Does harsh parenting increase the risk of self-harm and suicide in young people?

Parents’ treatment of their children, as well as family dynamics more broadly, are related to the children and young people’s (CYP) mental health. Parenting can involve many different behaviours, including positive parenting (e.g., emotional support, praise) and negative parenting (e.g., yelling, hostility). Both positive and negative parenting predict CYP’s mental health outcomes, such as depression, anxiety or substance misuse (Clayborne et al., 2021; Yap et al., 2017; find Natasha’s Mental Elf blog on Yap’s paper here). Meanwhile, family dynamics include family function and dysfunction, which are characteristics like cohesion and/or conflict.

Some previous studies have linked parenting practices to teenagers’ risk of self-harm and suicidality, but this research has never been synthesised or compared. As self-harm and suicidality are major public health concerns for CYP, and a leading cause of CYP deaths (WHO, 2021), there is a need to better understand the strength and direction of these patterns. Consequently, Hammond and colleagues (2025) aimed to summarise and synthesise the findings of prospective cohort studies of children and adolescents, where family dynamics and self-harm or suicidality were assessed with at least one year’s gap between each other.

Parenting practices and family dynamics have been linked to young people’s risk of self-harm and suicidality, but the prospective longitudinal literature has never before been summarised.

Methods

For the systematic review, five databases were searched with no language restrictions, and results were screened by a team of eight researchers, who also extracted data. To be included, papers needed to be representative cohort studies, focus on people under 20 years old, and measure family dynamics and self-harm or suicidality with a minimum 12-month gap between exposures (i.e., family dynamics) and outcomes (e.g., self-harm).

For the meta-analysis, odds ratios (ORs) and beta coefficients (β) were extracted to assess the associations between positive parenting, negative parenting, family function or dysfunction, and non-specific self-harm, non-suicidal self-harm, suicidal ideation, suicide attempt, and suicidality (combined suicidal ideation and attempt), in random-effects models.

Results

The systematic review identified 38 studies of 101,979 CYP. Twenty-four studies were included in the meta-analysis.

Most of the included studies were conducted in the USA (n = 12; 32%) or China (n = 11; 29%) and had exactly the minimum follow-up gap of 12 months between the exposure (family dynamics) and the outcome (self-harm or suicidality). Nearly all the studies (n = 36; 95%) measured the exposure and the outcome in adolescence; although, two studies measured the exposure before the age 10 years.

The researchers conducted quality assessment using the Newcastle-Ottawa scale for cohort studies, which revealed that most studies were moderate or low quality (58% 6 or below on the 1-10 scale, where 10 denotes highest quality).

Meta-analyses: Negative parenting

The meta-analyses of the associations between negative parenting and self-harm or suicidality were conducted on 16 studies and 19 associations.

Experiencing negative parenting was linked to combined self-harm and suicidal ideation (OR = 1.29, 95% CI [1.15 to 1.46]), and to non-suicidal self-harm (OR = 1.46, 95% CI [1.25 to 1.71]), when the outcomes were captured in a binary way (e.g., self-harm: yes or no).
Negative parenting was not linked to suicidal ideation alone (OR = 1.07, 95% CI [0.92 to 1.24]).

The authors also conducted meta-analyses between negative parenting and combined self-harm and suicidality captured in a continuous way (i.e., a spectrum of experiences) and found no statistically significant relationship (β = 0.07, 95% CI [–0.10 to 0.23].

Meta-analyses: Positive parenting

The meta-analyses of the associations between positive parenting and self-harm or suicidality were conducted on 10 studies and 13 associations. Enough estimates for a meta-analysis were only available for the following outcomes: combined self-harm and suicidal ideation, and suicidal ideation alone.

Experiencing positive parenting was not associated with combined self-harm and suicidal ideation (OR = 0.92, 95% CI [0.82 to 1.02]) or with suicidal ideation alone (OR = 0.99, 95% CI [0.84 to 1.17]), captured in binary ways.

The authors were unable to conduct meta-analyses between positive parenting and combined self-harm and suicidality captured in a continuous way.

Meta-analyses: Family function and dysfunction

The meta-analyses of the associations between family functioning and dysfunction, and self-harm or suicidality were conducted on 21 studies and 35 associations.

Family dysfunction was linked to combined self-harm and suicidality (OR = 1.29, 95% CI [1.13 to 1.48]), and to non-specific self-harm (OR = 1.70, 95% CI [1.10 to 2.63]), captured in binary ways.
Family dysfunction was not linked to suicide attempt (OR = 1.24 95% CI [0.93 to 1.66]), captured in binary ways.

Meta-analyses between family functioning and continuous outcomes were not possible.

Negative parenting practices, but not positive parenting practices, were significantly associated with combined self-harm and suicidal ideation in young people.

Conclusions

This systematic review and meta-analysis by Hammond et al. (2025) is the first to synthesise the relationship between family dynamics and self-harm and/or suicidality in CYP.
It found that negative parenting and family dysfunction were linked to subsequent self-harm and suicidality.
In contrast, positive parenting was not linked with subsequent decreased risk of self-harm or suicidality.
Further, the effects of negative parenting were only seen for binary measurement of self-harm and suicidality.

Negative parenting practices were linked to subsequent self-harm and suicidality, but only when they were measured in a binary way (i.e., ‘yes’ versus ‘no’).

Strengths and limitations

Strengths

One notable strength of this paper is that the review was restricted to longitudinal studies, which highlight the direction of a relationship. Although far from an experimental or even a quasi-experimental design, longitudinal research, which takes measures across more than one time point, is a better indication that links between exposure and outcome may be causal, in comparison to cross-sectional research. In the ideal circumstances, knowing an exposure causes an outcome makes the strongest cases for policy and practice to focus on targeting the exposure.
Another factor considered when thinking about cause-effect relationships is whether a relationship is free of confounders. Most of the studies included in the review and subsequent meta-analyses (n = 31; 82%) adjusted for at least one confounding factor, which is another strength of this evidence.
Furthermore, the evidence synthesised is more likely to be relevant to practitioners and policy makers, as many of the included studies are recent (published in the last 10 years).

Limitations

A notable limitation of the meta-analyses is the lack of investigation of moderators of the significant effects. For example, it would have been highly relevant to know whether study characteristics, such as geographical location or the average age of a sample, predicted whether the study found a significant effect.
Another important limitation is that longitudinal research only addresses the directionality criteria for causality, but it does not address other criteria such as thorough confounder-adjustment, unlike causal inference methods, such as propensity score matching and difference-in-difference study designs (Pearl, 2009). The strongest case for funding an intervention would come from a synthesis of studies that apply causal inference methods, although this study is a good beginning for evidence-based social policy.
Finally, the review and the included studies do not distinguish between gender and sex, despite the two representing different constructs (Gahagan et al., 2015), and most of the studies did not report on marginalisation by race or ethnicity of the participants, meaning that it is difficult for readers to evaluate whom the evidence represents and whom it does not. No studies were identified that investigated thoughts of self-harm, and in some studies, family dynamics were self-reported, which introduces social desirability bias.

A main strength of this paper is that the review and meta-analyses only included prospective longitudinal studies, which spotlights directionality of effects. However, longitudinal studies still do not mean causation.

Implications for practice

Clinical practice and social care implications

The review concludes that negative family dynamics are a modifiable exposure which increases the risk of self-harm, suicidal ideation, and suicide attempt. As such, the authors suggest that family interventions could contribute to the reduction of the rates of self-harm and suicidality in adolescence.

The meta-analyses only found associations between negative parenting and subsequent self-harm and suicidality, and not positive parenting. The clear implication of this is that interventions should focus on reducing and or preventing negative parenting behaviours. One of the family interventions may be family therapy available through local authorities (e.g., councils) or community health service providers, which explores the family dynamic and seeks to resolve underlying reasons (e.g., stress or distress, that may be contributing to negative parenting behaviours). Although, previous research showed mixed results in terms of the effectiveness of one type of family therapy (systemic) for CYP self-harm (Cottrell et al., 2018; find Udita’s Mental Elf blog on the paper here).

Additionally, awareness raising and psychoeducation around the outcomes of negative parenting or dysfunctional family dynamics could be made available to more parents. Some parents may not be aware of the potential repercussions of their parenting behaviours, such as harsh discipline, and may in fact believe they are positive. My friends or acquaintances who become parents have made me aware how little one can know about parenting until they do it. At the same time, easily accessible and digestible resources for parents, who are understandably often way too busy to read whole books about parenting, are not easy to come by, they say. Therefore, education around the negative child outcomes related to specific parenting behaviours is essential, and it should be available in digestible language, format and length. Further, social prescribing (Pescheny et al., 2019; see Stella and Dafni’s blog on the paper here) to address problematic family circumstances as well as welfare state policies may benefit families in preventing hardship, stress and self-medicating with drugs, all of which may trigger negative parenting or family conflict.

Research implications

Future research on the link between family dynamics and self-harm or suicidality could involve piloting randomised controlled trials of family interventions. At the same time, applying more causal inference methods (such as propensity score matching) to observational research would elucidate whether the links are causal and whether the interventions are likely to be successful.

Future research on the topic would also benefit from considering cultural differences in parenting. Notably, this does not need to mean a call for international research – research based on reported family culture or cultural background within multi-cultural societies, such as the UK, would be able to investigate cultural differences in these relationships.

Preventing negative family dynamics could contribute to the reduction of the rates of self-harm and suicidality in childhood and adolescence.

Statement of interests

I have worked closely with one of the authors of this study in the past. However, I was not involved in this project, nor have I ever spoken to them about this particular study.

Links

Primary paper

Hammond N.G., Semchishen S.N., Geoffroy M-C., Sikora L., Wafy G., Hsueh L., Khan H., Edwards J., Gravel C., Ferro M., Colman, I. (2025). Family dynamics and self-harm and suicidality in children and adolescents: a systematic review and meta-analysis. The Lancet Psychiatry. S2215-0366(25)00217-2

Other references

Cottrell, D. J., Wright-Hughes, A., Collinson, M., Boston, P., Eisler, I., Fortune, S., … & Farrin, A. J. (2018). Effectiveness of systemic family therapy versus treatment as usual for young people after self-harm: a pragmatic, phase 3, multicentre, randomised controlled trial. The Lancet Psychiatry, 5(3), 203-216.

Clarke, N. (2017). Parenting factors associated with adolescent alcohol misuse. The Mental Elf.

Clayborne, Z. M., Kingsbury, M., Sampasa-Kinyaga, H., Sikora, L., Lalande, K. M., & Colman, I. (2021). Parenting practices in childhood and depression, anxiety, and internalizing symptoms in adolescence: a systematic review. Social Psychiatry and Psychiatric Epidemiology, 56(4), 619-638.

Gahagan, J., Gray, K., & Whynacht, A. (2015). Sex and gender matter in health research: addressing health inequities in health research reporting. International Journal for Equity in Health, 14(1), 12.

Iyengar, U., & Ougrin, D. (2018). Family therapy for adolescent self-harm: SHIFT trial says it doesn’t reduce hospital visits and isn’t cost-effective. The Mental Elf.

Pescheny, J. V., Randhawa, G., & Pappas, Y. (2020). The impact of social prescribing services on service users: a systematic review of the evidence. European Journal of Public Health, 30(4), 664-673.

Pearl, J. (2009). Causal inference in statistics: An overview. Statistics Survey, 3, 96-146.

Tsoll, S., & Katsampa D. (2019). Social prescribing: we’re doing it more and more, but is there evidence that it works? The Mental Elf.

World Health Organization (2021). Suicide worldwide in 2019: global health estimates. World Health Organization.

Yap, M. B., Cheong, T. W., Zaravinos‐Tsakos, F., Lubman, D. I., & Jorm, A. F. (2017). Modifiable parenting factors associated with adolescent alcohol misuse: a systematic review and meta‐analysis of longitudinal studies. Addiction, 112(7), 1142-1162.