The purpose of this review was to estimate the global prevalence of self-harm during COVID-19. We combined the data from 16 studies on self-harm related COVID-19. The obtained pooled prevalence of self-harm suggested that the self-harm closely related to COVID-19 can not be ignored, and the characteristics included in the studies would affect the pooled prevalence of self-harm.
The pooled prevalence of self-harm
The pooled prevalence of self-harm for all participants in our meta-analysis was 15.8%, which was higher than the 8.2% rate of self-harm obtained by Moller et al.,[39] when they surveyed 4126 participants in 2013. The influence of COVID-19 can not be ignored in spite of the fact that the characteristics of the study place and population will play a role in the difference of the prevalence of self-harm among the two studies. As is known to all, the outbreak of COVID-19 has brought a great impact on people's work and life in a short time[4], which was acted as the source of recent stress greatly contributed to the individual psychological pressure[40], can be coordinated self-harm persistent risk factors (impulsivity, adverse childhood experiences, etc.) to strengthen the demand for self-harm to alleviate psychological pressure and achieve rapid emotional release[41, 42]. Moreover, the pooled prevalence of self-harm in this review was slightly lower than that reported in the meta-analysis by Gillies et al.[43], which may be related to the fact that Gillies’s study was conducted among adolescents. And what we need to note is that self-harm is more common in adolescents than in other age group[44]. Not only that, our pooled prevalence of self-harm was much lower than the prevalence of lifelong self-harm investigated by Muller et al.[45]. The studies included in this review were all related to COVID-19 outbreaks within a short period of time. However, the prevalence of lifelong self-harm means that other events in an individual's life rather than just a certain emergency may stimulate his desire to harm himself. In addition, the impact of a certain event on individuals may be revealed after a long period of time, which enlightened that we should concerned about the long-term impact of COVID-19 on self-harm[46].
Subgroup analysis of variables
Further, subgroup analysis was conducted on variables that may affect the heterogeneity of meta-analysis based on literature review and clinical experience, including study place, study time, age, gender, study design, purpose of self-harm, mental symptoms and restrictions. As expected, there were great differences in the pooled prevalence of self-harm between Asia and other continents, that is, the pooled prevalence of self-harm in Asia was significantly higher than that in other continents, which reflected the impact of COVID-19's sudden and explosive nature on self-harm. And our finding was consistent with previous studies found that the adverse mental health effect (i.e. self-harm) has been observed worldwide, notably in the Asia Pacific region, dominantly in countries such as China where the first COVID-19 case was reported in 2019[47–49]. As the first continent to discover COVID-19 and continue to spread, Asia is the first to feel the uncertainty and threatening nature of the epidemic[49]. People in Asian countries affected by the epidemic will undoubtedly shoulder the psychological burden caused by the epidemic[40], while people from other continents will buffer the psychological pressure for a certain period of time. This may explain the higher prevalence of self-harm in Asia during COVID-19. With regard to study time, we observed that studies conducted prior to July 2020 had a slightly higher prevalence of self-harm than studies conducted after July 2020. The psychological impact on individuals in the early stage of COVID-19 may be more significant than that in the later stage due to its uncertainty, severity, and persistence[50], increasing the possibility that individuals resort to self-harm to relieve negative pressure[24]. Besides, Patwary et al.[51] found that some social media in early COVID-19 may disclose unconfirmed COVID-19 information, which will not only eliminate public doubts about the epidemic, but also aggravate psychological burden, especially for adolescents with low ability to distinguish the authenticity of social media[52], who are the high-risk group of self-harm[44]. Unfortunately, due to the limitations of the included studies, we used July 2020 as a time dividing point to describe the different stages of of development of COVID-19, but this time may not be representative, so the prevalence of the two time periods did not show a particularly significant difference.
In terms of study design, the pooled prevalence of self-harm was higher when the included cross-sectional studies were combined. The descriptive data obtained from cross-sectional studies are collected at a certain time point or in a short time interval[53], which objectively reflects the data characteristics of this time point. In this study, cross-sectional design was used to collect data only during COVID-19, which indicated the change of the prevalence of self-harm due to COVID-19. It should be noted that the long-term impact of COVID-19 on self-harm also requires other types of study design (i.e. cohort studies). Furthermore, if the samples of the included study were recruited from hospitals or schools, a higher prevalence of self-harm was reported, which was confirmed by other study[54]. The reason for this may be discussed through the following explanations. The samples from hospitals may be affected by the disease, which may lead to produce anxiety in the recovery of the disease and the acquisition of regular treatment during COVID-19, especially the patients with mental disorders who are easy to be influenced by the outside world and are not likely to respond positively[55]. While the trend of the prevalence of self-harm in age can be reflected by the samples from schools, that is, adolescents may be more prone to have self-harm.
Complementally, there were significant differences in the prevalence of self-harm among different age groups. Specifically, studies that included only adolescents reported a higher prevalence of self-harm than studies that covered all age groups, which was supported by other studies[43, 56]. Adolescence is a vulnerable phase for developing self-harm, as elevated levels of impulsivity and emotional reactivity are present due to brain developmental processes[57], so adolescents have weak ability to control their own emotions and are prone to adopt self-harm due to external influences such as COVID-19. Consistent with previous studies[57, 58], our results indicated that the prevalence of self-harm among female was higher than that among male. Female are more likely to take self-harm due to inner emotional factors (e.g., “I felt very depressed”, “to escape painful memories”), while male are more likely to engage in self-harm for interpersonal reasons (e.g., “it makes me more gregarious”, “to makes me more masculine”)[59, 60]. It is clear that COVID-19 has brought more bad inner experiences and negative emotions to individuals. Additionally, self-harm among male focuses on social rather than emotional factors, they have other strategies to achieve their goals rather than self-harm, including aggression and alcoholism, etc[58].
In particular, we made subgroup analysis according to the purpose of self-harm. And the result revealed that NSSI had a higher prevalence. The self-harm literature is increasingly moving towards a separation of suicidal and NSSI, as outlined in the Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition (DSM-5)[61]. Possible explanations for the high prevalence of NSSI should start from the special functions of NSSI. On the one hand, NSSI has a low-cost and immediate effect in eliminating unpleasant emotional states[62], while COVID-19's outbreak and popularity are often accompanied by negative emotions, including fear, sadness, tension and anxiety and despair[63]. On the other hand, influenced by COVID-19's infectivity, remote life, virtual classroom and lockdown are the main lifestyle to maintain physical and social distance[64], which seriously affects social interaction and increases the feeling of emptiness and loneliness. Whereas NSSI can provide stimulation by experiencing strong emotions and get rid of the feeling of emptiness and loneliness[63].
In accord with other studies[54, 62], we found that respondents with mental symptoms (depressive, anxiety symptoms, etc.) reported a higher prevalence self-harm than that without mental symptoms. Likewise, previous studies have also shown that the main risk factors of self-harm include accompanying mental symptoms, especially mental illness involving mood disorders[54]. It’s well known that psychiatric patients with emotional regulation disorders have difficulty in regulating negative emotions caused by negative events, such as the inability to cope with negative emotions caused by COVID-19[55], while self-harm has been proved to be a coping strategy that can regulate emotions[54]. Hence, careful consideration by caregivers and healthcare system adaptations to allow for mental health support should be required to reduce the risk behaviors of patients with mental symptoms despite the restrictions of COVID-19. Finally, we separately estimated the pooled prevalence of self-harm in both groups based on whether restrictions were applied in post-COVID-19 studies, suggesting that the group with restrictions reported a higher prevalence of self-harm. To abate the rate of infection, global Governments have imposed restrictions to some extent, including restrictions on social activities, shopping, exercise[65]. However, we all know that social interaction is indispensable in providing psychological support and help, and the adoption of restrictions is likely to have a negative impact on mental health and well-being[65]. It should be emphasized that objective social isolation and subjective loneliness are associated with a higher prevalence of self-harm[66]. Notably, many patients and their families forgo or delay health care due to fear or the decreased access to medical services during the lockdown[67], which is not conducive to the rehabilitation of patients and increases the anxiety and worry of patients, especially patients with chronic illness such as mental disorders. In brief, the employment of restrictions during COVID-19 may exacerbate negative emotions and worsen them, which may be an incentive for individuals to take self-harm. And this enlightens us that multifunctional social software, home exercise programs, strategies to enhance relapse prevention and the use of alternative approaches as e-health technologies need to be implemented[68].
Sensitivity analysis and publication bias
Based on sensitivity analysis, the results of this review were robust and reliable. Yet it must be recognized that the quality of most included studies are indeed at a medium level. Since the emergence of COVID-19, more studies has focused on the reduction of infection rate and the treatment of diagnosed patients in a short period of time with limited resources, and the number of studies involving mental health (self-harm) was limited. Most of the studies in related fields was in its early stages, which may affect the study quality to a certain extent. In addition, the sudden outbreak of COVID-19 makes researchers eager to find the mental health outcomes of COVID-19 as soon as possible, so as to take targeted measures as far as possible. Due to the inadequate consideration of study design or study scheme, it is likely to be detrimental to the study quality. What’s more, since the outbreak of COVID-19 has only lasted for about two years, all studies have failed to explore the long-term impact of the epidemic on self-harm, which will undoubtedly have an impact on the study quality. The impact of COVID-19 on individuals is profound and lasting, especially on mental health. As an important manifestation that is not conducive to mental health, self-harm is likely to be used by individuals to quickly regulate emotions and alleviate negative emotions. There will be more research on the self-harm during COVID-19, especially in specific countries or groups. And this present study can reveal the research status to a certain extent, promoting the improvement of the quantity and quality of related studies in the future.
As for publication bias, the funnel plot of this meta-analysis showed that there was a certain publication bias, while Begg’s test and Egger’s test were opposite. Significantly, the results of the trim-and-fill analysis for self-harm displayed that there was no significant changes in the estimate of combined effect size. There were inconsistent results between funnel plot and Begg’s test, Egger’s test on the existence of publication bias, which may not be able to bypass the "confounding factor" of heterogeneity. Specifically, although the funnel plot in meta-analysis is an intuitive method to test publication bias, not all funnel plot asymmetries are caused by publication bias[69]. Some statisticians proposed that it is inappropriate to use funnel asymmetry to judge publication bias when there is great heterogeneity among the included studies (I2 > 75%)[70]. And the heterogeneity of the studies included in this review is relatively large (I2 > 99%) because the prevalence of self-harm estimates were based on heterogeneous populations. As a result, future meta-analysis in related fields may consider focusing on a certain type of NSSI in a certain population, a certain country or other targeted characteristics to minimize the source of heterogeneity and weaken its interference on publication bias. Not only heterogeneity, we need to point out that there may be some other reasons for the asymmetry of funnel plot that need to be further explored.
Limitations
Several limitations should be admitted. Firstly, due to the fact that COVID-19 has only emerged for about two years and there were only a few relevant studies, most of the studies included were observational studies, and inherent biases and differences in the design of observational studies tend to increase the risk of heterogeneity. Secondly, although we incorporate data from a significant period during COVID-19, it would be useful to conduct longitudinal study over longer time spans, as the antecedent factors for, and outcomes of, self-harm may change throughout the lifespan. Next, we performed subgroup analysis of relevant variables based on literature and clinical experience, which may not include some variables that affect heterogeneity. In the subgroup analysis of the study time, the division of the time span of different stages may reduce the accuracy of the findings. Besides, subgroup analysis did not completely solve or explain the obvious heterogeneity. Therefore, our findings should be cited with caution. Moreover, the quality assessment of the included studies were mostly at the medium level, which may affect the study results. Finally, this review was limited by language and region, and did not include non-English or non-Chinese studies. Maybe there is some available information.
Implications
Despite the above limitations, our findings have implications for policy and practice. Compared with other studies focusing on the impact of COVID-19 on mental health, this study focuses on specific topic (self-harm), to a certain extent, arousing the attention of governments around the world and promoting the rational allocation of resources. Most strikingly, the pooled prevalence of self-harm during COVID-19 in this review was not cheerful. Therefore, it also suggests that relevant departments should formulate relevant preventive measures in time to identify high-risk factors of self-harm as soon as possible, such as adolescents, female, groups with mental symptoms or groups with loneliness and emptiness after experiencing restrictive measures. For these risk factors, relevant departments can reduce their adverse effects on individual self mutilation by forming targeted public health intervention measures, including regular psychological assessment of the above-mentioned high-risk groups through the combination of online and offline, interventions to strengthen social interaction, etc. For instance, studies have confirmed that brief contact interventions (i.e. telephone/letter/postcard contact and emergency green cards) may contribute to enhance social support and social contact in a long-distance context to reduce the prevalence of self-harm[71]. More importantly, timely psychological counseling should not be neglected in response to self-harm that has already occurred, and the intervention of professional medical treatment can be considered when necessary.
Four directions for further research are emphasized. First of all, future studies can determine the comparable and long-term impact of COVID-19 on the prevalence of self-harm by establishing an appropriate control group and adequate follow-up. Second, the studies included in this study only represented the situation of self-harm in a few countries, which was considered to be jointly shaped by cultural and social environment. Therefore, there may be differences in the impact of COVID-19 on different countries, which requires reasonably designed studies in different countries in the future, especially in low- and middle-income countries. Third, we should not stop exploring the variables affecting heterogeneity for further subgroup analysis. Some variables that may be suitable for subgroup analysis have been reported in only one study or or have been not reported in any included study, and subgroup analysis cannot be carried out. For example, the study of Iob et al.[22] stratified the number of self-harm in the sample according to the COVID-19 diagnosis, which was the only study to report the COVID-19 diagnosis. Recent evidence indicated that individuals with the diagnosis of COVID-19 would have serious traumatic experience and adverse mental health[12, 22]. If conditions allow, it is necessary to conduct psychological evaluation on individuals diagnosed with COVID-19 in order to find more specific and special connections. Finally, large sample, high quality studies need to be conducted, which are not only limited to the estimation of the prevalence of self-harm, but also focus on the risk factors and prevention/intervention strategies of self-harm, so as to broaden the research field of relevant studies to suggest ways by which the levels self-harm can be reduced.