This page outlines the research methods used in the report 'One year on: how the COVID-19 pandemic is affecting wellbeing and suicidality'
The research in this report brings together:
- Analysis of anonymous service data that is routinely collected about Samaritans calls and emails
- Primary research with our listening volunteers through regular surveys
- Secondary analysis of findings from focus groups with health and social care helpline volunteers
- Analysis of a subset of data collected as part of the UK COVID-19 Mental Health & Wellbeing study (MHWS)).
Further details about the data and the analyses for each are detailed below.
1. Analysis of Samaritans service data
This analysis was carried out to understand trends in contacts to Samaritans, including the callers’ most common concerns and what changes our service had experienced compared to the previous year.
After every contact, Samaritans’ volunteers complete a log about the nature of the contact. This provides data to help us understand how people use our service and to some extent who they are, which can help us understand their needs.
Over the year since restrictions began, people could contact Samaritans by phone call, email, letter, online chat when pilots were running or face-to-face when branches are open. Throughout the report we use the term 'contacts’ to refer to conversations with people by any of these methods.
However, this data is somewhat limited, and are more focused on the conversations we have with callers rather than our callers themselves. Our listening volunteers do not specifically gather data about callers, and callers are not required to disclose details about themselves. This means that volunteers can focus on the caller’s concerns and data collection doesn’t get in the way of providing support to people when they need it. Therefore, volunteers only record the information that is voluntarily disclosed during the natural course of the conversation. For instance, while we record a caller’s gender when it is known, in 3% of our phone calls and 7% of all contacts this remains unknown.
In certain circumstances, such as a safeguarding issue, identifiable information may be recorded, but no identifiable information is ever used for research purposes.
This research involved the analysis of 3 sets of Samaritans service data, described below.
Helpline and email dataset:
For the purpose of this research, we used a dataset of over 4 million anonymous contact records. This included contacts where emotional support was provided, by telephone and email, across a two-year period (23 March 2019 to 22 March 2021). The dataset included:
- Broad topics that raised by the caller as a concern during the contact (eg, loneliness)
- Method of contact
- Gender of caller, where known
- Age of caller, where known
- If suicidal feelings or behaviours were expressed during the contact
- If self-harm was discussed during the contact
Healthcare workers dataset:
We used a subset of the ‘Helpline and email dataset’ above of over 17,000 anonymous contact records where:
- emotional support was provided by telephone and email
- The contact was in the 1-year period (3 April 2020 - 22 March 2021).
- The volunteer logged the contact as having been from a ’NHS Worker’.
This dataset includes most answered calls to the dedicated NHS Helpline in England and Wales, as well as calls to the main helpline where the caller mentioned that they were an NHS worker.
Prison contacts dataset:
We used a data set of 120,000 anonymous contact records. This included contacts where emotional support was provided by telephone or face-to-face within prisons over a 2-year period (23 March 2019 to 22 March 2021), across two sources of support:
- Contacts to the general UK Samaritans helpline which were known to be from a prison setting (a subset of the ‘Helpline and email dataset’ above)
- Face-to-face contacts within prisons settings across UK and Republic of Ireland, delivered by a peer support Listener.
- Descriptive analyses
We carried out a general, descriptive analysis of the variables listed above for the overall post-pandemic dataset (ie, the year since restrictions began). This provided weekly descriptions of the overall volume of contacts during this time, as well as the volume of specific types of contacts (ie, broad concerns, suicidal feelings or behaviours, and self-harm).
- Comparisons with pre-pandemic periods
Visual comparisons of graphs between the year pre- and post-pandemic (with ‘pandemic’ defined as 23 March 2020, when the UK went into lockdown for the first time) were conducted across the variables listed above, with any trends or differences tested for statistical significance using chi-square tests.
This was done for all three datasets: telephone and email contacts, healthcare workers contacts, and prison contacts.
Healthcare worker status was not routinely collected until after lockdown began, so this data covers the period from 3 April 2020 - 22 March 2021 and we do not have comparison data from the previous year. Similarly, coronavirus was only recorded as a caller concern from 4 April 2020 onwards.
- Comparisons between groups
Visual comparisons of graphs between the following sub-groups were conducted across the variables listed above, with any trends or differences tested for significance in line with the above. This was conducted for both the pre- and post-pandemic years and explored differences in week-by-week trends during the post-pandemic year.
Any differences for the analyses listed above were tested for significance using chi-square tests; all reported data is significant to p<.001. Results that were not significant are not reported. Analysis was conducted using RStudio.
2. Volunteer surveys
Surveys of volunteers were collected regularly throughout the year since the pandemic began to deepen our understanding of the service data and how common concerns were changing in nature as the pandemic progressed.
Qualitative data was collected through seven online surveys with Samaritans volunteers in the UK and ROI throughout the year since restrictions began. The surveys were distributed through Samaritans’ internal communications channels.
The surveys asked volunteers who had completed at least one listening shift in the relevant time period (eg, the month prior to the survey) to reflect on trends in what callers talk to us about and how this changed from month to month. We asked volunteers about:
- Discussion of suicidal thoughts and attempts
- The most common caller concerns and how these change over time
- The ways people used the Samaritans service
- Specific concerns of key groups, such as people with pre-existing mental health conditions, young people, men and healthcare workers.
We received 10,766 responses over the 7 surveys, which contained a mixture of open and closed text responses.
Descriptive analysis of closed-text questions was completed in SurveyMonkey. Open-ended responses were analysed using thematic analysis, initially using a deductive coding approach. Once a framework was established, an inductive coding approach was used to build on findings from each survey and gather a picture of trends throughout the year.
3. Qualitative research with volunteers of the NHS and social care worker helpline
As part of the internal evaluation of this new service, which is available in England and Wales, focus groups were conducted to identify early insights into how the helpline was working. The focus groups were carried out in November 2020 by Humankind Research on behalf of Samaritans. The topic guide for the focus groups was informed by a survey of 110 volunteers who provided dedicated support through the health and social care helpline.
This research used the briefings and reports summarising the findings from this internal evaluation activity to supplement the analysis of the healthcare workers dataset described above (a). The original qualitative research consisted of seven 90-minute online focus groups with volunteers on the helpline for health and social care workers (26 volunteers in total).
Secondary analysis of the findings and triangulation with findings from analyses from analysis of the quantitative dataset unique to this research.
4. Analysis of UK COVID-19 Mental Health & Wellbeing Study data
The UK COVID-19 Mental Health & Wellbeing Study (MHWS) is a national wellbeing tracker survey, which Samaritans has jointly funded and collaborated on with the Suicidal Behaviour Research Lab at The University of Glasgow and Scottish Association for Mental Health.
This survey involved a nationally representative baseline sample of 3,077 adults across the UK ran from 31 March throughout the year and is still ongoing at the time of writing. Its purpose is to understand the impact the coronavirus pandemic is having on key psychological factors related to suicide risk.
Additional analysis was conducted for this report to explore help-seeking behaviours during the pandemic and between sub-groups.
A subset of the data collected via the first six waves of this longitudinal survey were provided for analysis in this research. The data were originally obtained via a quota survey design and a sampling frame that permitted recruitment of a nationally representative sample. A total of 3,077 adults in the UK completed the first wave of the survey through an existing online UK panel (Panelbase.net), dropping to 2,283 adults by wave 6. Drop off in participation was highest among younger adults (aged 18-29 years) and non-white participants.
The variables within this dataset were related to sociodemographic characteristics (age, gender, sociodemographic status) and whether, and how often, individuals had sought help from:
- Friends or family
- Samaritans by telephone
- Other helplines or voluntary support services
- Professional counselling or therapy
- Community groups/clubs
- Social media/online
Analysis was conducted by Heather McClelland from the Suicidal Behaviour Research Lab at The University of Glasgow using Stata MP 16.
- Data was recoded as follows: all help-seeking answers were recoded to ‘ever’ vs ‘never’, gender was coded into binary data (male=1, female= 2), age was grouped into three categories (18-29, 30-59, 60+ years), socio-economic status was divided into high/ low income.
- Adjusted odds ratios (ORs) with 95% confidence intervals (95% CI) were used to report binary sociodemographic characteristics (gender, socio-economic status)
- Chi square (χ²) was used to report differences between categorical age groups.
- Bonferoni adjustments were used to report post-hoc analysis of chi-square data.