Intimate Partner Violence and Incident Depressive Symptoms and Suicide AttemptsA Systematic Review of Longitudinal Studies

Let's start with the idea, not the numbers. Violence in a relationship doesn't just bruise skin; it reshapes a mind. People who live with intimate partner violence—either physical or sexual harm from someone they're with—often describe fear that doesn't switch off, isolation that grows, and stress that never lets the body reset. Those are the ingredients of depression. In the most tragic cases, they're the backdrop to suicide attempts. That's the hunch clinicians have carried for years. Devries and colleagues set out to test it with time on their side, following people forward to see what happens first, what follows, and how often. Why lean so hard on time? Because three stories can fit the same snapshot. One: violence drives depression and suicide attempts. Two: being depressed, or already thinking about suicide, raises the risk that you'll enter or stay in a violent relationship. Three: some upstream forces—such as childhood adversity, alcohol use, and poverty—push both violence and depression together. To get beyond that tangle, Devries's team cast a very wide net. They searched twenty databases—from Medline and Embase to CINAHL—from the first records ever entered up to early 2009, and then they pushed an update for depression and suicide papers through February first, 2013. They only brought in longitudinal studies, which are those that measured intimate partner violence or the mental health outcomes at more than one time point and in either direction. Studies focused only on suicidal thoughts or plans, and those on antenatal or postpartum depression, were set aside. The measures varied—different scales and yes or no items—but they had to be valid and used properly. Who ended up in this big picture? Sixteen longitudinal studies, reported across seventeen papers, following thirty-six thousand one hundred sixty-three people. Most of the participants were women, and most of the studies were done in high-income countries. Ten were U.S. datasets; others came from Australia, Sweden, India, Nicaragua, and South Africa. Typical follow-up was about three years—the median was thirty-six months—and attrition hovered around one in five. Violence was usually recorded with act-based tools like the Conflict Tactics Scale or World Health Organization instruments. Depression was more often measured with symptoms on a scale—the Center for Epidemiologic Studies Depression Scale, or CES-D, or the Beck Depression Inventory—than through a clinical diagnosis. Now to the heart of it. Among women, when you step into a violent relationship—or stay in one—you're nearly doubling your odds of moving into depression later. Pooling six studies, the odds ratio was 1.97. Think of it this way: compared to women not reporting violence, those who did were almost twice as likely to develop new depressive symptoms over follow-up. Every single contributing study pointed in that direction. The size of the effect varied from study to study—that's what an I squared around 50 percent tells you—but the direction was consistent. It tracks with what survivors say: chronic fear and control drain the system. Sleep fractures. Social ties thin. Over months to years, mood sinks. The twist is that the arrow runs the other way too. Women with depressive symptoms were more likely to encounter violence down the line. Four studies could be pooled here, and the odds ratio sat at 1.93, with essentially no heterogeneity. That's a tight signal. It doesn't mean depression causes someone else's violence, but it does suggest a feed-forward loop: symptoms like hopelessness, impaired concentration, and low self-worth can make it harder to appraise risk, to leave, or to find support—conditions under which an abusive partner's control can take root or intensify. Temporality matters. When symptoms show up first, the risk of later violence rises. What about the gravest endpoint—suicide attempts? Here the evidence is smaller, but what exists points in one direction. In three studies of women, those exposed to intimate partner violence had higher odds of attempting suicide later. In a U.S. adolescent sample, Ackard and colleagues found an odds ratio around 3.2—uncertain because the confidence interval was extremely wide, but suggestive. In India, Chowdhary and Patel reported a striking odds ratio of 7.97; again, based on relatively few events, but the association was clear. Roberts and colleagues, using a large U.S. cohort, reported a small but statistically reliable positive link between violence and later attempts. No one had published a focused meta-analysis for suicide at that point, and the methods across those papers didn't line up cleanly enough to pool. Still, three independent datasets showed three pushes in the same direction. You may be wondering how much of this rests on symptom scales rather than diagnoses. Quite a lot. Nearly all the depression outcomes were based on symptoms, not major depressive disorder. That's a limitation if you want to map directly to diagnostic criteria. It's also a strength if you care about day-to-day suffering and function, because symptom scales catch change earlier and more sensitively. They are what most of these cohorts had in hand. Let's talk about men, because the pattern shifts. Four longitudinal studies that included male participants give us a window, and the view is hazier. In a U.S. adolescent and young adult cohort, Ackard and colleagues reported that males who experienced physical or sexual partner violence had nearly doubled odds of later depressive symptoms—an odds ratio of 1.92 after adjusting for baseline symptoms. Roberts's analysis of Add Health data also showed a positive association, with a beta coefficient around 0.18—modest, but clearly above zero. Foshee and colleagues, tracking whether depressive symptoms led to later victimization, reported a hazard ratio around 1.10 that was not statistically significant. A Swedish study by Jonsson and colleagues saw a small absolute difference—about two and a half percent of men with adolescent depressive symptoms later reported partner violence, versus none without those symptoms—but the numbers were tiny. In other words, there's a hint of a link from violence to later symptoms in men, but far less to say about symptoms forecasting violence. On suicide attempts among men, the studies don't cohere into a clear signal. One U.S. analysis reported a very imprecise odds ratio above seven for attempts after violence, but the confidence interval spanned from below one to sky-high, meaning the estimate was too unstable to rely on. Another study found no meaningful association. The take-home point is simple: we don't have reliable evidence yet that men exposed to partner violence show an elevated risk of later suicide attempts, at least not in these longitudinal datasets. All of this begs a question: how much confidence should we place in the size of these estimates? Devries and colleagues were careful about that. They used random-effects meta-analysis, which assumes the true effect can vary across settings, and they quantified heterogeneity with that I squared statistic I mentioned earlier. From each data source, they picked a single estimate for any given analysis—the one they judged least biased based on prespecified quality criteria—to avoid double counting. However, the underlying studies often adjusted only for demographics and for baseline levels of the outcome. Few accounted for childhood adversity, substance use, or other early-life risks that plausibly influence both violence and mental health. Emotional abuse was largely out of frame. Many violence exposures were coded as a simple yes or no, which can wash out severity and timing. In the original search, abstracts were screened by a single reviewer; the team didn't contact authors for extra data, and publication bias couldn't be probed meaningfully because the eligible pool was small. All of that pulls the estimates toward caution. Context also matters. Most samples were from high-income countries, and a fair number were adolescents or young adults in dating or non-cohabiting relationships. That shapes both exposure and impact. Some studies only followed women who stayed in partnerships across waves, a design quirk that can actually push findings toward the null by excluding those who left the most dangerous relationships. Attrition—people dropping out over time—sat around 22 percent on average. While not catastrophic, it's enough to introduce bias if those who left were systematically different. Despite the caveats, the pattern in women is hard to ignore. Violence today predicts depressive symptoms tomorrow. Symptoms today predict a higher risk of violence tomorrow. When violence is in the picture, the risk of a future suicide attempt rises, at least in the datasets that tracked it. That bidirectionality is the clinical headline. It means that when a woman presents with depression, you have to think about her safety—not just her serotonin. It also means that when a woman discloses partner violence, you should be alert to the onset or worsening of depression over the coming months and years. In numbers, the pooled odds ratios were just under two in both directions—1.97 from violence to later depressive symptoms and 1.93 from symptoms to later violence. Those are not small shifts in risk. For men, the message is different: we see some association from violence to later depressive symptoms, but the evidence is thin, and there's no stable signal yet for suicide attempts or for symptoms forecasting later violence. That's a data problem as much as a biological or social one. There simply haven't been enough well-designed, adequately powered longitudinal studies that include men and measure both directions with the same care. So what do we do with this? First, prevention and support for survivors are mental health interventions, whether or not they're billed that way. Reduce partner violence and you likely reduce new onsets of depressive symptoms and, in women, some suicide attempts. Second, clinicians should weave safety questions into depression care. Not as a checkbox, but as part of understanding risk and planning follow-up. Temporality doesn't prove causality, but it strengthens it and makes it actionable. Looking ahead—briefly, and with discipline—the research agenda is clear. Build cohorts that measure physical, sexual, and emotional violence with standardized tools; use both symptom scales and diagnostic interviews; and adjust for the confounders we know matter, like childhood adversity and alcohol use. Include men in sufficient numbers. When feasible, harmonize measures across studies so future meta-analyses aren't constrained to a sliver of what's been collected. If Devries and colleagues gave us the map of the terrain, the next wave of work can give us altitude—sharper estimates, better causal leverage, and interventions tuned to where the risk actually concentrates. Until then, the story you can carry with you is simple and sober. Intimate partner violence and depression are entangled across time, especially for women, and that entanglement sometimes extends to suicide attempts. Breaking the violence is a way to bend the mental health curve. Spotting depression can be a way to interrupt the next chapter of harm.