Evaluating subjective versus objective measures of changes in hair density using standardized scalp photography

Hair loss, complex due to individual and ethnic variations, remains difficult to visually quantify.1Rushton D.H. De Brouwer B. De Coster W. Van Neste D.J. Comparative evaluation of scalp hair by phototrichogram and unit area trichogram analysis within the same subjects.Acta Derm Venereol. 1993; 73: 150-153Crossref PubMed Google Scholar, 2Barman J.M. Astore I. Pecoraro V. The normal trichogram of the adult.J Invest Dermatol. 1965; 44: 233-236https://doi.org/10.1038/jid.1965.42Abstract Full Text PDF PubMed Scopus (96) Google Scholar, 3Jackson A.J. Price V.H. How to diagnose hair loss.Dermatol Clin. 2013; 31: 21-28https://doi.org/10.1016/j.det.2012.08.007Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar Previous research suggests hair loss becomes noticeable to patients only after a 50% reduction in total scalp hair; however, no study has defined the minimal change in hair counts which result in noticeable hair loss.1Rushton D.H. De Brouwer B. De Coster W. Van Neste D.J. Comparative evaluation of scalp hair by phototrichogram and unit area trichogram analysis within the same subjects.Acta Derm Venereol. 1993; 73: 150-153Crossref PubMed Google Scholar, 2Barman J.M. Astore I. Pecoraro V. The normal trichogram of the adult.J Invest Dermatol. 1965; 44: 233-236https://doi.org/10.1038/jid.1965.42Abstract Full Text PDF PubMed Scopus (96) Google Scholar, 3Jackson A.J. Price V.H. How to diagnose hair loss.Dermatol Clin. 2013; 31: 21-28https://doi.org/10.1016/j.det.2012.08.007Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar Our study sought to identify when hair loss becomes noticeable to professionals, thereby informing treatment initiation and adjustment and setting realistic patient expectations. In our NYU School of Medicine institutional review board–approved study, we aimed to identify the minimum discernable hair loss difference recognized by dermatologists. Using the Canfield HairMetrix device, we obtained standardized pretreatment and posttreatment global scalp photographs and hair counts of 100 patients with androgenetic alopecia, stratified into 7 density groups per Sturges’ rule, and developed a survey featuring 2 photos from each density group (Supplementary Fig 1, available via Mendeley at https://doi.org/10.17632/6kchv6h4dh.2). A sample size of 120 dermatologists was selected based on feasibility and the need for diverse perspectives. Of the participating 120 board-certified dermatologists, 100 responded (83.33% response rate), via email (85/105) and in-person (15/15). The data analysis, conducted using Python, used a logistic regression model to estimate the likelihood of dermatologists accurately detecting differences in hair density. A “75% threshold” was determined based on a logistic regression model, where it signifies the hair density difference at which 75% of the dermatologists were accurately able to identify the change. The logistic regression model significantly fit the data and was able to accurately determine the differences in absolute hair density and relative percent differences in hair densities at a 75% critical threshold. Further model parameters and fit indices are detailed in Table I and additional thresholds for subgroups are listed in the supplemental (Supplementary Table I, available via Mendeley at https://doi.org/10.17632/6kchv6h4dh.2). The estimated critical threshold for accurate identification of absolute hair density difference was 43.36 hairs/cm2, whereas for critical threshold for the relative percent difference in hair densities was a difference in hair density of 22.66%, much lower than previously cited 50% scalp hair reduction.Table ILogistic regression model parameters and fit indicesParameterValueStandard error95% CIP valueAbsolute counts R-squared0.832--- RMSE0.133--- Alpha30.023.04(23.40, 36.63)<.001 Beta12.142.8(6.04, 18.25)<.001 Critical threshold (hairs/cm2)43.36-(33.94, 52.78)-Percent difference R-squared0.873--- RMSE0.116--- Alpha15.561.36(12.59, 18.52)<.001 Beta6.471.23(3.79, 9.15)<.001 Critical threshold (%)22.66-(18.49, 26.84)-This table presents the critical parameters and indices of the logistic regression analyses. Both absolute hair density differences and relative percent differences were assessed. For each model, R-squared and RMSE values, reflecting model fit, are reported. The table includes the alpha (intercept) and beta (slope) coefficients for each model, with accompanying standard errors, 95% confidence intervals, and P values, facilitating statistical evaluation. The critical thresholds, corresponding to the hair density differences detectable with 75% accuracy, and their confidence intervals are also provided. Open table in a new tab This table presents the critical parameters and indices of the logistic regression analyses. Both absolute hair density differences and relative percent differences were assessed. For each model, R-squared and RMSE values, reflecting model fit, are reported. The table includes the alpha (intercept) and beta (slope) coefficients for each model, with accompanying standard errors, 95% confidence intervals, and P values, facilitating statistical evaluation. The critical thresholds, corresponding to the hair density differences detectable with 75% accuracy, and their confidence intervals are also provided. The results of our study elucidate the 75% critical threshold at which, notable differences in hair density become perceptible within our study population (Table II). Our study illuminates the importance of objective trichometric measurements as minor shifts in hair density may not be immediately apparent. We do urge caution in extrapolating these findings to nonrepresented populations, despite the diversity of our survey demographics, which span a broad spectrum of dermatologists.Table IIDemographic and professional characteristics of survey participantsGroup (ntotal = 100)No.Gender Female67 Male32 Nonbinary1Age (y)No. Less than 302 31 to 4049 41 to 5026 51 to 6015 61 to 706 Greater than 702Primary area of practiceNo. Medical dermatology49 Pediatric dermatology2 Cosmetic dermatology6 Dermatologic immunology1 Mohs and dermatologic surgery15 Hair loss7 General dermatology17 Other3Years in practiceNo. N/A or current resident1 0 to 536 6 to 1026 11 to 1511 More than 1526No. of hair patients per weekNo. 0 to 542 6 to 1031 11 to 1510 More than 1517This table outlines the demographic and professional characteristics of the 100 dermatologists who participated in our study. Dermatologists were selected in person at the 2023 AAD meeting from a previously compiled email list of both Academic and private practice university affiliates and peers. There were no regional bounds in the selection, encompassing professionals from Asia, Europe, North America, and South America, although exact locations were not specified in the survey data. Participant details are categorized into gender, age, primary area of practice, years in practice, and number of hair patients per week. The data presented can provide insights into the diverse range of perspectives considered in our study, and aid in understanding the potential biases in our findings due to the underrepresentation of certain groups. Open table in a new tab This table outlines the demographic and professional characteristics of the 100 dermatologists who participated in our study. Dermatologists were selected in person at the 2023 AAD meeting from a previously compiled email list of both Academic and private practice university affiliates and peers. There were no regional bounds in the selection, encompassing professionals from Asia, Europe, North America, and South America, although exact locations were not specified in the survey data. Participant details are categorized into gender, age, primary area of practice, years in practice, and number of hair patients per week. The data presented can provide insights into the diverse range of perspectives considered in our study, and aid in understanding the potential biases in our findings due to the underrepresentation of certain groups. This study provides preliminary data on perceptual thresholds for hair density changes, with relative percent differences much lower than previously cited, suggesting that dermatologists are better than expected at predicting the degree of hair loss based on just a clinical examination. As objective data are integral to effective hair loss management, we hope this study encourages further exploration into these perceptual thresholds and their contributing factors, advocating for its extension to diverse patient groups and visually oriented medical disciplines. The research emphasizes the value of objective trichometric measurements and the need for improved, personalized treatment strategies and patient outcomes in hair loss management. Dr. Shapiro is a consultant for Aclaris Therapeutics, Incyte, and Replicel Life Sciences. Drs. Shapiro and Lo Sicco have been investigators for Regen Lab and are investigators for Pfizer. Dr. Lo Sicco is a consultant for Pfizer and Aquis. Authors Buontempo, Oh, Alhanshali, Klein, and Karim have no conflicts to disclose.