Global Burden and Demographic Decomposition of Seborrheic Dermatitis, 1990-2023: A Secondary Analysis of GBD 2023 Estimates.

Save

Background

Case counts of seborrheic dermatitis have progressively increased, which may reflect population growth, changing age structure, and changing age-specific rates. We quantified these components within a consistent Global Burden of Disease (GBD) 2023 series.

Methods

We analyzed incidence, prevalence, and disability-adjusted life-years (DALYs) for 1990–2023 globally and by region, sex, age, and Socio-demographic Index group. Symmetric Das Gupta decomposition partitioned changes in global counts into population-size, age-structure, and age-specific-rate components. Replacement-order and age-group sensitivities assessed component directions.

Deterministic scenarios combined fixed 2023 age-specific rates with each UN World Population Prospects 2024 variant’s age-specific populations through 2040.

Results

In 2023, modeled incident cases totaled 183.94 million (95% uncertainty interval 170.17–197.03 million), prevalent cases 26.73 million (25.14–28.48 million), and DALYs 358,629 (206,994–558,580), increasing by 50.7%, 62.3%, and 61.3% from 1990. Age-standardized rates changed by 0.31%, 0.23%, and 0.38%, respectively. Population growth accounted for 100.8%, 85.1%, and 86.2% of the net increases.

Age-structure change contributed −1.7% to incident cases but +14.2% to prevalent cases and +12.8% to DALYs; age-specific-rate changes contributed 0.9%, 0.6%, and 1.0%. Age-structure component directions persisted across replacement orders and ten-year age grouping. The fixed-rate Medium scenario yielded 204.18 million incident cases in 2040.

Conclusion

Within GBD 2023 point estimates, population growth was the main accounting component of increasing absolute burden. Age-structure change contributed in opposite directions to incidence versus prevalence and DALYs. These descriptive results support reporting counts alongside standardized rates; they do not establish individual disease duration or biological mechanisms.

Introduction

Seborrheic dermatitis is a common, relapsing inflammatory dermatosis affecting the scalp, face, and trunk. Although it is rarely fatal, recurrent symptoms and repeated treatment can impair quality of life and create a sustained demand for dermatologic care.1,2 Its pathogenesis is associated with the interaction of Malassezia yeasts, sebaceous lipid metabolism, host inflammatory responses, and skin-barrier function.3–5 Its population burden is best described with complementary measures: absolute counts quantify the number of people affected, age-specific rates show how burden varies across the life course, and age-standardized rates facilitate comparisons across populations with different age structures.

Previous GBD analyses have reported increasing counts of seborrheic dermatitis alongside comparatively stable age-standardized rates.6,7 Huang et al also applied demographic decomposition and projections through 2040 to adults aged 55 years and older using GBD 2021.8 The question addressed here is how the components compare across incidence, prevalence, and DALYs over the full age range in GBD 2023. Using a single release keeps temporal comparisons internally consistent and avoids treating changes between GBD releases as changes in disease occurrence.

We therefore described global, regional, sex-specific, age-specific, and SDI-stratified burden from 1990 through 2023 and decomposed global count changes into population growth, age-structure change, and age-specific-rate change. Age-specific rates, population shares, age-level contributions, and decomposition sensitivities directly inform this comparison. We also reported the modeled prevalence-to-incidence ratio descriptively and constructed fixed-rate demographic scenarios through 2040.

Exploratory temporal and age-period-cohort analyses are reported in the supplement because their diagnostic and identification limitations restrict inference.

Materials and Methods

Data Sources

This secondary analysis used modeled estimates from the IHME Global Burden of Disease (GBD) Results Tool for seborrheic dermatitis (cause ID 979; exported label “Seborrhoeic dermatitis”, corresponding to ICD-10 category L21).9 We extracted annual incidence, prevalence, and DALYs from 1990 through 2023 for the global aggregate, 21 GBD regions, both sexes, females, males, and five GBD Socio-demographic Index (SDI) groups. Counts were taken from the “Number” metric and all-age category; age-standardized rates were taken directly from the “Rate” metric and expressed per 100,000 population. Global age-specific analyses used the 20 GBD age groups from 0–4 years through 95 years or older.

Broad age groups were taken directly from the corresponding GBD exports. GBD 2023 is a versioned, frozen release: exported values do not change after publication, so the release identifier and source-file checksums, rather than the calendar date of download, identify the exact data used. Detailed query fields, source files, checksums, and release checks are provided in the Supplementary Methods.

The analysis retained the exported point estimates and 95% uncertainty intervals (UIs). Records were keyed by measure, location, sex, age, cause, metric, and year; duplicate records were resolved using the prespecified numerical tolerance, and all retained records satisfied the ordering lower bound ≤ estimate ≤ upper bound. Country-level GBD 2021 files and previously derived tables were excluded.

SDI groups were treated as descriptive strata. The published GBD methods provide the modeling framework for the estimates used here.10

For the 2040 analysis, we used the official UN World Population Prospects (WPP) 2024 revision for the World.11 Each of the Medium, Low, and High variants supplied its own age-specific populations; the Low and High scenarios therefore incorporate their respective age structures as well as population sizes. WPP counts in thousands were converted to persons, and ages 95–99 and 100 years or older were combined to match the GBD 95+ group. The WPP source files, checksums, and age-group mapping are described in the Supplementary Methods, with provenance recorded in Supplementary Table S4g.

Statistical Analysis

Endpoint percentage change was calculated as 100 × (2023 estimate / 1990 estimate − 1). To summarize long-term trends, we regressed the logarithm of each annual age-standardized point estimate on calendar year and calculated the estimated annual percentage change (EAPC) as 100 × [exp(β) − 1]. Conditional 95% confidence intervals describe uncertainty in the fitted trend and are distinct from the GBD UIs; reporting of estimates and uncertainty followed the GATHER principles.12

We used the Das Gupta standardization and decomposition framework13 to partition the change in global counts between 1990 and 2023 into population-growth, age-structure, and age-specific-rate components. The latter two are conventionally termed population aging and epidemiological change, but here denote changes in population age shares and modeled age-specific rates. For each age group, the population denominator was reconstructed from the GBD count and age-specific rate.

With total population P, age share s_a, and age-specific rate r_a, the count was written as N = P × Σ(s_a × r_a) / 100,000. The 1990 values of P, s_a, and r_a were replaced by their 2023 values in each of the six possible orders, and each component was averaged across orders. Components sum to the modeled count change and describe its accounting allocation, not causal mechanisms.

We displayed age-level contributions to the age-structure component and assessed directions in each replacement order and after combining adjacent five-year into ten-year age groups.

For EAPC regressions only, conditional 95% confidence intervals used Newey–West heteroskedasticity- and autocorrelation-consistent covariance estimates with lag 3.14 Lag and time-window sensitivities are reported in Supplementary Table S1e. These conditional regression intervals do not propagate GBD source uncertainty.

For each global age group, we calculated the prevalence-to-incidence rate ratio in 1990 and 2023 as an exploratory comparison of two outputs from the same GBD model. Interpreting this ratio as duration would require steady-state and case-definition assumptions that are not established by these aggregate exports.15 We therefore did not interpret it as measured disease duration, persistence, or independent validation of a biological mechanism.

Supplementary analyses used the same GBD mean trajectories. Joinpoint-style segmented regression fitted log-rate models with zero to three candidate breakpoints and selected the lowest Bayesian information criterion.16 Conditional breakpoint intervals and model warnings were retained. Exploratory interrupted time-series (ITS) models used 2020 as a prespecified time marker, generalized least squares with AR(1) errors, and 2019/2021 sensitivity markers.17 Residual autocorrelation diagnostics were assessed.

For global incidence, a Gaussian penalized-spline age-period-cohort (APC) model fitted log age-specific rates using REML smoothing.18 Because age, period, and cohort are deterministically related, concurvity and basis-dimension sensitivities assess limitations of the fitted decomposition; this model does not adjust the demographic accounting for confounding or identify independent APC effects.

For the main 2024–2040 scenarios, each 2023 GBD age-specific rate was held constant and multiplied by its corresponding WPP population in each year and variant. Counts were summed across ages, and crude rates were computed by dividing scenario counts by the corresponding total population. No prediction intervals were assigned: the scenarios are deterministic consequences of specified rates and demographic inputs, and the Low–High span is not a probability interval.

Log-linear age-specific-rate extrapolations over 1990–2023, 2010–2023, and 2015–2023 were retained only as sensitivity scenarios. Historical comparisons used cutoffs in 2010, 2015, and 2018, predicting subsequent GBD point estimates through 2023 with reconstructed populations for those years; they assess rate assumptions conditional on known demography, not future population accuracy or independent clinical outcomes. All analyses used the archived R and Python code described in the Supplementary Methods.

Results

Unless otherwise noted, values are GBD 2023 modeled point estimates with the exported 95% UI in parentheses.

Overall Burden and Demographic Patterns

Absolute counts and age-standardized rates describe different aspects of modeled population burden. Incident cases rose from 122.08 million (95% UI 112.25–130.79 million) in 1990 to 183.94 million (170.17–197.03 million) in 2023, a 50.7% increase. Prevalent cases rose from 16.47 million (15.37–17.64 million) to 26.73 million (25.14–28.48 million), a 62.3% increase, and DALYs from 222,326 (127,839–350,122) to 358,629 (206,994–558,580), a 61.3% increase (Table 1, Figure 1 and Supplementary Tables S1a–S1c).Table 1Global Counts and Age-Standardized Rates of Seborrheic Dermatitis in 1990 and 2023Outcome1990 Estimate (95% UI)2023 Estimate (95% UI)1990 ASR (95% UI)2023 ASR (95% UI)Incidence122,083,803 (112,251,897–130,793,168)183,936,979 (170,173,984–197,028,604)2,258.75 (2,084.86–2,416.64)2,265.77 (2,093.78–2,423.63)Prevalence16,469,876 (15,368,473–17,641,858)26,728,921 (25,141,303–28,482,994)320.69 (300.19–343.12)321.44 (301.20–343.07)DALYs222,326 (127,839–350,122)358,629 (206,994–558,580)4.31 (2.48–6.74)4.32 (2.49–6.75)Notes: Values in parentheses are exported 95% GBD uncertainty intervals.

Counts are reported as numbers, with DALYs expressed in person-years of healthy life lost. Source: Institute for Health Metrics and Evaluation, GBD 2023 estimates. Rates and counts were extracted as separate metrics.

Percentage-change point estimates and exploratory regression CIs are reported separately in the supplementary tables.Abbreviations: ASR, age-standardized rate per 100,000; DALYs, disability-adjusted life-years.Figure 1Global modeled counts and age-standardized rates, 1990–2023. (A) all-age incident cases, (B) age-standardized incidence rate, (C) all-age prevalent cases, (D) age-standardized prevalence rate, (E) all-age DALYs, and (F) age-standardized DALY rate. Lines connect annual GBD 2023 point estimates; error bars at the 1990 and 2023 endpoints are exported 95% UIs.

Labels give the 2023 point estimate and percentage change from 1990. Rates are per 100,000.The data consists of six line graphs from 1990 to 2023. Image A shows ′Incidence count′ with cases rising from 122 million in 1990 to 184 million in 2023, a 50.7 percent increase.

Image B, ′Incidence rate′, remains nearly flat, ending at 2,266 per 100,000 in 2023, a 0.3 percent rise. Image C, ′Prevalence count′, shows cases increasing from 16 million in 1990 to 27 million in 2023, a 62.3 percent rise. Image D, ′Prevalence rate′, is nearly flat, ending at 321 per 100,000 in 2023, a 0.2 percent increase.

Image E, ′DALYs count′, rises from 222,000 in 1990 to 359,000 in 2023, a 61.3 percent increase. Image F, ′DALYs rate′, remains nearly flat, ending at 4 per 100,000 in 2023, a 0.4 percent rise. All graphs include vertical error bars at 1990 and 2023.A set of six line graphs showing incidence, prevalence and DALYs counts and rates from 1990 to 2023.

Age-standardized rates changed little over the same period. Incidence changed from 2,258.75 (2,084.86–2,416.64) to 2,265.77 (2,093.78–2,423.63) per 100,000, prevalence from 320.69 (300.19–343.12) to 321.44 (301.20–343.07), and the DALY rate from 4.31 (2.48–6.74) to 4.32 (2.49–6.75), corresponding to endpoint changes of 0.31%, 0.23%, and 0.38%.

The global EAPC was 0.005% per year for incidence (conditional HAC 95% CI −0.001% to 0.011%), −0.0004% for prevalence (−0.0069% to 0.0061%), and 0.0039% for the DALY rate (−0.0045% to 0.0123%). Sex-specific incidence EAPCs were 0.0237% per year (0.0177% to 0.0296%) in males and −0.0127% (−0.0181% to −0.0074%) in females. These changes were small in absolute terms (Figure 2A–C and Supplementary Table S1d).

Figure 2Subgroup comparisons of age-standardized rates, 1990 and 2023. (A–C) Rates by sex for incidence, prevalence, and DALYs; points and horizontal intervals are source point estimates and exported 95% UIs. (D–F) Rates by GBD Socio-demographic Index (SDI) group; points and horizontal intervals are source point estimates and exported 95% UIs; no continuous SDI relationship is fitted.

(G–I) Percentage change in modeled counts from 1990 to 2023 by SDI group; bars use endpoint point estimates, and marginal endpoint bounds were not used to construct uncertainty intervals for changes. SDI groups in D–I are ordered High, High-middle, Middle, Low-middle, and Low.Image A: Incidence rates per 100,000 by sex show females at ~2,350 in both 1990 and 2023, while males increased slightly from ~2,150 in 1990 to ~2,200 in 2023. Image B: Prevalence rates for females remained ~330 and for males, it rose slightly from ~315 in 1990 to ~320 in 2023.

Image C: DALYs for females stayed at ~4.5 and for males, it increased from ~4.0 in 1990 to ~4.2 in 2023. Image D: Incidence by SDI shows consistent rates across all levels, with high SDI at ~2,200 and others at ~2,300 or ~2,250 in both years. Image E: Prevalence by SDI remained ~320 across all levels.

Image F: DALYs by SDI stayed at ~4.2 for all levels. Image G: Incidence change by SDI shows increases, with low SDI at 127.7 percent. Image H: Prevalence change by SDI shows low SDI at 128.1 percent.

Image I: DALYs change by SDI shows low SDI at 128.9 percent. Legend: Year, 1990, 2023.A mixed chart showing incidence, prevalence and DALYs by sex and SDI, plus percent change in count.

Burden was concentrated in working-age adults. In 2023, people aged 15–49 years accounted for 130.08 million incident cases (95% UI 118.33–141.15 million), 14.56 million prevalent cases (13.09–15.95 million), and 198,462 DALYs (110,587–313,470), compared with 2.43 million incident cases (2.12–2.71 million), 1.91 million prevalent cases (1.67–2.19 million), and 23,480 DALYs (13,999–38,333) among those aged 75 years or older. The pattern reflects both population size and age-specific rates (Figure 3 and Supplementary Table S1f).

Figure 3Global age-specific counts and rates in 1990 and 2023 across 20 GBD age groups. (A) incident cases, (B) age-specific incidence rate, (C) prevalent cases, (D) age-specific prevalence rate, (E) DALYs, and (F) age-specific DALY rate. Points and horizontal intervals are source point estimates and exported 95% UIs.

Counts are shown in millions of cases or thousands of DALYs; age-specific rates are per 100,000.The image contains six dot-and-interval plots comparing data from 1990 and 2023 across 20 GBD age groups from less than 5 to 95 plus years. Image A shows incidence counts in millions, with the largest counts across the 15–49-year age groups. Image B shows incidence rates per 100,000, peaking at ages 35–39.

Image C shows prevalence counts in millions, with the largest counts across the 15–49-year age groups. Image D shows prevalence rates per 100,000, peaking at ages 75–79. Image E shows DALYs in thousands, with the largest counts across the 15–49-year age groups.

Image F shows DALY rates per 100,000, peaking at ages 75–79. Each plot compares 1990 and 2023 point estimates with horizontal uncertainty intervals.Six dot-and-interval plots comparing incidence, prevalence and DALYs by age group for 1990 and 2023.

In 2023, the highest age-specific incidence point estimate was at ages 35–39, 3,462.04 per 100,000 (95% UI 2,778.28–4,176.71), whereas prevalence and DALY rates peaked at ages 75–79, at 626.81 (519.56–754.20) and 7.83 (4.52–12.68), respectively. These distinct age profiles provide the rate weights for the demographic decomposition (Figure 3 and Supplementary Table S1f).

The numerical age-specific prevalence-to-incidence ratio was approximately 0.11 across ages 15–49 in both 1990 and 2023, increased from the mid-50s, and peaked at approximately 0.81 around ages 75–84 before declining slightly at the oldest ages (Supplementary Table S1g). This describes the relative age profiles of the two modeled rates and is not an estimate of individual disease duration.

Geographic and Socioeconomic Heterogeneity

Regional and SDI-stratified results help determine whether the overall pattern observed globally is broadly consistent across settings or concentrated in particular regions or levels of development. Regional rate estimates showed a modest but consistent gradient, with Central Asia and Eastern Europe generally ranking highest and high-income regions generally ranking lowest across outcomes. The highest 2023 incidence-rate estimates were in Eastern Europe, 2,456.63 (95% UI 2,261.21–2,625.71) per 100,000, and Central Asia, 2,417.39 (2,230.00–2,606.67); the lowest was in High-income North America, 1,993.90 (1,871.29–2,111.85).

Central Asia also had the highest prevalence-rate estimate, 354.41 (331.61–379.99), against a low of 305.11 (286.90–324.09) in High-income Asia Pacific, and the highest DALY-rate estimate, 4.81 (2.69–7.67), against a low of 4.09 (2.32–6.43) in Australasia (Table 2; Figure 4 and Supplementary Table S1a). Given the width and overlap of these UIs, particularly for DALYs, these regional orderings describe point estimates rather than statistically confirmed differences between regions.Table 2Regional Age-Standardized Rates of Seborrheic Dermatitis in 2023, by GBD RegionRegionIncidence ASR (95% UI)Prevalence ASR (95% UI)DALY ASR (95% UI)Andean Latin America2,259.60 (2,086.59–2,416.07)318.21 (299.14–341.07)4.29 (2.41–6.65)Australasia2,056.24 (1,904.19–2,203.31)306.35 (289.25–326.82)4.09 (2.32–6.43)Caribbean2,259.86 (2,086.88–2,416.47)318.26 (299.22–341.15)4.28 (2.43–6.67)Central Asia2,417.39 (2,230.00–2,606.67)354.41 (331.61–379.99)4.81 (2.69–7.67)Central Europe2,368.86 (2,203.87–2,513.30)351.36 (331.98–373.09)4.75 (2.72–7.53)Central Latin America2,281.82 (2,108.86–2,439.22)318.66 (298.39–340.77)4.26 (2.44–6.62)Central Sub-Saharan Africa2,259.45 (2,086.58–2,415.90)318.33 (299.34–341.25)4.26 (2.40–6.78)East Asia2,296.83 (2,116.40–2,465.42)318.36 (296.87–341.24)4.31 (2.46–6.72)Eastern Europe2,456.63 (2,261.21–2,625.71)354.11 (330.77–378.75)4.79 (2.75–7.53)Eastern Sub-Saharan Africa2,276.23 (2,102.67–2,434.02)318.52 (298.72–340.93)4.28 (2.46–6.69)High-income Asia Pacific2,068.39 (1,921.68–2,213.64)305.11 (286.90–324.09)4.11 (2.36–6.34)High-income North America1,993.90 (1,871.29–2,111.85)318.40 (308.10–328.36)4.21 (2.45–6.54)North Africa and Middle East2,262.27 (2,087.12–2,418.61)317.82 (298.83–340.83)4.27 (2.41–6.66)Oceania2,258.02 (2,085.36–2,414.35)317.86 (298.69–340.54)4.29 (2.42–6.61)South Asia2,295.26 (2,117.61–2,462.03)318.47 (296.69–341.22)4.28 (2.48–6.67)Southeast Asia2,282.74 (2,106.69–2,441.26)318.61 (298.27–340.77)4.32 (2.47–6.75)Southern Latin America2,052.32 (1,899.43–2,197.24)305.34 (287.94–326.40)4.10 (2.32–6.37)Southern Sub-Saharan Africa2,290.75 (2,115.79–2,456.07)318.87 (297.20–341.18)4.27 (2.44–6.68)Tropical Latin America2,300.53 (2,120.42–2,470.22)318.87 (297.12–341.85)4.28 (2.43–6.64)Western Europe2,058.61 (1,908.49–2,204.87)305.65 (288.54–325.61)4.10 (2.35–6.42)Western Sub-Saharan Africa2,280.99 (2,108.04–2,439.26)318.69 (298.49–340.79)4.28 (2.48–6.65)Notes: Values in parentheses are exported 95% GBD uncertainty intervals. Regions are ordered alphabetically; ranked comparisons are shown in Figure 4.

Regional 1990 estimates, counts, and EAPCs are in Supplementary Tables S1a and S1d.Abbreviations: ASR, age-standardized rate per 100,000; DALYs, disability-adjusted life-years.Figure 4Age-standardized rates across the 21 GBD regions in 2023. (A) incidence, (B) prevalence, and (C) DALYs. All panels use the same regional order, descending by the incidence-rate point estimate.

Points and horizontal intervals are source point estimates and exported 95% UIs, with separate rate scales per 100,000. Rankings do not establish statistically distinguishable regional differences.Three plots compare regional rates per 100,000 for incidence, prevalence and DALYs. Each plot lists 21 regions.

A. Incidence shows Eastern Europe and Central Asia with the highest rates around 2,456 and 2,417, while High-income North America is lowest at approximately 1,994. B.

Prevalence highlights Central Asia with the highest rate near 354 and High-income Asia Pacific lowest at about 305. C. DALYs indicate Central Asia highest at 4.81, Australasia lowest at 4.09.

Regions with higher incidence often show higher prevalence and DALYs, though some variations exist. Each plot uses distinct colors for differentiation. Horizontal intervals represent uncertainty, with some regions showing wider intervals, indicating variability in estimates.Three dot and interval plots show regional incidence, prevalence and DALYs rates per 100,000.

Across the five GBD SDI groups, counts increased in every group, with the largest percentage increases in the Low SDI group (127.7% for incidence, 128.1% for prevalence, and 128.9% for DALYs) and the smallest in the High SDI group (17.0%, 35.7%, and 34.0%, respectively) (Figure 2D–I). These are comparisons among the five supplied SDI groups.

Demographic Decomposition of Count Changes

Population growth was the dominant component of the increase in all three outcomes. Of the 61.85-million increase in incident cases, population growth contributed 62.36 million (100.8%), age-structure change −1.07 million (−1.7%), and age-specific-rate change 0.57 million (0.9%). For prevalence, the corresponding contributions were 8.74 million (85.1%), 1.46 million (14.2%), and 0.065 million (0.6%); for DALYs they were 117,551 (86.2%), 17,456 (12.8%), and 1,297 person-years (1.0%).

The age-structure component was therefore negative for incidence and positive for prevalence and DALYs within this accounting exercise (Table 3; Figure 5A–C; Supplementary Table S2).Table 3Symmetric Decomposition of Changes in Global Modeled Counts Between 1990 and 2023OutcomePopulation GrowthAge-Structure ChangeAge-Specific-Rate ChangeNet ChangeIncidence62,355,522 (100.8%)−1,069,403 (−1.7%)567,057 (0.9%)61,853,176Prevalence8,735,526 (85.1%)1,458,894 (14.2%)64,625 (0.6%)10,259,045DALYs117,551 (86.2%)17,456 (12.8%)1,297 (1.0%)136,304Notes: Cells show a component’s count contribution and its percentage of the net change. Incidence and prevalence contributions are in cases; DALY contributions are in person-years of healthy life lost. These are deterministic point estimates, without uncertainty intervals or tests of significance.

Components were averaged over all six replacement orders of population size, age structure, and age-specific rates. A component can exceed 100% of the net change when another component is negative. Because each cell is independently rounded to the nearest integer, summed components may differ from the corresponding rounded endpoint difference in Table 1 by a negligible amount (here, 1 person-year for DALYs).

Reconstructed population denominators come from corresponding GBD age-specific counts and rates and provide no independent demographic validation.Figure 5Demographic decomposition of global count changes, 1990–2023. (A–C) Symmetric decomposition of incidence, prevalence, and DALYs. Bars show population growth, age-structure change, and age-specific-rate change averaged across all six replacement orders; labels give each component’s percentage of the outcome’s net change.

These are deterministic accounting terms without propagated UIs or causal interpretation. (D) Global population age shares in 1990 and 2023, reconstructed from GBD counts and rates. (E–G) Age-level contributions to the age-structure component for incidence, prevalence, and DALYs, respectively; contributions average the age-specific change at the age-share replacement step across all six orders and sum to the age-structure component in Table 3.

They do not measure the causal effect of aging individuals, and no joint uncertainty intervals are available.The image contains seven panels. Panels A–C show decomposition contributions for incidence, prevalence and DALYs. Population size is the largest positive component in all three outcomes; age structure is negative for incidence and positive for prevalence and DALYs; age-specific-rate contributions are small and positive.

Panel D shows global population age shares in 1990 and 2023 across age groups from less than 5 to 95 plus years. Panels E–G show age-structure contributions by age group. For incidence, the largest negative contribution is at ages 20–24 and the largest positive contribution at ages 45–49.

For prevalence and DALYs, the largest negative contribution is at ages 20–24 and the largest positive contribution at ages 70–74.A multi plot chart showing incidence, prevalence and DALYs decomposition and age group contributions.

The reconstructed population share at ages 0–29 fell from 59.81% to 47.83%, while shares at ages 30–64 and 65 years or older increased from 34.07% to 41.91% and from 6.13% to 10.26%, respectively. The age-level accounting clarifies the negative incidence component. Age-structure contributions from ages 0–29, 30–64, and 65 years or older were approximately −15.28, +12.07, and +2.15 million incident cases, respectively, summing to −1.07 million.

Thus, the net negative component arose from the redistribution of shares across the full age range; it does not mean that the contribution allocated to ages 65 years or older was negative (Figure 5D–G and Supplementary Tables S2a–S2b). Incidence remained negative and prevalence and DALYs remained positive in all six replacement orders and with ten-year age grouping. These sensitivities establish stability to the specified accounting choices, not uncertainty bounds or statistical significance (Supplementary Table S2c).

Demographic Scenarios Through 2040

With age-specific rates fixed at their 2023 levels, the WPP Medium scenario yielded 204.18 million incident cases, 31.76 million prevalent cases, and 423,554 DALYs in 2040. Low–High scenarios yielded 199.07–209.29 million incident cases, 31.08–32.44 million prevalent cases, and 414,192–432,927 DALYs (Figure 6 and Supplementary Table S4e). These spans describe different demographic assumptions, not 95% prediction intervals.

Crude rates vary with scenario age structure even though age-specific rates are fixed. Figure 6Fixed-age-specific-rate demographic scenarios, 2024–2040. (A) incident cases, (B) crude incidence rate, (C) prevalent cases, (D) crude prevalence rate, (E) DALYs, and (F) crude DALY rate.

Each scenario combines fixed 2023 GBD age-specific rates with the age-specific populations of its own WPP 2024 Low, Medium, or High variant. Black points mark the separate 2023 GBD counts or crude rates; scenario curves start in 2024 and are not rescaled to the GBD baseline. Rates are per 100,000 and use each scenario’s current age structure, not a standard population.

Scenario differences are not prediction intervals.Image A: Incidence count (millions) over years. 2023: 184 cases. By 2040: Low 199, Medium 204, High 209.

Image B: Incidence rate per 100,000. 2023: 2,275. By 2040: Low 2,265, Medium 2,225, High 2,200.

Image C: Prevalence count (millions). 2023: 26.8 cases. By 2040: Low 31.1, Medium 31.8, High 32.4.

Image D: Prevalence rate per 100,000. 2023: 332. By 2040: Low 354, Medium 346, High 339.

Image E: DALYs count (thousands). 2023: 359. By 2040: Low 414, Medium 424, High 433.

Image F: DALYs rate per 100,000. 2023: 4.45. By 2040: Low 4.71, Medium 4.62, High 4.52.

Legend: GBD 2023, WPP variant, Low, Medium, High.A set of six line graphs showing incidence, prevalence and DALYs counts and crude rates by year.

The scenario calculation directly uses WPP populations without rescaling them to GBD counts. WPP’s 2023 total population is 0.321% higher than the GBD-reconstructed population; the separate baseline reconciliation in Supplementary Table S4i documents the resulting count differences. Historical fixed-rate and log-linear comparisons are reported in Supplementary Tables S4f and S4h.

For the 2010 cutoff, fixed-rate mean absolute percentage errors were 0.026%, 0.018%, and 0.040% for incidence, prevalence, and DALYs, respectively, compared with 0.214%, 0.151%, and 0.210% for the full-history log-linear model. Fixed rates had lower error than full-history extrapolation at all three cutoffs for all outcomes, but recent ten-year trends performed better for DALYs at two cutoffs. These comparisons support displaying sensitivity to rate assumptions, not a claim of validated future forecasts.

Supplementary Temporal and APC Analyses

Segmented models located the last prevalence and DALY breakpoints near 2017.9 and 2018.8; their conditional 95% intervals included 2020. ITS residuals remained autocorrelated despite AR(1) errors, and only four annual observations followed the 2020 marker. These limitations prevent the analyses from establishing or excluding an effect of the pandemic (Supplementary Figure S1 and Supplementary Tables S3a–S3g).

For APC, the baseline age smooth had a k-index of approximately 0.05 and approached its basis-dimension limit. Worst-case concurvity was 1, and changing age basis dimension materially changed allocation to the period and cohort smooths. Neither the baseline nor larger-basis models supported independent interpretation of the three components (Supplementary Figure S2 and Supplementary Tables S4a–S4d and S4j).

Discussion

This study used GBD 2023 estimates to describe global, regional, sex-specific, age-specific, and SDI-stratified seborrheic dermatitis burden from 1990 to 2023, and applied a Das Gupta decomposition to separate population growth, age-structure change, and age-specific-rate change across incidence, prevalence, and DALYs within a single, internally consistent release. Global counts of incident cases, prevalent cases, and DALYs increased by approximately one-half to two-thirds over this period, while age-standardized rates changed by less than 0.4%; population growth accounted for most of each net increase, and age-structure change contributed in opposite directions across outcomes — negatively for incidence and positively for prevalence and DALYs. These patterns extend earlier GBD descriptions of rising counts alongside stable standardized rates.6,7 Huang et al applied decomposition and projection methods to adults aged 55 years and older using GBD 2021.8 Our contribution is an all-age comparison of incidence, prevalence, and DALYs within a single GBD 2023 series, with age-level accounting and sensitivity checks.

Because age-standardized rates changed so little, the rising case counts reported here reflect a larger patient population rather than a higher per-person risk of developing seborrheic dermatitis; dermatologic service capacity should therefore be sized to population growth and ageing trajectories rather than to an assumed rise in disease risk. The opposite-signed age-structure contribution carries a further planning implication: population ageing reduced the accounted share of the incidence increase but enlarged the accounted share of the prevalence and DALY increases, consistent with a growing prevalent pool at older ages even without a rise in new-case risk, so demand may shift from managing new presentations toward longer-term and recurrent-disease care as populations age. This age-structure term summarizes shifts in all age shares weighted by their corresponding rates — for incidence, negative contributions from younger groups outweighed positive contributions from older groups — so its net negative sign should not be read as a protective effect of ageing.

Similarly, the higher modeled prevalence-to-incidence ratio at older ages is not independent evidence of longer disease duration, because both quantities share GBD modeling inputs and constraints that do not encode the steady-state, recurrence, remission, migration, or ascertainment assumptions a duration interpretation would require.

Several factors further qualify direct interpretation of these estimates. Absolute burden is sensitive to the GBD release itself: Hao and colleagues reported 135.71 million incident cases and an age-standardized incidence rate of 1,704.04 per 100,000 for 2021 using GBD 2021,7 whereas GBD 2023 estimates for the same year are 180.90 million and 2,265.88 — approximately 33.30% and 32.97% higher, a difference that reflects re-estimation between releases rather than disease growth within a calendar year. Case definitions and ascertainment differ further: a systematic review of clinician-diagnosed seborrheic dermatitis reported a pooled prevalence of 4.38% with substantial heterogeneity,19 and its sampling, settings, and ascertainment differ from GBD estimation, so the two measures should not be treated as interchangeable or used to validate one another without harmonization; nor do annual incident and prevalent counts directly measure visits, repeated flares, or treatment workload.

Regional and SDI results likewise provide descriptive context rather than a basis for resource allocation on their own — the largest proportional increases occurred in the Low SDI group, but this analysis did not match dermatology workforce or health-service data to these groups, SDI groups are not equivalent to World Bank income categories, and regional point-estimate rankings are not tests of regional contrasts, especially where UIs overlap. Finally, the deterministic 2040 scenarios show how demographic change alone would alter modeled burden if age-specific rates remained at 2023 levels, and are conditional calculations rather than calibrated probability forecasts; the supplementary temporal and APC analyses add no causal explanation to the main decomposition, since residual dependence, short follow-up after 2020, breakpoint uncertainty, and APC non-identifiability preclude conclusions about a pandemic effect or independent period and cohort effects.

Several limitations, and the directions that follow from them, define the scope of this study and its next steps. Diagnostic and coding practices and capture of mild or self-treated disease may vary across locations and over time; joint posterior draws were unavailable, so uncertainty in endpoint changes, component signs, contribution percentages, and ratios was not propagated, and the sensitivity checks reported here do not substitute for that uncertainty; and geographic analyses were limited to the global aggregate, 21 GBD regions, and five SDI groups. Addressing these constraints points toward several directions for future work: longitudinal, individual-level cohorts that follow incident cases through remission or recurrence would help confirm whether seborrheic dermatitis genuinely persists longer in older patients, a question the cross-sectional GBD exports cannot answer; replicating this decomposition as future GBD releases become available would help distinguish a stable demographic pattern from an artifact of GBD 2023’s specific modeling assumptions; studies that map GBD case definitions onto clinician-diagnosed or self-reported cohorts would help translate these modeled estimates into service-utilization terms; pairing regional and SDI burden estimates with local workforce and service-capacity data — particularly in Low SDI settings, where the proportional increase was largest — would translate this accounting into staffing and service-planning guidance; and extending the deterministic 2040 scenarios with joint probabilistic uncertainty in demographic and rate assumptions would turn them into decision-relevant forecasts.

Conclusions

Within GBD 2023 point estimates, population growth was the largest accounting component of the increase in global seborrheic dermatitis burden from 1990 to 2023, while age-structure change contributed negatively to incident cases but positively to prevalent cases and DALYs. Because age-standardized rates changed little over this period, rising case counts reflect a growing patient population rather than a rising per-person risk, and population ageing is shifting burden toward longer-term, recurrent-disease management even where new-case incidence is not increasing. Disease surveillance and health-service planning for seborrheic dermatitis should therefore report absolute counts alongside age-standardized rates and anticipate growing chronic-management demand as populations age; these descriptive, model-based results support that planning use but do not identify individual disease duration or biological mechanisms.

Related articles

PubMed

Association of Serum IgE and Peripheral Eosinophilia with Seborrheic Dermatitis: Evidence from a Case-Control Study.

Introduction It is thought that some shared mechanisms may play a role in the pathophysiology of seborrheic dermatitis (SD) and atopic conditions, both of which are chronic, inflammatory, and recurrent in nature. Objective This study aimed to investigate the presence of atopic diseases in SD patients without atopic dermatitis or other inflammatory skin disorders and to assess their serum IgE levels and peripheral eosinophil values as a marker of atopy. Methods This prospective cross-section...

PubMed

Psychological Differences Between Seborrheic Dermatitis Patients and Healthy Controls.

Introduction Seborrheic dermatitis (SD) is associated with significant psychological distress, including elevated anxiety and depression and quality-of-life impairment [1]. Psychological stress is frequently reported by patients as a precipitating factor for flareups, with symptom intensity often worsening during stressful periods [2]. While difficulties in emotion regulation have been linked to stress-related inflammatory processes in SD [3], trait anger and anger expression remain understudie...