Clean beauty gentrification: The role of historic and contemporary resource allocation practices on hair product safety in Boston, MA

Abstract

Background:

Personal care products are a well-documented source of endocrine-disrupting chemicals (EDCs). Differences in exposure to EDCs are reported, with Black women more commonly using products, such as hair products, containing EDCs. Our previous work identified differences in access to safer hair products between neighborhoods in Boston, Massachusetts; however, limited research has explored what may be driving these differences. Redlining and gentrification are two resource allocation practices that may impact neighborhood-level access to health-promoting services and goods, including access to safer hair products.

Objective:

We examined if there is an association between historic and contemporary practices related to resource allocation (i.e., historic redlining and gentrification) with hair product safety (i.e., containing EDCs and other chemicals of concern) among the neighborhoods in Boston, Massachusetts.

Methods:

We randomly selected and visited 117 stores in Boston to collect data on hair product availability and safety. We used the Environmental Working Group’s (EWG) Skin Deep database to evaluate hair product safety through hazard categories (low, moderate, or high). We collected data on historic and contemporary factors at the census tract level (redlining and gentrification). The associations between the historic and contemporary factors and EWG hazard categories were examined using multinomial logistic regressions.

Results:

A total of 39,421 hair products were cataloged in Boston, Massachusetts. Stores in previously Home Owners’ Loan Corporation (HOLC) graded C and D-not gentrified areas had increased access to high-hazard hair products (risk ratio [RR] HOLC C = 1.1; 95% confidence interval [CI] = 0.9, 1.4, RR HOLC D = 1.4; 95% CI = 1.1, 1.7). Decreased access to low-hazard hair products was reported among stores in HOLC D-gentrified or intensely gentrified (RR = 0.7; 95% CI = 0.5, 1.1), HOLC C-not gentrified (RR = 0.7; 95% CI = 0.5, 1.0), and HOLC D-not gentrified areas (RR = 0.5; 95% CI = 0.4, 0.8).

Discussion:

Differences in access to safer hair products were reported when examining the impacts of both redlining and gentrification. Our findings present evidence of the role of historic and contemporary resource allocation practices on current-day neighborhood-level safer hair product accessibility, which may have potential downstream health consequences.

What this study adds:

Areas in Boston, Massachusetts, experiencing prolonged resource deprivation (historically redlined and not gentrified) had significantly increased access to high-hazard hair products and decreased access to low-hazard hair products. Our findings present evidence of the role of resource allocation practices on current-day neighborhood-level safer hair product accessibility, which may have potential downstream health consequences. To our knowledge, this study is one of the first to consider the role of resource allocation on hair product safety using one of the largest databases of city-wide hair product accessibility (~40,000).

Introduction

Personal care products are a well-documented source of endocrine-disrupting chemicals (EDCs),^1,2 synthetic or natural chemicals that can interfere with the body’s hormone system.³ Some commonly used EDCs in personal care products include phthalates and parabens.⁴ Use of personal care products, specifically leave-in products, may increase exposure to EDCs due to the length of time products are left on the skin or scalp.^5,6

While exposure to EDCs is ubiquitous based on their widespread use in a variety of consumer and personal care products, inequities are reported, with communities of color more highly exposed to personal care product-associated EDCs.^4,7,8 These inequities in exposure have persisted over time.^9–11 A portion of these differences in exposure may be driven by differences in hair product use patterns—Black communities are found to more commonly use leave-in hair products (i.e., hair oil, pomade, and leave-in conditioners) and hair relaxers compared with other racial/ethnic groups.^12–14 Hair products are a mixture of chemicals and ingredients, with some of the more studied chemicals of concern including EDCs, formaldehyde, and formaldehyde-releasing preservatives.^1,2,15 Concentrations of these chemicals of concern may differ by hair product type based on their function—for example, anti-frizz products have higher concentrations of cyclosiloxanes (suspected EDCs used for a smooth or silky texture) compared with other products.¹ Furthermore, in addition to chemical composition, the frequency and duration of product use contribute to higher exposure to EDCs.^1,5,6,16 Previous studies demonstrate that hair products are an important contributor to socially driven health disparities.¹⁷ Exposure to these chemicals of concern may be associated with a range of health outcomes across the life course, including reproductive/developmental, cardiometabolic, and cancer.^18–21

Personal care product use is driven by a variety of factors, including those at the individual, interpersonal, community/neighborhood, and societal levels.^22,23 For example, the price and effectiveness of products are common factors that individuals consider when selecting products to use.^14,24 Interpersonal interactions, including those among family or in workplace settings, may also contribute to product use.^25,26 Additionally, the use of certain products, such as chemical hair straighteners, may be driven by racialized beauty standards that value “good hair” as “straight, silky, bouncy, manageable, healthy, and shiny.”²⁷ Considering neighborhood-level factors, our previous analysis was one of the first to explore differences in access to safer hair products, and observed, in eight sociodemographically diverse neighborhoods in Boston, Massachusetts, that lower-income communities and/or communities of color had an increased risk of finding high-hazard hair products (indicating less safe) compared with the predominately white and higher socioeconomic status reference neighborhood.²⁸ Differences in access to safer hair products may be an example of retailer redlining, which is the discriminatory practice of differential treatment or not serving certain customers or communities based on their sociodemographic characteristics.²⁹ Additionally, historical or contemporary practices that impact the distribution of investments and resources may contribute to differences in access to safer products (Supplemental Figure S1; https://links.lww.com/EE/A401).²⁸

In recent years, the role of discriminatory practices on access to neighborhood-level resources and health has received significant attention.^30–34 Notably, research has considered the role of historic residential redlining, a legalized form of structural racism in the 1930s by the Home Owners’ Loan Corporation (HOLC) that involved limiting lending and consequently home ownership and wealth among certain communities based on their perceived suitability for refinancing.^35,36 Neighborhoods that were predominantly lower-income residents of color and immigrants received a grade of D, indicating “hazardous,” and were shaded red on maps (hence the name redlining), while neighborhoods that were predominantly white and higher income received a grade of A, indicating “best.” Historically, redlining enforced racial residential segregation and led to disinvestment in areas with worse HOLC grades (C and D).³⁰

Contemporary practices may also impact residential mobility and the allocation of resources in certain communities.^37–39 Gentrification is a debated multidimensional process of neighborhood change that occurs over time and is noted by sociodemographic changes in residents, increased rent or property values, and physical neighborhood changes.^40,41 Typically, younger and higher-income individuals move into lower-income areas, and through reinvestment and rising housing prices, the original lower-income residents may be displaced.^42,43 This process is characterized by reinvestment in areas that are historically disadvantaged⁴²—for example, increased access to healthy food, greenspace, infrastructure that promotes physical activity, improved physical neighborhood changes, and housing investment.^40,44 Gentrification may be an important contemporary practice in reshaping the urban neighborhood landscape that has been impacted by the downstream and long-term effects of historic redlining with regard to access to health-promoting resources and residential sociodemographic composition. Specifically, longstanding residents may be displaced or priced out of accessing these health-promoting resources. While these practices may operate independently, patterns of resource allocation are increasingly being reported, where communities that were historically redlined are contemporarily being gentrified.^38,39 However, other communities may experience prolonged or persistent resource deprivation (areas that were historically redlined and not gentrified)—which may impact access to healthy goods, services, and other factors. These resource allocation practices may have distinct implications in terms of access to safer hair products. Specifically, gentrification may increase access to hair products due to the potential increase in health-promoting goods and stores in areas historically resource deprived, such as redlined communities.⁴⁵ While a greater number of health-promoting goods and stores, including safer hair products, may be available in gentrifying neighborhoods, the ability for longer-term residents to afford these products may be limited due to higher costs and displacement. In comparison, due to resource deprivation and decreased access to health-promoting stores and goods, there may be decreased access to safer hair products in areas that historically experienced redlining and are not undergoing gentrification.³⁶

Previous research has explored how these complex historical and contemporary processes impact access to other health-promoting resources.^{30,33,37,46,47} Yet, research has yet to evaluate how these resource allocation practices impact access to safer hair products, a potential health-promoting or harmful factor that may contribute to reported health inequities.^12,23 Specifically, areas that have and continue to experience resource deprivation (neighborhoods that were historically redlined and currently have not undergone gentrification) may have less access to health-promoting stores, services, and goods, including hair products, due to prolonged disinvestment in these neighborhoods. Resource allocation practices may be one of a complex combination of factors that contribute to product use patterns with downstream health consequences (Supplemental Figure S1; https://links.lww.com/EE/A401). Identifying potential drivers of inequities in access to safer hair products—both historical and contemporary—is an important step toward developing solutions or interventions. Thus, we cataloged hair products sold in a random selection of stores in each neighborhood in Boston, Massachusetts, to examine their safety. Next, we examined if there was an association between historic redlining and contemporary gentrification with hair product safety (i.e., containing EDCs and other chemicals of concern evaluated by the Environmental Working Group’s [EWG] Skin Deep Database) in Boston, Massachusetts. We hypothesized that stores located in areas that have experienced historical redlining would be more likely to sell hazardous hair products, while stores located in areas that have experienced gentrification would be more likely to sell safer hair products. Furthermore, we hypothesized that neighborhoods that were historically redlined and that had not undergone contemporary gentrification—an indicator of long-term, persistent low resource allocation—would be at higher risk of having higher hazard hair products compared to communities that were not redlined and contemporarily gentrified.

Methods

Study area

The study area included all 24 neighborhoods in Boston designated by the City of Boston’s Office of Neighborhood Services: Allston, Back Bay, Bay Village, Beacon Hill, Brighton, Charlestown, Chinatown-leather district, Dorchester, Downtown, East Boston, Fenway, Hyde Park, Jamaica Plain, Longwood, Mattapan, Mission Hill, North End, Roslindale, Roxbury, South Boston, South Boston Waterfront, South End, West End, and West Roxbury. In general, Boston is home to more than 600,000 residents (American Community Survey [ACS] 2014–2018 5-year estimates).⁴⁸ In terms of residential racial/ethnic composition, 47% identified as white, 22.4% identified as non-Hispanic Black, 17.4% identified as Hispanic, and 8.9% identified as Asian (Supplemental Table S1; https://links.lww.com/EE/A401).⁴⁸

Store identification and selection

Through market research (examining trends and patterns in consumption), we identified the most common store types where consumers in the United States (US), including those from diverse backgrounds, purchase personal care and cosmetic products (pharmacies, department stores, grocery stores, and beauty supply stores).^49–51 Next, we used 2017 North American Industry Classification System (NAICS) codes through Environmental Systems Research Institute (ESRI) ArcGIS Business Analyst to identify these four store types within each neighborhood (pharmacies NAICS: 44611009, department stores: 45221001, grocery stores: 44511003, and beauty supply stores: 44612004). A total of 399 stores were identified in the study area. We excluded any stores that were not located at the address provided or a specialty market that was unlikely to sell hair products (e.g., butcher) (n = 153). The remaining 246 stores were in our target population before stratified random sampling. The store identification and selection process occurred in 2021. Using the NAICS codes, no stores were identified in Bay Village (one of the smallest neighborhoods in Boston); thus, the neighborhood was excluded from our study area. Supplemental Figure S2; https://links.lww.com/EE/A401 documents the store selection and visitation process.

We used stratified random sampling to allow for the heterogeneity of store types that are commonly visited for the purchasing of personal care products by US consumers, with particular attention to store types visited by communities of color.⁵⁰ Chan et al²⁸ 2023 describes the stratified random sampling process in more depth. A total of 135 stores were randomly selected and visited for data collection. Out of the 135 stores that were visited in the Boston neighborhoods, 117 had hair products for sale, agreed to have photos taken, did not have a language barrier, and were at the provided address (Supplemental Figure S2; https://links.lww.com/EE/A401).

Hair product data collection

Photographs were taken of the hair products section of each store in two phases: January to March 2021 and September to March 2022. After documenting the hair products for sale, the products’ information (brand name, product name, and price) and the stores’ information (name, neighborhood, and date of data collection) were identified and added to our database. Based on the product name and branding, the hair products were categorized into 12 product types (e.g., shampoo, conditioner, and hair oil) as done in prior research^12,28—the full list of hair product types is presented in Supplemental Table S2; https://links.lww.com/EE/A401. All products marketed to women or girls/children on the label or based on branding were included in the analysis based on prior research that identified this subpopulation as highly exposed to personal care product EDCs.^9–11 The following products were excluded: products marketed to or branded for men, hair dyes/temporary hair color, packages with multiple hair products, essential oils, hair supplements, lice products, and travel/sample-sized products. Additional information and rationale for the exclusions are detailed elsewhere.²⁸

Environmental working group hazard score

To evaluate hair product safety, we used EWG’s Skin Deep Database, which is an online database that allows consumers to learn about the safety of their personal care products through a calculated numeric hazard score.⁵² To date, Skin Deep has more than 130,000 products from 6,154 brands and is one of the largest databases of personal care product safety available.⁵³ In summary, hazard scores are developed by comparing the ingredients to 60 toxicity and regulatory databases and considering 17 hazard categories.⁵⁴ A two-part score is developed based on ingredient hazard and data availability. More information about the development of the EWG score is reported elsewhere.^28,54 In terms of ingredients, the presence of chemicals of concern, including undisclosed fragrance and EDCs, may contribute to higher hazard scores, while the absence of these same ingredients may contribute to lower hazard scores. EWG hazard scores range from the lowest score of EWG-verified⁵⁵ to the highest score of 10. Products with an EWG-verified score were assigned a 0, representing the least harmful based on their listed ingredients. Furthermore, we categorized scores into low (hazard score 0–2), moderate (3–6), and high hazard (7–10), as done by EWG. We assigned all products found in the Skin Deep database the most recent hazard score available; products not found in the database were assigned a missing value. The latest entry of EWG scores into our database was completed in April 2024. We implemented three quality assurance and control (QA/QC) procedures to evaluate the accuracy and completeness of our database (Supplemental Table S3; https://links.lww.com/EE/A401). All neighborhoods in the analysis met our QA/QC criteria.

Historic and contemporary neighborhood variables

Historic residential redlining

Historic residential redlining (implemented in the 1930s and banned in the 1960s) was evaluated through HOLC security maps using 2020 census tract boundaries accessed through the University of Richmond’s Mapping Inequality Project.³⁵ HOLC distributed grades (A–D) based on an area’s perceived risk of financial investment determined by factors such as racial/ethnic composition, immigration status, and location of perceived “amenities” and “disamenities.”^34,35 Each store visited in Boston was geocoded based on the business’ address obtained from ESRI ArcGIS Business Analyst.⁵⁶ We assigned each store the redlining grade (A–D) of the HOLC polygon where it was located, and each product was assigned the HOLC grade of the store. Stores that were not located within a HOLC-graded polygon were excluded from this analysis (Supplemental Figure S3; https://links.lww.com/EE/A401).

Contemporary gentrification

In combination with historic redlining, we examined contemporary gentrification as a way of assessing sociohistorical and contemporary processes of resource allocation. Gentrification is a contemporary neighborhood change process characterized by a sociodemographic shift in residents driven by focused resource allocation to a previously disinvested area.^40,42,57 We used Drexel University Urban Health Collaborative’s measure of gentrification, which first identifies areas eligible for gentrification and second identifies areas that may be gentrified through changes in the percentage of residents with a college degree or higher and changes in rent or home value over time compared with the median change in the city or state. More details on this measure of gentrification are reported elsewhere.⁴⁰ We used data from the ACS 5-year estimates 2008–2012 and 2015–2019 as proxies for neighborhood characteristics in 2010 and 2020, as done in previous research.⁴¹ Furthermore, these data were integrated into 2010 census tract boundaries to account for changing tract boundaries over time using Brown University’s Longitudinal Tract Database.⁵⁸ Census tracts were categorized as not gentrified, gentrified (above median increase in percentage of residents with a college degree or higher and 50–75th percentile increase for median rent or 50–75th percentile increase for median home value), or intensely gentrified (above median increase in percentage of residents with a college degree or higher and 75th percentile or above increase for median rent or 75th percentile or above increase for median home value). Stores were assigned values for gentrification based on the census tract where they were located, and each hair product was assigned the value of gentrification for that store.

We created an additional variable to consider the role of both redlining and gentrification on neighborhood-level access to safer hair products (Supplemental Table S4; https://links.lww.com/EE/A401). In brief, this new variable ranged from 1 to 4. When looking at both redlining and gentrification, there were no stores located in HOLC grade B areas; thus, stores located in HOLC C-gentrified or intensely gentrified areas were categorized as the reference (category 1). The remaining categories included HOLC D-gentrified or intensely gentrified (2), HOLC C-not gentrified (3), and HOLC D-not gentrified (4).

Statistical analysis

The count and percentage of stores by store type (department, grocery, pharmacy, and beauty supply store) and by our exposures of interest (1) redlining, (2) gentrification, and (3) redlining and gentrification were calculated. We also reported the count and percentage of hair products with missing EWG scores by our exposures of interest. The total count of hair products in each neighborhood was assessed based on several variables, including EWG hazard category (low, moderate, or high), EWG hazard score (0–10), and product types (e.g., shampoo, hair gel, hair spray, etc.). We also examined whether there was a statistically significant difference in EWG hazard categories between each neighborhood and Beacon Hill, our reference neighborhood that is predominantly white and of higher socioeconomic status, using Pearson’s chi-square tests.

We first explored if there was an association between neighborhood (categorical) and EWG hazard category among all neighborhoods in Boston using a multinomial logistic regression. The moderate-hazard category was chosen as our reference due to the small number of low-hazard hair products across the neighborhoods in Boston, and driven by most products scoring in the moderate range (EWG scores 4–6) (Supplemental Table S2; https://links.lww.com/EE/A401). To consider the role of missingness in this analysis, we added a fourth category to the EWG hazard categories (low, moderate, high, and missing). We also explored whether there were differences in safer hair product availability between different neighborhood locations of the same pharmacy chain.

For our main analyses, we examined the associations between historic redlining (HOLC grades A–D), contemporary gentrification (not gentrified, gentrified, and intensely gentrified), and combined assessment of redlining and gentrification (HOLC C-gentrified or intensely gentrified, HOLC D-gentrified or intensely gentrified, HOLC C-not gentrified, and HOLC D-not gentrified) with EWG hazard categories (low, moderate, or high) using individual multinomial logistic regressions. Hair products found in Chinatown were excluded from these analyses based on the small sample size of products with available EWG scores; however, products found in Chinatown were included in the descriptive statistics (Figure 1, Supplemental Table S2; https://links.lww.com/EE/A401). We used a complete case analysis for all models. Statistical analyses were conducted using R version 4.1.1, and mapping was completed using ESRI ArcGIS Pro.

Figure 1.

Stacked bar chart of the percentages of low (0–2), moderate (3–6), and high (7–10) Environmental Working Group hazard score hair products in each neighborhood in Boston, Massachusetts. Each neighborhood was assigned a number from 1 to 23; 1 = Beacon Hill, 2 = Chinatown, 3 = Dorchester, 4 = Downtown, 5 = East Boston, 6 = Mattapan, 7 = Mission Hill, 8 = Roxbury, 9 = Fenway, 10 = Longwood, 11 = Allston, 12 = Back Bay, 13 = South Boston, 14 = Jamaica Plain, 15 = Hyde Park, 16 = Roslindale, 17 = West Roxbury, 18 = South Boston Waterfront, 19 = South End, 20 = Brighton, 21 = West End, 22 = North End, 23 = Charlestown.

Results

Neighborhood characteristics

Supplemental Table S1; https://links.lww.com/EE/A401 presents the neighborhood sociodemographic and geographic characteristics based on the American Community Survey 2014–2018 5-year estimates and 2010 Census at the census block group level. Differences in percent below the poverty line and educational attainment were reported between neighborhoods in Boston, Massachusetts. Furthermore, residential racial/ethnic composition differed by neighborhood—Beacon Hill and North End had the highest percentage of white residents. In comparison, Mattapan and Roxbury had the highest percentage of non-Hispanic Black residents. Finally, Chinatown had the highest percentage of Asian residents, and East Boston had the highest percentage of Hispanic residents.

Store and hair product characteristics

In total, 13 department stores, 52 grocery stores, 44 pharmacies, and 8 beauty supply stores were visited for data collection and included in our analytic sample (Supplemental Table S2; https://links.lww.com/EE/A401). Supplemental Table S5; https://links.lww.com/EE/A401 details the counts and percentages of store types by our exposures of interest. Supplemental Figures S4A–C; https://links.lww.com/EE/A401 present maps of the store locations and their assigned exposures of interest across Boston, Massachusetts. The majority of stores were located in HOLC grade C areas, regardless of store type. None of the stores randomly selected and visited in this analysis were located in HOLC grade A areas. In terms of missing EWG scores by HOLC grades, grades B, C, and D had 28.5%, 42.7%, and 31.8% missing EWG hazard scores, respectively. Furthermore, the majority of stores were located in areas that were not gentrified (n = 53 vs. n = 17 in gentrified, and n = 4 in intensely gentrified). Of the stores in gentrified or intensely gentrified areas, the most common type was grocery stores (52.9% and 75%, respectively). Stores in gentrified or intensely gentrified areas were found across the city, specifically in East Boston, Brighton, Allston, South Boston, Dorchester, Mattapan, Roxbury, and Mission Hill. The percentage of missing EWG scores also differed by gentrification classification (not gentrified: 41.8%; gentrified: 32.7%; and intensely gentrified: 32.9%). When accounting for both redlining and gentrification, regardless of store type, the highest percentage of stores were found in HOLC C-not gentrified areas (department: 66.7%; grocery: 46.7%; pharmacy: 45.5%; and beauty supply: 60.0%). Additionally, the highest percentage of missing EWG scores were found among stores in HOLC C-not gentrified areas (45.8%).

Figure 1 and Supplemental Table S2; https://links.lww.com/EE/A401 present the hair product characteristics by neighborhood. A total of 39,421 hair products were cataloged. Of the stores selected and visited, Dorchester had the most total products (n = 6,072), and Chinatown had the least (n = 159). Shampoo and conditioner were the most frequent product types found across all 23 neighborhoods. Hair relaxers and hair spray/mousse had the highest mean EWG scores (6.6 and 5.8, respectively), and hair oil and children’s hair products had the lowest mean EWG scores (4.2 and 4.6, respectively). As a note, the hair oil category consisted of both single-ingredient oils (e.g, coconut and almond), which generally had lower EWG scores, and oils that were a mixture of ingredients, which generally had higher EWG scores. Chinatown, North End, West Roxbury, and Roslindale had the highest percentage of high-hazard hair products (EWG hazard scores ≥7) (33.3%, 14.8%, 13.1%, and 12.8%, respectively). Of note, Chinatown had the highest percentage of hair products with missing EWG scores (94.3%), followed by Roxbury (54.6%), Jamaica Plain (52.8%), and Roslindale (52.2%)—all four of these neighborhoods are communities of color (Supplemental Table S2; https://links.lww.com/EE/A401). In comparison, North End and Charlestown, both predominantly white communities, had the lowest percentage of hair products with missing EWG scores (19.5% and 18.8%, respectively). Pearson’s chi-square tests examining differences in EWG hazard categories comparing each neighborhood to Beacon Hill reported statistically significant differences at the 0.05 significance level for Roxbury, Hyde Park, Roslindale, West Roxbury, and North End (Supplemental Table S2; https://links.lww.com/EE/A401).

Access to safer hair products between neighborhoods

Supplemental Table S6; https://links.lww.com/EE/A401 presents the risk ratios (RR) of finding hair products in stores with low, high, or missing EWG hazard scores relative to moderate EWG hazard scores in each neighborhood compared with Beacon Hill. Among certain lower-income communities of color (i.e., Roxbury and Roslindale) and higher-income white communities (i.e., North End and West Roxbury), a significantly increased risk of finding high versus moderate hazard hair products was reported compared with Beacon Hill. Furthermore, several communities of color (Roxbury, Jamaica Plain, Dorchester, Mattapan, and Roslindale) reported the highest significantly increased risk of finding products with missing EWG scores compared with moderate-hazard scores.

Differences in access to safer hair products were also reported among the pharmacy chain that was in 20 neighborhoods (Table 1). Specifically, Jamaica Plain (RR = 1.8; 95% confidence interval [CI] = 1.1, 2.9), Roslindale (RR = 1.7; 95% CI = 1.0, 2.7), and South End (RR = 1.7; 95% CI = 1.0, 2.8) had an increased risk of finding high versus moderate-hazard hair products in the pharmacy chain compared with the pharmacy chain in Beacon Hill. In comparison, almost every neighborhood had a decreased risk of finding low-hazard hair products. West End (RR = 0.3; 95% CI = 0.1, 0.9) and Charlestown (RR = 0.3; 95% CI = 0.1, 0.8) both had statistically significant decreased risks.

Table 1.

Risk ratio (RR) and 95% confidence intervals of finding hair products with low and high Environmental Working Group (EWG) hazard scores versus hair products with moderate EWG hazard scores in a pharmacy chain comparing each neighborhood to Beacon Hill in Boston, Massachusetts, 2021–2024 (n = 10,210)

Neighborhooda	Low	Moderateb	High
1: Beacon Hillc (n = 245)	Ref	Ref	Ref
3: Dorchester (n = 430)	1.0 (0.4, 2.4)	Ref	1.0 (0.6, 1.7)
4: Downtown (n = 591)	0.6 (0.3, 1.5)	Ref	1.1 (0.7, 1.8)
5: East Boston (n = 625)	0.8 (0.4, 2.0)	Ref	1.2 (0.7, 2.0)
6: Mattapan (n = 292)	0.9 (0.3, 2.3)	Ref	1.3 (0.7, 2.2)
8: Roxbury (n = 305)	0.7 (0.2, 1.9)	Ref	1.0 (0.5, 1.7)
9: Fenway (n = 701)	0.9 (0.4, 2.0)	Ref	0.9 (0.5, 1.5)
10: Longwood (n = 449)	0.7 (0.3, 1.9)	Ref	1.1 (0.6, 1.8)
11: Allston (n = 517)	0.5 (0.2, 1.4)	Ref	0.8 (0.5, 1.4)
12: Back Bay (n = 498)	0.6 (0.2, 1.5)	Ref	0.9 (0.5, 1.5)
13: South Boston (n = 414)	0.7 (0.3, 1.9)	Ref	0.9 (0.5, 1.6)
14: Jamaica Plain (n = 451)	0.4 (0.1, 1.1)	Ref	1.8 (1.1, 2.9)
15: Hyde Park (n = 546)	0.6 (0.2, 1.4)	Ref	1.0 (0.6, 1.7)
16: Roslindale (n = 710)	0.5 (0.2, 1.3)	Ref	1.7 (1.0, 2.7)
18: South Boston Waterfront (n = 585)	0.5 (0.2, 1.3)	Ref	0.7 (0.4, 1.2)
19: South End (n = 508)	0.4 (0.1, 1.1)	Ref	1.7 (1.0, 2.8)
20: Brighton (n = 510)	0.4 (0.1, 1.1)	Ref	0.9 (0.5, 1.5)
21: West End (n = 685)	0.3 (0.1, 0.9)	Ref	1.6 (1.0, 2.6)
22: North End (n = 426)	0.3 (0.1, 1.0)	Ref	1.6 (1.0, 2.7)
23: Charlestown (n = 722)	0.3 (0.1, 0.8)	Ref	1.3 (0.8, 2.1)

Risk ratio with 95% confidence intervals of finding hair products with low (0–2) and high (7–10) EWG hazard scores compared with hair products with moderate (3–6) EWG hazard scores in a pharmacy chain located in 20 of the 23 neighborhoods in Boston compared with the pharmacy chain in Beacon Hill, estimated by multinomial logistic regression. Chinatown (neighborhood = 2) was excluded based on the small number of hair products with EWG scores.

^aEach neighborhood was assigned a number from 1 to 23; 1 = Beacon Hill, 3 = Dorchester, 4 = Downtown, 5 = East Boston, 6 = Mattapan, 7 = Mission Hill, 8 = Roxbury, 9 = Fenway, 10 = Longwood, 11 = Allston, 12 = Back Bay, 13 = South Boston, 14 = Jamaica Plain, 15 = Hyde Park, 16 = Roslindale, 17 = West Roxbury, 18 = South Boston Waterfront, 19 = South End, 20 = Brighton, 21 = West End, 22 = North End, 23 = Charlestown.

^bThe moderate-hazard category (3–6) was the reference group.

^cBeacon Hill, which is predominantly non-Hispanic White and of higher socioeconomic status, was the reference neighborhood.

Historic and contemporary practices and hair product safety

Table 2 presents the RRs of finding hair products in the stores with low and high EWG hazard scores versus moderate EWG hazard scores in worse HOLC-graded areas compared with HOLC grade B areas. Among stores located in worse HOLC-graded areas, there was a significantly decreased risk of finding high-hazard hair products compared with moderate-hazard hair products (HOLC C RR = 0.6; 95% CI = 0.5, 0.8; HOLC D RR = 0.6; 95% CI = 0.5, 0.8). In comparison, increased RRs were reported for finding low-hazard hair products in stores located in worse HOLC-graded areas; however, the estimates were not statistically significant, and associated CIs were noticeably wider.

Table 2.

Risk ratio (RR) and 95% confidence intervals of finding hair products with low and high Environmental Working Group (EWG) hazard scores versus hair products with moderate EWG hazard scores in stores located in areas historically Home Owners’ Loan Corporation (HOLC) graded C and D versus HOLC B (reference group) in Boston, Massachusetts, 2021–2024 (n = 17,594)

	HOLC grade
EWG risk	Bb (n = 660)	C (n = 10,632)	D (n = 6,302)
Low	Ref	1.3 (0.8, 2.2)	1.2 (0.7, 2.2)
Moderatea	Ref	Ref	Ref
High	Ref	0.6 (0.5, 0.8)	0.6 (0.5, 0.8)

Risk ratio with 95% confidence intervals of finding hair products with low (0–2) and high (7–10) EWG hazard scores compared with hair products with moderate (3–6) EWG hazard scores in stores located in historically graded HOLC C and D areas versus HOLC B areas, estimated by multinomial logistic regression.

^aThe moderate-hazard category (3–6) was the reference group.

^bHOLC grade B was the reference group for HOLC grades.

Table 3 presents the RRs of finding hair products with low and high EWG hazard scores in stores located in areas classified as gentrified and intensely gentrified versus not gentrified. Stores in gentrified or intensely gentrified areas had a slightly decreased risk of selling high versus moderate-hazard hair products compared with stores located in not gentrified areas (gentrified RR = 0.9; 95% CI = 0.8, 1.1; intensely gentrified RR = 0.8; 95% CI = 0.6, 1.1). Stores in gentrified areas had a slightly increased risk of finding low-hazard hair products (RR = 1.2; 95% CI = 1.0, 1.5), while stores in intensely gentrified areas had a decreased risk of finding low-hazard hair products (RR = 0.5; 95% CI = 0.2, 1.1).

Table 3.

Risk ratio (RR) and 95% confidence intervals of finding hair products with low and high Environmental Working Group (EWG) hazard scores versus hair products with moderate EWG hazard scores among stores in gentrified or intensely gentrified areas versus not gentrified areas (n = 16,154)

EWG risk	Not gentrifiedb (n = 11,316)	Gentrified (n = 4,312)	Intensely gentrified (n = 526)
Low	Ref	1.2 (1.0, 1.5)	0.5 (0.2, 1.1)
Moderatea	Ref	Ref	Ref
High	Ref	0.9 (0.8, 1.1)	0.8 (0.6, 1.1)

Risk ratio with 95% confidence intervals of finding hair products with low (0–2) and high (7–10) EWG hazard scores compared with hair products with moderate (3–6) EWG hazard scores, comparing stores in gentrified or intensely gentrified areas to those in not gentrified areas, estimated by multinomial logistic regression.

^aThe moderate-hazard category (3–6) was the reference group.

^bStores in census tracts that were not gentrified were the reference group.

Table 4 presents the RRs of finding hair products with low and high EWG hazard scores compared with moderate EWG scores in stores located in areas classified by HOLC grade and gentrification. In general, stores in areas classified as HOLC D-gentrified or intensely gentrified and HOLC C or D-not gentrified had a decreased risk of finding low-hazard hair products compared with stores in HOLC C-gentrified or intensely gentrified areas. Additionally, a pattern was observed of increasing RRs of finding high-hazard hair products when examining stores in HOLC D-gentrified or intensely gentrified areas, HOLC C or D-not gentrified areas (HOLC D-gentrified or intensely gentrified RR = 1.0; 95% CI = 0.8, 1.3; HOLC C-not gentrified RR = 1.1; 95% CI = 0.9, 1.4; HOLC D-not gentrified RR = 1.4; 95% CI = 1.1, 1.7). Overall, stores in HOLC D-not gentrified areas had a statistically significant decreased risk of finding low-hazard hair products and a statistically significant increased risk of finding high-hazard hair products compared with stores in HOLC C-gentrified or intensely gentrified areas.

Table 4.

Risk ratio (RR) and 95% confidence intervals of finding hair products with low and high Environmental Working Group (EWG) hazard scores versus hair products with moderate EWG hazard scores among stores located in areas classified as Home Owners’ Loan Corporation (HOLC) D-gentrified or intensely gentrified, HOLC C-not gentrified, and HOLC D-not gentrified versus HOLC C-gentrified or intensely gentrified (n = 11,624).

	HOLC C-gentrified or intensely gentrifiedb (n = 1,185)	HOLC D-gentrified or intensely gentrified (n = 1,629)	HOLC C-not gentrified (n = 6,142)	HOLC D-not gentrified (n = 2,668)
Low	Ref	0.7 (0.5, 1.1)	0.7 (0.5, 1.0)	0.5 (0.4, 0.8)
Moderatea	Ref	Ref	Ref	Ref
High	Ref	1.0 (0.8, 1.3)	1.1 (0.9, 1.4)	1.4 (1.1, 1.7)

Risk ratio with 95% confidence intervals of finding hair products with low (0–2) and high (7–10) EWG hazard scores compared with hair products with moderate (3–6) EWG hazard scores comparing stores located in areas HOLC D-gentrified or intensely gentrified, HOLC C-not gentrified, and HOLC D-not gentrified versus HOLC C-gentrified or intensely gentrified, estimated by multinomial logistic regression.

^aThe moderate-hazard category (3–6) was the reference group.

^bStores in census tracts classified as HOLC C-gentrified or intensely gentrified were the reference group.

Discussion

This analysis explored the associations between historic (i.e., redlining) and contemporary (i.e., gentrification) resource allocation practices with hair product safety in Boston, Massachusetts. Differences in access to safer hair products were observed. Notably, differences were reported among the same pharmacy chain in 20 neighborhoods. Stores in areas with worse HOLC grades had decreased access to high-hazard hair products compared with stores in HOLC-graded B areas. Furthermore, stores in areas that were gentrified or intensely gentrified had decreased access to high-hazard hair products compared with stores in areas not gentrified. When looking at both redlining and gentrification, stores in HOLC D-not gentrified areas had significantly increased access to high-hazard hair products and decreased access to low-hazard hair products. We also reported that several communities of color had an increased risk of finding products with missing EWG scores. Given these missing data and the fact that many estimates had CIs that included the null, these findings should be interpreted with caution.

We reported differences in access to safer hair products among the same pharmacy chain, indicating that stores may target or market certain products to specific neighborhoods within the same city. This finding may be an example of retailer redlining, which has been considered in other research regarding the prevalence of dollar stores (a store type that targets lower-income communities) and pharmacies among communities of color.^59–61 However, research has reported that increased access to certain store types may not translate to product availability.⁶² This analysis may have important implications for the health of communities if products available in communities of color are more hazardous. Furthermore, these findings may in part explain the higher exposure to personal care product chemicals among Black communities,¹⁶ who more commonly use products containing chemicals of concern (e.g., EDCs, formaldehyde).^1,2,15

In the current analysis, neighborhoods with differing sociodemographic characteristics (lower-income communities of color and higher-income white communities) had a higher risk of finding high-hazard hair products. This finding differs from the predominantly lower-income and/or communities of color identified as having a higher risk of finding high-hazard hair products in our previous analysis.²⁸ However, the same neighborhoods previously identified were also noted in the current analysis. Three factors may contribute to these results. First, in our previous analysis, we intentionally selected neighborhoods based on the highest and lowest percentages of racial/ethnic groups and percentages below the poverty line to create contrast, while the current analysis includes all neighborhoods in the city of Boston and reflects more heterogeneity in sociodemographic characteristics and product availability. Second, in higher-income, predominantly white communities where a higher risk of high-hazard hair products was reported, there may be niche, smaller brands that are falsely marketed as natural or “greenwashed” (disinformation to present a product or program as environmentally responsible or healthy).⁶³ In recent years, increased consumer demand for sustainable, natural, and organic personal care products has resulted in the notable growth of this market space and the resulting revenue.^64,65 However, branding and marketing of personal care products in the United States are not regulated in terms of the use of words such as organic.⁶⁶ Future research should further explore the different mechanisms through which high-hazard hair products are sold in certain communities. Lastly, these results may be driven by missing data—as previously discussed, communities of color had higher percentages of missing data, and predominantly white communities had lower percentages of missing data.²⁸

Redlining and gentrification are both discriminatory resource allocation practices—meaning that decisions were/continue to be made by both public and private entities in terms of what communities should or should not receive investments or resources.^30,67 In the context of redlining, by limiting lending in lower-income, immigrant communities, and/or communities of color, these communities were barred from building wealth and receiving investments through homeownership and small businesses.^30,35 In comparison, for communities undergoing gentrification, the movement of typically higher-income, educated residents into lower-income communities is coupled with reinvestment into these previously disadvantaged areas.⁴⁰ Driven by rising rent and housing prices, the original residents may be displaced; however, residential mobility in the context of gentrification is debated.⁴³

Specifically, our analyses illustrate that these resource allocation practices may impact access to safer hair products both independently and jointly. We presented evidence that stores in areas that were not gentrified and had higher HOLC grades had decreased access to safer hair products and increased access to more harmful hair products in Boston, Massachusetts. We also reported that redlining may decrease access to high-hazard hair products. These redlining findings, which contradicted our hypothesis, were supported by a study of a different health-promoting factor in Baltimore, Maryland, from Sadler et al³⁷ that reported increased access to healthy food among areas that were redlined, which was hypothesized to be driven by gentrification. Separately, in their analysis, gentrification was reported to increase access to healthy foods. In comparison, Shaker et al⁴⁶ examined access to grocery stores across the United States and reported that areas that were redlined and in the process of gentrifying had reduced access to grocery stores. Of note, this analysis used the Gini index, which measures the distribution of income inequality, as a proxy for gentrification. While gentrification may provide resources to historically disadvantaged communities, displacement is a major concern in the context of who has access to these health-promoting resources, including safer hair products.

This study has several limitations. While we used NAICS codes for the most common store types where diverse communities shop for personal care products based on market research, and we visited a significant share of the stores selling hair products (n = 117), store types other than the four visited may sell hair products, including salons.^49–51 Next, EWG’s Skin Deep database is one of a growing number of safer consumer product databases that aim to provide information or suggestions on products, each of which may calculate hazard scores differently. However, EWG considers 17 hazard categories that are important in understanding ingredient risk. Our analyses also did not take into consideration contemporary factors impacting store locations, including population density, land use, etc. Additionally, this analysis examined access to safer products in Boston, Massachusetts, and did not consider where individuals shop for products or specific product use patterns. Furthermore, our analyses are limited by the large amount of missing data, with the missing EWG scores being more prevalent among communities of color. Communities of color, notably Black communities, are found to more commonly use leave-in hair products containing EDCs or are reported to be hormonally active.^1,2 These differences in available EWG scores may speak towards differences in access to publicly available information about personal care product chemicals of concern among different communities.^7,8 If products with missing EWG scores are more likely to contain chemicals of concern, the missing data may bias our results towards the null. However, we did include a category for missingness in our analysis to begin to explore these patterns, and future research will further evaluate the missingness presented in this dataset. Using Skin Deep, we were able to quantitatively assess and identify the hazard scores of almost 25,000 hair products—one of the largest studies of hair products to date. Furthermore, our analyses are limited by the small number of low-hazard hair products identified across Boston. Next, gentrification may occur at different time points and over different lengths of time; thus, our selection of ACS 5-year estimates as a proxy for 2010 and 2020, as done in prior research,⁴¹ reflect recent neighborhood changes and would not be able to identify gentrified areas in Boston before 2010 or areas currently undergoing the process of gentrification. However, we identified 74 stores located in census tracts that were eligible for gentrification (residential areas that historically experienced a period of disinvestment). Our redlining analyses are limited by the lack of heterogeneity in HOLC grades in the city of Boston, with the majority of stores located in HOLC C areas, while there were few to no stores selected in HOLC A and B areas. Furthermore, our analysis on the role of both redlining and gentrification is limited by our use of HOLC C as the redlining reference due to the lack of HOLC B stores when considering both resource allocation practices. Despite this limitation, these findings warrant future research in other historically redlined cities that may have greater variability with respect to stores selling hair products across HOLC grades. Also, gentrification and historic redlining are only two of many discriminatory practices that have targeted or impacted low-income communities of color (i.e., restricted covenants, the urban renewal section of the Federal Housing Act of 1949, the Federal Aid Highway Act of 1956, and blockbusting).^37,68 Lastly, personal care products are only one distinct source of exposure to environmental chemicals.^12,23 Research in exposomics is emerging to understand the totality of environmental exposures and social stressors experienced by communities.⁶⁹ While this is beyond the scope of this analysis, future work will aim to consider personal care product-associated exposures in combination with other environmental and social stressors.

This analysis also has several strengths. Through this study, we collected data on almost 40,000 hair products in the city of Boston, which, to our knowledge, is one of the largest datasets of city-wide product availability that includes safety. Personal care product use has also typically been considered an individual behavior—this narrative fails to consider how factors at the neighborhood or societal level may drive product use.^23,28 This analysis explored factors contributing to neighborhood-level product accessibility and, in part, product purchasing since consumers still purchase the majority of their personal care products in person.⁷⁰ It is also one of the first to our knowledge to consider both redlining and gentrification in the context of hair product accessibility. We also used EWG’s Skin Deep Database, which is a publicly available database where consumers can look up their products and shop for safer alternatives, and were able to identify the hazard score of close to 25,000 products in the city of Boston. Lastly, we considered different historic and contemporary resource allocation practices to broadly explore what may be contributing to differences in hair product safety in Boston, Massachusetts.

This analysis reported on our new dataset that considers access to safer hair products across neighborhoods in Boston, Massachusetts, and includes characteristics such as product type, product price, and store type. We observed differences in access to safer hair products between one pharmacy chain located in 87% of neighborhoods in Boston, which may point to retailer redlining and the targeted marketing and selling of certain products in communities. We also reported differences in access to safer hair products when considering historical redlining and gentrification—with stores located in areas previously graded D and not experiencing gentrification having increased access to high-hazard products and decreased access to low-hazard products. Future research is needed to consider other historic or contemporary factors that may impact access to safer products and to explore how business or supply chain decisions may drive differences in product availability. Finally, the use of unsafe hair products may contribute to reported inequities in exposure to EDCs and associated health outcomes reported in the United States,^4,11,12,71 thus, updated federal policies and regulations are needed to ensure there is equitable access to safe hair products for everyone.

Conflicts of interest statement

The authors declare that they have no conflicts of interest with regard to the content of this report.

ACKNOWLEDGEMENTS

We would like to thank Dr. Leopoldo Cabassa for his guidance in interpreting these findings. We would also like to thank Lily Claus, Shivani Parkih, Brian Stevens, and Noah Cott for their assistance in data collection and Lisa Frueh, Rachel Ford, Jacky Duong, Jasming Jones, Christine Alcindor, Jordyn Kerr, Kate Hardock, Azariah Boyd, Akansha Das, Brooke Baptista, Humza Irfan, Monica Malaka-Kajangu, and Shana Grant for their assistance in data entry.