The Price of Absence: Productivity Costs and Sickness-Absence Episodes Attributable to Mental and Behavioral Disorders in the Brazilian Federal Workforce — A Nationwide Time-Series Study (2012–2024) Authors: Douglas Paschoal dos Santos, Alfredo Chaoubah, Helian Nunes de Oliveira, Quéren Hapuque de Carvalho, João Henrique Gonçalves Mazzeu Published: 2026 ABSTRACT BackgroundMental and behavioral disorders (MDs) represent a leading cause of sickness absence among Brazilian public servants, generating substantial social and economic burdens. Despite their impact, comprehensive nationwide assessments of productivity costs attributable to MDs over extended periods remain scarce.ObjectivesThe aim of this study was to estimate the productivity costs associated with MDs (ICD-10 Chapter V) among Brazilian federal civil servants from 2012 to 2024 compared with the four other most prevalent non-mental causes of absence.MethodsWe conducted a retrospective longitudinal study using administrative data from SIAPE-SAÚDE (n = 471,507 workers/year average). Outcomes were classified according to ICD-10, focusing on MDs (F00-F99) and comparing them with the four other most prevalent non-mental causes. Productivity costs were estimated using the Human Capital Approach, calculating lost workdays (22 days per 30-day leave) multiplied by daily remuneration. Annual episode rates, average leave duration, and mean costs were stratified by region, education, gender, and age. All monetary values were adjusted to December 2024.ResultsMDs accounted for 3.94 billion Brazilian Real (BRL) (38% of total costs among leading causes), representing 13.9 million days lost. Costs were concentrated in mood disorders (F30–F39: BRL 2.04 billion) and neurotic/stress-related disorders (F40–F48: BRL 1.49 billion). Episode rates increased from 42.3 to 83.8 per 1000 employees (2012–2019), declined in 2020 (39.4), then recovered to 85.8 in 2023. Women showed higher episode rates (particularly those aged ≤ 30 years: ≈ 160 per 1000), while men exhibited longer absences (44.7 vs 37.5 days) and higher per-case costs (BRL 11,720 vs BRL 8905). Marked increases occurred in autism spectrum disorders (+ 1673%) and attention-deficit/hyperkinetic disorder (ADHD) (+ 331%).ConclusionsMDs impose a substantial and persistent burden of productivity costs on the Brazilian federal public service, with rising sickness-leave episode rates over the period and pronounced gender, educational, and regional disparities, requiring targeted, equity-oriented occupational health policies.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s41669-026-00673-8. Key Points for Decision Makers Mental health conditions were the largest single cause of productivity loss from sickness absence in the Brazilian federal workforce (2012–2024), costing about BRL 3.9 billion (≈ USD 636 million) and, unlike other major causes, did not decline over the period.The burden fell unevenly across the workforce: women (especially those aged under 40 years) took leave more often, men had longer and costlier absences, and workers with intermediate education carried disproportionately high absence rates;Embedding routine monitoring of mental health absence and its costs into occupational health services, alongside early detection and return-to-work support, could reduce these losses while protecting worker wellbeing. Mental health conditions were the largest single cause of productivity loss from sickness absence in the Brazilian federal workforce (2012–2024), costing about BRL 3.9 billion (≈ USD 636 million) and, unlike other major causes, did not decline over the period.The burden fell unevenly across the workforce: women (especially those aged under 40 years) took leave more often, men had longer and costlier absences, and workers with intermediate education carried disproportionately high absence rates;Embedding routine monitoring of mental health absence and its costs into occupational health services, alongside early detection and return-to-work support, could reduce these losses while protecting worker wellbeing. Introduction Mental and behavioral disorders (MDs) are a leading cause of staff absence due to sickness and productivity loss among Brazilian public servants, with significant social and economic repercussions. Studies across different states have consistently shown that MDs account for a substantial share of all medical leave and are particularly associated with prolonged sick leave and early retirement [1–4]. The growing amount of psychiatric sick leave among public servants has been associated with multiple interrelated factors, including escalated work, psychosocial stressors, organizational instability, and deteriorating working conditions, particularly in education and health sectors, where emotional demands and role overload are prevalent [4, 5]. Therefore, this condition represents not only a major source of individual suffering but also a significant and growing socioeconomic burden. The Brazilian social security framework is based on constitutional principles and ensures universal healthcare, social security, and social assistance [6]. Federal law provides civil servants with a dedicated social security system for support in cases of illness or incapacity [7, 8]. Through these structures, the public administration can diagnose and monitor patterns of work-related morbidity, enabling a better understanding of the leading causes of work incapacity, particularly those related to mental health, and informing preventive actions beyond mere sickness certification [9]. MDs rank among the leading causes of disability-adjusted life years (DALYs) and years of life lost (YLL), particularly within the working-age population [10]. Their growing prevalence entails substantial productivity losses and escalating economic costs [11–16]. In 2020 alone, these conditions generated an estimated annual expenditure of 7.4 billion Brazilian Real (BRL) (≈ USD 1.37 billion) for the Brazilian social security system. They ranked as the fourth leading cause of disability retirement and the third of temporary sick leave [17]. Among Brazilian public servants, the share of sickness leave due to MDs ranged from about 23 to 52%. Within mental causes, depression/mood disorders were the leading cause of mental-health leave. Anxiety/stress-related disorders were typically next, often comprising the majority (≈ 70%) of mental-health leave [1, 2, 4, 18, 19]. Only a subset of studies quantified indirect costs; in a nationwide federal series (2012–2017), MDs accounted for BRL 1.89 billion in indirect costs. The per-worker annual indirect cost increased from BRL 596.48 to BRL 737.76 (value adjusted to December 2018). Among diagnoses, mood disorders drove about 55% and anxiety/stress-related disorders contributed about 32% of total indirect costs [4]. In another sample, 4524 lost workdays corresponded to an estimated payroll-equivalent indirect cost of about BRL 0.90 million over about 21 months from 2011 to 2012 [19]. Furthermore, it remains unclear whether Brazilian public-sector workers take more sick leave than other workers, although this pattern is observed elsewhere. For example, according to Whittaker et al. [20], in the United Kingdom, public-sector workers have higher overall sickness-leave rates than private-sector employees, and public-sector employment independently increases the odds of taking sickness leave. Moreover, screening positive for a probable common MD is strongly associated with absence across all sectors, with a larger effect in the public sector. Brazilian studies on the productivity costs of MDs among public servants remain scarce [4, 18, 19], despite their economic relevance to the federal civil servants' social-protection system. Assessing this burden is crucial for evidence-based occupational health policies. In this context, this research aimed to estimate the productivity costs associated with MDs (ICD-10, Chapter V) among Brazilian federal public servants, focusing on work leave, considering the period from January 2012 to December 2024 and comparing them to other major causes of absence (ICD-10 S, Z, M, and C). Methods We conducted an observational, exploratory, retrospective longitudinal study of the productivity costs associated with MDs among federal civil servants. We extracted sickness-absence data from SIAPE-SAÚDE via the DW-SIASS data warehouse for January 2012 to December 2024 (Fig. 1). Denominators and remuneration profiles were supplemented with data from the Federal Government Transparency Portal (FGTP) and the Federal Personnel Statistical Panel (FPSP), using end-of-year population totals for 2012–2024. We excluded outsourced and temporary workers (under the General Social Security Regime). Administrative data were obtained from the Brazilian federal government upon formal request. These data are not publicly available due to confidentiality restrictions.Fig. 1Study population selection. Source: Prepared by the authors Study population selection. Source: Prepared by the authors No sample size calculation was performed, as the study included the entire eligible population of federal civil servants under the statutory regime during the study period. The analytic population averaged 471,507 workers per year (range: 436,920–499,322). Women comprised 45.2% and men comprised 54.8%. Age distribution was concentrated in mid-career strata: 31–40 years (27.9%), 41–50 years (26.7%), and 51–60 years (26.5%), with smaller proportions ≥ 60 years (11.9%) and ≤ 30 years (6.8%). Geographically, workers were distributed across all macro-regions, comprising the Southeast (35.5%), Northeast (23.0%), Center-West (16.0%), South (13.7%), and North (11.8%). Educational attainment was predominantly high, with almost 60% having completed higher education. The SIAPE-Saúde database includes both sick leave for the employee’s own health condition and leave granted to care for an ill family member, which we quantified separately. Because the ICD-10 code refers to the certified medical condition and not necessarily to the worker, results should be interpreted as causes of leave rather than strictly as morbidity of the federal workforce. We classified outcomes according to ICD-10, focusing on MDs, Chapter V (F00-F99), and, for comparison, the other four most prevalent non-mental causes of leave, an aggregated group including ICD-10 S (External causes), Z (Health status–related factors), M (Musculoskeletal diseases) and C (Neoplasms). Productivity costs were estimated using the Human Capital Approach (HCA). Daily remuneration equals the average gross monthly wage divided by 30. Lost workdays were computed as 22 days for every 30 days of leave [2, 21, 22]. The productivity cost for each leave spell was calculated as lost workdays times daily remuneration:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Productivity cost=\left(\frac{Leave days\times 22}{30}\right)\times \frac{Adjusted monthly remuneration}{30}.$$\end{document}Productivitycost=Leavedays×2230×Adjustedmonthlyremuneration30. For example, a 30-day leave with an adjusted monthly remuneration of BRL 6600 corresponds to 22 lost workdays (30 × 22/30) and a daily remuneration of BRL 220.00 (6600/30), resulting in a productivity cost of BRL 4840 for that spell. Remuneration was available only as salary bands. We used the median remuneration of each band as a proxy for individual remuneration in each leave episode. Median remuneration was used to minimize the influence of extreme values and better represent the central tendency of income among public servants. The median remuneration values were obtained directly from the SIAPE-SAÚDE database, except for those above BRL 8500, which were not specified and were instead calculated based on the annual median remuneration of civil servants earning above this threshold, as provided by the FGTP. Annual episode rates of sickness leave due to MDs were calculated per 1000 employees. The numerator was the number of new episodes recorded in each calendar year. The denominator was the total number of active employees in the same period. The same employee could contribute more than one episode within a year. Episode rates were stratified by region, education, gender, and age. In addition, the average number of days of leave granted and the mean cost per employee were calculated for each stratum and year. This allowed assessment of both the frequency and economic impact of MD–related absenteeism. Descriptive analyses included frequencies, means, and rates stratified by region, gender, age, and education. We reviewed temporal trends by visually inspecting time series and percentage changes from 2012 to 2024. We performed data management and statistical analyses in R (version 4.3.3) and Microsoft Excel 365. All monetary values were converted to December 2024 BRL using the National Consumer Price Index (INPC), applying year-specific December-to-December deflators \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left(\mathrm{I}\mathrm{N}{\mathrm{P}\mathrm{C}}_{Dec/2024}/{\mathrm{INPC}}_{Dec/\mathrm{y}}\right)$$\end{document}INPCDec/2024/INPCDec/y. Salaries refer to the remuneration recorded at leave onset and were adjusted using the same approach. For international comparability, the main aggregate costs are also reported in US dollars, converted at the exchange rate of 31 December 2024 (BRL 6.1923/USD). Results Among the five leading causes of sick leave for federal public servants, MDs (ICD-10 F) accounted for 13.9 million days of absence and BRL 3.94 billion (≈ USD 636 million) in costs (38% of the total), of which leave granted to care for an ill family member represented BRL 74.5 million (1.89%). Musculoskeletal diseases (ICD-10 M) totaled 9.0 million days and BRL 2.56 billion (25%), external causes (ICD-10 S) represented 4.8 million days and BRL 1.41 billion (14%), neoplasms (ICD-10 C) contributed 4.2 million days and BRL 1.29 billion (13%), and health status–related factors (ICD-10 Z) accounted for 3.6 million days and BRL 1.06 billion (10%). Altogether, these conditions generated 35.6 million days of absence and BRL 10.25 billion (≈ USD 1.66 billion) in costs. Productivity costs due to MDs were the highest among the groups analyzed throughout all years, ranging from BRL 197.5 million in 2012 to a peak of BRL 348.4 million in 2023. Musculoskeletal diseases were the second largest burden, ranging from BRL 178.4 million in 2012 to a peak of BRL 269.9 million in 2014, followed by external causes, with values from BRL 57.1 million in 2020 to BRL 145.7 million in 2017. Neoplasms showed costs ranging from BRL 63.3 million in 2020 to BRL 126.2 million in 2017, while factors influencing health status ranged from BRL 34.6 million in 2020 to BRL 150.5 million in 2015. All diagnostic groups showed a reduction in 2020, with partial recovery in subsequent years. Productivity costs declined across all groups except MDs, which remained comparatively stable with slight increases (Fig. 2).Fig. 2Annual productivity cost in Brazilian Real (BRL) for Mental and behavioral disorders (ICD-10 F), External causes (ICD-10 S), Musculoskeletal diseases (ICD-10 M), Neoplasms (ICD-10 C), Health status–related factors (ICD-10 Z), 2012–2024. Source: Prepared by the authors Annual productivity cost in Brazilian Real (BRL) for Mental and behavioral disorders (ICD-10 F), External causes (ICD-10 S), Musculoskeletal diseases (ICD-10 M), Neoplasms (ICD-10 C), Health status–related factors (ICD-10 Z), 2012–2024. Source: Prepared by the authors The overall episode rate of sickness leave episodes due to MDs increased from 42.3 per 1000 employees in 2012 to 83.8 per 1000 in 2019. A reduction occurred in 2020 (39.4 per 1000), followed by a recovery and a new peak in 2023 (85.8 per 1000). MDs evidenced higher average productivity costs and longer sick leave durations per case compared with other causes aggregated (ICD-10 S, Z, M, and C). For MDs, average costs ranged from BRL 8229.29 (28.78 days) in 2019 to BRL 13,404.18 (51.30 days) in 2021, whereas ICD-10 S, Z, M, and C showed costs ranging from BRL 4423.71 (15.44 days) in 2019 to BRL 6976.50 (27.07 days) in 2021. Both groups exhibited a peak in 2020–2021, followed by a decline and relative stabilization in subsequent years (Fig. 3).Fig. 3Correlation between productivity costs and average sick leave duration per case: ICD-10 F versus aggregated category (ICD-10 S, Z, M, and C), 2012–2024. BRL Brazilian Real, Source: Prepared by the authors Correlation between productivity costs and average sick leave duration per case: ICD-10 F versus aggregated category (ICD-10 S, Z, M, and C), 2012–2024. BRL Brazilian Real, Source: Prepared by the authors When stratified by income, sick leave due to MDs was heterogeneously distributed across workers’ wage ranges. The highest number of cases occurred among those earning 10–15 minimum wages (n = 145,765), followed by the group earning three to six minimum wages (n = 143,670). The average duration of leave ranged from 28.17 days (up to two minimum wages) to 36.09 days (two to three minimum wages), with relative stability across the categories. Regarding economic impact, the average cost per case showed a progressive increase according to wage level, rising from BRL 929.58 among workers earning up to two minimum wages to BRL 19,209.12 in the 15–25 minimum-wages group. The analysis by age group revealed a progressive increase in the average cost per case of sick leave due to MDs. The lowest mean cost occurred among workers aged 15–19 years (BRL 2045.00), while the highest was found in individuals aged 65 years or older (BRL 14,236.00). The number of episodes of sick leave peaked in the 35–39 years’ range (n = 66,495) and gradually declined in older age groups, despite the rising costs. Examining average adjusted remuneration by age group among those who were granted sick leave due to MDs, an increase was observed from BRL 4707.00 among workers aged 15–19 years to BRL 12,776.00 in those aged 60–64 years, followed by a decline in those aged 65 years or older (BRL 12,217.00). The distribution of costs by age group showed distinct patterns when comparing MDs with other causes aggregated. For MDs, costs were concentrated between 30 and 59 years, peaking at 40–44 years (15.31%) and 50–54 years (16.84%). In contrast, ICD-10 S, Z, M, and C categories reached their highest burden in later ages, particularly 50–54 years (17.28%) and 55–59 years (19.33%). While both groups displayed minimal cost participation at younger ages (< 25 years) and in older age groups (> 70 years), MDs showed earlier peaks, whereas ICD-10 S, Z, M, and C maintained higher percentages into older working ages (Fig. 4).Fig. 4Proportional cost (%) by age group for ICD-10 F versus aggregated category (ICD-10 S, Z, M, and C), 2012–2024. Source: Prepared by the authors Proportional cost (%) by age group for ICD-10 F versus aggregated category (ICD-10 S, Z, M, and C), 2012–2024. Source: Prepared by the authors Additionally, remuneration patterns were examined, and the average adjusted wage per year revealed similar patterns between MDs and the comparison group (ICD-10 S, Z, M, and C). In both groups, mean remuneration peaked in 2014, reaching BRL 13,330.33 for MDs and BRL 13,149.54 for ICD-10 S, Z, M, and C, followed by a gradual decline in subsequent years. The lowest values were observed in 2022, with BRL 9589.41 for ICD-10 F and BRL 9566.48 for ICD-10 S, Z, M, and C. Over the entire period (2012–2024), the averages were nearly identical between groups, at BRL 11,599.19 for ICD-10 F and BRL 11,618.62 for ICD-10 S, Z, M, and C (Fig. 5).Fig. 5Average inflation-adjusted remuneration (BRL) for ICD-10 F versus aggregated category (ICD-10 S, Z, M, and C), 2012–2024. BRL Brazilian Real. Source: Prepared by the authors Average inflation-adjusted remuneration (BRL) for ICD-10 F versus aggregated category (ICD-10 S, Z, M, and C), 2012–2024. BRL Brazilian Real. Source: Prepared by the authors Gender-stratified analysis revealed differences in both frequency and economic impact of MD-related absences. Women represented the largest share of the burden, with 57% of costs and 8.2 million days for ICD-10 F (compared with 56% of costs and 12.5 million days for ICD-10 S, Z, M, and C), with gender information unavailable for only 3% of records. Moreover, women accounted for the highest number of cases (214,941 leave episodes; average 37.5 days), while men had fewer cases (115,208) but longer absences (average 44.7 days). Male workers also displayed a higher mean expenditure per case (BRL 11,720 vs BRL 8905), consistent with both longer sick leave duration and higher average remuneration (BRL 12,356 vs BRL 11,261). Women exhibited substantially higher episode rates than men due to MDs, with the steepest rates among younger female groups, reaching ≈ 160 and ≈ 148 per 1000 employees in 2023, respectively, for those aged ≤ 30 and 31–40 years. Male peaks were much lower (≈ 65–70 per 1000 in the highest years). Although the 2019–2020 decline affected both genders, the post-2020 rebound was more pronounced among women, particularly younger ones. A higher average number of days lost was also identified among women who registered approximately 2.1 days of absence per employee per year, compared with 1.6 days among men. This difference was reflected in the average cost per employee, estimated at BRL 582.00 for women and BRL 511.00 for men (Fig. 6).Fig. 6Episode rate of mental and behavioral disorders (MDs) by gender and age group, 2012–2024. Source: Prepared by the authors Episode rate of mental and behavioral disorders (MDs) by gender and age group, 2012–2024. Source: Prepared by the authors The costs related to sick leave due to MDs were heterogeneous across diagnostic categories. The greatest impact was observed in mood disorders (F30–F39: 180,806 cases; BRL 2.04 billion) and neurotic/stress-related disorders (F40–F48: 187,298 cases; BRL 1.49 billion). These two groups accounted for 91% of all cases and nearly 89.5% of the total costs, while the remaining groups showed fewer cases and lower associated costs. The ICD-10 categories F80–F89 (specific developmental disorders) and F90–F99 (behavioral and emotional disorders with onset usually in childhood and adolescence) displayed the largest increases between 2012 and 2024, with cost growth rates of 1449% and 203%, respectively. Marked increases were also noted for F40–F48 (neurotic, stress-related, and somatoform disorders) with rises of 150%. In contrast, F70–F79 (mental retardation) exhibited the highest reductions costs (–85%). Among the developmental and behavioral groups, the most remarkable increases were observed in F84 (pervasive developmental disorders) and F90 (hyperkinetic disorders), with cost growth rates of 1673% and 331%, respectively. Both the number of requests for sick leave and the total number of leave days granted due to autism spectrum disorders (F84) and attention-deficit/hyperkinetic disorder (F90) showed an increase over the period. The largest increase was for F84, with higher levels after 2020, reaching the highest levels in 2024. In contrast, F90 increased more gradually over the same period (Fig. 7).Fig. 7Number of sick leave days granted due to medical certificates and expert examinations for Autism (ICD-10 F84) and ADHD (ICD-10 F90), 2012–2024. ADHD attention-deficit/hyperkinetic disorder. Source: Prepared by the authors Number of sick leave days granted due to medical certificates and expert examinations for Autism (ICD-10 F84) and ADHD (ICD-10 F90), 2012–2024. ADHD attention-deficit/hyperkinetic disorder. Source: Prepared by the authors Regarding the distribution of costs by region of the country for MDs, the Southeast accounted for the largest share of total expenditures (BRL 1,186,212,321; 36.1%), followed by the Northeast (BRL 682,112,619; 20.8%) and the Midwest (BRL 615,395,748; 18.7%). The South and North contributed lower proportions, corresponding to 13.2% (BRL 433,976,619) and 11.2% (BRL 368,980,323), respectively. Notably, state-level data (‘UF Servidor’) were compromised for the 2012–2014 period, particularly in 2013, when only one case contained this information. In subsequent years, missing values were negligible (max: 126 cases in 2019). Across macro-regions, episode rates rose until 2019, with the Midwest consistently showing the highest levels, peaking at approximately 118.7 per 1000 in 2019, dropping to 46.5 per 1000 in 2020, and rising again to nearly 100 per 1000 in 2023. The Midwest also exhibited the highest mean cost per employee (≈ BRL 771) and the longest average absence duration (≈ 2.6 days per employee per year), contrasting with the Northeast, which showed the lowest figures for both indicators (≈ BRL 598 and ≈ 2.1 days, respectively). In terms of educational attainment, the NI group (intermediate education; completed high school) consistently exhibited higher episode rates of sickness leave due to MDs than the NS group (higher education) across the entire series (mean episode rate: NI = 80.4 per 1000; NS = 51.0 per 1000). This educational gradient was reflected in a greater average number of days of absence among workers with intermediate education (2.7 vs 1.9 days per employee), although there were no corresponding differences in economic impact, as the mean cost per employee was higher among workers with higher education (BRL 715 for NS vs BRL 529 for NI). Discussion This study investigated the magnitude, composition, and time trends of sickness leave–related productivity costs due to MDs among Brazilian federal public servants from 2012 to 2024. MDs (ICD-10 F) comprise the largest productivity cost burden, accounting for approximately BRL 3.94 billion (≈ 38% of total costs among leading chapters) and 13.9 million days lost. Costs were highly concentrated in mood disorders (F30–F39) and neurotic, stress-related disorders (F40–F48), which together comprised the great majority of MD spending. Notable gender differences emerged; although women contributed a larger share of total MD costs, men exhibited longer mean durations and higher mean productivity cost per case. Compared with other health conditions, MD costs were persistently higher on a per-case basis and remained relatively stable, with a pandemic-era dip in 2020 followed by recovery, while other chapters tended to decline. Our findings corroborate and amplify previous Brazilian studies showing that MDs are the primary cause of absence among public servants, with similar proportions and predominance of mood and anxiety disorders [1–4, 18, 19]. However, since none of these studies systematically investigated the national economic burden over an extended period, our study represents the first comprehensive assessment of productivity costs attributable to MDs across the entire Brazilian federal workforce spanning more than a decade. The pronounced gender differences observed deserve careful interpretation. Brazilian research has consistently documented higher absence rates among women for most mental health categories, with the notable exception of substance use disorders, which show male predominance [3, 5]. International studies similarly report elevated sickness leave rates among women across developed countries, with female workers evidencing higher incidence rates [20], particularly in younger age groups [16]. Our findings extend this pattern, revealing that women aged ≤30 and 31–40 years exhibited the steepest episode rates, especially in recent years (≈ 160 and ≈ 148 per 1000 in 2023, respectively), substantially exceeding male peaks. However, men displayed longer average absence duration (44.7 vs 37.5 days) and higher mean expenditure per case (BRL 11,720 vs BRL 8905). Previous Brazilian research found that substance use disorders among public servants were associated with longer absence duration per case, which aligns with our observation of greater male involvement in these conditions and their extended leave periods [18]. The higher per-case costs among men also reflect their higher average remuneration (BRL 12,356 vs BRL 11,261), underscoring how both clinical and socioeconomic factors interact to shape the economic burden disparities between genders. The literature has extensively corroborated that men face stigma and barriers that delay their pursuit of psychiatric help, resulting in more severe conditions and longer work absences, which may partially explain our results [23]. On the other hand, women experience more frequent absences due to stress related to domestic demands, maternal responsibilities, and the burden of their dual workload, which may also explain the higher frequency of sick leave absences seen in our study [24]. We should underscore that other authors already found an association between all-cause sickness absence, including MDs among public sector employees [25], and future risk for developing common MDs after some years. Given our observation of earlier onset of MD-related absence among younger workers, particularly women under 30 years, it is reasonable to anticipate that future cohorts may experience even higher MD prevalence rates and, consequently, more absence due to this cause. Lavikainen et al. [16] showed that, especially among youngest age group, MDs were the most common condition causing burden on productivity-adjusted life years, while other chronic health conditions affect older groups, which aligns with our study and shows the MDs affecting younger age groups, while the other four health conditions increase their impact with age. The wage gradient in MD-related costs reveals important equity considerations. While the highest case frequency occurred among workers earning three to six minimum wages and those earning 10–15 minimum wages, the mean cost per case rose progressively with income level, from BRL 929.58 (up to two minimum wages) to BRL 19,209.12 (15–25 minimum wages). This pattern reflects both the longer absence durations in higher-paid positions and the HCA methodology, wherein productivity losses are proportional to earnings. From a policy perspective, this gradient suggests that MDs impose significant burdens across all income strata, though the economic valuation differs substantially. Educational level emerged as a significant factor, with workers having intermediate education (completed high school) exhibiting consistently higher episode rates (mean 80.4 per 1000) compared with those with higher education (51.0 per 1000). This educational gradient was reflected in greater average absence days among intermediate-education workers (2.7 vs 1.9 days per employee), although paradoxically, mean cost per employee was higher among those with higher education (BRL 715 vs BRL 529), again reflecting remuneration differences. International evidence similarly shows that lower occupational classifications are associated with increased sickness absence [20], suggesting that educational and occupational hierarchies intersect with health outcomes through multiple pathways including work conditions, psychosocial stressors, and access to resources. Regional heterogeneity was marked, with the Southeast accounting for the largest absolute costs. However, when examining episode rates and per-employee metrics, the Midwest consistently showed the highest burden. These regional disparities may reflect differences in workforce composition—particularly considering that the largest proportion of these workers are located in the Southeast—availability of mental health services, occupational health infrastructure, and local socioeconomic conditions. The overall episode rate showed an increase from 42.3 per 1000 employees (2012) to 83.8 per 1000 (2019), a near-doubling over 7 years. The reduction to 39.4 per 1000 in 2020, followed by rapid recovery and a new peak in 2023 (85.8 per 1000), clearly reflects pandemic-related disruptions. Multiple, well-documented events likely contributed to the 2020 decline: reduced access to healthcare services and medical certification during lockdowns, fear of seeking in-person care, shifts to remote work that may have temporarily accommodated certain mental health needs, and administrative interruption in leave processing [26, 27]. The subsequent rebound suggests pent-up demand and worsening mental health conditions due to pandemic-related stressors. Notably, while costs for musculoskeletal diseases, external causes, neoplasms, and health status factors showed downward trends across the series, MD costs remained relatively stable with slight increases, interrupted only by the pandemic dip, a pattern that had already been verified in groups of workers in the United Kingdom [28]. Findings from a study of Japanese public servants demonstrated that MD-related long-term sickness absence was persistent despite some declining trends and demonstrated longer duration than all physical disorders except circulatory system diseases, with mood disorders consistently representing the most prevalent diagnosis [29]. This persistence underscores the chronic and recurrent nature of mental health conditions and suggests that systemic factors driving MD burden have not been adequately addressed over the past decade. This contrasts with other conditions, potentially indicating improved management of physical health issues while mental health challenges remain insufficiently controlled. Studies from Sweden similarly found elevated sickness absence risk among public-sector workers, particularly for common mental disorders, with manual workers in the public sector showing especially increased risk [30]. While our study did not directly compare Brazilian public and private sectors, the patterns observed are consistent with international evidence suggesting that public-sector work environments may present unique mental health challenges related to bureaucratic structures, resource constraints, high public accountability, and limited organizational flexibility. The marked increases in costs associated with specific developmental disorders (F80–F89: + 1449%) and behavioral disorders with childhood onset (F90–F99: + 203%), particularly autism spectrum disorders (F84: + 1673%) and attention-deficit/hyperkinetic disorder (ADHD) (F90: + 331%), represent a striking epidemiological shift. This trend likely reflects multiple converging factors: increased clinical recognition and diagnostic capacity; reduced stigma facilitating help-seeking; broader diagnostic criteria or even diagnostic errors related to media and social media influence, and a less likely genuine increase in prevalence potentially linked to environmental, social, or occupational changes. Similarly, Zaleski et al. [31] found rising recognition and disproportionately high healthcare utilization and expenditures among individuals with ASD and ADHD, supporting the interpretation of shifting diagnostic and care patterns rather than purely prevalence changes. The increase of F84-related absences after 2020 may also relate to pandemic-induced disruptions that disproportionately affected individuals with autism and related conditions. From a workforce perspective, these findings could indicate that federal public service is increasingly accommodating neurodevelopmental diversity, which is positive from an inclusion standpoint. However, the rising absence rates also suggest potential gaps in workplace accommodations, support systems, and management practices tailored to the needs of neurodivergent employees. Our findings have direct implications for occupational health policy. First, the sustained rise in mental health–related leave and its costs support institutionalizing surveillance within the federal public servants’ health care structure, incorporating standardized indicators of incidence, duration, and productivity costs to guide resource allocation. Second, because mood and anxiety disorders accounted for most absences and costs, interventions should prioritize early identification, stepped care, and structured return-to-work pathways coordinating occupational health, primary care, and specialized services. Finally, the rising certification of neurodevelopmental conditions (F84/F90) highlights the need for workplace accommodation and manager training. All strategies should address the documented disparities by gender, education, and region. Limitations Several limitations must be acknowledged. First, administrative data depend on medical certification practices, which may vary over time and across regions, potentially introducing ascertainment bias. Temporal increases in certain categories (e.g., F84, F90) may reflect changes in recognition practices rather than true prevalence shifts. Under-reporting is likely, as not all mental health episodes result in formal sick leave, particularly given persistent stigma surrounding mental illness. Given that under-reporting underestimates the true economic burden, cross-study and cross-country comparisons should be interpreted with caution. Moreover, selection bias may affect our findings since severely affected individuals who departed through early retirement or resignation no longer appear in the administrative records. Second, the Human Capital Approach, while standard [14, 15], values productivity losses based on earnings, which raises equity concerns by assigning higher economic value to higher-paid workers’ health. Alternative methods such as friction cost approaches or quality-adjusted life years may provide complementary perspectives. Moreover, our assumption of 22 working days for every 30 days of leave [2, 21, 22] does not capture regional or annual holiday variation, introducing minor imprecision. Third, we lacked individual-level data on occupational exposures, work organization, and psychosocial factors, limiting causal inference regarding determinants of MD burden. Finally, as noted earlier, we did not conduct a comparison with private-sector workers, leaving open the question of whether Brazilian public servants experience higher MD-related absence rates than other worker populations. Conclusions MDs impose a substantial and persistent economic burden on the Brazilian federal public service, with an estimated BRL 3.94 billion (≈ USD 636 million) in productivity costs between 2012 and 2024. This burden is largely driven by mood and anxiety disorders, while the growth of neurodevelopmental disorder-related leave suggests shifting epidemiological patterns and emerging demands on occupational health policy. It is crucial to emphasize that absenteeism should not be interpreted as unproductivity, but may reflect a health indicator of organizational distress and a warning sign of systemic imbalance. Overall, our findings highlight pronounced and intersecting inequalities in the burden and economic impact of mental disorders across gender, income, education, and regions. Women consistently showed higher rates of sickness leave episodes, while men showed a higher concentration of longer episodes and higher per-case costs, reflecting both clinical profiles and wage structures. Socioeconomic and educational gradients indicated that workers in intermediate occupational positions experience a disproportionate volume of leave, whereas higher-paid and more highly educated groups generate greater monetary losses per episode and per employee. Marked regional heterogeneity also emerged, with some areas showing a concentration of the highest absolute expenditures and others bearing the greatest relative burden in terms of episodes and days lost. Taken together, these patterns underscore the need for targeted, equity-oriented policies and for strengthening the occupational health infrastructure to integrate productivity cost indicators into routine surveillance and guide more effective prevention and care strategies. The policy infrastructure is pivotal for surveillance, prevention, and clinical pathways, as already pointed out by other authors [32, 33], and permits the embedding of metrics, like productivity costs, into routine monitoring. It may therefore represent not only a public health priority but also a macroeconomically rational strategy to preserve the productive capacity of the federal workforce. Future research should compare mental disorder–related burden across public, private, and mixed employment regimes. It should also integrate administrative data with individual-level occupational and clinical information to identify modifiable risk factors and use longitudinal designs to clarify causal pathways between work organization, mental health outcomes, and absence trajectories. Economic evaluations of preventive interventions, together with qualitative studies on workers’ and managers’ perceptions of mental health, stigma, and organizational support, will be essential to guide contextually adapted, evidence-based strategies. The economic magnitude documented in this study provides a compelling rationale for such investments, which could reduce productivity losses while strengthening wellbeing, dignity, and productive capacity across the federal workforce. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 18 KB)Supplementary file2 (PDF 36 KB) Supplementary file1 (DOCX 18 KB) Supplementary file2 (PDF 36 KB)