COVID-19 Pandemic and Lockdown Measures Impact on Mental Health Among the General Population in Italy

Picture early March 2020 in Italy. Sirens, empty piazzas, a country shutting its doors to itself. On March ninth, the government locked down the nation. Three weeks later, as the first wave crested, a group led by Rossi asked a blunt, urgent question: how is everyone holding up? Not in a year, not after careful clinic visits, but right now, at scale, with enough detail to guide help where it’s needed most. They built a rapid, national snapshot. A web-based, cross-sectional survey ran from March twenty-seventh to April sixth, aligning with the contagion peak tracked by the World Health Organization. Participation was anonymous, consent was online, and the study had ethics approval from the University of L'Aquila. Recruitment leaned on sponsored social media and a snowball approach, pushing the questionnaire across all regions. That makes it fast and wide, but it also means they couldn’t calculate a response rate. They never knew exactly how many people saw the invite. The sample was big: 18,147 people completed the questionnaire. Women were the majority—about seventy-nine point six percent—and the median age was thirty-eight. Those demographics reflect the method; social media sampling often skews female and younger. The team accounted for that by including age, gender, and region in the models, grouping residence into Northern, Central, and Southern Italy to capture the virus’s uneven footprint across the map. What did they measure? Six outcomes that together paint the mental health canvas of a lockdown. Post-traumatic stress symptoms, the intrusive memories and hyperarousal you can feel after a shocking event. Depression and anxiety, the core mood and worry dimensions that tend to rise in crises. Insomnia, because stress scrambles sleep. Perceived stress, a broad sense of strain. And adjustment disorder symptoms, which capture preoccupation and impairment when life is upended. Each domain used a standard tool. Depression came from the nine-item Patient Health Questionnaire with a severe cut-off at fifteen; anxiety from the seven-item Generalized Anxiety Disorder scale, also with a fifteen cut-off; and insomnia from the Insomnia Severity Index at twenty-two for severe. For stress, they flagged the top quartile on the Perceived Stress Scale. Post-traumatic stress used the Global Psychotrauma Screen, counting more than three out of five key symptoms. Adjustment disorder relied on the International Adjustment Disorder Questionnaire, adapted so the stressor had to be COVID-related. One technical note: most of these had strong internal consistency—think of it as the items hanging together—with alphas around zero point eight seven to zero point nine one. The post-traumatic stress screen was lower, at zero point five four, a reminder to treat that particular signal with caution. They didn’t analyze these outcomes one by one. Instead, they fit them together using a seemingly unrelated multivariate logistic regression—imagine six linked equations, one for each outcome, estimated jointly so that errors that spill over from one to another are allowed to correlate. That matters when the same people are answering all the questions and symptoms travel in packs. The models adjusted for age, gender, region, education, occupation, and relationship status, and they did something crucial: they controlled for prior psychiatric diagnoses and childhood trauma. That helps separate the pandemic’s shock from background vulnerability. COVID-specific covariates captured quarantine status, changes in work—discontinuation, remote work, or working more than usual—whether a loved one was infected, hospitalized, or had died, and a checklist of COVID-related stressful events. Missing data were rare, under three percent, so they used listwise deletion. If you’re counting software, the team ran this in Stata sixteen, using a post-estimation routine called suest to knit the logistic models together. So what did Italy look like, three to four weeks into lockdown? Heavy. About thirty-seven percent screened positive for post-traumatic stress symptoms—six thousand six hundred four people out of the sample. Severe anxiety was reported by twenty point eight percent—that’s three thousand seven hundred individuals—and severe depression by seventeen point three percent, or three thousand eighty-four people. These are not clinical diagnoses, but they are meaningful red flags when you see them at this scale. Two other signals stood out. Severe insomnia affected seven point three percent, or one thousand three hundred one respondents, and high perceived stress—sitting in the top quarter of the stress scale—was present in twenty-one point eight percent, about three thousand eight hundred ninety-five people. Adjustment disorder symptoms were common too, at twenty-two point nine percent, which translates to four thousand ninety-two individuals. All of this emerged within a month of the shutdown. Early. Fast. Widespread. Who was most at risk? Across outcomes, women consistently showed higher odds than men. The gaps were largest for post-traumatic stress and perceived stress: women had about double the odds for each—two point twelve for post-traumatic stress and two point zero six for high perceived stress. Anxiety was higher too, with an odds ratio of one point seventy-seven. The pattern extended to depression, insomnia, and adjustment disorder as well, though the increases were smaller. If you’ve seen early reports from China’s first wave, this will sound familiar; women and younger adults often shouldered more of the psychological burden there too. Age mattered. Younger respondents had higher odds of most adverse outcomes. Anxiety rose the steepest with youth—an odds ratio of one point seventy-two—followed by perceived stress at one point seventy-six and depression at one point fifty-five. Adjustment disorder didn’t show a clear age pattern; its odds ratio hovered near one and wasn’t statistically convincing. That’s a nuance worth pausing on. Not every outcome moved in lockstep with age, even though the overall tilt was clear. Place mattered as well. Compared to the North, the South showed elevated odds across several domains: one point thirty-six for post-traumatic stress, one point twenty-nine for anxiety, and one point forty-one for insomnia. Depression and perceived stress were modestly higher too. Adjustment disorder didn’t track this regional gradient. This could reflect differences in healthcare resources, economic stress, or the timing of the outbreak’s spread. The models can’t tell you why; they just map where the signal was stronger. Now quarantine. Being under quarantine—because you were infected, exposed, or at risk—was not just an inconvenience. It amplified symptoms. Odds of adjustment disorder more than doubled to two point twenty-eight. Post-traumatic stress climbed to one point seventy-four, and anxiety to one point fifty-two. Depression and perceived stress also ticked up, though by smaller amounts. That’s the lived cost of isolation layered on top of fear. Separate from quarantine, people who reported COVID-related stressful events—the kinds of concrete blows the pandemic dealt to daily life—were more likely to endorse symptoms across the board. High perceived stress jumped the most, with an odds ratio of one point eighty-two. Anxiety and depression followed at one point sixty-four and one point fifty-eight, and insomnia rose to one point fifty-eight. You can think of these events as the sparks that keep reigniting distress even when the immediate health threat recedes. One modeling quirk here: because adjustment disorder was defined to require a COVID-related stressor, they couldn’t estimate the effect of stressors on adjustment disorder within the same framework without breaking that definition. Work, as always, was a hinge. Discontinuing work because of COVID—not working remotely, but stopping—was associated with higher odds of several outcomes. Depression showed the clearest link at one point forty, while insomnia and perceived stress rose to one point twenty-two and one point nineteen. Post-traumatic stress and anxiety nudged up too. On the flip side, working more than usual wasn’t a buffer; it came with its own strain. Odds of high perceived stress rose to one point seventy-one, post-traumatic stress to one point forty-one, and adjustment disorder to one point thirty-nine. Different routes, similar destination: distress. Bereavement left a stark imprint. If a loved one had died from COVID-19, odds of post-traumatic stress rose to one point sixty-eight, and insomnia to one point seventy-four. Depression increased to one point forty-one, and adjustment disorder mirrored insomnia at one point seventy-four. Perceived stress climbed as well. It’s intuitive, and still sobering, to hear the numbers align with the human cost. Education and occupation added another layer. Lower education was tied to higher risk in most domains, peaking for insomnia with an odds ratio of one point seventy-six, and for depression at one point sixty-two. Homemakers showed elevated odds across several outcomes—depression at one point thirty-five, for example—hinting at the pressure cooker of domestic work during lockdown. Unemployment sharpened risk for depression at one point fifty-nine and for adjustment disorder at one point thirty-three, with anxiety also elevated. Students were a mixed picture: they were less likely to report post-traumatic stress, with an odds ratio of zero point seventy-nine, but more likely to endorse depression at one point sixty and higher perceived stress. Different roles, different vulnerabilities. Two background factors were important and were explicitly adjusted for: childhood trauma and prior mental illness. Childhood trauma was most strongly tied to adjustment disorder and depression, with odds ratios of one point fifty and one point forty-one, and it also lifted anxiety and insomnia. Prior psychiatric diagnosis stood out as a potent predictor across the board—depression at two point nineteen and anxiety at two point ten were the largest—along with post-traumatic stress, perceived stress, insomnia, and adjustment disorder. Here’s the takeaway: even after accounting for these, COVID-specific factors—quarantine, bereavement, work disruption, acute stressors—were independently associated with worse outcomes. The pandemic wasn’t just revealing existing cracks; it was widening them. Every study has its seams, and Rossi and colleagues are clear about theirs. Social media sampling introduced selection bias and a heavy female skew. Self-report screens can over- or under-estimate prevalence compared with clinical interviews. Cross-sectional data can’t prove causality. The internal consistency of the post-traumatic stress screen was modest, suggesting some measurement noise there. They also couldn’t tell you how many people saw the survey and chose not to click. Against that, weigh the strengths: a very large, national sample gathered during the peak of the crisis, validated instruments for rapid surveillance, vanishingly low missing data—under three percent—and analyses that accounted for prior psychiatric history and childhood trauma. What does this add up to? Three to four weeks into lockdown, Italy was carrying a broad psychological burden: elevated trauma-related symptoms, anxiety, depression, sleep disruption, stress, and difficulty adjusting. Women and younger adults were at higher risk. Quarantine, work changes, and bereavement amplified distress in distinct ways, and those effects held even when you control for who had struggled before. Similar patterns showed up in early Chinese surveys, which adds confidence that these aren’t idiosyncrasies of one dataset but common features of how societies absorb a shock like COVID-19. If you’re a policymaker or a clinician, the implications are practical. Surveillance needs to be timely and repeated, because these signals can shift as restrictions change. Interventions should target groups and stressors we can now name: support for those in quarantine, outreach after bereavement, resources for people who lost work and for those suddenly working more, and special attention to women and younger adults who, on average, were carrying more of the load. And maybe the meta-lesson is about method. In a crisis, speed and scale matter. A web survey won’t replace a careful diagnostic interview, but as Rossi and colleagues showed, it can give you a high-resolution map while the earth is still shaking. That’s the kind of map you can act on.