Article Text

Download PDFPDF

2020 design and methods of the Population Assessment of Tobacco and Health (PATH) study during the COVID-19 pandemic
  1. Ralph DiGaetano1,
  2. Sylvia Dohrmann1,
  3. Ethel V Taylor2,
  4. Colm D Everard3,4,
  5. Victoria Castleman1,
  6. Ting Yan1,
  7. Heather L Kimmel3,
  8. Izabella Zandberg5,
  9. Andrea Piesse1,
  10. Jean D Opsomer1,
  11. Nicolette Borek2,
  12. Marushka L Silveira4,6,
  13. Frost Hubbard1,
  14. Kristie Taylor1,
  15. MeLisa R Creamer3,
  16. Anikah H Salim2,
  17. Eva Sharma1,
  18. Yu-Ching Cheng2,
  19. Victoria Vignare1,
  20. Tammy Cook1,
  21. Wioletta Szeszel-Fedorowicz2,7,
  22. Yumiko Siegfried1,
  23. Charles Carusi1,
  24. Debra Stark2,
  25. Silvana Skara2,
  26. Andrew Hyland8
  1. 1Westat Inc, Rockville, Maryland, USA
  2. 2Center for Tobacco Products, Food and Drug Administration, Silver Spring, Maryland, USA
  3. 3National Institute on Drug Abuse, National Institutes of Health, Bethesda, Maryland, USA
  4. 4Kelly Government Solutions, Rockville, Maryland, USA
  5. 5Center for Scientific Review, National Institutes of Health, Bethesda, Maryland, USA
  6. 6National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, Maryland, USA
  7. 7Health Resources and Services Administration, Rockville, Maryland, USA
  8. 8Roswell Park Comprehensive Cancer Center, Buffalo, New York, USA
  1. Correspondence to Sylvia Dohrmann, Westat Inc, Rockville, Maryland, USA; sylviadohrmann{at}


The Population Assessment of Tobacco and Health (PATH) Study is a nationally representative, longitudinal study of the US population on tobacco use and its effects on health, collecting data annually since 2013. The COVID-19 pandemic interrupted in-person survey data collections around the world. In the USA, this included a PATH Study data collection focused on youth (13–17) and young adults (18–19) as well as other US surveys on tobacco use. Given that it was necessary to pause data collection and considering that tobacco-use behaviours could be expected to change along with pandemic-related changes in the social environment, the original design for the 2020 PATH Study data collection for youth and young adults was modified. Also, the PATH Study Adult Telephone Survey was developed to address the need for adult tobacco use monitoring in this unprecedented time. This article describes the modifications made to the 2020 PATH Study design and protocol to provide nationally representative data for youth and adults after the onset of the COVID-19 pandemic as well as the implications of these modifications for researchers.

  • COVID-19
  • Surveillance and monitoring
  • Electronic nicotine delivery devices

Data availability statement

Data may be obtained from a third party and are not publicly available. No data are available. Data used to create the timings and response rates shown in Table 3 are not available. Data used to create estimates in Table 4 are available in restricted-use format via the National Addiction and HIV Data Archive Program at the Inter-university Consortium for Political and Social Research.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • The Population Assessment of Tobacco and Health (PATH) Study is an ongoing representative study of tobacco use and health in the USA, with data collected since 2013.


  • Data collection for the PATH Study was temporarily suspended in March 2020 due to the COVID-19 pandemic. Several design modifications were made to support the resumption of data collection starting in July 2020. This study describes those modifications and provides research implications.


  • This study will assist researchers and policymakers analysing PATH Study data collected in 2020, particularly in comparison with those from other years.


The Population Assessment of Tobacco and Health (PATH) Study is an ongoing, nationally representative, longitudinal cohort study of adults and youth in the USA, collecting information on tobacco-use patterns and associated health behaviours. A stratified, address-based, area probability sampling design was implemented at Wave 1 (2013–2014), oversampling several subgroups of adults (adults who use tobacco, young adults (ages 18–24) and African American adults) as well as including youth (ages 12–17) and a shadow sample (ages 9–11) at disproportionately high rates, which resulted in 32 320 adult and 13 651 youth interview respondents. (Whenever the PATH Study sample is replenished with newly sampled cases to be interviewed, youth under age 12 are also sampled and parents provide consent allowing them to be interviewed at later waves when they are ages 12 and older. These ‘under age 12’ samples are referred to as ‘shadow’ samples. The Wave 1 weighted response rates were 54.0%, 74.0%, 78.4% and 80.2% for the household screener, adult interview, youth interview and parent consent for shadow youth, respectively.) Using in-person interviews, the PATH Study collected data annually for the full sample (interviewing youth ages 12–17 and adults) through Wave 4, after which full sample data collections, or primary waves, were scheduled every 2 years (see table 1 for data collection dates specific to each study wave).

Table 1

PATH Study data collection dates, types, ages interviewed, methods and numbers of interviews by wave*

Additional documentation on the PATH Study design and implementation through Wave 4 can be found in Hyland et al,1 Piesse et al,2 Opsomer et al3 and the PATH Study Restricted Use Files User Guide4 at

After the primary waves became biennial, special data collections were conducted in intervening years. Wave 4.5 (2017–2018) followed youth ages 12–17 only. Wave 5.5, originally scheduled for 2019–2020, was developed to follow the same youth 2 years later as youth and young adults. 10-year-old ‘shadow sample’ members at Wave 4, who were age 11 and thus not interviewed at Wave 4.5, were interviewed at Wave 5.5, resulting in a target population ages 13–19. (Prior to completing an interview, all respondents 18 and older provide informed consent and youth respondents ages 12–17 provide assent after a parent/legal guardian provides consent.)

The COVID-19 pandemic interrupted the Wave 5.5 data collection for the PATH Study and other US national surveys on tobacco use and created an environment in which tobacco-use behaviours could be expected to change, including among youth with the closure of in-person schools. Moreover, in late 2019, the Federal minimum age for the sale of tobacco in the USA changed through legislation amending the minimum age for the sale of tobacco products from 18 to 21 years of age (also called ‘Tobacco 21’).5 In 2020, the Food and Drug Administration issued a policy prioritising enforcement against certain flavoured cartridge-based e-cigarettes that were not authorised for marketing in the USA, as well as prioritising ongoing enforcement against products marketed to youth.6 As a result of the pandemic, the original design for Wave 5.5 was modified and a second study, the PATH Study Adult Telephone Survey (PATH-ATS), was developed to collect data from adults ages 20 and older. The addition of the PATH-ATS allowed the PATH Study to become one of the few sources of adult tobacco-use data for 2020.

This article describes modifications made to the 2020 PATH Study design and protocol to provide nationally representative data for youth and adults after the data collection suspension as well as the implications of these modifications for researchers.


Two major design modifications were made after the onset of the pandemic. The first was specific to Wave 5.5, which began 1 December 2019, using the usual in-person interviewing protocols. The second was the addition of a new data collection effort to survey adults ages 20 and older.

Wave 5.5 data collection was planned to occur over 12 months using audio computer-assisted self-interviewing (ACASI) for adult (ages 18–19) and youth (ages 13–17) interviews and computer-assisted personal interviewing (CAPI) for interviews of youth’s parents. When data collection was suspended on 17 March 2020, because of safety concerns stemming from the pandemic, the immediate questions were, ‘When and how can data collection best resume?’ The PATH Study team identified five major issues requiring attention before data collection could resume; these are summarised in table 2.

Table 2

Issues, modifications and reasons for modifications to the original 2020 sample design and study protocol

Items 1 and 2: Uncertainty about the resumption of data collection and the interpretation of data

With uncertainty about when data collection would resume, it was clear that there would be a significant time gap in data collection for Wave 5.5, preventing that wave’s data from covering the original 12-month period as designed. Moreover, most in the targeted age range were in school at the time in-person data collection was suspended. Pandemic-related school closures affected the social life of this group, with a potential effect on tobacco-use opportunities and exposure. Thus, it was expected that tobacco use among those who had been in school would change during the time when no PATH Study data were collected. Rather than pooling data collected prior to mid-March with data collected later in the year, the Wave 5.5 time frame was ‘reset’ to cover the latter half of 2020 only. (Extending data collection into 2021 was considered, but ultimately rejected as Wave 6 was already planned for early 2021.) This avoided both confounding the interpretation of Wave 5.5 estimates arising from data collected from two radically different social environments and the absence of data for a portion of the year due to the data collection hiatus, when tobacco-use behaviour could be expected to change. Participants scheduled for interviews in the latter portion of the year were not a random subsample of those eligible for Wave 5.5, so participants who completed a Wave 5.5 interview in person on or before 17 March 2020, and were still age-eligible for Wave 5.5 (ie, those under 20) were recontacted for an interview in the new time frame (see table 1). (Most of those fielded for Wave 5.5 and whose in-person interview status had been finalised as ‘nonrespondent’ prior to the hiatus had been classified specifically as ‘refusals’. Because it was expected that a high percentage of these nonrespondents would also refuse a request to complete a subsequent telephone interview, it was decided that no one finalised as a nonrespondent prior to the hiatus would be recontacted for a Wave 5.5 telephone interview. A total of 15 793 study participants younger than 20 on 31 August 2020, including those who were a Wave 5.5 nonrespondent during the in-person survey period, were included in the weighting process.)

Item 3: Expansion of the target population

With uncertainty about how the pandemic would affect the full adult population, it was of analytic interest to expand the original 2020 target population ages 13–19 to cover the full range of adults ages 18 and older. The PATH-ATS was therefore developed to obtain data for adults ages 20 and older to complement the original target population (ages 13–19 at Wave 5.5). (The reference date used to distinguish between 19 and 20 years old was 31 August 2020.) Because the data collection period was shorter than most PATH Study data collection efforts to coincide with the revised Wave 5.5 time frame, only a subsample of adult participants were interviewed. A stratified random sample of 18 601 PATH Study participants was selected for the PATH-ATS from a total of 31 343 eligible participants, oversampling adults who used particular tobacco products of high interest to researchers (eg, electronic nicotine delivery systems (ENDS), or both ENDS and cigarettes).

Item 4: Transition to telephone

After the suspension of the original data collection effort, it became clear that it would be some time before in-person interviewing would be feasible. The PATH Study routinely collects telephone numbers from participants. This additional method of contact was available for over 95% of the sample, so it was decided to change from in-person interviewing to telephone interviewing. Because the samples consisted solely of past respondents, their familiarity with PATH Study procedures was expected to ease the transition to the new mode of data collection. Although the instruments could in principle have been programmed for web administration, the shift to telephone could proceed more quickly. The Wave 5.5 ACASI instruments already supported in-person interviewer administration in instances where respondents did not want to enter their own answers into the computer, making it straightforward for the interviewers to administer the instruments by telephone.

To transition from in-person administration to telephone data collection and to capture timely information related to the pandemic, several modifications were made to the Wave 5.5 instruments originally developed for ACASI administration. For example, visual aids or ‘show cards’ were created to support telephone administration of the adult, youth and parent interviews for questions that contained visual imagery, sensitive content or long or complex response options. These show cards were made available electronically on the internet or mailed in paper form to participants unable to access the internet. (Days prior to the interview, the field interviewers notified the participant (or parent) that the interview would be conducted by telephone and verified internet access. If the participant did not have internet access, the field interviewer scheduled an appointment for at least 5 days later to allow time for the paper materials to be delivered to the participant. Participants with internet access were directed to the PATH Study participant website and the location of the show cards at the start of the interview. Researchers can view show cards and the questions for which they were referenced in the annotated instruments available at ( The average interview administration times for Wave 5.5, PATH-ATS, and comparable timings for Wave 5 are shown in table 3.

Table 3

Interview timings and weighted response rates for the Population Assessment of Tobacco and Health (PATH) Study Waves 5 and 5.5 and the Adult Telephone Survey (PATH-ATS)*

Item 5: Obtain data about the impact of COVID-19

New questions were added to the re-launched Wave 5.5 adult, youth and parent interviews to capture detailed information about the impact of the pandemic on respondent health and tobacco use. The added questions included: social distancing practices; impact of the pandemic on tobacco use; perceptions of COVID-19 severity among those who smoke cigarettes and use ENDS compared with those who do not; impact of the pandemic on respondents’ stress; and questions related to COVID-19 symptoms and diagnosis. (Social distancing and tobacco-use questions were adapted from the International Tobacco Control Policy Evaluation Project—April 2020 Four Country Smoking and Vaping Survey. Impact on stress question was adapted from National Institutes of Health Coronavirus Impact Scale. COVID-19 symptoms were informed by those listed on the Centers for Disease Control (CDC) website. All questions related to COVID-19 were worded the same across the three instruments; additional questions relevant to older adults were included in the PATH-ATS instrument. The timeline for developing these instruments was such that these items were not subject to cognitive testing.)

The data collection instrument for the PATH-ATS was designed to capture core measures associated with tobacco use and health. To determine tobacco-use prevalence, the PATH-ATS asked many standard PATH Study questions including current and former use of cigarettes, ENDS, cigars, smokeless tobacco, snus, pipes and hookah. (The use of standard PATH Study questions, with the same or similar wording across instruments, facilitates the combined analysis of Wave 5.5 and PATH-ATS data. Researchers are encouraged to review the respective instruments and codebooks for more information.) Detailed information on nicotine dependence and cessation was captured among adults who smoke cigarettes and use ENDS. Generally, the PATH-ATS instrument was a shortened version of the previous adult instruments to better accommodate telephone administration, but additional questions were added as deemed necessary to meet the analytical needs of the study.

Pandemic-related questions developed for Wave 5.5 were included in the PATH-ATS instrument along with new questions that were more relevant for the older adult population such as the impact of the pandemic on unemployment and questions about COVID-19 severity for those testing positive. (Question on impact of COVID-19 on unemployment was adapted from the Census 2020 Household Pulse Survey. The specific question wording used in the Wave 5.5 and PATH-ATS instruments can be found on the NAHDAP website: Similar to Wave 5.5 implementation, electronic or paper show cards were used to support telephone administration of questions containing visual, sensitive or complex content in the PATH-ATS instrument.

Existing PATH Study field interviewers conducted telephone interviewing for the Wave 5.5 sample. Changes were made to protocols to aid in scheduling appointments and to verify internet access. Field interviewers were trained to administer questions with show cards and assist the respondent in identifying appropriate show cards and forms. They also learnt how to administer sensitive questions that were previously self-administered and were instructed on the pronunciation of medical terminology and prescription drug and tobacco product names. Trained telephone interviewers not previously associated with the PATH Study conducted the PATH-ATS interviews using specialised scheduling, management and CATI systems.

The Wave 5.5 modifications were made in approximately 3.5 months, and telephone interviewing began on 3 July 2020. The PATH-ATS was conceived after plans to resume Wave 5.5 were in place. Development of this new survey was done quickly to permit interviewing to begin as soon as possible, about 2 months after Wave 5.5 data collection resumed. Both data collections continued until late December 2020. Table 1 provides the number of Wave 5.5 and PATH-ATS telephone interviews completed. The data collected through these efforts can be used together to create estimates representing the US population ages 13 and older (or some subset) in the latter portion of 2020 who were in the civilian, non-institutionalised population at the time of Wave 4 and in the resident (not incarcerated) US population after that time. (See the PATH Study Restricted Use Files User Guide for instructions on combining the Wave 5.5 youth and adult data files with the PATH-ATS data file and further discussion related to such analyses.) We consider estimates for adults ages 18–24 in the discussion.


The longitudinal nature of the study allows researchers to examine how the pandemic and resulting study design modifications may have affected PATH Study participants’ interview responses and propensity to respond. Interview responses to Wave 5.5 and PATH-ATS can be compared with responses to the same or similar questions from previous waves that are not expected to change much wave to wave (eg, being diagnosed with cancer). One would not expect the 2020 data to show many differences in responses to such questions.

To assess response propensities to the 2020 data collection efforts, we examined weighted response rates7 for Wave 5 compared with those for Wave 5.5 and PATH-ATS. These appear in table 3. (All response rates were computed using the RR3 response rate formula provided by the American Association for Public Opinion Research. Weighted response rates estimate the proportion of the survey population that would have responded if asked to participate in the study, providing a measure of the potential impact of nonresponse on study estimates. The weighted response rates for Wave 5 and Wave 5.5 condition on participation at Wave 4; the PATH-ATS response rates condition on selection into the PATH-ATS sample. Differential response rates were accounted for by the weighting process.) The ages of those eligible for youth interviews at Wave 5 (ages 12–17) and Wave 5.5 (ages 13–17), and those eligible for adult interviews at Wave 5 (ages 18+), Wave 5.5 (ages 18–19) and PATH-ATS (ages 20+) do not match exactly. Nevertheless, the comparisons remain informative, assessing response propensities overall, by study recruitment wave, and for demographic and tobacco-use categories as of Wave 4. (Participants were recruited into the PATH Study either at Wave 1 or through the replenishment effort at Wave 4.) Note that participants were generally 3 years older in 2020 than they were at Wave 4. In addition, all Wave 5 interviews were conducted in person, whereas all Wave 5.5 and PATH-ATS interviews were by telephone; the adult and youth interview incentives were the same for Wave 5, Wave 5.5 and PATH-ATS.

The Wave 5 response rates were noticeably higher than corresponding rates for Wave 5.5 and PATH-ATS, as expected, since telephone surveys generally achieve lower response rates than in-person surveys.8 Other large, in-person Federal surveys that switched to telephone data collection in 2020 also experienced decreases in response rates. (These include the 2020 American Community Survey, the 2020 Current Population Survey Annual and Social Economic Supplement, the 2020 Consumer Expenditure Surveys and the 2020 National Health Interview Survey. The pre-pandemic and post-pandemic response rates for these surveys are summarised in Krieger et al.)9–13 The PATH-ATS response rates were likely further reduced due to the compressed field period (less than 15 weeks) and modified contact protocol for the PATH-ATS, eliminating some standard follow-up procedures used in the PATH Study, including Wave 5.5. (Data collection was limited to this period to coincide with the revised Wave 5.5 time frame.)

PATH-ATS response rates appeared correlated with two indicators of socioeconomic status: educational attainment and health insurance status. Response rates by level of education as of Wave 4 range from 38.2% for those with no high school degree up to 70.0% for those with at least a bachelor’s degree, whereas response rates for those with health insurance were 17 points higher than for those without (57.9% vs 40.7%). The finding that adults of higher socioeconomic status were more likely than adults with lower socioeconomic status to respond also aligns with the results of other government-sponsored surveys conducted during 2020.10

Differences in response rates alone do not indicate nonresponse bias in the Wave 5.5 or PATH-ATS estimates. The Wave 5.5 and PATH-ATS sample weights include adjustments for nonresponse designed to limit or eliminate such bias. Full nonresponse bias analyses were performed for Wave 5.5 and PATH-ATS,14 15 and no serious concerns were identified regarding potential nonresponse bias in estimates from these data collection efforts. Generally, the response rates achieved in 2020 reflect the challenges of data collection during the pandemic and the changes to the protocol.


The changes made to the PATH Study design for 2020 resulted in the continuation of annual data collection and provided the opportunity to learn about health and tobacco-use behaviours during the pandemic for youth and all ages of adults, but this required a change in mode and a shortened data collection period.

The PATH Study was not alone in having to adjust to the pandemic. Internationally, many surveys suspended data collection in March 2020 and then modified designs and contact protocols. These include the Survey of Health, Ageing and Retirement in Europe, the German Family Panel (pairfam), the Generations and Gender Survey and Understanding Society, all in Europe,16–18 as well as major US Federal surveys such as the Current Population Survey (CPS), the American Community Survey (ACS) and the National Health Interview Survey (NHIS). The four European surveys are panel surveys, whereas in the USA the ACS and NHIS are cross-sectional and the CPS has a rotating panel design with both new and previously participating households interviewed each month. We consider the three US surveys for direct comparison with the PATH Study. (The National Youth Tobacco Survey (NYTS), another data source for tobacco use among youth in the USA, also encountered data collection issues in 2020. The NYTS collects data from youth in the school setting and was expected to be fielded until mid-May in 2020. The onset of the pandemic caused all NYTS data collection efforts to end as of 16 March 2020. Thus, the NYTS data for 2020 reflected tobacco use and related behaviour among youth prior to any impact due to the pandemic with a substantially lower school participation rate (49.9%) than in 2019 (77.2%). For 2021 the NYTS did undertake changes in the fielding procedures, including changes in the mode of data collection and in its documentation indicated that, due to these changes, ‘the 2021 NYTS results cannot be compared with results from previous NYTS survey waves that were primarily conducted on school campuses’.)19

All three surveys reported concerns about the data collected in 2020 being of lower than usual quality due to reduced response rates and associated evidence of nonresponse bias. Specifically, the three surveys reported evidence of nonresponse bias associated with socioeconomic status, with households in higher socioeconomic levels relatively more likely to respond but with sample weighting unable to fully compensate for this. Although adults with higher socioeconomic status were also more likely to respond to the PATH-ATS, there is no evidence of nonresponse bias in the weighted estimates. (Socioeconomic status could not be directly assessed in the Wave 5.5 nonresponse bias analysis in the same manner as in the PATH-ATS analysis given the types of questions asked of Wave 5.5 respondents at Wave 4.) The PATH-ATS appears to have benefitted from the longitudinal nature of the PATH Study, with a wealth of information from prior waves available for adjusting weights for nonresponse.

Nevertheless, there are important considerations for researchers to take into account when using 2020 data from the PATH Study. In particular, the shift to telephone interviewing for the PATH Study raises issues that analysts should recognise. PATH Study data were collected via telephone whereas data for all prior waves were collected through ACASI, resulting in greater involvement of interviewers in the 2020 data collection process. For example, telephone administration requires interviewers to read the questions to the respondent and record the responses, whereas, with ACASI, respondents read the questions themselves and directly enter the responses into the survey instrument. This difference in interviewer involvement in data collection has been shown to affect answers to sensitive questions.20–22

A second difference between telephone administration and ACASI is the mode of communication. With telephone administration, questions are read and answers are provided orally, an auditory interaction. However, for ACASI, respondents see questions and answers on the computer screen, an interaction incorporating visual cues. This difference in communication mode has been shown to affect answers to questions with a long list of response options, producing response order effects.23 As noted earlier, an effort was made to provide show cards via mail or the internet to help retain at least some of the visual component of the ACASI methodology.

It was important to the PATH Study that the data collected at Wave 5.5 be, to the extent possible, the same as originally planned. Thus, the Wave 5.5 interview was essentially an ACASI instrument administered by telephone. Even with modifications to ease the transition to telephone administration, the resulting Wave 5.5 interviews tended to be much longer than for prior waves (see table 3). The longer interview completion time could affect the level of comparability of Wave 5.5 estimates to estimates from prior waves. On the other hand, since there was originally no intention to collect data for those ages 20 and older in 2020, the PATH-ATS interview could be designed specifically for telephone administration and tailored to the special circumstances of the pandemic, in addition to retaining some questions comparable to those administered in past waves. As a result, fewer questions were asked than in the usual adult interview so that the administration time was roughly half the length of adult interviews at prior waves.

One limitation for both surveys was the shorter time period for data collection, with perhaps an impact on response rates and shifting of the data collection period to the latter portion of the year. To the extent that there may be a seasonal component to tobacco use (eg, related to school attendance), comparisons over time may be affected. However, the pandemic affected different localities in different ways and at different times of the year. Thus, its impact was not generally uniform across regions or time periods, possibly resulting in comparability issues with past PATH Study waves.

One final factor to keep in mind when comparing 2020 PATH Study estimates to those of other waves involves newly amended Federal tobacco sales laws and issued policy changes. Tobacco 21 legislation and the enforcement policy on unauthorised flavoured cartridge-based e-cigarettes enacted in 2020 were designed to impact tobacco-use behaviour. With the 2020 data collected in the latter portion of the year, it is possible that such policy changes would have been captured in PATH Study estimates.

Because of the issues of comparability discussed above, it may be helpful to consider a simplified example of an exploration of 2020 PATH Study data. For illustrative purposes, table 4 includes cross-sectional descriptive statistics (estimates and corresponding 95% CIs) on the past 30-day use of ENDS among young adults (ages 18–24) by subgroups based on sex, race/ethnicity and educational attainment using 2020 data, in comparison to estimates from prior waves available from Restricted Use Files.

Table 4

Weighted estimates of past 30-day use of electronic nicotine delivery systems (ENDS) among young adult (ages 18–24) participants in the Population Assessment of Tobacco and Health (PATH) Study, by socio-demographic characteristics over time†

There was an increase in the current use of ENDS between Wave 1 and Wave 5, most of it occurring in the 2-year interval between Waves 4 and 5 across all demographic categories considered. However, in the time between Wave 5 and the data collected in 2020 (roughly a year and a half), there were sizeable decreases for all categories. Even with the comparability issues identified earlier (mode, data collection timing, new tobacco-related policies), these differences indicate evidence of an important change in 2020 in the use of ENDS products among young adults, an age group in which such use had been increasing. There remains uncertainty about the actual magnitude of such change due to the comparability issues.

The estimates in table 4 are cross-sectional. However, because the PATH Study is a longitudinal study, analysts can examine data from the same individuals over time in models that more fully assess potential factors influencing tobacco-use behaviour. For example, patterns of tobacco use for adults ages 18–20 in 2020 could be examined across prior years back to Wave 4 when they were youth.

Unfortunately, it may not be possible to clearly identify the impact of any single factor on 2020 estimates with so many potential confounders. Because of the differences in the data collection protocol, the period of data collection and the impact of the pandemic, researchers should exercise caution when comparing estimates from Wave 5.5 and the PATH-ATS to estimates from other waves. Researchers are advised to first examine differences in demographic composition and key variables of analytic interest (eg, tobacco use) by mode of data collection and to consider sensitivity analyses before interpreting results.

Nevertheless, while caution in terms of comparability is warranted, the PATH Study was able to avoid many of the problems other studies encountered involving data collected in 2020. The sample design was adapted to produce estimates using a single mode of data collection for well-defined target populations. Questions related to the pandemic were incorporated into the survey instrument, providing researchers with an opportunity to examine the impact of the pandemic on tobacco use and related behaviour. Although care is needed in assessing comparisons over time, the PATH Study data collected in 2020 provide interesting information for researchers on tobacco use during the COVID-19 pandemic.

Data availability statement

Data may be obtained from a third party and are not publicly available. No data are available. Data used to create the timings and response rates shown in Table 3 are not available. Data used to create estimates in Table 4 are available in restricted-use format via the National Addiction and HIV Data Archive Program at the Inter-university Consortium for Political and Social Research.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by Federal-wide assurance (FWA) identification # is 00005551, IRB registration # is 0000695. Participants gave informed consent to participate in the study before taking part.



  • Contributors RD conceived the presented redesign with input from JO, SD, CC, KT, AH, ET, Y-CC, CE, HLK and NB. VC adapted the questionnaires, and TC and VV led data collection operations with CC and KT. KT and ES designed the PATH-ATS sampling strata and the sample was selected by YS under the supervision of RD, SD and JO. The response rates and nonresponse bias analyses were created by AP and FH. The development of ENDS estimates was led by ES. RD and SD co-wrote the manuscript with direct contributions from TY, VC, TC, AP, FH, KT and AH. ET, CE, HLK, IZ, JO, NB, MLS, MC, AHS, ES, Y-CC, WS-F, CC, DS and SS provided feedback on the manuscript. SD is the guarantor.

  • Funding This manuscript is supported with Federal funds from the National Institute on Drug Abuse (NIDA), National Institutes of Health (NIH), and the Center for Tobacco Products (CTP), Food and Drug Administration (FDA), Department of Health and Human Services, under contract to Westat (contract nos. HHSN271201100027C and HHSN271201600001C) and through an interagency agreement between NIH NIDA and FDA CTP. HLK and MC were substantially involved in the scientific management of and providing scientific expertise for contract nos. HHSN271201100027C and HHSN271201600001C.

  • Disclaimer The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US Department of Health and Human Services or any of its affiliated institutions or agencies. This article work was prepared while MLS was employed at NIDA, NIH via Kelly Government Solutions and while IZ and WS-F were employed at the CTP, FDA.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.