Mirjam J. Knol

A Tutorial on Interaction

Abstract In this tutorial, we provide a broad introduction to the topic of interaction between the effects of exposures. We discuss interaction on both additive and multiplicative scales using risks, and we discuss their relation to statistical models (e.g. linear, log-linear, and logistic models). We discuss and evaluate arguments that have been made for using additive or multiplicative scales to assess interaction. We further discuss approaches to presenting interaction analyses, different mechanistic forms of interaction, when interaction is robust to unmeasured confounding, interaction for continuous outcomes, qualitative or “crossover” interactions, methods for attributing effects to interactions, case-only estimators of interaction, and power and sample size calculations for additive and multiplicative interaction.

Overestimation of risk ratios by odds ratios in trials and cohort studies: alternatives to logistic regression

Logistic regression analysis, which estimates odds ratios, is often used to adjust for covariables in cohort studies and randomized controlled trials (RCTs) that study a dichotomous outcome. In case–control studies, the odds ratio is the appropriate effect estimate, and the odds ratio can

Enteric-coated mycophenolate sodium versus cyclosporin A as long-term treatment in adult patients with severe atopic dermatitis: A randomized controlled trial

BACKGROUND Cyclosporin A (CsA) is frequently used in the treatment of severe atopic dermatitis (AD). Enteric-coated mycophenolate sodium (EC-MPS) may be an alternative with equal efficacy and fewer side effects. OBJECTIVE The aim of this observer-blinded randomized controlled trial was to compare EC-MPS with CsA as long-term treatment in adult patients with severe AD. METHODS Fifty five patients with AD were treated with CsA (5 mg/kg) in a 6-week run-in period. Thereafter, patients either received CsA (3 mg/kg; n = 26) or EC-MPS (1440 mg; n = 24) during a maintenance phase of 30 weeks and there was a 12-week follow-up period. Disease activity was measured using the objective SCORAD and serum thymus and activation-regulated chemokine (TARC) levels and side effects were registered. RESULTS During the first 10 weeks the objective SCORAD and serum TARC levels in the EC-MPS study arm were higher in comparison with the CsA study arm. In addition, 7 of the 24 patients treated with EC-MPS required short oral corticosteroid courses. During maintenance phase disease activity was comparable in both study arms. Side effects in both study arms were mild and transient. After study medication withdrawal, disease activity of the patients in the CsA study arm significantly increased compared with the EC-MPS study arm. LIMITATION The nonblinding of patients and prescriber of rescue medication are limitations. CONCLUSIONS This study shows that EC-MPS is as effective as CsA as maintenance therapy in patients with AD. However, clinical improvement with EC-MPS is delayed in comparison with CsA. Clinical remission after stopping EC-MPS lasts longer compared with CsA.

Depressive symptoms in subjects with diagnosed and undiagnosed type 2 diabetes

Objective: To investigate if disturbed glucose homeostasis or known diagnosis of diabetes was associated with depressive symptoms. The reason for the increased prevalence of depression in patients with Type 2 diabetes mellitus (DM2) is unknown. Methods: Within the Utrecht Health Project, an ongoing longitudinal study among inhabitants of a residential area of a large city in The Netherlands, 4747 subjects (age: 39.4 ± 12.5 years) were classified into four mutually exclusive categories: normal fasting plasma glucose (FPG) (<5.6 mmol/l), impaired FPG (≥5.6 and <7.0 mmol/l), undiagnosed DM2 (FPG ≥7.0 mmol/l), and diagnosed DM2. Presence of depressive symptoms was defined as a score of ≥25 on the depression subscale of the Symptom Check List (SCL-90) or self-reported use of antidepressants. Results: Diagnosed DM2 was associated with an increased risk of depressive symptoms (odds ratio (OR) = 1.69; 95% confidence interval (CI) 1.06–2.72) after adjustment for demographic and lifestyle variables. Additional adjustment for number of chronic diseases reduced the OR to 1.36 (95% CI 0.83–2.23). Impaired fasting glucose and undiagnosed DM2 were not associated with depressive symptoms. Conclusions: Our findings suggest that disturbed glucose homeostasis is not associated with depressive symptoms. The increased prevalence of depressive symptoms among patients with diagnosed DM2 suggests that depressive symptoms might be a consequence of the burden of diabetes. The number of chronic diseases seems to explain part of the association between DM2 and depressive symptoms. DM2 = Type 2 diabetes mellitus; CI = confidence interval; DSM-IV = Diagnostic and Statistical Manual of Mental Disorders, fourth revision; FPG = fasting plasma glucose; OR = odds ratio; SCID = Structured Clinical Interview for DSM-IV; SCL-90 = Symptom Checklist; SD = standard deviation; SPSS = Statistical Package for the Social Sciences; UHP = Utrecht Health Project.

What Do Case-Control Studies Estimate? Survey of Methods and Assumptions in Published Case-Control Research

To evaluate strategies used to select cases and controls and how reported odds ratios are interpreted, the authors examined 150 case-control studies published in leading general medicine, epidemiology, and clinical specialist journals from 2001 to 2007. Most of the studies (125/150; 83%) were based on incident cases; among these, the source population was mostly dynamic (102/125; 82%). A minority (23/125; 18%) sampled from a fixed cohort. Among studies with incident cases, 105 (84%) could interpret the odds ratio as a rate ratio. Fifty-seven (46% of 125) required the source population to be stable for such interpretation, while the remaining 48 (38% of 125) did not need any assumptions because of matching on time or concurrent sampling. Another 17 (14% of 125) studies with incident cases could interpret the odds ratio as a risk ratio, with 16 of them requiring the rare disease assumption for this interpretation. The rare disease assumption was discussed in 4 studies but was not relevant to any of them. No investigators mentioned the need for a stable population. The authors conclude that in current case-control research, a stable exposure distribution is much more frequently needed to interpret odds ratios than the rare disease assumption. At present, investigators conducting case-control studies rarely discuss what their odds ratios estimate.

When one depends on the other: reporting of interaction in case-control and cohort studies

Background: Interaction refers to the situation in which the effect of 1 exposure on an outcome differs across strata of another exposure. We did a survey of epidemiologic studies published in leading journals to examine how interaction is assessed and reported. Methods: We selected 150 case-control and 75 cohort studies published between May 2001 and May 2007 in leading general medicine, epidemiology, and clinical specialist journals. Two reviewers independently extracted data on study characteristics. Results: Of the 225 studies, 138 (61%) addressed interaction. Among these, 25 (18%) presented no data or only a P value or a statement of statistical significance; 40 (29%) presented stratum-specific effect estimates but no meaningful comparison of these estimates; and 58 (42%) presented stratum-specific estimates and appropriate tests for interaction. Fifteen articles (11%) presented the individual effects of both exposures and also their joint effect or a product term, providing sufficient information to interpret interaction on an additive and multiplicative scale. Reporting was poorest in articles published in clinical specialist articles and most adequate in articles published in general medicine journals, with epidemiology journals in an intermediate position. Conclusions: A majority of articles reporting cohort and case-control studies address possible interactions between exposures. However, in about half of these, the information provided was unsatisfactory, and only 1 in 10 studies reported data that allowed readers to interpret interaction effects on an additive and multiplicative scale.

The diagnostic value of specific IgE to Ara h 2 to predict peanut allergy in children is comparable to a validated and updated diagnostic prediction model.

BACKGROUND A diagnostic prediction model for peanut allergy in children was recently published, using 6 predictors: sex, age, history, skin prick test, peanut specific immunoglobulin E (sIgE), and total IgE minus peanut sIgE. OBJECTIVES To validate this model and update it by adding allergic rhinitis, atopic dermatitis, and sIgE to peanut components Ara h 1, 2, 3, and 8 as candidate predictors. To develop a new model based only on sIgE to peanut components. METHODS Validation was performed by testing discrimination (diagnostic value) with an area under the receiver operating characteristic curve and calibration (agreement between predicted and observed frequencies of peanut allergy) with the Hosmer-Lemeshow test and a calibration plot. The performance of the (updated) models was similarly analyzed. RESULTS Validation of the model in 100 patients showed good discrimination (88%) but poor calibration (P < .001). In the updating process, age, history, and additional candidate predictors did not significantly increase discrimination, being 94%, and leaving only 4 predictors of the original model: sex, skin prick test, peanut sIgE, and total IgE minus sIgE. When building a model with sIgE to peanut components, Ara h 2 was the only predictor, with a discriminative ability of 90%. Cutoff values with 100% positive and negative predictive values could be calculated for both the updated model and sIgE to Ara h 2. In this way, the outcome of the food challenge could be predicted with 100% accuracy in 59% (updated model) and 50% (Ara h 2) of the patients. CONCLUSIONS Discrimination of the validated model was good; however, calibration was poor. The discriminative ability of Ara h 2 was almost comparable to that of the updated model, containing 4 predictors. With both models, the need for peanut challenges could be reduced by at least 50%.

Does physical activity modify the risk of obesity for type 2 diabetes: a review of epidemiological data

Obesity and physical inactivity are both risk factors for type 2 diabetes. Since they are strongly associated, it has been suggested that they might interact. In this study, we summarized the evidence on this interaction by conducting a systematic review. Two types of interaction have been discerned, statistical and biological interaction, which could give different results. Therefore, we calculated both types of interaction for the studies in our review. Cohort studies, published between 1999 and 2008, that investigated the effects of obesity and physical activity on the risk of type 2 diabetes were included. We calculated both biological and statistical interaction in these studies. Eight studies were included of which five were suitable to calculate interaction. All studies showed positive biological interaction, meaning that the joint effect was more than the sum of the individual effects. However, there was inconsistent statistical interaction; in some studies the joint effect was more than the product of the individual effects, in other studies it was less. The results show that obesity and physical inactivity interact on an additive scale. This means that prevention of either obesity or physical inactivity, not only reduces the risk of diabetes by taking away the independent effect of this factor, but also by preventing the cases that were caused by the interaction between both factors. Furthermore, this review clearly showed that results can differ depending on what method is used to assess interaction.

No increased incidence of diabetes in antidepressant users.

This study investigated whether the association between depression and diabetes was influenced by the presence of chronic somatic disease. To distinguish between depression and other psychosocial complaints, we studied the onset of diabetes in antidepressant (AD) users and benzodiazepine (BD) users, respectively. From the PHARMO database, which includes complete drug prescription data, we identified subjects using (i) no ADs and no BDs; (ii) AD but no BD; (iii) BD but no AD; and (iv) AD and BD. A total of 60 516 individuals (age: 45.5±17 years; 42.1% men) were followed from their first prescription for AD or BD until end of registration or a first prescription for antidiabetic drugs. The crude incidence rate in AD but no BD users was not increased compared with no AD and no BD users. After adjustment for age, sex and chronic diseases, the hazard ratios (95% confidence interval) were 1.05 (0.88–1.26) for AD but no BD users, 1.21 (1.02–1.43) for BD but no AD users and 1.37 (1.12–1.68) for AD and BD users compared with no AD and no BD users. We did not find an increased risk of diabetes in individuals using ADs. The association between BD use and diabetes was partly explained by chronic somatic comorbidity.

E-health in caring for patients with atopic dermatitis: A randomized controlled cost-effectiveness study of internet-guided monitoring and online self-management training

Background  The Dermatology Department of the University Medical Centre Utrecht, the Netherlands, developed an e‐health portal for patients with atopic dermatitis (AD), consisting of e‐consultation, a patient‐tailored website, monitoring and self‐management training.