Gender-based violence (GBV)—that is, violence directed against a person because of their gender or violence that affects persons of a particular gender disproportionately—can take various forms, including, most notoriously, domestic violence. Here, we are concerned specifically with physical or sexual violence directed against adult women, who are, jointly with girls, the main victims of GBV. According to the World Health Organisation (WHO), 30% of women worldwide have suffered physical and/or sexual violence at some point in their lives, and tackling this issue is one of the specific development goals of its 2030 agenda. It is clearly also a significant healthcare policy concern, as victims of sexual and physical GBV are more likely to require health services than the general population to address the physical, gynaecological, and psychological consequences of the aggressions suffered. Especially troublingly, under some circumstances, GBV is fatal. In particular, the global prevalence of ever-partnered women aged 15–49 years who have experienced physical or sexual intimate partner violence or both at least once in their lifetime has been recently estimated to be around 27% in a comprehensive systematic review and meta-analysis.

According to the macro survey on GBV violence conducted by the Spanish government in 2019, 21.5% of women in Spain aged 16 and older report having been victims of physical violence while 13.7% report having suffered acts of sexual violence. The findings of this survey stress both the ubiquity and severity of physical and sexual GBV in our society. However, it also showed that only 32% of the victims of physical and sexual GBV had reported the episode of violence to the police or justice system and only 6.78% had gone to primary health services. Indeed, the latter should have a key role to play in the detection of GBV, given that women typically visit a primary healthcare unit at some point in their lives when seeking advice about sexual and reproductive health, and the recently implemented public health policies in Catalonia and Spain are intended to reach the ultimate goal of transforming primary healthcare into the focal point for GBV case screening and detection.

To address the GBV problem more effectively, we need to identify as many cases of GBV as possible. To do so, one of the first questions we must address is why victims of GBV are reluctant to report episodes of violence. Two main reasons are typically forwarded: personal (e.g., shame, fear of retaliation, economic dependence) and societal (e.g., the imbalance in male/female power relations, protecting family privacy, victim blaming). Whatever the reasons for the underreporting of GBV, the principal concern of this article is quantifying the magnitude of cases of gender violence reported in the primary care system and in attempting to reconstruct the complete GBV iceberg. Reducing GBV is at the top of the policy agenda of many governments and organisations and the analysis of these policies and associated campaigns highlights the complexity of the issue and the data biases that beset any discussion of the problem.

In Spain, few studies have attempted to evaluate the impact of policies or political interventions on the occurrence of episodes of GBV and, as far as the authors are aware, no analyses have been undertaken of the impact of non-political interventions (such as training campaigns) on the recording of GBV episodes. Here, in seeking to reconstruct the GBV iceberg, the importance of considering specific events and interventions that have an impact on GBV becomes apparent. Specifically, the article evaluates the impact of the implementation of awareness campaigns among primary healthcare professionals, initiated in late 2019.

Within the period of study, two in-person training activities were conducted in the analysed region. These two activities had an overall duration of 10 hours (6+4), and the focus was on training primary healthcare professionals in detection and prevention of GBV. Given that these and other campaigns and policies have been implemented in recent years, it is crucial to assess their impact on the actual reporting of cases of GBV.

In order to assess the potential underreporting of GBV cases in the primary health care system, the weekly number of cases recorded in the Barcelona Northern Metropolitan Health Region have been used. This article defines underreporting as the difference between the number of cases recorded in the system and the number of cases estimated by the model.

The first question to consider is whether the theoretical model is able to reproduce the process observed accurately, or equivalently, if the assumptions underlying in the model hold. It is evident that the simulated processes are able to capture the variability of the observations accurately.

The proposed modelling approach demonstrates that awareness training for professionals significantly reduces underreporting from an early stage. However, it is crucial to conduct these campaigns repeatedly to sustain their effectiveness.

The following figure shows how the phenomenon has changed over time. It also includes a reconstruction of what most likely happened, based on the median values calculated from the estimated data.

The primary health care system in Catalonia aims to be the main point of support for identifying cases of gender-based violence. A remarkable effort has been made in recent 
years to train professionals to improve their proficiency in detecting cases. However, some specific subpopulations seem to be more difficult to reach. One of these groups is women below age 25, who are still reluctant to seek professional health assistance after suffering an assault. This work suggests that this is the case by comparing the different impact of age on the probability of suffering GBV on the basis of data sources from primary care and from the general population.

To evaluate the impact of age on the risk of experiencing GBV, three supervised learning algorithms were applied to different data sources (on one hand, primary health care, and on the other, macrosurveys in 2015 and 2019): a random forest, a support vector machine, and a neural network.

Although researchers have been using novel data science approaches to better understand GBV, most focus primarily on risk factors. To the best of the authors’ knowledge, this is the first time that advanced statistical modelling techniques have been used to estimate the real magnitude of GBV. The results of the study show that, on average, only around 11% of the victims of GBV between January 2010 and December 2021 in Barcelona’s Northern Metropolitan Health Area who requested primary health care were properly identified as GBV victims. That is, considering the whole period, 3,110 victims were detected while the model estimates that there were a total of 27,264. The severity of this underreporting issue was especially dramatic at the beginning of this period—that is, starting in 2010 and before the professionals responsible for detecting such cases underwent awareness training in 2019—when only 3% of cases were recorded. After the training until the end of the time period under study, 42% of cases were recorded. The impact of this awareness-raising activity can, therefore, be quantified as a 14-fold reduction in the underreporting issue. The results of this and previous studies clearly show how important these training activities are. However, it seems that these activities should be paired with additional efforts to increase awareness about the true scale of GBV.

The results reported herein cannot be considered exclusive to the specific geographical area for which actual data were analysed. On the contrary, these conclusions might be generalisable to any other developed country of similar social characteristics—considering the impact on GBV rates of differences in culture, generation, and population demographics—and may be even worse in developing countries.

GOICOLEA, I. (2023). «What a critical public health perspective can add to the analysis of healthcare responses to gender-based violence that focus on asking». En: BMC Public Health, 23(1), 1738. DOI: 10.1186/s12889-023-16641-4.

GRACIA, E. (2004). «Unreported cases of domestic violence against women: towards an epidemiology of social silence, tolerance, and inhibition». En: Journal of Epidemiology and Community Health, 58(7), 536-537. DOI: 10.1136/jech.2003.019604.

O’LEARY, K. (2022). «Tackling under-reporting of violence against women». En: Nature Medicine. DOI: 10.1038/d41591-022-00083-3.

MORIÑA, D.; MILLÁN, I.; FERNÁNDEZ-FONTELO, A.; PUIG, P.; TORAN, P.; GÓMEZ-MALDONADO, M.;
FALGUERA, G. (2024). «Exploring what lies beneath the tip of the gender-based violence iceberg». En: MedRxiv, preimpresión. DOI: 10.1101/2024.02.26.24303373.

MUÑOZ-SELLÉS, E.; PUJOLAR-DÍAZ, G.; FUSTER-CASANOVAS, A.; MIRÓ CATALINA, Q. (2023). «Detection of gender-based violence in primary care in Central Catalonia: a descriptive cross-sectional study». En: BMC Health Services Research, 23(1), 110. DOI: 10.1186/s12913-023-09091-6.

World Health Organization (2018). «Violence against Women Prevalence Estimates».World Health Organization.