Metal Concentrations in E-Cigarette Aerosol Samples: A Comparison by Device Type and Flavor

Abstract

Background:

The rapid evolution of electronic cigarette (e-cigarette) products warrants surveillance of the differences in exposure across device types—modifiable devices (MODs), cartridge (“pod”)-containing devices (PODs), disposable PODs (d-PODs)—and flavors of the products available on the market.

Objective:

This study aimed to measure and compare metal aerosol concentrations by device type and common flavors.

Methods:

We collected aerosol from 104 MODs, 67 PODs (four brands: JUUL, Bo, Suorin, PHIX), and 23 d-PODs (three brands: ZPOD, Bidi, Stig) via droplet deposition in a series of conical pipette tips. Metals and metalloids [aluminum (Al), arsenic (As), cobalt (Co), chromium (Cr), copper (Cu), iron (Fe), manganese (Mn), nickel (Ni), lead (Pb), antimony (Sb), tin (Sn), and zinc (Zn)] were measured using inductively coupled plasma mass spectrometry (ICP-MS), results were log-transformed for statistical analysis, and concentrations are reported in aerosol units (${\mathrm{mg}/\mathrm{m}}^3$).

Results:

Of the 12 elements analyzed, concentrations were statistically significantly higher in MOD devices, except for Co and Ni, which were higher in PODs and d-PODs. Of the POD brands analyzed, PHIX had the highest median concentrations among four metals (Al, Ni, Pb, and Sn) compared to the rest of the POD brands. According to POD flavor, seven metals were three to seven orders of magnitude higher in tobacco-flavored aerosol compared to those in mint and mango flavors. Among the d-POD brands, concentrations of four metals (Al, Cu, Ni, and Pb) were higher in the ZPOD brand than in Bidi Stick and Stig devices. According to d-POD flavor, only Cr concentrations were found to be statistically significantly higher in mint than tobacco-flavored d-PODs.

Discussion:

We observed wide variability in aerosol metal concentrations within and between the different e-cigarette device types, brands, and flavors. Overall, MOD devices generated aerosols with higher metal concentrations than PODs and d-PODs, and tobacco-flavored aerosols contained the highest metal concentrations. Continued research is needed to evaluate additional factors (i.e., nicotine type) that contribute to metal exposure from new and emerging e-cigarette devices in order to inform policy. https://doi.org/10.1289/EHP11921

Introduction

Electronic cigarette (e-cigarette) use has continued to increase, despite growing evidence of their toxicity and adverse health effects.^1–4 As of 2021, 11.3% of U.S. high school students (1.72 million) and 2.8% of middle school students (320,000) reported currently using e-cigarettes.⁵ E-cigarettes are primarily composed of a battery, a cartridge or tank, a liquid solution [e-liquid, composed of propylene glycol (PG) and vegetable glycerin (VG) with varying concentrations of nicotine and flavorings], and a heating element, typically a metallic coil, to generate an aerosol that is inhaled by the user. Within the past decade, these devices have rapidly evolved. First-generation devices ($\sim 2005–2010$) are known as cig-a-likes, as they resemble conventional tobacco cigarettes; they contain disposable prefilled cartridges (cartomizers) and low-powered rechargeable batteries with no ability to change the power.⁶ Second- and third-generation devices (since $\sim 2010$) have been classified as “open” devices and include e-pen models and tank-style or modifiable devices (MODs), which are refillable, reusable systems and common among former smokers.⁷ Compared to cig-a-likes, MOD devices are typically larger and allow users to modify their devices by replenishing the e-liquid (varying flavors and tobacco concentrations), changing heating coils, and adjusting power output. Fourth-generation devices (since $\sim 2015$), referred to as PODs (i.e., JUUL), were initially shaped like USB thumb drives and, similar to cig-a-likes, had low-capacity rechargeable batteries and single-use cartridges (“pods”) filled with nicotine-containing e-liquid.^8,9 These PODs quickly evolved to offer options not resembling USB drives (i.e., Suorin, Vuse) that also offered refillable pods. What differentiates PODs from cig-a-likes is the e-liquid formulation; instead of using freebase nicotine, which is commonly found in e-liquids of cig-a-likes (and earlier MODs), PODs use nicotine salts, a protonated form of nicotine with lower pH, which allows for a smoother “throat hit” that can deliver higher nicotine concentration.^10,11 Although the Food and Drug Administration (FDA) implemented a partial ban on JUUL flavors early in 2020 to curb use among youth, newer disposable pods (d-PODs) (i.e., Puff Bar, Bidi Stick) were introduced in 2019, eclipsing JUUL in sales as of April 2020¹²; d-PODs have since gained popularity due to their attractive colors, low price, and fruity flavors.¹³ While the perception of safety and variety of appealing flavors contribute to their popularity,^14–16 e-cigarettes are not toxicant-free. Metals and metalloids linked to cardiovascular and kidney disease,^17–19 neurotoxicity,²⁰ and lung cancer^21,22 have been detected in the e-liquids and aerosols of cig-a-like, MODs, and PODs.^23–28 Studies have shown sources of metal exposure may be from the heating coil used to aerosolize the e-liquid,^14,24 the e-liquid itself,²⁴ as well as soldered joints and other parts of the device.^29,30 Across all device generations, heating coils are commonly made up of metal alloys, such as Kanthal (chromium, aluminum, iron), Nichrome (nickel and chromium), and stainless steel (iron, nickel, chromium),^14,24,31 while solder joints are made up of tin and lead.^29,30

Evidence of toxic metal exposure from e-cigarettes is growing, with a majority of studies to date focused on cig-a-likes^{25,27,29,30,32–35} and MODs.^{24,26,36–38} Studies investigating the constituents of POD devices^39–42 have focused on e-liquid⁴³ and aerosol samples.^26,44–47 To our knowledge, only one other study has looked at exposure from the more recent and popularly used d-PODs that were released into the market in 2019.⁴⁸ With the rapid evolution of e-cigarette products in recent years, differences in exposure across all three popular device generations—MODs, PODs, d-PODs—and e-liquid flavors have yet to be characterized. The aims of this study were two-fold: first, to compare metal concentrations in the aerosol of several popular brands and flavors of PODs and d-PODs and, second, to compare the metal aerosol concentrations between device types (generations) in order to inform regulatory agencies and the public of the differential risk of exposure to toxic metals.

Methods

A convenience sample of commercially available e-cigarette device types (194 devices: 104 MODs, 67 PODs, 23 d-PODs), brands (including six POD and seven d-POD brands), and flavors (including tobacco, mint, and mango) were analyzed for this project. All d-POD ($n=23$) and some POD devices ($n=44$) were purchased online, while all MODs ($n=104$) and some POD devices ($n=23$) were brought to the lab by e-cigarette users participating in the Exposure to Metals from E-Cigarettes (EMIT) study (R01ES030025, PI: Dr. Ana Rule). The study recruited participants from April 2015 until March 2020 through vaping conventions and flyers posted in e-cigarette shops, newspapers, college campuses, and social media platforms. Participants were instructed to bring their regular e-cigarette device and refilling dispenser of e-liquid or a replacement POD (as applicable) on the day of the interview; information relevant to this paper included type of device, brand name, and e-liquid flavor; nicotine strength, puff frequency and duration, and other demographic and behaviors not relevant to this paper were also obtained. The study was approved by the institutional review board of the Johns Hopkins Bloomberg School of Public Health. All participants provided informed consent.

Aerosol Collection

For all e-cigarettes, we collected $0.25–0.50\text{\hspace{0.17em}ml}$ of aerosol generated by the e-cigarette devices by connecting the mouthpiece via flexible tubing to a peristaltic pump inside a fume hood. The aerosol collection is based on the methodology described in Olmedo et al.²⁴ and Hilpert et al.,⁴⁹ which has been used in several other studies.^26,31 To optimally generate and collect e-cigarette aerosol based on the device’s power settings, the pump’s flow rate was $0.90\mathrm{L}/\text{min}$ for MODs and $0.70\mathrm{L}/\text{min}$ for PODs (including d-PODs); flow rates were measured using a Primary Standard Calibrator (Bios Defender 520; Mesa Laboratories, Inc.). The pump operated at 4 s/puff and 30 s interpuff time.⁵⁰ Briefly, on the downstream side of the pump, the generated aerosol was collected in a 1.5-mL centrifuge tube via impaction into a series of conical pipette tips and Tygon tubing L/S (inner radius, 4.8 mm).⁴⁹ MODs were activated between 13 and 65 puffs (average of 28) to collect the required aerosol volume, while 35 to 500 puffs (average of 178) were needed for the PODs. The number of puffs required to collect from each device was recorded. The large puff range is due to the varying device, power settings, and aerosol generation characteristics of each device, and corresponds with whatever settings our participants were using in case of MODs, and with the only setting available in case of PODs and d-PODs.

To improve the recovery of condensed aerosol for metal analysis, the tubing was manually flicked to dislodge the remaining droplets and connected back to the peristaltic pump. The pump was then activated for 30 s with room air, with the e-cigarette device disconnected to push the remaining droplets. Aerosol recovery using this method has been previously calculated as 72–83% for PODs and 78–83% for MODs.²⁶ All samples were stored at room temperature before analysis of metal concentrations.

To account for potential contamination from room air and/or the collection device, five samples of “blank e-liquid,” consisting of a solution of 70% propylene glycol (high purity grade; Amresco) and 30% glycerol (ultrapure; ICN Biochemicals) were poured into a 1-mL pipette tip and passed through the plastic tubing and conical pipette tips, using the peristaltic pump, drawing laboratory room air. The median concentration from these five blanks was used to correct aerosol sample results. More details on quality control and assurance are reported in Olmedo et al.²⁴

E-Cigarette Devices Used for the Study

MODs.

From 2015 until 2017, we collected aerosols from 104 MODs of daily e-cigarette users who were recruited as part of the EMIT study.^31,51

PODs.

From 2019 until 2021, several brands of rechargeable (nondisposable) POD devices were selected based on popularity in the United States.^52–54 Aerosol samples ($n=67$) were collected from devices purchased online ($n=44$) or from devices brought to the lab by e-cigarette users ($n=23$) participating in the EMIT Study. Brands included JUUL (JUUL Labs) ($n=45$), BO (BO Vaping) ($n=6$), PHIX (PHIX Vapor) ($n=6$), Suorin ($n=6$), and other brands brought by participants ($n=4$) [Nautilus (Shenzen Eigate Technology), Leap (E-Alternative Solutions), Vuse (R.J. Reynolds Vapor Company), and one unknown]. To facilitate comparability between brands, an available tobacco flavor ($n=28$) was selected for each brand (Virginia tobacco for JUUL, Cut tobacco for Bo, Original tobacco for PHIX, and tobacco Mi-Salt E-liquids for Suorin). To compare metal generation by flavor, mint ($n=20$) and mango JUUL flavors ($n=14$) as well as “other” flavors ($n=5$) from other devices (i.e., Nautilus, Leap) such as strawberry and artic berry were included; these flavor groups were included based on popularity⁵⁵ as well as what participants brought to the study. Nicotine concentrations across brands ranged from 35 to $59\mathrm{mg}/\text{ml}$ as indicated by the cartridge packaging or online website.

d-PODs.

Three brands of d-PODs were selected based on popularity and availability in the United States in March 2020 and purchased online. Although Puff Bars were the most popular d-PODs at the time,¹³ we were unable to purchase these devices, as they were unavailable online and at local vape shops due to FDA restrictions. The brands selected were Z-POD (Plus Distribution, Inc.), Bidi Stick (Bidi Vapor, LLC), and Stig (Stig, Inc.). To facilitate comparability with other device types and between brands, tobacco flavor ($n=12$) was selected (Tobacco for Z-Pod, Classic tobacco for Bidi Stick, and Cubano for Stig). To compare metal generation by flavor within a brand, mint flavors ($n=11$) were also selected for Stig (Mighty Mint) and Bidi stick (Mint Freeze). Although a total of 29 d-PODs were purchased, there were six d-PODS that did not generate aerosol and were not included in the analysis. Twenty-three d-PODs were successfully sampled: five of Z-Pods (tobacco), nine of Bidi Stick (three Classic tobacco and six mint), and nine of Stig (four Cubano and five Mighty Mint). Nicotine concentrations across brands ranged from 50 to $60\mathrm{mg}/\text{ml}$ as indicated by the cartridge packaging or online website.

Metal Analyses

All aerosol samples were sent to the Institute for Chemistry, University of Graz (Graz, Austria) for metal analysis. Twelve elements were analyzed because of their human health toxicity and likelihood of presence based on coil composition and on previously conducted studies^23,24,26: aluminum (Al), arsenic (As), cobalt (Co), chromium (Cr), copper (Cu), iron (Fe), manganese (Mn), nickel (Ni), lead (Pb), antimony (Sb), tin (Sn), and zinc (Zn). Methods for metal analysis have previously been reported in detail.²⁴ In brief, multielement analysis in all samples and calibration standards were performed on an Agilent 8800 or 8900 triple quadrupole inductively coupled plasma mass spectrometer (ICPQQQMS) (Agilent Technologies). To ensure the accuracy of the results, an internal standard and a reference standard were used. The multielement internal standard consisted of a solution containing $200\mathrm{mg}/\mathrm{L}$ of beryllium (Be), germanium (Ge), indium (In), and lutetium (Lu) and was added online to the samples prior to the nebulizer of the ICP-MS via a T-piece to compensate for instrumental instabilities and possible matrix effects. In addition, a reference standard [Reference Material NIST SRM 1640a—Trace Elements in Natural Water; National Institute of Standards and Technology (NIST)] and two blanks were analyzed at the beginning of each sequence and after every X sample, where X is the square root of the total number of samples in the sequence. All elements of the reference standard were found within 5% of the NIST-certified concentrations. Altogether, the standard was analyzed 12 times, with a mean recovery of $98\%\pm 2\%$ standard deviation (SD), suggesting a very stable measurement. Concentrations were reported in a weight/weight basis ($\mathrm{\mu }\mathrm{g}/\mathrm{kg}$) due to the difficulty to measure the volumes of thick and sticky samples. Limits of detection (LOD) were calculated by dividing the SD of the response of the ICP-MS by the slope of the calibration curve and multiplying the quotient by a factor of 3. LODs in solution in $\mathrm{\mu }\mathrm{g}/\mathrm{kg}$ varied across three analytical batches. The average LODs were $2.00\times {10}^{-1}$ for Al, $3.87\times {10}^{-3}$ for As, $2.80\times {10}^{-4}$ for Co, $8.48\times {10}^{-3}$ for Cr, $2.80\times {10}^{-4}$ for Cu, $3.33\times {10}^{-2}$ for Fe, $7.38\times {10}^{-3}$ for Mn, $6.83\times {10}^{-3}$ for Ni, $2.22\times {10}^{-3}$ for Pb, $5.25\times {10}^{-3}$ for Sn, $1.92\times {10}^{-3}$ for Sb, and $1.60\times {10}^{-1}$ for Zn. Concentrations below the LOD were replaced with the LOD divided by the square root of 2 for analysis.⁵⁶ Metals with $>50\%$ of samples below LOD were excluded from statistical analyses.

Conversion to Air Concentrations and Comparison to Exposure Limits

Metal concentrations reported by the lab in a weight/weight basis ($\mathrm{\mu }\mathrm{g}/\mathrm{kg}$) were converted into aerosol concentrations (${\mathrm{mg}/\mathrm{m}}^3$) as described in Olmedo et al.²⁴ Between 13 and 500 puffs were collected for analysis; based on past recruitment, e-cigarette users reported the number of puffs per day ranged from 15 to 1,000 puffs,⁵¹ so collected condensed aerosol was assumed to be less than or equivalent to daily consumption and representative of daily exposure. We assumed that metal concentrations emitted per puff is not homogenous and can increase or decrease, depending on several factors such as the type of device, condition of the coil, battery life, and device settings; as the conversion from $\mathrm{\mu }\mathrm{g}/\mathrm{kg}$ to ${\mathrm{mg}/\mathrm{m}}^3$ can be highly dependable on the aerosol production of each device/setting, results vary depending on the device. However, since the overall conclusions hold independent of the units used (Table S1), we chose to report in ${\mathrm{mg}/\mathrm{m}}^3$ to be able to compare to health-based limits. In brief, the conversion factored in the total weight of the sample collected (mg) and the volume of air required to obtain each sample, which was calculated by multiplying the puffing flow rate Q (L/min), the puffing duration (4 s/puff or 0.066 min/puff), and the number of puffs required to collect the desired volume of aerosol (Equations 1–3). Since aerosol production is not constant along sampling, this conversion is an estimate.

$$\text{Volume\hspace{0.17em}of\hspace{0.17em}air\hspace{0.17em}}\left({\mathrm{m}}^3\right)=\mathrm{Q\hspace{0.17em}}\left(\mathrm{L}/\text{min}\right)\times \text{puff\hspace{0.17em}time\hspace{0.17em}}\left(\text{min}\right)\times \left(1{\mathrm{m}}^3/\mathrm{1,000}\mathrm{L}\right),$$ (1)

$$\text{Puffing\hspace{0.17em}time\hspace{0.17em}}\left(\text{min}\right)=\#\text{puffs}\times \text{puff\hspace{0.17em}duration\hspace{0.17em}}\left(\text{min}/\text{puff}\right),$$ (2)

$$\text{Air\hspace{0.17em}concentration\hspace{0.17em}}\left({\mathrm{mg}/\mathrm{m}}^3\right)=\left[\text{metal\hspace{0.17em}concentration\hspace{0.17em}}\left(\mathrm{\mu }\mathrm{g}/\mathrm{kg}\right)\times \left(1\times {10}^{-3}\mathrm{mg}/1\mathrm{\mu }\mathrm{g}\right)\times \text{total\hspace{0.17em}mass\hspace{0.17em}of\hspace{0.17em}}\text{condensate\hspace{0.17em}}\left(\mathrm{kg}\right)\right]/\text{volume\hspace{0.17em}of\hspace{0.17em}air\hspace{0.17em}}\left({\mathrm{m}}^3\right).$$ (3)

This conversion allows for comparison to aerosol standards and health-based exposure limits of metal concentrations a person would typically inhale (minute ventilation: $6\mathrm{L}/\text{min}$) during a window of exposure. Air concentrations of Cr, Mn, and Ni were compared to the Agency for Toxic Substances and Disease Registry (ATSDR) chronic minimum risk levels (MRLs),⁵⁷ and Pb concentrations were compared to the U.S. Environmental Protection Agency (EPA) National Ambient Air Quality Standard (NAAQS).⁵⁸ Both limits are 24-h exposure averages. As there are no acute, intermediate, or chronic duration inhalation MRLs for As, sample concentrations were compared to the California EPA (CalEPA) chronic inhalation reference exposure level (REL) for total As.⁵⁹

In addition, aerosol metal concentration from each of our tested devices was compared to the U.S. FDA permissible daily exposure (PDE) limits for elemental impurities of As, Cr, Ni, and Pb in inhalation medications in units of $\mathrm{\mu }\mathrm{g}/\text{day}$.⁶⁰ Metal aerosol exposure per day ($\mathrm{\mu }\mathrm{g}/\text{day}$) was estimated by multiplying the aerosol concentrations ($\mathrm{\mu }\mathrm{g}/\mathrm{g}$) by the daily e-liquid dose in grams. MOD e-liquid consumption has been reported at a $32.5\text{-mL}$ median weekly⁵¹ or $5\mathrm{mL}/\text{day}$.⁶¹ Based on an e-liquid density of $1.14\pm 0.06$ ($\text{mean}\pm \text{SD}$) g/mL, previously measured in our laboratory, $5\mathrm{mL}$ of e-liquid is approximately equivalent to $5.7\mathrm{g}$ consumed per day. Typical consumption of four to ten cartridges per month ($\sim 0.1–0.3\mathrm{g}$ per day) for JUUL has been reported.⁶² The daily dose was therefore assumed to be $5.7\mathrm{g}/\text{day}$ for MODs and $0.2\mathrm{g}/\text{day}$ for PODs and d-PODs.

Statistical Analyses

Metal concentrations were right skewed and were log-transformed to perform the statistical analyses. Medians and interquartile ranges (IQR) were calculated to summarize metal concentrations. We graphically described metal concentrations by device type (MODs, PODs, and d-PODs), by POD brand and flavor, and by d-POD brand and flavor using box plots. To test whether certain POD and d-POD brands and certain flavors generate more metals in the aerosol, differences in log-transformed metal concentrations across different POD and d-POD brands and flavors were analyzed by one-way analysis of variance (ANOVA), and mean differences in log-transformed metal concentrations were compared by estimating geometric mean ratios (GMR) [95% confidence interval (CI)], where the mean difference (equivalent to the $\mathrm{\beta }\text{-coefficient}$) and corresponding 95% CI are both exponentiated; POD and d-POD aerosol concentrations were converted from $\mathrm{\mu }\mathrm{g}/\mathrm{kg}$ to ${\mathrm{mg}/\mathrm{m}}^3$ using Equations 1–3. For MODs, we did not have statistical power to compare metal levels by device brand and flavor, as more than 20 different brand/flavor combinations were reported by participants. We also estimated GMRs using log-transformed metal concentrations in aerosol samples of PODs and d-PODs compared to those of MODs. All analyses were performed using Stata 15.1 (StataCorp) and R software (version 3.6.2; R Project for Statistical Computing). The level of statistical significance ($\mathrm{\alpha }$) was set at 0.05, and all tests were two-sided.

Results

Median (IQR) aerosol concentrations found by device types are summarized using boxplots in Figure 1. A total of 194 aerosol samples were collected from MOD ($n=104$), POD ($n=67$), and d-POD devices ($n=23$). Aerosol metal concentrations spanned several orders of magnitude within and between e-cigarette device types (Figure 1; Table S1). For example, Sn levels ($\mathrm{\mu }\mathrm{g}/\mathrm{kg}$) ranged from $<\text{LOD}$ ($\text{LOD}=5.25\times {10}^{-3}$) to $2.62\times {10}^3$, and As levels ranged from $<\text{LOD}$ ($\text{LOD}=3.87\times {10}^{-3}$) to $1.10\times {10}^3$ for all devices. Cu levels ranged from $<\text{LOD}$ ($\text{LOD}=2.8\times {10}^{-4}$) to $2.81\times {10}^4$ for MODs and $<\text{LOD}$ ($\text{LOD}=2.8\times {10}^{-4}$) to $2.12\times {10}^5$ for PODs, while Sn levels ranged from $<\text{LOD}$ ($\text{LOD}=5.25\times {10}^{-3}$) to 10 for d-PODs. Among all devices, 19% of Ni ($n=37$), 32% of Pb ($n=62$), 34% of Co ($n=66$), 38% of Mn ($n=74$), 39% of Cu ($n=76$), 41% of Sn ($n=80$), 45% of Al ($n=87$), 49% of Fe ($n=95$), 50% of Cr ($n=97$), 57% of Sb ($n=111$), 73% of Zn ($n=142$), and 85% of As ($n=165$) were $<\text{LOD}$. Summary data is reported in Table S1.

Figure 1.

Metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in e-cigarette aerosol samples by type of device (MODs, PODs, and d-PODs) ($n=194$). Box plot of metal aerosol concentrations measured in and stratified by type of e-cigarette device (i.e., MODs, PODs, d-PODs). The horizontal lines within boxes indicate medians, boxes indicate interquartile ranges, whiskers indicate values within 1.5 times the interquartile range from boxes, and solid circles outside the boxes indicate outlier data values. Summary data can be found in Table S1. Note: Al, aluminum; As, arsenic; Co, cobalt; Cr, chromium; Cu, copper; d-PODs, disposable POD devices; e-cigarette, electronic cigarette; Fe, iron; Mn, manganese; MODs, modifiable devices; Ni, nickel; Pb, lead; PODs, cartridge (“pod”)-containing devices; Sb, antimony; Sn, tin; Zn, zinc.

Metal Concentrations by Device Type: MODs vs. PODs vs. d-PODs

Seven metal concentrations (Al, Cr, Cu, Fe, Mn, Pb, Sn) were statistically significantly higher for MOD devices than for both disposable and nondisposable PODs ($p$-value $<0.001$) (Table 1). Both PODs and d-PODs had statistically significantly higher Co concentrations than MODs ($p$-value $<0.001$); PODs had statistically significantly higher Ni concentrations than MODs ($p$-value $<0.001$) (Table 1).

Table 1.

Crude GM and GMR (95% CI) of metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in aerosol samples by device (MODs, PODs, d-PODs).

	$n$	Crude GM	GMR (95% CI)	$p$-Value
Aluminum
MODs	104	$4.07\times {10}^{-4}$	1.00 (Ref)	$<0.001$
PODs	67	$2.49\times {10}^{-5}$	$6.10\times {10}^{-2}$ ($2.70\times {10}^{-2}$, 0.14)	—
d-PODs	23	$1.07\times {10}^{-5}$	$2.60\times {10}^{-2}$ ($8.60\times {10}^{-3}$, $8.10\times {10}^{-2}$)	—
Cobalt
MODs	104	$1.85\times {10}^{-7}$	1.00 (Ref)	$<0.001$
PODs	67	$1.56\times {10}^{-4}$	844 (404, $1.76\times {10}^3$)	—
d-PODs	23	$1.07\times {10}^{-4}$	579 (279, $1.20\times {10}^3$)	—
Chromium
MODs	104	$5.49\times {10}^{-6}$	1.00 (Ref)	$<0.001$
PODs	67	$8.13\times {10}^{-7}$	0.15 ($6.40\times {10}^{-2}$, 0.34)	—
d-PODs	23	$8.38\times {10}^{-7}$	0.15 ($3.70\times {10}^{-2}$, 0.64)	—
Copper
MODs	104	$2.83\times {10}^{-5}$	1.00 (Ref)	$<0.001$
PODs	67	$5.59\times {10}^{-5}$	1.97 (0.50, 7.78)	—
d-PODs	23	$5.41\times {10}^{-6}$	0.19 ($4.20\times {10}^{-2}$, 0.86)	—
Iron
MODs	104	$1.48\times {10}^{-4}$	1.00 (Ref)	$<0.001$
PODs	67	$5.07\times {10}^{-6}$	$3.40\times {10}^{-2}$ ($1.10\times {10}^{-2}$, 0.10)	—
d-PODs	23	$8.97\times {10}^{-7}$	$6.00\times {10}^{-3}$ ($2.20\times {10}^{-3}$, $1.60\times {10}^{-2}$)	—
Manganese
MODs	104	$\begin{array}{c}1.52\times {10}^{-5}\end{array}$	1.00 (Ref)	$<0.001$
PODs	67	$1.34\times {10}^{-6}$	$8.80\times {10}^{-2}$ ($3.50\times {10}^{-2}$, 0.22)	—
d-PODs	23	$1.31\times {10}^{-6}$	$8.60\times {10}^{-2}$ ($2.20\times {10}^{-2}$, 0.34)	—
Nickel
MODs	104	$5.77\times {10}^{-5}$	1.00 (Ref)	$<0.001$
PODs	67	$6.85\times {10}^{-4}$	11.8 (4.16, 33.9)	—
d-PODs	23	$5.57\times {10}^{-5}$	0.97 (0.13, 7.06)	—
Lead
MODs	104	$3.69\times {10}^{-6}$	1.00 (Ref)	$<0.001$
PODs	67	$1.62\times {10}^{-6}$	0.44 (0.13, 1.53)	—
d-PODs	23	$1.54\times {10}^{-6}$	0.42 ($7.70\times {10}^{-2}$, 2.24)	—
Tin
MODs	104	$2.89\times {10}^{-5}$	1.00 (Ref)	$<0.001$
PODs	67	$8.61\times {10}^{-7}$	$3.00\times {10}^{-2}$ ($8.40\times {10}^{-3}$, 0.110)	—
d-PODs	23	$3.71\times {10}^{-7}$	$1.30\times {10}^{-2}$ ($2.30\times {10}^{-3}$, $6.90\times {10}^{-2}$)	—

Note: The GMRs of metal concentrations in e-cigarette aerosol of MODs compared with that of PODs and that of d-PODs were obtained by exponentiating the corresponding mean difference (95% CI) in log-transformed metal aerosol concentrations. The $p$-values were obtained from the linear regression model used to calculate the GMRs. —, no data; CI, confidence interval; d-PODs, disposable POD devices; e-cigarette, electronic cigarette; GM, geometric mean; GMR, geometric mean ratio; MODs, modifiable devices; PODs, cartridge (“pod”)-containing devices; Ref, reference.

Metal Concentrations by POD and d-POD Brands

By POD brands.

We found statistically significant differences in aerosol metal concentrations among brands of nondisposable POD devices (Figure 2; Table S2). For example, PHIX devices had the highest median Al, Ni, Pb, and Sn compared to JUUL, Bo, and Suorin. Linear regression of geometric means of seven metal concentrations (Al, Co, Cu, Mn, Ni, Pb, and Sn) were three to seven orders of magnitude higher in PHIX compared to JUUL devices (Table 2). Compared to JUUL, three metal concentrations (Al, Fe, and Sn) were three to five orders of magnitude higher in Suorin devices, and Cu and Mn were three to seven orders of magnitude higher in Bo devices.

Figure 2.

Metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in e-cigarette aerosol of PODs by brand ($n=67$). Box plot of metal aerosol concentrations measured in PODs and stratified by four POD brands: JUUL, Bo, Phix, Suorin, and other. The horizontal lines within boxes indicate medians, boxes indicate interquartile ranges, whiskers indicate values within 1.5 times the interquartile range from boxes, and solid circles outside the boxes indicate outlier data values. Summary data can be found in Table S2. Note: Al, aluminum; As, arsenic; Co, cobalt; Cr, chromium; Cu, copper; e-cigarette, electronic cigarette; Fe, iron; Mn, manganese; Ni, nickel; Pb, lead; PODs, cartridge (“pod”)-containing devices; Sb, antimony; Sn, tin; Zn, zinc.

Table 2.

Geometric mean ratios (95% CI) of metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in aerosol by nondisposable and disposable POD brand devices.

Nondisposable POD brands					Disposable POD brands
	$n$	Crude GM	GMR (95% CI)	$p$-value		$n$	Crude GM	GMR (95% CI)	$p$-Value
Aluminum
JUUL	45	$9.53\times {10}^{-6}$	1.00 (Ref)	$<0.001$	ZPOD	5	$1.21\times {10}^{-4}$	1.00 (Ref)	0.11
Bo	6	$1.05\times {10}^{-4}$	10.9 (0.93, 129)	—	BIDI	9	$3.84\times {10}^{-6}$	$3.20\times {10}^{-2}$ ($1.20\times {10}^{-3}$, 0.85)	—
Phix	6	$1.22\times {10}^{-3}$	127 (63.6, 129)	—	STIG	9	$7.76\times {10}^{-6}$	$6.40\times {10}^{-2}$ ($2.60\times {10}^{-3}$, 1.57)	—
Suorin	6	$8.42\times {10}^{-5}$	8.83 (1.04, 75.1)	—	—	—	—	—	—
Other	4	$6.58\times {10}^{-5}$	6.90 (0.79, 60.2)	—	—	—	—	—	—
Cobalt
JUUL	45	$1.60\times {10}^{-4}$	1.00 (Ref)	0.009	ZPOD	5	$7.56\times {10}^{-5}$	1.00 (Ref)	0.079
Bo	6	$1.32\times {10}^{-5}$	$8.10\times {10}^{-2}$ ($3.77\times {10}^{-3}$, 1.77)	—	BIDI	9	$6.38\times {10}^{-5}$	0.84 (0.10, 6.98)	—
Phix	6	$6.43\times {10}^{-4}$	3.96 (1.44, 10.9)	—	STIG	9	$2.20\times {10}^{-4}$	2.91 (0.45, 18.7)	—
Suorin	6	$3.70\times {10}^{-4}$	2.28 (0.96, 5.45)	—	—	—	—	—	—
Other	4	$1.43\times {10}^{-4}$	0.88 (0.24, 3.22)	—	—	—	—	—	—
Chromium
JUUL	45	$6.17\times {10}^{-7}$	1.00 (Ref)	0.26	ZPOD	5	$5.47\times {10}^{-7}$	1.00 (Ref)	0.044
Bo	6	$2.48\times {10}^{-6}$	4.01 (0.46, 35.4)	—	BIDI	9	$5.73\times {10}^{-6}$	10.4 (0.23, 467)	—
Phix	6	$8.34\times {10}^{-7}$	1.35 (0.51, 3.56)	—	STIG	9	$1.55\times {10}^{-7}$	0.28 ($1.70\times {10}^{-2}$, 4.74)	—
Suorin	6	$2.80\times {10}^{-6}$	4.54 (0.80, 25.7)	—	—	—	—	—	—
Other	4	$5.14\times {10}^{-7}$	0.83 (0.25, 2.78)	—	—	—	—	—	—
Copper
JUUL	45	$4.13\times {10}^{-6}$	1.00 (Ref)	$<0.001$	ZPOD	5	$1.02\times {10}^{-5}$	1.00 (Ref)	0.002
Bo	6	0.46	$6.49\times {10}^4$ ($2.13\times {10}^4$, $1.97\times {10}^5$)	—	BIDI	9	$4.73\times {10}^{-5}$	4.64 (0.16, 133)	—
Phix	6	$7.00\times {10}^{-2}$	$9.76\times {10}^3$ ($2.13\times {10}^3$, $4.47\times {10}^4$)	—	STIG	9	$4.35\times {10}^{-7}$	$4.30\times {10}^{-2}$ ($3.93\times {10}^{-3}$, 0.46)	—
Suorin	6	$5.20\times {10}^{-6}$	0.73 (0.26, 2.06)	—	—	—	—	—	—
Other	4	$6.72\times {10}^{-4}$	94.1 (0.83, $1.06\times {10}^4$)	—	—	—	—	—	—
Iron
JUUL	45	$2.03\times {10}^{-6}$	1.00 (Ref)	0.007	ZPOD	5	$1.24\times {10}^{-6}$	1.00 (Ref)	0.53
Bo	6	$3.04\times {10}^{-5}$	14.9 (0.75, $3.00\times {10}^2$)	—	BIDI	9	$5.76\times {10}^{-7}$	0.46 (0.10, 2.06)	—
Phix	6	$7.13\times {10}^{-6}$	3.51 (0.32, 38.8)	—	STIG	9	$1.16\times {10}^{-6}$	0.94 (0.26, 3.41)	—
Suorin	6	$2.78\times {10}^{-4}$	137 (7.29, $2.57\times {10}^3$)	—	—	—	—	—	—
Other	4	$1.52\times {10}^{-5}$	7.50 (0.39, 142)	—	—	—	—	—	—
Manganese
JUUL	45	$5.37\times {10}^{-7}$	1.00 (Ref)	$<0.001$	ZPOD	5	$2.51\times {10}^{-6}$	1.00 (Ref)	0.49
Bo	6	$1.42\times {10}^{-5}$	26.5 (2.81, 249)	—	BIDI	9	$2.05\times {10}^{-6}$	0.82 ($2.30\times {10}^{-2}$, 28.8)	—
Phix	6	$1.29\times {10}^{-5}$	23.9 (4.21, 136)	—	STIG	9	$5.78\times {10}^{-7}$	0.23 ($1.20\times {10}^{-2}$, 4.24)	—
Suorin	6	$9.44\times {10}^{-6}$	17.5 (2.31, 133)	—	—	—	—	—	—
Other	4	$2.19\times {10}^{-6}$	4.07 (0.19, 87.0)	—	—	—	—	—	—
Nickel
JUUL	45	$2.80\times {10}^{-4}$	1.00 (Ref)	$<0.001$	ZPOD	5	$1.39\times {10}^{-3}$	1.00 (Ref)	$<0.001$
Bo	6	$4.32\times {10}^{-3}$	15.4 (8.23, 28.9)	—	BIDI	9	$7.75\times {10}^{-4}$	0.55 ($5.00\times {10}^{-2}$, 6.11)	—
Phix	6	$5.10\times {10}^{-2}$	181 (72.4, 452)	—	STIG	9	$6.69\times {10}^{-7}$	$4.79\times {10}^{-4}$ ($2.95\times {10}^{-5}$, $7.78\times {10}^{-3}$)	—
Suorin	6	$8.74\times {10}^{-4}$	3.12 (0.74, 13.1)	—	—	—	—	—	—
Other	4	$1.12\times {10}^{-3}$	3.99 ($2.20\times {10}^{-2}$, 723)	—	—	—	—	—	—
Lead
JUUL	45	$3.71\times {10}^{-7}$	1.00 (Ref)	$<0.001$	ZPOD	5	$6.88\times {10}^{-5}$	1.00 (Ref)	$<0.001$
Bo	6	$5.81\times {10}^{-5}$	156 (27.8, $8.80\times {10}^2$)	—	BIDI	9	$3.68\times {10}^{-6}$	$5.30\times {10}^{-2}$ ($2.05\times {10}^{-3}$, 1.39)	—
Phix	6	$3.71\times {10}^{-3}$	$1.00\times {10}^4$ ($1.24\times {10}^3$, $8.05\times {10}^4$)	—	STIG	9	$7.76\times {10}^{-8}$	$1.12\times {10}^{-3}$ ($2.51\times {10}^{-4}$, $5.08\times {10}^{-3}$)	—
Suorin	6	$9.96\times {10}^{-8}$	0.27 ($9.60\times {10}^{-2}$, 0.75)	—	—	—	—	—	—
Other	4	$7.67\times {10}^{-5}$	206 (0.28, $1.54\times {10}^5$)	—	—	—	—	—	—
Tin
JUUL	45	$2.24\times {10}^{-7}$	1.00 (Ref)	$<0.001$	ZPOD	5	$5.40\times {10}^{-7}$	1.00 (Ref)	0.46
Bo	6	$5.36\times {10}^{-7}$	2.39 (0.38, 15.2)	—	BIDI	9	$7.48\times {10}^{-7}$	1.38 ($1.70\times {10}^{-2}$, 114)	—
Phix	6	$2.96\times {10}^{-3}$	$1.31\times {10}^4$ ($5.18\times {10}^3$, $3.35\times {10}^4$)	—	STIG	9	$1.50\times {10}^{-7}$	0.28 ($1.60\times {10}^{-2}$, 4.83)	—
Suorin	6	$8.07\times {10}^{-7}$	3.60 (1.87, 6.94)	—	—	—	—	—	—
Other	4	$3.62\times {10}^{-5}$	161 (0.91, $2.86\times {10}^4$)	—	—	—	—	—	—

Note: The GMRs of metal concentrations in e-cigarette aerosol of POD brands (i.e., JUUL compared with that of Bo, Phix, Suorin, and other brands) were obtained by exponentiating the corresponding mean difference (95% CI) in log-transformed metal aerosol concentrations. GMRs of metal concentrations in e-cigarette aerosol of d-POD brands (i.e., ZPOD compared with that of Bidi and Stig brands) were also calculated. The $p$-values were obtained from the linear regression model used to calculate the GMRs. —, no data; CI, confidence interval; d-PODs, disposable POD devices; e-cigarette, electronic cigarette; GM, geometric mean; GMR, geometric mean ratio; PODs, cartridge (“pod”)-containing devices; Ref, reference.

By d-POD brands.

We also found statistically significant differences in aerosol metal concentrations among the tested d-POD brand devices (Figure 3; Table S3). Linear regression of geometric means of four metals (Al, Cu, Ni, and Pb) was found to be between three and five orders of magnitude higher in ZPOD devices than Bidi Stick and Stig devices (Table 2). Conversely, Bidi Stick devices had the highest median Cr and Cu concentrations, while Stig devices had the highest median Co concentrations compared to ZPOD devices (Table S3).

Figure 3.

Metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in e-cigarette aerosol of d-PODs by brand ($n=23$). Box plot of metal aerosol concentrations measured in d-POD devices and stratified by three d-POD brands: Z-Pod, Bidi, and Stig. The horizontal lines within boxes indicate medians, boxes indicate interquartile ranges, whiskers indicate values within 1.5 times the interquartile range from boxes, and solid circles outside the boxes indicate outlier data values. Summary data can be found in Table S3. Note: Al, aluminum; As, arsenic; Co, cobalt; Cr, chromium; Cu, copper; d-PODs, disposable POD devices; e-cigarette, electronic cigarette; Fe, iron; Mn, manganese; Ni, nickel; Pb, lead; PODs, cartridge (“pod”)-containing devices; Sb, antimony; Sn, tin; Zn, zinc.

Metal Concentrations by POD and d-POD Flavors

By POD flavors.

Tobacco noticeably had the highest median metal aerosol concentrations compared to mint, mango, and other flavors (Figure 4; Table S4); linear regression of geometric means of seven metal concentrations (Al, Cr, Cu, Fe, Mn, Ni, and Pb) was found to be statistically significantly higher, by two to three orders of magnitude, in tobacco flavored aerosol compared to mint, mango, and other flavored aerosols (Table 3).

Figure 4.

Metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in e-cigarette aerosol of PODs by flavor ($n=67$). Box plot of metal aerosol concentrations measured in POD devices and stratified by flavor: tobacco, mint, mango, and other. The horizontal lines within boxes indicate medians, boxes indicate interquartile ranges, whiskers indicate values within 1.5 times the interquartile range from boxes, and solid circles outside the boxes indicate outlier data values. Summary data can be found in Table S4. Note: Al, aluminum; As, arsenic; Co, cobalt; Cr, chromium; Cu, copper; e-cigarette, electronic cigarette; Fe, iron; Mn, manganese; Ni, nickel; Pb, lead; PODs, cartridge (“pod”)-containing devices; Sb, antimony; Sn, tin; Zn, zinc.

Table 3.

Geometric mean ratios (95% CI) of metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in aerosol by POD and disposable POD flavors.

Nondisposable POD brands					Disposable POD brands
	$n$	Crude GM	GMR (95% CI)	$p$-Value		$n$	Crude GM	GMR (95% CI)	$p$-Value
Aluminum
Tobacco	28	$7.43\times {10}^{-5}$	1.00 (Ref)	0.013	Tobacco	12	$1.86\times {10}^{-5}$	1.00 (Ref)	0.24
Mint	20	$9.73\times {10}^{-6}$	0.13 ($3.80\times {10}^{-2}$, 0.45)	—	Mint	11	$5.88\times {10}^{-6}$	0.32 ($4.20\times {10}^{-2}$, 2.32)	—
Mango	14	$1.17\times {10}^{-5}$	0.16 ($4.50\times {10}^{-2}$, 0.54)	—	—	—	—	—	—
Other	5	$1.94\times {10}^{-5}$	0.26 ($2.50\times {10}^{-2}$, 2.67)	—	—	—	—	—	—
Cobalt
Tobacco	28	$1.67\times {10}^{-4}$	1.00 (Ref)	0.97	Tobacco	12	$8.52\times {10}^{-5}$	1.00 (Ref)	0.44
Mint	20	$1.40\times {10}^{-4}$	0.84 (0.24, 2.98)	—	Mint	11	$1.39\times {10}^{-4}$	1.63 (0.46, 5.80)	—
Mango	14	$1.55\times {10}^{-4}$	0.93 (0.11, 7.89)	—	—	—	—	—	—
Other	5	$1.79\times {10}^{-4}$	1.07 (0.36, 3.24)	—	—	—	—	—	—
Chromium
Tobacco	28	$1.35\times {10}^{-6}$	1.00 (Ref)	0.05	Tobacco	12	$2.00\times {10}^{-7}$	1.00 (Ref)	0.024
Mint	20	$3.03\times {10}^{-7}$	0.23 ($7.70\times {10}^{-2}$, 0.66)	—	Mint	11	$4.00\times {10}^{-6}$	20.0 (1.55, 259)	—
Mango	14	$1.03\times {10}^{-6}$	0.77 (0.12, 5.10)	—	—	—	—	—	—
Other	5	$1.28\times {10}^{-6}$	0.95 ($2.70\times {10}^{-2}$, 33.0)	—	—	—	—	—	—
Copper
Tobacco	28	$8.99\times {10}^{-4}$	1.00 (Ref)	$<0.001$	Tobacco	12	$1.55\times {10}^{-6}$	1.00 (Ref)	0.062
Mint	20	$3.52\times {10}^{-6}$	$3.92\times {10}^{-3}$ ($3.77\times {10}^{-4}$, $4.10\times {10}^{-2}$)	—	Mint	11	$2.11\times {10}^{-5}$	13.5 (0.86, 213)	—
Mango	14	$7.34\times {10}^{-6}$	$8.15\times {10}^{-3}$ ($5.08\times {10}^{-4}$, 0.13)	—	—	—	—	—	—
Other	5	$1.81\times {10}^{-4}$	0.20 ($9.91\times {10}^{-4}$, 40.7)	—	—	—	—	—	—
Iron
Tobacco	28	$1.07\times {10}^{-5}$	1.00 (Ref)	0.008	Tobacco	12	$9.11\times {10}^{-7}$	1.00 (Ref)	0.96
Mint	20	$1.33\times {10}^{-6}$	0.12 ($3.20\times {10}^{-2}$, 0.48)	—	Mint	11	$8.82\times {10}^{-7}$	0.97 (0.27, 3.42)	—
Mango	14	$4.25\times {10}^{-6}$	0.40 ($4.90\times {10}^{-2}$, 3.21)	—	—	—	—	—	—
Other	5	$2.68\times {10}^{-5}$	2.50 ($4.50\times {10}^{-2}$, 139)	—	—	—	—	—	—
Manganese
Tobacco	28	$3.20\times {10}^{-6}$	1.00 (Ref)	0.024	Tobacco	12	$1.01\times {10}^{-6}$	1.00 (Ref)	0.67
Mint	20	$4.22\times {10}^{-7}$	0.13 ($3.60\times {10}^{-2}$, 0.48)	—	Mint	11	$1.73\times {10}^{-6}$	1.71 (0.12, 23.8)	—
Mango	14	$1.01\times {10}^{-6}$	0.31 ($5.70\times {10}^{-2}$, 1.74)	—	—	—	—	—	—
Other	5	$2.45\times {10}^{-6}$	0.77 ($2.30\times {10}^{-2}$, 25.1)	—	—	—	—	—	—
Nickel
Tobacco	28	$1.64\times {10}^{-3}$	1.00 (Ref)	0.024	Tobacco	12	$8.03\times {10}^{-5}$	1.00 (Ref)	0.69
Mint	20	$2.07\times {10}^{-4}$	0.13 ($3.20\times {10}^{-2}$, 0.50)	—	Mint	11	$3.74\times {10}^{-5}$	0.47 ($8.60\times {10}^{-3}$, 25.0)	—
Mango	14	$5.89\times {10}^{-4}$	0.36 ($7.70\times {10}^{-2}$, 1.68)	—	—	—	—	—	—
Other	5	$9.38\times {10}^{-4}$	0.57 ($2.10\times {10}^{-2}$, 15.2)	—	—	—	—	—	—
Lead
Tobacco	28	$9.00\times {10}^{-6}$	1.00 (Ref)	0.024	Tobacco	12	$1.37\times {10}^{-6}$	1.00 (Ref)	0.89
Mint	20	$2.66\times {10}^{-7}$	$3.00\times {10}^{-2}$ ($3.04\times {10}^{-3}$, 0.29)	—	Mint	11	$1.74\times {10}^{-6}$	1.27 ($4.10\times {10}^{-2}$, 39.7)	—
Mango	14	$5.39\times {10}^{-7}$	$6.00\times {10}^{-2}$ ($3.42\times {10}^{-3}$, 1.05)	—	—	—	—	—	—
Other	5	$3.41\times {10}^{-6}$	0.38 ($1.58\times {10}^{-3}$, 90.6)	—	—	—	—	—	—
Tin
Tobacco	28	$1.92\times {10}^{-6}$	1.00 (Ref)	0.40	Tobacco	12	$1.95\times {10}^{-7}$	1.00 (Ref)	0.37
Mint	20	$3.72\times {10}^{-7}$	0.19 ($2.70\times {10}^{-2}$, 1.39)	—	Mint	11	$7.52\times {10}^{-7}$	3.87 (0.18, 82.9)	—
Mango	14	$4.95\times {10}^{-7}$	0.26 ($2.90\times {10}^{-2}$, 2.30)	—	—	—	—	—	—
Other	5	$1.31\times {10}^{-6}$	0.69 ($6.37\times {10}^{-3}$, 73.6)	—	—	—	—	—	—

Note: The GMRs of metal concentrations in e-cigarette aerosol of POD flavors (i.e., tobacco compared with that of mint, mango, and other) were obtained by exponentiating the corresponding mean difference (95% CI) in log-transformed metal aerosol concentrations. GMRs of metal concentrations in e-cigarette aerosols of d-POD flavors (i.e., tobacco compared with that of mint) were also calculated. The $p$-values were obtained from the linear regression model used to calculate the GMRs. —, no data; CI, confidence interval; d-PODs, disposable POD devices; e-cigarette, electronic cigarette; GM, geometric mean; GMR, geometric mean ratio; PODs, cartridge (“pod”)-containing devices; Ref, reference.

By d-POD flavors.

Cr concentrations were statistically significantly higher in mint-flavored d-PODs than tobacco flavored d-PODs (Table 3 and Figure 5; Table S4).

Figure 5.

Metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in e-cigarette aerosol of d-PODs by flavor ($n=23$). Box plot of metal aerosol concentrations measured in d-POD devices and stratified by two flavors: tobacco and mint. The horizontal lines within boxes indicate medians, boxes indicate interquartile ranges, whiskers indicate values within 1.5 times the interquartile range from boxes, and solid circles outside the boxes indicate outlier data values. Summary data can be found in Table S4. Note: Al, aluminum; As, arsenic; Co, cobalt; Cr, chromium; Cu, copper; d-PODs, disposable POD devices; e-cigarette, electronic cigarette; Fe, iron; Mn, manganese; Ni, nickel; Pb, lead; PODs, cartridge (“pod”)-containing devices; Sb, antimony; Sn, tin; Zn, zinc.

Comparison to Exposure Limits

Concentrations for each of the detected metals spanned several orders of magnitude (Table 4; Excel Table S1). Due to their toxicity when found in aerosols, Ni, Cr, Pb, Mn, and As have inhalation health-based limit concentrations. Approximately half [$n=100$ (52%)] of e-cigarette aerosol samples exceeded the ATSDR⁵⁷ daily chronic MRL for Ni of $2.00\times {10}^{-4}{\mathrm{mg}/\mathrm{m}}^3$^63,64; 45 (43%) of the samples that exceeded the Ni MRL were from MOD samples and another 45 (67%) were from POD samples. We did not determine the valence state of Cr or As in our samples and do not know what proportion was Cr(VI) (hexavalent) vs. trivalent. If all Cr in our samples were Cr(VI), 45 (23%) of the samples would exceed the daily MRL for Cr(VI) in mist ($5.00\times {10}^{-6}{\mathrm{mg}/\mathrm{m}}^3$). Conversely, if Cr in our samples was Cr(III), 32 (7%) of the samples would exceed daily MRL for soluble Cr(III) ($1.00\times {10}^{-4}{\mathrm{mg}/\mathrm{m}}^3$).⁶⁵ Thirty-five aerosol samples (18%) exceeded the U.S. EPA NAAQS⁵⁸ for Pb of $1.50\times {10}^{-4}{\mathrm{mg}/\mathrm{m}}^3$, and 14 samples (7%) exceeded the standard in nonattainment areas of $1.50\times {10}^{-3}{\mathrm{mg}/\mathrm{m}}^3$. Twenty-eight of the samples (14%) exceeded the daily Mn MRL of $3.00\times {10}^{-4}{\mathrm{mg}/\mathrm{m}}^3$,⁶⁶ and 39% exceeded the U.S. EPA daily cancer reference concentration (RfC) of $6.00\times {10}^{-6}{\mathrm{mg}/\mathrm{m}}^3$.⁶⁷ Nineteen (10%) of samples (all of those above the detection limit) exceeded CalEPA’s total As REL of $1.50\times {10}^{-5}{\mathrm{mg}/\mathrm{m}}^3$.⁵⁹ None of the individual samples exceeded the FDA PDE limits ($\mathrm{Cr}=2.9\mathrm{\mu }\mathrm{g}/\text{day}$; $\mathrm{Co}=2.9\mathrm{\mu }\mathrm{g}/\text{day}$; $\mathrm{Cu}=34\mathrm{\mu }\mathrm{g}/\text{day}$; $\mathrm{Pb}=5\mathrm{\mu }\mathrm{g}/\text{day}$; $\mathrm{Ni}=6\mathrm{\mu }\mathrm{g}/\text{day}$; $\mathrm{Sn}=64\mathrm{\mu }\mathrm{g}/\text{day}$) of any of these selected metals (Excel Table S2).

Table 4.

Median (range) of daily metal concentrations (${\mathrm{mg}/\mathrm{m}}^3$) in collected aerosol samples in comparison to regulatory and health-based limits for Ni, Cr, Pb, Mn, and As.

Value	Ni	Cr	Pb	Mn	As
Median	$2.82\times {10}^{-4}$	$9.05\times {10}^{-7}$	$5.97\times {10}^{-7}$	$1.13\times {10}^{-6}$	$5.24\times {10}^{-7}$
Range	$8.55\times {10}^{-9}$ to 0.450	$8.55\times {10}^{-9}$ to $9.90\times {10}^{-3}$	$4.27\times {10}^{-9}$ to 0.110	$1.03\times {10}^{-8}$ to $5.33\times {10}^{-2}$	$4.27\times {10}^{-9}$ to $3.04\times {10}^{-2}$

		Cr (VI)	Cr (III)	3-Month average	NAA	MRL	RfC
Regulatory health-based limitsa	$2.00\times {10}^{-4}$ b	$5.00\times {10}^{-6}$ c	$1.00\times {10}^{-4}$ d	$1.50\times {10}^{-4}$ e	$1.50\times {10}^{-3}$ f	$3.00\times {10}^{-4}$ g	$6.00\times {10}^{-6}$ h	$1.50\times {10}^{-5}$ i
Number of samples (%) exceeding the limit	—	—	—	—	—	—	—	—
All devices	100 (52)	45 (23)	32 (7)	35 (18)	14 (7)	28 (14)	75 (39)	19 (10)
MODs	45 (43)	31 (34)	25 (10)	18 (17)	7 (7)	26 (25)	50 (48)	16 (15)
PODs	45 (67)	8 (12)	5 (3)	12 (18)	7 (10)	2 (2)	17 (25)	3 (4)
d-PODs	10 (43)	6 (26)	2 (9)	5 (22)	0	0	8 (35)	0

Note: None of the samples exceeded the FDA PDE limits (Excel Table S2). —, no data; As, total arsenic; ATSDR, Agency for Toxic Substances and Disease Registry; Cr, chromium; d-PODs, disposable POD devices; FDA, U.S. Food and Drug Administration; Mn, manganese; MODs, modifiable devices; MRL, minimum risk level; NAAQS, National Ambient Air Quality Standard; Ni, nickel; Pb, lead; PDA, permissible daily limit; PODs, cartridge (“pod”)-containing devices; RfC, cancer reference concentration.

^a U.S. EPA NAAQS are regulatory, and all other limits are health-based.

^b ATSDR MRL for Ni.^63,64

^c MRL for Cr (VI) in mists.⁶⁵ MRLs are daily averages.

^d MRL for soluble Cr (III).⁶⁵ MRLs are daily averages.

^e U.S. EPA NAAQS (rolling 3-month average).⁵⁸

^f U.S. EPA NAAQS for nonattainment areas.⁵⁸

^g MRL for MN. MRLs are daily averages.⁶⁶

^h U.S. EPA RfC, daily values.⁶⁷

ⁱ California EPA REL for total As.⁵⁹

Discussion

In the assessment of aerosol samples collected from MODs, PODs, and d-PODs, we found that aerosol metal concentrations spanned several orders of magnitude within and among device types. Of the 12 metals analyzed, all concentrations were statistically significantly higher for MOD devices than for both PODs and d-PODs, except for Co and Ni. These differences in metal concentrations could be related to coil composition or power.^26,68 For instance, coils in MODs may either be stainless steel (Cr-Ni-Fe-Mn alloy), Kanthal (Fe-Cr-Al alloy), or Nichrome (Ni-Cr alloy),^14,25 and we have previously found higher aerosol Cr, Fe, Mn, Ni, Pb, and Sn concentrations in MOD devices using Kanthal coils compared to others coils²⁴; conversely, coils in PODs such as JUUL, are predominantly made of Nichrome, an alloy consisting of 80% Ni and 20% Cr.³⁷ Although the coil compositions of the d-POD brands analyzed in this study have not been disclosed online or on the device/packaging, both d-PODs and PODs were found to have higher Co metal concentrations than MODs. In a brief report by Talih et al.⁴⁸ that looked at five d-POD brands and compared to one POD brand (JUUL), metal concentrations varied widely and no significant difference was found between d-PODs and JUUL, with the exception of one d-POD brand (Puff Bar) having significantly higher Ni and Sb levels than JUUL. Another recent study by Halstead et al.⁴⁷ found single-use devices (Blu, Logic Power, and NJOY) that were purchased in 2016 and 2017 to have quantifiable levels of Ni in aerosol as well as the highest Cu and Zn concentrations compared to PODs (JUUL) and MODs (Fin, Mystic, Vuse, and Markten). The authors attributed the higher aerosol concentrations to the internal metal content in contact with the e-liquid. Several studies have found heating coils, such as Nichrome or Kanthal,^{24,25,29,30,37,38} that are in contact with e-liquids in the tank or POD facilitate metal leaching into the liquid. Consistently, aerosol Al, Cr, Fe, and Mn concentrations were higher in MODs than in PODs and d-PODs, and the possibility to modify the power and heating temperature could be an additional factor for metal release and transfer to e-liquid in the tank of the device.^26,68

In addition to the coil, other parts may release or contain metals. Williams et al.⁶⁹ showed the presence of Cu, Ni, Zn, and Sn in thick wires, wicks, sheaths, and joints of e-cigarette devices. The use of brass clamps and copper wires with silver coatings for instance has been associated with higher Zn, Cu, and Ag in the aerosol.^25,29,30 Moreover, a recent study on elemental analysis of POD components showed that electrical connectors and wires contained Zn, which may explain its presence in the aerosol.^46,68 Furthermore, metals may be present in the e-liquid that have not been in contact with the coil, as previously shown in our early work.²⁴

Within each device type (i.e., PODs, d-PODs), some brands were found to generate higher concentrations of certain metals than others. For example, among the POD brands, the majority of the metals were observed to be statistically significantly higher in PHIX devices, while few metals were higher in Suorin and Bo devices. The varied metal concentrations by POD brand are consistent with previous studies on PODs as well as previous device generations.^23,30 Among d-POD brands, several metals (Al, Cu, Ni, and Pb) analyzed were higher in ZPOD devices compared to Stig and Bidi Stick. Of note, some of the d-PODs purchased for our study did not work at all (i.e., did not generate aerosol) and were excluded from the study, highlighting concerns with quality control and poor manufacturing techniques, which can potentially contribute to metal impurities found in the liquids and aerosols. More research is needed to better understand the differences in device composition and how they operate as well as how these levels compare to other popular brands such as Puff Bars, which were not available for this study.

Similarly, differences in aerosol concentrations were found between flavors of a particular device type. In PODs, for example, the majority of metals concentrations measured were statistically significantly higher in tobacco-flavored aerosols compared to mint-flavored aerosols. In d-PODs, no differences were found between flavored aerosol concentrations except for Cr, which was statistically significantly higher in mint-flavored than tobacco-flavored d-PODs. One other study found higher concentrations of As, Fe, and Mn in tobacco-flavored aerosols than in menthol-flavored aerosols.⁷⁰ With so much variability between flavors and even under the same flavor group (i.e., tobacco), using flavor groups may not serve as a predictor for aerosol metal concentrations, given that e-liquids in different brands are manufactured with different ingredients and formulas to produce the same/similar flavor. Additionally, even PODs from the same manufacturer purchased at the same time were found to have aerosol metal concentrations with orders of magnitude difference from pod to pod.⁴⁵

Directly comparing the metal concentrations emitted in e-cigarettes to concentrations in combustible cigarettes was not the purpose of this study; however, if we assume that 15 puffs are equivalent to one cigarette⁷¹ and base calculations on respective flow rates (MODs: $0.9\mathrm{L}/\text{min}$; PODs/dPODs: $0.7\mathrm{L}/\text{min}$) to convert to $\text{ng}/15$ puffs, we find certain metal concentrations to fall within the same range of concentrations as that of combustible tobacco cigarettes while other metals to be higher in concentration in e-cigarettes in our study sample (Table S5). In mainstream smoke from combustible tobacco cigarettes in the United States, Cd and Pb ranged from $<5$ to 80 and $<5$ to $23\text{\hspace{0.17em}ng}/\text{cigarettes}$, respectively, while other metals (Co, Cr, Mn, and Ni) were below $10\text{\hspace{0.17em}ng}/\text{cigarette}$.⁷² In our study sample, the range (median) of Cr and Co metal concentrations ($\text{ng}/15$ puffs) for MODs, PODs, and dPODs was comparable to the range found in combustible cigarettes, while a portion of our e-cigarette samples (12%) had higher Ni concentrations and some samples had higher concentrations of Mn and Pb (1% Mn and 2.1% for Pb) (Table S5). It should be noted that Cd was not measured in our study sample, as the literature has shown significantly lower concentrations in e-cigarettes relative to combustible cigarettes.⁷⁰

Findings from this study are similar to our previous work where we found metal concentrations to be higher in aerosol samples from two brands of MODs (iStick and SMOK) compared to samples from two brands of PODs (JUUL and Blu).²⁶ More than half (67%) of the POD aerosol samples in the present study also exceeded Ni health-based limits. The MRL exceedance observed among the different device types may be due to several factors, including the large variability between brands, the sources of constituents to formulate e-liquids, and how these devices are stored,⁶⁸ when these devices were purchased (i.e., different manufacturing, composition) and how soon from the time of purchase was the aerosol sample collected and analyzed.

Absorption of metals depends on the solubility of the metal and whether the metal is in particle or soluble phase; even though to date, data is scant on the condition of each metal in e-cigarette aerosols, our previous study³¹ and others^73,74 have shown associations between metals in the inhaled aerosol and use frequency and intensity as measured in biospecimens. Therefore, it is reasonable to expect that most toxic metals will be absorbed to some extent through the respiratory tract. Concerningly, metals that were detected in our study have been associated with adverse health effects.^75,76 This is significant given the increase in POD and d-POD use over the past few years, particularly among youth and young adults^77,78; in 2020, it was determined that one in five U.S. high school students used e-cigarettes.⁷⁹ Although the FDA placed restrictions on certain flavored PODs, d-POD use among youth rapidly increased from 2.4% in 2019 to 26.5% in 2020.⁸⁰ Our results show that aerosols from MOD devices generally had higher metal concentrations than either PODs or d-PODS; however, a percentage of d-POD samples exceeded Ni, Cr, Pb, and Mn health-based limits, suggesting youth and young adults can be exposed to metal concentrations that could result in detrimental health effects.

A case study reported lung scarring (metal pneumoconiosis) typically found in metal workers following exposure to Co from e-cigarette use,⁸¹ while other studies have reported several cases of Ni-induced allergic dermatitis from e-cigarette use.^82,83 The lung is the most sensitive target to Ni toxicity, and even at the lowest adverse effect levels, lung inflammatory changes have been observed in animal models (mice, rats) as well as in nickel sinter plant workers.⁸⁴ In our collected sample, 43% of MODs, 67% of PODs, and 43% of d-PODs exceeded the minimum risk levels for Ni inhalation.²⁶ This is concerning, as Ni is an established inhalation carcinogen as Cr(VI)^21,22; exposure to Ni and Cr has also been associated with decreased lung function and increased risk of asthma, bronchitis,⁷⁵ and cardiovascular disease.⁸⁵ While several studies have reported total Cr in e-cigarette aerosol,²⁶ there is particular concern regarding the species [Cr(III) vs. Cr(VI)] that reaches the lungs. To our knowledge, no studies have assessed Cr speciation from e-cigarette use.

Arsenic is highly toxic to several organ systems. The presence of arsenic in e-cigarette aerosols has previously been reported.^{24,26,86–88} Similar levels of As have been reported in e-liquids and aerosols,²⁶ suggesting that e-liquid impurities rather than device components may be the primary source of As. One study on As species in e-cigarette liquids and aerosols⁸⁷ detected six As species in aerosol samples derived from reusable POD and MOD devices, although no d-POD devices were analyzed. Although only 15% of our samples were above the detection limit, 10% ($n=19$) of the samples exceeded the CalEPA REL for total As, which is the only health-based limit available. Of the 10% of samples that exceeded the CalEPA REL, the majority were from MODs ($n=16$), a few from PODs ($n=3$), and none from d-PODs (Table 4). Pb exceeded health-based limits in 7–18% of our e-cigarette aerosol samples. Inhalation of Pb may result in lead accumulation in the blood, soft tissue, lung, and bones⁸⁹ and has been associated with an increased risk of cardiovascular and kidney disease^17,19; furthermore, Pb is a major neurotoxicant, particularly for children, and it has been reported to decrease concentration and impair memory in adults.⁸⁹ Other metals of concern found in our samples include Al, Co, Cu, Fe, and Mn, and the health effects associated with inhalation exposure have been previously reported.^24,26,31

Even though most known health effects related to metal inhalation come from occupational exposures, which are higher than ambient concentrations and are typically limited to single metal assessments, samples in our study exceeded health-based limits for Cr, Mn, Ni, and Pb (MODs: 7–48%; d-PODs: 0–43%; PODs: 2–67%); these metals should be of particular concern given that these health-based limits apply to the general population. With a growing body of evidence showing the adverse synergistic effects of metal mixtures exposures^90,91 and the functional pulmonary impairment associated with metal exposure at ambient concentrations in young adults,^92,93 further research to characterize the potential interactions of metals found in e-cigarettes is needed to quantify the true risk of usage.

Some of the limitations of this study include a smaller sample of d-PODs relative to MODs and PODs investigated. Second, PODs and d-PODs have fixed device settings in terms of power, temperature, and coil type, while these settings varied for the MODs in our sample as each of the sampled MOD devices were set to the participants’ preferences. The effect of device settings was not part of the scope of this study as other studies have already assessed^24,26 and noted that metal emissions reach a threshold. Although device component analysis (i.e., coils, wires, joints) may provide insight into potential metal leaching sources, this was beyond the scope of our study. Also, the pH of e-liquid, shelf-life (i.e., how long the device was stored before actual use) and device age were not assessed. We are aware that shelf life and pH may change the release of metals from the device and/or e-liquid⁹⁴; however, since we collected the aerosol at the time the participants would have used their devices, or within a few weeks after purchase, which is when a typical user would consume their e-liquid, we believe the metal concentrations reported in our results reflect what users are exposed to.

Another limitation is that MRLs are calculated using resting breathing parameters and assume daily continuous exposure (24 h). Similar to our previous studies,^24,26 we assumed our collected condensed aerosol was representative of daily exposure, which may overestimate the risk. Based on our previous study describing e-cigarette use behaviors and characteristics among daily exclusive e-cigarette users,⁵¹ the hourly puff rate averages for both PODs and MODs were 14.3 puffs/h, with a maximum of 100 puffs/h and equal use throughout the day (average of 10 h). This means that our participants would vape an average of 140 puffs and up to 1,000 puffs per day. Since we collected an average of 178 puffs per sample for PODs and d-PODs, our assumption of daily consumption is reasonable for those devices. However, the average of 28 puffs for MODs means our assumption results in an overestimate of the exposure of up to 5x, which is considered small since our data spans several orders of magnitude. While our calculations and comparisons to continuous MRLs may overestimate the risk based on average use and even at the maximum of 1,000 puffs per day,⁵¹ we believe it is important to communicate to the public that metals with known toxicity and carcinogenicity are being inhaled. Furthermore, it provides salient information for regulators as they make decisions on risk for a product that is used regularly for extended periods of time.

By the time we finished the metals analysis from the collected devices, e-liquids with synthetic nicotine had not been introduced into the market, and we were unable to assess whether nicotine formulation (i.e., synthetic vs. tobacco-derived) can impact retention and/or release of metal concentrations from the e-liquid or device. Lastly, metal speciation for Cr was not performed in our analyses; future studies evaluating metal exposure from e-cigarette devices should consider classifying Cr exposure [Cr(III) vs. Cr(VI)] given that Cr(VI) is a known inhalation carcinogen.²¹

Notwithstanding these limitations, our findings can inform policymakers, health care providers, and users about the exposures associated with e-cigarettes and help design strategies to reduce the risk of metal exposure from e-cigarette use. Strengths of our study include analysis of new and emerging d-POD e-cigarette devices that are popular among youth and young adults. It also includes the collection of aerosol samples that have not been filtered or diluted during the collection process, likely reflecting what users are inhaling. Since we are collecting the aerosol at the settings used by each vaper, our results are in line with how users use these devices and reflect actual exposures. Moreover, our results provide a direct comparison of aerosol metal levels between the different device types (MODs, PODs, d-PODs), within device types particularly PODs and d-PODs, as well as between flavors.

Conclusion

Wide variability in aerosol metal concentrations was observed among and within the different e-cigarette devices and flavors. MODs and tobacco flavors were generally found to release higher metal concentrations. Findings from this study add to the existing literature that e-cigarettes are a relevant source of metal exposure, with certain devices and flavors potentially placing users at increased exposure risks. As newer devices emerge on the market, continued research is needed to evaluate the factors (manufacturing, device and e-liquid components, user vaping regimen) that contribute to metal exposure to inform regulation.