Burden of Skin Diseases in Thailand: Analysis of Prevalence and DALYs Using Real-world Healthcare Service Data

Background There is little evidence to describe the burden of skin diseases in developing countries and its accuracy remained questionable. The objective of this study was to examine prevalence and disability adjusted life years (DALYs) of skin diseases in a Thai general population in Ubonratchathani. Methods Based on real-world healthcare service data (diagnoses, prevalence, and cause-specic mortality) retrieved from the National Health Security Oce reimbursement database, we used a simplied prevalence-based approach adopted in the Global Burden of Diseases to compute disease burden, measured as DALYs, of skin diseases. DALYs was calculated as the sum of years lost due to disability and years of life lost due to skin diseases, with adoption of previously published averaged disability weights and a 95% uncertainty interval (UI) estimated using a Bayesian bootstrap technique (normal distribution assumed).

diseases was not improved or even increased, with varying burdens across types of skin diseases and sexes. This was similar for both developed and developing countries. (4) However, methods and parameters used to inform computations of DALYs in the previous studies, although being consecutively updated and improved, were largely based on expert panels, epidemiological studies, speci c analytical methods and disability weighting techniques, which may be somewhat problematic. (5,7) This would particularly have considerable impact on DALYs estimates in developing countries, where the quality and completeness of data sources (prevalence, incidence, disability and cause-speci c mortality) may be unsatisfactorily low. This merits further efforts to improve these country-speci c data to obtain more accurate and reliable estimates of disease burden. This study aimed to examine the prevalence and DALYs of skin diseases in a Thai general population using real-world healthcare service data.
This would help inform policy decision and planning to address skin and subcutaneous diseases in tropical and developing countries.

Methods
This study examined the burden of skin diseases in a Thai general population in Ubonratchathani using data from health services and reimbursement database. Ubonratchathani locates in the Northeastern region of Thailand and is the country's fourth largest province with a total population of 1,503,945 in 2018. Health service data have been standardized and stored in electronic health records in all levels of healthcare services, primary to tertiary care, for more than 15 years. Service data include personal and medical history, physical examination, investigations, diagnoses, medical treatments and procedures, follow-up, health promotion and disease prevention activities and vital status. All the service data are entered and stored in 43 separated folders, each of which is linked with unique national identi cation numbers. The data are then retrieved through a centralized database system called 'Health Data Center' and assembled and audited in a database of the Sample size determination was based on research questions to examine the prevalence of different skin diseases according to the GBD 2017. Based on prevalence data reported in a study by Fasih S,et al.(9) with a 95% con dence and 10% margin of error allowed, a minimum of 63,641 participants would be needed for this study.  Table 1). These included dermatitis (atopic dermatitis, seborrheic dermatitis, contact dermatitis), viral skin diseases (wart, molluscum contagiosum), fungal skin diseases (tinea capitis, other fungal skin diseases), bacterial skin diseases (impetigo, pyoderma, cellulitis, abscess and other bacterial skin diseases), acne vulgaris, psoriasis, alopecia areata, scabies, melanoma, squamous cell carcinoma and basal cell carcinoma, carcinoma in situ, pruritus, urticaria, decubitus ulcer and other skin and subcutaneous diseases. According to current ICD coding systems, squamous cell carcinoma and basal cell carcinoma are not coded as separate diagnoses as the two malignancies are captured under a single diagnostic code of non-melanoma skin cancers.

Estimation of prevalence of skin diseases and disease-speci c mortality
Based on the NHSO database, the overall and cause-speci c prevalence of skin diseases was calculated as the number of individuals diagnosed with skin diseases divided by a total population of Ubonratchathani, with 95% uncertainty interval (UI) estimated using the Bayesian bootstrap technique (normal distribution assumed). The prevalence estimates were computed for the overall population and by sexes and districts. According to Karimkhani C. et at, (11) age-and sex-speci c mortality was obtained for six skin and subcutaneous diseases: cellulitis, decubitus ulcer, abscess and other bacterial skin diseases, melanoma, squamous cell carcinoma, and other skin and subcutaneous diseases. This was based on mortality observed in individuals diagnosed with the above skin diseases in 2018.

Estimation of DALYs attributable to skin diseases
Using a simpli ed approach to calculation of DALYs (prevalence-based) adopted in the GBD 2010, (12) we computed DALYs as the sum of years lost due to disability (YLDs) and years of life lost (YLLs) due to skin diseases for each sex, age group, and type of skin diseases. The prevalence estimates were multiplied by averaged disability weights to calculate YLDs resulting from each type of skin disease. The averaged disability weights were derived from the GBD 2017.(13) Due to disease severity not being captured in electronic health records for most skin diseases, a single constant value of averaged disability weight was applied for each individual skin disease. YLLs were calculated as the sum of each death multiplied by the standard life expectancy at each age. As described earlier, deaths due to the selected skin diseases were identi ed through the NHSO database. The standard life expectancy for each sex and 5-year age group was taken from the Public Health Statistic A.D. 2018 Report convened by Thailand's Ministry of Public Health (Strategy and Planning Division). (14) Corresponding 95% UI for these measures of disease burden were calculated using a parametric bootstrap technique. (15) All analyses were performed using STATA 14.2 statistical package.

Prevalence of skin diseases
Out of 1,503,945 people residing in Ubonratchathani, 110,205 were diagnosed with at least one of skin diseases, a prevalence of 7.33% (95%UI 7.29 -7.37). Table 1 shows the number and prevalence of skin diseases diagnosed and treated in Ubonratchathani's healthcare services in 2018. The prevalence of skin diseases varied substantially from lower than 0.01% to 2.35%. The highest prevalence was observed for dermatitis, bacterial skin infection and urticaria (2.35%, 2.21% and 0.89% respectively). Contact dermatitis constituted most cases of all dermatitis, with the prevalence of 0.98%, Considering skin malignancies, 435 cases of malignant melanoma were treated in Ubonratchathani's healthcare services including a special cancer hospital, accounting for a prevalence of 0.03% or 30 per 100,000 population. There were 222 cases of non-melanoma skin cancers (squamous and basal cell carcinoma combined), a prevalence of 10 per 100,000 population. Figure 1 shows the prevalence of skin diseases by district. There was great discrepancy in the prevalence of skin diseases across 25 districts of Ubonratchathani, ranging from 5.35% to 15.85%. The districts close to the border between Thailand and Lao PDR appeared to have lower skin disease prevalence than those located further from the border. The highest prevalence were observed in Muang Sam Sip, Nam Khun and Thung Si Udom Districts (15.85%, 10.53%, 9.30% respectively). The districts with the lowest skin disease prevalence were Khong Chiam, Khemmarat and Warin Chamrap Districts (5.35%, 5.51%, 5.57% respectively).

YLLs, YLD and DALYs attributable to skin diseases
Numbers of deaths and YLLs for six skin diseases overall and by sex are shown in Supplementary Table 2 Table 3). The top 3 diseases that contributed most YLDs were contact dermatitis, urticaria and cellulitis (YLDs 1,598, 1,445 and 985 years). These together contributed to 65% of all YLDs.
Melanoma resulted in a total YLD of 144 years. When using a disability weight of squamous cell carcinoma, nonmelanoma skin cancers contributed to 48 YLDs. In a sensitivity analysis using a disability weight of basal cell carcinoma, the diseases caused only 2.4 YLDs, Of note, females had higher YLDs than males for all skin diseases, except for cellulitis, impetigo, psoriasis and decubitus ulcer.  (11,416.7 -13,399.1) respectively) The largest contribution of s single cause to DALYs in Ubonratchathani province was from cellulitis, which accounted for up to 45% of total DALYs. This was followed by decubitus ulcer and contact dermatitis, which together contributed to another 25% of the total DALYs. For certain skin diseases including cellulitis, decubitus ulcer and skin cancers, they contributed to more DALYs in males than females, while other diseases caused more DALYs in females than males. Figures 2 and 3 show

Discussion
In the present study, we described the burden of skin and subcutaneous diseases in a Thai population in Ubonratchathani, using real-world healthcare service data. The overall prevalence of skin diseases was as high as 7% and varied across sex, age group and geographic areas. Dermatitis, bacterial skin infection and urticaria were the most common skin diseases.
DALYs attributable to skin diseases was 26,125, and generally higher in males than females. Cellulitis, decubitus ulcer and contact dermatitis were the main causes of DALYs in this population.
Skin diseases were common worldwide, although the overall prevalence varied greatly across different countries. The overall prevalence of skin diseases in the present study was approximately 7%. This was considerably lower than the prevalence reported in other population-based survey studies, the overall prevalence of 15% and 27% reported in Bangladesh and European countries respectively. (16,17) The discrepancy may be explained by that the Bangladesh study was conducted in one rural community without systemic sampling of participants and therefore prone to biased selectionthose with symptoms and signs of skin diseases were likely participating, hence high prevalence. Additionally, a populationbased survey in European countries described a lifetime prevalence, whereas our study reported the prevalence of any skin diseases diagnosed and recorded in electronic health records in 2018. Differences in study settings, computation methods and de nitions may make it di cult to compare skin disease prevalence across studies.
Comparison in skin disease prevalence is even more challenging when considering pattern or types of skin diseases. Based on the GBD of skin diseases 2010, fungal skin diseases, acne vulgaris and other skin and subcutaneous diseases were among the top 10 most prevalent diseases globally,(2) Dermatitis was one of the most prevalent skin diseases worldwide and the prevalence of dermatitis and its subtypes varied across populations. Dermatitis constituted 21-32% of all skin diseases. (9,16,18,19) While atopic dermatitis represented most cases of dermatitis in developed countries, (6,17,(20)(21)(22) contact dermatitis constituted most cases of dermatitis in our study and some European countries.(17) Skin infestation from scabies was prevalent in developing countries, (9,16,18) but rarely occurred in developed countries. For example, the prevalence of scabies was as high as 15% in Pakistan population,(9) while the gure was lower than 1% in this Thai population and most developed countries.(2) Interestingly, warts and acne topped the ranking of the most prevalent skin diseases in European countries,(17) while they were less prevalent in developing countries.(2) These differences may suggest true discrepancies in the pattern of skin diseases between countries and geographic regions or simply re ect different age groups of study participants and methods to obtain the prevalence estimates. with disproportionately high use of agricultural chemicals. This may also suggest possible differences in healthcare service systems and quality of healthcare between countries. Besides, this may be explained by differences between these studies in the quality and completeness of data on diagnoses, related consequences and cause-speci c mortality used to calculate DALYs.
Sex differences in DALYs due to skin diseases exist in many populations. This may be attributable to differences between sexes in the pattern of skin disease burden. Consistent to previous studies,(2, 6) our study found that YLLs constituted higher relative contribution to DALYs in men than women. Men may be more likely to be affected by fatal skin diseases such as skin cancers, cellulitis, and other serious skin infections, while women may be more likely to suffer from non-fatal disabling skin conditions such as dermatitis, pruritus and urticaria. Interestingly, while the GBD 2017 and GBD of Skin Diseases 2010 Reports suggested that woman had higher overall burden from skin and subcutaneous diseases than men, (2) our study showed opposite results. This might re ect true differences in the patterns of disease burden caused by skin disorders in the Thai population and other populations, or might be explained by a difference in data sources and methods to compute these indices of disease burden.
Skin cancers contribute to signi cant burden of skin diseases, albeit with disproportionate contributions of different skin cancers in different countries. While non-melanoma skin cancers (both basal and squamous cell carcinoma) were reportedly the most common skin malignancies in most Caucasian and Asian populations, (24)(25)(26) our study showed different results with malignant melanoma predominated in this Thai population. This could be due to differences between countries in climates and sun exposure and protection behaviors or that the Thais may have skin type that protects them from Ultraviolet B radiation than Caucasians and other Asians. (25) When considering DALYs, our study is consistent with many previous studies suggesting that malignant melanoma caused the higher disease burden than non-melanoma skin cancers, compared to that of melanoma, and disproportionate rises in skin cancers incidence in light-skinned populations. (23,27) This study was among the rst to examine prevalence and burden of skin diseases measured as DALYs in people of all ages in developing countries, using the real-world healthcare service and mortality data and standard computation techniques. However, our study had some limitations. First, diagnoses of skin and subcutaneous diseases based on respective ICD-10 codes from electronic health records did not contain data on disease severity. This could have altered the estimation of YLDs and hence DALYs in our study. However, this may at least represent the burden in real-world settings. Besides, the current ICD-10 coding system did not allow separate ICD codes for squamous and basal cell carcinomas.
Applying the averaged disability weight of squamous cell carcinoma to the computation of YLDs and DALYs for patients with either squamous or basal cell carcinomas would lead to underestimation of these measures of disease burden.
However, a sensitivity analysis using the averaged disability weight of basal cell carcinoma would re ect the possible upper-most estimates of the disease burden of both diseases. In addition, diagnoses of skin diseases were made by various health professions from primary care nurse and medical practitioners to tertiary care medical specialists and dermatologists. Therefore, there may be uncertainty about the accuracy and completeness of diagnostic information.
However, all hospital diagnoses were routinely checked and coded by trained coder and then validated by the NHSO. Therefore, this method of cases ascertainment was likely to accurately capture diagnoses and burden of skin diseases in real-world clinical practices. A veri cation study to examine the accuracy of skin disease diagnoses may also be needed. Lastly, as our study was based on data from mixed urban-rural communities in Thailand, the generalizability of our ndings to other populations and countries may be limited.

Conclusions
Skin disease caused considerable disease burden, measured as both prevalence and DALYs, in this Thai population and its contribution to the disease burden varied across sexes, age group and types of skin diseases. These estimates of disease burden will be useful in policy development and decision, health system planning and resource allocation with regards to skin diseases in Thailand and other developing countries.

Declarations
Ethics approval and consent to participate: The present study was approved by the Ethics Committee of Sanpasitthiprasong Hospital (ref no. 043/2557). Need for participant's informed consent was waived off by the Ethics Committee of Sanpasitthiprasong Hospital as this study retrospectively used anonymous data from electronic health records and reimbursement data. All methods in this study were performed in compliance with the International Conference on Harmonization (ICH) and the principles of the Declaration of Helsinki.

Consent for publication:
Not applicable Availability of data and materials: The data that support the ndings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Competing interests:
The authors declare that they have no competing interests