Schizophrenia susceptibility and age of diagnosis — A frailty approach Elisabeth Svensson a, b, ⁎, Maria Rogvin c , Christina M. Hultman d , Ted Reichborn-Kjennerud a, e , Sven Sandin d , Tron A. Moger c, f a Division of Mental Health, Norwegian Institute of Public Health, Norway b Department Clinical Epidemiology, Aarhus University, Denmark c Department of Health Economics and Health Management, Institute of Health and Society, University of Oslo, Norway d Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Sweden e Institute of Psychiatry, University of Oslo, Norway f Department of Biostatistics, Institute of Basal Medical Sciences, University of Oslo, Norway abstract article info Article history: Received 16 November 2012 Received in revised form 11 February 2013 Accepted 5 March 2013 Available online 27 March 2013 Keywords: Age factors Genetic predisposition to disease/epidemiology Incidence Risk factors Schizophrenia Susceptibility Sweden Background: Using a frailty model approach, we aim to evaluate the effect of early-life risk factors on suscep- tibility and age at diagnosis of schizophrenia. We assume paternal age and familial schizophrenia influence the susceptibility, while these and several early risk factors influence the age of diagnosis. Method: Schizophrenia incidence data were derived from the population-based Swedish Patient Registry; in- cluding individuals aged 18 to 45 years, diagnosed between 1974 and 2008. Data were analyzed by a frailty model, a random effects model in survival analysis, using a compound Poisson model. Results: 15,340 incident schizophrenia cases were included. For individuals without familial schizophrenia, a protective effect was seen across most ages of diagnosis for females, low paternal age, born in rural areas, and being born in later cohorts. For individuals with familial schizophrenia, a protective effect is found for females diagnosed between ages 18 and 30 years, corresponding values were 18–25 years for low paternal age. Being born in rural areas and in the last birth cohort was protective for all. The estimated proportion of susceptible was 5% for those without familial schizophrenia and 18% for individuals with familial schizophrenia. There was no statistically significant effect of paternal age on the proportion of susceptible. Discussion: To our knowledge, this is the first regression modeling of time to schizophrenia diagnosis allowing for a non-susceptible fraction of the population, including age dependent modeling of covariate effects and an inter- action. Applying frailty model to schizophrenia provide etiological clues, elucidating patterns of susceptibility and age-at-diagnosis for which early-life factors are of importance. © 2013 Elsevier B.V. All rights reserved. 1. Introduction One important epidemiological feature of schizophrenia is that the age-specific incidence curve increases until it reaches its highest point in the mid-20s, then declines (Suvisaari et al., 1999; Osby et al., 2001; Laursen et al., 2007). A frailty model describes such a downturn as the population effect of mixing two sub-groups: one highly susceptible with high risk, and perhaps early onset, and one non-susceptible, carry- ing almost no risk. By applying a frailty model to schizophrenia, we aim to further advance the understanding of the etiology of this complex disorder. Previously, such model has successfully elucidated the etiolo- gy of other complex disorders, such as for testis cancer (Aalen and Tretli, 1999; Moger et al., 2004) and colorectal cancer (Svensson et al., 2006). As a consequence of the interplay between genetic background (Sullivan et al., 2003) and environmental exposures occurring early in life (Murray and Lewis, 1987), only a minority of the population is sus- ceptible (or frail) to the disorder while the remainder of the population has almost no risk of disease. There are several examples of genetic fac- tors conferring increased susceptibility in schizophrenia(Owen, 2012): individuals with familial schizophrenia have a ten-fold higher risk (Mortensen et al., 1999), and GWAS studies have found a number of common gene variants (SNPs) associated with increased risk (The International Schizophrenia Consortium, 2008; Lee et al., 2012). Addi- tionally, it is hypothesized that the association between advancing paternal age and schizophrenia can be explained by a higher number of rare de novo mutations in paternal germ cells (Malaspina, 2001; Kong et al., 2012), but the direct evidence is lacking (Petersen et al., 2011). Based on the proposed genetic component of these factors, we include them in the frailty distribution, allowing them to influence the proportion of susceptible at birth. This enables us to test whether they influence susceptibility, and also enables us to examine an in- teraction effect, allowing the effect of the other covariates to be different, depending on whether familial schizophrenia or high paternal age is present or not. Several early-life risk factors, for example place of birth, Schizophrenia Research 147 (2013) 140–146 ⁎ Corresponding author at: Division of Mental Health, Norwegian Institute of Public Health, P.O. Box 4404, Nydalen, 0403 Oslo, Norway. Tel.: +45 29883626. E-mail address: elisabeth.svensson@dce.au.dk (E. Svensson). 0920-9964/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.schres.2013.03.004 Contents lists available at SciVerse ScienceDirect Schizophrenia Research journal homepage: www.elsevier.com/locate/schres