Rethinking Google Searches for Suicide-Related Keywords and Their Association with Suicide Rates, Attempts, and Self-Harm Hospitalisation: An IMV-Model Approach

Rethinking Google Searches for Suicide-Related Keywords and Their Association with Suicide Rates, Attempts, and Self-Harm Hospitalisation: An IMV-Model Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Rethinking Google Searches for Suicide-Related Keywords and Their Association with Suicide Rates, Attempts, and Self-Harm Hospitalisation: An IMV-Model Approach Sandersan Onie, Matthew J. Coleshill, Michelle Tye, Venkatesha Venkatesha, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6137446/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 29 Aug, 2025 Read the published version in npj Digital Medicine → Version 1 posted 9 You are reading this latest preprint version Abstract Background According to the recent WHO Global Health Estimates, the globe is not on track to meet the UN Sustainable Development Goal 3.4.2 of the reduction of suicide, with suicide monitoring being a key issue. Past research has found an association between Google searches for suicide-related keywords and suicide rates, offering a potential tool for rapid monitoring of population suicide rates – although recent findings call this relationship into question. However, the relationship between Google searches and suicide attempts or self-harm has not been investigated. Across three studies, we aimed to ascertain the associations between search volumes for suicide-related keywords and suicide rates, suicide attempts, and self-harm hospitalisation rates, within the IMV-Model of Suicidal Behaviour. Methods Study 1 investigated the relationship between provincial relative search volumes for suicide-related keywords with attempt and suicide rates across Indonesian provinces in 2021. Study 2 investigated the relationship between national relative search volumes for suicide-related keywords with attempt and self-harm hospitalisation rates in Australia between 2008 and 2020. Study 3 investigated the relationship between categories of suicide-related keywords grouped according to the IMV-Model of Suicidal Behaviour, and their relationship with attempt and suicide rates across provinces in Indonesia. Results In studies 1 and 2, we did not observe a significant association between relative search volumes for suicide-related keywords and suicide rates. However, relative search volumes for suicide-related keywords were positively associated with provincial suicide attempt rates in Indonesia and yearly self-harm hospitalisation rates in Australia. Study 3 revealed that keywords associated with distress showed no relationship with attempt or suicide rate, while keywords associated with explicit suicide ideation showed a relationship with attempt only. Keywords associated with specific methods – the volitional component of the IMV-Model - were uniquely associated with both attempt and suicide, with the relationship with suicide rate driven by high-lethality methods keywords. Limitations Ascertaining the quality of data on suicides, suicide attempts, and self-harm incidents is challenging. Moreover, Google Trends has limitations regarding the granularity of data it provides, and it may not fully represent the entire population. Conclusions Our findings suggest that the relationship between search volumes and suicide and attempt rates may depend on the category of keyword, with ‘suicide’ and ‘suicide method’ being associated with suicide rate and self-harm hospitalisation, but not suicide rate. The findings show promise for improved suicide monitoring. Biological sciences/Psychology Health sciences/Health care/Disease prevention Health sciences/Health care/Health policy Health sciences/Health care/Health services Health sciences/Health care/Public health/Epidemiology suicide self-harm Google Trends Australia Indonesia Figures Figure 1 Introduction Suicide is a global health issue, responsible for over 720,000 deaths annually, with over one in every 100 deaths globally being a suicide 1 . Critically, the number of attempts are estimated to be up to 30 times higher than the number of suicides 2 . In response, reducing suicide mortality has been highlighted as a target under United Nations Sustainable Development Goal (UN SDG) 3.4.2 3 . Similarly, the World Health Organization (WHO) Comprehensive Mental Health Action Plan, which supports UN SDG 3.4.2, aims to reduce global suicide deaths by one-third 4 . However, recent WHO Global Health Estimates (GHE) indicate that the world is not on track to meet these goals. To achieve the 2030 target, the rate of reduction would need to double across countries from 1.3% per year to 3.2% per year. A major challenge in addressing suicide mortality is the lack of reliable suicide monitoring. Out of 183 member states, only 60 countries had high-quality vital registration data, with remaining countries relying on modelling using lower quality data if data was provided at all 1 . Without accurate and timely data, countries cannot respond effectively to emerging suicide trends and risk factors. Therefore, innovative approaches are urgently needed to strengthen suicide monitoring—not only to track their progress towards the SDG 3.4.2, but also to implement responsive suicide prevention programs. Previous research has identified a link between suicide rates and Google searches for suicide-related keywords. These studies typically are epidemiological in nature, investigating the association between relative search volumes extracted using Google Trends, commonly for ‘suicide’ and ‘suicide methods’ keywords, and suicide rates across a period of time or across geographic regions 4 – 12 . These findings found correlations that indicated that tracking internet search trends could offer a rapid and cost-effective method for monitoring trends in suicidal behaviour. However, conflicting evidence exists, with other studies failing to find this association 13 – 15 . In particular, a reanalysis of previous studies examining the association between suicide keywords and suicide rate observed mixed findings between countries and concluded that Google Trends data were unreliable in predicting suicide rate 14 . Such disparate findings are perhaps unsurprising due to how factors affecting the transition from ideation to an attempt, and the fatality of the attempt, vary both geographically and temporally. For example, methods used for suicide may vary across urban and rural locations due to the accessibility of different means. As a result, this may obscure the statistical relationship at a macro level. Despite initial enthusiasm for using Google Trends to predict suicide rate, these contradictory findings have raised doubts about its reliability. While most studies focus on suicide rates, very few studies have investigated the relationship between search volumes and self-harm and suicide attempts, and how these keywords map onto existing models of suicidal behaviour. The Integrated-Motivational Volitional Model of Suicidal Behaviour (IMV-Model) 16 suggests that there are three distinct phases to suicide: a pre-motivational phase with background and triggering events; a motivational phase in which the individual experiences defeat and entrapment, and subsequently a suicidal ideation and intent subphase; followed by a volitional phase in which an attempt is made. For every individual, myriad factors may influence whether they transition from ideation to attempt, as well as whether that attempt is fatal. The model, however, does not outline factors pertaining to the fatality of these attempts. Given that internet searches could occur throughout the motivational phase (during distress and ideation), which are more proximal to an attempt than to a suicide death, one hypothesis is that Google searches and self-harm and attempt rates may have a more robust statistical relationship. Across three studies, we investigated the relationship between Google searches and rates of suicidal behaviours in Indonesia and Australia. In the first study, we assessed the relationship between provincial relative search volumes for suicide-related keywords with suicide rate and attempt rate on a geographic dimension in Indonesia for 2021. In the second study, we conceptually replicated the first study by assessing the relationship between national relative search volumes with suicide deaths and hospitalisations due to self-harm in Australia between 2008 and 2020. In the third study, we investigated whether keywords mapping onto phases of the IMV-Model are differentially associated with attempt/self-harm and suicide rate. Study 1 Methods Design We conducted a retrospective, secondary data analysis investigating the relationship between relative search volume for suicide-related keywords across the provinces in Indonesia and attempt and suicide rates. This study was pre-registered on Open Science Framework 17 , with the hypothesis that if suicide rate and suicide attempt rate are positively correlated, then Google search volume would positively correlate with suicide rate through suicide attempt. However, if suicide rate and suicide attempt rate are not positively correlated, then Google search volume would positively correlate with suicide attempt rate but not suicide rate. Given this is secondary data use, Indonesian Psychology Research Ethics through the National Research and Innovation Centre does not require ethics for this study. This study was registered at the University of New South Wales Human Research Ethics Committee portal for secondary data use. Setting The study was conducted using data from Indonesia, an archipelago with over 273.8 million people 18 . Data were analysed on a provincial level, the country’s largest sub-national administrative area grouping. Prior to 2022, the country consisted of 34 provinces, which then became 37 19 . Given the study data in Indonesia was from 2021, we used the 34 provinces in our analysis. Google Searches In line with prior research, we utilised Google Trends 20 to gather Google search volume data. Google Trends is a freely available online tool that provides a relative search volume index ranging from 0 to 100 for specific terms, rather than raw Google search volumes. We examined Google Trends data for the year 2021 to investigate the search patterns related to suicide keywords across all 34 Indonesian provinces. Building on prior studies, we selected a set of keywords related to suicide, suicide methods, and painlessness 12 . We also attempted to include a wider range of terms on suicide methods. These terms were translated to Bahasa Indonesia, the national language of Indonesia, and inputted into Google Trends. The selected suicide keyword was ‘bunuh diri’ (which translates to ‘suicide’ and ‘die by suicide’). The selected suicide methods keywords were ‘cara bunuh diri’ (‘how to suicide’ and ‘suicide methods’). The painlessness keyword was ‘bunuh diri tidak sakit’ (‘painless suicide’ or ‘how to kill myself without pain’). In addition to these keyword categories based on prior work 12 , an additional keyword related to a specific suicide method – ‘gantung diri’ (‘hang myself’) – was added. Trend data for ‘bunuh diri’ was available for all 34 provinces, while ‘cara bunuh diri’ data was available for 31 provinces, and ‘gantung diri’ data was available for 32 provinces and thus was included in the analysis. However, search volume data for ‘bunuh diri tidak sakit’ was only available for 8 provinces, and thus this term was omitted from subsequent analyses due to limited data availability. We also averaged the relative search volumes for ‘bunuh diri’ (suicide) and ‘cara bunuh diri’ (suicide methods) across each province, to provide an aggregated metric. While this follows methods of prior work 12 , the aggregated metrics are not directly comparable due to the different data availability for individual terms. Suicide and Attempt Data Sources The data used for suicide and attempt rate were taken from a previous study we conducted to develop Indonesia’s suicide statistical profile. Full information on data preparation is reported elsewhere 21,22 , and is briefly described here. We obtained province-level suicide attempt and suicide rates for the year 2021. Crude provincial suicide rates were sourced from official police data, which has traditionally served as the authoritative suicide statistics reference 21,22 , given that suicides must be processed within the police system. This data was made available through a collaborative agreement with the Indonesian Ministry of Health and was not open to the public. Provincial suicide attempt rates were extracted from a publicly accessible report issued by the National Bureau of Statistics (Badan Pusat Statistik) 23 . This report stems from a triannual data collection process known as the Village Potential survey, which covers various domains, including economic, agricultural, and health indicators. The report includes the count of total suicide attempts per province. Further details regarding the survey, its methodology, and major findings are available elsewhere 22 . Analysis Plan First, we examined the correlation between annual provincial suicide and attempt rates. Subsequently, we investigated the correlation between search volumes for each of the individual search terms (‘bunuh diri’ [suicide], ‘cara bunuh diri’ [suicide methods], and ‘gantung diri’ [hang myself] and the suicide and attempt rates at the provincial level. The analyses were then repeated using the averaged data from ‘bunuh diri’ and ‘cara bunuh diri’ keywords. Pearson correlation coefficients were calculated in each case for one-way (positive) significance test. All statistical analyses were performed using JASP Statistical Software 24 . Results The analyses showed no evidence of a relationship between provincial suicide and attempt rates (r=0.031, p=0.862). There was mixed evidence of a relationship between provincial relative search volumes and suicide rates, with only ‘gantung diri’ (hang myself) demonstrating a significant positive correlation with suicide rates (r=0.420, p=.008). However, there was a significant positive relationship between the provincial suicide attempt rates and the relative search volumes for ‘bunuh diri’ (suicide; r=0.333, p=.027), ‘cara bunuh diri’ (suicide methods; r=0.649, p<.001), the combined index (r=0.506, p=.001), and ‘gantung diri’ (hang myself; r=0.375, p=.017). Table 1: Correlation of relative search volumes with suicide and attempt rate (at the provincial level in Indonesia, 2021) Keyword Correlation Coefficient Suicide Rate Attempt Rate Bunuh Diri (Suicide, commit suicide) 0.189 0.333* Cara Bunuh Diri (How to suicide/ how to kill yourself/ suicide methods) -0.116 0.649*** Combined 0.021 0.506** Gantung diri (Hang myself) 0.420** 0.375* Note: * indicates p < .05, ** indicates p <.01, *** indicates p<.001 Discussion In Study 1, we examined the relationship between annual provincial relative search volumes for suicide-related keywords and annual suicide and attempt rates. Our analysis revealed no evidence of a relationship between provincial suicide and attempt rates, consistent with the idea that the fatality rate of suicide attempts differs across provinces. Our analyses also revealed consistently significant relationships between the relative search volumes for suicide-related keywords and attempt rates, but not suicide rates – with only ‘gantung diri’ (hang myself) correlating with suicide rates. This pattern of results is consistent with our hypothesis that internet searches, which reflect suicidality, may be closer in proximity and thus have a stronger association with suicide attempts than to suicide rates. Study 2 In Study 2 we attempted to conceptually replicate the findings of Study 1 by assessing the relationship between relative search volumes, and suicide and self-harm rate data across time in Australia. Whereas Study 1 examined a single year of data at a provincial level, Study 2 examines national-level data across 13 years from 2008 to 2020. Methods Design We conducted a retrospective, secondary data analysis investigating the relationship between hospitalisations for self-harm (as a proxy for attempts), suicide rates, and the volume of searches for suicide-related keywords across multiple years in Australia. This study was registered at the University of New South Wales Human Research Ethics Committee portal for secondary data use. Setting The study was conducted using data from Australia 25 , a nation continent with over 26.4 million people. Data were analysed on a national level. Google Searches We extracted national relative search volumes from 2008 to 2020, for the terms ‘suicide’, ‘how to commit suicide’, ‘how to suicide’, ‘how to kill yourself’ ‘painless suicide’, and ‘hang myself’ from Google Trends. Due to a lack of data, Google Trends could not report data for the ‘hang myself’ keyword. Akin to Study 1, we constructed a combined index by averaging the relative search volumes for all extracted keywords. However, given that in Study 1, the index did not include ‘painless suicide’, we also constructed a second combined index to allow comparison with findings from Study 1. Suicide and Self-Harm Hospitalisation Data Sources National rates for self-harm hospitalisation and suicide were obtained per year from 2008 to 2020 from the Australian Institute for Health and Welfare website 26,27 . For hospital-presenting self-harm cases, the Australian Institute for Health and Welfare utilises data from the National Hospital Morbidity Database 26 , which records data on hospitalisations for self-harm with or without suicidal intent. This database does not include individuals who present to community mental health services or general practitioners for self-harm or suicide attempts 27 . For suicide rates, the Australian Institute for Health and Welfare utilises suicide rates obtained from the Australian Bureau of Statistics. If the coronial process concludes that a suicide has occurred, it is recorded by the Australian Bureau of Statistics 27 . Data Preparation Given that correlating temporal data may yield spurious effects, we performed pre-whitening or the Box-Jenkins adjustment on all the yearly relative search volumes, national suicide, and attempt rates as required, following previous studies 14 . An autoregressive integrated moving average (ARIMA) approach was used to examine temporal correlations. For each of the variables, we visually assessed any patterns in the data plot for trends, aiding in the selection of the differencing parameter (d). Following this, correlograms for the autocorrelation function (ACF) and the partial autocorrelation function (PACF) were inspected to inform the candidate model for the autoregressive (AR) and moving average (MA) parameters of the ARIMA model. The candidate model was then compared against similar alternative model specifications using the normalized Bayesian Information Criterion (BIC). To further assess the model adequacy, ACF and PACF correlograms of residuals were inspected to ensure that the residual serial correlations from the selected model specification do not exceed the 95% confidence bands. Once the best-fitting model was identified based on the lowest BIC value and satisfactory residual diagnostics, the residuals were extracted for further analysis. Following that, we conducted cross-correlations on the pre-whitened data, which allowed us to assess the temporal dynamics between the variables. For ARIMA parameters, please see supplementary material. Analysis Plan We first investigated the correlation between national suicide and hospitalisation for self-harm rates per year. We then assessed whether the relative search volume for each keyword correlated with the yearly suicide and self-harm hospitalisation rates. Finally, we repeated the analyses combinations of search terms, with and without ‘painless suicide’. Each cross-correlation was performed on the pre-whitened data, and coefficients were obtained at lag 0. For cross-correlation analyses, p-values are not given and thus not used as a threshold for significance; however, significance is determined on whether the cross-correlation coefficient exceeds the 95% confidence interval surrounding a coefficient of 0. All analyses in this study were conducted in SPSS. Results The cross-correlation analysis showed no evidence of a relationship between yearly suicide and hospitalisation for self-harm rates (r = -0.303; not significant). There was no evidence of a relationship between suicide rate and the yearly relative search volumes for any of the individual or combined keywords (see Table 2). However, there was a significant positive relationship between the yearly self-harm hospitalisation rates and the relative search volumes for ‘suicide’ (r=0.596), ‘how to suicide’ (r=0.801), and the combined index that excluded painless suicide (0.600; see Table 2). Table 2: Correlation of relative search volumes with suicide rate and self-harm hospitalisation rate (2008-2020) Keyword Cross-Correlation Coefficient Suicide Rate Self-Harm Hospitalisation Rate Suicide -0.129 0.596* How to commit suicide 0.217 0.038 How to suicide -0.408 0.801* How to kill yourself -0.154 0.389 Painless suicide -0.156 0.122 Commit Suicide 0.487 -0.188 Combined -0.254 0.500 Combined (excluding ‘Painless suicide’) 0.259 0.600* * Indicates a significant cross-correlation coefficient at time 0 indicated by when the cross-correlation coefficient exceeds the 95% confidence interval around the null. Discussion In Study 2, we investigated the connection between annual relative search volumes for keywords related to suicide and the annual rates for both self-harm hospitalisations and suicide. Our findings did not find a relationship between self-harm hospitalisation and suicide rates, nor between search volumes for any of the keyword categories and suicide rates. However, our results demonstrated significant associations between the search volumes for two keywords and one combination of keywords with self-harm hospitalisation rates. These findings align with our initial hypothesis from Study 1, suggesting that internet searches reflecting feelings of suicidality may be more closely linked to self-harm and suicide attempts rather than to rates of completed suicide. Study 3 While overall, suicide-related search volumes from Google Trends have shown a more reliable association with attempts rather than deaths, the Indonesian keyword translating to ‘hang myself’ in Study 1 showed a relationship with both attempt and suicide. ‘Hang myself’ was the only method-specific keyword included, based on the availability of Google Trends data, which in the IMV-Model 16 would make it more closely associated with the volitional and behavioural phase of the model. Therefore, one possibility is that the association between relative search volume and attempt and suicide rate may depend on what stage the individual is at in the IMV-Model. We used three keyword groupings based on the IMV-Model, with an additional postvention group. The first “distress” group represented the defeat and humiliation, and entrapment subphase of the IMV-Model, and consisted of keywords surrounding distress and proximal factors. The second “ideation” group represented the suicidal ideation and intent subphase of the motivational phase, and consisted of keywords with explicit suicidal ideation. The third “methods” group focused on the transition between the motivational and the volitional phases, specifically on suicide methods mentioned in the volitional moderators, and consisted of keywords about access and use of suicide methods. The last “postvention” group consisted of keywords related to suicide loss and bereavement. Given that each group reflects a stage of cognition with different proximities to attempt and suicide, we hypothesised that keywords in the distress group would correlate with neither attempt nor suicide rates; keywords in the methods group would correlate with both attempt and suicide rates; and postvention keywords would correlate only with suicide rate. No additional hypothesis was tested for ideation keywords, as these keywords were explored in Studies 1 and 2. Methods Design We conducted a retrospective, secondary data analysis investigating the relationships between relative Google Trends search volumes for suicide-related keywords across the provinces in Indonesia, and attempt and suicide rates. The setting, and suicide and attempt data were identical to Study 1. This study, including keywords, extraction methods, and analyses was pre-registered in the Open Science Framework 17 Given this is secondary data use, Indonesian Psychology Research Ethics through the National Research and Innovation Centre does not require ethics for this study. This study was registered at the University of New South Wales Human Research Ethics Committee portal for secondary data use. Google Searches We examined Google Trends data for the year 2021 to investigate the search patterns related to suicide keywords across all 34 Indonesian provinces. The keywords were selected after consulting an advisory group consisting of clinicians and lived-experience advisors in Indonesia. Table 3 outlines the English keywords and their Indonesian counterparts, indicating which were included in the analysis based on data availability (see Data Preparation, below). Due to language differences, some English keywords may have multiple Indonesian translations and vice versa. Table 3: Selection of Keywords used in Study 3 Category English Keyword Indonesian Keyword Number of Datapoints Included Groups Distress - 34 Yes Ideation - 34 Yes Methods - 33 Yes Postvention - 8 No Distress Depression Depresi 34 Yes Sedih banget 21 Yes Stress Stress 34 Yes Stress banget 2 No Giving up Menyerah 34 Yes Menyerah saja 23 Yes Fatigue and Burnout Cape 34 Yes Cape banget 6 No Lelah 34 Yes Aku lelah 33 Yes Ideation Suicide and Commit suicide Bunuh diri 34 Yes How to suicide Cara bunuh diri 31 Yes Suicide methods Want to Die Ingin mati 34 Yes Methods Hanging Gantung diri 32 Yes Hang myself Drink poison Minum racun 24 Yes Poisoning Jumping from Loncat dari 33 Yes Suicide meds Obat bunuh diri 20 Yes Suicide drugs Suicide pills Pil bunuh diri 0 No Overdose Overdosis 31 Yes Postvention Suicide reasons Alasan bunuh diri 8 No Why did they suicide Kenapa mereka bunuh diri 1 No Why do people suicide Kenapa orang bunuh diri 2 No Why did my family member suicide Kenapa keluarga saya bunuh diri 0 No Why did my friend suicide Kenapa teman saya bunuh diri 2 No Why did my parent suicide Kenapa orangtua saya bunuh diri 0 No Note: Number of datapoints refers to how many of the 34 provinces Google Trends was able to provide data for. Data Preparation Similar to Study 1, we also aggregated the individual keywords by averaging the relative search volumes within each group. This yielded relative search volumes for distress, ideation, methods, and postvention keyword groups. Individual or aggregated keywords were retained for analysis if search volume data was available for 20 or more provinces. Missing data were handled using pairwise deletion method, given that the keywords had been selected on data availability, and listwise deletion would lead to too little data for analysis. Insufficient Google Trends data were available for all postvention keywords. As such, this keyword group was excluded from analyses. Analysis Plan The main outcome was to investigate the relationship between the combined search volumes for each keyword category, and attempt and suicide rates. For each keyword group, we investigated whether there was a positive correlation between the search volume and the attempt and suicide rates. If the search volumes for any of these keyword groups were found to correlate with each other, we would assess their relationship with suicide and attempt rate while controlling for the other to encapsulate their unique relationship with the suicide metrics. The secondary outcome was to similarly investigate whether individual keywords were positively associated with attempt and suicide rates, using Pearson correlations testing a one-way positive relationship. All statistical analyses were performed using JASP Statistical Software 23 . Follow-up analysis 1 – method lethality The main analysis identified that the methods keyword group showed a relationship with both suicide rate and attempt rate. We therefore investigated whether ‘high lethality methods’ and ‘low lethality methods’ had different relationships with attempts and suicide rate. We combined the keyword search volumes for ‘high lethality methods’ (hanging/‘gantung diri’, jumping from height/‘loncat dari’) and ‘low lethality methods’ (self-poisoning/‘minim racun’, ‘obat bunuh diri’, and ‘overdosis’), and assessed their positive relationships with suicide and attempt rates, while controlling for the other. Follow-up analysis 2 – loneliness Given that loneliness is a key factor in suicidal thoughts and behaviours 28 but was not included as a term in the pre-registration, we repeated the main analysis, adding loneliness to the distress category. These keywords were ‘kesepian’ and ‘aku kesepian’, translated to ‘loneliness’ and ‘I am lonely’ respectively. Each met criteria for inclusion with ‘kesepian’ having 34 data points and ‘aku kesepian’ having 24 data points. Results The aggregated search volumes for the distress and ideation keyword groups were significantly correlated (r = 0.451, p = .004), therefore their respective correlations with suicide and attempts rates controlled for each other. Neither the distress group or ideation group correlated with the methods group (r=0.179, p=.155; r=0.207, p=.120; respectively). The main analyses revealed that distress keywords, whether combined or individually, did not show a significant positive relationship with attempt or suicide rate (see Table 4). As was already shown in Study 1, the individual ‘bunuh diri’ and ‘cara bunih diri’ ideation keywords showed a relationship with attempt rate, but not suicide rate. The same pattern was observed for the combined ideation group, despite the addition of the ‘ingin mati’ keyword which did not correlate with either attempt or suicide rate. The combined methods keyword group showed a relationship with both suicide and attempt rate. However, there is a mixed pattern with individual keywords showing a relationship with both attempt and suicide, only one outcome, or neither. The results of the correlation analyses can be found in Table 4. Table 4: Correlation of the relative search volumes of IMV category keywords with attempt and suicide rates in Indonesia 2021 Keyword Correlation Coefficient Groups Suicide Rate Attempt Rate Distress (controlling for Ideation) 0.140 -0.189 Ideation (controlling for Distress) -0.061 0.462** Methods 0.317* 0.435** Distress Depresi (Depression) 0.108 -0.077 Sedih banget (Depression) -0.370 -0.024 Stress (Stress) 0.216 -0.231 Menyerah (Giving up) 0.367 0.132 Menyerah saja (Giving up) 0.286 0.17 Cape (Fatigue and burnout) -0.018 -0.289 Lelah (Fatigue and burnout) 0.011 0.157 Aku lelah (Fatigue and burnout) -0.120 0.185 Ideation Bunuh diri (Suicide, commit suicide)† 0.189 0.333* Cara bunuh diri (How to suicide, how to kill yourself, suicide methods) † -0.116 0.649*** Ingin mati (want to die) 0.057 0.153 Methods Gantung diri (Hanging, hang myself)† 0.420** 0.375* Minum racun (Drink poison, poisoning) -0.394 0.459* Loncat dari (Jumping from) 0.500** 0.078 Obat bunuh diri (Suicide meds, suicide drugs) 0.003 0.549** Overdosis (Overdose) 0.139 0.218 Note: * indicates p < .05, ** indicates p <.01, *** indicates p<.001. † Analyses for individual keywords are repeated from Study 1. The follow-up analysis revealed that search volumes for low lethality and high lethality methods were significantly correlated (r = 0.301, p = .05). When controlling for low-lethality methods, high-lethality methods were uniquely associated with both suicide rate (r = 0.562, p < .001) and attempt rate (r = 0.334, p = .031). When controlling for high-lethality methods, low-lethality method search volumes were positively associated with attempt rate (r = 0.409, p = .011), but not suicide rate (r = -0.011, p = .524). Neither of the loneliness search terms significantly correlated with attempts or deaths (kesepian – suicides: r = -0.192; kesepian – attempts: r = 0.057; aku kesepian – suicides: r = 0.061; aku kesepian – attempts: -0.448). Including these terms in the distress keyword group also did not affect the pattern of findings (distress – suicides: r = 0.111; distress – attempts: -0.129). Discussion In Study 3, we investigated the relationship between categories of keywords, grouped according to which phase they broadly map onto in the IMV-Model, and attempt and suicide rates. In line with our hypothesis, the distress keywords showed no relationship with attempt or suicide rates, and method keywords were uniquely associated with both suicide and attempt rates. Due to the lack of data for postvention keywords, it was not possible to validate the hypothesis that this would also be related to suicide rates. Furthermore, when considering the methods keywords based on the method lethality, it appeared that both low and high-lethality keywords had a positive association with attempt rate, but only high-lethality keywords had a positive association with suicide rate. Thus, this suggests that for method-related keywords, the relationship with suicide rate may be driven primarily by high lethality keywords. In sum, the results suggest that keyword categories according to the IMV-Model differentially predict attempt and suicide rate depending on their proximity to the volitional stage of the IMV-Model or attempt and their lethality. General Discussion Across three studies, we investigated the relationship between relative Google Trends search volumes of suicide-related keywords and rates of suicidal behaviours. Our findings provided evidence suggesting that correlations between search volumes and suicide outcomes may differ across stages of the IMV-Model. Specifically for methods-related keywords, this relationship is dependent on the lethality of the method searched. Figure 1 outlines the synthesised findings. Previous studies have reported conflicting findings for whether Google Trends search volumes are associated with suicide rates. A key paper which strongly argues for the absence of this association was published by Tran and colleagues in 2017 14 . The authors reanalysed several past studies investigating temporal associations between suicide keywords and suicide rates, using best practice time series analyses. While the authors found an association between the relative search volume of ‘suicide methods’ and suicide rates in the US, overall, they did not find an association between ‘suicide’, ‘suicide methods’, and other commonly used keywords, and suicide rates across several countries. The authors concluded that Google Trends data were not reliable in predicting suicide rates. We raise three key points, outlining how our findings and the findings from the study do not contradict one another. First, the authors only examined suicide rates and did not include attempt or self-harm rates. Consistent with their findings, we did not find a relationship between ‘suicide’ and ‘suicide methods’ keywords and suicide rate, as Studies 1 and 2 demonstrated a relationship with attempt/self-harm hospitalisation rates. Secondly, the authors only investigated time series data, whereas we include analyses for single time points across multiple geographic regions. Our time series analysis in Study 2 replicated the finding of a relationship with self-harm but not suicide rate, and also found this association using regional data for a single time period (Study 1 and 3). Third, our findings suggest that the relationship between relative search volumes and suicide rates depends on the type of keyword, whereas the earlier study did not categorise the keywords into semantic groupings. Thus, our findings support the previous findings, but also provide new insights. Together, these previous and new findings point towards a non-uniform relationship between relative search volumes of suicide-related keywords and suicide and attempt rates, in which not every suicide-related keyword will have a relationship with suicide and attempt rates. However, our study suggests certain relationships reliably emerge across settings and dimensions. Future research needs to carefully consider which keywords to use, informed by what stage of their ideation the individual is likely to be, and whether a temporal or geographic association is being investigated. Together, these findings suggest that relative search volume may still be a useful method to monitor suicide outcomes at an aggregate level; however, we must carefully consider what suicide outcomes are of interest, which will inform what keywords are used. Further, our findings suggest that keywords commonly used in past studies 12,14 , that is ‘suicide’ and ‘suicide method’, reliably show an association with attempt rather than suicide rate, and are also consistent with previous findings that these keywords were not reliably associated with suicide rate. Future studies may seek to further understand which combination of keywords are best correlated with attempt or suicide rate and validate the keywords using novel data. Given that a suicide is a fatal attempt, it may be reasonable to assume that suicide attempt and death rates would be correlated. However, we did not find this association. One possibility is that the regional and temporal factors which affect the fatality of an attempt – such as availability of lethal means or accessibility of emergency services – are so diverse and varied across time and regions 22 that the added statistical variability obscures this relationship. Furthermore, in Study 3, the relative search volume for methods-related keywords is correlated to attempt rate, with high lethality keywords also correlated to suicide rate. A previous study found that searching for methods-related keywords corresponded to seriously considering suicide 29 , thus having a relationship with attempt rate. However, the relationship between methods and suicide rate may be driven by relative search volumes of particularly lethal methods. Nonetheless, it is still unclear whether higher searches for more lethal methods reflects higher use of these methods. Future research may seek to investigate whether searches for more lethal methods reflects higher use of these methods. Linking search terms and the IMV-Model may help us design better interventions for individuals searching for suicide-related keywords. Our recent work has examined the use of search engine adverts for suicide prevention, in which a person searching for suicide-related keywords is presented with a tailored advert, which when clicked leads to a landing page with resources and help-seeking links 30,31 . One study investigated the effect of using explicit suicide wording in the advert, based on different groups of search keywords including ‘distress’ and ‘ideation’ keywords (called ‘low risk’ and ‘high risk’ respectively in previous study). Using an advert which explicitly mentioned suicide was associated with increased engagement for individuals searching keywords explicitly communicating suicidality 31 (ideation/high-risk). One possible approach is to tailor these campaigns depending on what keywords the individual uses, indicative of which stage the IMV-Model the individual is likely to be. Tailoring based on the IMV-Model may increase the likelihood that a person is met with relevant and helpful information. Our studies have several limitations. Firstly, there are concerns regarding data quality and comprehensiveness. Suicides and suicide attempts are known to be underreported in Indonesia due to stigma and data recording issues 21,22 . In Australia, the self-harm hospitalisation data includes only those admitted to a hospital ward, and thus represents only a proportion of self-harm episodes. Nevertheless, this still represents a high quality and standardised dataset describing self-harm on a national scale. Secondly, the data from Google Trends represents a subset of the total population, in which there may be regional differences in internet accessibility 31 or search engine preferences. Future studies should examine whether, despite these shortcomings, online engagement is widespread enough to allow for representative data and whether adjustments are needed to account for these variations. Third, there are reservations concerning the utility of relative search volumes. Previous studies have highlighted that data from Google Trends might not be completely aligned with intention in relation to a given behaviour 14 . For instance, individuals may conduct suicide-related searches for reasons unrelated to distress, such as academic research. Further, Google Trends data is normalised, and thus does not provide the raw volume of searches, also preventing calculation of weighted averages for combination of keywords. Future studies could aim to replicate this analysis using Google Ads data, which offers ad presentation data which can be multiplied by impression share percentage (the proportion of times the ad was shown compared to its eligible presentations) as a proxy for searches, provides higher temporal and geographic resolution, allows grouping keywords, and permits the exclusion of specific keywords to focus on search intent. Another possibility is using Google’s symptom dataset, which provides greater resolution relative to search volumes for health-related keywords but only for Australia, Ireland, New Zealand, the United Kingdom, and the United States. Fourth, we do not yet know how repeat attempts or self-harm influence information-seeking online. For example, an individual may engage in more information seeking prior to an index attempt, compared with subsequent attempts, leading to the possibility of additional confounders in the data and analysis. Fifth, self-harm and suicide attempts are separate outcomes with different underlying methodologies, and thus Studies 1 and 3, and Study 2 are not direct replications. However, of existing data sources, these are the optimal data sources available to test our hypothesis. Subsequent studies should look to replicating these findings using identical outcomes. Finally, our time series analyses did not investigate different lags in Study 2 as has been done in previous studies. In conclusion, our findings suggest that the relationship between relative search volumes and suicide attempts and deaths is not universal. Instead, keywords correlate differently with suicide outcomes at various stages, broadly aligning with the IMV Model. Future research and prevention efforts should carefully consider the primary outcome of interest (suicide or attempt), the likely state of individuals conducting searches, and the specific keywords used. Nevertheless, across different geographic regions and time periods, and in both Indonesia and Australia, the relationship between search volumes for ‘suicide’ and ‘suicide methods’ and attempt rates appears stable. Since suicide monitoring is a crucial step in addressing SDG 3.4.2, the use of a tool that provides ongoing data, covers regions with limited high-quality suicide data, and maintains a stable relationship with suicide attempts at a macro level warrants further exploration. This is particularly important to improve monitoring accuracy and achieve the rate of reduction needed to meet SDG 3.4.2’s target of reducing mortality by one-third. Declarations Author Contributions Sandersan Onie led the conceptualization, formal analysis, funding acquisition, methodology, writing- original draft, and writing – review & editing. Mark Larsen led supervision, supported conceptualization, funding acquision, and led writing – review & editing. Matthew Coleshill led writing – review & editing. Fiona Shand supported supervision, methodology, funding acquision, and writing – review & editing. Michelle Tye supported funding acquisition, and writing – review & editing. Venkatesha supported methodology and analysis. Funding This research was supported by the Australian Government Department of Health–funded National Suicide Prevention Research Fund, managed by Suicide Prevention Australia to Sandersan Onie; the NHMRC Centre of Research Excellence in Suicide Prevention (APP1152952) to Mark Larsen, Fiona Shand, and Michelle Torok; and Australian Department of Foreign Affairs and Trade, Australian-Indonesian Institute (AII2020322) to Sandersan Onie. Competing Interest Statement SO has provided unpaid consultation to tech companies on suicide prevention policies. All other authors declare no competing financial or non-financial interest. Data Availability The Google search datasets generated or analysed during the current study are available on Google Trends. The Australian suicide and self-harm hospitalisation data is accessible through the Australian Institute of Health and Welfare website. The Indonesian suicide and attempt data is not accessible, and was provided under strict requirements by the Indonesian government. References World Health Organization. Global Health Estimates (2024). Available at: https://www.who.int/data/global-health-estimates. Institute of Medicine (US) Committee on Pathophysiology and Prevention of Adolescent and Adult Suicide. Reducing Suicide: A National Imperative . Edited by Goldsmith, S.K., Pellmar, T.C., Kleinman, A.M. & Bunney, W.E. (National Academies Press, 2002). PMID: 25057611. United Nations. Sustainable Development Goals (2025). Available at: https://sdgs.un.org/goals/goal3. World Health Organization. Comprehensive Mental Health Action Plan 2013–2030 (2021). Available at: https://www.who.int/publications/i/item/9789240031029. Areán, P. A. et al. Perceived utility and characterization of personal Google search histories to detect data patterns proximal to a suicide attempt in individuals who previously attempted suicide: Pilot cohort study. J. Med. Internet Res. 23, e27918 (2021). Sueki, H. Does the volume of internet searches using suicide-related search terms influence the suicide death rate: Data from 2004 to 2009 in Japan. Psychiatry Clin. Neurosci. 65, 392–394 (2011). Yang, A. C., Tsai, S., Huang, N. E. & Peng, C. Association of internet search trends with suicide death in Taipei City, Taiwan, 2004–2009. J. Affect. Disord. 132, 179–184 (2011). Lee, J. Y. Search trends preceding increases in suicide: A cross-correlation study of monthly Google search volume and suicide rate using transfer function models. J. Affect. Disord. 262, 155–164 (2020). Parker, J., Cuthbertson, C., Loveridge, S., Skidmore, M. & Dyar, W. Forecasting state-level premature deaths from alcohol, drugs, and suicides using Google Trends data. J. Affect. Disord. 213, 9–15 (2017). Arora, V. S., Stuckler, D. & McKee, M. Tracking search engine queries for suicide in the United Kingdom. Public Health 137, 147–153 (2016). Kristoufek, L., Moat, H. S. & Preis, T. Estimating suicide occurrence statistics using Google Trends. EPJ Data Sci. 5, 32 (2016). Ma-Kellams, C., Or, F., Baek, J. H. & Kawachi, I. Rethinking suicide surveillance: Google search data and self-reported suicidality differentially estimate completed suicide risk. Clin. Psychol. Sci. 4, 480–484 (2016). Page, A., Chang, S. S. & Gunnell, D. Surveillance of Australian suicidal behaviour using the internet? Aust. N. Z. J. Psychiatry 45, 1020–1022 (2011). Tran, U. S. et al. Low validity of Google Trends for behavioral forecasting of national suicide rates. PLoS ONE 12, e0183149 (2017). Knipe, D., Gunnell, D., Evans, H., John, A. & Fancourt, D. Is Google Trends a useful tool for tracking mental and social distress during a public health emergency? A time-series analysis. J. Affect. Disord. 294, 737–744 (2021). O'Connor, R. C. & Kirtley, O. J. The Integrated Motivational-Volitional Model of suicidal behaviour. Philos. Trans. R. Soc. B Biol. Sci. 373, 20170268 (2018). Onie, S. Google searches and suicide attempt. OSF Pre-registration (2023). https://doi.org/10.17605/OSF.IO/SQPNY Badan Pusat Statistik. Jumlah penduduk hasil proyeksi menurut provinsi dan jenis kelamin (ribu jiwa), 2018–2020. Badan Pusat Statistik (2022). https://www.bps.go.id/indicator/12/1886/1/jumlah-penduduk-hasil-proyeksi-menurut-provinsi-dan-jenis-kelamin.html Kompas.com. Jumlah provinsi di Indonesia ada 38, ini daftar dan ibu kotanya. Kompas.com (2023). https://www.kompas.com/tren/read/2023/04/11/074500965/jumlah-provinsi-di-indonesia-ada-38-ini-daftar-dan-ibu-kotanya Google. Google Trends. Google (2024). https://trends.google.com/trends/ Onie, S. et al. Indonesian first national suicide prevention strategy: Key findings from the qualitative situational analysis. Lancet Reg. Health Southeast Asia 2023, 100245 (2023). Onie, S. et al. Indonesia’s first suicide statistics profile: An analysis of suicide and attempt rates, underreporting, geographic distribution, gender, method, and rurality. Lancet Reg. Health Southeast Asia (in press). Badan Pusat Statistik. Village Potential Statistics of Indonesia (2021). https://www.bps.go.id/publication/2022/03/24/ceab4ec9f942b1a4fdf4cd08/statistik-potensi-desa-indonesia-2021.html JASP Team. JASP (Version 0.17.3) (2023). Australian Bureau of Statistics. Population. Australian Bureau of Statistics (2023). https://www.abs.gov.au/statistics/people/population Australian Bureau of Statistics. Intentional self-harm deaths (suicide) in Australia. Australian Bureau of Statistics (2023). https://www.abs.gov.au/statistics/health/causes-death/causes-death-australia/latest-release#intentional-self-harm-deaths-suicide-in-australia Australian Bureau of Statistics. Causes of death Australia. Australian Bureau of Statistics (2024). https://www.abs.gov.au/statistics/health/causes-death/causes-death-australia/latest-release McClelland, H., Evans, J. J., Nowland, R., Ferguson, E. & O'Connor, R. C. Loneliness as a predictor of suicidal ideation and behaviour: A systematic review and meta-analysis of prospective studies. J. Affect. Disord. 2024, in press. Additional Declarations Competing interest reported. SO has provided unpaid consultation to tech companies on suicide prevention policies. All other authors declare no competing financial or non-financial interest. Supplementary Files AttemptManuscriptSupp.pdf Cite Share Download PDF Status: Published Journal Publication published 29 Aug, 2025 Read the published version in npj Digital Medicine → Version 1 posted Editorial decision: Revision requested 30 Apr, 2025 Reviews received at journal 28 Apr, 2025 Reviewers agreed at journal 17 Apr, 2025 Reviews received at journal 17 Mar, 2025 Reviewers agreed at journal 08 Mar, 2025 Reviewers invited by journal 07 Mar, 2025 Editor assigned by journal 05 Mar, 2025 Submission checks completed at journal 05 Mar, 2025 First submitted to journal 01 Mar, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6137446","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":424627265,"identity":"02f42b0f-8e1a-4fe2-8fee-b240176f39cd","order_by":0,"name":"Sandersan Onie","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9ElEQVRIiWNgGAWjYLCCBAabBD4GBmYGHhK0pCWwkaaFgeEwCVrkG5iPPXhQcT6Pjf3wY4M3FbUM/NLHLzD8bMOtxeAAW7pBwpnbxWw8acaJc84cZ5Dsyylg7MWnhYHHTCKx7XZiG0MO82HetmMMBmd4Ehh48WiRbwBp+XcusY3/DUIL4188WhgOgLQ0HEhsk8hhTuZtqwFqYT/AjM8Wg8NsaRIJx5KL2SSeGRvOOXOAR7KHh+GwzDk8DmtvPib5o8Yuj58/+bHEm4o6OX4e9ocP35ThcRgzKvcwMGp4DA7g0YAB6oCY/QEpOkbBKBgFo2D4AwDSvEqREsRf/AAAAABJRU5ErkJggg==","orcid":"","institution":"Black Dog Institute","correspondingAuthor":true,"prefix":"","firstName":"Sandersan","middleName":"","lastName":"Onie","suffix":""},{"id":424627266,"identity":"d17a410a-e6cd-4862-a5ef-fed4ac441d49","order_by":1,"name":"Matthew J. Coleshill","email":"","orcid":"","institution":"Black Dog Institute","correspondingAuthor":false,"prefix":"","firstName":"Matthew","middleName":"J.","lastName":"Coleshill","suffix":""},{"id":424627267,"identity":"a9052ac0-2693-48a1-b0a7-d0fd4d1ea6f8","order_by":2,"name":"Michelle Tye","email":"","orcid":"","institution":"Black Dog Institute","correspondingAuthor":false,"prefix":"","firstName":"Michelle","middleName":"","lastName":"Tye","suffix":""},{"id":424627268,"identity":"90e05b8a-2f99-4c0e-a50e-9b7b4168e1c2","order_by":3,"name":"Venkatesha Venkatesha","email":"","orcid":"","institution":"UNSW Sydney","correspondingAuthor":false,"prefix":"","firstName":"Venkatesha","middleName":"","lastName":"Venkatesha","suffix":""},{"id":424627269,"identity":"935ae592-5066-48dd-96cc-88bd5705be1f","order_by":4,"name":"Fiona Shand","email":"","orcid":"","institution":"Black Dog Institute","correspondingAuthor":false,"prefix":"","firstName":"Fiona","middleName":"","lastName":"Shand","suffix":""},{"id":424627270,"identity":"a9aa2d84-3693-42fd-9526-8788c4753d8e","order_by":5,"name":"Mark Larsen","email":"","orcid":"","institution":"Centre for Big Data Research in Health, UNSW Sydney","correspondingAuthor":false,"prefix":"","firstName":"Mark","middleName":"","lastName":"Larsen","suffix":""}],"badges":[],"createdAt":"2025-03-02 05:08:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6137446/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6137446/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41746-025-01916-4","type":"published","date":"2025-08-29T15:57:28+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":77984744,"identity":"33566afc-efde-47f5-9b79-91c3f8e29a69","added_by":"auto","created_at":"2025-03-07 13:33:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":111019,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic of the synthesised findings from Study 3. The schematic illustrates how keyword groups were derived from corresponding IMV-Model components and correlated with attempt and suicide rates (* indicates p \u0026lt; .05, ** indicates p \u0026lt;.01, *** indicates p\u0026lt;.001).\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6137446/v1/38e3560f4a34dac39285d48f.png"},{"id":90344980,"identity":"11365208-b0f2-439b-998e-2c233fd15a0a","added_by":"auto","created_at":"2025-09-01 16:08:45","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1065926,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6137446/v1/27291d1b-a1ed-4077-9bf5-465fe8b63dca.pdf"},{"id":77984738,"identity":"4e5b142c-d683-4ea0-8da6-1c51dbbd9682","added_by":"auto","created_at":"2025-03-07 13:32:59","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":33618,"visible":true,"origin":"","legend":"","description":"","filename":"AttemptManuscriptSupp.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6137446/v1/94a8235860f23990cd8063ec.pdf"}],"financialInterests":"Competing interest reported. SO has provided unpaid consultation to tech companies on suicide prevention policies. All other authors declare no competing financial or non-financial interest.","formattedTitle":"Rethinking Google Searches for Suicide-Related Keywords and Their Association with Suicide Rates, Attempts, and Self-Harm Hospitalisation: An IMV-Model Approach","fulltext":[{"header":"Introduction","content":"\u003cp\u003eSuicide is a global health issue, responsible for over 720,000 deaths annually, with over one in every 100 deaths globally being a suicide\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Critically, the number of attempts are estimated to be up to 30 times higher than the number of suicides\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. In response, reducing suicide mortality has been highlighted as a target under United Nations Sustainable Development Goal (UN SDG) 3.4.2\u003csup\u003e3\u003c/sup\u003e. Similarly, the World Health Organization (WHO) Comprehensive Mental Health Action Plan, which supports UN SDG 3.4.2, aims to reduce global suicide deaths by one-third\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. However, recent WHO Global Health Estimates (GHE) indicate that the world is not on track to meet these goals. To achieve the 2030 target, the rate of reduction would need to double across countries from 1.3% per year to 3.2% per year. A major challenge in addressing suicide mortality is the lack of reliable suicide monitoring. Out of 183 member states, only 60 countries had high-quality vital registration data, with remaining countries relying on modelling using lower quality data if data was provided at all\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Without accurate and timely data, countries cannot respond effectively to emerging suicide trends and risk factors. Therefore, innovative approaches are urgently needed to strengthen suicide monitoring\u0026mdash;not only to track their progress towards the SDG 3.4.2, but also to implement responsive suicide prevention programs.\u003c/p\u003e \u003cp\u003ePrevious research has identified a link between suicide rates and Google searches for suicide-related keywords. These studies typically are epidemiological in nature, investigating the association between relative search volumes extracted using Google Trends, commonly for \u0026lsquo;suicide\u0026rsquo; and \u0026lsquo;suicide methods\u0026rsquo; keywords, and suicide rates across a period of time or across geographic regions\u003csup\u003e\u003cspan additionalcitationids=\"CR5 CR6 CR7 CR8 CR9 CR10 CR11\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. These findings found correlations that indicated that tracking internet search trends could offer a rapid and cost-effective method for monitoring trends in suicidal behaviour. However, conflicting evidence exists, with other studies failing to find this association\u003csup\u003e\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. In particular, a reanalysis of previous studies examining the association between suicide keywords and suicide rate observed mixed findings between countries and concluded that Google Trends data were unreliable in predicting suicide rate\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Such disparate findings are perhaps unsurprising due to how factors affecting the transition from ideation to an attempt, and the fatality of the attempt, vary both geographically and temporally. For example, methods used for suicide may vary across urban and rural locations due to the accessibility of different means. As a result, this may obscure the statistical relationship at a macro level. Despite initial enthusiasm for using Google Trends to predict suicide rate, these contradictory findings have raised doubts about its reliability.\u003c/p\u003e \u003cp\u003eWhile most studies focus on suicide rates, very few studies have investigated the relationship between search volumes and self-harm and suicide attempts, and how these keywords map onto existing models of suicidal behaviour. The Integrated-Motivational Volitional Model of Suicidal Behaviour (IMV-Model)\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e suggests that there are three distinct phases to suicide: a pre-motivational phase with background and triggering events; a motivational phase in which the individual experiences defeat and entrapment, and subsequently a suicidal ideation and intent subphase; followed by a volitional phase in which an attempt is made. For every individual, myriad factors may influence whether they transition from ideation to attempt, as well as whether that attempt is fatal. The model, however, does not outline factors pertaining to the fatality of these attempts. Given that internet searches could occur throughout the motivational phase (during distress and ideation), which are more proximal to an attempt than to a suicide death, one hypothesis is that Google searches and self-harm and attempt rates may have a more robust statistical relationship.\u003c/p\u003e \u003cp\u003eAcross three studies, we investigated the relationship between Google searches and rates of suicidal behaviours in Indonesia and Australia. In the first study, we assessed the relationship between provincial relative search volumes for suicide-related keywords with suicide rate and attempt rate on a geographic dimension in Indonesia for 2021. In the second study, we conceptually replicated the first study by assessing the relationship between national relative search volumes with suicide deaths and hospitalisations due to self-harm in Australia between 2008 and 2020. In the third study, we investigated whether keywords mapping onto phases of the IMV-Model are differentially associated with attempt/self-harm and suicide rate.\u003c/p\u003e"},{"header":"Study 1","content":"\u003ch2\u003eMethods\u003c/h2\u003e\n\u003ch3\u003eDesign\u003c/h3\u003e\n\u003cp\u003eWe conducted a retrospective, secondary data analysis investigating the relationship between relative search volume for suicide-related keywords across the provinces in Indonesia and attempt and suicide rates. This study was pre-registered on Open Science Framework\u003csup\u003e17\u003c/sup\u003e, with the hypothesis that if suicide rate and suicide attempt rate are positively correlated, then Google search volume would positively correlate with suicide rate through suicide attempt. However, if suicide rate and suicide attempt rate are not positively correlated, then Google search volume would positively correlate with suicide attempt rate but not suicide rate. Given this is secondary data use, Indonesian Psychology Research Ethics through the National Research and Innovation Centre does not require ethics for this study. This study was registered at the University of New South Wales Human Research Ethics Committee portal for secondary data use.\u003c/p\u003e\n\u003ch3\u003eSetting \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/h3\u003e\n\u003cp\u003eThe study was conducted using data from Indonesia, an archipelago with over 273.8 million people\u003csup\u003e18\u003c/sup\u003e. Data were analysed on a provincial level, the country\u0026rsquo;s largest sub-national administrative area grouping. Prior to 2022, the country consisted of 34 provinces, which then became 37\u003csup\u003e\u0026nbsp;19\u003c/sup\u003e. Given the study data in Indonesia was from 2021, we used the 34 provinces in our analysis.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eGoogle Searches\u003c/h3\u003e\n\u003cp\u003eIn line with prior research, we utilised Google Trends\u003csup\u003e20\u003c/sup\u003e to gather Google search volume data. Google Trends is a freely available online tool that provides a relative search volume index ranging from 0 to 100 for specific terms, rather than raw Google search volumes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe examined Google Trends data for the year 2021 to investigate the search patterns related to suicide keywords across all 34 Indonesian provinces. Building on prior studies, we selected a set of keywords related to suicide, suicide methods, and painlessness\u003csup\u003e12\u003c/sup\u003e. We also attempted to include a wider range of terms on suicide methods. These terms were translated to Bahasa Indonesia, the national language of Indonesia, and inputted into Google Trends.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe selected suicide keyword was \u0026lsquo;bunuh diri\u0026rsquo; (which translates to \u0026lsquo;suicide\u0026rsquo; and \u0026lsquo;die by suicide\u0026rsquo;). The selected suicide methods keywords were \u0026lsquo;cara bunuh diri\u0026rsquo; (\u0026lsquo;how to suicide\u0026rsquo; and \u0026lsquo;suicide methods\u0026rsquo;). The painlessness keyword was \u0026lsquo;bunuh diri tidak sakit\u0026rsquo; (\u0026lsquo;painless suicide\u0026rsquo; or \u0026lsquo;how to kill myself without pain\u0026rsquo;). In addition to these keyword categories based on prior work\u003csup\u003e12\u003c/sup\u003e, an additional keyword related to a specific suicide method \u0026ndash; \u0026lsquo;gantung diri\u0026rsquo; (\u0026lsquo;hang myself\u0026rsquo;) \u0026ndash; was added.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTrend data for \u0026lsquo;bunuh diri\u0026rsquo; was available for all 34 provinces, while \u0026lsquo;cara bunuh diri\u0026rsquo; data was available for 31 provinces, and \u0026lsquo;gantung diri\u0026rsquo; data was available for 32 provinces and thus was included in the analysis. However, search volume data for \u0026lsquo;bunuh diri tidak sakit\u0026rsquo; was only available for 8 provinces, and thus this term was omitted from subsequent analyses due to limited data availability.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe also averaged the relative search volumes for \u0026lsquo;bunuh diri\u0026rsquo; (suicide) and \u0026lsquo;cara bunuh diri\u0026rsquo; (suicide methods) across each province, to provide an aggregated metric. While this follows methods of prior work\u003csup\u003e12\u003c/sup\u003e, the aggregated metrics are not directly comparable due to the different data availability for individual terms. \u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eSuicide and Attempt Data Sources\u003c/h3\u003e\n\u003cp\u003eThe data used for suicide and attempt rate were taken from a previous study we conducted to develop Indonesia\u0026rsquo;s suicide statistical profile. Full information on data preparation is reported elsewhere\u003csup\u003e\u0026nbsp;21,22\u003c/sup\u003e, and is briefly\u003csup\u003e\u0026nbsp;\u003c/sup\u003edescribed here.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe obtained province-level suicide attempt and suicide rates for the year 2021. Crude provincial suicide rates were sourced from official police data, which has traditionally served as the authoritative suicide statistics reference\u003csup\u003e21,22\u003c/sup\u003e, given that suicides must be processed within the police system. This data was made available through a collaborative agreement with the Indonesian Ministry of Health and was not open to the public.\u003c/p\u003e\n\u003cp\u003eProvincial suicide attempt rates were extracted from a publicly accessible report issued by the National Bureau of Statistics (Badan Pusat Statistik)\u003csup\u003e23\u003c/sup\u003e. This report stems from a triannual data collection process known as the Village Potential survey, which covers various domains, including economic, agricultural, and health indicators. The report includes the count of total suicide attempts per province. Further details regarding the survey, its methodology, and major findings are available elsewhere\u003csup\u003e22\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eAnalysis Plan\u003c/h3\u003e\n\u003cp\u003eFirst, we examined the correlation between annual provincial suicide and attempt rates. Subsequently, we investigated the correlation between search volumes for each of the individual search terms (\u0026lsquo;bunuh diri\u0026rsquo; [suicide], \u0026lsquo;cara bunuh diri\u0026rsquo; [suicide methods], and \u0026lsquo;gantung diri\u0026rsquo; [hang myself] and the suicide and attempt rates at the provincial level. \u0026nbsp; The analyses were then repeated using the averaged data from \u0026lsquo;bunuh diri\u0026rsquo; and \u0026lsquo;cara bunuh diri\u0026rsquo; keywords. Pearson correlation coefficients were calculated in each case for one-way (positive) significance test. All statistical analyses were performed using JASP Statistical Software\u003csup\u003e24\u003c/sup\u003e.\u003c/p\u003e\n\u003ch2\u003eResults\u003c/h2\u003e\n\u003cp\u003eThe analyses showed no evidence of a relationship between provincial suicide and attempt rates (r=0.031, p=0.862).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThere was mixed evidence of a relationship between provincial relative search volumes and suicide rates, with only \u0026lsquo;gantung diri\u0026rsquo; (hang myself) demonstrating a significant positive correlation with suicide rates (r=0.420, p=.008). However, there was a significant positive relationship between the provincial suicide attempt rates and the relative search volumes for \u0026lsquo;bunuh diri\u0026rsquo; (suicide; r=0.333, p=.027), \u0026lsquo;cara bunuh diri\u0026rsquo; (suicide methods; r=0.649, p\u0026lt;.001), the combined index (r=0.506, p=.001), and \u0026lsquo;gantung diri\u0026rsquo; (hang myself; r=0.375, p=.017).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 1: Correlation of relative search volumes with suicide and attempt rate (at the provincial level in Indonesia, 2021)\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"642\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 444px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKeyword\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 198px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCorrelation Coefficient\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 444px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eSuicide Rate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eAttempt Rate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 444px;\"\u003e\n \u003cp\u003eBunuh Diri (Suicide, commit suicide)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.333*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 444px;\"\u003e\n \u003cp\u003eCara Bunuh Diri (How to suicide/ how to kill yourself/ suicide methods)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e-0.116\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.649***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 444px;\"\u003e\n \u003cp\u003eCombined\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.506**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 444px;\"\u003e\n \u003cp\u003eGantung diri (Hang myself)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.420**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.375*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eNote: * indicates p \u0026lt; .05, ** indicates p \u0026lt;.01, *** indicates p\u0026lt;.001\u003c/p\u003e\n\u003ch2\u003eDiscussion\u003c/h2\u003e\n\u003cp\u003eIn Study 1, we examined the relationship between annual provincial relative search volumes for suicide-related keywords and annual suicide and attempt rates. Our analysis revealed no evidence of a relationship between provincial suicide and attempt rates, consistent with the idea that the fatality rate of suicide attempts differs across provinces. Our analyses also revealed consistently significant relationships between the relative search volumes for suicide-related keywords and attempt rates, but not suicide rates \u0026ndash; with only \u0026lsquo;gantung diri\u0026rsquo; (hang myself) correlating with suicide rates. This pattern of results is consistent with our hypothesis that internet searches, which reflect suicidality, may be closer in proximity and thus have a stronger association with suicide attempts than to suicide rates. \u0026nbsp;\u0026nbsp;\u003c/p\u003e"},{"header":"Study 2","content":"\u003cp\u003eIn Study 2 we attempted to conceptually replicate the findings of Study 1 by assessing the relationship between relative search volumes, and suicide and self-harm rate data across time in Australia. Whereas Study 1 examined a single year of data at a provincial level, Study 2 examines national-level data across 13 years from 2008 to 2020.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eMethods\u003c/h2\u003e\n\u003ch3\u003eDesign\u003c/h3\u003e\n\u003cp\u003eWe conducted a retrospective, secondary data analysis investigating the relationship between hospitalisations for self-harm (as a proxy for attempts), suicide rates, and the volume of searches for suicide-related keywords across multiple years in Australia. This study was registered at the University of New South Wales Human Research Ethics Committee portal for secondary data use.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eSetting\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/h3\u003e\n\u003cp\u003eThe study was conducted using data from Australia\u003csup\u003e25\u003c/sup\u003e, a nation continent with over 26.4 million people. Data were analysed on a national level. \u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eGoogle Searches\u003c/h3\u003e\n\u003cp\u003eWe extracted national relative search volumes from 2008 to 2020, for the terms \u0026lsquo;suicide\u0026rsquo;, \u0026lsquo;how to commit suicide\u0026rsquo;, \u0026lsquo;how to suicide\u0026rsquo;, \u0026lsquo;how to kill yourself\u0026rsquo; \u0026lsquo;painless suicide\u0026rsquo;, and \u0026lsquo;hang myself\u0026rsquo; from Google Trends. Due to a lack of data, Google Trends could not report data for the \u0026lsquo;hang myself\u0026rsquo; keyword.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAkin to Study 1, we constructed a combined index by averaging the relative search volumes for all extracted keywords. However, given that in Study 1, the index did not include \u0026lsquo;painless suicide\u0026rsquo;, we also constructed a second combined index to allow comparison with findings from Study 1.\u003c/p\u003e\n\u003ch3\u003eSuicide and Self-Harm Hospitalisation Data Sources\u003c/h3\u003e\n\u003cp\u003eNational rates for self-harm hospitalisation and suicide were obtained per year from 2008 to 2020 from the Australian Institute for Health and Welfare website\u003csup\u003e26,27\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFor hospital-presenting self-harm cases, the Australian Institute for Health and Welfare utilises data from the National Hospital Morbidity Database\u003csup\u003e26\u003c/sup\u003e, which records data on hospitalisations for self-harm with or without suicidal intent. This database does not include individuals who present to community mental health services or general practitioners for self-harm or suicide attempts\u003csup\u003e27\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor suicide rates,\u0026nbsp;the Australian Institute for Health and Welfare utilises\u0026nbsp;suicide rates obtained from the Australian Bureau of Statistics. If the coronial process concludes that a suicide has occurred, it is recorded by the Australian Bureau of Statistics\u003csup\u003e27\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eData Preparation\u003c/h3\u003e\n\u003cp\u003eGiven that correlating temporal data may yield spurious effects, we performed pre-whitening or the Box-Jenkins adjustment on all the yearly relative search volumes, national suicide, and attempt rates as required, following previous studies\u003csup\u003e14\u003c/sup\u003e.\u0026nbsp;\u0026nbsp;An autoregressive integrated moving average (ARIMA) approach was used to examine temporal correlations. For each of the variables, we visually assessed any patterns in the data plot for trends, aiding in the selection of the differencing parameter (d). Following this, correlograms for the autocorrelation function (ACF) and the partial autocorrelation function (PACF) were inspected to inform the candidate model for the autoregressive (AR) and moving average (MA) parameters of the ARIMA model. The candidate model was then compared against similar alternative model specifications using the normalized Bayesian Information Criterion (BIC). To further assess the model adequacy, ACF and PACF correlograms of residuals were inspected to ensure that the residual serial correlations from the selected model specification do not exceed the 95% confidence bands. Once the best-fitting model was identified based on the lowest BIC value and satisfactory residual diagnostics, the residuals were extracted for further analysis. Following that, we conducted cross-correlations on the pre-whitened data, which allowed us to assess the temporal dynamics between the variables. For ARIMA parameters, please see supplementary material.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eAnalysis Plan\u003c/h3\u003e\n\u003cp\u003eWe first investigated the correlation between national suicide and hospitalisation for self-harm rates per year. We then assessed whether the relative search volume for each keyword correlated with the yearly suicide and self-harm hospitalisation rates. Finally, we repeated the analyses combinations of search terms, with and without \u0026lsquo;painless suicide\u0026rsquo;. Each cross-correlation was performed on the pre-whitened data, and coefficients were obtained at lag 0. For cross-correlation analyses, p-values are not given and thus not used as a threshold for significance; however, significance is determined on whether the cross-correlation coefficient exceeds the 95% confidence interval surrounding a coefficient of 0. All analyses in this study were conducted in SPSS.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eResults\u003c/h2\u003e\n\u003cp\u003eThe cross-correlation analysis showed no evidence of a relationship between yearly suicide and hospitalisation for self-harm rates (r = -0.303; not significant).\u003c/p\u003e\n\u003cp\u003eThere was no evidence of a relationship between suicide rate and the yearly relative search volumes for any of the individual or combined keywords (see Table 2). However, there was a significant positive relationship between the yearly self-harm hospitalisation rates and the relative search volumes for \u0026lsquo;suicide\u0026rsquo; (r=0.596), \u0026lsquo;how to suicide\u0026rsquo; (r=0.801), and the combined index that excluded painless suicide (0.600; see Table 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2: Correlation of relative search volumes with suicide rate and self-harm hospitalisation rate (2008-2020)\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"623\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKeyword\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 368px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCross-Correlation Coefficient\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eSuicide Rate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eSelf-Harm Hospitalisation Rate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eSuicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e-0.129\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.596*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eHow to commit suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e0.217\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eHow to suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e-0.408\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.801*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eHow to kill yourself\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e-0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.389\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003ePainless suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e-0.156\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.122\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eCommit Suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e0.487\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e-0.188\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eCombined\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e-0.254\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eCombined (excluding \u0026lsquo;Painless suicide\u0026rsquo;)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 161px;\"\u003e\n \u003cp\u003e0.259\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e0.600*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e* Indicates a significant cross-correlation coefficient at time 0 indicated by when the cross-correlation coefficient exceeds the 95% confidence interval around the null.\u003c/p\u003e\n\u003ch2\u003eDiscussion\u003c/h2\u003e\n\u003cp\u003eIn Study 2, we investigated the connection between annual relative search volumes for keywords related to suicide and the annual rates for both self-harm hospitalisations and suicide. Our findings did not find a relationship between self-harm hospitalisation and suicide rates, nor between search volumes for any of the keyword categories and suicide rates. However, our results demonstrated significant associations between the search volumes for two keywords and one combination of keywords with self-harm hospitalisation rates. These findings align with our initial hypothesis from Study 1, suggesting that internet searches reflecting feelings of suicidality may be more closely linked to self-harm and suicide attempts rather than to rates of completed suicide.\u0026nbsp;\u003c/p\u003e"},{"header":"Study 3","content":"\u003cp\u003eWhile overall, suicide-related search volumes from Google Trends have shown a more reliable association with attempts rather than deaths, the Indonesian keyword translating to \u0026lsquo;hang myself\u0026rsquo; in Study 1 showed a relationship with both attempt and suicide. \u0026lsquo;Hang myself\u0026rsquo; was the only method-specific keyword included, based on the availability of Google Trends data, which in the IMV-Model\u003csup\u003e16\u003c/sup\u003e would make it more closely associated with the volitional and behavioural phase of the model. Therefore, one possibility is that the association between relative search volume and attempt and suicide rate may depend on what stage the individual is at in the IMV-Model.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe used three keyword groupings based on the IMV-Model, with an additional postvention group. The first \u0026ldquo;distress\u0026rdquo; group represented the defeat and humiliation, and entrapment subphase of the IMV-Model, and consisted of keywords surrounding distress and proximal factors. The second \u0026ldquo;ideation\u0026rdquo; group represented the suicidal ideation and intent subphase of the motivational phase, and consisted of keywords with explicit suicidal ideation. The third \u0026ldquo;methods\u0026rdquo; group focused on the transition between the motivational and the volitional phases, specifically on suicide methods mentioned in the volitional moderators, and consisted of keywords about access and use of suicide methods. The last \u0026ldquo;postvention\u0026rdquo; group consisted of keywords related to suicide loss and bereavement.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGiven that each group reflects a stage of cognition with different proximities to attempt and suicide, we hypothesised that keywords in the distress group would correlate with neither attempt nor suicide rates; keywords in the methods group would correlate with both attempt and suicide rates; and postvention keywords would correlate only with suicide rate. No additional hypothesis was tested for ideation keywords, as these keywords were explored in Studies 1 and 2.\u003c/p\u003e\n\u003ch2\u003eMethods\u003c/h2\u003e\n\u003ch3\u003eDesign\u003c/h3\u003e\n\u003cp\u003eWe conducted a retrospective, secondary data analysis investigating the relationships between relative Google Trends search volumes for suicide-related keywords across the provinces in Indonesia, and attempt and suicide rates. The setting, and suicide and attempt data were identical to Study 1. This study, including keywords, extraction methods, and analyses was pre-registered in the Open Science Framework\u003csup\u003e17\u003c/sup\u003e Given this is secondary data use, Indonesian Psychology Research Ethics through the National Research and Innovation Centre does not require ethics for this study. This study was registered at the University of New South Wales Human Research Ethics Committee portal for secondary data use.\u003c/p\u003e\n\u003ch3\u003eGoogle Searches\u003c/h3\u003e\n\u003cp\u003eWe examined Google Trends data for the year 2021 to investigate the search patterns related to suicide keywords across all 34 Indonesian provinces. The keywords were selected after consulting an advisory group consisting of clinicians and lived-experience advisors in Indonesia. Table 3 outlines the English keywords and their Indonesian counterparts, indicating which were included in the analysis based on data availability (see Data Preparation, below). Due to language differences, some English keywords may have multiple Indonesian translations and vice versa.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 3: Selection of Keywords used in Study 3\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"685\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCategory\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEnglish Keyword\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIndonesian Keyword\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber of Datapoints\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 67px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIncluded\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003eGroups\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eDistress\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eIdeation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eMethods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003ePostvention\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"10\" valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003eDistress\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eDepression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eDepresi\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eSedih banget\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eStress\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eStress\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eStress banget\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eGiving up\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eMenyerah\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eMenyerah saja\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"4\" valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eFatigue and Burnout\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCape\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCape banget\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eLelah\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAku lelah\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"4\" valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003eIdeation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eSuicide and Commit suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eBunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eHow to suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCara bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eSuicide methods\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eWant to Die\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eIngin mati\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"9\" valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003eMethods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eHanging\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eGantung diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eHang myself\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eDrink poison\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eMinum racun\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003ePoisoning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eJumping from\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eLoncat dari\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eSuicide meds\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eObat bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eSuicide drugs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eSuicide pills\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003ePil bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eOverdose\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eOverdosis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"4\" valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003ePostvention\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eSuicide reasons\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAlasan bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eWhy did they suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eKenapa mereka bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eWhy do people suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eKenapa orang bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eWhy did my family member suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eKenapa keluarga saya bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eWhy did my friend suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eKenapa teman saya bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 228px;\"\u003e\n \u003cp\u003eWhy did my parent suicide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eKenapa orangtua saya bunuh diri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNote: Number of datapoints refers to how many of the 34 provinces Google Trends was able to provide data for.\u003c/p\u003e\n\u003ch3\u003eData Preparation\u003c/h3\u003e\n\u003cp\u003eSimilar to Study 1, we also aggregated the individual keywords by averaging the relative search volumes within each group. This yielded relative search volumes for distress, ideation, methods, and postvention keyword groups. Individual or aggregated keywords were retained for analysis if search volume data was available for 20 or more provinces. Missing data were handled using pairwise deletion method, given that the keywords had been selected on data availability, and listwise deletion would lead to too little data for analysis. Insufficient Google Trends data were available for all postvention keywords. As such, this keyword group was excluded from analyses.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eAnalysis Plan\u003c/h3\u003e\n\u003cp\u003eThe main outcome was to investigate the relationship between the combined search volumes for each keyword category, and attempt and suicide rates. For each keyword group, we investigated whether there was a positive correlation between the search volume and the attempt and suicide rates. If the search volumes for any of these keyword groups were found to correlate with each other, we would assess their relationship with suicide and attempt rate while controlling for the other to encapsulate their unique relationship with the suicide metrics. The secondary outcome was to similarly investigate whether individual keywords were positively associated with attempt and suicide rates, using Pearson correlations testing a one-way positive relationship. All statistical analyses were performed using JASP Statistical Software\u003csup\u003e23\u003c/sup\u003e.\u003c/p\u003e\n\u003ch4\u003eFollow-up analysis 1 \u0026ndash; method lethality\u003c/h4\u003e\n\u003cp\u003eThe main analysis identified that the methods keyword group showed a relationship with both suicide rate and attempt rate. We therefore investigated whether \u0026lsquo;high lethality methods\u0026rsquo; and \u0026lsquo;low lethality methods\u0026rsquo; had different relationships with attempts and suicide rate. We combined the keyword search volumes for \u0026lsquo;high lethality methods\u0026rsquo; (hanging/\u0026lsquo;gantung diri\u0026rsquo;, jumping from height/\u0026lsquo;loncat dari\u0026rsquo;) and \u0026lsquo;low lethality methods\u0026rsquo; (self-poisoning/\u0026lsquo;minim racun\u0026rsquo;, \u0026lsquo;obat bunuh diri\u0026rsquo;, and \u0026lsquo;overdosis\u0026rsquo;), and assessed their positive relationships with suicide and attempt rates, while controlling for the other.\u003c/p\u003e\n\u003ch4\u003eFollow-up analysis 2 \u0026ndash; loneliness\u003c/h4\u003e\n\u003cp\u003eGiven that loneliness is a key factor in suicidal thoughts and behaviours\u003csup\u003e28\u003c/sup\u003e but was not included as a term in the pre-registration, we repeated the main analysis, adding loneliness to the distress category. These keywords were \u0026lsquo;kesepian\u0026rsquo; and \u0026lsquo;aku kesepian\u0026rsquo;, translated to \u0026lsquo;loneliness\u0026rsquo; and \u0026lsquo;I am lonely\u0026rsquo; respectively. Each met criteria for inclusion with \u0026lsquo;kesepian\u0026rsquo; having 34 data points and \u0026lsquo;aku kesepian\u0026rsquo; having 24 data points.\u003c/p\u003e\n\u003ch2\u003eResults\u003c/h2\u003e\n\u003cp\u003eThe aggregated search volumes for the distress and ideation keyword groups were significantly correlated (r = 0.451, p = .004), therefore their respective correlations with suicide and attempts rates controlled for each other. Neither the distress group or ideation group correlated with the methods group (r=0.179, p=.155; r=0.207, p=.120; respectively).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe main analyses revealed that distress keywords, whether combined or individually, did not show a significant positive relationship with attempt or suicide rate (see Table 4). As was already shown in Study 1, the individual \u0026lsquo;bunuh diri\u0026rsquo; and \u0026lsquo;cara bunih diri\u0026rsquo; ideation keywords showed a relationship with attempt rate, but not suicide rate. The same pattern was observed for the combined ideation group, despite the addition of the \u0026lsquo;ingin mati\u0026rsquo; keyword which did not correlate with either attempt or suicide rate. The combined methods keyword group showed a relationship with both suicide and attempt rate. However, there is a mixed pattern with individual keywords showing a relationship with both attempt and suicide, only one outcome, or neither. The results of the correlation analyses can be found in Table 4.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 4: Correlation of the relative search volumes of IMV category keywords with attempt and suicide rates in Indonesia 2021\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"640\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKeyword\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 295px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCorrelation Coefficient\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eGroups\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eSuicide Rate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eAttempt Rate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eDistress (controlling for Ideation)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.140\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e-0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eIdeation (controlling for Distress)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e-0.061\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.462**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eMethods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.317*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.435**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eDistress\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eDepresi (Depression)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.108\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e-0.077\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eSedih banget (Depression)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e-0.370\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e-0.024\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eStress (Stress)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.216\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e-0.231\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eMenyerah (Giving up)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.367\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.132\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eMenyerah saja (Giving up)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.286\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eCape (Fatigue and burnout)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e-0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e-0.289\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eLelah (Fatigue and burnout)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.157\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eAku lelah (Fatigue and burnout)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e-0.120\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.185\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eIdeation\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eBunuh diri (Suicide, commit suicide)\u0026dagger;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.333*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eCara bunuh diri (How to suicide, how to kill yourself, suicide methods) \u0026dagger;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e-0.116\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.649***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eIngin mati (want to die)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.153\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eMethods\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eGantung diri (Hanging, hang myself)\u0026dagger;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.420**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.375*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eMinum racun (Drink poison, poisoning)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e-0.394\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.459*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eLoncat dari (Jumping from)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.500**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.078\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eObat bunuh diri (Suicide meds, suicide drugs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.549**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 344px;\"\u003e\n \u003cp\u003eOverdosis (Overdose)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003e0.139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e0.218\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNote: * indicates p \u0026lt; .05, ** indicates p \u0026lt;.01, *** indicates p\u0026lt;.001. \u0026dagger; Analyses for individual keywords are repeated from Study 1.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe follow-up analysis revealed that search volumes for low lethality and high lethality methods were significantly correlated (r = 0.301, p = .05). When controlling for low-lethality methods, high-lethality methods were uniquely associated with both suicide rate (r = 0.562, p \u0026lt; .001) and attempt rate (r = 0.334, p = .031). When controlling for high-lethality methods, low-lethality method search volumes were positively associated with attempt rate (r = 0.409, p = .011), but not suicide rate (r = -0.011, p = .524).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNeither of the loneliness search terms significantly correlated with attempts or deaths (kesepian \u0026ndash; suicides: r = -0.192; kesepian \u0026ndash; attempts: r = 0.057; aku kesepian \u0026ndash; suicides: r = 0.061; aku kesepian \u0026ndash; attempts: -0.448). Including these terms in the distress keyword group also did not affect the pattern of findings (distress \u0026ndash; suicides: r = 0.111; distress \u0026ndash; attempts: -0.129).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eDiscussion\u003c/h2\u003e\n\u003cp\u003eIn Study 3, we investigated the relationship between categories of keywords, grouped according to which phase they broadly map onto in the IMV-Model, and attempt and suicide rates. In line with our hypothesis, the distress keywords showed no relationship with attempt or suicide rates, and method keywords were uniquely associated with both suicide and attempt rates. Due to the lack of data for postvention keywords, it was not possible to validate the hypothesis that this would also be related to suicide rates. Furthermore, when considering the methods keywords based on the method lethality, it appeared that both low and high-lethality keywords had a positive association with attempt rate, but only high-lethality keywords had a positive association with suicide rate. Thus, this suggests that for method-related keywords, the relationship with suicide rate may be driven primarily by high lethality keywords. In sum, the results suggest that keyword categories according to the IMV-Model differentially predict attempt and suicide rate depending on their proximity to the volitional stage of the IMV-Model or attempt and their lethality.\u0026nbsp;\u003c/p\u003e"},{"header":"General Discussion","content":"\u003cp\u003eAcross three studies, we investigated the relationship between relative Google Trends search volumes of suicide-related keywords and rates of suicidal behaviours. Our findings provided evidence suggesting that correlations between search volumes and suicide outcomes may differ across stages of the IMV-Model. Specifically for methods-related keywords, this relationship is dependent on the lethality of the method searched. Figure 1 outlines the synthesised findings.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePrevious studies have reported conflicting findings for whether Google Trends search volumes are associated with suicide rates. A key paper which strongly argues for the absence of this association was published by Tran and colleagues in 2017\u003csup\u003e14\u003c/sup\u003e. The authors reanalysed several past studies investigating temporal associations between suicide keywords and suicide rates, using best practice time series analyses. While the authors found an association between the relative search volume of \u0026lsquo;suicide methods\u0026rsquo; and suicide rates in the US, overall, they did not find an association between \u0026lsquo;suicide\u0026rsquo;, \u0026lsquo;suicide methods\u0026rsquo;, and other commonly used keywords, and suicide rates across several countries. The authors concluded that Google Trends data were not reliable in predicting suicide rates. We raise three key points, outlining how our findings and the findings from the study do not contradict one another. First, the authors only examined suicide rates and did not include attempt or self-harm rates. Consistent with their findings, we did not find a relationship between \u0026lsquo;suicide\u0026rsquo; and \u0026lsquo;suicide methods\u0026rsquo; keywords and suicide rate, as Studies 1 and 2 demonstrated a relationship with attempt/self-harm hospitalisation rates. Secondly, the authors only investigated time series data, whereas we include analyses for single time points across multiple geographic regions. Our time series analysis in Study 2 replicated the finding of a relationship with self-harm but not suicide rate, and also found this association using regional data for a single time period (Study 1 and 3). Third, our findings suggest that the relationship between relative search volumes and suicide rates depends on the type of keyword, whereas the earlier study did not categorise the keywords into semantic groupings. Thus, our findings support the previous findings, but also provide new insights. Together, these previous and new findings point towards a non-uniform relationship between relative search volumes of suicide-related keywords and suicide and attempt rates, in which not every suicide-related keyword will have a relationship with suicide and attempt rates. However, our study suggests certain relationships reliably emerge across settings and dimensions. Future research needs to carefully consider which keywords to use, informed by what stage of their ideation the individual is likely to be, and whether a temporal or geographic association is being investigated.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTogether, these findings suggest that relative search volume may still be a useful method to monitor suicide outcomes at an aggregate level; however, we must carefully consider what suicide outcomes are of interest, which will inform what keywords are used. Further, our findings suggest that keywords commonly used in past studies\u003csup\u003e12,14\u003c/sup\u003e, that is \u0026lsquo;suicide\u0026rsquo; and \u0026lsquo;suicide method\u0026rsquo;, reliably show an association with attempt rather than suicide rate, and are also consistent with previous findings that these keywords were not reliably associated with suicide rate. Future studies may seek to further understand which combination of keywords are best correlated with attempt or suicide rate and validate the keywords using novel data.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGiven that a suicide is a fatal attempt, it may be reasonable to assume that suicide attempt and death rates would be correlated. However, we did not find this association. One possibility is that the regional and temporal factors which affect the fatality of an attempt \u0026ndash; such as availability of lethal means or accessibility of emergency services \u0026ndash; are so diverse and varied across time and regions\u003csup\u003e22\u003c/sup\u003e that the added statistical variability obscures this relationship. Furthermore, in Study 3, the relative search volume for methods-related keywords is correlated to attempt rate, with high lethality keywords also correlated to suicide rate. A previous study found that searching for methods-related keywords corresponded to seriously considering suicide\u003csup\u003e29\u003c/sup\u003e, thus having a relationship with attempt rate. However, the relationship between methods and suicide rate may be driven by relative search volumes of particularly lethal methods. Nonetheless, it is still unclear whether higher searches for more lethal methods reflects higher use of these methods. Future research may seek to investigate whether searches for more lethal methods reflects higher use of these methods.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLinking search terms and the IMV-Model may help us design better interventions for individuals searching for suicide-related keywords. Our recent work has examined the use of search engine adverts for suicide prevention, in which a person searching for suicide-related keywords is presented with a tailored advert, which when clicked leads to a landing page with resources and help-seeking links\u003csup\u003e30,31\u003c/sup\u003e. One study investigated the effect of using explicit suicide wording in the advert, based on different groups of search keywords including \u0026lsquo;distress\u0026rsquo; and \u0026lsquo;ideation\u0026rsquo; keywords (called \u0026lsquo;low risk\u0026rsquo; and \u0026lsquo;high risk\u0026rsquo; respectively in previous study). Using an advert which explicitly mentioned suicide was associated with increased engagement for individuals searching keywords explicitly communicating suicidality\u003csup\u003e31\u003c/sup\u003e (ideation/high-risk). One possible approach is to tailor these campaigns depending on what keywords the individual uses, indicative of which stage the IMV-Model the individual is likely to be. Tailoring based on the IMV-Model may increase the likelihood that a person is met with relevant and helpful information.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur studies have several limitations. Firstly, there are concerns regarding data quality and comprehensiveness. Suicides and suicide attempts are known to be underreported in Indonesia due to stigma and data recording issues\u003csup\u003e21,22\u003c/sup\u003e. In Australia, the self-harm hospitalisation data includes only those admitted to a hospital ward, and thus represents only a proportion of self-harm episodes. Nevertheless, this still represents a high quality and standardised dataset describing self-harm on a national scale. Secondly, the data from Google Trends represents a subset of the total population, in which there may be regional differences in internet accessibility\u003csup\u003e31\u003c/sup\u003e or search engine preferences. Future studies should examine whether, despite these shortcomings, online engagement is widespread enough to allow for representative data and whether adjustments are needed to account for these variations. Third, there are reservations concerning the utility of relative search volumes. Previous studies have highlighted that data from Google Trends might not be completely aligned with intention in relation to a given behaviour\u003csup\u003e14\u003c/sup\u003e. For instance, individuals may conduct suicide-related searches for reasons unrelated to distress, such as academic research. Further, Google Trends data is normalised, and thus does not provide the raw volume of searches, also preventing calculation of weighted averages for combination of keywords. Future studies could aim to replicate this analysis using Google Ads data, which offers ad presentation data which can be multiplied by impression share percentage (the proportion of times the ad was shown compared to its eligible presentations) as a proxy for searches, provides higher temporal and geographic resolution, allows grouping keywords, and permits the exclusion of specific keywords to focus on search intent. Another possibility is using Google\u0026rsquo;s symptom dataset, which provides greater resolution relative to search volumes for health-related keywords but only for Australia, Ireland, New Zealand, the United Kingdom, and the United States. Fourth, we do not yet know how repeat attempts or self-harm influence information-seeking online. For example, an individual may engage in more information seeking prior to an index attempt, compared with subsequent attempts, leading to the possibility of additional confounders in the data and analysis. Fifth, self-harm and suicide attempts are separate outcomes with different underlying methodologies, and thus Studies 1 and 3, and Study 2 are not direct replications. However, of existing data sources, these are the optimal data sources available to test our hypothesis. Subsequent studies should look to replicating these findings using identical outcomes. Finally, our time series analyses did not investigate different lags in Study 2 as has been done in previous studies.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn conclusion, our findings suggest that the relationship between relative search volumes and suicide attempts and deaths is not universal. Instead, keywords correlate differently with suicide outcomes at various stages, broadly aligning with the IMV Model. Future research and prevention efforts should carefully consider the primary outcome of interest (suicide or attempt), the likely state of individuals conducting searches, and the specific keywords used. Nevertheless, across different geographic regions and time periods, and in both Indonesia and Australia, the relationship between search volumes for \u0026lsquo;suicide\u0026rsquo; and \u0026lsquo;suicide methods\u0026rsquo; and attempt rates appears stable. Since suicide monitoring is a crucial step in addressing SDG 3.4.2, the use of a tool that provides ongoing data, covers regions with limited high-quality suicide data, and maintains a stable relationship with suicide attempts at a macro level warrants further exploration. This is particularly important to improve monitoring accuracy and achieve the rate of reduction needed to meet SDG 3.4.2\u0026rsquo;s target of reducing mortality by one-third.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSandersan Onie led the conceptualization, formal analysis, funding acquisition, methodology, writing- original draft, and writing \u0026ndash; review \u0026amp; editing. Mark Larsen led supervision, supported conceptualization, funding acquision, and led writing \u0026ndash; review \u0026amp; editing. Matthew Coleshill led writing \u0026ndash; review \u0026amp; editing. Fiona Shand supported supervision, methodology, funding acquision, and writing \u0026ndash; review \u0026amp; editing. Michelle Tye supported funding acquisition, and writing \u0026ndash; review \u0026amp; editing. Venkatesha supported methodology and analysis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was supported by the Australian Government Department of Health\u0026ndash;funded National Suicide Prevention Research Fund, managed by Suicide Prevention Australia to Sandersan Onie; the NHMRC Centre of Research Excellence in Suicide Prevention (APP1152952) to Mark Larsen, Fiona Shand, and Michelle Torok; and Australian Department of Foreign Affairs and Trade, Australian-Indonesian Institute (AII2020322) to Sandersan Onie.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interest Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSO has provided unpaid consultation to tech companies on suicide prevention policies. All other authors declare no competing financial or non-financial interest.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Google search datasets generated or analysed during the current study are available on Google Trends. The Australian suicide and self-harm hospitalisation data is accessible through the Australian Institute of Health and Welfare website. The Indonesian suicide and attempt data is not accessible, and was provided under strict requirements by the Indonesian government.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWorld Health Organization. \u003cem\u003eGlobal Health Estimates\u003c/em\u003e (2024). Available at: https://www.who.int/data/global-health-estimates.\u003c/li\u003e\n\u003cli\u003eInstitute of Medicine (US) Committee on Pathophysiology and Prevention of Adolescent and Adult Suicide. \u003cem\u003eReducing Suicide: A National Imperative\u003c/em\u003e. Edited by Goldsmith, S.K., Pellmar, T.C., Kleinman, A.M. \u0026amp; Bunney, W.E. (National Academies Press, 2002). PMID: 25057611.\u003c/li\u003e\n\u003cli\u003eUnited Nations. \u003cem\u003eSustainable Development Goals\u003c/em\u003e (2025). Available at: https://sdgs.un.org/goals/goal3.\u003c/li\u003e\n\u003cli\u003eWorld Health Organization. \u003cem\u003eComprehensive Mental Health Action Plan 2013\u0026ndash;2030\u003c/em\u003e (2021). Available at: https://www.who.int/publications/i/item/9789240031029.\u003c/li\u003e\n\u003cli\u003eAre\u0026aacute;n, P. A. et al. Perceived utility and characterization of personal Google search histories to detect data patterns proximal to a suicide attempt in individuals who previously attempted suicide: Pilot cohort study. J. Med. Internet Res. 23, e27918 (2021).\u003c/li\u003e\n\u003cli\u003eSueki, H. Does the volume of internet searches using suicide-related search terms influence the suicide death rate: Data from 2004 to 2009 in Japan. Psychiatry Clin. Neurosci. 65, 392\u0026ndash;394 (2011).\u003c/li\u003e\n\u003cli\u003eYang, A. C., Tsai, S., Huang, N. E. \u0026amp; Peng, C. Association of internet search trends with suicide death in Taipei City, Taiwan, 2004\u0026ndash;2009. \u003cem\u003eJ. Affect. Disord.\u003c/em\u003e 132, 179\u0026ndash;184 (2011).\u003c/li\u003e\n\u003cli\u003eLee, J. Y. Search trends preceding increases in suicide: A cross-correlation study of monthly Google search volume and suicide rate using transfer function models. \u003cem\u003eJ. Affect. Disord.\u003c/em\u003e 262, 155\u0026ndash;164 (2020).\u003c/li\u003e\n\u003cli\u003eParker, J., Cuthbertson, C., Loveridge, S., Skidmore, M. \u0026amp; Dyar, W. Forecasting state-level premature deaths from alcohol, drugs, and suicides using Google Trends data. \u003cem\u003eJ. Affect. Disord.\u003c/em\u003e 213, 9\u0026ndash;15 (2017).\u003c/li\u003e\n\u003cli\u003eArora, V. S., Stuckler, D. \u0026amp; McKee, M. Tracking search engine queries for suicide in the United Kingdom. \u003cem\u003ePublic Health\u003c/em\u003e 137, 147\u0026ndash;153 (2016).\u003c/li\u003e\n\u003cli\u003eKristoufek, L., Moat, H. S. \u0026amp; Preis, T. Estimating suicide occurrence statistics using Google Trends. \u003cem\u003eEPJ Data Sci.\u003c/em\u003e 5, 32 (2016).\u003c/li\u003e\n\u003cli\u003eMa-Kellams, C., Or, F., Baek, J. H. \u0026amp; Kawachi, I. Rethinking suicide surveillance: Google search data and self-reported suicidality differentially estimate completed suicide risk. \u003cem\u003eClin. Psychol. Sci.\u003c/em\u003e 4, 480\u0026ndash;484 (2016).\u003c/li\u003e\n\u003cli\u003ePage, A., Chang, S. S. \u0026amp; Gunnell, D. Surveillance of Australian suicidal behaviour using the internet? \u003cem\u003eAust. N. Z. J. Psychiatry\u003c/em\u003e 45, 1020\u0026ndash;1022 (2011).\u003c/li\u003e\n\u003cli\u003eTran, U. S. \u003cem\u003eet al.\u003c/em\u003e Low validity of Google Trends for behavioral forecasting of national suicide rates. \u003cem\u003ePLoS ONE\u003c/em\u003e 12, e0183149 (2017).\u003c/li\u003e\n\u003cli\u003eKnipe, D., Gunnell, D., Evans, H., John, A. \u0026amp; Fancourt, D. Is Google Trends a useful tool for tracking mental and social distress during a public health emergency? A time-series analysis. \u003cem\u003eJ. Affect. Disord.\u003c/em\u003e 294, 737\u0026ndash;744 (2021).\u003c/li\u003e\n\u003cli\u003eO\u0026apos;Connor, R. C. \u0026amp; Kirtley, O. J. The Integrated Motivational-Volitional Model of suicidal behaviour. \u003cem\u003ePhilos. Trans. R. Soc. B Biol. Sci.\u003c/em\u003e 373, 20170268 (2018).\u003c/li\u003e\n\u003cli\u003eOnie, S. Google searches and suicide attempt. \u003cem\u003eOSF Pre-registration\u003c/em\u003e (2023). https://doi.org/10.17605/OSF.IO/SQPNY\u003c/li\u003e\n\u003cli\u003eBadan Pusat Statistik. Jumlah penduduk hasil proyeksi menurut provinsi dan jenis kelamin (ribu jiwa), 2018\u0026ndash;2020. \u003cem\u003eBadan Pusat Statistik\u003c/em\u003e (2022). https://www.bps.go.id/indicator/12/1886/1/jumlah-penduduk-hasil-proyeksi-menurut-provinsi-dan-jenis-kelamin.html\u003c/li\u003e\n\u003cli\u003eKompas.com. Jumlah provinsi di Indonesia ada 38, ini daftar dan ibu kotanya. \u003cem\u003eKompas.com\u003c/em\u003e (2023). https://www.kompas.com/tren/read/2023/04/11/074500965/jumlah-provinsi-di-indonesia-ada-38-ini-daftar-dan-ibu-kotanya\u003c/li\u003e\n\u003cli\u003eGoogle. Google Trends. \u003cem\u003eGoogle\u003c/em\u003e (2024). https://trends.google.com/trends/\u003c/li\u003e\n\u003cli\u003eOnie, S. \u003cem\u003eet al.\u003c/em\u003e Indonesian first national suicide prevention strategy: Key findings from the qualitative situational analysis. \u003cem\u003eLancet Reg. Health Southeast Asia\u003c/em\u003e 2023, 100245 (2023).\u003c/li\u003e\n\u003cli\u003eOnie, S. \u003cem\u003eet al.\u003c/em\u003e Indonesia\u0026rsquo;s first suicide statistics profile: An analysis of suicide and attempt rates, underreporting, geographic distribution, gender, method, and rurality. \u003cem\u003eLancet Reg. Health Southeast Asia\u003c/em\u003e (in press).\u003c/li\u003e\n\u003cli\u003eBadan Pusat Statistik. \u003cem\u003eVillage Potential Statistics of Indonesia\u003c/em\u003e (2021). https://www.bps.go.id/publication/2022/03/24/ceab4ec9f942b1a4fdf4cd08/statistik-potensi-desa-indonesia-2021.html\u003c/li\u003e\n\u003cli\u003eJASP Team. \u003cem\u003eJASP (Version 0.17.3)\u003c/em\u003e (2023).\u003c/li\u003e\n\u003cli\u003eAustralian Bureau of Statistics. Population. \u003cem\u003eAustralian Bureau of Statistics\u003c/em\u003e (2023). https://www.abs.gov.au/statistics/people/population\u003c/li\u003e\n\u003cli\u003eAustralian Bureau of Statistics. Intentional self-harm deaths (suicide) in Australia. \u003cem\u003eAustralian Bureau of Statistics\u003c/em\u003e (2023). https://www.abs.gov.au/statistics/health/causes-death/causes-death-australia/latest-release#intentional-self-harm-deaths-suicide-in-australia\u003c/li\u003e\n\u003cli\u003eAustralian Bureau of Statistics. Causes of death Australia. \u003cem\u003eAustralian Bureau of Statistics\u003c/em\u003e (2024). https://www.abs.gov.au/statistics/health/causes-death/causes-death-australia/latest-release\u003c/li\u003e\n\u003cli\u003eMcClelland, H., Evans, J. J., Nowland, R., Ferguson, E. \u0026amp; O\u0026apos;Connor, R. C. Loneliness as a predictor of suicidal ideation and behaviour: A systematic review and meta-analysis of prospective studies. \u003cem\u003eJ. Affect. Disord.\u003c/em\u003e 2024, in press.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"suicide, self-harm, Google Trends, Australia, Indonesia","lastPublishedDoi":"10.21203/rs.3.rs-6137446/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6137446/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e \u003cp\u003eAccording to the recent WHO Global Health Estimates, the globe is not on track to meet the UN Sustainable Development Goal 3.4.2 of the reduction of suicide, with suicide monitoring being a key issue. Past research has found an association between Google searches for suicide-related keywords and suicide rates, offering a potential tool for rapid monitoring of population suicide rates \u0026ndash; although recent findings call this relationship into question. However, the relationship between Google searches and suicide attempts or self-harm has not been investigated. Across three studies, we aimed to ascertain the associations between search volumes for suicide-related keywords and suicide rates, suicide attempts, and self-harm hospitalisation rates, within the IMV-Model of Suicidal Behaviour.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e \u003cp\u003eStudy 1 investigated the relationship between provincial relative search volumes for suicide-related keywords with attempt and suicide rates across Indonesian provinces in 2021. Study 2 investigated the relationship between national relative search volumes for suicide-related keywords with attempt and self-harm hospitalisation rates in Australia between 2008 and 2020. Study 3 investigated the relationship between categories of suicide-related keywords grouped according to the IMV-Model of Suicidal Behaviour, and their relationship with attempt and suicide rates across provinces in Indonesia.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e \u003cp\u003eIn studies 1 and 2, we did not observe a significant association between relative search volumes for suicide-related keywords and suicide rates. However, relative search volumes for suicide-related keywords were positively associated with provincial suicide attempt rates in Indonesia and yearly self-harm hospitalisation rates in Australia. Study 3 revealed that keywords associated with distress showed no relationship with attempt or suicide rate, while keywords associated with explicit suicide ideation showed a relationship with attempt only. Keywords associated with specific methods \u0026ndash; the volitional component of the IMV-Model - were uniquely associated with both attempt and suicide, with the relationship with suicide rate driven by high-lethality methods keywords.\u003c/p\u003e\u003cp\u003e\u003cb\u003eLimitations\u003c/b\u003e\u003c/p\u003e \u003cp\u003eAscertaining the quality of data on suicides, suicide attempts, and self-harm incidents is challenging. Moreover, Google Trends has limitations regarding the granularity of data it provides, and it may not fully represent the entire population.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e\u003c/p\u003e \u003cp\u003eOur findings suggest that the relationship between search volumes and suicide and attempt rates may depend on the category of keyword, with \u0026lsquo;suicide\u0026rsquo; and \u0026lsquo;suicide method\u0026rsquo; being associated with suicide rate and self-harm hospitalisation, but not suicide rate. The findings show promise for improved suicide monitoring.\u003c/p\u003e","manuscriptTitle":"Rethinking Google Searches for Suicide-Related Keywords and Their Association with Suicide Rates, Attempts, and Self-Harm Hospitalisation: An IMV-Model Approach","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-07 13:32:53","doi":"10.21203/rs.3.rs-6137446/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-04-30T12:21:44+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-04-28T10:55:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"184581601686022399092949414624944944371","date":"2025-04-17T11:58:56+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-03-17T05:32:06+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"319232484035587348116388871371900194531","date":"2025-03-08T13:10:20+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-03-07T11:04:14+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-03-05T16:07:18+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-03-05T06:21:15+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Digital Medicine","date":"2025-03-02T04:59:16+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"cca93fbc-cd6c-4264-b7b8-8b5f34b7ad54","owner":[],"postedDate":"March 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":45251424,"name":"Biological sciences/Psychology"},{"id":45251425,"name":"Health sciences/Health care/Disease prevention"},{"id":45251426,"name":"Health sciences/Health care/Health policy"},{"id":45251427,"name":"Health sciences/Health care/Health services"},{"id":45251428,"name":"Health sciences/Health care/Public health/Epidemiology"}],"tags":[],"updatedAt":"2025-09-01T16:04:45+00:00","versionOfRecord":{"articleIdentity":"rs-6137446","link":"https://doi.org/10.1038/s41746-025-01916-4","journal":{"identity":"npj-digital-medicine","isVorOnly":false,"title":"npj Digital Medicine"},"publishedOn":"2025-08-29 15:57:28","publishedOnDateReadable":"August 29th, 2025"},"versionCreatedAt":"2025-03-07 13:32:53","video":"","vorDoi":"10.1038/s41746-025-01916-4","vorDoiUrl":"https://doi.org/10.1038/s41746-025-01916-4","workflowStages":[]},"version":"v1","identity":"rs-6137446","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6137446","identity":"rs-6137446","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}