Seasonal Trends of Hair Loss in the Southern Hemisphere: An Analysis of Google Search Patterns

A proportion of humans experience an increase in the rate of daily hair shedding at certain times of the year—a phenomenon referred to as seasonal shedding. Mechanisms of seasonal shedding are incompletely understood; changes in available light and variations in temperature during the year may influence hair shedding [1, 2]. The vast majority of studies of seasonal shedding have focused on the northern hemisphere, with few comprehensive, large-scale investigations specifically studying trends in the southern hemisphere. Previous studies have demonstrated a seasonal pattern of hair loss peaking in summer and early autumn [1-3]. We examined monthly seasonal Google Trends [4, 5] search interest in the southern hemisphere from January 2004 to January 2020 using the search terms "hair loss" and "hair shedding." Seven countries—Peru, Argentina, Chile, Brazil, Australia, New Zealand, and South Africa—met our predefined inclusion criteria: a proportion of English-speaking residents, located greater than 10° south of the equator, and having adequate search volume index (SVI) data in > 90% of months. Adequate SVI was defined as greater than one SVI. We used the SVI of keywords for "hair loss" and "hair shedding" as a proxy measurement for hair loss [4, 5]. We hypothesized that SVI would be highest in late summer/early autumn [1, 2]. Using R, the Prais–Winsten procedure for an AR (1) process (autoregressive, order 1) was applied. SVI was used as the outcome variable, time as a continuous predictor and then as a categorical predictor, and the mean monthly temperature of the seven countries as a covariate. A quarterly model was fit, and data were converted from monthly to quarterly. The Durbin–Watson statistic before and after transformation confirmed first-order autocorrelation. The results demonstrated statistically significant seasonality. The overall F-test for the model with time as a continuous variable was highly significant (p < 1×10−8); the adjusted R-squared was 18.8%. The time index was a significant predictor (p < 0.001), but temperature was not (p = 0.65). Treating time as a categorical predictor, the overall F-test remained substantial (p < 1.6×10−6) with an adjusted R-squared of 18.6%. Using our quarterly model, which examined each quarter (season), we found that the trend reached significance in autumn. SVI rates in autumn months (March, April, and May in the southern hemisphere) were 2.90 times higher (95% CI 1.27, 4.53) compared to spring months (September, October, and November) (Figure 1; Table 1). The seasonality of hair loss needs to be considered when assessing potential causes of a reduction in a patient's hair density. Our findings support the notion that shedding may increase in autumn compared to spring. While our model explains only a modest portion of the variance in SVI, the effect of time remains highly significant whether treated as a continuous or categorical variable. This underscores the importance of seasonality in hair-shedding patterns, despite the influence of other factors. Studies have not uncovered the precise mechanisms of seasonal shedding in humans and have relied mainly on animal studies. Seasonal molting in mammals appears to be controlled by a variety of endogenous and exogenous factors, including changes in air temperature and sunlight [2]. It has been postulated that the photoperiod during summer months and temperature increases may mediate hair shedding through optic and hypothalamic–pituitary pathways [1, 2]. In turn, this may have a regulatory effect on several hormones influencing the hair cycle. In humans, it has been suggested that exogenous factors such as temperature, humidity, and even seasonal viral infection trends may impact seasonal shedding. It will also be important to evaluate whether there is a different susceptibility to shedding across different genetic backgrounds. This is particularly important given that the greatest body of current literature on the subject comes from the northern hemisphere. The authors declare no conflicts of interest.