The present study was conducted to determine the patterns and predictors of a history of self-harm in patients with substance use disorders attending the outpatient clinic in a tertiary drug dependence treatment center. It found a history of deliberate self-harm in almost a third of the patients with substance use disorders and found several significant sociodemographic and clinical associations including a younger age, being unmarried/separated, unemployed, living in an urban area, a shorter duration of substance use, a greater severity of substance use disorder, history of injecting drug use and high-risk sexual behavior, of being caught by police and of being incarcerated, as well as cannabis dependence. Age, duration of substance use, and being ever caught by police were the independent predictors of self-harm in this sample.

The mean age of the patients was in the thirties. Majority of the patients were married, were employed, and were living in an urban area in nuclear families. This demographic profile is similar to those reported by other studies from India.[2829] Majority of the patients had opioid use disorder, followed by alcohol and cannabis use disorder. Most of the patients were also dependent on nicotine.

The prevalence figures for a lifetime wish to be dead (39%) and suicidal ideation (22%) are comparable to those found by other authors in the substance-using population from other cultures, as are the rates of actual self-harm (32.7%).[1030] These rates are significantly higher than those observed in the general population in India.[31] This difference from the general population is expected, as patients with substance use are known to be at a higher risk of suicide.[10] World literature reveals that patients with substance use disorders are more likely to suffer from personality disorders and other axis I psychiatric disorders and are more likely to have greater life stresses and poorer psychosocial support than the general population, all of which have been shown to significantly increase suicidal ideation.[30]

A majority of the patients harming themselves did so using means with potentially serious or even fatal outcomes. The most common means of self-harm was cutting oneself. This carries the risk of irreversible nerve damage, leading to lifetime handicap, and death in case of major vessel injury. Hitting one's head against the wall can potentially lead to a concussion/bleed, which may result in lifetime neurological deficit. A frontal lobe lesion, for example, may lead to further disinhibition and personality changes, leading to further loss of control of impulsive behavior. Thus, even if committed as nonsuicidal acts, such means of self-harm are potentially lethal, and should be considered in this light.

In our study, upon analyzing the effect of sociodemographic variables, it was found that patients with a history of self-harm were younger and had a shorter duration of substance use. This finding has been replicated in previous studies,[30] which may be explained by various factors. First of all, it has been shown that suicide in India is more common in the young population, which itself may contribute to this finding.[32] Another reason could be that patients with self-harm represent a subclass of substance users who have a more severe disease, and thus present earlier to the outpatient for management. Third, it may be possible that because those patients with suicidal ideation or attempts are more likely to die prematurely from suicide, a reduced life expectancy of this population may lead to the lower mean age observed. The association with a shorter duration of substance use may be confounded with lower age of those with self-harm. Indeed, this association is reversed when controlling for age, as seen in the results of the multivariable binary logistic regression.

We found that a significantly greater proportion of patients with a history of self-harm were unmarried/separated, as compared to those without such a history. Furthermore, living alone or in a nuclear family afforded a greater risk for harming oneself. Those who were unemployed were at a significantly greater risk of committing self-harm as well. All these findings are in line with previous world literature.[32] It is known that poor psychosocial support and unemployment are associated with a greater risk of suicide.[3334] The finding that living in an urban area increased a person's chances of having committed self-harm is different from those of other authors, who have found that living in rural regions carries a greater risk of suicide.[35] This may be because previous studies have investigated a community sample, whereas ours is a hospital-based one.

Among clinical variables, the presence of injecting drug use, high-risk sexual behavior, and cannabis dependence were found to be significantly associated with the presence of self-harm. Studies from India have previously demonstrated that patients with injecting drug use are at a significantly greater risk of suicide, with more than half of this population experiencing suicidal ideation and more than a third making an actual attempt.[20] This may be explicable by the fact that patients with injecting drug use represent a population with a more severe dependence, and either the determinants or the consequences of a more severe addiction (e.g., personality factors and poor socioeconomic status) may also contribute to this higher rates of suicidal ideation and attempts. The association of high-risk sexual behavior and greater legal complications may also be explained by personality factors, as impulsivity has been shown to increase the risk of legal complications, high-risk sexual behavior, as well as self-harm behavior.[18]

On attempting to devise a model for predicting self-harm using binary logistic regression, it is seen that a younger age, a longer duration of substance use, and the history of being caught by the police are significant predictors. This would suggest that sociodemographic factors, in addition to other factors (e.g., personality traits), are the primary determinants of self-harm behavior, rather than the substance use profile. This is also suggested by other studies in this area.[17]

The presence of cannabis dependence was significantly associated with a history of self-harm. Previous literature has shown that the increase in suicidality conferred by cannabis is usually due to another psychiatric disorder.[36] Because this is not the case here, other reasons need to be explored. Previous data have suggested that cannabis use is associated with a poorer psychosocial status, which may be the causative factor for a higher suicide rate.[37] On exploring the relationship of cannabis dependence with sociodemographic predictors, it was observed that having a status of “not married” significantly mediated (partially) the effect of cannabis use on the history of self-harm. Because studies have demonstrated that cannabis use is indeed associated with an unmarried status,[38] and that being unmarried/separated increases the risk of self-harm, it seems intuitive that cannabis use does predispose to patients remaining unmarried or being separated, which in turn leads to a greater risk of self-harm.

Given these findings, the clinical implications of this study include the identification of a more vulnerable group among patients with substance use disorders (i.e., young unmarried unemployed men) who may require greater vigilance for self-harm and possibly more intense intervention. Clinicians need to be aware of the history of self-harm among substance users and understand the reasons of the same. Moreover, they would need to deal with the acute risk of further self-harm, and deal with the consequences of a previous attempt, if that is the case. Health policy too may be guided by these results, as demographics are a viable target for intervention, and considering their role in increasing self-harm, such an intervention may go a long way in reducing the occurrence of self-harm.

The strengths of this study include a fairly large sample size, the consideration of all substance use disorders together, and the investigation of mediation effects among the clinical and sociodemographic variables. The understanding of mediation effects of factors effecting self-harm better equips the mental health professional to intervene and nullify the effects of any such factor, even if the factor itself may not be removed. There are some limitations of the study which require mention. One limitation of the study is that personality was not assessed. Because personality factors have been shown to increase the risk of self-harm, and substance use as well, it is possible that the effects which have been observed in this study may, in fact, be moderated by personality factors. Second, as this was a retrospective cross-sectional study, it was subject to recall bias. Third, being a cross-sectional study, it is difficult to differentiate causality from association, i.e., it is difficult to say whether the associations observed in this study are causal in nature. Fourth, there could be chances of response bias in divulging on this topic, despite adequate privacy being ensured to the patients. Finally, a detailed structured assessment for other axis-I mental illnesses was not performed, as it is extremely difficult to establish in retrospect whether a diagnosable disorder was present at the time of commitment of self-harm. An assessment along these lines, though highly informative, would be subject to extensive recall bias, limiting its reliability.