Portal de Periódicos da CAPES

Suicide‐related Internet search queries in India following media reports of a celebrity suicide: an interrupted time series analysis

2021; Wiley; Volume: 20; Issue: 1 Linguagem: Inglês

10.1002/wps.20832

2051-5545

Gregory Armstrong, Tilahun Haregu, Vikas Arya, Lakshmi Vijayakumar, Mark Sinyor, Thomas Niederkrotenthaler, Matthew J. Spittal,

HIV, Drug Use, Sexual Risk

On June 14, 2020, media reported that Sushant Singh Rajput, a 34-year-old male Bollywood star, had died by suicide. He had starred in television and cinema for over a decade, with highly prominent roles that reached a broad cross-section of the community, including playing the Indian cricket team captain M.S. Dhoni in a blockbuster 2016 biographical film. This suicide has generated widespread media coverage, including reports about hanging as the suicide method and the actor's struggle with depression. The event has sparked immediate concerns about possible imitation suicides. Due to the long delays in the release of suicide data, it may take some time before we are able to assess any impact on suicidal behaviour. However, it is possible to assess how the public responded in terms of online searching behaviours. Social learning plays an important role in suicide. The Werther effect hypothesizes that a suicide can become a stimulus for subsequent imitation suicides, which can be exacerbated in cases of celebrity suicide and by irresponsible media reporting. A recent meta-analysis1 estimated that the risk of suicide increased by 13% in the period after media reports of celebrity suicide; when the suicide method was reported in the media, there was an associated 30% increase in deaths by the same method, highlighting that media ought to be highly cautious in this regard. On the other hand, the Papageno effect hypothesizes that media can report on suicide in ways that stimulate protective effects. Research has primarily examined the protective effects of reporting on people who have been suicidal and have drawn on internal or external resources to avoid progressing to an attempt2, but it is recommended that media can assist in more ways, such as highlighting the role that mental ill-health can play in suicide crises and the supports that are available3. A substantial amount of social learning now happens online, and suicide research has recently been focusing on Internet search queries as one component of understanding online social learning exposures and interest in a topic4. For example, one study observed a 19% increase in worldwide Google suicide queries in the 19 days following the release of the TV series 13 Reasons Why 5, and that event was associated with a 13% increase in youth suicides in the US6. Given these findings, we might expect that Internet search query volumes could yield insights into suicide trends over time in the context of celebrity suicides that receive a high level of media coverage. We examined changes in Internet search queries in India in the three weeks following the suicide, compared to the two preceding years. We obtained weekly data on relative search volumes from Google Trends (https://trends.google.com/trends) for a series of search terms for the time period from June 13, 2018 to July 4, 2020. English search terms were adapted from those documented in a review of suicide-related Google Trends studies4. We included putatively harmful search terms ("suicide", "commit suicide", "how to suicide", "hanging", "how to hang") as well as protective terms that may reflect a stimulation of suicide/depression awareness ("suicide prevention", "depression", "suicide help­line", "depression doctor", "psychiatrist", "antidepressants"). We also collected data for a selection of Hindi search terms, including "aatmahatya" (suicide), "khudkushi" (suicide) and "avsaad" (depression). Due to small day-to-day variations in data obtained through Google Trends, we followed validated methods4 and repeated our search on seven consecutive days between August 3 and 9, 2020 and used the average for our analyses. To investigate whether weekly search volumes had changed since the suicide of Sushant Singh Rajput, relative to the two preceding years, we conducted interrupted time series regression for each search term. The outcome was the relative weekly search volume for the search term, which is generated in Google Trends as an index ranging between 0 and 1004, and analyses were repeated using binary predictor variables representing each of the three weeks since the suicide. Models were fit using a generalized linear model from the Poisson family, with a loglink function and a scale parameter to account for overdispersion. Models controlled for long-term trends (entered as a fractional polynomial to account for nonlinearity) and short-term seasonality trends (entered as Fourier terms). At week 1, we observed large relative risk (RR) increases for "suicide" (RR=11.53, 95% CI: 10.01-13.27), "commit suicide" (RR=16.46, 95% CI: 14.20-19.07), "how to suicide" (RR=10.15, 95% CI: 7.38-13.97), "hanging" (RR=2.08, 95% CI: 1.65-2.62), "how to hang" (RR=10.80, 95% CI: 6.33-18.44), "suicide hanging" (RR=2.53, 95% CI: 2.00-3.19), "aatmahatya" (RR=2.70, 95% CI: 1.76-4.14) and "khudkushi" (RR=7.56, 95% CI: 5.02-11.38). Significant increases persisted at week 2 for the search terms "suicide" (RR=3.11, 95% CI: 2.55-3.80), "commit suicide" (RR=2.96, 95% CI: 2.39-3.67), "how to suicide" (RR=6.50, 95% CI: 4.58-9.22), "hanging" (RR=1.43, 95% CI: 1.09-1.88), "how to hang" (RR=4.00, 95% CI: 1.95-8.17), and "suicide hanging" (RR=1.67, 95% CI: 1.27-2.20), and at week 3 for the search terms "suicide" (RR=1.61, 95% CI: 1.25-2.08), "commit suicide" (RR=1.48, 95% CI: 1.13-1.95), "how to suicide" (RR=3.05, 95% CI: 1.98-4.69) and "suicide hanging" (RR=1.82, 95% CI: 1.39-2.37). At week 1, we also observed large increases for the putatively protective search terms "suicide prevention" (RR=12.64, 95% CI: 5.01-31.89), "suicide helpline" (RR=5.63, 95% CI: 4.57-6.94), "depression" (RR=6.40, 95% CI: 5.93-6.92), "depression doctor" (RR=4.99, 95% CI: 3.10-8.03), "psychiatrist" (RR=1.86, 95% CI: 1.65-2.10), "antidepressants" (RR=1.47, 95% CI: 1.19-1.81) and "avsaad" (RR=5.79, 95% CI: 3.22-10.39). Significant increases persisted at week 2 for the search terms "suicide helpline" (RR=1.86, 95% CI: 1.36-2.54), "depression" (RR=1.79, 95% CI: 1.59-2.02), and "psychiatrist" (RR=1.41, 95% CI: 1.23-1.62), and at week 3 for "depression doctor" (RR=2.61, 95% CI: 1.74-3.93) and "avsaad" (RR=2.87, 95% CI: 1.16-7.13). The suicide of Sushant Singh Rajput and the subsequent widespread media coverage appears to have activated large increases in both harmful and protective Google search queries in India. This may have been exacerbated by the COVID-19 lockdown, with more time spent at home on online devices resulting in increased media exposure and greater opportunities to search online. Whether Sushant Singh Rajput's death and these search trends will be associated with an increase in suicides should be investigated when data become available. Regardless, the current analysis clearly shows that his suicide was associated with strong increases in suicide-related Internet search behaviours, highlighting the need to promote media recommendations3. Furthermore, the monitoring of Internet search queries after celebrity suicides and other large-scale media phenomena could inform suicide prevention.