Cross-national epidemiology of DSM-IV major depressive episode

Evelyn J. Bromet, Laura Helena Andrade, Irving Hwang, Nancy A. Sampson, Jordi Alonso, Giovanni de Girolamo, Ron de Graaf, Koen Demyttenaere, Chiyi Hu, Noboru Iwata, Aimée Karam, Jagdish Kaur, Stanislav Kostyuchenko, Jean‐Pierre Lépine, Daphna Levinson, Herbert Matschinger, María Elena Medina Mora, Mark Oakley Browne, José Posada‐Villa, María Carmen Viana, David R. Williams, Ronald C. Kessler•View original

OverviewBalancedalloy voice

If you want to know how common depression really is around the world, you run into a simple problem: most countries don't have the same yardstick. There are different questions, different interviews, and different thresholds. Bromet and colleagues set out to fix that. They built a single, careful system—the World Mental Health surveys—that they could take to 18 countries and put in front of eighty-nine thousand and thirty-seven people. Same questions. Same training. Same rules. Now you can start making apples-to-apples comparisons. First, what exactly are they measuring? They focus on a major depressive episode as defined in the fourth edition of the Diagnostic and Statistical Manual. Think five out of nine hallmark symptoms—like depressed mood, loss of interest, sleep and appetite changes, and fatigue—clustered together for at least two weeks, most of the day, nearly every day, with real distress or impairment. The team used the World Health Organization's Composite International Diagnostic Interview, or CIDI, a structured interview that lay interviewers can deliver. They opened with three screening stems—sadness, discouragement, and loss of interest lasting several days. If you said yes to any, you went into the full module. And because words matter, bilingual clinicians in each country translated and back-translated key phrases—what "sadness" means, how "loss of interest" lands—so the questions felt native while the criteria stayed fixed. Clinical reappraisals in several sites showed that CIDI-based diagnoses lined up well with blinded clinician assessments, which is the bedrock you want for a project like this. Methodologically, they did the unglamorous work that makes the numbers credible. Interviews were face-to-face, done by trained laypeople. To keep things manageable, the survey had two parts: everyone answered core questions about common mental disorders, and a subset went deeper on additional disorders and correlates. They weighted that second part by the inverse of the selection probability so estimates reflect the full population, not just the oversampled group. Samples were drawn to be nationally representative where possible or representative of major urban areas, then reweighted to match each country's demographics. The average weighted response rate was about seventy-two percent, which is solid by cross-national standards. And when it came time to analyze the data, they used design-based standard errors—Taylor series linearization, for the methods fans—to respect clustering and weights, and set a conventional two-sided significance level. So what did they find when they held the yardstick steady? Lifetime depression was more commonly reported in high-income settings than in low- and middle-income ones: approximately fifteen percent versus eleven percent. That's a real gap. But when you look at the past twelve months, those averages converge—about five point five percent in high-income countries and five point nine percent in low- and middle-income countries. If you zoom in even further to the past thirty days, the picture flips: one point eight percent in high-income versus two point six percent in lower-income contexts. What that says is nuanced. People in wealthier countries are more likely to report having ever had a major episode across their lives, but recent episodes are just as common, if not a touch more common, outside the wealthiest settings. Persistence, recall, and help-seeking could all be playing roles here. Hold that thought. The spread across individual countries is wide, and that's where it gets interesting. On the twelve-month measure, Japan sits near the low end around two percent, while Brazil is near the high end just above ten percent. For lifetime rates, several high-income countries—France, the Netherlands, New Zealand, and the United States—cluster above eighteen percent. At the low end of lifetime prevalence are India, Mexico, China, and South Africa, each below ten percent. A single definition and uniform methods didn't collapse the world into sameness; they revealed a map with peaks and valleys. Now, someone might ask, are these differences just a quirk of who says yes to those screening stems? The team checked. Across countries, the proportion of people who said they had a stretch of sadness or loss of interest varied by a factor of three point three from the lowest to the highest. And among those who screened positive, the fraction who actually met full criteria also varied, by a factor of three point zero. In other words, both the tendency to endorse the first gate and the chance of meeting full criteria once inside varied, and those differences were not all pointing in the same direction. That argues against a single explanation like a soft or hard threshold being the whole story. Timing matters too. If you take the twelve-month prevalence and divide it by the lifetime prevalence, you get a crude sense of how much recent depression there is relative to the lifetime pool. That ratio was lower in high-income countries—about thirty-eight percent—than in low- and middle-income countries, where it was closer to fifty-three percent. That could reflect more past cases being remembered and reported in wealthier settings, more enduring or recurrent episodes in lower-income settings, or some mix. The data can't by themselves untangle memory from course, but the contrast is there. Age of onset tells you when depression usually shows up. Across all surveys, the median fell in the mid-twenties, a little later in high-income contexts than in lower-income ones—about twenty-six versus twenty-four. In the United States, it clustered around twenty-three; in Japan it was closer to thirty; in China nearer to nineteen. That spread, from late teens into the early thirties, says something simple and sobering: this is mostly a disorder of the first half of adulthood. It arrives when education, work, and family formation are underway. Numbers about burden can be abstract until you ask, what does it do to a person's day? The team used the World Health Organization Disability Assessment Schedule to quantify functional impairment in the thirty days before the interview. People with current depression were far more limited than those without. In high-income countries, the average impairment score was about fifteen among those with current episodes and about three among those without. That's a big gap—on the order of one and a half standard deviations—and the pattern was similarly large in low- and middle-income settings. Not surprisingly, impairment eased as episodes receded into the past. People who had a depression in the past year but not in the past month still showed more restriction—often two to four times more—than those with no history, and by the time you reach people whose episodes were more than a year ago, the average gap narrowed and, in several countries, no longer reached statistical significance. Here's a cross-national twist. Countries with more depression also tended to report more impairment, and that link was strongest outside the wealthiest settings. Across all eighteen countries, the correlation between twelve-month prevalence and impairment hovered around zero point five. Split the sample, and it was modest in high-income countries but very strong—about zero point eight—in low- and middle-income ones. When the team decomposed prevalence into stem-question endorsement and the chance of meeting criteria among those who screened in, the impairment signal lined up more with the stem endorsement, especially in lower-income countries. In plainer language: in places where more people say they've had extended sadness or lost interest, you also see more disability. Who's at higher risk? Two patterns are remarkably consistent. Women have roughly twice the odds of a twelve-month episode as men, a ratio that held in most sites. And then there's age, which flips depending on context. In many high-income countries, younger adults carry the heaviest load; in several lower-income contexts, the risk shifts toward older adults. That divergence matters for planning. A clinic geared to university students makes sense in one setting; door-to-door outreach for older adults might be smarter in another. Marital status also divides along income lines. In high-income countries, being separated or never having married often shows the strongest association with depression. In lower-income countries, it's divorce and widowhood that stand out. Some country snapshots bring the point home. In Japan, being divorced is linked to a markedly higher risk—odds ratios around five in the reported estimates. In China, the divorce signal looks even stronger, closer to six. Those aren't just sociological footnotes; they point to the pathways by which social disruption and support, or the lack of it, can drive mood disorders. Economic position and education add another layer. In wealthier countries, the poorest groups tend to face about double the odds of depression compared with the highest-income stratum. In lower-income countries, income itself was not a consistent predictor. Education is even more context-dependent. In India, having the least education was associated with dramatically higher risk—an order of magnitude larger than the highest education group, in some analyses about fourteen-fold. Yet in Japan and China, the least-educated groups sometimes showed the lowest risk. Same variable, very different worlds. That's a reminder that schooling can be a marker of status, stress, or opportunity depending on where you look. All of this rests on a standardized analytic approach. The team treated each country's complex sample design seriously—weights to restore population structure, clustering to capture how interviews were grouped, and variance estimation that respects both. They ran logistic regressions to estimate associations and simple F tests to compare impairment across people with current, recent, or more remote episodes. The goal wasn't the most exotic model; it was comparability. Nigeria, by the way, was left out of some analyses because its reported prevalence was extremely low, raising worries about measurement or response that could distort pooled results. Of course, there are caveats. These are cross-sectional surveys; they can't tell you whether divorce caused the depression or the depression strained the marriage. Lifetime reporting is vulnerable to memory—especially for milder past episodes that never saw treatment—and recall may vary by culture. Response rates varied across sites, and some groups were simply not in the sampling frame: people who were institutionalized, incarcerated, on active military duty, or living with severe disability. Even with careful translation, words like "despair" or "guilt" don't carry identical connotations everywhere. The good news is that clinical reappraisal studies in multiple countries found good agreement between the CIDI diagnoses and clinician judgment, which helps anchor the patterns we're seeing. If you zoom out, two big messages come into focus. First, depression is common everywhere and deeply impairing when it's present. The exact numbers shift across borders, but the public-health footprint does not. Second, the shape of risk—the ages most affected, the social statuses that carry the heaviest load—depends on context. In high-income settings where youth are at higher risk, you might prioritize campus screening and early-career supports. In places where divorce or widowhood maps onto risk among older adults, social protection and grief-focused services might pay larger dividends. The point of building a single yardstick wasn't to flatten the world. It was to let us see the differences clearly enough to act on them. Bromet and the World Mental Health team gave us that tool. The next step is less glamorous than a new molecule or a bright new therapy, but it's arguably just as important: keep measuring with the same rigor, watch how these patterns move as societies change, and, country by country, point prevention and care toward the people most likely to need it. That's how surveillance becomes service, and how a global map turns into local action.

They built a single, careful system—the World Mental Health surveys—that they could take to 18 countries and put in front of eighty-nine thousand and thirty-seven people. Same questions. Same training. Same rules. Now you can start making apples-to-apples comparisons.

First, what exactly are they measuring? They focus on a major depressive episode as defined in the fourth edition of the Diagnostic and Statistical Manual. Think five out of nine hallmark symptoms—like depressed mood, loss of interest, sleep and appetite changes, and fatigue—clustered together for at least two weeks, most of the day, nearly every day, with real distress or impairment.

The team used the World Health Organization's Composite International Diagnostic Interview, or CIDI, a structured interview that lay interviewers can deliver. They opened with three screening stems—sadness, discouragement, and loss of interest lasting several days. If you said yes to any, you went into the full module.

And because words matter, bilingual clinicians in each country translated and back-translated key phrases—what "sadness" means, how "loss of interest" lands—so the questions felt native while the criteria stayed fixed. Clinical reappraisals in several sites showed that CIDI-based diagnoses lined up well with blinded clinician assessments, which is the bedrock you want for a project like this.

Methodologically, they did the unglamorous work that makes the numbers credible. Interviews were face-to-face, done by trained laypeople. To keep things manageable, the survey had two parts: everyone answered core questions about common mental disorders, and a subset went deeper on additional disorders and correlates.

They weighted that second part by the inverse of the selection probability so estimates reflect the full population, not just the oversampled group. Samples were drawn to be nationally representative where possible or representative of major urban areas, then reweighted to match each country's demographics. The average weighted response rate was about seventy-two percent, which is solid by cross-national standards.

And when it came time to analyze the data, they used design-based standard errors—Taylor series linearization, for the methods fans—to respect clustering and weights, and set a conventional two-sided significance level.

So what did they find when they held the yardstick steady? Lifetime depression was more commonly reported in high-income settings than in low- and middle-income ones: approximately fifteen percent versus eleven percent. That's a real gap.

But when you look at the past twelve months, those averages converge—about five point five percent in high-income countries and five point nine percent in low- and middle-income countries. If you zoom in even further to the past thirty days, the picture flips: one point eight percent in high-income versus two point six percent in lower-income contexts. What that says is nuanced.

People in wealthier countries are more likely to report having ever had a major episode across their lives, but recent episodes are just as common, if not a touch more common, outside the wealthiest settings. Persistence, recall, and help-seeking could all be playing roles here. Hold that thought.

The spread across individual countries is wide, and that's where it gets interesting. On the twelve-month measure, Japan sits near the low end around two percent, while Brazil is near the high end just above ten percent. For lifetime rates, several high-income countries—France, the Netherlands, New Zealand, and the United States—cluster above eighteen percent.

At the low end of lifetime prevalence are India, Mexico, China, and South Africa, each below ten percent. A single definition and uniform methods didn't collapse the world into sameness; they revealed a map with peaks and valleys.

Now, someone might ask, are these differences just a quirk of who says yes to those screening stems? The team checked. Across countries, the proportion of people who said they had a stretch of sadness or loss of interest varied by a factor of three point three from the lowest to the highest.

And among those who screened positive, the fraction who actually met full criteria also varied, by a factor of three point zero. In other words, both the tendency to endorse the first gate and the chance of meeting full criteria once inside varied, and those differences were not all pointing in the same direction. That argues against a single explanation like a soft or hard threshold being the whole story.

Timing matters too. If you take the twelve-month prevalence and divide it by the lifetime prevalence, you get a crude sense of how much recent depression there is relative to the lifetime pool. That ratio was lower in high-income countries—about thirty-eight percent—than in low- and middle-income countries, where it was closer to fifty-three percent.

That could reflect more past cases being remembered and reported in wealthier settings, more enduring or recurrent episodes in lower-income settings, or some mix. The data can't by themselves untangle memory from course, but the contrast is there.

Age of onset tells you when depression usually shows up. Across all surveys, the median fell in the mid-twenties, a little later in high-income contexts than in lower-income ones—about twenty-six versus twenty-four. In the United States, it clustered around twenty-three; in Japan it was closer to thirty; in China nearer to nineteen.

That spread, from late teens into the early thirties, says something simple and sobering: this is mostly a disorder of the first half of adulthood. It arrives when education, work, and family formation are underway.

Numbers about burden can be abstract until you ask, what does it do to a person's day? The team used the World Health Organization Disability Assessment Schedule to quantify functional impairment in the thirty days before the interview. People with current depression were far more limited than those without.

In high-income countries, the average impairment score was about fifteen among those with current episodes and about three among those without. That's a big gap—on the order of one and a half standard deviations—and the pattern was similarly large in low- and middle-income settings. Not surprisingly, impairment eased as episodes receded into the past.

People who had a depression in the past year but not in the past month still showed more restriction—often two to four times more—than those with no history, and by the time you reach people whose episodes were more than a year ago, the average gap narrowed and, in several countries, no longer reached statistical significance.

Here's a cross-national twist. Countries with more depression also tended to report more impairment, and that link was strongest outside the wealthiest settings. Across all eighteen countries, the correlation between twelve-month prevalence and impairment hovered around zero point five.

Split the sample, and it was modest in high-income countries but very strong—about zero point eight—in low- and middle-income ones. When the team decomposed prevalence into stem-question endorsement and the chance of meeting criteria among those who screened in, the impairment signal lined up more with the stem endorsement, especially in lower-income countries. In plainer language: in places where more people say they've had extended sadness or lost interest, you also see more disability.

Who's at higher risk? Two patterns are remarkably consistent. Women have roughly twice the odds of a twelve-month episode as men, a ratio that held in most sites.

And then there's age, which flips depending on context. In many high-income countries, younger adults carry the heaviest load; in several lower-income contexts, the risk shifts toward older adults. That divergence matters for planning.

A clinic geared to university students makes sense in one setting; door-to-door outreach for older adults might be smarter in another.

Marital status also divides along income lines. In high-income countries, being separated or never having married often shows the strongest association with depression. In lower-income countries, it's divorce and widowhood that stand out.

Some country snapshots bring the point home. In Japan, being divorced is linked to a markedly higher risk—odds ratios around five in the reported estimates. In China, the divorce signal looks even stronger, closer to six.

Those aren't just sociological footnotes; they point to the pathways by which social disruption and support, or the lack of it, can drive mood disorders.

Economic position and education add another layer. In wealthier countries, the poorest groups tend to face about double the odds of depression compared with the highest-income stratum. In lower-income countries, income itself was not a consistent predictor.

Education is even more context-dependent. In India, having the least education was associated with dramatically higher risk—an order of magnitude larger than the highest education group, in some analyses about fourteen-fold. Yet in Japan and China, the least-educated groups sometimes showed the lowest risk.

Same variable, very different worlds. That's a reminder that schooling can be a marker of status, stress, or opportunity depending on where you look.

All of this rests on a standardized analytic approach. The team treated each country's complex sample design seriously—weights to restore population structure, clustering to capture how interviews were grouped, and variance estimation that respects both. They ran logistic regressions to estimate associations and simple F tests to compare impairment across people with current, recent, or more remote episodes.

The goal wasn't the most exotic model; it was comparability. Nigeria, by the way, was left out of some analyses because its reported prevalence was extremely low, raising worries about measurement or response that could distort pooled results.

Of course, there are caveats. These are cross-sectional surveys; they can't tell you whether divorce caused the depression or the depression strained the marriage. Lifetime reporting is vulnerable to memory—especially for milder past episodes that never saw treatment—and recall may vary by culture.

Response rates varied across sites, and some groups were simply not in the sampling frame: people who were institutionalized, incarcerated, on active military duty, or living with severe disability. Even with careful translation, words like "despair" or "guilt" don't carry identical connotations everywhere. The good news is that clinical reappraisal studies in multiple countries found good agreement between the CIDI diagnoses and clinician judgment, which helps anchor the patterns we're seeing.

If you zoom out, two big messages come into focus. First, depression is common everywhere and deeply impairing when it's present. The exact numbers shift across borders, but the public-health footprint does not.

Second, the shape of risk—the ages most affected, the social statuses that carry the heaviest load—depends on context. In high-income settings where youth are at higher risk, you might prioritize campus screening and early-career supports. In places where divorce or widowhood maps onto risk among older adults, social protection and grief-focused services might pay larger dividends.

The point of building a single yardstick wasn't to flatten the world. It was to let us see the differences clearly enough to act on them. Bromet and the World Mental Health team gave us that tool.

The next step is less glamorous than a new molecule or a bright new therapy, but it's arguably just as important: keep measuring with the same rigor, watch how these patterns move as societies change, and, country by country, point prevention and care toward the people most likely to need it. That's how surveillance becomes service, and how a global map turns into local action.

Related lectures