Umeå University's logo

Introduction: This study aimed to investigate if individuals with childhood-onset type 1 diabetes having a parent with the same condition (parental diabetes) had worse metabolic control and an increased risk of death and renal failure compared with those with parents without type 1 diabetes (sporadic diabetes).

Research design and methods: We conducted a population-based cohort study using data from the Swedish Childhood Diabetes Register, including cases with onset of type 1 diabetes before the age of 15 and recorded between 1977 and 2010. The cohort was linked to national registers to compare mortality, renal failure, and glycated hemoglobin (HBA1c) levels.

Results: We identified 16 572 incident cases of childhood-onset type 1 diabetes. Of these, 15 701 had data on parental diabetes status, with 1390 (8.9%) having at least one parent with this condition. HbA1c data were available in 9105 individuals at 20-30 years of age, with the parental group showing higher levels compared with the sporadic diabetes group (8.4% (68 mmol/mol) vs 8.2% (66 mmol/mol), p=0.004). The Cox proportional HR for death in parental diabetes was 1.33 (95% CI 1.00 to 1.75), and the competing risk HR for renal failure was 1.27 (95% CI 1.08 to 1.50). Women in the parental diabetes group had a higher risk of early death (HR 1.79, 95% CI 1.17 to 2.72) compared with the sporadic diabetes group.

Conclusions: Individuals with parental diabetes had slightly higher HbA1c and elevated risks of renal failure and death compared with those with sporadic diabetes, especially pronounced in women. Although the exact mechanisms behind these differences are unclear, we suggest that individualized care may benefit individuals with parental type 1 diabetes.

Aims/hypothesis: In persons with type 1 diabetes, the risk of cancer remains controversial. We wanted to examine the excess risk of cancer in a large population-based cohort diagnosed with type 1 diabetes before 15 years of age.

Study population and methods: From 1 July 1977 to 31 December 2013, we prospectively and on a national scale included 18,724 persons (53% men) with childhood-onset type 1 diabetes. For each person with type 1 diabetes, we selected four referents, matched for the date at birth and municipality of living at the time when the case developed diabetes. Cases and referents were linked to national registers of cancer and of the cause of death.

Results: A total of 125 persons (61% women) with diabetes had 135 different cancers, all diagnosed after the diabetes diagnosis. The median duration from diabetes diagnosis to first cancer diagnosis was 19 years (interquartile range 10-26). The median age at cancer diagnosis in the diabetes group was 28 years (interquartile range 20-35). The overall standardized incidence ratio (95%), using the Swedish general population as referents for women with diabetes was 1.28 (1.02, 1.58) and when comparing women with diabetes with matched referents, we found a hazard ratio of 1.42 (1.10, 1.85). No elevated risk was seen for men. Cancers of the breast and testis were the most common types in women and men respectively.

Conclusions: Women with childhood-onset type 1 diabetes had a small but significantly elevated risk of cancer. No such tendency was seen for men. The reason behind this is unclear.

This paper investigates if the effect of type 1 diabetes mellitus (T1DM) on school performance, documented in prior research, has changed in more recent birth cohorts of children using national Swedish population register data. The issue is of interest because management and treatment of the disease have improved over the last decades and, furthermore, because of changes in the educational grading system. Despite these changes, data indicate a persistent negative effect of T1DM on compulsory and upper secondary school grades with a standardized effect size of −0.109 and −0.070, respectively, and the results appear only marginally smaller compared to earlier findings in cohorts completing school under the previous grading system. Moreover, the results are consistent for alternative model specifications and econometric estimation strategies. Whereas access to new treatment technologies and improved diabetes management strategies has reduced the burden of diabetes in daily life, the results from this study indicate that continued efforts are needed to improve the situation in school for children with T1DM to prevent potential long-term socio-economic consequences.

In observational studies, the causal effect of a treatment may be confounded with variables that are related to both the treatment and the outcome of interest. In order to identify a causal effect, such studies often rely on the unconfoundedness assumption, i.e., that all confounding variables are observed. The choice of covariates to control for, which is primarily based on subject matter knowledge, may result in a large covariate vector in the attempt to ensure that unconfoundedness holds. However, including redundant covariates can affect bias and efficiency of nonparametric causal effect estimators, e.g., due to the curse of dimensionality. In this paper, data-driven algo- rithms for the selection of sufficient covariate subsets are investigated. Under the assumption of unconfoundedness we search for minimal subsets of the covariate vector. Based on the framework of sufficient dimension reduction or kernel smoothing, the algorithms perform a backward elim- ination procedure testing the significance of each covariate. Their performance is evaluated in simulations and an application using data from the Swedish Childhood Diabetes Register is also presented.

When an initial case-control study is performed, data can be used in a secondary analysis to evaluate the effect of the case-defining event on later outcomes. In this paper, we study the example in which the role of the event is changed from a response variable to a treatment of interest. If the aim is to estimate marginal effects, such as average effects in the population, the sampling scheme needs to be adjusted for. We study estimators of the average effect of the treatment in a secondary analysis of matched and unmatched case-control data where the probability of being a case is known. For a general class of estimators, we show the components of the bias resulting from ignoring the sampling scheme and demonstrate a design-weighted matching estimator of the average causal effect. In simulations, the finite sample properties of the design-weighted matching estimator are studied. Using a Swedish diabetes incidence register with a matched case-control design, we study the effect of childhood onset diabetes on the use of antidepressant medication as an adult.

We describe the R package CovSel, which reduces the dimension of the covariate vector for the purpose of estimating an average causal effect under the unconfoundedness assumption. Covariate selection algorithms developed in De Luna, Waernbaum, and Richardson (2011) are implemented using model-free backward elimination. We show how to use the package to select minimal sets of covariates. The package can be used with continuous and discrete covariates and the user can choose between marginal co-ordinate hypothesis tests and kernel-based smoothing as model-free dimension reduction techniques.

This thesis contributes to the research area of causal inference, where estimation of the effect of a treatment on an outcome of interest is the main objective. Some aspects of the estimation of average causal effects in observational studies in general, and case-control studies in particular, are explored.

An important part of estimating causal effects in an observational study is to control for covariates. The first paper of this thesis concerns the selection of minimal covariate sets sufficient for unconfoundedness of the treatment assignment. A data-driven implementation of two covariate selection algorithms is proposed and evaluated.

A common sampling scheme in epidemiology, and when investigating rare events, is the case-control design. In the second paper we study estimators of the marginal causal odds ratio in matched and independent case-control designs. Estimators that, under a logistic regression model, utilize information about the known prevalence of being a case is examined and compared through simulations.

The third paper investigates the particular situation where case-control sampled data is reused to estimate the effect of the case-defining event on an outcome of interest. The consequence of ignoring the design when estimating the average causal effect is discussed and a design-weighted matching estimator is proposed. The performance of the estimator is evaluated with simulation experiments, when matching on the covariates directly and when matching on the propensity score.

The last paper studies the effect of type 1 diabetes mellitus (T1DM) on school achievements using data from the Swedish Childhood Diabetes Register, a population-based incidence register. We apply theoretical results from the second and third papers in the estimation of the average causal effect within the T1DM population. A matching estimator that accounts for the matched case-control design is used.

Estimation of marginal causal effects from case-control data has two complications: (i) confounding due to the fact that the exposure under study is not randomized, and (ii) bias from the case-control sampling scheme. In this paper, we study estimators of the marginal causal odds ratio, addressing these issues for matched and unmatched case-control designs when utilizing the knowledge of the known prevalence of being a case. The estimators are implemented in simulations where their finite sample properties are studied and approximations of their variances are derived with the delta method. Also, we illustrate the methods by analyzing the effect of low birth weight on the risk of type 1 diabetes mellitus using data from the Swedish Childhood Diabetes Register, a nationwide population-based incidence register.