Skip to main content
Cambridge Prisms: Global Mental Health logoLink to Cambridge Prisms: Global Mental Health
. 2026 Jan 19;13:e11. doi: 10.1017/gmh.2025.10104

Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran

Sohrab Amiri 1,✉, Jannat Mashayekhi 1
PMCID: PMC12835953  PMID: 41608309

Abstract

This study focuses on the national and subnational estimation of prevalence, incidence, disability-adjusted life years (DALYs) related to self-harm and suicide mortality in Iran. These indicators of disease burden were analyzed over the period from 1990 to 2021, with stratifications based on sex, age and geographic location. Additionally, the percentage change observed between 1990 and 2021 was documented. The age-standardized prevalence rate (per 100,000) of self-harm decreased from 173.92 (95% UI: 146.13–208.75) in 1990 to 131.2 (95% UI: 110.55–156.67) in 2021, reflecting a percentage change of −0.25% over the period. In terms of self-harm prevalence in 2021, males had a higher rate (137.62 per 100,000) compared to females (124.82 per 100,000). The findings of the current study revealed that, despite significant challenges such as demographic shifts, economic instability and the impacts of war, the trends in self-harm incidents and suicide mortality rates in Iran have generally been on the decline. Additionally, it was observed that suicide-related deaths were more prevalent among males when compared to females. However, when examining self-harm behaviors over previous decades, these acts appeared to be more frequent among females.

Keywords: prevalence, incidence, self-harm, suicide mortality, global burden of disease

Impact statement

One of the important factors in self-harm and suicide mortality is geographical differences, which lead to cultural and religious differences. Consequently, during these years, the nation grappled with a range of issues stemming from the aftermath of war, including an elevated incidence of mental health disorders, economic instability and pressures associated with international policies. Recognizing the significant role of mental health in enhancing overall community well-being, health policies should prioritize expanding access to mental health care services. This can be achieved through proactive initiatives that focus on education at various levels, including community programs, schools, universities and health centers. Efforts should aim to improve mental health literacy while addressing cases requiring psychological interventions, thereby reducing the economic and social impact of mental health issues.

Introduction

Self-harm and suicide represent significant health and social challenges on a global scale (Knipe et al., 2022). The World Health Organization (WHO) conceptualizes self-harm as the act of intentionally causing injury to oneself as a means of managing or expressing severe emotional distress and internal conflict. While individuals engaging in such behavior typically do not aim to end their lives, the potential consequences can nonetheless be fatal. Common manifestations of self-harm include deliberate poisoning through the ingestion of an excessive amount of medication or harmful substances, self-inflicted wounds through cutting or burning, head-banging against solid objects or inflicting physical pain by punching or striking oneself against hard surfaces. Although suicidal intent is generally absent among those who engage in self-harm, the risks associated with these actions remain profoundly serious (World Health Organization, 2019). Suicide can be defined as “the act of intentionally carrying out an action to kill oneself” (World Health Organization, 2019).

The WHO estimates that over 700,000 individuals lose their lives to suicide annually worldwide (World Health Organization, 2023). In 2019, the global age-standardized suicide rate stood at 9.0 per 100,000 people, with men exhibiting a higher rate of 12.6 per 100,000 compared to women, who had a rate of 5.4 per 100,000 (World Health Organization, 2021). Although suicide mortality rates are more common in males, self-harm is more common in females (Knipe et al., 2022). Self-harm and suicide result from the interaction of complex factors and are influenced by variables such as age, sex, ethnicity and geography (Knipe et al., 2022).

The region encompassing North Africa and the Middle East, including Iran, has experienced significant challenges in recent decades, marked by rapid population growth, ongoing conflicts and persistent warfare. The nations within the Middle East exhibit substantial ethnic and demographic commonalities, as noted (Moradinazar et al., 2022). Islam serves as the predominant religion across these countries. While historically, the rate of suicide has generally been low in most Islamic societies, contemporary evidence suggests a concerning upward trajectory in this phenomenon (Pritchard and Amanullah, 2007; Mirhashemi et al., 2016). In Iran, the overall suicide rate is 5.3 per 100,000, with rates of 7 and 3.6 for males and females respectively (World Health Organization, 2014). Some determinants of suicide in Iran include family conflict, marital problems, economic constraints and educational failures (Nazarzadeh et al., 2013).

During the past decades, Iran has been affected by economic turmoil, war and international politics (Danaei et al., 2019). After the end of the war in 1988, Iran entered a period of construction in health and nonhealth infrastructures, which led to an increase in gross domestic product (Bank, 2021). Demographic transition has led to population growth, a change in fertility, increased life expectancy, aging and epidemiological transition (Farzadfar et al., 2022; Bhattacharjee et al., 2024; GBD 2021 Causes of Death Collaborators, 2024). Access to mental health care services has not been developed in response to the growth of mental disorders (Farzadfar et al., 2022). The prevalence of mental health issues has gained considerable attention since the conclusion of the Iran–Iraq war. This concern is further underscored by the findings of the Iranian Mental Health Survey, which indicates a substantial occurrence of psychiatric disorders within the country (Sharifi et al., 2015; Danaei et al., 2019). This study focuses on estimating the prevalence, incidence and disability-adjusted life years (DALYs) associated with self-harm and suicide mortality in Iran at both national and subnational levels, using data from the Global Burden of Disease (GBD) 2021.

Methods

Protocol

This manuscript was produced as part of the GBD Collaborator Network and following the GBD Protocol.

Data source

This research was based on the GBD 2021 (Ferrari et al., 2024). The extracted burden of disease indicators included prevalence, incidence, DALYs, years lived with disability (YLDs), years of life lost (YLLs) and death for 371 diseases and injuries, along with estimates of healthy life expectancy. These estimates are provided for sex and age groups, and for 204 countries and territories, including subnational estimates for 21 countries (Ferrari et al., 2024). Data sources used in GBD 2021 included 100,983 data sources (19,189 new data sources for DALYs), 12 new causes and other important methodological updates (Ferrari et al., 2024). These indicators were examined at the national and subnational levels in Iran. More details of GBD 2021 are presented elsewhere (Ferrari et al., 2024).

Case definitions

Self-harm in GBD 2021 is “deliberate bodily damage inflicted on oneself resulting in death or injury. ICD-9: E950-E959; ICD-10: X60-X64.9, X66-X84.9, Y87.0” (2021, 2024; Ferrari et al., 2024). This contains two subclasses: (i) Self-harm by firearm, defined as “Death or disability inflicted by the intentional use of a firearm on oneself. ICD-9: E955-E955.9; ICD-10: X72-X74.9” (2021, 2024; Ferrari et al., 2024); and (ii) Self-harm by other specified means, defined as “Death or occurrence of deliberate bodily damage inflicted on oneself resulting in death by means of self-poisoning, medication overdose, transport, falling from height, hanging or strangulation or other mechanisms not including firearms. ICD9: E950-E954, E956-E959; ICD10: X60-X64.9, X66-X67.9, X69-X71.9, X75-X75.9, X77-X84.9, Y87.0” (2021, 2024; Ferrari et al., 2024).

Estimation framework

YLDs were calculated “with a microsimulation process that used estimated age-sex-location-year-specific prevalent counts of nonfatal disease sequelae (consequences of a disease or injury) for each cause and disability weights for each sequela as the input estimates at varying levels of severity by an appropriate disability weight” (Ferrari et al., 2024). YLLs were calculated as “the product of estimated age-sex-location-year-specific deaths and the standard life expectancy at the age death occurred for a given cause” (Ferrari et al., 2024). DALYs were calculated as the sum of YLDs and YLLs (Ferrari et al., 2024).

Statistics

All-age count estimates and age-standardized rate prevalence (per 100,000) were calculated for prevalence, incidence, DALYs and death. Each of the disease burden indicators was examined in the period of 1990–2021, stratified by sex, age and location, and the % change between 1990 and 2021 was reported. The 95% uncertainty interval (UI) was reported for each of the reported estimates. More details about data, data processing and modeling are provided elsewhere and are related to GBD 2021 (Ferrari et al., 2024). GBD 2021 complies with the Guidelines for Accurate and Transparent Health Estimates Reporting (Stevens et al., 2016) and analyses were completed using Python (version 3.10.4), Stata (version 13.1) and R (version 4.2.1).

Results

Prevalence of self-harm in Iran from 1990 to 2021

Age-standardized rate prevalence (per 100,000) of self-harm in 1990 was 173.92 (95% UI: 146.13–208.75) versus 131.2 (95% UI: 110.55–156.67) in 2021, thus indicating a decrease; the percentage change from 1990 to 2021 was −25%. All-ages count estimates of self-harm in 1990 were 65,834 (95% UI: 55,038–79,967) versus 122,781 (95% UI: 103,040–147,165) in 2021; the percentage change from 1990 to 2021 was 0.86%. While the age-standardized rate has shown a decline, the absolute count estimates have increased over recent decades, primarily due to population growth (Table 1 and Figure 1).

Table 1.

All-ages counts and age-standardized rate (per 100,000) prevalence of self-harm in Iran, stratified by provinces, 1990–2021

Location Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age-standardized rate prevalence (per 100,000)
Iran 173.92 146.13 208.75 131.2 110.55 156.67 −0.25 −0.28 −0.22
Alborz 178.5 149.19 214.24 152.64 127.58 181.88 −0.14 −0.17 −0.12
Ardebil 203.51 171.14 243.77 139.42 117.74 166.11 −0.31 −0.35 −0.28
Bushehr 184.88 155.03 221.06 125.95 107.24 149.5 −0.32 −0.36 −0.28
Chahar Mahaal and Bakhtiari 212.45 177.76 253.55 132.24 110.95 157.88 −0.38 −0.41 −0.35
East Azarbayejan 175.65 146.19 215.15 137.76 115.82 164.44 −0.22 −0.26 −0.18
Fars 211.5 175.97 255.93 150.55 125.4 180.46 −0.29 −0.32 −0.26
Gilan 158.59 131.6 191.18 133.92 112.27 159.34 −0.16 −0.19 −0.12
Golestan 228.88 189.83 279.77 147.88 123.62 177.4 −0.35 −0.39 −0.32
Hamadan 209.33 173.99 255.23 160.45 134.13 191 −0.23 −0.28 −0.19
Hormozgan 165.42 139.31 196.97 126.22 106.7 151.08 −0.24 −0.27 −0.21
Ilam 276.54 230.25 334.34 188.33 157.11 223.85 −0.32 −0.35 −0.29
Isfahan 133.54 112.54 160.13 117.3 98.45 140.14 −0.12 −0.15 −0.1
Kerman 165.18 137.85 199.46 131.3 110.3 157.52 −0.21 −0.24 −0.18
Kermanshah 301.39 251.47 370.07 178.76 148.3 213.86 −0.41 −0.44 −0.38
Khorasan-e-Razavi 169.05 141.02 204.91 125.02 105.77 148.83 −0.26 −0.3 −0.22
Khuzestan 185.52 155.27 222.52 140.34 117.2 169.14 −0.24 −0.27 −0.22
Kohgiluyeh and Boyer-Ahmad 226.6 188.32 278.46 169.1 140.05 203.16 −0.25 −0.29 −0.22
Kurdistan 217.73 182.21 262.47 131.14 110.26 156.09 −0.4 −0.43 −0.37
Lorestan 251.81 210.01 303.26 174.12 144.77 205.66 −0.31 −0.34 −0.27
Markazi 162.87 136.18 196.35 114.55 96.64 136.53 −0.3 −0.33 −0.26
Mazandaran 134.46 112.75 161.84 122.26 102.44 146.53 −0.09 −0.12 −0.06
North Khorasan 205.6 172.25 250.6 140.01 117.22 167.14 −0.32 −0.35 −0.29
Qazvin 122.68 103.93 146.4 116.23 98.39 138.28 −0.05 −0.09 −0.02
Qom 123.74 105.3 147.14 108.72 92.02 129.28 −0.12 −0.15 −0.09
Semnan 123.04 103.9 147.66 109.19 92.72 129.22 −0.11 −0.15 −0.08
Sistan and Baluchistan 115.68 99.28 137.06 121.66 103.55 144.79 0.05 0.02 0.08
South Khorasan 138.54 117.01 165.56 122.12 103.87 146.15 −0.12 −0.16 −0.09
Tehran 146.96 125.22 173.76 109.65 93.76 129.7 −0.25 −0.28 −0.23
West Azarbayejan 209.73 174.08 258.42 155.46 129.71 185.95 −0.26 −0.3 −0.22
Yazd 139.26 118.03 165.83 112.4 94.88 133.54 −0.19 −0.23 −0.17
Zanjan 124.69 104.09 150.7 106.84 90.84 127.12 −0.14 −0.19 −0.11
All-ages counts estimates
Iran 65,834.87 55,038.75 79,967.61 122,781.76 103,040.92 147,165.99 0.86 0.75 0.97
Alborz 1,696.82 1,405.40 2,081.74 5,259.88 4,377.23 6,314.42 2.1 1.93 2.27
Ardebil 1,436.58 1,198.00 1,739.51 2,019.58 1,696.22 2,413.46 0.41 0.31 0.5
Bushehr 804.56 667.29 983.66 1,573.10 1,329.39 1,881.32 0.96 0.81 1.08
Chahar Mahaal and Bakhtiari 914.4 760.57 1,108.48 1,413.89 1,179.86 1,700.93 0.55 0.44 0.64
East Azarbayejan 4,103.35 3,397.43 5,076.82 6,465.85 5,421.25 7,712.64 0.58 0.45 0.69
Fars 4,935.51 4,059.97 6,077.20 8,718.66 7,230.23 10,491.51 0.77 0.65 0.88
Gilan 2,643.18 2,187.74 3,231.66 4,440.05 3,727.77 5,293.25 0.68 0.58 0.79
Golestan 1,899.27 1,564.14 2,373.57 3,029.73 2,520.14 3,653.02 0.6 0.46 0.71
Hamadan 2,341.79 1,934.59 2,891.69 3,232.73 2,694.26 3,851.66 0.38 0.27 0.48
Hormozgan 934.97 782.04 1,127.97 2,362.41 1,971.94 2,852.77 1.53 1.39 1.63
Ilam 714.73 584.95 884.35 1,267.53 1,049.95 1,523.53 0.77 0.64 0.89
Isfahan 3,447.57 2,892.63 4,196.77 7,361.04 6,172.07 8,820.49 1.14 1.01 1.25
Kerman 1,943.91 1,620.02 2,375.54 4,561.80 3,801.49 5,503.58 1.35 1.22 1.47
Kermanshah 3,285.21 2,701.59 4,109.75 4,128.72 3,416.37 4,953.89 0.26 0.16 0.34
Khorasan-e-Razavi 5,164.58 4,300.56 6,309.27 8,880.43 7,477.54 10,626.20 0.72 0.59 0.83
Khuzestan 3,535.05 2,942.60 4,346.78 7,056.76 5,847.49 8,562.09 1 0.9 1.09
Kohgiluyeh and Boyer-Ahmad 635.97 521.02 801.32 1,333.77 1,093.08 1,631.27 1.1 0.95 1.23
Kurdistan 1,726.69 1,432.93 2,120.29 2,496.20 2,088.58 2,990.30 0.45 0.34 0.54
Lorestan 2,365.37 1,961.41 2,913.33 3,347.92 2,773.13 3,989.78 0.42 0.32 0.51
Markazi 1,327.51 1,111.45 1,616.79 1,950.96 1,643.72 2,325.07 0.47 0.36 0.57
Mazandaran 2,368.55 1,972.35 2,909.53 5,241.43 4,394.54 6,269.77 1.21 1.09 1.35
North Khorasan 855.31 712.79 1,061.82 1,256.70 1,047.33 1,508.13 0.47 0.36 0.55
Qazvin 735.99 617.77 893.05 1,748.59 1,475.38 2,088.49 1.38 1.24 1.51
Qom 578.08 487.18 702.7 1,618.84 1,361.07 1,939.42 1.8 1.64 1.97
Semnan 423.16 356.45 510.08 945.74 799.38 1,128.53 1.23 1.13 1.33
Sistan and Baluchistan 967.64 822.49 1,157.74 2,888.69 2,427.41 3,464.85 1.99 1.88 2.09
South Khorasan 614.73 519.05 739.81 1,076.21 911.1 1,291.98 0.75 0.66 0.83
Tehran 8,903.83 7,516.98 10,694.68 18,736.82 15,979.39 22,234.83 1.1 0.99 1.2
West Azarbayejan 3,193.78 2,621.41 4,007.53 5,687.11 4,723.61 6,828.00 0.78 0.65 0.9
Yazd 635.95 536.06 769.85 1,381.60 1,160.92 1,651.76 1.17 1.06 1.28
Zanjan 700.84 583.61 860.52 1,299.00 1,099.52 1,550.85 0.85 0.74 0.96

Figure 1.

Figure 1.

Trend in prevalence of self-harm in Iran, 1990–2021.

Incidence of self-harm in Iran from 1990 to 2021

The age-standardized incidence rate of self-harm per 100,000 population was 57.34 (95% UI: 46.67–69.95) in 1990, compared to 46.01 (95% UI: 36.07–57.53) in 2021. This reflects a percentage change of −0.20% over the period, indicating a downward trend. All-ages count estimates increased from 32,083.13 (95% UI: 25,531–40,402) in 1990 to 40,908.57 (95% UI: 32,066.60–50,434.56) in 2021; the percentage change from 1990 to 2021 was 0.28%. Between 2000 and 2010, the estimated incidence of self-harm showed an overall increase, peaking in 2005. Nevertheless, when adjusted for age, the incidence rates indicate a declining trend (Table 2 and Figure 2).

Table 2.

All-ages counts and age-standardized rate (per 100,000) incidence of self-harm in Iran, stratified by provinces, 1990–2021

Location Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age-standardized rate incidence (per 100,000)
Iran 57.34 46.67 69.95 46.01 36.07 57.53 −0.2 −0.26 −0.15
Alborz 58.75 47.85 71.94 54.12 42.58 67.02 −0.08 −0.15 −0.01
Ardebil 66.26 53.35 81.24 48.96 38.34 61.13 −0.26 −0.32 −0.2
Bushehr 62.26 50.4 76.03 45.3 35.66 56.3 −0.27 −0.34 −0.21
Chahar Mahaal and Bakhtiari 71.82 58.3 88.23 48.59 38.34 61.09 −0.32 −0.38 −0.27
East Azarbayejan 55.48 44.64 68.17 47.45 36.97 59.62 −0.14 −0.22 −0.06
Fars 69.82 56.66 85.59 52.91 41.66 66.62 −0.24 −0.3 −0.18
Gilan 55.97 45.52 68.17 47.72 37.45 59.53 −0.15 −0.23 −0.08
Golestan 75.49 60.77 92.1 51.35 40.35 63.55 −0.32 −0.38 −0.26
Hamadan 66.09 53.63 80.8 55.26 42.9 68.57 −0.16 −0.24 −0.09
Hormozgan 52.5 42.54 64.54 42.75 33.5 53.47 −0.19 −0.25 −0.11
Ilam 94.06 76.65 114.12 67.71 53.48 83.73 −0.28 −0.34 −0.22
Isfahan 43.51 35.1 53.68 41 32.37 51.27 −0.06 −0.13 0.01
Kerman 52.94 42.66 65.05 44.85 34.99 56.4 −0.15 −0.23 −0.07
Kermanshah 98.99 81.37 121.18 62.73 49.26 77.69 −0.37 −0.42 −0.31
Khorasan-e-Razavi 54.11 43.74 67.21 42.95 33.59 53.64 −0.21 −0.27 −0.14
Khuzestan 61.47 49.24 75.94 49.07 38.3 61.32 −0.2 −0.26 −0.15
Kohgiluyeh and Boyer-Ahmad 75.31 60.75 92.62 60.5 47.55 76.12 −0.2 −0.26 −0.12
Kurdistan 71.16 57.59 86.6 45.7 35.79 57.71 −0.36 −0.42 −0.3
Lorestan 83.76 67.77 102.28 63.28 49.55 78.07 −0.24 −0.3 −0.19
Markazi 51.66 41.77 62.96 39.46 31.08 49.8 −0.24 −0.3 −0.17
Mazandaran 44.86 36.35 54.98 43.56 34.06 54.22 −0.03 −0.1 0.05
North Khorasan 66.26 53.58 80.45 48.4 38.25 60.81 −0.27 −0.33 −0.21
Qazvin 39.42 31.94 48.2 40.52 31.88 50.76 0.03 −0.05 0.13
Qom 38.71 31.52 47.94 37.23 29.02 46.83 −0.04 −0.12 0.05
Semnan 39.16 31.78 48.14 38.01 29.58 47.79 −0.03 −0.11 0.04
Sistan and Baluchistan 36.32 29.18 44.62 39.48 30.79 50.01 0.09 0.02 0.16
South Khorasan 44.36 36.08 54.6 42.09 33.09 53.21 −0.05 −0.13 0.03
Tehran 48.8 39.64 59.27 39.32 30.69 49.46 −0.19 −0.26 −0.14
West Azarbayejan 67.54 54.44 83.27 54.2 42.62 67.46 −0.2 −0.26 −0.13
Yazd 45.25 36.24 55.14 39.36 30.72 49.83 −0.13 −0.2 −0.06
Zanjan 42.44 34.25 51.99 37.43 29.29 46.57 −0.12 −0.19 −0.05
All-ages counts estimates
Iran 32,083.13 25,531.07 40,402.50 40,908.57 32,066.60 50,434.56 0.28 0.13 0.45
Alborz 852.84 677.61 1,065.37 1,738.63 1,360.77 2,121.95 1.04 0.79 1.34
Ardebil 726.2 571.52 906.22 674.68 530.38 833.61 −0.07 −0.2 0.08
Bushehr 428.98 339.05 541.65 618.96 485.3 763.99 0.44 0.28 0.64
Chahar Mahaal and Bakhtiari 484.87 381.39 608.4 507.2 402.47 632.45 0.05 −0.09 0.21
East Azarbayejan 1,905.05 1,495.14 2,426.58 1,990.47 1,551.77 2,469.60 0.04 −0.1 0.21
Fars 2,468.30 1,952.91 3,084.60 2,866.83 2,245.32 3,581.91 0.16 0.02 0.34
Gilan 1,290.10 1,030.44 1,602.01 1,280.72 1,020.46 1,574.78 −0.01 −0.15 0.16
Golestan 962.84 761.62 1,221.29 1,026.12 811.59 1,266.94 0.07 −0.06 0.21
Hamadan 1,108.85 875.84 1,399.66 988.45 775.47 1,221.09 −0.11 −0.23 0.03
Hormozgan 448.97 355.86 563.39 901.75 699.31 1,116.00 1.01 0.76 1.24
Ilam 399.7 319.8 502.48 449.74 355.03 561.26 0.13 −0.04 0.32
Isfahan 1,645.50 1,299.43 2,083.79 2,271.98 1,780.54 2,781.57 0.38 0.2 0.6
Kerman 928.84 731.49 1,170.48 1,594.94 1,231.80 1,995.61 0.72 0.5 0.97
Kermanshah 1,634.78 1,304.50 2,037.65 1,315.87 1,040.43 1,633.22 −0.2 −0.3 −0.07
Khorasan-e-Razavi 2,429.10 1,920.42 3,086.05 3,057.46 2,402.54 3,775.67 0.26 0.1 0.43
Khuzestan 1,843.14 1,443.24 2,343.91 2,517.80 1,960.87 3,115.10 0.37 0.21 0.55
Kohgiluyeh and Boyer-Ahmad 357.38 282.13 458.36 506.33 392.96 638.21 0.42 0.22 0.65
Kurdistan 853.51 672.77 1,065.77 841.78 660.22 1,047.65 −0.01 −0.15 0.14
Lorestan 1,248.81 988.24 1,581.47 1,189.21 934.28 1,464.29 −0.05 −0.17 0.1
Markazi 612.27 485.14 771.58 601.49 474.71 749.9 −0.02 −0.15 0.13
Mazandaran 1,169.66 926.07 1,457.30 1,577.38 1,235.91 1,932.35 0.35 0.18 0.58
North Khorasan 406.36 319.96 508.8 429.77 339.25 534.98 0.06 −0.06 0.2
Qazvin 361.49 283.59 457.61 580.87 457.06 719.76 0.61 0.41 0.88
Qom 277.67 219.6 354.37 566.56 437.16 706 1.04 0.77 1.32
Semnan 183.86 146.88 232.81 328.01 256.54 410.98 0.78 0.56 1.03
Sistan and Baluchistan 468.26 366.97 603.82 1,248.74 958.94 1,614.82 1.67 1.45 1.92
South Khorasan 268.34 213.38 339.46 383.37 300.94 478.68 0.43 0.27 0.61
Tehran 4,091.57 3,277.09 5,103.61 5,986.92 4,666.01 7,380.59 0.46 0.3 0.66
West Azarbayejan 1,580.07 1,241.56 1,994.11 1,948.27 1,544.81 2,397.27 0.23 0.08 0.42
Yazd 291.25 229.19 363.95 488.96 380.83 613.11 0.68 0.49 0.89
Zanjan 354.59 279.96 448.1 429.31 337.33 532.67 0.21 0.04 0.41

Figure 2.

Figure 2.

Trend in incidence of self-harm in Iran, 1990–2021.

DALYs of self-harm in Iran from 1990 to 2021

Age-standardized rate DALYs (per 100,000) of self-harm in 1990 was 350.18 (95% UI: 287.04–382.31) versus 226.63 (95% UI: 202.91–250.46) in 2021; the percentage change from 1990 to 2021 was −0.35% and showed a decreased DALYs. All-ages count estimates increased from 188,787 (95% UI: 148,375–206,949) in 1990 to 203,257 (95% UI: 183,865–225,136) in 2021vs. ; the percentage change from 1990 to 2021 was 0.08 (Table 3 and Figure 3).

Table 3.

All-ages counts and age-standardized rate (per 100,000) DALYs of self-harm in Iran, stratified by provinces, 1990–2021

Location Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age-standardized rate DALYs (per 100,000)
Iran 350.18 287.04 382.31 226.63 202.91 250.46 −0.35 −0.42 −0.22
Alborz 493.12 252.38 631.4 356.58 190.9 454.22 −0.28 −0.47 0.11
Ardebil 408.97 295.12 508.74 291.96 204.31 342.93 −0.29 −0.44 −0.01
Bushehr 341.13 239.41 431.86 171.32 142.1 204.44 −0.5 −0.62 −0.27
Chahar Mahaal and Bakhtiari 401.81 259.98 502.95 184.18 140.85 232.39 −0.54 −0.66 −0.33
East Azarbayejan 415.46 313.52 513.48 266.1 211.07 324.22 −0.36 −0.51 −0.14
Fars 406.88 311.49 501.78 319.17 254.46 393.1 −0.22 −0.4 0.05
Gilan 361.58 276.02 451.34 244.75 193.76 299.89 −0.32 −0.49 −0.11
Golestan 489.57 300.89 616.74 309.1 243.07 366.57 −0.37 −0.51 −0.13
Hamadan 555.83 344.06 691.64 416.68 255.41 513.88 −0.25 −0.44 −0.01
Hormozgan 356.45 275.81 454.42 260.18 207.84 320.75 −0.27 −0.47 0
Ilam 880.89 320.57 1,132.86 566.95 253.28 691.14 −0.36 −0.51 −0.08
Isfahan 178.11 139.92 268.33 144.55 114.9 198.61 −0.19 −0.4 0.1
Kerman 377.8 301.1 471.96 233.57 192.98 291.97 −0.38 −0.54 −0.16
Kermanshah 785.01 440.5 977.94 477.91 271.32 582.46 −0.39 −0.54 −0.18
Khorasan-e-Razavi 355.79 273.46 438.88 198.85 161.85 242.65 −0.44 −0.57 −0.24
Khuzestan 346.58 276.61 431.72 253.28 212.23 299.43 −0.27 −0.44 −0.04
Kohgiluyeh and Boyer-Ahmad 573.56 335.88 710.31 434.18 298.98 520.95 −0.24 −0.41 0
Kurdistan 421.82 305.53 534.14 225.02 181.42 270.28 −0.47 −0.6 −0.25
Lorestan 724.79 305.66 933.62 377.43 192.99 464.12 −0.48 −0.62 −0.25
Markazi 294.71 231.32 379.33 175.1 141.12 228.91 −0.41 −0.56 −0.19
Mazandaran 198.17 155.78 260.8 158.73 129.76 217.98 −0.2 −0.39 0.04
North Khorasan 374.07 282.56 457.11 238.5 200.23 280.85 −0.36 −0.51 −0.12
Qazvin 223.39 175.79 292.66 149.51 119.74 189.36 −0.33 −0.51 −0.1
Qom 231.49 178.3 335.87 179.69 144.44 231.96 −0.22 −0.46 0.09
Semnan 215.17 171.28 280.5 121.24 94.55 163.1 −0.44 −0.59 −0.25
Sistan and Baluchistan 190 142.26 283.89 198.58 155.68 330.4 0.05 −0.22 0.45
South Khorasan 223.24 174.56 288 158.68 133.85 196.89 −0.29 −0.45 −0.05
Tehran 170.45 130.14 232.72 106.46 83.78 153.4 −0.38 −0.53 −0.1
West Azarbayejan 539.35 371.79 675.15 334.31 234.64 396.99 −0.38 −0.52 −0.18
Yazd 200.33 160.86 283.47 120.79 95.01 167.71 −0.4 −0.56 −0.18
Zanjan 233.32 183.92 287.05 130.58 106.69 169.15 −0.44 −0.57 −0.19
All-ages counts estimates
Iran 188,787.36 148,375.22 206,949.05 203,257.54 183,865.89 225,136.26 0.08 −0.04 0.34
Alborz 6,687.32 3,383.84 8,655.34 11,694.80 6,202.08 14,982.35 0.75 0.29 1.66
Ardebil 4,298.47 3,001.06 5,401.70 4,095.63 2,884.61 4,772.64 −0.05 −0.25 0.35
Bushehr 2,226.34 1,495.73 2,824.44 2,342.69 1,941.82 2,791.26 0.05 −0.2 0.56
Chahar Mahaal and Bakhtiari 2,572.76 1,586.08 3,295.07 1,958.68 1,497.03 2,464.93 −0.24 −0.44 0.14
East Azarbayejan 13,840.83 10,046.49 17,409.55 11,301.21 9,024.35 13,630.56 −0.18 −0.38 0.12
Fars 13,854.68 10,114.14 17,323.50 17,182.30 13,656.97 21,111.36 0.24 −0.07 0.69
Gilan 7,986.56 5,896.88 10,049.17 6,744.62 5,337.00 8,212.65 −0.16 −0.35 0.12
Golestan 6,024.01 3,505.53 7,683.93 6,186.09 4,854.80 7,301.54 0.03 −0.21 0.5
Hamadan 8,840.85 5,310.55 11,115.27 7,578.78 4,638.73 9,382.85 −0.14 −0.36 0.14
Hormozgan 2,869.23 2,165.39 3,655.39 5,477.61 4,365.06 6,742.94 0.91 0.36 1.69
Ilam 3,497.29 1,202.58 4,504.89 3,808.27 1,716.84 4,634.00 0.09 −0.18 0.64
Isfahan 6,601.83 5,114.99 9,719.58 8,094.64 6,453.58 11,114.41 0.23 −0.11 0.67
Kerman 6,300.35 4,955.29 7,911.67 8,304.06 6,854.86 10,420.84 0.32 −0.04 0.78
Kermanshah 12,517.38 6,747.37 15,817.70 10,157.17 5,823.33 12,417.83 −0.19 −0.4 0.11
Khorasan-e-Razavi 15,211.79 11,724.65 18,831.81 14,243.37 11,589.26 17,306.93 −0.06 −0.28 0.31
Khuzestan 9,906.02 7,774.19 12,392.73 12,949.45 10,892.92 15,186.51 0.31 −0.01 0.73
Kohgiluyeh and Boyer-Ahmad 2,701.46 1,511.41 3,379.37 3,666.89 2,593.28 4,395.43 0.36 0.04 0.89
Kurdistan 4,885.55 3,329.27 6,228.69 4,189.00 3,395.40 4,981.09 −0.14 −0.36 0.22
Lorestan 10,132.03 3,964.18 13,262.95 7,291.35 3,693.00 9,030.03 −0.28 −0.47 0.1
Markazi 3,375.52 2,647.18 4,259.36 2,716.78 2,183.41 3,555.95 −0.2 −0.41 0.1
Mazandaran 4,995.86 3,910.74 6,458.09 5,829.47 4,793.05 8,074.73 0.17 −0.11 0.52
North Khorasan 2,228.20 1,673.37 2,742.59 2,124.05 1,784.06 2,507.51 −0.05 −0.29 0.31
Qazvin 1,920.77 1,505.80 2,447.16 2,193.83 1,767.28 2,804.78 0.14 −0.16 0.55
Qom 1,565.97 1,188.84 2,266.52 2,768.63 2,235.96 3,585.56 0.77 0.24 1.5
Semnan 958.21 762.6 1,229.19 1,061.99 835.54 1,432.08 0.11 −0.19 0.49
Sistan and Baluchistan 2,329.67 1,674.97 3,142.36 6,234.16 4,879.94 10,396.56 1.68 0.96 2.94
South Khorasan 1,297.16 1,005.52 1,660.15 1,452.93 1,223.74 1,797.36 0.12 −0.13 0.52
Tehran 13,837.11 10,528.42 18,855.65 16,561.57 13,082.46 23,630.01 0.2 −0.11 0.73
West Azarbayejan 12,249.07 8,154.51 15,321.40 12,016.01 8,433.89 14,271.46 −0.02 −0.24 0.33
Yazd 1,230.79 976 1,682.76 1,492.94 1,178.11 2,081.85 0.21 −0.11 0.66
Zanjan 1,844.30 1,411.32 2,281.83 1,538.57 1,257.01 1,997.43 −0.17 −0.37 0.24

Figure 3.

Figure 3.

Trend in DALYs of self-harm in Iran, 1990–2021.

Suicide mortality in Iran from 1990 to 2021

Age-standardized rate (per 100,000) of suicide mortality in 1990 was 6.26 (95% UI: 5.26–6.9) versus 4.12 (95% UI: 3.72–4.58) in 2021; the percentage change from 1990 to 2021 was −0.34% and showed a decreased trend. The incidence of suicide-related mortality in 2021 amounted to 3,708 individuals. This figure indicates an increase when compared to 1990, during which the recorded number of suicide deaths was 3,069 (Table 4 and Figure 4).

Table 4.

All-ages counts and age-standardized rate (per 100,000) of suicide mortality in Iran, stratified by provinces, 1990–2021

Location Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age-standardized rate suicide mortality (per 100,000)
Iran 6.26 5.26 6.9 4.12 3.72 4.58 −0.34 −0.41 −0.19
Alborz 9.26 4.67 11.95 6.64 3.5 8.49 −0.28 −0.47 0.1
Ardebil 7.3 5.28 9.1 5.52 3.86 6.46 −0.24 −0.41 0.04
Bushehr 6.15 4.36 7.74 3.29 2.74 3.89 −0.46 −0.59 −0.23
Chahar Mahaal and Bakhtiari 7.28 4.78 9.08 3.49 2.65 4.43 −0.52 −0.64 −0.28
East Azarbayejan 7.38 5.65 9.13 4.89 3.88 5.92 −0.34 −0.49 −0.12
Fars 7.21 5.61 8.9 5.62 4.44 6.88 −0.22 −0.41 0.06
Gilan 6.71 5.22 8.29 4.56 3.58 5.56 −0.32 −0.49 −0.11
Golestan 8.59 5.36 10.68 5.67 4.44 6.69 −0.34 −0.49 −0.1
Hamadan 9.97 6.17 12.47 7.56 4.54 9.39 −0.24 −0.43 0.03
Hormozgan 6.47 4.95 8.2 4.65 3.73 5.75 −0.28 −0.48 −0.02
Ilam 16.24 5.73 20.62 10.68 4.58 12.93 −0.34 −0.49 −0.07
Isfahan 3.12 2.4 4.86 2.57 2.03 3.59 −0.18 −0.4 0.13
Kerman 6.75 5.4 8.45 4.2 3.44 5.35 −0.38 −0.54 −0.15
Kermanshah 14.09 7.97 17.46 8.64 4.81 10.57 −0.39 −0.54 −0.16
Khorasan-e-Razavi 6.4 5.06 7.95 3.67 2.98 4.46 −0.43 −0.56 −0.23
Khuzestan 6.14 4.88 7.63 4.53 3.82 5.35 −0.26 −0.43 −0.03
Kohgiluyeh and Boyer-Ahmad 9.71 5.77 12.05 7.41 5.11 8.88 −0.24 −0.4 0.01
Kurdistan 7.42 5.4 9.47 4.09 3.3 4.94 −0.45 −0.59 −0.22
Lorestan 13.2 5.59 16.99 7.28 3.57 8.93 −0.45 −0.58 −0.22
Markazi 5.23 4.14 6.98 3.14 2.51 4.19 −0.4 −0.56 −0.18
Mazandaran 3.5 2.74 4.74 2.86 2.33 3.96 −0.18 −0.37 0.08
North Khorasan 6.52 4.87 7.99 4.28 3.61 5.04 −0.34 −0.5 −0.07
Qazvin 4.06 3.22 5.58 2.83 2.27 3.63 −0.3 −0.48 −0.06
Qom 4.27 3.27 6.3 3.28 2.62 4.23 −0.23 −0.47 0.07
Semnan 3.95 3.14 5.32 2.27 1.79 3.09 −0.43 −0.58 −0.23
Sistan and Baluchistan 3.4 2.55 5.67 3.47 2.72 5.93 0.02 −0.23 0.42
South Khorasan 3.98 3.12 5.25 2.88 2.42 3.64 −0.28 −0.44 −0.02
Tehran 3.11 2.34 4.33 1.97 1.55 2.87 −0.37 −0.53 −0.07
West Azarbayejan 9.5 6.74 11.83 6.06 4.22 7.2 −0.36 −0.5 −0.15
Yazd 3.63 2.92 5.36 2.19 1.74 3.11 −0.4 −0.56 −0.17
Zanjan 4.23 3.39 5.23 2.46 2.01 3.22 −0.42 −0.55 −0.17
All-ages counts estimates
Iran (Islamic Republic of) 3,069.66 2,463.49 3,362.63 3,708.56 3,354.15 4,117.61 0.21 0.07 0.5
Alborz 112.25 56.14 145.25 217.09 112.52 278.66 0.93 0.43 2
Ardebil 69.12 48.96 86.94 77.61 54.53 90.84 0.12 −0.11 0.58
Bushehr 35.93 24.4 45.81 42.91 35.72 50.92 0.19 −0.1 0.79
Chahar Mahaal and Bakhtiari 41.83 25.94 53.2 36.79 28.31 46.41 −0.12 −0.35 0.32
East Azarbayejan 222.88 163.9 278.89 210.69 168.22 256.38 −0.05 −0.29 0.29
Fars 223.18 166.22 277.95 305.49 243.09 375.63 0.37 0.02 0.86
Gilan 135.4 101.72 169.4 131.37 103.42 160.69 −0.03 −0.26 0.28
Golestan 96.3 56.94 121.95 112.94 89.32 134.02 0.17 −0.09 0.67
Hamadan 144.66 87.42 180.94 141.58 84.61 177.2 −0.02 −0.27 0.31
Hormozgan 47.24 35.7 61.18 94.83 75.59 117 1.01 0.45 1.8
Ilam 57.15 19.34 73.91 70.78 30.48 86.22 0.24 −0.06 0.83
Isfahan 105.42 80.84 159.8 147.13 116.7 204.98 0.4 0.01 0.93
Kerman 102.64 80.31 128.48 146.82 120.13 186.11 0.43 0.05 0.94
Kermanshah 203.33 109.99 256.36 186.67 104.82 228.98 −0.08 −0.31 0.26
Khorasan-e-Razavi 247.99 190.24 307.48 259.84 209.17 314.53 0.05 −0.21 0.44
Khuzestan 159.37 126.21 199.72 228.89 192.68 270.19 0.44 0.09 0.9
Kohgiluyeh and Boyer-Ahmad 41.58 23.27 52.13 62.25 43.42 74.27 0.5 0.15 1.08
Kurdistan 78.09 54.4 99.75 76.24 61.66 90.81 −0.02 −0.26 0.39
Lorestan 165.74 66.09 216.64 139.13 67.9 171.91 −0.16 −0.39 0.26
Markazi 54.87 42.7 70.94 49.92 39.55 66.64 −0.09 −0.33 0.26
Mazandaran 80.41 62.45 106.15 108.66 88.26 152.86 0.35 0.03 0.79
North Khorasan 35.51 26.21 43.78 37.92 31.95 44.84 0.07 −0.21 0.49
Qazvin 31.46 24.44 40.78 41.29 33.04 53.25 0.31 −0.02 0.8
Qom 25.75 19.64 37.82 49.67 39.95 64.19 0.93 0.34 1.75
Semnan 16.21 12.86 21.11 19.59 15.53 26.77 0.21 −0.12 0.63
Sistan and Baluchistan 37.63 27.62 54.31 103.54 80.93 176.23 1.75 1.02 2.96
South Khorasan 21.33 16.44 27.63 26.09 21.8 32.83 0.22 −0.07 0.65
Tehran 231.21 173.85 319.4 312.09 243.11 451.58 0.35 −0.01 1
West Azarbayejan 194.55 131.62 243.98 214.92 149.49 257.61 0.1 −0.14 0.49
Yazd 20.4 16.21 28.9 26.7 21.14 37.83 0.31 −0.05 0.81
Zanjan 30.27 23.66 37.37 29.13 23.64 38.12 −0.04 −0.27 0.39

Figure 4.

Figure 4.

Trend in suicide mortality in Iran, 1990–2021.

The sex-specific burden of self-harm and suicide mortality in Iran

There are differences between males and females in the prevalence, incidence, DALYs and suicide mortality of self-harm. In 1990, the prevalence of self-harm was lower in males than in females. For males, the age-standardized rate (per 100,000) was 136.66 (95% UI: 116.78–162.01); for females, it was 212.79 (95% UI: 176.25–258.02). In 2021, the prevalence of self-harm was higher in males (137.62 per 100,000) than in females (124.82 per 100,000). Since 2015, a notable shift in the prevalence rates of self-harm among men and women has been observed, signaling an epidemiological transformation. By 2021, the incidence of self-harm reached a total of 64,913 cases among males (95% UI: 54,867–76,963) and 57,867 cases among females (95% UI: 48,020–70,468) (Table 5 and Figure 5).

Table 5.

Sex-specific prevalence, incidence, and DALYs of self-harm in Iran, stratified by provinces, 1990–2021

Location Sex Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age-standardized rate prevalence (per 100,000)
Iran (Islamic Republic of) Males 136.66 116.78 162.01 137.62 116.74 162.53 0.01 −0.03 0.04
Alborz Males 163.83 138.08 193.6 170.83 143.83 203.19 0.04 0.01 0.08
Ardebil Males 147.11 125.39 174.27 146.43 123.87 173.2 0 −0.05 0.04
Bushehr Males 108.32 93.34 127.56 115.08 99.08 134.96 0.06 0.03 0.1
Chahar Mahaal and Bakhtiari Males 141.49 121.28 167.54 122.41 104.08 144.75 −0.13 −0.17 −0.11
East Azarbayejan Males 146.14 122.95 175.12 150.01 126.79 177.18 0.03 −0.04 0.09
Fars Males 156.62 132.69 186.4 154.82 130.19 182.87 −0.01 −0.06 0.03
Gilan Males 129.48 109.09 154.08 142.85 121.01 168.34 0.1 0.05 0.16
Golestan Males 132.94 112.99 157.58 138.12 116.27 163.85 0.04 0 0.08
Hamadan Males 183.67 154.66 217.44 178.33 150.46 210.48 −0.03 −0.09 0.02
Hormozgan Males 136.49 116.77 160.3 143.66 120.47 170.41 0.05 0.02 0.09
Ilam Males 174.94 147.95 207.73 185.15 155.1 219.47 0.06 0.01 0.11
Isfahan Males 120.93 104.21 142.67 131.21 110.32 156.01 0.08 0.05 0.13
Kerman Males 130.06 110.03 155.79 135.89 114.57 161.69 0.04 0 0.09
Kermanshah Males 213.47 180.1 252.98 171.91 144.46 202.43 −0.19 −0.24 −0.16
Khorasan-e-Razavi Males 120.01 101.73 142.93 131.06 111.92 154.32 0.09 0.04 0.15
Khuzestan Males 125.51 107.76 147.84 135.26 114.71 160.87 0.08 0.04 0.12
Kohgiluyeh and Boyer-Ahmad Males 152.92 128.2 185.68 161.33 133.97 191.2 0.05 0 0.11
Kurdistan Males 144.67 124.12 171.59 133.17 112.12 157.88 −0.08 −0.12 −0.04
Lorestan Males 170.14 144.18 200.88 162.63 136.65 191.88 −0.04 −0.08 −0.01
Markazi Males 153.06 129.43 183.8 131.17 110.5 154.49 −0.14 −0.19 −0.1
Mazandaran Males 106.78 91.57 125.97 129.96 110.33 154.04 0.22 0.17 0.26
North Khorasan Males 123.89 106.34 147.15 125.66 107.34 149.21 0.01 −0.03 0.06
Qazvin Males 94.46 81.32 111.76 127.18 107.98 149.52 0.35 0.3 0.41
Qom Males 129.43 110.25 153.63 130.34 109.48 155.05 0.01 −0.04 0.06
Semnan Males 108.7 93.72 127.54 123.29 104.92 144.83 0.13 0.1 0.17
Sistan and Baluchistan Males 98.52 85.6 115.04 127.44 108.69 150.3 0.29 0.25 0.35
South Khorasan Males 99.77 85.85 116.51 124.17 106.32 146.05 0.24 0.19 0.3
Tehran Males 140.72 121.56 164.86 120.41 103.19 141.56 −0.14 −0.17 −0.12
West Azarbayejan Males 136.83 114.97 165.12 152.28 128.35 178.97 0.11 0.05 0.18
Yazd Males 106.22 91.71 125.49 118.59 100.71 140.09 0.12 0.08 0.17
Zanjan Males 94.22 81.02 111.67 112.97 96.35 132.83 0.2 0.16 0.25
Iran (Islamic Republic of) Females 212.79 176.25 258.02 124.82 104.03 151.66 −0.41 −0.44 −0.39
Alborz Females 194.62 160.52 237.67 133.38 109.69 162.29 −0.31 −0.34 −0.29
Ardebil Females 262.38 218.03 320.49 132.92 111.12 159.89 −0.49 −0.53 −0.46
Bushehr Females 263.56 218.14 319.91 137.95 115.98 164.82 −0.48 −0.51 −0.44
Chahar Mahaal and Bakhtiari Females 287.49 237.05 346.59 142.37 117.99 172.09 −0.5 −0.54 −0.48
East Azarbayejan Females 206.17 168.84 255.19 125.54 104.22 153.25 −0.39 −0.42 −0.36
Fars Females 268.48 220.89 329.47 146.49 121.12 178.87 −0.45 −0.48 −0.43
Gilan Females 187.13 153.98 227.69 125.22 103.71 152.34 −0.33 −0.36 −0.3
Golestan Females 325.86 267.47 409.23 157.58 131.42 190.88 −0.52 −0.55 −0.48
Hamadan Females 235.8 193.77 293.79 142.45 117.46 173.31 −0.4 −0.43 −0.37
Hormozgan Females 196.78 163.41 237.7 108.38 91.43 131.41 −0.45 −0.48 −0.42
Ilam Females 391.63 322.86 477.01 191.76 159.36 229.14 −0.51 −0.54 −0.48
Isfahan Females 147.17 122.35 179.41 102.96 86.62 124.11 −0.3 −0.34 −0.27
Kerman Females 201.38 166.13 245.48 126.81 105.55 155.48 −0.37 −0.4 −0.34
Kermanshah Females 396.03 327.12 494.24 186.02 152.78 228.87 −0.53 −0.56 −0.5
Khorasan-e-Razavi Females 219.32 181.16 269.09 119.19 99.32 143.53 −0.46 −0.49 −0.42
Khuzestan Females 247.81 204.57 301.81 145.62 119.92 177.73 −0.41 −0.44 −0.38
Kohgiluyeh and Boyer-Ahmad Females 307.75 252.04 385.9 177.37 145.18 217.53 −0.42 −0.45 −0.39
Kurdistan Females 295.53 243.08 367.49 129.3 107.95 156.02 −0.56 −0.59 −0.53
Lorestan Females 339.58 281.1 417.86 185.75 153.94 224.89 −0.45 −0.48 −0.42
Markazi Females 172.34 143.2 209.3 97.89 82.65 117.82 −0.43 −0.47 −0.4
Mazandaran Females 161.91 133.47 198.03 114.71 94.21 139.21 −0.29 −0.32 −0.26
North Khorasan Females 288.87 239.4 358.38 154.63 128.55 187.04 −0.46 −0.49 −0.44
Qazvin Females 151.48 126.66 182.54 105.33 88.65 127.22 −0.3 −0.33 −0.28
Qom Females 117.62 99.64 140.8 85.98 73.07 103.62 −0.27 −0.3 −0.24
Semnan Females 137.72 114.97 167.14 95.06 80.27 114.2 −0.31 −0.34 −0.28
Sistan and Baluchistan Females 134.15 112.85 160.44 115.94 97.34 139.55 −0.14 −0.16 −0.11
South Khorasan Females 178.25 148.6 216.38 120.68 101.08 147.06 −0.32 −0.35 −0.3
Tehran Females 153.7 129.42 183.73 98.65 83.33 118.84 −0.36 −0.39 −0.33
West Azarbayejan Females 285.56 234.25 358.33 159.06 131.37 194.74 −0.44 −0.47 −0.42
Yazd Females 174.02 145.29 210.98 105.78 88.33 127.36 −0.39 −0.43 −0.37
Zanjan Females 155.04 127.61 189.14 101.03 84.81 121.27 −0.35 −0.39 −0.32
All-ages prevalence counts estimates
Iran (Islamic Republic of) Males 25,980.61 21,921.03 31,313.86 64,913.99 54,867.53 76,963.05 1.5 1.36 1.63
Alborz Males 799.12 666.04 960.89 3,004.47 2,512.81 3,580.00 2.76 2.54 2.98
Ardebil Males 526.13 442.47 628.02 1,066.37 896.7 1,266.74 1.03 0.9 1.15
Bushehr Males 232.98 197.84 280.1 746.01 634.99 883.77 2.2 2.03 2.37
Chahar Mahaal and Bakhtiari Males 308.66 261.85 369.18 658.52 554.24 787.27 1.13 1.01 1.24
East Azarbayejan Males 1,727.24 1,439.06 2,099.40 3,566.44 3,006.81 4,224.22 1.06 0.89 1.23
Fars Males 1,842.56 1,541.73 2,227.87 4,527.43 3,782.49 5,373.18 1.46 1.28 1.62
Gilan Males 1,064.00 889.05 1,286.54 2,381.49 2,017.29 2,802.77 1.24 1.08 1.4
Golestan Males 543.91 459.05 656.48 1,396.89 1,169.92 1,669.31 1.57 1.45 1.71
Hamadan Males 1,037.36 870.46 1,245.50 1,819.30 1,531.40 2,154.08 0.75 0.61 0.89
Hormozgan Males 395.57 333.8 470.5 1,367.40 1,133.36 1,646.31 2.46 2.29 2.63
Ilam Males 233.76 196.32 281.15 623.48 518.14 745.61 1.67 1.48 1.86
Isfahan Males 1,583.45 1,347.54 1,896.86 4,184.50 3,512.20 4,998.22 1.64 1.51 1.79
Kerman Males 771.08 646.96 936.16 2,417.17 2,016.64 2,890.05 2.13 1.95 2.33
Kermanshah Males 1,188.21 993.95 1,432.18 1,974.55 1,651.23 2,333.74 0.66 0.54 0.78
Khorasan-e-Razavi Males 1,833.49 1,550.81 2,204.80 4,647.60 3,941.63 5,499.90 1.53 1.37 1.7
Khuzestan Males 1,194.35 1,014.86 1,439.53 3,419.90 2,874.35 4,087.69 1.86 1.72 2.02
Kohgiluyeh and Boyer-Ahmad Males 219.1 182.4 270.9 646.6 532.49 777.9 1.95 1.74 2.16
Kurdistan Males 578.62 491.99 692.14 1,280.78 1,072.53 1,526.19 1.21 1.09 1.34
Lorestan Males 811.01 682.51 967.64 1,543.82 1,288.03 1,833.60 0.9 0.8 1.01
Markazi Males 617.56 518.61 748.5 1,128.17 948.56 1,336.84 0.83 0.7 0.96
Mazandaran Males 919.91 780.92 1,100.78 2,800.21 2,371.82 3,320.58 2.04 1.88 2.23
North Khorasan Males 255.36 216.95 307.29 562.07 477.11 670.29 1.2 1.06 1.35
Qazvin Males 283.28 241.67 340.29 970.2 817.79 1,146.22 2.42 2.23 2.65
Qom Males 309.33 260.62 375.88 992.11 825.86 1,188.43 2.21 2 2.46
Semnan Males 186.59 160 221.95 537.86 454.96 635.85 1.88 1.75 2.03
Sistan and Baluchistan Males 419.79 362.38 494.65 1,494.14 1,261.12 1,780.24 2.56 2.42 2.73
South Khorasan Males 223.13 192.09 261.35 542.42 460.96 643.26 1.43 1.31 1.56
Tehran Males 4,320.45 3,699.32 5,140.16 10,377.91 8,885.78 12,242.71 1.4 1.28 1.51
West Azarbayejan Males 1,044.44 873.45 1,276.68 2,796.59 2,335.12 3,312.87 1.68 1.48 1.9
Yazd Males 245.46 210.04 292.23 751.43 634.95 893.43 2.06 1.92 2.24
Zanjan Males 264.71 226.32 316.18 688.16 584.17 815.08 1.6 1.48 1.76
Iran (Islamic Republic of) Females 39,854.26 32,912.27 49,170.48 57,867.77 48,020.96 70,468.16 0.45 0.36 0.53
Alborz Females 897.71 735.44 1,117.21 2,255.41 1,843.73 2,758.72 1.51 1.38 1.64
Ardebil Females 910.45 751.28 1,133.08 953.21 794.49 1,149.59 0.05 −0.03 0.12
Bushehr Females 571.59 468.55 710.8 827.09 689.69 993.68 0.45 0.32 0.56
Chahar Mahaal and Bakhtiari Females 605.73 496.13 746.28 755.37 622.38 918.81 0.25 0.15 0.33
East Azarbayejan Females 2,376.11 1,938.71 2,969.95 2,899.41 2,405.31 3,533.16 0.22 0.13 0.3
Fars Females 3,092.95 2,512.98 3,879.40 4,191.23 3,453.43 5,132.54 0.36 0.27 0.44
Gilan Females 1,579.18 1,291.76 1,943.30 2,058.57 1,709.06 2,493.13 0.3 0.23 0.38
Golestan Females 1,355.35 1,099.45 1,738.65 1,632.84 1,356.21 1,986.89 0.2 0.09 0.32
Hamadan Females 1,304.44 1,062.72 1,644.82 1,413.44 1,166.56 1,719.03 0.08 0.01 0.16
Hormozgan Females 539.4 446.17 662.17 995.01 829.73 1,212.26 0.84 0.73 0.96
Ilam Females 480.97 388.86 602.86 644.05 532.2 776.04 0.34 0.23 0.45
Isfahan Females 1,864.13 1,541.57 2,294.25 3,176.54 2,668.91 3,838.50 0.7 0.58 0.8
Kerman Females 1,172.83 966.64 1,449.53 2,144.63 1,771.82 2,645.44 0.83 0.73 0.91
Kermanshah Females 2,097.00 1,707.56 2,682.03 2,154.17 1,763.87 2,652.71 0.03 −0.05 0.11
Khorasan-e-Razavi Females 3,331.09 2,728.24 4,131.31 4,232.83 3,510.21 5,115.22 0.27 0.17 0.36
Khuzestan Females 2,340.69 1,917.52 2,916.10 3,636.86 2,971.03 4,474.76 0.55 0.47 0.63
Kohgiluyeh and Boyer-Ahmad Females 416.87 337.4 533.86 687.17 556.58 853.49 0.65 0.54 0.76
Kurdistan Females 1,148.07 937.16 1,447.30 1,215.41 1,010.46 1,471.07 0.06 −0.04 0.14
Lorestan Females 1,554.36 1,273.86 1,954.09 1,804.10 1,488.74 2,188.69 0.16 0.08 0.25
Markazi Females 709.95 589.45 871.35 822.79 694.96 990.55 0.16 0.06 0.23
Mazandaran Females 1,448.63 1,184.90 1,810.57 2,441.21 2,001.92 2,965.36 0.69 0.58 0.79
North Khorasan Females 599.95 494.62 756.91 694.63 575.09 843.12 0.16 0.08 0.23
Qazvin Females 452.71 377.39 554.24 778.39 652.81 944.25 0.72 0.62 0.81
Qom Females 268.75 226.52 323.73 626.73 529.19 758.85 1.33 1.2 1.46
Semnan Females 236.57 197.03 289.77 407.88 342 493.68 0.72 0.62 0.81
Sistan and Baluchistan Females 547.85 460.26 667.46 1,394.56 1,160.24 1,705.03 1.55 1.45 1.63
South Khorasan Females 391.6 327.67 478.47 533.79 445.96 652.73 0.36 0.29 0.42
Tehran Females 4,583.38 3,826.56 5,549.04 8,358.90 7,024.47 10,099.26 0.82 0.71 0.91
West Azarbayejan Females 2,149.33 1,741.77 2,737.79 2,890.52 2,378.10 3,553.10 0.34 0.25 0.44
Yazd Females 390.49 324.71 477.85 630.17 523.85 762.77 0.61 0.51 0.7
Zanjan Females 436.14 357.65 541.26 610.84 510.35 735.01 0.4 0.29 0.49
Age-standardized rate incidence (per 100,000)
Iran (Islamic Republic of) Males 39.72 32.38 48.02 42.9 33.59 53.18 0.08 0 0.15
Alborz Males 48.1 38.96 59.03 54.42 42.69 68.08 0.13 0.03 0.23
Ardebil Males 41.99 34.41 51.28 45.82 35.69 57.3 0.09 −0.02 0.19
Bushehr Males 31.21 25.45 38.25 36.19 28.39 44.41 0.16 0.05 0.28
Chahar Mahaal and Bakhtiari Males 41.53 33.68 50.62 39.02 30.82 48.73 −0.06 −0.15 0.03
East Azarbayejan Males 41.09 33.28 50.27 46.58 36.29 58.7 0.13 0.02 0.27
Fars Males 45.45 37 55.34 48.4 37.83 60.3 0.06 −0.03 0.16
Gilan Males 40.3 32.91 48.8 45.19 35.38 56.99 0.12 0 0.23
Golestan Males 37.94 30.66 46.1 42.16 32.92 53.06 0.11 0.02 0.23
Hamadan Males 52.45 42.43 64.45 55.56 43.28 68.82 0.06 −0.04 0.16
Hormozgan Males 38.5 31.11 46.72 43.79 34.36 54.93 0.14 0.03 0.24
Ilam Males 50.27 41.26 61.28 58.1 45 72.6 0.16 0.04 0.27
Isfahan Males 35.49 28.47 43.13 41.06 32.74 50.77 0.16 0.06 0.26
Kerman Males 36.98 29.85 44.73 41.55 32.1 52.04 0.12 0.01 0.24
Kermanshah Males 61.34 50.34 74.75 53.38 41.59 66.15 −0.13 −0.21 −0.04
Khorasan-e-Razavi Males 33.79 27.52 40.97 40.27 31.53 50.11 0.19 0.08 0.31
Khuzestan Males 36.08 28.98 44.22 41.66 32.7 51.78 0.15 0.07 0.24
Kohgiluyeh and Boyer-Ahmad Males 43.15 34.7 53.23 50.59 38.96 64.31 0.17 0.05 0.31
Kurdistan Males 41.32 33.83 50.2 40.97 31.75 51.37 −0.01 −0.11 0.09
Lorestan Males 48.83 39.9 59.32 51.73 40.34 63.26 0.06 −0.03 0.15
Markazi Males 44 35.67 52.97 40.57 31.82 51.58 −0.08 −0.16 0.01
Mazandaran Males 31.42 25.57 38.43 41.07 32.35 51.67 0.31 0.21 0.43
North Khorasan Males 34.68 28.02 42.38 38.13 30 47.22 0.1 −0.01 0.2
Qazvin Males 26.83 21.79 32.55 39.76 31.53 49.6 0.48 0.35 0.64
Qom Males 36.96 30.05 44.49 40.46 31.41 50.45 0.09 0 0.2
Semnan Males 31.35 25.23 38.06 38.6 29.97 48.37 0.23 0.13 0.33
Sistan and Baluchistan Males 27.74 21.92 34.25 37.47 28.93 47.15 0.35 0.25 0.46
South Khorasan Males 28.03 22.76 33.88 38.28 30.21 47.44 0.37 0.24 0.49
Tehran Males 41.75 33.7 50.54 38.31 29.94 47.4 −0.08 −0.17 0
West Azarbayejan Males 38.43 31.11 47.37 47.1 36.98 58.71 0.23 0.12 0.36
Yazd Males 30.86 25.05 37.32 36.93 28.81 46.11 0.2 0.08 0.32
Zanjan Males 28.29 22.73 34.5 35.28 28.15 43.95 0.25 0.12 0.35
Iran (Islamic Republic of) Females 75.31 60.81 92.21 49.28 38.52 62 −0.35 −0.39 −0.31
Alborz Females 69.93 56.65 86.89 53.88 41.9 68.12 −0.23 −0.31 −0.16
Ardebil Females 90.98 73.02 112.32 52.35 40.71 65.42 −0.42 −0.48 −0.37
Bushehr Females 93.4 75.67 114.33 55.48 43.55 69.35 −0.41 −0.46 −0.35
Chahar Mahaal and Bakhtiari Females 102.83 83.08 126.35 58.52 46.08 73.73 −0.43 −0.48 −0.37
East Azarbayejan Females 70.01 55.34 86.87 48.49 37.42 62.21 −0.31 −0.38 −0.24
Fars Females 94.77 76.48 116.55 57.71 44.92 73.28 −0.39 −0.44 −0.33
Gilan Females 71.35 57.66 87.03 50.4 39.7 63.99 −0.29 −0.36 −0.23
Golestan Females 112.34 90.13 138.03 60.77 47.93 75.37 −0.46 −0.51 −0.41
Hamadan Females 80.68 65.52 99.61 55.05 42.47 69.87 −0.32 −0.39 −0.25
Hormozgan Females 66.64 53.85 82.32 41.61 32.16 52.86 −0.38 −0.43 −0.3
Ilam Females 139.23 113.16 168.87 77.78 61.73 95.92 −0.44 −0.49 −0.39
Isfahan Females 52.12 41.67 65.11 40.98 31.81 51.61 −0.21 −0.29 −0.15
Kerman Females 68.94 55.26 85.3 48.36 37.41 61.5 −0.3 −0.37 −0.22
Kermanshah Females 138.04 112.06 169.9 72.75 57.16 90.73 −0.47 −0.52 −0.42
Khorasan-e-Razavi Females 74.15 59.57 92.46 45.74 35.58 57.57 −0.38 −0.44 −0.33
Khuzestan Females 87.24 69.38 108.88 56.74 43.87 71.17 −0.35 −0.41 −0.29
Kohgiluyeh and Boyer-Ahmad Females 107.69 87.03 132.6 71.09 55.83 89.33 −0.34 −0.4 −0.28
Kurdistan Females 101.26 81.7 122.94 50.66 39.87 64.11 −0.5 −0.55 −0.45
Lorestan Females 119.17 96.36 147.34 75.39 59.05 93.69 −0.37 −0.42 −0.32
Markazi Females 59.36 47.63 73.22 38.29 30.12 48.37 −0.35 −0.41 −0.29
Mazandaran Females 58 46.51 71.38 46.12 35.57 57.86 −0.2 −0.28 −0.13
North Khorasan Females 97.64 78.89 119.7 59.16 46.52 74.68 −0.39 −0.45 −0.34
Qazvin Females 52.13 41.71 64.24 41.25 32.29 52.19 −0.21 −0.29 −0.13
Qom Females 40.48 32.78 50.57 33.92 25.83 43.45 −0.16 −0.26 −0.08
Semnan Females 47.48 38.05 59.29 37.41 28.89 47.26 −0.21 −0.3 −0.14
Sistan and Baluchistan Females 44.55 35.73 54.64 41.62 32 53.52 −0.07 −0.15 0.03
South Khorasan Females 60.06 48.63 74.43 46.46 36.04 58.83 −0.23 −0.3 −0.16
Tehran Females 56.12 45.41 69.69 40.44 31.08 51.82 −0.28 −0.35 −0.21
West Azarbayejan Females 97.25 78.63 119.79 61.7 48.35 76.94 −0.37 −0.43 −0.3
Yazd Females 60.92 48.29 74.42 41.81 32.46 53.5 −0.31 −0.37 −0.25
Zanjan Females 56.57 45.39 69.97 39.65 30.62 50.19 −0.3 −0.37 −0.23
All-ages incidence counts estimates
Iran (Islamic Republic of) Males 10,863.93 8,683.10 13,557.19 19,891.51 15,639.20 24,678.36 0.83 0.61 1.09
Alborz Males 349.13 277.04 435.04 919.83 722.14 1,136.48 1.63 1.3 2.04
Ardebil Males 227.44 178.89 286.2 335.04 260.71 417.72 0.47 0.24 0.75
Bushehr Males 103.53 81.91 129.82 265.07 206.41 327.44 1.56 1.22 1.96
Chahar Mahaal and Bakhtiari Males 135.96 107.26 171.83 212.83 168.2 266.23 0.57 0.35 0.82
East Azarbayejan Males 696.7 552.65 887.62 1,035.48 810.01 1,287.90 0.49 0.26 0.77
Fars Males 787.85 623.31 989.99 1,385.60 1,080.34 1,737.76 0.76 0.52 1.04
Gilan Males 451.4 360.52 562.08 637.71 504 794.34 0.41 0.19 0.66
Golestan Males 229.91 179.83 287.42 431.8 338.18 538.01 0.88 0.64 1.13
Hamadan Males 439.32 344.46 555.95 523.41 408.46 649.82 0.19 0.02 0.39
Hormozgan Males 160.93 126.8 200.21 480.39 370.69 606.76 1.99 1.59 2.38
Ilam Males 105.95 83.72 133.72 200.72 155.85 251.21 0.89 0.6 1.23
Isfahan Males 671.02 528.33 836.46 1,181.98 935.71 1,457.97 0.76 0.51 1.05
Kerman Males 317.11 246.88 391.76 785.62 604.24 981.57 1.48 1.13 1.88
Kermanshah Males 503.87 403.65 633.65 587.22 460.29 726.85 0.17 0 0.38
Khorasan-e-Razavi Males 730.31 582.84 914.52 1,480.67 1,164.70 1,829.01 1.03 0.74 1.36
Khuzestan Males 527.11 411.93 672.32 1,109.54 861.75 1,382.74 1.1 0.85 1.41
Kohgiluyeh and Boyer-Ahmad Males 100.74 78.91 129.96 223.81 171.48 285.39 1.22 0.86 1.62
Kurdistan Males 239.46 190.86 298.92 395.72 304.63 494.97 0.65 0.4 0.95
Lorestan Males 355.86 281.4 447.12 506.42 395.45 624.25 0.42 0.25 0.65
Markazi Males 253.53 200.4 317.66 321.6 252.86 400.58 0.27 0.09 0.47
Mazandaran Males 390.13 308.76 487.34 780.42 615.57 967.57 1 0.72 1.35
North Khorasan Males 102.75 81.65 129.89 175.36 136.51 218.26 0.71 0.47 0.98
Qazvin Males 119.73 94.42 151.06 301.01 237.57 374.02 1.51 1.14 2
Qom Males 132.81 105.17 163.95 316.62 245.77 394.79 1.38 1.06 1.73
Semnan Males 73.01 57.85 90.05 173.62 133.57 217.61 1.38 1.1 1.7
Sistan and Baluchistan Males 169.5 130.86 215.52 595.14 452.38 764.92 2.51 2.15 2.88
South Khorasan Males 81.25 64.94 100.04 181.89 142.49 226.33 1.24 0.98 1.55
Tehran Males 1,752.87 1,391.40 2,166.57 3,005.67 2,368.74 3,694.62 0.71 0.51 0.95
West Azarbayejan Males 441.43 347.17 562.15 889.18 698.07 1,109.11 1.01 0.74 1.35
Yazd Males 99.38 78.97 125 239.72 185.31 297.94 1.41 1.12 1.75
Zanjan Males 113.93 89.73 142.92 212.42 168.78 265.49 0.86 0.54 1.18
Iran (Islamic Republic of) Females 21,219.21 16,839.31 26,855.28 21,017.07 16,503.98 26,128.95 −0.01 −0.11 0.12
Alborz Females 503.71 396.1 628.71 818.8 639.42 1,013.54 0.63 0.43 0.87
Ardebil Females 498.76 389.38 629.6 339.64 268.61 422.02 −0.32 −0.41 −0.21
Bushehr Females 325.46 258.8 412.46 353.89 276.8 443.47 0.09 −0.04 0.24
Chahar Mahaal and Bakhtiari Females 348.91 273 436.97 294.38 232 367.34 −0.16 −0.27 −0.03
East Azarbayejan Females 1,208.35 934.52 1,536.14 954.99 740.22 1,186.94 −0.21 −0.31 −0.1
Fars Females 1,680.46 1,330.32 2,108.27 1,481.23 1,167.81 1,854.87 −0.12 −0.22 0.01
Gilan Females 838.7 668.31 1,040.75 643.01 510.03 785.47 −0.23 −0.33 −0.12
Golestan Females 732.93 577.22 929.77 594.32 468.88 729.33 −0.19 −0.29 −0.07
Hamadan Females 669.53 528.38 850.1 465.04 363.84 578.97 −0.31 −0.4 −0.19
Hormozgan Females 288.03 227.28 365.29 421.36 327.17 525.13 0.46 0.29 0.67
Ilam Females 293.75 234.37 372.99 249.02 197.19 307.94 −0.15 −0.28 0
Isfahan Females 974.47 765.58 1,248.17 1,090.00 852.83 1,341.85 0.12 −0.03 0.3
Kerman Females 611.74 482.85 780.36 809.32 632.07 1,023.44 0.32 0.16 0.53
Kermanshah Females 1,130.91 888.73 1,409.42 728.65 576.29 904.08 −0.36 −0.44 −0.26
Khorasan-e-Razavi Females 1,698.79 1,340.32 2,177.33 1,576.79 1,231.59 1,974.26 −0.07 −0.19 0.04
Khuzestan Females 1,316.03 1,032.12 1,684.08 1,408.26 1,089.00 1,753.13 0.07 −0.06 0.21
Kohgiluyeh and Boyer-Ahmad Females 256.64 201.84 328.19 282.52 219.99 354.32 0.1 −0.05 0.28
Kurdistan Females 614.05 482.59 767.4 446.05 354.05 556.64 −0.27 −0.38 −0.16
Lorestan Females 892.95 704.49 1,138.94 682.78 535.33 843.42 −0.24 −0.34 −0.12
Markazi Females 358.74 281.78 458.42 279.89 221.22 348.1 −0.22 −0.33 −0.09
Mazandaran Females 779.54 616.65 980.67 796.95 622.31 975.06 0.02 −0.11 0.19
North Khorasan Females 303.61 238.02 379.34 254.41 201.06 316.96 −0.16 −0.26 −0.05
Qazvin Females 241.75 189.38 305.89 279.86 219.89 347.65 0.16 0 0.36
Qom Females 144.86 112.99 185.03 249.94 191.72 314.54 0.73 0.49 0.98
Semnan Females 110.84 87.3 141.23 154.39 118.55 194.87 0.39 0.2 0.59
Sistan and Baluchistan Females 298.76 233.08 384.36 653.6 495.96 847.86 1.19 0.97 1.45
South Khorasan Females 187.09 148.51 238.84 201.48 157.76 253.72 0.08 −0.03 0.2
Tehran Females 2,338.70 1,866.34 2,988.74 2,981.26 2,316.35 3,738.27 0.27 0.12 0.47
West Azarbayejan Females 1,138.65 893.93 1,434.19 1,059.09 840.84 1,303.36 −0.07 −0.19 0.06
Yazd Females 191.87 149.79 241.73 249.24 193.39 314.46 0.3 0.16 0.45
Zanjan Females 240.65 189.64 307.1 216.89 168.89 269.18 −0.1 −0.22 0.05
Age-standardized rate DALYs (per 100,000)
Iran (Islamic Republic of) Males 394.97 339.53 445.59 325.29 277.26 357.39 −0.18 −0.28 −0.03
Alborz Males 694.07 289.42 940.95 568.33 270.5 740.65 −0.18 −0.44 0.29
Ardebil Males 437.62 324.74 559.98 427.56 283.87 514.44 −0.02 −0.31 0.35
Bushehr Males 255.66 188.19 363.54 185.59 145.2 236.09 −0.27 −0.48 0.02
Chahar Mahaal and Bakhtiari Males 378.01 278.62 484.44 233.23 183.4 295.33 −0.38 −0.56 −0.08
East Azarbayejan Males 529.26 367.79 700.6 400.04 291.82 502.15 −0.24 −0.45 0.07
Fars Males 427.21 321.88 568.62 452.77 340.94 578.21 0.06 −0.27 0.5
Gilan Males 446.55 325.39 591.94 374.08 278.67 478.36 −0.16 −0.41 0.18
Golestan Males 374.4 278.02 490.89 389.87 310.2 486.92 0.04 −0.24 0.42
Hamadan Males 740.03 449.09 976.54 667.57 366.81 849.16 −0.1 −0.37 0.23
Hormozgan Males 443.71 318.94 591.52 426.21 321.3 542.79 −0.04 −0.33 0.42
Ilam Males 803.7 358.86 1,068.51 783.79 363.66 981.94 −0.02 −0.3 0.39
Isfahan Males 226.08 166.18 348.91 227.97 177.84 302.1 0.01 −0.31 0.5
Kerman Males 431.89 331.87 578.97 321.77 255.42 417.4 −0.25 −0.47 0.09
Kermanshah Males 831.54 543.57 1,084.82 643.3 361.75 802.83 −0.23 −0.45 0.07
Khorasan-e-Razavi Males 380.64 283.59 508.14 280.69 222.07 351.07 −0.26 −0.49 0.05
Khuzestan Males 327.9 248.49 444.45 332.82 265 409.69 0.01 −0.26 0.43
Kohgiluyeh and Boyer-Ahmad Males 550.17 389.45 714.78 538.36 386.37 669.28 −0.02 −0.31 0.42
Kurdistan Males 396.96 283.42 525.24 313.95 244.99 393.81 −0.21 −0.45 0.19
Lorestan Males 706.38 373.4 925.06 506.66 268.65 637.51 −0.28 −0.48 0.02
Markazi Males 420.77 313.47 544.22 286.03 224.74 361.11 −0.32 −0.53 −0.05
Mazandaran Males 213.73 158.55 329.37 225.5 178.65 310.47 0.06 −0.26 0.54
North Khorasan Males 310.01 232.57 425.74 279.62 224.36 350.7 −0.1 −0.35 0.3
Qazvin Males 238.23 180.7 360.25 220.53 169.43 282.36 −0.07 −0.35 0.34
Qom Males 367.47 276.03 484.11 325.16 251.69 413.66 −0.12 −0.39 0.27
Semnan Males 289.04 219.31 384.44 192.44 148.03 256.73 −0.33 −0.53 −0.03
Sistan and Baluchistan Males 224.44 160.99 362.99 294.01 223.81 421.41 0.31 −0.12 0.88
South Khorasan Males 236.57 173.87 343.12 221.15 178.26 280.61 −0.07 −0.33 0.34
Tehran Males 252.38 184.91 333.36 163.33 127.01 219.3 −0.35 −0.54 −0.03
West Azarbayejan Males 509.17 372.31 662.57 408.53 289.07 502.87 −0.2 −0.43 0.14
Yazd Males 221.89 164.94 358.27 173.3 129.73 239.73 −0.22 −0.45 0.14
Zanjan Males 253.86 193.33 330.9 184.33 144.28 242.46 −0.27 −0.48 0.06
Iran (Islamic Republic of) Females 303.11 202.92 343.37 124.44 108.73 155.18 −0.59 −0.66 −0.42
Alborz Females 273.44 153.68 390.8 139.71 98.2 182.74 −0.49 −0.68 −0.08
Ardebil Females 380.15 226.79 510.12 146.79 106.84 178.62 −0.61 −0.72 −0.4
Bushehr Females 426.38 194.12 599.11 155.23 97.02 193.26 −0.64 −0.75 −0.42
Chahar Mahaal and Bakhtiari Females 425.11 178.38 591.25 132.98 71.17 182.46 −0.69 −0.8 −0.49
East Azarbayejan Females 298.01 190.39 411.07 124.41 93.01 167.63 −0.58 −0.73 −0.32
Fars Females 385.15 210.57 522.91 178.03 132.2 233.03 −0.54 −0.68 −0.2
Gilan Females 276.55 171.73 378.93 110.85 84.48 144.12 −0.6 −0.74 −0.36
Golestan Females 601.83 227.7 837.06 226.11 121.39 296.84 −0.62 −0.75 −0.42
Hamadan Females 364.15 203.55 488.33 154.19 112.64 196.73 −0.58 −0.71 −0.34
Hormozgan Females 262.49 178.8 363.78 85.31 64.19 140.85 −0.67 −0.78 −0.42
Ilam Females 954.67 239.66 1,313.03 338.08 122.5 432.97 −0.65 −0.75 −0.38
Isfahan Females 126.42 89.13 223.27 59.08 41.49 128.59 −0.53 −0.69 −0.29
Kerman Females 321.43 218.47 460.37 137.46 101.5 206.1 −0.57 −0.73 −0.31
Kermanshah Females 733.54 273.89 1,038.41 307.61 146.28 414.09 −0.58 −0.71 −0.34
Khorasan-e-Razavi Females 328.9 211.97 445.44 114.95 87.06 160.73 −0.65 −0.77 −0.39
Khuzestan Females 364.66 243.34 504.77 171.33 129.64 219.08 −0.53 −0.68 −0.28
Kohgiluyeh and Boyer-Ahmad Females 594.83 226.95 820.87 323.95 175.16 419.69 −0.46 −0.62 −0.17
Kurdistan Females 445.34 228.6 604.68 131.16 91.25 167.77 −0.71 −0.8 −0.52
Lorestan Females 740.93 186.48 1,033.89 243.63 90.3 318.12 −0.67 −0.77 −0.43
Markazi Females 169.85 115.91 290.92 59.69 42.24 141.3 −0.65 −0.77 −0.46
Mazandaran Females 182.45 126.12 267.71 90.27 67.65 148.97 −0.51 −0.69 −0.21
North Khorasan Females 437.35 262.32 584.12 196.46 142.49 252.85 −0.55 −0.7 −0.27
Qazvin Females 207.87 150.08 286.34 75.01 53.73 111.44 −0.64 −0.76 −0.39
Qom Females 86.97 53.75 229.17 31.34 19.87 107.53 −0.64 −0.77 −0.47
Semnan Females 139.16 99.41 210.76 46.91 32.94 95.89 −0.66 −0.78 −0.45
Sistan and Baluchistan Females 152.18 100.86 227.23 100.09 68.32 256.9 −0.34 −0.61 0.27
South Khorasan Females 209.04 140.87 286.51 92.16 72.32 139.86 −0.56 −0.68 −0.32
Tehran Females 83.99 52.92 187.24 49.94 34.47 118.61 −0.41 −0.62 −0.07
West Azarbayejan Females 568.94 298.05 785.6 257.42 173.48 326.7 −0.55 −0.69 −0.34
Yazd Females 177.79 129.8 246.98 65.08 46.07 111.03 −0.63 −0.76 −0.44
Zanjan Females 212.63 148.03 290.36 74.54 56.97 111.17 −0.65 −0.77 −0.4
All-ages DALYs counts estimates
Iran (Islamic Republic of) Males 105,109.67 88,800.06 118,654.48 150,642.85 129,386.56 165,530.23 0.43 0.25 0.68
Alborz Males 4,785.14 1,977.87 6,546.54 9,620.22 4,583.93 12,427.56 1.01 0.39 2.16
Ardebil Males 2,238.31 1,566.58 2,962.58 3,140.48 2,135.42 3,791.58 0.4 0.02 0.95
Bushehr Males 792.73 581.01 1,105.82 1,353.29 1,056.52 1,733.32 0.71 0.21 1.43
Chahar Mahaal and Bakhtiari Males 1,168.91 852.52 1,513.74 1,278.57 1,001.65 1,637.58 0.09 −0.23 0.65
East Azarbayejan Males 8,819.41 5,859.44 11,866.60 8,886.08 6,519.33 11,162.29 0.01 −0.29 0.44
Fars Males 7,240.32 5,363.35 9,567.00 12,688.20 9,659.50 16,177.91 0.75 0.22 1.47
Gilan Males 4,827.20 3,470.78 6,426.42 5,302.00 3,935.33 6,774.20 0.1 −0.22 0.55
Golestan Males 2,158.03 1,596.81 2,788.46 3,929.64 3,140.74 4,904.34 0.82 0.33 1.48
Hamadan Males 5,909.33 3,462.24 7,788.94 6,295.59 3,462.19 7,975.75 0.07 −0.26 0.44
Hormozgan Males 1,757.61 1,257.63 2,384.00 4,619.52 3,488.14 5,866.08 1.63 0.81 2.94
Ilam Males 1,562.63 656.45 2,106.58 2,700.39 1,275.84 3,377.95 0.73 0.23 1.41
Isfahan Males 4,300.99 3,121.74 6,405.25 6,536.09 5,108.41 8,657.84 0.52 0.05 1.26
Kerman Males 3,543.70 2,688.95 4,708.34 6,014.05 4,802.83 7,730.71 0.7 0.2 1.44
Kermanshah Males 6,617.79 4,107.53 8,759.14 7,130.22 4,071.72 8,845.19 0.08 −0.24 0.5
Khorasan-e-Razavi Males 7,922.77 5,805.93 10,472.49 10,315.00 8,102.14 12,809.88 0.3 −0.09 0.84
Khuzestan Males 4,573.24 3,421.02 6,045.35 8,760.33 7,013.53 10,703.06 0.92 0.39 1.68
Kohgiluyeh and Boyer-Ahmad Males 1,283.36 875.53 1,714.53 2,415.38 1,762.24 3,025.18 0.88 0.31 1.76
Kurdistan Males 2,213.84 1,575.17 2,953.90 3,032.71 2,365.45 3,793.67 0.37 −0.05 1.06
Lorestan Males 4,735.00 2,375.03 6,287.15 5,044.98 2,681.32 6,399.49 0.07 −0.24 0.53
Markazi Males 2,384.36 1,755.07 3,084.16 2,283.17 1,785.47 2,874.05 −0.04 −0.32 0.36
Mazandaran Males 2,627.34 1,935.90 3,921.82 4,261.76 3,402.84 5,691.96 0.62 0.15 1.37
North Khorasan Males 887.39 663.14 1,252.67 1,290.11 1,028.98 1,638.99 0.45 0.04 1.06
Qazvin Males 988.27 736.08 1,434.88 1,677.70 1,287.92 2,131.81 0.7 0.2 1.45
Qom Males 1,269.80 952.6 1,663.77 2,540.10 1,979.52 3,210.06 1 0.4 1.87
Semnan Males 643.45 482.9 858.41 867.68 669.82 1,149.98 0.35 −0.04 0.99
Sistan and Baluchistan Males 1,350.61 950.06 2,021.48 4,677.26 3,526.83 6,687.79 2.46 1.35 4.16
South Khorasan Males 665.21 479.13 944.35 1,054.71 848.67 1,327.53 0.59 0.14 1.27
Tehran Males 10,485.67 7,640.32 13,685.69 12,997.11 10,090.36 17,426.77 0.24 −0.13 0.84
West Azarbayejan Males 5,702.28 4,097.11 7,618.92 7,695.77 5,446.80 9,502.46 0.35 −0.05 0.95
Yazd Males 691.59 505.54 1,090.36 1,107.88 837.76 1,536.14 0.6 0.1 1.34
Zanjan Males 963.33 716.54 1,284.21 1,126.86 886.07 1,474.55 0.17 −0.18 0.72
Iran (Islamic Republic of) Females 83,677.70 53,304.15 95,584.67 52,614.69 45,850.79 66,073.28 −0.37 −0.49 −0.07
Alborz Females 1,902.17 1,030.34 2,777.59 2,074.58 1,468.43 2,719.96 0.09 −0.31 1.03
Ardebil Females 2,060.16 1,172.19 2,818.16 955.16 695.95 1,171.13 −0.54 −0.68 −0.27
Bushehr Females 1,433.61 634.1 2,033.64 989.4 623.31 1,229.33 −0.31 −0.53 0.12
Chahar Mahaal and Bakhtiari Females 1,403.85 548.8 1,991.43 680.1 367.91 918.53 −0.52 −0.69 −0.19
East Azarbayejan Females 5,021.42 3,137.44 6,997.12 2,415.13 1,825.18 3,279.25 −0.52 −0.68 −0.19
Fars Females 6,614.35 3,453.55 9,141.91 4,494.10 3,334.20 5,960.97 −0.32 −0.54 0.26
Gilan Females 3,159.35 1,882.83 4,445.75 1,442.61 1,100.67 1,878.44 −0.54 −0.71 −0.25
Golestan Females 3,865.98 1,413.45 5,434.44 2,256.45 1,225.93 2,975.43 −0.42 −0.62 −0.06
Hamadan Females 2,931.52 1,608.62 3,991.56 1,283.19 963.08 1,618.84 −0.56 −0.7 −0.3
Hormozgan Females 1,111.61 736.35 1,567.15 858.09 644.95 1,432.46 −0.23 −0.5 0.43
Ilam Females 1,934.65 456.1 2,695.96 1,107.88 403.16 1,441.32 −0.43 −0.6 0.03
Isfahan Females 2,300.84 1,605.67 3,919.77 1,558.55 1,098.17 3,480.22 −0.32 −0.56 0.04
Kerman Females 2,756.65 1,823.94 4,029.96 2,290.01 1,702.28 3,449.73 −0.17 −0.48 0.37
Kermanshah Females 5,899.59 2,165.86 8,473.20 3,026.95 1,455.77 4,065.35 −0.49 −0.65 −0.14
Khorasan-e-Razavi Females 7,289.02 4,593.40 9,929.28 3,928.37 2,964.65 5,520.13 −0.46 −0.65 −0.01
Khuzestan Females 5,332.78 3,422.79 7,422.09 4,189.12 3,217.65 5,371.54 −0.21 −0.48 0.22
Kohgiluyeh and Boyer-Ahmad Females 1,418.10 520.69 1,949.37 1,251.51 698.94 1,614.59 −0.12 −0.41 0.46
Kurdistan Females 2,671.71 1,306.05 3,624.85 1,156.29 829.01 1,478.96 −0.57 −0.7 −0.28
Lorestan Females 5,397.03 1,277.35 7,642.97 2,246.37 850.28 2,932.28 −0.58 −0.71 −0.24
Markazi Females 991.16 668.58 1,656.46 433.62 309.85 1,035.31 −0.56 −0.71 −0.3
Mazandaran Females 2,368.52 1,587.49 3,453.79 1,567.71 1,176.28 2,641.40 −0.34 −0.59 0.06
North Khorasan Females 1,340.82 771.66 1,829.43 833.94 613.13 1,061.11 −0.38 −0.58 0.02
Qazvin Females 932.49 651.71 1,276.84 516.13 369.8 773.08 −0.45 −0.64 −0.06
Qom Females 296.17 177.94 780.29 228.53 144.94 791.67 −0.23 −0.53 0.17
Semnan Females 314.76 220.79 462.81 194.31 136.6 401.39 −0.38 −0.61 0.03
Sistan and Baluchistan Females 979.05 613.79 1,443.46 1,556.90 1,054.02 4,003.73 0.59 −0.09 2.55
South Khorasan Females 631.96 423.94 864.7 398.22 312.79 610.75 −0.37 −0.54 0
Tehran Females 3,351.44 2,067.81 7,478.63 3,564.46 2,481.47 8,516.25 0.06 −0.33 0.74
West Azarbayejan Females 6,546.79 3,219.58 9,091.93 4,320.23 3,012.45 5,442.08 −0.34 −0.56 −0.01
Yazd Females 539.19 392.81 755.47 385.07 271.44 662.05 −0.29 −0.55 0.11
Zanjan Females 880.97 582.36 1,226.49 411.7 316.38 625.77 −0.53 −0.7 −0.17

Figure 5.

Figure 5.

Sex-specific prevalence of self-harm in Iran, 1990–2021.

The incidence of self-harm in males and females has had a different trend. Since 1990, the incidence of self-harm in males has shown a relatively stable trend with a slight increase. In contrast, a decreasing trend can be seen in females; however, in 2020, the incidence was close to that of males, and in 2021, it shows an increasing trend again (Table 5 and Figure 6). In 2021, age-standardized rate of suicide mortality (per 100,000) in males was 6.04 (95% UI: 5.14–6.67), and in females was 2.14 (95% UI: 1.85–2.74). The suicide mortality rate in males compared to females has been higher since 1990 and has increased over time. In 2021, there were 2,398 suicide mortality among males, compared with 906 suicide mortality among females (Table 6 and Figure 7).

Figure 6.

Figure 6.

Sex-specific incidence of self-harm in Iran, 1990–2021.

Table 6.

Sex-specific suicide mortality in Iran, stratified by provinces, 1990–2021

Location Sex Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age-standardized rate suicide mortality (per 100,000)
Iran (Islamic Republic of) Males 7.42 6.35 8.39 6.04 5.14 6.67 −0.19 −0.29 −0.03
Alborz Males 13.42 5.54 18.33 10.67 4.94 13.88 −0.21 −0.44 0.26
Ardebil Males 8.26 6.12 10.58 8.33 5.52 9.96 0.01 −0.28 0.41
Bushehr Males 4.94 3.66 7.21 3.69 2.9 4.71 −0.25 −0.47 0.04
Chahar Mahaal and Bakhtiari Males 7.3 5.36 9.34 4.57 3.6 5.77 −0.37 −0.55 −0.05
East Azarbayejan Males 9.76 6.97 12.94 7.57 5.48 9.5 −0.22 −0.45 0.11
Fars Males 7.92 5.98 10.55 8.09 6.11 10.31 0.02 −0.29 0.46
Gilan Males 8.66 6.43 11.48 7.06 5.2 8.98 −0.18 −0.42 0.15
Golestan Males 7.09 5.29 9.49 7.21 5.75 9.1 0.02 −0.26 0.4
Hamadan Males 13.75 8.37 18.23 12.31 6.64 15.66 −0.1 −0.38 0.22
Hormozgan Males 8.47 6.18 11.37 7.68 5.8 9.75 −0.09 −0.36 0.32
Ilam Males 15.56 6.84 20.51 14.67 6.6 18.34 −0.06 −0.32 0.34
Isfahan Males 4.12 3.01 6.56 4.12 3.21 5.4 0 −0.32 0.5
Kerman Males 8.09 6.17 10.97 5.93 4.69 7.67 −0.27 −0.48 0.07
Kermanshah Males 15.59 10.24 20.15 11.89 6.73 14.85 −0.24 −0.46 0.06
Khorasan-e-Razavi Males 7.28 5.46 9.64 5.34 4.18 6.68 −0.27 −0.49 0.05
Khuzestan Males 6.13 4.62 8.38 6.09 4.89 7.4 −0.01 −0.28 0.39
Kohgiluyeh and Boyer-Ahmad Males 9.57 6.86 12.39 9.35 6.73 11.6 −0.02 −0.31 0.39
Kurdistan Males 7.49 5.36 9.98 5.77 4.52 7.28 −0.23 −0.47 0.16
Lorestan Males 13.56 7.25 17.69 10.1 5.3 12.58 −0.26 −0.46 0.05
Markazi Males 7.74 5.72 10.18 5.22 4.1 6.63 −0.33 −0.52 −0.03
Mazandaran Males 3.94 2.9 6.3 4.15 3.3 5.64 0.05 −0.26 0.53
North Khorasan Males 5.79 4.34 8.22 5.16 4.14 6.48 −0.11 −0.36 0.3
Qazvin Males 4.63 3.48 7.16 4.3 3.35 5.47 −0.07 −0.35 0.33
Qom Males 6.96 5.26 9.26 5.96 4.6 7.54 −0.14 −0.42 0.21
Semnan Males 5.59 4.21 7.53 3.71 2.87 4.89 −0.34 −0.53 −0.03
Sistan and Baluchistan Males 4.22 3.02 7.45 5.28 4 7.7 0.25 −0.15 0.82
South Khorasan Males 4.45 3.28 6.51 4.13 3.32 5.23 −0.07 −0.35 0.35
Tehran Males 4.76 3.48 6.41 3.11 2.4 4.14 −0.35 −0.54 −0.01
West Azarbayejan Males 9.49 6.91 12.27 7.7 5.45 9.44 −0.19 −0.42 0.15
Yazd Males 4.28 3.17 7.09 3.22 2.39 4.49 −0.25 −0.47 0.09
Zanjan Males 4.89 3.72 6.53 3.6 2.85 4.76 −0.26 −0.47 0.06
Iran (Islamic Republic of) Females 5.04 3.47 5.71 2.14 1.85 2.74 −0.58 −0.65 −0.4
Alborz Females 4.71 2.58 6.69 2.49 1.74 3.27 −0.47 −0.66 0.01
Ardebil Females 6.29 3.65 8.47 2.59 1.88 3.2 −0.59 −0.71 −0.35
Bushehr Females 7.35 3.15 10.38 2.86 1.73 3.55 −0.61 −0.73 −0.37
Chahar Mahaal and Bakhtiari Females 7.21 2.78 9.91 2.38 1.18 3.23 −0.67 −0.78 −0.45
East Azarbayejan Females 4.89 3.13 6.83 2.1 1.55 2.87 −0.57 −0.72 −0.27
Fars Females 6.45 3.53 8.82 3.02 2.22 4.01 −0.53 −0.69 −0.19
Gilan Females 4.78 2.98 6.55 1.99 1.48 2.61 −0.58 −0.73 −0.33
Golestan Females 10.02 3.54 13.97 4.1 2.09 5.35 −0.59 −0.72 −0.36
Hamadan Females 6.03 3.22 8.16 2.61 1.93 3.33 −0.57 −0.7 −0.31
Hormozgan Females 4.3 2.84 6.02 1.48 1.08 2.62 −0.66 −0.77 −0.38
Ilam Females 16.74 3.82 23.06 6.54 2.21 8.38 −0.61 −0.72 −0.29
Isfahan Females 2.05 1.4 3.89 0.98 0.66 2.34 −0.52 −0.69 −0.27
Kerman Females 5.36 3.61 7.65 2.33 1.71 3.69 −0.57 −0.73 −0.29
Kermanshah Females 12.39 4.33 17.53 5.33 2.43 7.17 −0.57 −0.71 −0.3
Khorasan-e-Razavi Females 5.44 3.49 7.41 1.98 1.5 2.77 −0.64 −0.76 −0.36
Khuzestan Females 6.12 4.09 8.5 2.94 2.27 3.78 −0.52 −0.67 −0.25
Kohgiluyeh and Boyer-Ahmad Females 9.77 3.55 13.42 5.37 2.89 6.96 −0.45 −0.61 −0.13
Kurdistan Females 7.28 3.58 10.06 2.34 1.61 3.04 −0.68 −0.78 −0.47
Lorestan Females 12.72 2.92 17.86 4.46 1.48 5.85 −0.65 −0.76 −0.39
Markazi Females 2.74 1.84 5.02 0.98 0.68 2.56 −0.64 −0.77 −0.43
Mazandaran Females 3.05 2.07 4.58 1.55 1.14 2.7 −0.49 −0.68 −0.19
North Khorasan Females 7.22 4.35 9.72 3.39 2.48 4.36 −0.53 −0.68 −0.2
Qazvin Females 3.48 2.51 4.82 1.3 0.93 2.01 −0.63 −0.76 −0.35
Qom Females 1.41 0.83 4.09 0.52 0.31 2.05 −0.63 −0.78 −0.44
Semnan Females 2.28 1.59 3.71 0.78 0.53 1.72 −0.66 −0.79 −0.42
Sistan and Baluchistan Females 2.47 1.58 4.04 1.61 1.06 4.46 −0.35 −0.62 0.2
South Khorasan Females 3.48 2.36 4.77 1.58 1.22 2.45 −0.55 −0.69 −0.3
Tehran Females 1.37 0.81 3.29 0.82 0.55 2.1 −0.4 −0.63 −0.07
West Azarbayejan Females 9.46 4.98 13.12 4.38 3 5.59 −0.54 −0.68 −0.32
Yazd Females 2.98 2.12 4.2 1.1 0.77 1.99 −0.63 −0.76 −0.43
Zanjan Females 3.55 2.46 4.77 1.3 0.98 2.03 −0.63 −0.76 −0.38
All-ages counts estimates
Iran (Islamic Republic of) Males 1,779.40 1,522.33 2,017.04 2,802.33 2,398.13 3,090.06 0.57 0.38 0.87
Alborz Males 82.52 33.68 112.88 181.11 84.22 235.1 1.19 0.52 2.43
Ardebil Males 37.7 26.98 49.02 60.47 40.73 73.02 0.6 0.16 1.24
Bushehr Males 13.5 9.83 19.07 25.43 19.86 32.9 0.88 0.36 1.67
Chahar Mahaal and Bakhtiari Males 20.06 14.73 25.74 24.59 19.31 31.53 0.23 −0.13 0.85
East Azarbayejan Males 146.65 99.22 196.33 169.72 123.17 213.62 0.16 −0.18 0.68
Fars Males 120.67 90.2 159.89 228.59 173.45 290.49 0.89 0.33 1.68
Gilan Males 84.84 61.25 113.09 104.35 76.56 132.17 0.23 −0.13 0.73
Golestan Males 36.67 27.04 48.18 72.12 57.49 90.58 0.97 0.43 1.7
Hamadan Males 99.5 59 131.23 119.22 64.24 151.64 0.2 −0.17 0.64
Hormozgan Males 30.28 21.58 40.85 80.55 60.46 101.58 1.66 0.85 3.02
Ilam Males 26.7 11.28 35.64 49.62 22.89 62.18 0.86 0.33 1.62
Isfahan Males 70.69 51.48 108.4 121.2 94.64 159.93 0.71 0.17 1.53
Kerman Males 60 45.49 80.58 108.66 85.91 140.07 0.81 0.27 1.64
Kermanshah Males 111.58 70.41 146.73 133.1 75.68 165.75 0.19 −0.15 0.68
Khorasan-e-Razavi Males 136.15 100.55 182.87 193.01 150.75 239.63 0.42 −0.01 1
Khuzestan Males 76.57 57.33 103.55 158.01 127.39 191.2 1.06 0.5 1.89
Kohgiluyeh and Boyer-Ahmad Males 20.23 14.03 26.55 41.58 30.32 52 1.06 0.44 2
Kurdistan Males 37.57 26.45 50.24 55.57 43.37 70.26 0.48 0.01 1.25
Lorestan Males 81.28 41.66 107.26 97.83 51.14 123.76 0.2 −0.15 0.71
Markazi Males 39.89 29.49 51.65 42.59 33.34 54.4 0.07 −0.24 0.52
Mazandaran Males 43.78 31.84 67.74 80.84 64.1 108.27 0.85 0.31 1.71
North Khorasan Males 14.97 11.18 20.68 23.6 18.8 29.88 0.58 0.11 1.26
Qazvin Males 17.05 12.75 25.37 32.35 24.8 41.1 0.9 0.34 1.74
Qom Males 21.34 16.02 28.15 46.05 35.76 58.16 1.16 0.49 2.16
Semnan Males 11.33 8.53 15.14 16.38 12.63 21.61 0.45 0.03 1.14
Sistan and Baluchistan Males 22.87 16.19 36.01 79.56 59.88 114.86 2.48 1.31 4.18
South Khorasan Males 11.56 8.39 16.9 19.29 15.55 24.44 0.67 0.18 1.39
Tehran Males 180.39 130.55 240.71 252.6 194.38 338.41 0.4 −0.04 1.09
West Azarbayejan Males 94.39 68.78 124.41 142.16 100.73 176.06 0.51 0.06 1.17
Yazd Males 11.97 8.82 19.34 20.3 15.31 28.42 0.7 0.17 1.48
Zanjan Males 16.7 12.55 22.13 21.85 17.21 28.7 0.31 −0.07 0.89
Iran (Islamic Republic of) Females 1,290.26 820.94 1,468.85 906.23 779.55 1,182.62 −0.3 −0.43 0.04
Alborz Females 29.73 15.9 43.27 35.98 24.9 48.23 0.21 −0.26 1.39
Ardebil Females 31.41 17.52 42.92 17.14 12.48 21.23 −0.45 −0.62 −0.12
Bushehr Females 22.43 9.57 32.18 17.49 10.71 21.9 −0.22 −0.47 0.3
Chahar Mahaal and Bakhtiari Females 21.77 8.15 31.02 12.2 6.09 16.57 −0.44 −0.64 −0.03
East Azarbayejan Females 76.23 46.51 106.41 40.97 30.08 57.76 −0.46 −0.66 −0.06
Fars Females 102.51 52.84 141.62 76.9 56.31 101.42 −0.25 −0.5 0.42
Gilan Females 50.56 30.4 70.31 27.02 20.1 35.61 −0.47 −0.67 −0.1
Golestan Females 59.63 20.74 83.94 40.82 21.09 53.83 −0.32 −0.55 0.11
Hamadan Females 45.16 24.1 61.99 22.37 16.4 28.08 −0.5 −0.67 −0.19
Hormozgan Females 16.96 11.04 24 14.27 10.33 25.47 −0.16 −0.46 0.58
Ilam Females 30.44 6.76 42.42 21.16 7.34 27.69 −0.3 −0.51 0.27
Isfahan Females 34.73 23.73 63.22 25.92 17.38 64.09 −0.25 −0.54 0.17
Kerman Females 42.64 28.13 62.27 38.15 27.94 61.45 −0.11 −0.45 0.48
Kermanshah Females 91.75 32.05 132.64 53.57 24.88 71.51 −0.42 −0.61 0
Khorasan-e-Razavi Females 111.84 70.05 153.61 66.83 50.25 94.77 −0.4 −0.61 0.1
Khuzestan Females 82.8 52.89 117.11 70.88 54.94 92.86 −0.14 −0.43 0.35
Kohgiluyeh and Boyer-Ahmad Females 21.36 7.56 29.49 20.67 11.29 27 −0.03 −0.35 0.6
Kurdistan Females 40.52 19.06 55.57 20.66 14.58 26.79 −0.49 −0.66 −0.11
Lorestan Females 84.45 18.4 119.71 41.3 13.97 54.39 −0.51 −0.67 −0.12
Markazi Females 14.98 10.03 26.49 7.33 5.06 19.61 −0.51 −0.69 −0.19
Mazandaran Females 36.63 24.02 54.47 27.81 20.4 49.95 −0.24 −0.54 0.22
North Khorasan Females 20.54 11.74 27.95 14.32 10.5 18.6 −0.3 −0.54 0.17
Qazvin Females 14.41 10.14 19.95 8.94 6.4 14.01 −0.38 −0.61 0.1
Qom Females 4.41 2.54 12.46 3.61 2.12 14.4 −0.18 −0.54 0.32
Semnan Females 4.88 3.34 7.71 3.22 2.15 7.12 −0.34 −0.61 0.14
Sistan and Baluchistan Females 14.75 9.32 22.19 23.98 15.76 65.51 0.62 −0.09 2.53
South Khorasan Females 9.77 6.52 13.48 6.79 5.27 10.69 −0.3 −0.52 0.1
Tehran Females 50.82 29.55 122.56 59.5 38.87 155.34 0.17 −0.3 0.96
West Azarbayejan Females 100.16 49.29 139.81 72.76 50.63 92.15 −0.27 −0.51 0.12
Yazd Females 8.42 5.98 11.93 6.39 4.42 11.53 −0.24 −0.53 0.21
Zanjan Females 13.57 9.2 18.95 7.28 5.45 11.52 −0.46 −0.65 −0.05

Figure 7.

Figure 7.

Sex-specific suicide mortality in Iran, 1990–2021.

The age-specific burden of self-harm and suicide mortality in Iran

In 2021, the age-specific rate of suicide mortality (per 100,000) was highest among individuals aged 15–39 years, at 6.82 (95% UI: 6.09–7.48). For both males and females, the highest age-specific rate of suicide mortality (per 100,000) was among individuals aged 15–39 years, at 10.01 (95% UI: 8.57–11.06) and 3.51 (95% UI: 3–4.42), respectively (Table 7 and Figure 8).

Table 7.

Age-specific suicide mortality in Iran, stratified by provinces, 1990–2021

Age Sex Year
1990 2021 Percentage change 1990–2021
Value Lower Upper Value Lower Upper Value Lower Upper
Age− rate suicide mortality (per 100,000)
10−14 years Both sexes 1.99 1.26 2.36 0.66 0.55 0.81 −0.67 −0.74 −0.51
15−39 years Both sexes 10.46 8.07 11.6 6.82 6.09 7.48 −0.35 −0.43 −0.19
40−44 years Both sexes 7.15 5.87 8.26 4.83 4.3 5.55 −0.32 −0.42 −0.1
45−49 years Both sexes 6.76 5.65 7.76 4.34 3.85 5.09 −0.36 −0.45 −0.13
50−69 years Both sexes 6.26 5.39 6.99 3.77 3.37 4.38 −0.4 −0.46 −0.23
70+ years Both sexes 6.35 5.63 7.09 4.89 4.26 5.67 −0.23 −0.31 −0.06
10−14 years Males 2.02 1.26 2.49 0.71 0.55 0.94 −0.65 −0.74 −0.46
15−39 years Males 11.28 9.44 12.95 10.01 8.57 11.06 −0.11 −0.24 0.05
40−44 years Males 9.41 7.51 11.14 7.53 6.37 8.65 −0.2 −0.34 0.06
45−49 years Males 9.36 7.73 11.13 6.71 5.73 7.88 −0.28 −0.41 −0.01
50−69 years Males 8.87 7.57 10.02 5.96 5.08 6.83 −0.33 −0.42 −0.14
70+ years Males 8 7.02 9.41 6.81 5.8 7.86 −0.15 −0.27 0.04
10−14 years Females 1.96 1.03 2.44 0.61 0.47 0.76 −0.69 −0.76 −0.52
15−39 years Females 9.61 5.92 11.14 3.51 3 4.42 −0.63 −0.72 −0.44
40−44 years Females 4.84 3.55 5.61 2.05 1.71 2.79 −0.58 −0.66 −0.36
45−49 years Females 4.1 3.01 5.02 1.85 1.56 2.78 −0.55 −0.64 −0.33
50−69 years Females 3.25 2.49 3.77 1.58 1.34 2.29 −0.52 −0.6 −0.33
70+ years Females 4.64 3.61 5.63 3.03 2.54 4.09 −0.35 −0.45 −0.18
Age count estimates
10−14 years Both sexes 144.58 91.89 171.23 43.7 36.68 53.76 −0.7 −0.76 −0.55
15−39 years Both sexes 2,270.81 1,752.85 2,519.34 2,367.98 2,112.17 2,594.32 0.04 −0.09 0.3
40−44 years Both sexes 155.59 127.84 179.74 344.75 307.21 396.52 1.22 0.91 1.95
45−49 years Both sexes 110.47 92.3 126.89 240.73 213.44 282.62 1.18 0.87 1.95
50−69 years Both sexes 326.58 280.94 364.8 532.74 475.98 618.61 0.63 0.46 1.09
70+ years Both sexes 61.63 54.63 68.8 178.66 155.69 207.09 1.9 1.61 2.52
10−14 years Males 74.96 46.65 92.31 24.02 18.61 31.84 −0.68 −0.76 −0.51
15−39 years Males 1,236.31 1,034.02 1,418.77 1,770.98 1,515.40 1,957.52 0.43 0.23 0.69
40−44 years Males 103.3 82.38 122.23 272.67 230.54 313.32 1.64 1.17 2.49
45−49 years Males 77.35 63.89 92.02 190.51 162.71 223.73 1.46 1.04 2.41
50−69 years Males 247.96 211.57 280.17 421.68 359.75 483.25 0.7 0.46 1.18
70+ years Males 39.51 34.7 46.47 122.46 104.41 141.45 2.1 1.65 2.8
10−14 years Females 69.62 36.64 86.84 19.68 15.19 24.7 −0.72 −0.79 −0.57
15−39 years Females 1,034.50 636.5 1,199.05 596.99 509.97 751.49 −0.42 −0.55 −0.12
40−44 years Females 52.29 38.33 60.57 72.08 60.19 98.39 0.38 0.1 1.07
45−49 years Females 33.12 24.32 40.59 50.22 42.21 75.52 0.52 0.19 1.25
50−69 years Females 78.62 60.26 91.32 111.05 94.64 161.64 0.41 0.16 0.94
70+ years Females 22.11 17.21 26.82 56.2 47.14 75.68 1.54 1.13 2.19

Figure 8.

Figure 8.

Age-specific suicide mortality in Iran, 1990–2021.

The burden of self-harm and suicide mortality stratified by provinces

The subnational burden of self-harm and suicide mortality showed that the highest age-standardized prevalence rate per 100,000 in 2021 was in Ilam (188.33 [95% UI: 157.11–223.85]), and the lowest was in Zanjan (106.84 [95% UI: 90.84–127.12]) (Table 1 and Figure 9). The highest self-harm counts were in Tehran (18,736 [95% UI: 15,979–22,234]), and the lowest was in Semnan (945 [95% UI: 799–1,128]) (Table 1 and Figure 10). The highest age-standardized suicide mortality rate per 100,000 was in Ilam (10.68 [95% UI: 4.58–12.93]) (Table 4 and Figure 11).

Figure 9.

Figure 9.

Age-standardized rate (per 100,000) prevalence of self-harm in Iran, stratified by provinces, 1990–2021.

Figure 10.

Figure 10.

All-ages prevalence of self-harm in Iran, stratified by provinces, 1990–2021.

Figure 11.

Figure 11.

Age-standardized suicide mortality rate (per 100,000) in Iran, stratified by provinces, 1990–2021.

Discussion

The present study aimed to investigate the impact of self-harm and suicide mortality across different sex and age groups in Iran from 1990 to 2021. It additionally sought to estimate the prevalence, incidence, disability and mortality rates associated with these issues, analyzing data at both national and subnational levels to provide a comprehensive understanding of their distribution and burden.

Over the past three decades, the patterns of self-harm and suicide mortality in Iran have exhibited a decline. The prevalence, incidence, disability attributed to self-harm and the rates of suicide mortality have shown noticeable reductions compared to figures from 1990. The conclusion of the war against Iran in 1988 marked a period when mental health disorders escalated due to the psychological impacts of war-related trauma. Consequently, the rates of self-harm and suicide mortality in 1990 and the subsequent years were significantly elevated. However, over time, these rates transitioned into a sustained downward trajectory, with notable decreases in the prevalence, incidence and disability associated with self-harm, as well as reductions in suicide mortality observed up to 2021.

Studies indicate that war has a profound impact on the mental health of civilians, leading to a noticeable rise in both the prevalence and incidence of mental disorders (Baingana et al., 2005; Lopez-Ibor et al., 2005; Murthy and Lakshminarayana, 2006). A newly published meta-analysis reveals a notable rise in mental health disorders in regions affected by conflict and war (Lim et al., 2022). Wars often lead to a rise in displaced populations, heightened migration, economic instability and the spread of physical illnesses (Lim et al., 2022). Consequently, this can result in a rise in mental health disorders within the community, potentially contributing to an increase in instances of self-harm and suicide-related fatalities.

During the past decades, Iran has been affected by economic turmoil and international politics (Danaei et al., 2019). During the 1980s, Iran witnessed a marked increase in the birth rate, which led to a significant increase in population growth. This increase in population growth affected all aspects of the health system, so there was an increased demand for health infrastructure, health care and especially mental health-related care. Effective and efficient measures were taken to increase health and health care, but in the area related to mental health care, access to mental health services did not grow proportionately (Farzadfar et al., 2022), and this was one of the factors that affected mental health.

On the other hand, the stigma related to mental disorders is one of the issues that, although it has faded over time, still stands; therefore, the request related to mental health care has always been associated with stigma (Corrigan et al., 2014; Taghva et al., 2017). Another effective factor in mental disorders in Iran during recent decades is due to economic turmoil. Studies have shown that mental disorders, as well as self-harm and suicide, increase as a result of economic crises (World Health Organization. Regional Office for E., 2011; Marazziti et al., 2021). Some of the consequences of economic crises, such as unemployment, poverty and hyperinflation, are associated with an increase in mental disorders, such as depression and suicide, and disrupt mental health (Paul and Moser, 2009; Amiri, 2022a, 2022b).

A part of the economic turmoil was caused by the sanctions imposed on Iran (Kokabisaghi, 2018), which intensified especially since 2011. Although the factors affecting mental disorders, especially self-harm and suicide mortality, have had cumulative effects in Iran over the past decades, unlike depression and anxiety disorders, the prevalence of self-harm and suicide mortality is low in Iran, and it is lower than the world average and many countries with a high level of social welfare. The dominant religion of Iran is Islam, and in most Islamic countries, the suicide rate is low (Pritchard and Amanullah, 2007; Mirhashemi et al., 2016; Lew et al., 2022). This could be partly due to the role of religion. In Islam, suicide and self-harm are prohibited, and the sanctity of life is emphasized in Islamic teachings (Ali, 2000). Of course, less access to mental health care alone is not the determining factor. There is a stigma attached to suicide for families (Wyllie et al., 2025). The stigma associated with suicide is important in Iranian culture (Masoomi et al., 2022). One significant obstacle to accessing mental health services in Iran is the social stigma surrounding psychiatric issues (Taghva et al., 2017). Research conducted in Iran indicates that women often attempt to conceal their suicide attempts due to the fear of being stigmatized by society (Azizpour et al., 2018), and the burden of this stigma causes many individuals and families to not seek mental health care despite the need for mental health professionals and mental health care providers (Tay et al., 2018). Suicide is heavily stigmatized in predominantly Muslim countries like Iran, and similarly, other religious communities are not immune to this issue (Shoib et al., 2022).

The differences between sexes in self-harm and suicide mortality have consistently drawn the attention of researchers. Studies exploring these disparities reveal that suicide mortality is notably higher among males compared to females. Data published by the WHO further supports this finding, highlighting that, on a global scale, the age-standardized suicide rate for males is 2.3 times greater than that of females (World Health Organization, 2021). Major depressive disorder constitutes a significant factor contributing to suicide. While its prevalence is observed to be twice as high among females compared to males, the likelihood of suicide among females is markedly lower, amounting to only one-fourth of the probability seen in males (Murphy, 1998). The higher incidence of suicide among men has been attributed to several factors, including contrasting behavioral and emotional tendencies between genders. Men often prioritize independence and decisiveness, perceiving the admission of needing assistance as a sign of weakness, which leads them to avoid seeking help. On the other hand, women tend to emphasize interdependence and are more likely to seek support from friends and accept help readily, contributing to differing responses in times of emotional distress (Murphy, 1998).

This study represents the first comprehensive endeavor to analyze trends in self-harm and suicide in Iran since 1990, offering gender- and age-specific estimates. However, it is subject to certain limitations extensively discussed within the framework of the GBD study. Challenges include issues related to the quality and collection of primary data, as well as inconsistencies in data availability. Moreover, the risk factors assessed are constrained by the exclusion of various potentially significant risk factors and covariates. The impact of the coronavirus disease 2019 pandemic further complicates the interpretation and analysis of findings. A more detailed discussion on the methodological constraints inherent in the GBD study is available in other sources (Brauer et al., 2024; Ferrari et al., 2024). GBD estimates are often derived from modeled data, particularly in contexts where the quality of surveillance systems is insufficient or constrained

Health implications

The gathered data highlights the trends of suicide and suicide mortality in Iran over the past three decades. Despite notable advancements in physical health within the Iranian healthcare system during this period, demographic shifts and lifestyle changes have exposed gaps in addressing mental health. Efforts to improve awareness around mental health and ensure broader access to mental health services have remained insufficient. To address these challenges effectively, it is crucial to prioritize mental health literacy and expand access to care as key components in health policy development.

Conclusion

The present study showed that despite demographic changes, economic turmoil and war, self-harm and suicide mortality have had a decreasing trend in Iran. Suicide mortality in males was higher than in females, self-harm was more common in females in past decades, and in recent years, it was lower than in males.

Acknowledgments

The data sources of this study were taken from GBD 2021, which is publicly available.

Abbreviations

DALYs

disability-adjusted life years

GBD

global burden of disease, encapsulating a comprehensive assessment of health losses due to diseases and injuries worldwide

ICD

International Classification of Diseases, providing a standard framework for categorizing health conditions

SEV

summary exposure value, quantifying exposure levels to specific risk factors

YLDs

years lived with a disability

YLLs

years of life lost due to premature death

Open peer review

To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2025.10104.

Data availability statement

The data sources of this study were taken from GBD 2021, which is publicly available. In this way, the data can be accessed through the links below: https://vizhub.healthdata.org/gbd-results/ https://vizhub.healthdata.org/gbd-compare/

Author contribution

Sohrab Amiri: Conceptualization, extraction, analysis, writing and revision.

Jannat Mashayekhi: Writing and revision.

Financial support

The study received no financial support.

Competing interests

The authors declare none.

Ethics statement

The data sources of this study were taken from GBD 2021, which is publicly available.

References

  1. Ali AY (2000) The holy Qur’an: Wordsworth Editions.
  2. Amiri S (2022a) Unemployment and suicide mortality, suicide attempts, and suicide ideation: A meta-analysis. International Journal of Mental Health 51(4), 294–318. [Google Scholar]
  3. Amiri S (2022b) Unemployment associated with major depression disorder and depressive symptoms: A systematic review and meta-analysis. International Journal of Occupational Safety and Ergonomics 28(4), 2080–2092. [DOI] [PubMed] [Google Scholar]
  4. Azizpour M, Taghizadeh Z, Mohammadi N and Vedadhir A (2018) Fear of stigma: The lived experiences of Iranian women after suicide attempt. Perspectives in Psychiatric Care 54(2), 293–299. 10.1111/ppc.12237. [DOI] [PubMed] [Google Scholar]
  5. Baingana FK, Bannon I and Thomas R (2005) Mental Health and Conflicts: Conceptual Framework and Approaches. World Bank Washington. https://documents.worldbank.org/en/publication/documents-reports/documentdetail/829381468320662693 [Google Scholar]
  6. Bank TW (2021) GDP per capita (current US$)—Iran, Islamic Rep. Available at https://data.worldbank.org/indicator/NY.GDP.PCAP.CD?end=2018&locations=IR&start=1960
  7. Bhattacharjee NV, Schumacher AE, Aali A, Abate YH, Abbasgholizadeh R, Abbasian M, … Vollset SE (2024) Global fertility in 204 countries and territories, 1950–2021, with forecasts to 2100: A comprehensive demographic analysis for the Global Burden of Disease Study 2021. The Lancet 403(10440), 2057–2099. 10.1016/S0140-6736(24)00550-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brauer M, Roth GA, Aravkin AY, Zheng P, Abate KH, Abate YH, … Gakidou E (2024) Global burden and strength of evidence for 88 risk factors in 204 countries and 811 subnational locations, 1990–2021: A systematic analysis for the global burden of disease study 2021. The Lancet 403(10440), 2162–2203. 10.1016/S0140-6736(24)00933-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Corrigan PW, Druss BG and Perlick DA (2014) The impact of mental illness stigma on seeking and participating in mental health care. Psychological Science in the Public Interest 15(2), 37–70. 10.1177/1529100614531398. [DOI] [PubMed] [Google Scholar]
  10. Danaei G, Farzadfar F, Kelishadi R, Rashidian A, Rouhani OM, Ahmadnia S, … Arhami M (2019) Iran in transition. The Lancet 393(10184), 1984–2005. [DOI] [PubMed] [Google Scholar]
  11. Farzadfar F, Naghavi M, Sepanlou SG, Saeedi Moghaddam S, Dangel WJ, Davis Weaver N, … Larijani B (2022) Health system performance in Iran: A systematic analysis for the global burden of disease study 2019. The Lancet 399(10335), 1625–1645. 10.1016/S0140-6736(24)00757-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Ferrari AJ, Santomauro DF, Aali A, Abate YH, Abbafati C, Abbastabar H, … Murray CJL (2024) Global incidence, prevalence, years lived with disability (YLDs), disability-adjusted life-years (DALYs), and healthy life expectancy (HALE) for 371 diseases and injuries in 204 countries and territories and 811 subnational locations, 1990–2021: A systematic analysis for the global burden of disease study 2021. The Lancet 403(10440), 2133–2161. 10.1016/S0140-6736(24)00757-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. GBD 2021 Causes of Death Collaborators (2024) Global burden of 288 causes of death and life expectancy decomposition in 204 countries and territories and 811 subnational locations, 1990-2021: A systematic analysis for the global burden of disease study 2021. Lancet 403(10440), 2100–2132. 10.1016/s0140-6736(24)00367-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Global Health Metrics G. (2024) Self-harm—Level 3 cause. Available at https://www.healthdata.org/research-analysis/diseases-injuries-risks/factsheets/2021-self-harm-level-3-disease
  15. Knipe D, Padmanathan P, Newton-Howes G, Chan LF and Kapur N (2022) Suicide and self-harm. The Lancet 399(10338), 1903–1916. 10.1016/S0140-6736(22)00173-8. [DOI] [PubMed] [Google Scholar]
  16. Kokabisaghi F (2018) Assessment of the effects of economic sanctions on Iranians’ right to health by using human rights impact assessment tool: A systematic review. International Journal of Health Policy and Management 7(5), 374–393. 10.15171/ijhpm.2017.147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lew B, Lester D, Kõlves K, Yip PSF, Chen YY, Chen WS, … Ibrahim N (2022) An analysis of age-standardized suicide rates in Muslim-majority countries in 2000-2019. BMC Public Health 22(1), 882. 10.1186/s12889-022-13101-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lim ICZY, Tam WWS, Chudzicka-Czupała A, McIntyre RS, Teopiz KM, Ho RC and Ho CSH (2022) Prevalence of depression, anxiety and post-traumatic stress in war- and conflict-afflicted areas: A meta-analysis. Frontiers in Psychiatry 13, 978703. 10.3389/fpsyt.2022.978703. PMID: 36186881; PMCID: PMC9524230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lopez-Ibor JJ, Christodoulou G, Maj M, Sartorius N and Okasha A (2005) Disasters and Mental Health. Wiley Online Library. [Google Scholar]
  20. Marazziti D, Avella MT, Mucci N, Della Vecchia A, Ivaldi T, Palermo S and Mucci F (2021) Impact of economic crisis on mental health: A 10-year challenge. CNS Spectrums 26(1), 7–13. 10.1017/S1092852920000140. [DOI] [PubMed] [Google Scholar]
  21. Masoomi M, Hosseinikolbadi S, Saeed F, Sharifi V, Jalali Nadoushan AH and Shoib S (2022) Stigma as a barrier to suicide prevention efforts in Iran. Frontiers in Public Health 10, 1026451. 10.3389/fpubh.2022.1026451. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Mirhashemi S, Motamedi MHK, Mirhashemi AH, Taghipour H and Danial Z (2016) Suicide in Iran. The Lancet 387(10013), 29. 10.1016/S0140-6736(15)01296-9. [DOI] [PubMed] [Google Scholar]
  23. Moradinazar M, Mirzaei P, Moradivafa S, Saeedi M, Basiri M and Shakiba M (2022) Epidemiological status of depressive disorders in the Middle East and North Africa from 1990 to 2019. Health Promotion Perspective 12(3), 301–309. 10.34172/hpp.2022.39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Murphy GE (1998) Why women are less likely than men to commit suicide. Comprehensive Psychiatry 39(4), 165–175. 10.1016/s0010-440x(98)90057-8. [DOI] [PubMed] [Google Scholar]
  25. Murthy RS and Lakshminarayana R (2006) Mental health consequences of war: A brief review of research findings. World Psychiatry 5(1), 25–30. [PMC free article] [PubMed] [Google Scholar]
  26. Nazarzadeh M, Bidel Z, Ayubi E, Asadollahi K, Carson KV and Sayehmiri K (2013) Determination of the social related factors of suicide in Iran: A systematic review and meta-analysis. BMC Public Health 13, 4. 10.1186/1471-2458-13-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Paul KI and Moser K (2009) Unemployment impairs mental health: Meta-analyses. Journal of Vocational Behavior 74(3), 264–282. [Google Scholar]
  28. Pritchard C and Amanullah S (2007) An analysis of suicide and undetermined deaths in 17 predominantly Islamic countries contrasted with the UK. Psychological Medicine 37(3), 421–430. 10.1017/s0033291706009159. [DOI] [PubMed] [Google Scholar]
  29. Sharifi V, Amin-Esmaeili M, Hajebi A, Motevalian A, Radgoodarzi R, Hefazi M and Rahimi-Movaghar A (2015) Twelve-month prevalence and correlates of psychiatric disorders in Iran: The Iranian mental health survey, 2011. Archives of Iranian Medicine 18(2), 76–84. [PubMed] [Google Scholar]
  30. Shoib S, Armiya’u AY u, Nahidi M, Arif N and Saeed F (2022) Suicide in Muslim world and way forward. Health science reports 5(4), e665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Stevens GA, Alkema L, Black RE, Boerma JT, Collins GS, Ezzati M, … Horton R (2016) Guidelines for accurate and transparent health estimates reporting: The GATHER statement. The Lancet 388(10062), e19–e23. [DOI] [PubMed] [Google Scholar]
  32. Taghva A, Farsi Z, Javanmard Y, Atashi A, Hajebi A and Khademi M (2017) Stigma barriers of mental health in Iran: A qualitative study by stakeholders of mental health. Iranian Journal of Psychiatry 12(3), 163–171. [PMC free article] [PubMed] [Google Scholar]
  33. Tay S, Alcock K and Scior K (2018) Mental health problems among clinical psychologists: Stigma and its impact on disclosure and help-seeking. Journal of Clinical Psychology 74(9), 1545–1555. 10.1002/jclp.22614. [DOI] [PubMed] [Google Scholar]
  34. World Health Organization (2019) Suicide and Self-Harm. Cairo: Regional Office for the Eastern, M [Google Scholar]
  35. World Health Organization. (2021). Suicide worldwide in 2019: global health estimates.
  36. World Health Organization (2023) Suicide. Regional Office for the Eastern, M [Google Scholar]
  37. World Health Organization (2014). Preventing suicide: A global imperative: World Health Organization.
  38. World Health Organization. Regional Office for, E. (2011). Impact of economic crises on mental health. Available at Copenhagen: https://iris.who.int/handle/10665/370872
  39. Wyllie JM, Robb KA, Sandford D, Etherson ME, Belkadi N and O’Connor RC (2025) Suicide-related stigma and its relationship with help-seeking, mental health, suicidality and grief: Scoping review. BJPsych Open 11(2), e60. 10.1192/bjo.2024.857. [DOI] [PMC free article] [PubMed] [Google Scholar]
Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr2

Review: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R0/PR2

EDUARD KYLE GOCOTANO 1

Percentage change was use for establishing trend. While percentage change is great for doing the trend in the data. I encourage adding a moving average as the data seem to be noisy. There were instance that the percentage change was negative and some positive. The trends in the data seems to have a lot of noise even though it is from the Global survey.

If this comment is not deemed relevant, will accept the paper as it is still well discussed.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr3

Review: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R0/PR3

Daman Bogati 1

The manuscript is very important and timely, as there is more attention to the issue of mental health and prevention of suicide, globally and also in the area. The GBD 2021 data were employed to analyze long-term trends and estimated national and subnational burden of self-harm and suicide in Iran, it is a solid foundation for future studies.

The authors succinctly point to important trends, such as an overall reduction in the prevalence of self-harm across each time period (31 years) and gender inequalities in the rates of both self-harm and suicide This manuscript does a good job of identifying some of the clear findings such as the decline in the prevalence of self-harm across the 31-y time frame and enduring disparities in both the rates of self-harm and suicide mortality by gender. Sex, age and.isSuccessful.Patient sex, age and

location adds significant depth to the analysis and allows for more targeted interpretation. However, I believe the paper could be further strengthened by elaborating on some contextual factors that may have influenced these trends, such as changes in health service availability, sociopolitical developments, stigma, or policy shifts in Iran.

The observed increase in absolute suicide deaths, despite a reduction in age-standardized rates, is an important point that reflects population growth and demographic transitions. Including more interpretation of this dynamic would add value to the discussion.

Overall, the manuscript makes a valuable contribution to the literature on mental health and mortality in the Eastern Mediterranean region. With a few additions related to contextual interpretation and policy implications, the paper would be even more impactful for public health professionals and decision-makers.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr4

Recommendation: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R0/PR4

Editor: Gonzalo Martínez-Alés Garcia1

No accompanying comment.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr5

Decision: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R0/PR5

Editor: Jerome Galea1

No accompanying comment.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr7

Review: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R1/PR7

EDUARD KYLE GOCOTANO 1

Line 34, 35: As the population increases, the need for mental health care will also increase.

Who said this, what made you say this, while suicide is mental health related indicator of a country, the quick assumption that increasing population has an increase ot the needs of mental health care is too quick to jump without reference to the data. Also the data is nto even talking about mental health care needs. The article premise may be coming from suicide population but it should not be said in an academic article. This may be said in an opinion piece, but in an academic article, it is best to cite or try to make an appropriate and coherent discussion as to why you say this. Because after this line you proceed to say that the population needs will also increase, indicating education, employment, welfare service etc. While there are papers highlighting its relationship, your paper just bluntly says it. It would be best to cite at this point. Also, that’s not your topic.

I’m not sold out on the highlighted part. it feels an opinion piece rather than an academic discussion on the implications of the data. It gives an idea what could be its effect and cause but it feels lacking, maybe an article specifying this things you’ve said.

In addition, the highlighted part again, "Of course, less access to mental

health care alone is not the determining factor, the stigma associated with mental

disorders is very important in Iranian culture, and the burden of this stigma causes

many individuals and families to not seek mental health care despite the need for

mental health professionals and mental health care providers. Especially in cases of suicide, many families try to hide the cause of death to avoid the stigma it

causes"

what made you say this, was there a qualitative study that explored those cases of suicide where the family had stigma about it and failed to seek help or assistance with regards to it.

I agree that this things are true but the thing is, it feels more of an opinion rather than an academic discussion. To improve it, I suggest find papers or articles that explore these things you have said and this would allow you explain it properly and have you backed up.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr8

Recommendation: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R1/PR8

Editor: Gonzalo Martínez-Alés Garcia1

The interpretation of demographic effects is important, especially the observation that absolute suicide deaths increased despite declining age-standardized rates, due to population growth and demographic transition. This point deserves deeper quantitative illustration, for example by presenting population growth alongside mortality trends in a supplementary figure. The discussion mentions war, economic turmoil, sanctions, and stigma, but these factors could be integrated more systematically, perhaps in a subsection linking major socio-political events to observed changes in self-harm and suicide. Although the introduction references global suicide rates, the discussion could benefit from direct comparison with other Middle Eastern countries to clarify whether Iran’s trends are unique or consistent with regional patterns. While the manuscript follows the GBD protocol, readers unfamiliar with the framework may struggle with terminology such as YLDs, YLLs, and DALYs. A brief paragraph simplifying these terms would improve accessibility.

The results are thorough but at times repetitive. Multiple sections restate that age-standardized rates decreased while absolute counts increased. This point could be consolidated to avoid redundancy. Figures are informative but could be enhanced with clearer labeling of years, uncertainty intervals, and inflection points such as the 2005 peak in incidence. Tables 5–7 on sex- and age-specific differences could be more impactful if the main message was summarized graphically, for example with a ratio of male-to-female suicide mortality over time.

The discussion section situates findings in historical and cultural context but could be more structured. It would help to first summarize key findings, then provide interpretation in relation to demographics, conflict, and economic pressures, and finally highlight policy implications. The role of religion is mentioned briefly but could be supported with empirical studies. The limitations section should more explicitly acknowledge that GBD estimates rely on modeled data in contexts where surveillance quality is limited.

The abstract could state more clearly the public health significance of findings. Currently it emphasizes descriptive trends but does not fully connect to policy relevance. The impact statement repeats several elements of the abstract and should instead focus on how the findings could inform interventions or health system planning. References need careful correction as some are inconsistently formatted, for example “Bank, T. W. (2 .(021GDP per capita …”. Multiple WHO references list “Regional Office for the Eastern” rather than “World Health Organization. Regional Office for the Eastern Mediterranean.” Figure captions should be expanded to include sample size, data source, and a one-sentence interpretation.

The manuscript is understandable but requires substantial language polishing for clarity and professionalism. Grammar and syntax should be corrected, for example “The number of suicide mortality in 2021 was 3,708 individuals” should be “The number of suicide deaths in 2021 was 3,708,” and “percentage change from 1990-2021 was -0.25” should be “The percentage change from 1990 to 2021 was -25%.” Repetition should be reduced, for example the phrase “prevalence, incidence, disability, and mortality rates” is overused. Typos include “compeletly teransperant” instead of “completely transparent” and “Demographic translation” instead of “demographic transition.” Some references are truncated or inconsistently capitalized. Formatting errors such as missing spaces before parentheses (for example “world(Knipe…”) should be fixed. Many sentences are too long and should be split for clarity, for example the sentence on population increase and suicide deaths in the discussion could be shortened into two simpler sentences.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr9

Decision: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R1/PR9

Editor: Jerome Galea1

No accompanying comment.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr11

Review: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R2/PR11

EDUARD KYLE GOCOTANO 1

line 127 and 128 “This manuscript was produced as part of the GBD Collaborator Network and following the GBD Protocol.”

It would make the paper better if this summarized rather than referred.

Case definitions (line 142 to 152)

In the title, it states self-harm and suicide, however, in the case definition, it lacks definition on what suicide is and how suicide and self-harm differentiates from each other. To improve, I suggest adding discussion on its difference in the case definition.

This is great work; I look forward to having it done published. Iran’s mental health (suicide and self-harm statistics) trend has to be in the journals. Goodluck and keep it up! You’re doing a great work!

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr12

Recommendation: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R2/PR12

Editor: Gonzalo Martínez-Alés Garcia1

Thanks for addressing all concerns, the manuscript is of great value.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr13

Decision: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R2/PR13

Editor: Jerome Galea1

No accompanying comment.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr15

Recommendation: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R3/PR15

Editor: Gonzalo Martínez-Alés Garcia1

No accompanying comment.

Glob Ment Health (Camb). doi: 10.1017/gmh.2025.10104.pr16

Decision: Sex- and age-specific burden of self-harm and suicide mortality: A national and subnational study in Iran — R3/PR16

Editor: Jerome Galea1

No accompanying comment.


Articles from Cambridge Prisms: Global Mental Health are provided here courtesy of Cambridge University Press

RESOURCES