Explore highly relevant questions in the Baumann skin type questionnaire through the digital skin analyzer: A retrospective single‐center study in South Korea

Baumann skin type questionnaire (BSTQ) has been widely used for evaluating skin types in dermatology. However, it requires excessive assessment time and lacks sufficient clinical validation for the Asian population.


| INTRODUC TI ON
5][6][7][8][9] However, such an approach has drawbacks that it requires expensive equipment and lacks clear reference points to categorize types of skin.Alternatively, the most widely used method is the Baumann Skin type indicator (BSTI), which divides the skin types into four fundamental parameters and classifies them into binary as following; oily or dry (O-D), sensitive or resistant (S-R), pigmented or non-pigmented (P-N), and wrinkled or tight (W-T). 4BSTI has a total of 16 skin type variations and is widely used in dermatology field that it can be linked to appropriate skin management such as cosmetics suitable for each skin type.BSTI is evaluated using the Baumann skin type questionnaire (BSTQ), which is consisted of 64 questions.Its excessive number of items often lowers patients' convenience and takes a long time to be used in the clinical field.What makes the matters worse, BSTQ may not adequately evaluate certain population groups, such as non-Caucasians (e.g., Asian population) as some of the questions are unnecessary or do not sufficiently reflect their skin conditions.
To address these issues, Choi et al. 10 proposed the modified version of BSTQ that consisted of 33 items after applying the original questionnaire to 202 Korean women.Still, there have not been many studies on corroborating questionnaire result with the assessment from a skin analyzer which evaluates the dermatological condition quantitatively.
In this study, we explored the correlation level between the items of modified BSTQ and the result from the skin analyzer, which checks the oiliness, redness, porphyrin, wrinkles, and pigmentation of the participating patients.Based on the correlation level, highly relevant questions were selected from the modified BSTQ based on two proposed selection strategies.

| Study design, data source, and study population
Presented work was a single-center retrospective study, in which the patients of WYNE dermatology clinic (Cheongju-si, Chungcheongbuk-do, and Republic of Korea) who completed both BSTQ and clinical photography assessment from January 2013 to May 2021 were included.Demographic information including age and gender was collected, and any cases where the difference between two evaluation dates exceeded more than 30 days were excluded from the analysis.This study was approved by the electronic Institutional Review Board of the Korea National Institute for Bioethics Policy (IRB no.: 2021-1552-002), and the informed consent was waived.

| Assessment of skin profile by Baumann skin type questionnaire and digital photography analyzer
The patients visiting WYNE dermatology clinic routinely filled out the modified BSTQ implemented by Choi et al., 10 which is composed of six questions on O-D, nine questions on S-R, seven questions on P-N, and eleven questions on W-T respectively.Differences from the original questionnaire are described in the Table S1.The answers from patients were used to categorize their skin types, which are determined based on the previously reported threshold level from Choi et al. 10 Digital photography was taken by MarkVu (PSI plus, Suwon-si, Gyeonggi-do, Republic of Korea), which emits four different lights (normal, specular, polarized, and ultraviolet) to analyze the level of sebum, redness, porphyrin, melanin pigmentation, and wrinkle and so on. 11

| Analysis of relationship between the modified Baumann test and the result of digital photography analyzer
We analyzed the data between the modified BSTQ and the results from the skin analyzer as follows; O-D questions were matched to sebum and porphyrin in each, S-R questions were matched to redness and porphyrin in each, P-N questions were matched to melanin pigmentation, and W-T questions were matched to wrinkle.Subgroup analysis was conducted according to age (less than 29 years or more than 30 years) and gender.
Patients answered the questionnaire by choosing one of the responses that relate to the level of the severity of their symptoms.
And we checked whether the skin measurement value of the corresponding patients match their responses, except for the fifth item which is "unknown".Any items where the number of respondents was less than 30, were excluded from the analysis and were further checked to see if there are any statistical differences between each item.

| Highly relevant question selection from the modified Baumann test based on the result of digital photography analyzer
Highly relevant questions were selected as follows; if the measured value increased with the degree of the response (a) or showed significant statistical difference between the items (b), the following question is assigned as a highly relevant question.We selected based on two strategies: Model 1 included questions fulfilled both

| Compare the result of digital photography analyzer according to the skin type from modified version of BSTQ or from highly relevant questions
The skin type was classified according to the threshold level of modified version of BSTQ or to the threshold level from highly relevant questions above mentioned.The difference of the result of digital photography analyzer between the skin types was evaluated by each threshold level.In addition, the correlation between the score total from questionnaire and the result of digital photography was analyzed.

| Statistical analysis
The comparison analysis was done using two different methods (student t-test or one-way analysis of variance [ANOVA]) depending on the number of groups that were collated with each other.Student t-test was employed only when two groups are compared, and if the number of groups was equal to or greater than three, ANOVA was used, followed by Tukey's honest significant difference test for its post-hoc analysis.Subgroup analysis was conducted by gender (male vs. female) or age (age 30 years or more vs.less than 30 years).
Pearson correlation coefficient (PCC) was calculated to evaluate the linear correlation between the questionnaire and the value of the digital skin analyzer.p-Value < 0.05 was considered statistically significant in this study.All the statistical analyses were carried out using R version 4.0.3software (R Foundation for Statistical Computing).

| Demographics of population
We first found 1191 patients who completed both questionnaire and photographic assessment using a skin analyzer.Among them, we excluded 154 cases where their demographic information was not recorded properly, some questionnaire questions were omitted, or skin analyzer measurement results were not fully recorded (Figure S1).
The measurement results of sebum, redness, melanin pigmentation, and wrinkle were higher in male compared to female (for sebum,

| Highly relevant question selection from the modified Baumann test based on the result of digital photography analyzer
Highly relevant questions are selected based on the correlation level between the questionnaire and skin measurement results.Among O-D questions that are related to sebum level, three and four out of six items satisfied the Models 1 and 2 criteria respectively and were assigned as highly relevant questions.For porphyrin level, three and five questions satisfied the criteria of Models 1 and 2 and were assigned as highly relevant questions (Figure 1A).For S-R questions that are related to redness, two and four questions out of nine items satisfied the criteria of Models 1 and 2. For porphyrin, no questions satisfied the criteria of Model 1 but six questions satisfied the criteria of Model 2 and set as highly relevant questions (Figure 1B).In P-N questions, three and six questions out of seven items satisfied the criteria of Models 1 and 2, in each for melanin pigmentation (Figure 1C).In W-T questions, four and nine questions out of eleven items satisfied the criteria of both Models 1 and 2, in each for wrinkle (Figure 1D).Selected questions for Models 1 and 2 are listed in Tables S1 and S3.Detailed evaluation values for each question and items are shown in Tables S4-S9.
The concordance of skin type by the modified version of BSTQ and by Model 1 or 2 is described in Table S10.Model 1 showed concordance rate ranged from 0.763 (S-R for redness) to 0.862 (P-N), while Model 2 ranged from 0.837 (S-R for redness) to 0.898 (W-T).
F I G U R E 1 Questions of each skin type and fulfillment of selection criteria in this study.
| 3163 The threshold levels to maximize the concordance were described in Table S11.

| The result of digital photography analyzer and skin type from the modified version of BSTQ or highly relevant questions in this study
Based the skin type assessment from the modified version of BSTQ, oily group showed higher value of sebum and porphyrin compared to dry group, sensitive group showed higher value of redness compared to resistant group, pigmented group showed higher value of melanin pigmentation compared to non-pigmented group, and wrinkled group showed higher value of wrinkle compared to tight group (Table 2).However, there was no significant difference in porphyrin between sensitive group and resistant group.
Then the skin type was classified based on highly relevant questions mentioned beforehand.The distribution of skin type according to the models is described in the Table 1.Based on the skin type from selected questions, oily group showed higher value of sebum and porphyrin compared to dry group (Figure 2).Sensitive group showed higher value of redness compared to resistant group but did not show significant difference in porphyrin (Figure 3).Pigmented group showed higher value of melanin pigmentation compared to non-pigmented group, and wrinkled group showed higher value of wrinkle compared to tight group (Figure 4).Detailed values of each group according to threshold levels of Models 1 and 2 is described in Table 2.  S7).Even if the questions were reduced in Models 1 and 2, the PCC value was not lower than the modified version of BSTQ (Table 3).

| DISCUSS ION
In this study, the relationship between the modified BSTQ and the measured digital photography analyzer was evaluated for 1037 Koreans, followed by selecting of highly relevant questions from the questionnaire based on proposed strategies (Models 1 and 2).It was found that the score acquired only from highly relevant questions matched well with that of modified version of BSTQ and skin measurement result, proving the effectiveness of proposed methodology.
Since the skin has various characteristics depending on the condition of individual, there have been attempts to classify skin type. 12ong proposed classification methods, the BSTQ is widely adapted technique in dermatology especially for skin care purpose. 4With the BSTI identified based on the BSTQ, physicians can choose procedures and recommend products that are right for patient's skin.The previous studies showed that the questionnaire was also applied to Korean population. 10,13,14Interestingly, this study showed OSNW was the most common in males and OSNT was the most common in females, similar to the previous studies that reported BSTI in Korean population. 13,14However, the distribution on skin type could show differences depending whether the study was conducted with the general population or with patients who visited the dermatology clinic.
In this study, skin measurement results and score from the questionnaires showed positive correlation in some degree.This suggests that although BSTQ was basically created for the purpose of binary classification, its score can be utilized to determine the degree of the corresponding skin type characteristics.Likewise, Baumann

TA B L E 2
The comparison of skin measurement value according to skin type by modified version of Baumann skin type questionnaire or by two models from this study (Models 1 and 2).et al. 15 reported that the sebum secretion level changed according to the score from the questionnaire.Skin evaluation through questionnaire is subjective and does not necessarily match the results of measurement equipment. 5,16However, the evaluation using professional equipment for skin characteristic such as multiprobe adapter (MPA) system is time-consuming and costly, and some skin characteristics such as sensitive skin are subjective, so it is difficult to evaluate solely with equipment.Thus, the method using a skin type questionnaire might serve as quick and cost-effective alternative or supplement to evaluate the skin.

Modified version of
The original questionnaire consists of 64 questions, which takes a long time to finish.Choi et al. 10 presented the shortened version of BSTQ, which has 33 questions, and WYNE dermatology clinic has been adopted the following questionnaire in their routine clinical setting.Large number of questions in questionnaire can lower the participation rate and cause fatigue from the patient, causing inaccurate result especially via online. 17 Highly relevant questions selection process through two strategies (Models 1 and 2) showed that the result distribution for each question of BSTQ was not even, meaning that they had different weight each other.This is mainly because the original questionnaire was developed based on Western population and does not solely account for Asian population.For example, the questions related to tanning are unnecessary as it is not much preferred custom in the Asian population. 18This study showed a significant inverse relationship between question W-T-11 ('What is your natural skin color [without tanning or self-tanners]?') and the measured wrinkle value.
Most Korean respondents would answer in consideration of their usual skin color regardless of tanning, as.In Koreans, a decrease in skin lightness was reported to be related to an increase in sun exposure, darker skin could be a sign of prolonged sun exposure. 19cluding these questions with low relevance could improve the accuracy of skin type evaluation.
In this study, based on the result of digital photography analyzer, questions that well reflect skin conditions were selected.Although the score was calculated only by a few selected questions, it showed similar or better correlation to the skin measurement value compared to the modified BSTQ.This means that skin types can be classified with less effort and time with highly relevant questions.However, the score from S-R question shows less correlation with the measurement results on skin redness and porphyrin.Only two questions directly linked to redness showed correlation to redness (Table S1).
In acnes for the manifestation or severity of acne. 21,22Thus, as long as the balance between C. acnes subtype, skin barrier, and microbiome maintained, the expression level of porphyrin by C. acnes itself does not mean the sensitive skin. 23In S-R, various conditions including sensitive skin, rosacea and allergic along with acnes were mixed in the score. 4Many skin diseases improve or worsen temporarily by time, so there is a limitation on judging the sensitiveness based on photographic images at certain point in time.Furthermore, sensitive skin itself usually does not show objective signs. 24Therefore, it is necessary to develop a more sophisticated and validated methods integrating subjective and objective properties that can evaluate sensitive skin. 25is study has several limitations.First, as the modified BSTQ was used, other questions that were included only in original BSTQ were not assessed using proposed methods.Since around half of questions were excluded in the modified BSTQ, some significant questions could be excluded in this study.Second, the skin measurements were conducted only by analyzing images, which is an indirect method.If direct method (e.g., skin contact measurement) is used for accurate evaluation, the result could be different in certain degree.In a follow-up study, further analysis using device such as MPA system is needed.Lastly, as the assessment were done for patients who visited a dermatology clinic, there may be differences from the general population.Also, racial differences could be present as the study is carried out for Asians, especially for Koreans.
Still, this study has significance that it compared the modified BSTQ obtained from 1037 patients with the result from a skin measurement device.

| CON CLUS ION
This study investigated the relationship between the result from modified BSTQ and that of digital photography analyzer, and the questions that predominantly determines the value of skin measurement were found for each skin type evaluation (i.e., highly relevant questions).This suggests that skin condition can be evaluated using questionnaire with fewer but essential items.Further research is needed to develop accurate and concise questionnaires for skin condition using precise skin measurement equipment in more diverse and larger population groups.
(a) and (b) conditions in the entire group, and Model 2 included questions fulfilled any of (a) or (b) conditions in the entire group or any subgroup.That is, Model 1 selects highly relevant questions strictly whereas Model 2 chooses lesser strict way.Afterwards, O-D, S-R, P-N, and W-T scores were assigned only by highly relevant questions.The threshold levels of models 1 and 2 were set so that it generates the most similar skin type classification ratio distribution as that of modified version of BSTQ using original threshold values.
Sebum, O-D and porphyrin, and W-T and wrinkle showed low correlation between the score of modified versions of BSTQ and skin measurement value (0.232, 0.230, and 0.217 in each).PCC of S-R and redness, S-R and porphyrin, and P-N and melanin pigmentation showed very low correlation (0.095, 0.051, and 0.150 in each) (Figures S5-

F I G U R E 2
Skin measurement values of sebum and porphyrin according to the oily-dry group.The skin type group was classified by the threshold level of modified version of Baumann skin type questionnaire, Model 1 and Model 2 in this study.F I G U R E 3 Skin measurement values of redness and porphyrin according to the sensitive-resistant group.The skin type group was classified by the threshold level of modified version of Baumann skin type questionnaire, Models 1 and 2 in this study.There was no question selection in the Model 1 for porphyrin.
Therefore, it is necessary to develop a more abbreviated version of the skin type questionnaire, especially for online use.Moreover, as the BSTQ is developed based on the Western population, it has not been fully validated for Asian population.
addition, porphyrin level was associated with the O-D questions, but no clear relationship was observed for the S-R questions.The porphyrin expression under ultraviolet light is not only related to the presence of Cutibaterium acnes (C.acnes) but also the level of serum secretion. 20Besides, C. acnes can act as an opportunistic pathogen, thus porphyrin level is not as important as the type of C. F I G U R E 4 Skin measurement value of melanin pigmentation according to the pigmented-non-pigmented group (upper) and value of wrinkle according to the wrinkled-tight group (lower).The skin type group was classified by the threshold level of modified version of Baumann skin type questionnaire, Models 1 and 2 in this study.
The distribution of patients in skin type classification by modified version of Baumann skin type questionnaire and two models from this study (Models 1 and 2) based on the threshold level.