Kristian Hindberg

C‐reactive protein, obesity, and the risk of arterial and venous thrombosis

Essentials We performed repeated measurements of C‐reactive protein (CRP) and obesity in a cohort study. CRP was associated with risk of myocardial infarction and venous thromboembolism. CRP was a mediator for risk of myocardial infarction in obese men and women. CRP was a partial mediator for risk of venous thromboembolism in obese women, but not in men.

Predicting colorectal surgical complications using heterogeneous clinical data and kernel methods

OBJECTIVE In this work, we have developed a learning system capable of exploiting information conveyed by longitudinal Electronic Health Records (EHRs) for the prediction of a common postoperative complication, Anastomosis Leakage (AL), in a data-driven way and by fusing temporal population data from different and heterogeneous sources in the EHRs. MATERIAL AND METHODS We used linear and non-linear kernel methods individually for each data source, and leveraging the powerful multiple kernels for their effective combination. To validate the system, we used data from the EHR of the gastrointestinal department at a university hospital. RESULTS We first investigated the early prediction performance from each data source separately, by computing Area Under the Curve values for processed free text (0.83), blood tests (0.74), and vital signs (0.65), respectively. When exploiting the heterogeneous data sources combined using the composite kernel framework, the prediction capabilities increased considerably (0.92). Finally, posterior probabilities were evaluated for risk assessment of patients as an aid for clinicians to raise alertness at an early stage, in order to act promptly for avoiding AL complications. DISCUSSION Machine-learning statistical model from EHR data can be useful to predict surgical complications. The combination of EHR extracted free text, blood samples values, and patient vital signs, improves the model performance. These results can be used as a framework for preoperative clinical decision support.

Bayesian multiscale feature detection of log-spectral densities

A fully-automatic Bayesian visualization tool to identify periodic components of evenly sampled stationary time series, is presented. The given method applies the multiscale ideas of the SiZer-methodology to the log-spectral density of a given series. The idea is to detect significant peaks in the true underlying curve viewed at different resolutions or scales. The results are displayed in significance maps, illustrating for which scales and for which frequencies, peaks in the log-spectral density are detected as significant. The inference involved in producing the significance maps is performed using the recently developed simplified Laplace approximation. This is a Bayesian deterministic approach used to get accurate estimates of posterior marginals for latent Gaussian Markov random fields at a low computational cost, avoiding the use of Markov chain Monte Carlo techniques. Application of the given exploratory tool is illustrated analyzing both synthetic and real time series.

Sun-induced migraine attacks in an Arctic population.

Aims: The main purpose of the study was to evaluate the impact of weather, and especially sun exposure, on migraine. Methods: Data from a previous prospective 12-month diary study was compared with meteorological data. We retrospectively evaluated 1250 migraine attacks recorded by a group of 40 women with a mean age of 37.1 years who fulfilled the IHS criteria for migraine with and without aura. Results: The patients reported more sun-induced migraine attacks on sunny days, but the total distribution of migraine attacks was constant throughout the year. Also, no seasonal variation of migraine, nor any relationships between weather parameters and onset of migraine attacks, were found. An analysis of a subgroup of patients with ‘sun-induced’ migraine showed a significant increase in frequency of migraine attacks in the summer compared to the winter (p = 0.04). Conclusion: This study confirms that sunlight might be a trigger for migraine, but a risk for increased impact of light on the total ailment of migraine headache should be searched for in a subgroup of sensitive migraineurs.

Joint effects of cancer and variants in the Factor 5 gene on the risk of venous thromboembolism

Venous thromboembolism occurs frequently in cancer patients. Two variants in the factor 5 gene (F5), rs6025 encoding for the factor V Leiden mutation R506Q, and rs4524 encoding K858R, have been found to be associated with venous thromboembolism. We assessed the joint effect of active cancer and these two F5 variants on venous thromboembolism risk in a case-cohort study. Cases with a first venous thromboembolism (n=609) and a randomly selected age-weighted cohort (n=1,691) were sampled from the general population in Tromsø, Norway. Venous thromboembolism was classified as cancer-related if it occurred in the period 6 months before to 2 years after a diagnosis of cancer. Active cancer was associated with an 8.9-fold higher risk of venous thromboembolism (95% CI 7.2–10.9). The risk of cancer-related venous thromboembolism was 16.7-fold (95% CI 9.9–28.0) higher in subjects heterozygous for rs6025 compared with non-carriers of this variant without active cancer. In subjects with active cancer the risk of venous thromboembolism was 15.9-fold higher (95% CI 9.1–27.9) in those with one risk allele at rs4524, and 21.1-fold (95% CI 12.4–35.8) higher in those with two risk alleles compared with non-carriers without active cancer. A synergistic interaction was observed between active cancer and factor V Leiden (relative excess risk due to interaction 7.0; 95% CI 0.5–14.4) and rs4524 (relative excess risk due to interaction 15.0; 95% CI 7.5–29.2). The incidence of venous thromboembolism during the initial 6 months following a diagnosis of cancer was particularly high in subjects with risk alleles at these loci. This implies that the combination of cancer and F5 variants synergistically increases venous thromboembolism risk.

Computer-Aided Decision Support for Melanoma Detection Applied on Melanocytic and Nonmelanocytic Skin Lesions: A Comparison of Two Systems Based on Automatic Analysis of Dermoscopic Images

Commercially available clinical decision support systems (CDSSs) for skin cancer have been designed for the detection of melanoma only. Correct use of the systems requires expert knowledge, hampering their utility for nonexperts. Furthermore, there are no systems to detect other common skin cancer types, that is, nonmelanoma skin cancer (NMSC). As early diagnosis of skin cancer is essential, there is a need for a CDSS that is applicable to all types of skin lesions and is suitable for nonexperts. Nevus Doctor (ND) is a CDSS being developed by the authors. We here investigate NDs ability to detect both melanoma and NMSC and the opportunities for improvement. An independent test set of dermoscopic images of 870 skin lesions, including 44 melanomas and 101 NMSCs, were analysed by ND. Its sensitivity to melanoma and NMSC was compared to that of Mole Expert (ME), a commercially available CDSS, using the same set of lesions. ND and ME had similar sensitivity to melanoma. For ND at 95% melanoma sensitivity, the NMSC sensitivity was 100%, and the specificity was 12%. The melanomas misclassified by ND at 95% sensitivity were correctly classified by ME, and vice versa. ND is able to detect NMSC without sacrificing melanoma sensitivity.

Recurrence and mortality after first venous thromboembolism in a large population-based cohort

Essentials Reports on recurrence and mortality after a first venous thromboembolism (VTE) vary considerably. We describe rates of recurrence and mortality in patients with a first VTE from the Tromsø study. The overall recurrence rate was 3.9 per 100 person‐years, but this varied widely with time. Despite advances in VTE management, the rates of adverse events are still fairly high.

Atherosclerotic Risk Factors and Risk of Myocardial Infarction and Venous Thromboembolism; Time-Fixed versus Time-Varying Analyses. The Tromsø Study.

Background Single measurements of modifiable risk factors may underestimate associations with outcomes in cohorts. We aimed to compare risk estimates of myocardial infarction (MI) and venous thromboembolism (VTE) by atherosclerotic risk factors during long follow-up using time-fixed analyses without and with correction for regression dilution and time-varying analyses. Methods The study included 5970 subjects enrolled in the fourth survey of the Tromsø Study (1994/95). Blood pressure, lipid levels, body mass index (BMI), diabetes and smoking status were measured at baseline, and subjects still alive at the fifth (2001/02, n = 5179) and sixth (2007/08, n = 4391) survey were re-measured. Incident events of MI (n = 714) and VTE (n = 214) were recorded until December 2010. Time-fixed and time-varying Cox regression models were used to estimate hazard ratios (HR) for MI and VTE adjusted for age and sex. Results Variations in BMI, blood pressure and lipid levels were small, and did not alter the risk estimates when time-varying analyses were compared to time-fixed analyses. For MI, variables that changed considerably over time yielded the greatest changes in risk estimates (HR for smoking changed from 1.80 (95% CI 1.55–2.10) to 2.08 (95% CI 1.78–2.42)). For VTE, only BMI was associated with increased risk in both time-fixed and time-varying analysis, but the risk estimates weakened in the time-varying analysis. Correction of time-fixed HRs with Rosner´s method tended to overestimate risk estimates compared to time-varying analysis. Comment For MI and VTE, risk estimates based on baseline and repeated measures corresponded well, whereas correction for regression dilution tended to overestimate risks.

Impact of Venous Thromboembolism on the Formation and Progression of Carotid Atherosclerosis: The Tromsø Study

Venous thromboembolism (VTE) is associated with increased risk of arterial cardiovascular diseases (CVD), and development of atherosclerosis secondary to VTE may be an intermediate between VTE and CVD. Therefore, we aimed to investigate whether incident VTE was associated with subsequent carotid atherosclerosis formation and progression in a population-based observational study. Subjects attending two or more ultrasound examinations of the right carotid artery, with measurement of total plaque area (TPA), in the Tromsø Study in 1994–1995, 2001–2002, and/or 2007–2008 were eligible. We identified 150 subjects diagnosed with first-lifetime VTE between the initial and follow-up visit, and randomly selected 600 age- and sex-matched subjects without VTE between the visits. Subjects with VTE and carotid plaque(s) at the first visit had 4.1 mm 2 (β: 4.13, 95% CI: −1.72 to 9.98) larger change in TPA between the first and second visit compared with subjects without VTE after adjustment for change in high-sensitivity C-reactive protein (hs-CRP) and traditional atherosclerotic risk factors. The association remained after restricting the analyses to VTE events diagnosed in the first half of the time interval between the carotid ultrasounds (β: 4.02, 95% CI: −3.66 to 11.70), supporting that the change in TPA occurred subsequent to the VTE. No association was found between VTE and novel carotid plaque formation. In conclusion, we found a possible association between VTE and atherosclerosis progression in those with already established carotid plaques, but not between VTE and novel plaque formation. The association between VTE and carotid plaque progression was not mediated by low-grade inflammation assessed by hs-CRP.

Using anchors from free text in electronic health records to diagnose postoperative delirium

OBJECTIVES Postoperative delirium is a common complication after major surgery among the elderly. Despite its potentially serious consequences, the complication often goes undetected and undiagnosed. In order to provide diagnosis support one could potentially exploit the information hidden in free text documents from electronic health records using data-driven clinical decision support tools. However, these tools depend on labeled training data and can be both time consuming and expensive to create. METHODS The recent learning with anchors framework resolves this problem by transforming key observations (anchors) into labels. This is a promising framework, but it is heavily reliant on clinicians knowledge for specifying good anchor choices in order to perform well. In this paper we propose a novel method for specifying anchors from free text documents, following an exploratory data analysis approach based on clustering and data visualization techniques. We investigate the use of the new framework as a way to detect postoperative delirium. RESULTS By applying the proposed method to medical data gathered from a Norwegian university hospital, we increase the area under the precision-recall curve from 0.51 to 0.96 compared to baselines. CONCLUSIONS The proposed approach can be used as a framework for clinical decision support for postoperative delirium.