Theses Doctoral

Major societal crises and suicide

Martinez-Ales, Gonzalo

Suicide is the leading cause of violent death and a major public health and clinical concern. Globally, suicide mortality has decreased over the last three decades, largely due to dramatic declines in pesticide poisonings in Asia. Recent suicide mortality trends, however, have been heterogeneous, and there have been increases in suicide in several countries and regions (e.g., the United States, Jamaica, Cameroon). Monitoring suicide rates is important for surveillance reasons as well as to generate causal hypotheses, two key components of suicide prevention efforts. Suicide increases following major societal crises, such as economic recessions, are often characterized by heterogeneity across population subgroups – with larger increases among vulnerable groups. Examining subgroups, even if evidence of an increase in suicide overall is absent, can guide identification of at-risk groups and development and implementation of targeted prevention strategies. In Spain, a country with one of the lowest suicide rates across Europe, there has been scientific debate regarding whether suicide increased following the 2008 economic recession. Most recent research suggests that suicide remained largely unchanged, but data are scarce on vulnerable groups among whom the downstream economic effects of the recession might have been more intense than in the general population.

Following the initial COVID-19 pandemic outbreak, there was generalized concern that suicide rates would go up due to increases in bereavement and loss of loved ones, fear of contagion and death, increases in prevalence of mental health conditions, and negative economic effects of the pandemic and contagion control measures. Initial examinations of suicide trends, however, indicated that suicide mortality either remained unchanged or decreased in most locations across the globe during the initial months following the pandemic onset. Subsequent evidence of delayed increases in suicide in specific places (e.g., Japan), however, pointed out the importance of continued monitoring of suicide rates. In addition, there is increasing evidence that suicide rates during the COVID-19 era have changed heterogeneously across sociodemographic groups with higher vulnerability to specific pandemic-related stressors (e.g., higher suicide risk among minoritized people in the United States or women in Japan). There are no systematic reviews examining suicide during the COVID-19 era beyond the initial 6 months of the pandemic, and there has been no systematic assessment of the variation in suicide changes after the onset of the pandemic across place, over time, and across population subgroups. In Spain, there has also been substantial debate regarding the impact of the pandemic on suicide rates: two studies using a suboptimal methodological approach found somewhat contradictory results. No studies have examined suicide among population subgroups during the pandemic in Spain.

The aim of this dissertation is to examine variations in suicide across population groups as defined by sociodemographic characteristics during major societal crises (i.e., the 2008 recession and the COVID-19 pandemic) in Spain, and variations across place, over time, and across sociodemographic groups globally.
The first chapter uses two different approaches to age-period-cohort modelling to examine suicide between 2000 and 2019 in Spain, stratifying analyses by foreign-born status – the most salient marker of disadvantage in Spain, and further analyzing suicide among foreign-born individuals without Spanish citizenship – a proxy for lack of residency permit. I found that, while suicide following the recession remained stable among native-born men, it increased slightly among native-born women – largely due to cohort effects affecting middle-aged women, and markedly among foreign-born individuals – largely due to period effects. Suicide increased especially among foreign-born individuals without Spanish citizenship. Notably, access to specialized healthcare and welfare was interrupted for migrants without residency permit shortly following onset of the recession, in the context of austerity politics undertaken across Europe. These results highlight the moderating role of socioeconomic vulnerability on suicide risk during major economic crises.

The second chapter is a systematic integrative review of the variation of population-based suicide estimates following the initial pandemic outbreak globally. In this review, I examine methodological features of all published studies examining suicide during the COVID-19 pandemic, highlighting the importance of addressing autocorrelation, non-stationarity, and seasonality in studies using an interrupted time-series analysis (ITSA) approach to test a causal question (i.e., to compare observed vs. expected or counterfactual suicide counts or rates). I also provide rationale to expect substantial heterogeneity in a so-called effect of the pandemic on suicide, given multiple versions of the exposure of interest that make it impossible to estimate a sole causal effect. I critically summarize the results overall with a focus on variation across place, over time, and across population subgroups.

My findings indicate substantial geographical heterogeneity; a variable initial period of decreased suicide followed, in several locations, by delayed suicide increases – underscoring the importance of sustained monitoring of rates; and heterogeneity across population subgroups with larger suicide increases among groups at higher risk of suicide contagion and mortality (e.g., older adults, racially minoritized residents) and groups vulnerable to negative economic effects of the pandemic (e.g., groups overrepresented in hospitality and tourism jobs).

The third chapter uses Seasonal Autoregressive Integrated Moving Average (SARIMA) prediction models, an approach to ITSA that can adequately deal with autoregression, non-stationarity, and seasonality, to predict monthly suicide counts between April and December 2020 in Spain had the pandemic not taken place. I do so overall and by sex- and age-group, and by foreign-born status in a set of sensitivity analyses, and I then compare observed vs. predicted suicides to determine if suicide increased. I find generalized higher-than-expected suicide rates during spring and summer of 2020, overall and across subgroups – especially among older males during the summer months. I discuss potential explanations and implications for decision-making of these findings considering the theoretical framework developed in chapters 1 and 2.

Geographic Areas


  • thumnail for MartinezAles_columbia_0054D_17543.pdf MartinezAles_columbia_0054D_17543.pdf application/pdf 3.78 MB Download File

More About This Work

Academic Units
Thesis Advisors
Susser, Ezra S.
Ph.D., Columbia University
Published Here
October 12, 2022