Post-pandemic shocks and non-suicidal self-injury risk in working-age adults: differential pathways via psychological distress and social support Authors: Xueyan Yang, Ziwei Qiang, Haoran Luo, Mandi Yuan Published: 2026 ABSTRACT AbstractBackgroundDuring the post COVID 19 transition, psychological challenges have shifted from short term reactions to enduring psychosocial strain that depletes individual coping resources and fosters maladaptive responses, with non-suicidal self-injury (NSSI) showing a sharp rise in prevalence. Despite its clinical severity and links to suicide risk, research on NSSI in the context of large scale public health emergencies remains limited, particularly among working-age adults who are especially vulnerable to pandemic related disruptions.MethodsData were collected via a nationwide online survey of 1,901 Chinese adults between January and March 2023. Partial Least Squares Structural Equation Modeling was employed to test associations among post-pandemic shocks(Objective resource loss and Subjective resource threat), psychological distress, perceived social support, and NSSI risk.ResultsObjective resource loss is associated with NSSI risk both directly and indirectly through psychological distress; Subjective resource threat influences NSSI risk primarily through a full mediation path via psychological distress; perceived social support (PSS) serves as a dual buffer, directly mitigating the impact of objective resource loss on NSSI and moderating the relationship between psychological distress and NSSI in the subjective threat pathway.ConclusionsPost-pandemic shocks were related to NSSI risk through distinct pathways. Visible losses were linked to more immediate risk, whereas perceived threats were linked to risk primarily through psychological distress. Early distress screening and strengthened social support may help reduce NSSI risk among working-age adults.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-28021-9.Keywords: Pandemic shocks, Non-suicidal self-injury (NSSI), Psychological distress, Social support Introduction In the post-pandemic transition period, mental health challenges have extended beyond acute reactions to more persistent social, economic, and psychological strain. Beyond the direct health threat posed by COVID-19, containment measures and their aftermath disrupted daily routines, social relationships, employment, and financial stability [2, 7, 28, 85]. Major disasters and public health emergencies often elicit adaptive coping responses, including proactive problem-solving and deliberate help-seeking behaviors [73, 88]. However, when stressors persist or individual resources become exhausted, adaptive responses may be displaced by maladaptive behaviors, such as increased alcohol and substance use or self-injury, which may provide short-term emotional relief but lead to substantial long-term harm [9, 22, 92]. Among these maladaptive responses, non-suicidal self-injury (NSSI) warrants distinct attention in the post-pandemic context. Defined as the deliberate destruction of body tissue without suicidal intent, NSSI may function as an immediate but harmful strategy for regulating intense negative affect or communicating distress [36, 64]. Unlike substance use, which involves consuming external agents, NSSI involves direct and intentional harm to one's own body, indicating a more severe breakdown in intrapersonal emotion regulation. Its clinical significance is further highlighted by its established link to subsequent suicidal thoughts and behaviors [34]. The rising prevalence of NSSI across various demographics during the pandemic is alarming [77, 81]. Meta-analytic evidence indicates that approximately 15.7% of adults reported NSSI during this period, substantially exceeding pre-pandemic estimates [14]. Nevertheless, disaster and pandemic mental health research has focused predominantly on internalizing symptoms, such as anxiety, depression, and post-traumatic stress disorder (PTSD), while paying limited attention to NSSI as a high-risk behavioral manifestation of maladaptation. This gap is particularly important among working-age adults, who often carry intersecting employment, financial, and family responsibilities that were substantially disrupted during the pandemic [68, 91]. In December 2022, China adjusted its COVID-19 management policies, producing a rapid transition in social conditions and a compressed period of resource disruption. This period provides a meaningful context for examining how abrupt environmental changes and pandemic-related shocks may be associated with NSSI risk. Drawing on data from a nationwide survey of Chinese working-age adults conducted during this transition, the present study investigates the associations between post-pandemic shocks and NSSI risk and examines the psychological and social mechanisms through which these associations may unfold. Theoretical background and hypotheses development Theoretical framework The present study employs the Stressor-Strain-Outcome (SSO) framework as its main organizational structure, drawing on the Conservation of Resources (COR) theory and the stress-buffering model, to investigate the impact of post-pandemic shocks on the risk of Non-Suicidal Self-Injury (NSSI) (Fig. 1). The SSO framework, initially established by Koeske [56], was primarily concerned with organizational dynamics, particularly how job demands and role ambiguity contribute to emotional exhaustion and turnover intention [11, 82]. In recent years, this framework has been widely extended to individual psychological stress research, especially in studies examining stress responses and maladaptive outcomes following large-scale risk events and disasters [48, 86]. The SSO model posits that environmental stimuli or stressors disrupt an individual’s equilibrium, leading to psychological or physiological strain, which subsequently manifests in specific behavioral outcomes. In the present post-pandemic disaster context, we conceptualize post-pandemic shocks as multidimensional stressors, psychological distress as the resulting strain, and NSSI risk as the maladaptive outcome. Research model To further delineate the nature of these stressors, we incorporate the Conservation of Resources (COR) theory [40]. COR theory proposes that psychological stress occurs when individuals experience actual resource loss, perceive a threat of resource loss, or fail to gain expected resources after significant investment. Public health crises are particularly disruptive as they simultaneously threaten multiple forms of valued resources. While some individuals experienced direct and observable losses, many others who had not yet faced direct impact still perceived a high risk of future depletion of key resources. Therefore, we categorize the impact of the pandemic into two distinct yet interconnected components: Objective Resource Loss (Objective Shocks): This dimension encompasses realized damage, defined as exposure to identifiable and recordable adverse events indicating disruptions in health, economic, or social conditions. This includes bereavement [61], housing instability [13, 23], and significant setbacks in employment or income [45, 72]. It represents the actual severity of the impact—the tangible losses already incurred. Objective Resource Loss (Objective Shocks): This dimension encompasses realized damage, defined as exposure to identifiable and recordable adverse events indicating disruptions in health, economic, or social conditions. This includes bereavement [61], housing instability [13, 23], and significant setbacks in employment or income [45, 72]. It represents the actual severity of the impact—the tangible losses already incurred. Subjective Resource Threat (Subjective Shocks): This dimension captures anticipated or perceived damage, pertaining to an individual’s subjective experience of vulnerability and uncertainty regarding key resources. In disaster psychology, perceived threat is a critical chronic stressor, especially when the hazard is invisible, contagious, or difficult to control [43, 60]. Consequently, the present study conceptualizes subjective resource threat through three domains: perceived infection threat [4, 44], perceived financial insecurity [52, 54, 93], and perceived social isolation [8, 67]. These dimensions respectively capture the profound uncertainty surrounding health and safety, economic stability, and relational belonging. Subjective Resource Threat (Subjective Shocks): This dimension captures anticipated or perceived damage, pertaining to an individual’s subjective experience of vulnerability and uncertainty regarding key resources. In disaster psychology, perceived threat is a critical chronic stressor, especially when the hazard is invisible, contagious, or difficult to control [43, 60]. Consequently, the present study conceptualizes subjective resource threat through three domains: perceived infection threat [4, 44], perceived financial insecurity [52, 54, 93], and perceived social isolation [8, 67]. These dimensions respectively capture the profound uncertainty surrounding health and safety, economic stability, and relational belonging. Post-pandemic shocks and NSSI risk Within the Stressor-Strain-Outcome (SSO) framework, acute public health crises function as powerful environmental stimuli that disrupt an individual’s internal equilibrium. In the context of the COVID-19 transition, pandemic shocks represent a concentrated burst of stressors that threaten an individual’s resource reservoir. Drawing on Conservation of Resources (COR) theory, the sudden realization of resource loss and the acute perception of future threat create an immediate state of psychological vulnerability [39]. The “shocks” associated with the post-pandemic transition are distinguished from chronic stressors by their intensity and perceived lack of control. According to the integrated theoretical model of NSSI [64], such acute and uncontrollable events serve as significant precipitating factors for self-injurious behavior. When individuals encounter rapid environmental changes, such as the sudden removal of health protections or abrupt economic disruptions, the ensuing psychological strain can overwhelm existing adaptive coping mechanisms. In the context of resource depletion, NSSI may be employed as a functional, albeit maladaptive, strategy to manage a surge of negative affect or to regain a sense of sensory control amid overwhelming uncertainty [36]. Research conducted during the pandemic has consistently associated infection-related anxiety, social isolation, and financial insecurity with heightened emotional distress and suicidal ideation [30, 41]. For individuals of working age, these disruptions are particularly consequential as they directly impact essential resources, notably employment and social belonging, which are necessary for daily functioning [68]. Given that acute stress serves as a primary catalyst for self-injury, the severity of post-pandemic disruptions is anticipated to be positively associated with NSSI risk. Hypothesis 1 Post-pandemic shocks (including objective resource loss and subjective resource threat) are positively associated with NSSI risk. The mediating role of psychological distress Psychological distress plays a crucial role in NSSI risk [63, 79]. Although NSSI is not always accompanied by psychiatric symptoms, individuals who engage in NSSI tend to exhibit higher levels of emotional regulation difficulties. Prospective research has demonstrated that mental disorders can predict the onset of NSSI [21, 51]. In a survey conducted as part of the World Mental Health International College Student (WMH-ICS) initiative across 24 colleges in nine countries (with a sample of 20,842 first-year students), it was found that 59.6% of individuals with a history of NSSI met the criteria for at least one mental disorder [50]. Psychological distress, characterized by a combination of negative emotions such as anxiety and depression, often serves as a precursor to self-injurious behavior [78]. This distress functions as a critical intermediary, potentially leading to NSSI when individuals feel overwhelmed and lack effective coping mechanisms. Emotion Regulation Model posits that NSSI provides temporary emotional relief, serving as a maladaptive strategy for managing emotional fluctuations and stress [26]. Psychological distress is not merely a trigger but also a bridge in the chain linking post-pandemic-induced stressors to NSSI [21, 42, 87]. This distress indicates a compromised emotional regulation system, increasing the likelihood of NSSI as an unhealthy coping response. Based on this, the following hypothesis is proposed: Hypothesis 2 Psychological distress mediates the relationship between post-pandemic shocks and NSSI risk. The moderating role of social support Across disaster mental health research, social support is recognized as a crucial social resource [27, 49, 84]. Low social support has been repeatedly identified as an important risk factor for the development of post-traumatic stress [59, 76]. In the contexts of disasters and post-pandemic scenarios, researchers are increasingly focusing not only on whether individuals have received social support but also on whether they perceive this support as accessible. While the effects of merely receiving support have produced inconsistent findings, perceived social support, which is defined as an individual's subjective belief that one is cared for and part of a cohesive network, has emerged as a more robust and stable predictor of successful psychological adjustment [47]. According to the stress-buffering model, perceived social support effectively mitigates the negative impact of stressors by directly promoting positive cognitive reappraisal, enabling effective emotional regulation, and maintaining a crucial sense of social belonging [12], thereby reducing the risks of psychological distress, depression, and anxiety [62, 71]. Perceived social support serves not only as a protective factor for mental health following a disaster but also influences whether individuals engage in self-injurious coping strategies under conditions of high stress [35, 89]. Individuals who engage in repetitive NSSI frequently report lower levels of perceived family support and have fewer individuals to whom they can turn for assistance [46]. However, there remains a lack of direct evidence demonstrating that perceived social support functions as a protective factor against the post-pandemic effects leading to an increased risk of NSSI, particularly within the general adult population. Thus, the following hypothesis was suggested: Hypothesis 3 Perceived social support buffers the positive association between post-pandemic shocks and NSSI risk. Methods Sampling and data collection Data were collected through a nationwide online survey conducted between January and March 2023 to assess the psychological distress and social support experiences of working-age adults during the initial surge of infections following the lifting of COVID-19 containment measures in China. The survey was administered via Wenjuanxing, a widely utilized online survey platform in China, and was organized and supervised by the Institute for Population and Development Studies at Xi’an Jiaotong University. In light of the infection risk, mobility barriers, and social avoidance behaviors prevalent during the peak infection period, the present study employed a nationwide online snowball sampling strategy. Drawing on special-period survey methodologies, the questionnaire link and a scannable QR code were initially disseminated through WeChat communities and WeChat “Moments.” [66]. Initial respondents were encouraged to forward the survey to individuals within their social networks who met the eligibility criteria. This chain-referral process was employed to reach working-age adults with specific eligibility characteristics during the early post-containment period. The online mode was deemed appropriate given the sensitive and private nature of the mental health topics addressed in the survey. A completely anonymous and depersonalized survey environment was anticipated to mitigate social desirability bias and encourage participants to provide more candid responses [6, 19]. Eligible participants were required to be working-age adults aged 18–60 years, currently residing in mainland China, and fully aware of their COVID-19 infection status. All participants provided informed consent through the Wenjuanxing platform prior to completing the questionnaire and received a cash reward of 5 RMB as a token of appreciation for their time and contribution. To improve data quality, a multi-stage quality control protocol was implemented. First, technical restrictions limited submissions to one response per unique IP address and device identifier. Second, internal logic checks, reverse-coded items, and specific lie-detection questions were embedded in the questionnaire to identify inconsistent or inattentive responses. Third, based on pilot testing, questionnaires completed in less than 180 s were considered rushed responses and were excluded. Among the 2,019 initially collected questionnaires, 118 were excluded because they failed to meet the pre-specified quality control criteria. Finally, 1,901 valid questionnaires were retained for analysis, yielding a valid questionnaire rate of 94.16%. The final sample skewed slightly female (55.55%), with most respondents in the 26–35 age group (40.35%), and 62.65% were married. A large proportion of respondents had recovered from COVID-19 (82.22%), providing information on post-infection psychological experiences during the early post-containment period. The survey adhered to ethical standards and received approval from the Shaanxi Normal University Ethics Committee. Informed consent was obtained from all participants, and all data were anonymized. Measurement instrument Guided by Conservation of Resources theory, post-pandemic stress exposure was operationalized through two related but distinct constructs: objective resource loss and subjective resource threat. Objective resource loss captured actual, externally anchored resource-loss events. Respondents indicated whether they had experienced any of the following events after the COVID-19 restrictions were lifted: death of a family member or friend due to COVID-19, loss of accommodation because of inability to pay rent, inability to pay bills or buy food, and unemployment, pay reduction, or inability to work due to childcare responsibilities. The “none” option was coded as 0 and was not counted as an event. The four adverse events were summed to create a cumulative shock-count index ranging from 0 to 4, with higher scores indicating greater exposure to objective resource loss [68]. Subjective resource threat is defined as the subjective assessment by participants that valued resources are at risk in the post-pandemic context. It is conceptualized as a formative construct consisting of three distinct indicators: perceived social isolation, perceived financial difficulties, and perceived infection risk. These indicators represent separate domains of social, economic, and health/safety resource threats and were measured using a 1 to 5 scale [49]. For assessing non-suicidal self-injury (NSSI) risk, our dependent variable, two dimensions were considered: ideation and behavior. Participants reported their self-harm thoughts and behaviors post-restrictions on scales from 1 to 4, with the instrument's reliability indicated by a Cronbach's alpha of 0.709 [15]. Psychological distress was operationalized as a latent construct reflected by anxiety and depressive symptoms, consistent with prior studies that have conceptualized psychological distress using combined anxiety–depression indicators [16, 57, 58]. Anxiety symptoms were assessed using the Generalized Anxiety Disorder-7 scale (GAD-7), and depressive symptoms were assessed using the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10), with Cronbach’s alphas of 0.935 and 0.877, respectively. The GAD-7 captured anxiety symptoms during the two weeks following the lifting of COVID-19 restrictions, whereas the CES-D-10 assessed depressive symptoms during the first week after restrictions were eased. Higher scores indicated greater symptom severity and, jointly, higher levels of psychological distress. Perceived social support was measured using the Chinese adaptation of the Multidimensional Scale of Perceived Social Support (MSPSS). This instrument assesses support from family, friends, and other sources in the context of COVID-19, and it is rated on a seven-point Likert scale. Higher scores on this scale, which demonstrated a Cronbach’s alpha of 0.797, indicate greater perceived social support [17, 53]. Common method variance Relying solely on survey questionnaires raised potential common method variance (CMV) concerns. To address this in Partial Least Squares Structural Equation Modeling (PLS-SEM), two methods were employed: a full collinearity assessment using Variance Inflation Factors (VIFs) [55], and the correlation matrix procedure. Regression of variables against a common factor showed VIFs between 1.000 and 2.704, below the 3.3 bias threshold, and correlation coefficients among constructs remained under 0.9. Thus, CMV was not deemed a significant issue in the data. Data analysis The present study employed Partial Least Squares Structural Equation Modeling (PLS-SEM) using SmartPLS 4.0.9 to evaluate the hypothesized relationships. PLS-SEM was selected because the study was theory-guided but explanatory–predictive in orientation, aiming to examine the associations among post-pandemic shocks, psychological distress, perceived social support, and NSSI risk while also assessing the model’s explanatory and predictive performance [31, 32, 74]. This approach is also suitable for models involving multiple pathways, mediation, and moderation, as well as for maximizing the explained variance of key endogenous constructs. In addition, the model included both reflective psychological constructs and composite exposure variables. Psychological distress, perceived social support, and NSSI risk were specified as reflective constructs, whereas subjective resource threat and objective resource loss were modeled as composite constructs. Specifically, objective resource loss was specified as a single-indicator composite construct, operationalized as an observed cumulative event-count index. When a construct is defined as the accumulation of external objective conditions, such as event counts or equipment counts, single-indicator measurement can be theoretically appropriate [65]. In PLS-SEM, the construct score of a single-indicator construct is equivalent to the standardized value of the observed indicator, and its reliability is typically assumed to be 1.0 by default [75]. Subjective resource threat was defined as a composite construct capturing the most theoretically salient dimensions of subjective threat: health and safety, economic stability, and relational belonging. Consistent with the pragmatic perspective on construct operationalization in PLS-SEM, such a composite indicator was used not to represent a latent psychological trait in the reflective sense, but to provide the best available approximation of the theoretical concept within the specific empirical context of this study [29, 37]. This mixed measurement specification further supports the use of PLS-SEM, which is particularly suitable for models that combine reflective constructs with composite constructs and observed exposure indices [31, 75]. Model assessment followed established PLS-SEM procedures. All variables were standardized prior to analysis. Reflective constructs were evaluated using standardized loadings, composite reliability, average variance extracted (AVE), and heterotrait–monotrait ratios (HTMT). The formative construct was assessed using indicator collinearity diagnostics, bootstrapped outer weights, and outer loadings. The structural model was evaluated using inner-model collinearity diagnostics, bootstrapped path coefficients, R2, and f2. SRMR was reported as supplementary evidence of no major model misspecification. Predictive relevance was assessed using PLSpredict, including Q²predict and comparisons with both indicator-average and linear regression benchmarks where appropriate [3, 94]. Results Evaluating the measurement model Table 1 reports the measurement results for both reflective and composite constructs. For the reflective constructs (Psychological distress, Perceived social support, and NSSI risk), internal consistency and convergent validity were confirmed, as all outer loadings exceeded 0.70, composite reliability (CR) values surpassed the 0.70 benchmark [20], and average variance extracted (AVE) values were all above the 0.50 threshold, with the lowest observed AVE being 0.686 [5]. Table 1Measurement modelConstructItemsOuter LoadingsP valuesCRAVEOuter weightsP valuesVIFSubjective resource threatPerceived social isolation0.850 < 0.0010.8530.6600.450 < 0.0011.587Perceived financial difficulties0.796 < 0.0010.376 < 0.0011.494Perceived risk of infection0.789 < 0.0010.403 < 0.0011.398Objective resource lossShocks count1.000 < 0.0011.000 < 0.0011.000Psychological distressAnxiety (GAD-7)0.945 < 0.0010.9460.8970.591 < 0.0012.704Depression (CES-D-10)0.949 < 0.0010.537 < 0.0012.704Perceived social supportFamily (MSPSS-12)0.895 < 0.0010.8680.6860.609 < 0.0011.444Friends (MSPSS-12)0.820 < 0.0010.300 < 0.0012.209Others (MSPSS-12)0.762 < 0.0010.273 < 0.0011.981NSSI riskIdeation0.894 < 0.0010.8720.7740.599 < 0.0011.432Behavior0.865 < 0.0010.536 < 0.0011.432Outer Loadings: Contribution of each indicator to its construct; Outer Weights: Weight assigned to each indicator in the model; VIF Variance Inflation Factor (Measure of multicollinearity among indicators); P Values: Statistical significance level of outer loads or weights (< 0.05). CR composite reliability, AVE average variance extracted, VIF variance inflation factor. CR and AVE are reported only for reflective constructs. Objective resource loss is an observed single-item event-count index; therefore, internal consistency statistics are not applicable [18] Measurement model Outer Loadings: Contribution of each indicator to its construct; Outer Weights: Weight assigned to each indicator in the model; VIF Variance Inflation Factor (Measure of multicollinearity among indicators); P Values: Statistical significance level of outer loads or weights (< 0.05). CR composite reliability, AVE average variance extracted, VIF variance inflation factor. CR and AVE are reported only for reflective constructs. Objective resource loss is an observed single-item event-count index; therefore, internal consistency statistics are not applicable [18] For the composite construct (Subjective resource threat), the measurement model was evaluated based on indicator redundancy and contribution [31, 65, 95]. As shown in Table 1, all indicator collinearity values were within acceptable limits, with variance inflation factors (VIF) well below the 3.3 threshold (the highest being 1.587). Furthermore, the statistical significance of the indicators was confirmed, as all outer weights were significant (p < 0.001). Finally, Objective resource loss was treated as a single-item observed event-count index. Consequently, internal consistency and AVE statistics were not applicable for this construct. Subsequently, to empirically assess the distinctiveness of all constructs, the zero-order correlations among the key constructs were first examined. As shown in Table 2, the correlation coefficients ranged from −0.237 to 0.600, all of which remained well below the common threshold of 0.85, providing preliminary evidence for discriminant validity. To further ensure distinctiveness, discriminant validity was rigorously evaluated using the well-regarded HTMT criterion [74]. Table 2Construct correlationsVariableSubjective resource threatObjective resource lossPsychological distressPerceived social supportNSSI riskSubjective resource threat1Objective resource loss0.411***1Psychological distress0.600***0.362***1Perceived social support−0.175***−0.113***−0.237***1NSSI risk0.282***0.328***0.382***−0.189***1***p < 0.001. All values are Pearson correlation coefficients Construct correlations ***p < 0.001. All values are Pearson correlation coefficients According to Henseler et al. [38], HTMT values should ideally be below 0.85 or 0.90 for acceptable discriminant validity. With the more conservative threshold of 0.85 applied, the highest observed HTMT value was 0.741, confirming that the discriminant validity of the data is within acceptable limits (see Table 3). Table 3Discriminant validity (HTMT Values)VariableSubjective resource threatObjective resource lossPsychological distressPerceived social supportNSSI riskSubjective resource threat—Objective resource loss0.475—Psychological distress0.7410.385—Perceived social support0.2300.1290.282—NSSI risk0.3860.3910.4800.256—HTMT heterotrait–monotrait ratio. All HTMT values were below the conservative threshold of 0.85, supporting discriminant validity Discriminant validity (HTMT Values) HTMT heterotrait–monotrait ratio. All HTMT values were below the conservative threshold of 0.85, supporting discriminant validity Evaluating the structural model The structural model was evaluated using SmartPLS with bootstrapping (5,000 resamples). Five constructs were included in the model: subjective resource threat, objective resource loss, psychological distress, perceived social support, and NSSI risk. The structural model results are depicted in Fig. 2. Structural model results. Notes: ***p < 0.001; ns =Not significant Following Hair et al. [32], the structural model was assessed using standardized path coefficients (β), t-values, p-values, confidence intervals, R2 values, and f2 effect sizes. All hypotheses except H1a and H3a were accepted. For the direct effects, H1a and H1b proposed the impact of subjective resource threat and objective resource loss on NSSI risk, respectively. The results of the structural model (see Table 4) show that with the exception of the H1a path, all main-effect paths were positive and significant. The path coefficient for H1a was not significant (β = 0.037, p > 0.05), suggesting insufficient evidence to support this hypothesis. This implies that the direct effect of subjective resource threat on NSSI risk is limited, with its primary impact being exerted indirectly through the mediating role of psychological distress (H2). Table 4Results of hypothesis testingHypotheses and pathsStandardized path coefficient (β)t-valueConfidence intervalResultsMain effects H1a: SRT → NSSI0.037 ns1.146[-0.026, 0.099]Unsupported H1b: ORL → NSSI0.177***6.147[0.120, 0.234]Supported H2: PD → NSSI0.286***10.782[0.235, 0.337]SupportedModerating effects H3a: PSS x SRT → NSSI0.047 ns1.504[-0.014, 0.108]Unsupported H3b: PSS x ORL → NSSI-0.131***3.616[-0.165, -0.047]Supported H3c: PSS x PD → NSSI-0.108***4.961[-0.185, -0.081]SupportedSRT subjective resource threat, ORL objective resource loss, PD psychological distress, NSSI Non-suicidal self-injury, PSS perceived social support, ns not significant; ***p < 0.001 Results of hypothesis testing SRT subjective resource threat, ORL objective resource loss, PD psychological distress, NSSI Non-suicidal self-injury, PSS perceived social support, ns not significant; ***p < 0.001 When examining the mediating role of psychological distress in the associations between post-pandemic shocks and NSSI risk, the results indicate two distinct patterns. Subjective resource threat was strongly associated with psychological distress (β = 0.546, p < 0.001), and psychological distress was positively associated with NSSI risk (β = 0.286, p < 0.001). However, the direct path from resource threat to NSSI risk was not significant (β = 0.037, 95% CI [−0.026, 0.099]). This pattern suggests that the association between subjective resource threat and NSSI risk is mainly accounted for by psychological distress, consistent with a full mediation pattern. In contrast, objective resource loss was positively associated with both psychological distress (β = 0.141, p < 0.001) and NSSI risk (β = 0.177, 95% CI [0.120, 0.234], p < 0.001). Given that both the direct and indirect paths are significant, this indicates a partial mediation pattern, in which psychological distress complements but does not fully account for the effect of objective resource loss on NSSI risk. With regard to the moderating effect of social support, evidence was found supporting both H3b and H3c. Specifically, H3b indicates that perceived social support significantly moderates the relationship between objective resource loss and NSSI risk, reducing the direct impact of these shocks. H3c indicates that perceived social support significantly moderates the relationship between psychological distress and NSSI risk, thereby buffering the impact of distress on NSSI risk. The simple slope analysis further illustrates these moderation effects. As shown in Fig. 3, the positive relationship between objective resource loss and NSSI risk becomes weaker as perceived social support increases. Individuals with lower levels of perceived social support exhibit a steeper increase in NSSI risk following resource loss, whereas this increase is substantially attenuated among those with higher levels of perceived social support. As shown in Fig. 4, a similar buffering pattern is observed for psychological distress. The association between psychological distress and NSSI risk is stronger at lower levels of perceived social support and weaker at higher levels of perceived social support. Simple slopes of objective resource loss predicting NSSI risk at low (−1 SD), mean, and high (+ 1 SD) levels of perceived social support Simple slopes of psychological distress predicting NSSI risk at low (−1 SD), mean, and high (+ 1 SD) levels of perceived social support The R2 values, indicating the proportion of variance in the dependent variables explained by the independent variables, were 0.380 for Psychological distress and 0.233 for NSSI risk, suggesting a moderate explanatory power of the model. Further analysis using the f2 effect sizesf to evaluate the influence of each variable revealed that objective resource loss had a small effect on NSSI risk (f = 0.032), whereas subjective resource threat had a significant effect on psychological distress (f = 0.401). The approximate fit of the structural model was assessed using the standardized root mean square residual (SRMR). The resulting SRMR value was 0.059, which is below the recommended threshold of 0.08, indicating an adequate model fit [65, 75]. PLSpredict results supported the model’s predictive relevance and comparative predictive performance, and ancillary dimension-specific single-mediator analyses were also provided for transparency (see Supplementary file-Additional Analysis Tables S1–S4). Discussion Utilizing the Stressor-Strain-Outcome (SSO) framework, the present study investigated associations between two qualitatively distinct post-pandemic shocks (objective resource loss and subjective resource threat) and the risk of non-suicidal self-injury (NSSI) among working-age adults in China. Consistent with the literature documenting links between pandemic-related stressors and heightened psychosocial risk outcomes [69, 92], our findings indicate that post-pandemic shocks are associated with increased NSSI risk. However, our findings further refine this relationship by showing that post-pandemic shocks are not homogeneous stressors. Objective resource loss retained a direct association with NSSI risk, whereas subjective resource threat was associated with NSSI risk primarily through psychological distress. Further, perceived social support demonstrated a pathway-specific buffering effect, attenuating the direct associations of objective resource loss and psychological distress with NSSI risk. Differential pathways from post-pandemic shocks to NSSI risk A key contribution of the current research is its disaggregation of pandemic-related shocks, revealing their differential pathways to NSSI risk. While prior literature often operationalized pandemic stress as a global exposure, the present findings underscore the importance of distinguishing between stressor types. Specifically, objective resource loss was directly and positively associated with NSSI risk. This effect persisted even after accounting for the mediating role of psychological distress. The existence of such a direct pathway suggests that acute, significant losses may impose an immediate and substantial burden on an individual's regulatory capacity, potentially precipitating self-injurious behavior as a coping mechanism, particularly under conditions of overwhelming external pressure. In the context of a global health crisis, these externally anchored stressors may not require a prolonged process of psychological accumulation to manifest as behavioral risk; their very severity and abruptness appear sufficient to directly elevate this risk. In contrast, subjective resource threat exhibited a distinctly different, indirect-only pattern. The association between this form of shock and NSSI risk operated primarily through psychological distress and showed no direct association. Specifically, a higher subjective threat strongly predicted greater psychological distress, which, in turn, was associated with an elevated risk of NSSI. The indirect-only effect can be comprehended within the framework of emotion regulation theory. From this perspective, stressors that are internal and appraisal-based, such as perceived uncertainty and anticipated loss, operate by gradually depleting an individual’s regulatory resources [1, 70]. As these resources gradually deplete over time, the ability to manage emotional turmoil diminishes, thereby increasing susceptibility to maladaptive coping mechanisms, such as NSSI, when distress becomes overwhelming. Our findings are consistent with this process, illustrating that subjective threat may operate more insidiously by first fueling internal distress before manifesting as behavioral risk. Notably, the analytic sample comprised working-age adults in China, where collectivist social norms may shape emotional restraint and self-reliance [10, 83]. These norms may shape how individuals internalise threats to subjective resources and whether they seek or mobilise social support when distressed. In this cultural context, internalised concerns may be downplayed or expressed less openly, potentially intensifying psychological distress and strengthening its association with maladaptive coping. Moderating effects of perceived social support The moderation results clarify the conditional protective role of perceived social support. Perceived social support weakened the association between objective resource loss and NSSI risk, supporting H3b. It also weakened the association between psychological distress and NSSI risk, supporting H3c. The findings show that perceived social support is associated with the pathways linking post-pandemic shocks to NSSI risk in different ways, rather than being universally protective. In the progression from objective resource loss to the risk of NSSI, perceived social support may operate through a compensatory resource mechanism. Consistent with COR theory, objective resource loss represents the actual depletion of valued resources, which can directly impair an individual’s ability to cope adaptively and increase their vulnerability to maladaptive behavioral responses [80]. Perceived social support may attenuate this pathway, as it serves as a vital reservoir of external resources [24]. Support from family, friends, and significant others may provide emotional reassurance, practical assistance, informational guidance, and a sense that additional coping resources are available [33]. In this context, perceived social support may partially compensate for resource depletion and may help prevent tangible losses from manifesting as self-directed harmful behavior. In the context of the pathway from psychological distress to NSSI risk, perceived social support may function through an emotion-regulation resource mechanism. Psychological distress can elevate the risk of NSSI when individuals lack adaptive strategies to manage intense negative emotions or to communicate their distress. A higher level of perceived social support may mitigate the likelihood that psychological distress leads to NSSI by providing alternative avenues for emotional regulation. These avenues include verbal expression, emotional validation, cognitive reappraisal, help-seeking, and interpersonal connection [25, 90]. Consequently, perceived social support may attenuate the distress–NSSI relationship not by substituting lost material resources, but by broadening the individual’s repertoire of adaptive emotion-regulation strategies. Limitations and future research directions First, the online snowball sampling strategy was employed to reach working-age adults within overlapping social networks and to reduce social desirability bias through an anonymous online survey. However, this approach may have introduced coverage bias and self-selection bias, particularly because individuals without smartphones, reliable Internet access, regular WeChat use, or sufficient digital literacy may have been underrepresented. Future research could employ probability-based sampling, stratified online panels, or mixed-mode recruitment strategies combining online and in-person data collection to enhance representativeness. Second, the cross-sectional design limits the ability to establish causal relationships among objective resource loss, subjective resource threat, psychological distress, and NSSI risk. Future longitudinal or repeated-measures studies are needed to clarify temporal ordering and examine how these factors interact over time. Third, the operationalization of post-pandemic shocks as non-reflective exposure measures has limitations. These measures should be understood as pragmatic proxies intended to approximate the theoretical concept of post-pandemic disruption rather than exhaustive measures of the construct [29, 75]. Although the indicators were theoretically guided by the domains of health and safety, economic stability, and relational belonging, they may not capture all relevant post-pandemic shock experiences. Future research should aim to develop a more comprehensive index of post-pandemic shocks to test the robustness and generalizability of the relationships identified in this study. Conclusion Based on the SSO framework, COR theory, and stress-buffering model, the present study shows that post-pandemic shocks were associated with NSSI risk among working-age adults in China through differentiated pathways. Objective resource loss was associated with NSSI both directly and indirectly via psychological distress, whereas subjective resource threat was primarily associated with NSSI risk through psychological distress. Perceived social support selectively buffered the associations of objective resource loss and psychological distress with NSSI risk, illustrating pathway-specific protective effects. While prior NSSI research has focused on adolescents, our findings indicate that in the context of disasters and major public health crises, adults experiencing acute stressors are also at risk of maladaptive behaviors. These results highlight the importance of considering adult populations and stressor specificity when designing mental health interventions and support strategies during crises. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements The authors extend their sincere appreciation to every participant in this study. Our heartfelt gratitude to all our research collaborators for their invaluable contributions. Abbreviations Non-Suicidal Self-Injury Perceived Social Support Authors’ contributions X.Y. conceptualized the study, acquired funding, provided resources, supervised the project, and reviewed and edited the manuscript. Z.Q. handled the methodology, software, investigation, data curation, and drafted the original manuscript. H.L. and M.Y. contributed to the investigation and data curation. All authors reviewed the manuscript. Funding No financial support was received from any person or institution for the research. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The present study was approved by the Ethics Committee of Shaanxi Normal University (Approval No. HR203-01-09). Written informed consent was obtained from all participants prior to their inclusion in the study. Participants were assured of the confidentiality and anonymity of their data, and they were informed that participation was voluntary. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement No datasets were generated or analysed during the current study.