Distress and Anxiety during the COVID-19 Pandemic as Revealed through Helpline Calls


 The COVID-19 crisis affects mental health. We use helpline calls to quantify this effect. Helpline calls offer a real-time measure of ‘revealed’ anxiety across a range of topics. We have collected data on 7 million calls from from 19 countries. We find that calls peaked at 35% above pre-pandemic levels, six weeks after the initial outbreak. The increase was mainly driven by fear (including fear of infection), loneliness, and, later in the pandemic, physical-health concerns. Relationship issues, economic problems, violence, and suicidal ideation, however, were less prevalent than before the pandemic. This pattern is apparent both during the first wave and during subsequent waves of the pandemic. Issues directly tied to COVID-19 therefore seem to have substituted rather than exacerbated underlying anxieties. Conditional on infection rates, suicide-related calls increased when containment policies became more stringent and decreased when income support was extended. This implies that financial relief can allay the distress triggered by lockdown measures.


B) Around the introduction of SIP orders
Figure 1: Evolution of total helpline call numbers during the first wave Event study models; dependent variable is ln(daily helpline calls+1). The sample includes daily data for 21 helplines during the period from 4 weeks before to 12 weeks after the event date in early 2020, and, for 17 of the 21 helplines, the corresponding days of 2019. Coefficients on week indicators, plotted with 95% confidence intervals, show the average percentage change in call volumes relative to reference week 0, which represents the week when (A) the cumulative number of COVID-19 infections exceeded 1 per 100,000 inhabitants [31], or (B) shelter-in-place orders were introduced [32]. Models include helpline fixed effects interacted with year, week-of-year and day-of-week indicators. See Methods, equation (1). Standard errors are clustered at the helpline-week level. Results on data weighted by total number of calls recorded for each helpline during the sample period (black); and unweighted models (gray). Supplementary Table S1 and Fig. S1 show the trends in call volumes and country-specific event dates separately for the 21 helplines. Table S2 shows detailed regression results underlying the graphs.  Supplementary Tables S3,  S4, and S5 show descriptive statistics for the estimation sample overall and separately by helpline, Table S6 shows detailed estimation results.
Overall, our results suggest that the observed increase in helpline calls during the first wave of , and the seven-day moving average of newly confirmed COVID-19 cases per million population and day in red (left axis) [31]. Shaded areas indicate first-and subsequent-wave periods. Plots on the right show estimated coefficients for binary variables denoting the two periods with 95% confidence intervals, based on separate linear probability regression models with dependent variable set to one for calls related to the respective topic. Models include year, week-of-year, and day-of-week fixed effects. See Methods,equation (4). Standard errors are clustered at the week level. Supplementary Table S7 shows detailed regression results.
Calls related to suicidal ideation decreased when case numbers rose, increased when contain- diverse sets of helplines and nations are less of a concern in this dataset. Moreover, calls to the 156 Lifeline have been used to predict the incidence of actual suicides, e.g. by [12]. As a helpline focused 157 on suicide, however, the Lifeline does not allow us to track changes in the composition of mental 158 health problems.  Table S8 for the precise 169 model specification). The left-hand chart shows estimates from data that are pooled across the entire 170 duration of the pandemic. We find that increases in SARS-CoV-2 infections, ceteris paribus, were  and more generous income support policies were followed by drops in the number of helpline calls 197 due to suicidal ideation, whereas more stringent NPIs were followed by more suicide-related calls.

198
These effects are estimated to be statistically significant, and consistent with those based on the U.S.

199
Lifeline. Our findings suggest that public compensation payments for pandemic-induced losses not 200 only reduce economic hardship but also have broader benefits: more generous income support is 201 found to lead to fewer calls because of fear, loneliness, suicidal ideation, physical health concerns 202 and, as expected, economic anxiety ('livelihood', see Fig. 5).

204
Helpline call data Our sample of helplines includes large, general crisis helplines, dedicated suicide preven-205 tion helplines, as well as some helplines that focus on specific groups such as children, parents, or immigrants.

206
The selection of sample headlines was based (a) on an internet search of well-documented helplines and 207 (b) on receiving data from those helplines. Of 154 helplines contacted, we received data from 37 helplines.

208
The information from 23 of these was of sufficiently detailed coverage and consistency to be included in our 209 pooled analyses. Table 1 lists the included helplines, classified based on the format of the available data. The 2019. To summarize the overall dynamics, we estimate the following model: The dependent variable is the natural logarithm of the number of calls to helpline h recorded on day t . We

227
The model includes helpline fixed effects ξ h , interacted with year, week-of-year and day-of-week indicators, 228 summarized in the vector Θ t . The reference category is the week of the pandemic outbreak, and the coefficients 229 γ τ allow us to track the percentage deviation in daily calls compared to week 0. See Figure S1 and applies to the group of callers older than 60, and the middle category in some cases includes also individuals 250 whose age is slightly below 30 years or above 60.

251
For Fig. 2   indicators, so that the coefficients represent the group-specific changes in topic shares: For the main effects of caller sex and age groups, indicators for the reference group of male callers in the As we analyze the two helplines separately, we do not include helpline fixed effects here, but capture secular  targeting to arrive at a value between 0 and 100. As we rely on the sub-national information on policies in 299 US states, it is important to note that we use the total index scores: whenever national policies were more 300 restrictive than those of individual states, the higher score is imputed. Data on the daily number of newly 301 confirmed COVID-19 cases and deaths are taken from the JHU CSSE data repository [31].

302
The panel structure allows us to exploit the idiosyncratic variation within states j over time (weeks w ) 303 while controlling for overall trends. We estimate a two-way error component model as illustrated in equation 5: The dependent variable is the natural logarithm of the number of calls plus one, COVID-19 is defined as

311
To investigate the extent to which the relationship changed over time, we re-estimate the model as in equation 6. Here, we include the three main explanatory variables interacted with two indicators variables that are set to 1 for the time period from January to July 2020, and for September 2020 to March 2021 respectively.
For the analysis in Fig. 5, we combine the previous approaches and estimate the relationship between call volumes and the three variables as illustrated in equation 7, based on topic-specific call volumes to Telefonseelsorge (Germany) and S.O.S. Amitié (France) during the time from January 1, 2019 to March 31, 2021.    Note: Evolution of total helpline call numbers during the first wave. The dependent variable is the natural logarithm of daily helpline call volumes. The sample includes the period from 4 weeks before to 12 weeks after the event date in 2020, as well as the corresponding days of the year 2019. Coefficients show the average percentage change in daily calls compared to reference week 0, which represents the week of a) the pandemic outbreak, defined as the date when the cumulative number of COVID-19 infections exceeded 1 per 100,000 inhabitants, or b) the date when shelter-in-place orders first entered into force. Standard errors in parentheses are clustered at the helpline-month level.  Note: Overall distribution of calls included in the estimation sample (Fig. 2).

0.35
Note: Overall shares of topics in the estimation sample, January 2019-June 2020 (Fig. 2), by helpline.         Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S3: Daily calls to Telefonseelsorge, Covid-19 incidence and government response in Austria  Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S5: Daily calls to Tele-Onthaal, Covid-19 incidence and government response in Belgium  Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S8: Daily calls to the Hope 24 Line, COVID-19, and government response in China S19     S.O.S. Amitié consists of a federation of several regional charitable organizations that was founded 1974, two 504 years before the helpline was established. The aim is to provide emotional support to anyone in emotional 505 distress, struggling to cope, or at risk of suicide via listening service by telephone, messaging and chat. S.O.S.

506
Amitié is a federation of 44 regional associations bringing together 50 listing stations. The anonymous listening 507 on the phone is offered 24 hours a day, 7 days a week and in chat every day from 1 p.m. to 3 a.m. For further 508 information, visit www.sos-amitie.com.

509
We received data on individual calls received for the time from January 1, 2019 to March 31, 2021, including 510 information on caller's sex, age group, and conversation topics. Counselors select up to three topics for each 511 the conversation, and topic categories are therefore non-exclusive.   Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S14: Daily calls to Telefonseelsorge, COVID-19, and government response in Germany S27  160 hours. Further information is available online at www.mutes.de.

528
We have received data on individual calls received during the time from 1 January 2020 to 8 December 529 2020. Data collection on conversation topics follows the classification of Telefonseelsorge, see Table S17.  We have received data on indivudual calls, covering the time from 23 January 2020 to 6 October 2020, 550 including the sex of callers. outbreak, the capacity to answer calls was at first lowered due to working-from-home requirements, but then 556 increased as an additional line was introduced. For further information, visit sos116-123.hu.

557
We received time series of daily call volumes, covering the time from January 1, 2019 to May 31, 2020. (from 9pm until midnight, 6 days a week). Further information can be found online at www.sahar.org.il.

567
We received anonymous chat data from with information on gender, age category, and the content that Index score/cases rate Jan 2020 Apr 2020 Jul 2020 Oct 2020 Jan 2021 Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S26: Daily calls to Sahar, Covid-19, and government response in Israel S40  established. There are 20 centers across Italy providing free and anonymous web, chat and telephone service 574 from 10 am to 12 pm each day. Further information is available onlne at www.telefonoamico.it 575 We received data on individual calls (incl. chats) from January 1, 2019 to June 1, 2020, including information 576 on callers' sex and age group, as well as problems of callers and conversation topics. Additionally, the data 577 includes the occupational status, living situation, call duration, and the sub-national region of origin. Calls Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S28: Daily calls to Telefono Amico, COVID-19, and government response in Italy      Amiga operated from 4 pm -24 pm. Due to updated communications equipment at the telephone operator, 616 capacity increased after the pandemic, and working hours were prolonged to range from 3.30pm until 00.30.

617
Further information is available online at www.sosvozamiga.org) 618 We received data on individual calls from 2 January 2019 -15 June 2020 with information on caller sex and 619 age group. Index score/cases rate Jan 2020 Apr 2020 Jul 2020 Oct 2020 Jan 2021 Daily helpline calls (7-day moving average) Government response stringency index Newly confirmed Covid-19 infections/million population Figure S37: Daily calls to Zaupni Telefon Samarijan, COVID-19, and government response in Slovenia S52