The impact of the COVID-19 pandemic on presentations to health services following self-harm: systematic review

Background Evidence on the impact of the pandemic on healthcare presentations for self-harm has accumulated rapidly. However, existing reviews do not include studies published beyond 2020. Aims To systematically review evidence on presentations to health services following self-harm during the COVID-19 pandemic. Method A comprehensive search of databases (WHO COVID-19 database; Medline; medRxiv; Scopus; PsyRxiv; SocArXiv; bioRxiv; COVID-19 Open Research Dataset, PubMed) was conducted. Studies published from 1 January 2020 to 7 September 2021 were included. Study quality was assessed with a critical appraisal tool. Results Fifty-one studies were included: 57% (29/51) were rated as ‘low’ quality, 31% (16/51) as ‘moderate’ and 12% (6/51) as ‘high-moderate’. Most evidence (84%, 43/51) was from high-income countries. A total of 47% (24/51) of studies reported reductions in presentation frequency, including all six rated as high-moderate quality, which reported reductions of 17–56%. Settings treating higher lethality self-harm were overrepresented among studies reporting increased demand. Two of the three higher-quality studies including study observation months from 2021 reported reductions in self-harm presentations. Evidence from 2021 suggests increased numbers of presentations among adolescents, particularly girls. Conclusions Sustained reductions in numbers of self-harm presentations were seen into the first half of 2021, although this evidence is based on a relatively small number of higher-quality studies. Evidence from low- and middle-income countries is lacking. Increased numbers of presentations among adolescents, particularly girls, into 2021 is concerning. Findings may reflect changes in thresholds for help-seeking, use of alternative sources of support and variable effects of the pandemic across groups.

The COVID-19 pandemic has led to a deterioration in population mental health and has placed considerable additional strains on health systems. 1,2 The pandemic has also heightened many of the risk factors for suicidal behaviour, such as job insecurity and unemployment; access to food, education and healthcare; and the availability of family and community support. 3 Understanding and quantifying trends in help-seeking for self-harm is a vital part of the public mental health response to COVID-19. It could help to expound the apparent paradox observed during the early stages of the pandemic, whereby although population mental health deteriorated, 4 fewer people sought help for their mental health from primary and secondary care services. 5 Examining self-harm presentations across health settings could help understand longer-term population effects and inform planning of services and interventions in the future phases of the pandemic. Numerous studies from high-income countries reported marked reductions in health service utilisation during the second quarter of 2020, following the start of the COVID-19 pandemic. For example, considerable reductions in diagnoses for acute physical and mental illnesses were found in the UK after introduction of the national lockdown in March 2020, with only partial recovery by July 2020. 6 In another UK study, reductions of around a third in health service contacts specifically for self-harm were found. 7 Focusing specifically on hospital admission for self-harm, overall reductions of just over 8% were reported in France, although increases in more serious potentially lethal acts of self-harm were observed. 8 Evidence relating to the indirect health impacts resulting from the pandemic in lowand middle-income countries also suggests that care for non-communicable diseases and mental disorders has been severely disrupted. 9 A systematic review on the impact of the pandemic on suicide and self-harm in low-and middle-income countries found mixed evidence, with either a decrease or no discernible impact in reported self-harm episodes, along with increases in certain age groups. 10 In 2020, a living systematic review was established to provide an up-to-date resource and data synthesis of evidence on the impact of the COVID-19 pandemic on self-harm and suicidal behaviour. 11 The most recent update of the review included studies up to 19 October 2020 and included 20 health service utilisation studies, including 11 focusing specifically on health service presentations following self-harm/suicide attempts. 12 The review reported that most studies reported a decrease in presentations to health services for self-harm during the early months of the COVID-19 pandemic. However, all 20 studies were on high-income countries, and the latest month of observation was August 2020. [13][14][15] In the subsequent months, many health services adapted and 'stay-at-home' orders have eased, although these restrictions later returned in *Joint contributors.
The British Journal of Psychiatry (2022) 221, 603-612. doi: 10.1192/bjp.2022.79 many countries and regions. Although studies suggest service utilisation had returned to expected volumes in some countries by the third quarter of 2020, 12 it is not known how subsequent restrictions and ongoing pressures on health systems in response to further waves of COVID-19 have affected help-seeking and access to healthcare for self-harm. In this article, we report on evidence concerning the frequency (reported incident or prevalent episode counts or rates) of presentations to health services following selfharm after the onset of the pandemic, compared with before the pandemic. There has been no synthesis of studies published since October 2020, some of which would be expected to include the later observation periods covering the latter months of 2020 and first half of 2021, as the pandemic continued to affect populations globally. Our aim was to systematically identify, review and synthesise evidence relating to presentations to health services for selfharm since the COVID-19 pandemic began in the first quarter of 2020.

Method
The protocol for the methodology applied in conducting the systematic review is registered within a living systematic review of the impact of the COVID-19 pandemic on self-harm and suicidal behaviour (PROSPERO identifier CRD42020183326; registered on 1 May 2020). 5,11,12 Additional inclusion and exclusion criteria specific to our research question were applied, and further screening, data extraction and study quality assessments were conducted. To address our research question, 'Did the frequency of health service presentation for self-harm during the pandemic change compared with antecedent periods?', we applied the following inclusion and exclusion criteria.

Inclusion criteria
The following inclusion criteria were applied for each study: (a) published from 1 January 2020 to 7 September 2021; (b) written in any language; (c) investigation of health service utilisation among the general population, including presentations to general hospital emergency departments, primary healthcare services, specialist mental healthcare services (accessible to general population), other secondary healthcare services that treat people who have selfharmed/attempted suicide (e.g. surgery) and admission to hospitals; (d) outcomes were presentations for self-harm, including broad definitions of self-harm (defined as non-fatal intentional selfinjury, intentional self-poisoning involving drugs or non-ingestible substances, including non-suicidal acts) or attempted suicide (including hospital attendance and/or admission for these reasons) 11 and narrower definitions, e.g. studies focused only on suicide attempts or specific methods of self-harm; and (e) comparisons in health service presentation frequencies (including incident or prevalent episode counts or rates) for self-harm before and after the beginning of the COVID-19 pandemic, considering specific time periods separately (e.g. both initial and subsequent lockdown periods).

Exclusion criteria
The following exclusion criteria were applied for each study: (a) studies without pre-pandemic observation periods or measurements, including those reporting use of service initiatives implemented in response to the pandemic, with no pre-pandemic comparison period; (b) reports where only an abstract was available; (c) studies focusing on specific groups, such as those with a specific physical or psychiatric diagnosis (including COVID-19), or where the baseline population was existing patients within a specialist service, such as psychiatric in-patients; (d) studies reporting selfharm and suicidal thoughts as a combined measure; (e) studies reporting proportions of self-harm presentations, without reporting absolute figures; and (f) studies of suicides.

Data analysis
The list of studies used for screening was obtained from the main living systematic review database. 12 This database is updated automatically, using daily electronic searches of multiple databases (World Health Organization COVID-19 database; Medline; medRxiv; Scopus; PsyRxiv; SocArXiv; bioRxiv; COVID-19 Open Research Dataset, PubMed) (for the search strategy for each database, K. and R.T.W.) completed daily assessments of the automated results, which included basic data extraction and assigning studies manually to a study design category, along with a description of the study design.
Identification and screening of studies for the current review was conducted with a methodology developed as part of an existing living systematic review (Fig. 1). 12 Studies included publications identified in the living systematic review from 1 January 2020 up to 7 September 2021. Screening was conducted according to the inclusion and exclusion criteria for the current review. The list of studies was extracted from the main living systematic review database on 14 September 2021. Categories assessed for inclusion in the current review were 'service utilisation', 'before/after studies', 'time trends analysis' and 'examination of electronic health records' (Fig. 1). 16 As part of the identification and screening procedure, further screening and data extraction was completed for the current systematic review, using a proforma designed to collect standardised information from each study (Supplementary Table 1). Study quality and risks of bias were assessed with an adapted version of an existing National Institute for Health (NIH) quality assessment tool, designed specifically for studies using before and after designs. 18 The NIH tool was adapted by authors D.K., J.P.T.H. and D.G. to include consideration of the pandemic and associated lockdown periods and other societal restrictions as the intervention of interest, and to account for the use of health service data sources in the study designs. The overall assessment tool was used to judge the quality of evidence relating to frequency of presentation to health services following self-harm within the studies, rather than the overall study quality, with predefined criteria established for evidence to be rated as high or moderate quality. Studies were assessed according to all criteria listed in the tool, although a study was only assessed as being of high or moderate quality if questions 3, 6, 7 and 8 all scored yes. Screening, data extraction and quality assessments were conducted by S.S. A second rater (D.D) assessed eligibility for 18% (26 out of 144) of the studies sought for retrieval, and conducted independent data extraction and analysis on 10% (five out of 51) of the included studies. There was agreement on all eligibility assessments (the independent reviewer reached the same decision to include or exclude the 26 studies in the systematic review) and study quality ratings (the quality ratings of five out five studies independently reviewed were the same). If a source was not available in English, data extraction was conducted by expert reviewers fluent in the language that the article was written in. Where included studies were preprints, searches for peer-reviewed version were conducted and the updated peer-reviewed version was used for data extraction where available. Data synthesis was conducted by extracting, assessing and tabulating key aspects of the studies, including setting, study design, data sources, outcome measures, follow-up and comparison periods, main findings and study quality. The main effect measure of interest was percentage difference in presentation frequency during a defined COVID-19 period compared with a pre-COVID-19 comparison period. If this data were missing, the overall direction of change (e.g. increase/no change/decrease) was recorded. Higher-quality studies were prioritised and reported separately during data synthesis and presentation of results.

Findings of included studies
Almost half (47%, n = 24) of the studies reported reductions in presentation frequency (Fig. 2) for the duration of the period studied, the majority of which included months no later than August 2020. All six studies rated as high-moderate quality (including one preprint) found decreases in frequency of presentations during the early months of the pandemic, with reductions of between 17 and 56% reported. 6,7,29-32 These studies were of  16,17 Self-harm during the COVID-19 pandemic primary and secondary care settings combined (four studies), emergency department presentations among ages 18-25 years (one study) and self-poisoning presentations to hospital (one study). Four studies used healthcare records in the UK to compare expected versus observed primary and secondary care-recorded episodes of self-harm, and found reductions of between 26 and 44%. 6,7, 30,32 Another study based in Sri Lanka found a 32% reduction in hospital presentations for self-poisoning compared with pre-pandemic numbers. However, these estimates included months no later than August 2020. Five studies (including one preprint) used national or nationally representative data. Four of these were assessed as high-moderate quality and reported decreases in presentations to health services of between 26 and 56%. One moderate quality study reported a 6% increase in emergency department presentations. 33 This USA-based study only included self-harm episodes classified as suicide attempts, and therefore may not reflect service use for self-harm more broadly.
Increases were reported in 15 (29%) studies, none of which were assessed as being of high-moderate quality and five of which were rated as moderate quality. An examination of the number of people admitted to a surgical department following self-harm by ingestion of corrosive substances was found to increase by 55% in one Bangkok hospital, although numbers in the study were relatively low. 34 Other moderate quality studies reporting increased patient numbers included emergency department, ambulance and surgery services, which are settings that are likely to be encountering patients with more medically severe episodes of self-harm.
Twelve out of 51 (24%) studies (two were preprints) reported no change in frequency of presentations to health services, including no high-moderate quality studies and two assessed as moderate quality. These were both conducted in emergency department settings, with one New Zealand emergency department reporting no change in self-harm presentations 35 and a UK-based study reporting no change in hospital admission following emergency department presentations for self-harm. 36 A further six studies were conducted in trauma settings, although all were rated as low quality.
Most studies (n = 46) included up to a maximum of 8 months of follow-up from the first wave of the pandemic (March to October 2020). Among the four studies including months from 2021 in their observation period (up to May 2021), three were rated as high-moderate quality. Among these, two studies of primary and secondary care-recorded self-harm reported longer-term reductions of between 8 and 30%, respectively, 30,32 and another study of emergency department presentations by young people aged 12-25 years found no overall change. 29 Studies including follow-up months beyond 2020 were limited to those originating from high-income countries.

Findings by study settings and subgroups
Seven out of 51 (14%) studies were conducted in upper-middle income (n = 3), middle-income (n = 1) and lower-middle income (n = 3) countries, one of which was rated as high-moderate quality. Four studies found a decrease in service use and three reported an increase. The study rated as high-moderate quality reported on self-poisoning episodes in a lower-middle income setting; using health record data from a toxicology unit in a Sri Lankan hospital, a 32% reduction in hospital presentations for self-poisoning was found, compared with pre-pandemic numbers. 31 A study of moderate quality conducted in one Nepalese emergency department found an increase of 44% in presentations for self-harm during the lockdown period compared with the same period the previous year, with indications that severity of self-harm was higher, although the numbers of presentations in both the lockdown and comparison periods were relatively small. 37 Eighteen studies included examination of service use for selfharm specifically among children and/or young people, with five rated as high-moderate quality. One high-moderate quality study including approximately 71% of emergency departments in the USA, across 49 states, examined presentations among ages 18-25 years and found reductions of 26% among ages 12-17 years and   Fig. 2 Percentage change in presentations to health services ordered by country income level and latest month of study observation period, grouped by study quality. Studies are grouped by country income level and ordered alphabetically within income level. The size of the data points corresponds to study quality, with larger bubbles denoting higher-quality studies. Some studies report percentage changes for more than one time period and therefore have more than one data point. *Zero at baseline. ¶ −26% among ages 12-17 years and by −17% among ages 18-25 years. § −31% for all episodes and −41% for incident episodes.
17% among ages 18-25 years in April 2020. However, when examining presentation rates over the longer term through to March 2021, increases compared with equivalent weeks in 2019 were found for girls aged 12-17 years. Among boys aged 12-17 years and all adults aged 18-25 years, rates through to March 2021 were in line with those in 2019. 29 Another high-moderate quality study, based in the UK, reported increased numbers of presentations to primary and secondary care among all adolescents aged 10-17 years, up to May 2021. 30 These findings are in contrast to those reported in other moderate-quality studies that used earlier COVID-19 observation periods (up to June 2020), where younger people were found to have significantly fewer self-harm presentations than in the equivalent period in 2019. 28,38 Discussion

Main findings
All of the studies assessed as high-moderate quality reported decreases in numbers of presentations to health services following self-harm, and were conducted in settings reflecting a broad spectrum of self-harm with higher frequency of presentations, such as primary care. We found that settings treating episodes of selfharm with lower frequency and higher lethality, such as trauma admissions and ambulance calls, were overrepresented among studies that reported increased or no change in demand. Among higher-quality studies that included months from 2021 in their observation period, numbers of people seeking help from health services were found to be either closer to pre-pandemic levels, although still lower than expected, or in line with expected numbers. Evidence from 2021 also suggested there was increased utilisation of health services following self-harm among adolescents, particularly so for girls. However, there were relatively few studies including followup months from 2021, and they were limited to those originating from high-income countries.

Strengths and limitations
This systematic review is the first to examine up-to-date evidence regarding associations between the COVID-19 pandemic and frequency of health service presentations for self-harm. An established, peer-reviewed, living systematic review methodology, 11 with ongoing data extraction by a panel of suicide prevention experts, was used as the basis for this review. This approach, along with a specific focus on studies comparing frequency in presentation to health services following self-harm in different settings during the COVID-19 pandemic versus antecedent pre-pandemic periods, enables timely synthesis of the evolving evidence base.
The findings of our study should be interpreted with some important caveats in mind. We excluded six studies that reported self-harm and suicidal thoughts as a combined measure, as it was not possible to make a like-for-like comparison with findings pertaining specifically to acts of self-harm. However, we included studies using a broad range of definitions of self-harm, including those that measured and reported on suicide attempts or self-poisoning methods only. We also did not include temporal trends in the proportion of all presentations that were for self-harm as a primary outcome, because of the limitation that this outcome would be affected by changes in the overall number of presentations for reasons other than self-harm. A minority (n = 4) of studies included in our systematic review were preprints and therefore not peer-reviewed. We considered it important to include preprints to capture the rapidly evolving evidence-base during the COVID-19 pandemic. Finally, one of our aims was to synthesise evidence on health service presentations beyond 2020. Although only four of the reviewed studies included follow-up time from 2021, the evidence relating to 2021 was considerably higher quality, with three out of four rated as high-moderate qualityhalf of all the highmoderate studies that were included in the whole review.
We conducted a comprehensive narrative synthesis of the data rather than a meta-analysis, because of heterogeneity in the pandemic and antecedent comparison periods, definitions of selfharm applied and healthcare settings that studies were conducted in. Performing a meta-analysis will be considered for future updates of the living systematic review. The studies included in our review are of mixed quality and are greatly underrepresentative of middle-and low-income countries. Although we have reported findings according to these characteristics, overall findings should be interpreted in light of these considerations.

Implications and comparison with existing evidence
This systematic review includes an additional 39 studies reporting on health service presentation frequencies since a previous published synthesis. 1 Our findings relating to a fall in presentation frequencies following self-harm during the early months of the pandemic strengthen this evidence base. Furthermore, findings from higherquality studies suggested either there were continued reductions in health service presentations into 2021, although to a lesser extent than earlier months of the pandemic, or that service use had broadly returned to pre-pandemic levels. However, most of the studies came from high-income countries, and these findings cannot necessarily be generalised to low-and middle-income countries. For example, allocation of COVID-19 vaccinations has been disproportionately skewed toward high-income countries. 63 Consequently, many low-and middle-income countries have experienced major subsequent waves of COVID-19 well into 2021. 64 The effects of these further waves of infection on many of the factors associated with self-harm (e.g. unemployment, mental and physical ill health, poor access to healthcare) are likely to be considerable. 65 Subsequent waves of COVID-19 have also been experienced by high-income countries into the latter half of 2021. For example, from November 2021, some European countries introduced further societal restrictions. 66 Continued surveillance is therefore needed in all settings.
Our findings are consistent with reports of increased acuity of presentations in some mental health services. 67,68 The increases in presentation frequency reported by studies that were conducted in healthcare settings treating more potentially lethal episodes of self-harm, such as ambulance calls and trauma admissions, indicates that the pandemic has affected the threshold for help-seeking. Evidence also shows that non-statutory mental health services, such as charities, experienced increased demand in the months following the onset of the pandemic. 69 This may explain the apparent paradox observed during the first year of the pandemic, where deterioration in population mental health alongside reductions in health services utilisation was observed. This indicates that reductions seen in settings capturing a broader spectrum of self-harm do not simply reflect decreased incidence of self-harm or reduced clinical need. For example, a systematic review found increases in prevalence and global burden of depressive and anxiety disorders, both of which are risk factors for self-harm, in 2020 as a result of the pandemic. 4 There is some evidence that, following initial deteriorations in 2020, some people's mental health improved following easing of lockdown measures. 70 However, the subsequent COVID-19 waves and the broader economic consequences of the pandemic have continued to adversely affect the mental health of a large proportion of the population. 70 People who have harmed themselves non-fatally have a markedly elevated suicide risk subsequently, irrespective of self-harm method at the index episode, and degrees of suicidal intent can fluctuate between different self-harm episodes Self-harm during the COVID-19 pandemic by the same person. 71 Therefore, it is vital that people harming themselves receive clinical intervention, and that health services across the world work to ensure services are available to provide timely and accessible care. 72,73 Studies examining changes in proportions of groups presenting with certain characteristics, and those examining combined 'suicidal thoughts and self-harm' outcomes, were not included in this systematic review as we were interested in absolute numbers of people using health services for self-harm. However, such studies can provide valuable information about help-seeking behaviour in different groups. For example, a study of hospital attendance for suicidal ideation and self-harm in Australia's Gold Coast region identified a number of groups with particularly reduced likelihood of presentation during March to August 2020, including Indigenous Australians and individuals with less severe suicidal and self-harm, whereas people younger than 18 years had increased numbers of presentations. 74 Another study conducted in a paediatric emergency department in New York City, USA, found that although overall there were significant decreases in emergency attendances, visits for suicidal ideation and self-harm among young people increased. 75 Increases in numbers of adolescents referred to mental health services in Ireland were found from September 2020, following an initial decline in April 2020. 76 Our findings of increased numbers of presentations to health services for self-harm into the early months of 2021 among adolescents, particularly girls, within this context, are concerning and warrant urgent attention.
In conclusion, all high-quality studies reported a fall in attendance frequency for self-harm during the early months of the pandemic, strengthening earlier evidence. New evidence relating to the first and second quarters of 2021 indicated that longer-term impacts on health services were less marked than during the first wave of the pandemic, although reductions in frequency of presentation persisted compared with expected levels. These patterns likely reflect changes in thresholds for help-seeking, increases in frequency of higheracuity episodes of self-harm and increased use of non-statutory health services. The increased numbers of health services presentations among adolescents, particularly girls, into the early months of 2021 warrants particular attention. However, evidence from lowand middle-income countries is still limited. High-quality, multicentre studies examining the longer-term impacts on health service utilisation for self-harm, particularly in low-and middle-income countries, including observation periods into 2021 and among children and young people, are urgently needed.