fraction, as well as serum gamma-glutamyltransferase (GGT) and aminotransferase concentrations by multivariable regression models.
Results
We observed significant inverse associations of VO2peak with LFC and serum GGT, but not with serum aminotransferase levels. Specifically, a 1 L/min lower VO2peak was associated with a 1.09% (95% confidence interval [CI]: 0.45-1.73; P = .002) higher LFC and a 0.18 μkatal/L (95% CI: 0.09-0.26; P < .001) higher GGT levels. The adjusted odds ratio (OR) for the risk of prevalent hepatic steatosis (HS) by a 1 L/min decrease in VO2peak was 1.61 (95% CI: 1.22-2.13; P = .001). Compared to subjects with high VO2peak, obese and overweight individuals with low VO2peak had 1.78% (95% CI: 0.32-3.25; P = .017) and 0.94% (95% CI: 0.15-1.74; P = .021) higher mean LFC, respectively. Compared to those with high VO2peak, low VO2peak was independently associated with a higher risk of prevalent HS in the obese (adjusted-OR 2.29, 95% CI=1.48-3.56; P < .001) and overweight (adjusted OR 1.57, 95% CI=1.16-2.14; P = .04) groups.
Conclusions
Lower VO2peak was significantly associated with greater LFC and higher serum GGT levels in a population-based cohort of adult individuals. Our results suggest that low VO2peak might be a risk factor for HS.
Objectives
During the corona pandemic, dental practices temporarily closed their doors to patients except for emergency treatments. Due to the daily occupational exposure, the risk of SARS-CoV-2 transmission among dentists and their team is presumed to be higher than that in the general population. This study examined this issue among dental teams across Germany.
Materials and methods
In total, 2784 participants provided usable questionnaires and dry blood samples. Dry blood samples were used to detect IgG antibodies against SARS-CoV-2. The questionnaires were analyzed to investigate demographic data and working conditions during the pandemic. Multivariable logistic mixed-effects models were applied.
Results
We observed 146 participants with positive SARS-CoV-2 IgG antibodies (5.2%) and 30 subjects with a borderline finding (1.1%). Seventy-four out of the 146 participants with SARS-CoV-2 IgG antibodies did not report a positive SARS-CoV-2 PCR test (50.7%), while 27 participants without SARS-CoV-2 IgG antibodies reported a positive SARS-CoV-2 PCR test (1.1%). Combining the laboratory and self-reported information, the number of participants with a SARS-CoV-2 infection was 179 (6.5%). Though after adjustment for region, mixed-effects models indicated associations of use of rubber dams (OR 1.65; 95% CI: 1.01–2.72) and the number of protective measures (OR 1.16; 95% CI: 1.01–1.34) with increased risk for positive SARS-CoV-2 status, none of those variables was significantly associated with a SARS-CoV-2 status in fully adjusted models.
Conclusions
The risk of SARS-CoV-2 transmission was not higher among the dental team compared to the general population.
Clinical relevance.
Following hygienic regulations and infection control measures ensures the safety of the dental team and their patients.
To evaluate the influence of the blood–brain barrier on neuronal gadolinium deposition in a mouse model after multiple intravenous applications of the linear contrast agent gadodiamide. The prospective study held 54 mice divided into three groups: healthy mice (A), mice with iatrogenic induced disturbance of the blood–brain barrier by glioblastoma (B) or cerebral infarction (C). In each group 9 animals received 10 iv-injections of gadodiamide (1.2 mmol/kg) every 48 h followed by plain T1-weighted brain MRI. A final MRI was performed 5 days after the last contrast injection. Remaining mice underwent MRI in the same time intervals without contrast application (control group). Signal intensities of thalamus, pallidum, pons, dentate nucleus, and globus pallidus-to-thalamus and dentate nucleus-to-pons ratios, were determined. Gadodiamide complex and total gadolinium amount were quantified after the last MR examination via LC–MS/MS and ICP-MS. Dentate nucleus-to-pons and globus pallidus-to-thalamus SI ratios showed no significant increase over time within all mice groups receiving gadodiamide, as well as compared to the control groups at last MR examination. Comparing healthy mice with group B and C after repetitive contrast administration, a significant SI increase could only be detected for glioblastoma mice in globus pallidus-to-thalamus ratio (p = 0.033), infarction mice showed no significant SI alteration. Tissue analysis revealed significantly higher gadolinium levels in glioblastoma group compared to healthy (p = 0.013) and infarction mice (p = 0.029). Multiple application of the linear contrast agent gadodiamide leads to cerebral gadolinium deposition without imaging correlate in MRI.
Background: The aim of our study was to investigate associations of spleen volume with blood count markers and lipid profile in the general population.
Materials & methods: Cross-sectional data from 1,106 individuals aged 30–90 years from the population-based Study of Health in Pomerania (SHIP-START-2) were analyzed. Blood count markers included red blood cell (RBC) counts, hemoglobin, platelet count, and white blood cell (WBC) counts. Lipid profile included total-cholesterol, high-density lipoprotein-cholesterol (HDL-C), and low-density lipoprotein-cholesterol (LDL-C) as well as triglycerides. Linear regression models adjusted for age, sex, body height, and weight were used to associate standardized spleen volume with blood counts and lipid profile markers.
Results: Spleen volume was positively associated with RBC (β = 0.05; 95% confidence interval [CI] = 0.03 to 0.08) and hemoglobin (β = 0.05; 95% CI = 0.01 to 0.09) but inversely with platelet count (β = −16.3; 95% CI = –20.5 to −12.1) and WBC (β = −0.25; 95% CI = −0.37 to −0.14). Furthermore, spleen volume showed inverse associations with total cholesterol (β = −0.17; 95% CI = −0.24 to −0.09), HDL-C (β = −0.08; 95% CI = −0.10 to −0.05), and LDL-C (β = −0.12; 95% CI = −0.17 to −0.06). There was no significant association of spleen volume with triglycerides.
Conclusion: Our study showed that the spleen volume is associated with markers of the blood count and lipid profile in the general population.
Background: The global obesity epidemic is a major public health concern, and accurate diagnosis is essential for identifying at-risk individuals. Three-dimensional (3D) body scanning technology offers several advantages over the standard practice of tape measurements for diagnosing obesity. This study was conducted to validate body scan data from a German population-based cohort and explore clinical implications of this technology in the context of metabolic syndrome. Methods: We performed a cross-sectional analysis of 354 participants from the Study of Health in Pomerania that completed a 3D body scanning examination. The agreement of anthropometric data obtained from 3D body scanning with manual tape measurements was analyzed using correlation analysis and Bland–Altman plots. Classification agreement regarding abdominal obesity based on IDF guidelines was assessed using Cohen’s kappa. The association of body scan measures with metabolic syndrome components was explored using correlation analysis. Results: Three-dimensional body scanning showed excellent validity with slightly larger values that presumably reflect the true circumferences more accurately. Metabolic syndrome was highly prevalent in the sample (31%) and showed strong associations with central obesity. Using body scan vs. tape measurements of waist circumference for classification resulted in a 16% relative increase in the prevalence of abdominal obesity (61.3% vs. 52.8%). Conclusions: These results suggest that the prevalence of obesity may be underestimated using the standard method of tape measurements, highlighting the need for more accurate approaches.
Homoarginine (hArg) is a non-essential cationic amino acid which inhibits hepatic alkaline phosphatases to exert inhibitory effects on bile secretion by targeting intrahepatic biliary epithelium. We analyzed (1) the relationship between hArg and liver biomarkers in two large population-based studies and (2) the impact of hArg supplementation on liver biomarkers. We assessed the relationship between alanine transaminase (ALT), aspartate aminotransferase (AST), γ-glutamyltransferase (GGT), alkaline phosphatases (AP), albumin, total bilirubin, cholinesterase, Quick’s value, liver fat, and Model for End-stage Liver Disease (MELD) and hArg in appropriately adjusted linear regression models. We analyzed the effect of L-hArg supplemention (125 mg L-hArg daily for 4 weeks) on these liver biomarkers. We included 7638 individuals (men: 3705; premenopausal women: 1866, postmenopausal women: 2067). We found positive associations for hArg and ALT (β 0.38 µkatal/L 95% confidence interval (CI): 0.29; 0.48), AST (β 0.29 µkatal/L 95% CI 0.17; 0.41), GGT (β 0.033 µkatal/L 95% CI 0.014; 0.053), Fib-4 score (β 0.08 95% CI 0.03; 0.13), liver fat content (β 0.016% 95% CI 0.006; 0.026), albumin (β 0.030 g/L 95% CI 0.019; 0.040), and cholinesterase (β 0.003 µkatal/L 95% CI 0.002; 0.004) in males. In premenopausal women hArg was positively related with liver fat content (β 0.047% 95%CI 0.013; 0.080) and inversely with albumin (β − 0.057 g/L 95% CI − 0.073; − 0.041). In postmenopausal women hARG was positively associated with AST (β 0.26 µkatal/L 95% CI 0.11; 0.42). hArg supplementation did not affect liver biomarkers. We summarize that hArg may be a marker of liver dysfunction and should be explored further.
Background. The German maternity guidelines require regular medical checkup (MC) during pregnancy as a measure of prevention. Socioeconomic factors such as education, profession, income and origin, but also age and parity may influence the preventive and health behavior of pregnant women. The aim was to investigate the influence of these factors on the participation rate in MC of pregnant women. Method. The current analysis is based on the prospective population-based birth cohort study Survey of Neonates in Pomerania, which was conducted in Western Pomerania, Germany. The data of 4092 pregnant women from 2004 to 2008 were analyzed regarding the antenatal care and health behavior. Up to 12 MC were regularly offered; participation in 10 MC is defined as standard screening according to maternity guidelines. Results. Women participated in the first preventive MC on average in the 10th (±3.8 SD) week of pregnancy. 1343 (34.2%) women participated in standard screening and 2039 (51.9%) took a screening above standard. 547 (13.92%) women participated in less than the 10 standard MCs. In addition, about one-third of the pregnancies investigated in this study were unplanned. Bivariate analyses showed an association between better antenatal care behavior and higher maternal age, stabile partnerships and mother born in Germany, p < 0.05. On the contrary antenatal care below standard were more often found by women with unplanned pregnancies, less educational women and women with lower equivalent income, p < 0.001. Health behaviors also influenced antenatal care. Whereas the risk of antenatal care below standard increased by smoking during pregnancy (RRR 1.64; 95% CI 1.25, 2.14) and alcohol consumption (RRR 1.31; 95% CI 1.01, 1.69), supplementation intake was associated with decreased risk (iodine—RRR 0.66; 95% CI 0.53, 0.81; folic acid—RRR 0.56; 95% CI 0.44, 0.72). The health behavior of pregnant women also differs according to their social status. Higher maternal income was negatively correlated with smoking during pregnancy (OR 0.2; 95% CI 0.15, 0.24), but positively associated with alcohol consumption during pregnancy (OR 1.3; 95% CI 1.15, 1.48) and lower pre-pregnancy BMI (Coef. = 0.083, p < 0.001). Lower maternal education was positively correlated with smoking during pregnancy (OR 59.0; 95% CI 28.68, 121.23). Conclusions. Prenatal care according to maternity guidelines is well established with a high participation rate in MC during pregnancy of more than 85%. However, targeted preventive measures may address younger age, socioeconomic status and health-damaging behaviors (smoking, drinking) of the pregnant women because these factors were associated with antenatal care below standard.
Body surface scan anthropometrics are related to cardiorespiratory fitness in the general population
(2022)
The assessment of cardiorespiratory fitness (CRF) is an important tool for prognosis evaluation of cardiovascular events. The gold standard to measure CRF is cardiopulmonary exercise testing (CPET) to determine peak oxygen uptake (VO2peak). However, CPET is not only time consuming but also expensive and is therefore not widely applicable in daily practice. The aim of our study was to analyze, whether and which anthropometric markers derived from a 3D body scanner were related to VO2peak in a general population-based study. We analyzed data (SHIP-START-3) from 3D body scanner and CPET of 1035 subjects (529 women; 51.1%, age range 36–93). A total of 164 anthropometric markers were detected with the 3D body scanner VITUS Smart XXL using the software AnthroScan Professional. Anthropometric measurements were standardized and associated with CRF by sex-stratified linear regression models adjusted for age and height. Anthropometric markers were ranked according to the − log- p values derived from these regression models. In men a greater left and right thigh-knee-ratio, a longer forearm-fingertip length, a greater left thigh circumference and greater left upper arm circumference were most strongly associated with a higher VO2peak. In women a greater left and right thigh circumference, left calf circumference, thigh thickness and right calf circumference were most strongly associated with a higher VO2peak. The detected VO2peak-related anthropometric markers could be helpful in assessing CRF in clinical routine. Commonly used anthropometric markers, e.g. waist and hip circumference, were not among the markers associated with VO2peak.
The Apolipoprotein E (APOE) gene polymorphism (rs429358 and rs7412) shows a well-established association with lipid profiles, but its effect on cardiovascular disease is still conflicting. Therefore, we examined the association of different APOE alleles with common carotid artery intima-media thickness (CCA-IMT), carotid plaques, incident myocardial infarction (MI) and stroke. We analyzed data from 3327 participants aged 20–79 years of the population-based Study of Health in Pomerania (SHIP) from Northeast Germany with a median follow-up time of 14.5 years. Linear, logistic, and Cox-regression models were used to assess the associations of the APOE polymorphism with CCA-IMT, carotid plaques, incident MI and stroke, respectively. In our study, the APOE E2 allele was associated with lower CCA-IMT at baseline compared to E3 homozygotes (β: − 0.02 [95% CI − 0.04, − 0.004]). Over the follow-up, 244 MI events and 218 stroke events were observed. APOE E2 and E4 allele were not associated with incident MI (E2 HR: 1.06 [95% CI 0.68, 1.66]; E4 HR: 1.03 [95% CI 0.73, 1.45]) and incident stroke (E2 HR: 0.79 [95% CI 0.48, 1.30]; E4 HR: 0.96 [95% CI 0.66, 1.38]) in any of the models adjusting for potential confounders. However, the positive association between CCA-IMT and incident MI was more pronounced in E2 carriers than E3 homozygotes. Thus, our study suggests that while APOE E2 allele may predispose individuals to lower CCA-IMT, E2 carriers may be more prone to MI than E3 homozygotes as the CCA-IMT increases. APOE E4 allele had no effect on CCA-IMT, plaques, MI or stroke.
Background
Missing data are ubiquitous in randomised controlled trials. Although sensitivity analyses for different missing data mechanisms (missing at random vs. missing not at random) are widely recommended, they are rarely conducted in practice. The aim of the present study was to demonstrate sensitivity analyses for different assumptions regarding the missing data mechanism for randomised controlled trials using latent growth modelling (LGM).
Methods
Data from a randomised controlled brief alcohol intervention trial was used. The sample included 1646 adults (56% female; mean age = 31.0 years) from the general population who had received up to three individualized alcohol feedback letters or assessment-only. Follow-up interviews were conducted after 12 and 36 months via telephone. The main outcome for the analysis was change in alcohol use over time. A three-step LGM approach was used. First, evidence about the process that generated the missing data was accumulated by analysing the extent of missing values in both study conditions, missing data patterns, and baseline variables that predicted participation in the two follow-up assessments using logistic regression. Second, growth models were calculated to analyse intervention effects over time. These models assumed that data were missing at random and applied full-information maximum likelihood estimation. Third, the findings were safeguarded by incorporating model components to account for the possibility that data were missing not at random. For that purpose, Diggle-Kenward selection, Wu-Carroll shared parameter and pattern mixture models were implemented.
Results
Although the true data generating process remained unknown, the evidence was unequivocal: both the intervention and control group reduced their alcohol use over time, but no significant group differences emerged. There was no clear evidence for intervention efficacy, neither in the growth models that assumed the missing data to be at random nor those that assumed the missing data to be not at random.
Conclusion
The illustrated approach allows the assessment of how sensitive conclusions about the efficacy of an intervention are to different assumptions regarding the missing data mechanism. For researchers familiar with LGM, it is a valuable statistical supplement to safeguard their findings against the possibility of nonignorable missingness.
Background
Approaching epidemiological data with flexible machine learning algorithms is of great value for understanding disease-specific association patterns. However, it can be difficult to correctly extract and understand those patterns due to the lack of model interpretability.
Method
We here propose a machine learning workflow that combines random forests with Bayesian network surrogate models to allow for a deeper level of interpretation of complex association patterns. We first evaluate the proposed workflow on synthetic data. We then apply it to data from the large population-based Study of Health in Pomerania (SHIP). Based on this combination, we discover and interpret broad patterns of individual serum TSH concentrations, an important marker of thyroid functionality.
Results
Evaluations using simulated data show that feature associations can be correctly recovered by combining random forests and Bayesian networks. The presented model achieves predictive accuracy that is similar to state-of-the-art models (root mean square error of 0.66, mean absolute error of 0.55, coefficient of determination of R2 = 0.15). We identify 62 relevant features from the final random forest model, ranging from general health variables over dietary and genetic factors to physiological, hematological and hemostasis parameters. The Bayesian network model is used to put these features into context and make the black-box random forest model more understandable.
Conclusion
We demonstrate that the combination of random forest and Bayesian network analysis is helpful to reveal and interpret broad association patterns of individual TSH concentrations. The discovered patterns are in line with state-of-the-art literature. They may be useful for future thyroid research and improved dosing of therapeutics.
The associations of thyroid function parameters with non-alcoholic fatty liver disease (NAFLD) and hepatic iron overload are not entirely clear. We have cross-sectionally investigated these associations among 2734 participants of two population-based cross-sectional studies of the Study of Health in Pomerania. Serum levels of thyroid-stimulating hormone (TSH), free tri-iodothyronine (fT3), and free thyroxine (fT4) levels were measured. Liver fat content (by proton-density fat fraction) as well as hepatic iron content (by transverse relaxation rate; R2*) were assessed by quantitative MRI. Thyroid function parameters were associated with hepatic fat and iron contents by median and logistic regression models adjusted for confounding. There were no associations between serum TSH levels and liver fat content, NAFLD, or hepatic iron overload. Serum fT4 levels were inversely associated with liver fat content, NAFLD, hepatic iron contents, and hepatic iron overload. Serum fT3 levels as well as the fT3 to fT4 ratio were positively associated with hepatic fat, NAFLD, hepatic iron contents, but not with hepatic iron overload. Associations between fT3 levels and liver fat content were strongest in obese individuals, in which we also observed an inverse association between TSH levels and NAFLD. These findings might be the result of a higher conversion of fT4 to the biologically active form fT3. Our results suggest that a subclinical hyperthyroid state may be associated with NAFLD, particularly in obese individuals. Furthermore, thyroid hormone levels seem to be more strongly associated with increased liver fat content compared to hepatic iron content.
Variability of Thyroid Measurements from Ultrasound and Laboratory in a Repeated Measurements Study
(2020)
Background: Variability of measurements in medical research can be due to different sources. Quantification of measurement errors facilitates probabilistic sensitivity analyses in future research to minimize potential bias in epidemiological studies. We aimed to investigate the variation of thyroid-related outcomes derived from ultrasound (US) and laboratory analyses in a repeated measurements study. Subjects and Methods: Twenty-five volunteers (13 females, 12 males) aged 22–70 years were examined once a month over 1 year. US measurements included thyroid volume, goiter, and thyroid nodules. Laboratory measurements included urinary iodine concentrations and serum levels of thyroid-stimulating hormone (TSH), free triiodothyronine (fT3), free thyroxine (fT4), and thyroglobulin. Variations in continuous thyroid markers were assessed as coefficient of variation (CV) defined as mean of the individual CVs with bootstrapped confidence intervals and as intraclass correlation coefficients (ICCs). Variations in dichotomous thyroid markers were assessed by Cohen’s kappa. Results: CV was highest for urinary iodine concentrations (56.9%), followed by TSH (27.2%), thyroglobulin (18.2%), thyroid volume (10.5%), fT3 (8.1%), and fT4 (6.3%). The ICC was lowest for urinary iodine concentrations (0.42), followed by fT3 (0.55), TSH (0.64), fT4 (0.72), thyroid volume (0.87), and thyroglobulin (0.90). Cohen’s kappa values for the presence of goiter or thyroid nodules were 0.64 and 0.70, respectively. Conclusion: Our study provides measures of variation for thyroid outcomes, which can be used for probabilistic sensitivity analyses of epidemiological data. The low intraindividual variation of serum thyroglobulin in comparison to urinary iodine concentrations emphasizes the potential of thyroglobulin as marker for the iodine status of populations.