Content uploaded by Floe Foxon
Author content
All content in this area was uploaded by Floe Foxon on Jul 17, 2021
Content may be subject to copyright.
For Peer Review Only
Estimating E-Cigarette Use Prevalence Among US
Adolescents Using Vaping-Related Online Search Trends
Journal:
Substance Use and Misuse
Manuscript ID
LSUM-2020-0685.R1
Manuscript Type:
Research Note
Keywords:
Adolescent, E-Cigarette, Vaping, Zero-Normalized Cross-Correlation,
NYTS, Google Trends
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
For Peer Review Only
Abstract
Introduction: Adolescent e-cigarette use is a developing phenomenon. Greater surveillance of
underage use is necessary to inform e-cigarette policy and mitigate adolescent e-cigarette use.
Accurate prevalence estimates for adolescent e-cigarette use are provided by large national
surveys. However, these surveys are costly and provide only annual estimates. To obtain more
affordable estimates faster and more frequently, novel methods are required.
Methods: Online search term popularity data were taken from Google Trends. Interest in vaping-
related search terms were followed monthly from January 2011 to November 2020. Time-lagged
zero-normalized cross-correlations were performed between the Google data and current (past 30
day) high-school e-cigarette use prevalence estimates from the National Youth Tobacco Survey
(NYTS). The search interest data were then calibrated to the NYTS data to estimate adolescent e-
cigarette use prevalence using online searches.
Results: Maximum correlation coefficients of 0.979 for ‘vapes’ and 0.938 for ‘vape’ were
obtained when search interest lagged use prevalence by one month, and 0.970 for ‘vape pen’
when the lag was two months (p<0.001 for all). Calibrating the search term data to NYTS
provided a high-school current e-cigarette use prevalence estimate of 12.1–18.4% for November
2020, suggesting adolescent use of e-cigarettes has continued to decline since the NYTS estimate
of 19.6% for January–March 2020.
Conclusions: Online search trend data may provide reasonably reliable and more frequent
estimates of adolescent e-cigarette use prevalence at substantially lower costs than traditional
surveys. Such additional data may help to assess immediate impacts of policies and events.
Page 1 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Key Words: Adolescent; E-Cigarette; Vaping; Zero-Normalized Cross-Correlation; NYTS;
Google Trends
Page 2 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Introduction
E-cigarettes have emerged as an alternative nicotine source and possible future smoking-
cessation device for adult smokers (American College of Cardiology, 2020; Hajek et al., 2019),
but adolescent use of e-cigarettes (Gentzke et al., 2019) presents a public health challenge.
Tobacco control policies are shaped by available data on prevalence of product use.
Existing data sources for adolescent e-cigarette use prevalence include Monitoring the Future
(University of Michigan, 2021), the Youth Risk Behavior Survey (YRBS) (US Centers for
Disease Control and Prevention, 2020) and the National Youth Tobacco Survey (NYTS) (US
Centers for Disease Control and Prevention, 2019a), with YRBS and NYTS administered by the
US Centers for Disease Control and Prevention, as well as various state-wide surveys such as the
state Youth Tobacco Surveys (YTS) (US Centers for Disease Control and Prevention, 2019b)
and Healthy Kids surveys (California Department of Education, 2020; Colorado Department of
Public Health & Environment, 2019).
These surveys utilize large sample sizes and complex sample designs which provide the
best existing estimates of adolescent e-cigarette use prevalence. However, administering state-
and nation-wide surveys is a costly endeavor in terms of the time, money, and labor involved.
Consequently, these surveys are administered at most annually or biannually.
Because adolescent e-cigarette use is evolving, data with higher temporal resolution (i.e.
less time between point estimates) and which require fewer resources to collect are necessary;
waiting an entire year to determine the impact of a policy may risk irreversible consequences if
that policy fails or backfires.
Adolescent e-cigarette use intersects with the online world. Half of US adolescents may
be exposed to tobacco or e-cigarette related social media (Hebert et al., 2017), and about 1 in 10
adolescent e-cigarette users source the e-cigarettes they use from the internet (Merianos et al.,
Page 3 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
2019). E-cigarette related discussion forums have also been identified as a rich source for
information on adolescent e-cigarette use behavior (Zhan et al., 2019). Given then that
adolescent e-cigarette use is associated with online content, this raises the question of whether
vaping-related internet use predicts real-world e-cigarette use.
Google Trends (Google, 2020), an online, search-term-popularity data generator, has
been demonstrated to have modest reliability in reflecting geographical and temporal patterns in
epidemiological settings (Cervellin et al., 2017). Extant research has applied temporal Google
Trends data to predict outbreaks of influenza (Carneiro & Mylonakis, 2009) and COVID-19
(Ahmad et al., 2020), the latter with correlation as high as 0.998. In nicotine and tobacco science,
relative Google search volumes have been compared for heat-not-burn products and e-cigarettes
to investigate heat-not-burn market growth (Caputi et al., 2017). Most relevant to the present
study, statistically significant but weak-to-moderate strength geographical correlation was
demonstrated between Google Trend data and state use prevalence for cigars in 2011 (Cavazos-
Rehg et al., 2015).
This study aims to determine whether Google Trends data correlate temporally with
adolescent e-cigarette use prevalence in the US using NYTS prevalence estimates from 2011–
2020. This presents novel research which may provide faster and higher temporal resolution
estimates of adolescent e-cigarette use prevalence.
Materials & Methods
Sample
National estimates for the prevalence of past 30 day (current) e-cigarette use among US
high-schoolers (here ‘adolescents’) from 2011–2020 were taken from resources published by the
US Department of Health and Human Services (US Food and Drug Administration, 2019b;
Page 4 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Wang et al., 2020). These prevalence estimates are derived from ten waves of NYTS, which is an
annual, nationally representative survey of US adolescents with a multi-stage probabilistic
cluster sample design. Approximately 17,000–25,000 adolescents were surveyed at each wave.
The exact dates through which these surveys were administered were taken from their annual
methodology reports (US Centers for Disease Control and Prevention, 2019a).
Online search popularity data were taken from Google Trends. Google Trends provides
normalized ‘search interest’ values which represent the popularity of a given search term through
Google web searches on a scale from zero to 100. A value of 100 represents the peak popularity
for that term, which falls on some month. A value of 50 on another month means the search term
was half as popular on that month than it was in the month when the value was 100. Monthly
data for the vaping-related search terms ‘e-cigarettes’, ‘e-cigs’, ‘mods’, ‘e-hookahs’, ‘vapes’,
‘vape’, and ‘vape pen’ were collected from January 2011 to November 2020. The above search
terms were selected because they are listed as alternative names for electronic cigarettes in the
2019 NYTS questionnaire definition (which precedes the e-cigarette-related questions in this
survey) (US Centers for Disease Control and Prevention, 2019a).
The range of dates was selected to coincide with NYTS data collection, and the data were
restricted geographically to ensure only searches from within the United States were included for
analysis.
Analyses
The NYTS prevalence estimates were normalized, and for visualization purposes the
Google web search interest data were re-normalized to the same scale by dividing each
popularity score in each set of search term data by the maximum popularity score for that search
Page 5 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
term such that for all search terms, a value of one represented the maximum popularity score for
that search term. These data were then plotted to observe the similarities and differences in their
trends over time.
The midpoint dates between the beginning and end of survey administration for each
NYTS wave were found, and the months these dates fell on were taken to be the months which
correspond to the prevalence estimate for that year. The NYTS data were then paired with the
Google web search data for the same months (zero lag). Treating the data as signals over time,
zero-normalized cross-correlations were performed between NYTS and each of the Google web
searches with these month-pairs, following the methods of Ahmad et al. (2020). Conceptually, it
is plausible that a change in search popularity may precede a corresponding change in current
use prevalence and vice versa. For example, an adolescent who has not used an e-cigarette in the
past 30 days may first search for online vaping-related content out of curiosity, become
influenced by that content over time, and then initiate e-cigarette use at some later date when an
opportunity arises or when their interest reaches a threshold to do so. Conversely, an adolescent
who is already a current user may after some time using e-cigarettes seek information on how to
quit. In both cases there will occur some time between online searching and meeting the
definition of a current user. Since there is no published literature on the duration of this lag,
cross-correlation as described above were repeated with exploratory time lags of months up to
three months in either direction, for example a one-month lag was implemented by cross-
correlating the NYTS data with Google Trends data which correspond to the NYTS months
minus one month, and a ‘-1’ month lag was implemented by cross-correlating the NYTS data
with Google Trends data which correspond to the NYTS months plus one month. The use of lags
accounts for the time taken for online interest to develop into real-world use and vice versa.
Page 6 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Corresponding p-values were calculated for the cross-correlation coefficients with statistical
significance at alpha=0.05.
Finally, the peaks of the search term popularity trends were calibrated to the peak NYTS
e-cigarette use prevalence (27.5%) such that e-cigarette use prevalence estimates could be
obtained from the search term data for any given month. In this way, prevalence estimates were
obtained for November 2020.
All analyses were conducted in Python version 3.7.6 with the packages NumPy version
1.18.5, Matplotlib version 3.2.2, and Pandas version 1.0.5. Code will be made available upon
reasonable request.
Results
Figure 1 shows the normalized trends in high school e-cigarette use prevalence and
Google web search popularity for the search terms ‘vapes’, ‘vape’, and ‘vape pen’ over time. It
can be readily seen than all three trends display very similar shapes.
Table 1 shows the zero-normalized cross-correlations between the NYTS prevalence
estimates and corresponding Google web search interest for a range of time lags. The maximum
cross-correlation coefficients were 0.979 for ‘vapes’ and 0.938 for ‘vape’ with search interest
lagging NYTS prevalence by one month, and 0.970 for ‘vape pen’ with search interest lagging
NYTS prevalence by two months. Across the entire range of lags, all cross-correlations for all
terms were statistically significant. In contrast, correlations were not significant for the search
terms ‘e-cigarettes’ (cross-correlation coefficient=0.423; p=0.2234), ‘e-cigs’ (-0.215; p=0.5506),
‘mods’ (-0.545; p=0.1032), and ‘e-hookahs’ (-0.174; p=0.6305) (trends not shown).
Page 7 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Calibrating the search term popularities to NYTS prevalence data for January–March
2020 gave current e-cigarette use prevalence estimates of 19.3% (‘vape’ search term with one-
month lag), 17.3% (‘vapes’ search term with one-month lag), and 12.7% (‘vape pen’ search term
with two-month lag). Since the actual prevalence estimate for January–March 2020 from NYTS
was 19.6%, this suggests good agreement between NYTS and the ‘vape’ search term with one-
month lag model, and less agreement with NYTS for the other two models.
Using these models, prevalence estimates for November 2020 of 18.4% (‘vape’ search
term with one-month lag), 16.8% (‘vapes’ search term with one-month lag), and 12.1% (‘vape
pen’ search term with two-month lag) were obtained.
Discussion
This research aimed to determine whether online vaping-related search data from Google
Trends correlate with adolescent e-cigarette use prevalence in the US. The results presented here
show very high and statistically significant correlation between NYTS prevalence estimates and
interest in the search terms ‘vapes’, ‘vape’, and ‘vape pen’, but no statistically significant
correlations were observed for the search terms ‘e-cigarettes’, ‘e-cigs’, ‘mods’, and ‘e-hookahs’.
This may suggest that the prior terms are more commonly used by adolescents to describe vaping
than the latter terms.
Correlations were strongest with 1–2 month lags, suggesting changes in vaping-related
search trends are shortly followed by changes in adolescent e-cigarette use trends, however
whether this relationship is causal is not determined.
Using the search trend data to predict high school current e-cigarette use prevalence
provided a prevalence estimate for November 2020 of 12.1–18.4%, which is lower than the
19.6% estimated by NYTS for January–March 2020. These findings may show that the number
Page 8 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
of adolescent e-cigarette users has continued to decline since NYTS 2020 was administered. This
suggests that recent adolescent e-cigarette use prevention efforts such as Tobacco 21 (US Food
and Drug Administration, 2019a), along with likely effects from pandemic-mitigation efforts,
have had some impact on reducing adolescent e-cigarette use.
Although a proof-of-concept, this research shows that monitoring online interest in
vaping may prove to be a cost-free and reliable means of generating more frequent prevalence
estimates than large national surveys.
Future research may apply these methods to specific geographies and patterns of use.
Strengths
Strengths of this research include the novel application of temporal online search data to
predict trends in e-cigarette use prevalence among US adolescents, and rigorous statistical
analysis.
Limitations
The primary limitation of this research is that the mechanism which associates vaping
search interest with e-cigarette use prevalence is complex, therefore an increase in search
popularity does not necessitate an increase in use prevalence. Online dissemination of
information regarding the health risks of nicotine product use may result in greater search
popularity but not greater use prevalence. Never e-cigarette users may search vaping-related
terms out of curiosity without initiating e-cigarette use, and former users may search for support
to sustain their abstinence. Furthermore, Google Trends does not provide age-filtering of Google
web searchers, therefore adolescent vaping-related search trends are confounded by other age
groups making the same searches. However, US internet usage is disproportionately dominated
Page 9 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
by younger Americans (Jones & Fox, 2009), and the trends in search term popularity presented
in this work do not resemble trends in adult e-cigarette use from the literature (see Figure 1(A) of
Dai and Leventhal (2019)), which may suggest that most searches involving these search terms
are made by adolescents and not by adults. An interesting extension of this work would be to
identify which online search trends most strongly correlate with adult e-cigarette use trends.
Finally, the search terms used in this work were sourced from the NYTS questionnaire, but this
selection may not be optimal and arbitrarily many other vaping-related search terms and
combinations of terms could also be investigated.
Acknowledgements
The author would like to acknowledge Dr Arielle Selya for proof-reading this article, and
Joe Gitchell for providing insightful comments and suggestions.
Declaration of interest
In 2020, Floe Foxon became a consultant to PinneyAssociates, Inc. PinneyAssociates
provides consulting services on tobacco harm reduction on an exclusive basis to Juul Labs, Inc.
In recent years, PinneyAssociates has consulted for British American Tobacco and Reynolds
American Inc and subsidiaries on tobacco harm reduction. Juul Labs, Inc. did not sponsor this
paper or participate in the design, study execution, data analysis, writing, or publication.
References
Ahmad, I., Flanagan, R., & Staller, K. (2020, Jul 3). Increased Internet Search Interest for GI Symptoms
May Predict COVID-19 Cases in U.S. Hotspots. Clin Gastroenterol Hepatol.
https://doi.org/10.1016/j.cgh.2020.06.058
American College of Cardiology. (2020). E3 Trial: E-Cigarettes Plus Counseling More Effective Than
Counseling Alone in Smoking Cessation https://www.acc.org/latest-in-
cardiology/articles/2020/03/24/16/41/mon-1045-e3-efficacy-safety-e-cigarettes-acc-2020
Page 10 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
California Department of Education. (2020). California Healthy Kids Survey. Retrieved October 4 from
https://www.cde.ca.gov/ls/he/at/chks.asp
Caputi, T. L., Leas, E., Dredze, M., Cohen, J. E., & Ayers, J. W. (2017). They're heating up: Internet search
query trends reveal significant public interest in heat-not-burn tobacco products. PLoS One,
12(10), e0185735. https://doi.org/10.1371/journal.pone.0185735
Carneiro, H. A., & Mylonakis, E. (2009, Nov 15). Google trends: a web-based tool for real-time
surveillance of disease outbreaks. Clin Infect Dis, 49(10), 1557-1564.
https://doi.org/10.1086/630200
Cavazos-Rehg, P. A., Krauss, M. J., Spitznagel, E. L., Lowery, A., Grucza, R. A., Chaloupka, F. J., & Bierut, L.
J. (2015, May). Monitoring of non-cigarette tobacco use using Google Trends. Tob Control, 24(3),
249-255. https://doi.org/10.1136/tobaccocontrol-2013-051276
Cervellin, G., Comelli, I., & Lippi, G. (2017, Sep). Is Google Trends a reliable tool for digital epidemiology?
Insights from different clinical settings. J Epidemiol Glob Health, 7(3), 185-189.
https://doi.org/10.1016/j.jegh.2017.06.001
Colorado Department of Public Health & Environment. (2019). Healthy Kids Colorado Survey. Retrieved
October 4 from https://www.colorado.gov/cdphe/hkcs
Dai, H., & Leventhal, A. M. (2019, Sep 16). Prevalence of e-Cigarette Use Among Adults in the United
States, 2014-2018. JAMA. https://doi.org/10.1001/jama.2019.15331
Gentzke, A. S., Creamer, M., Cullen, K. A., Ambrose, B. K., Willis, G., Jamal, A., & King, B. A. (2019, Feb
15). Vital Signs: Tobacco Product Use Among Middle and High School Students - United States,
2011-2018. MMWR Morb Mortal Wkly Rep, 68(6), 157-164.
https://doi.org/10.15585/mmwr.mm6806e1
Google. (2020). Google Trends. Retrieved October 3 from https://trends.google.com/trends/?geo=US
Hajek, P., Phillips-Waller, A., Przulj, D., Pesola, F., Myers Smith, K., Bisal, N., Li, J., Parrott, S., Sasieni, P.,
Dawkins, L., Ross, L., Goniewicz, M., Wu, Q., & McRobbie, H. J. (2019, Feb 14). A Randomized
Trial of E-Cigarettes versus Nicotine-Replacement Therapy. N Engl J Med, 380(7), 629-637.
https://doi.org/10.1056/NEJMoa1808779
Hebert, E. T., Case, K. R., Kelder, S. H., Delk, J., Perry, C. L., & Harrell, M. B. (2017, Sep). Exposure and
Engagement With Tobacco- and E-Cigarette-Related Social Media. J Adolesc Health, 61(3), 371-
377. https://doi.org/10.1016/j.jadohealth.2017.04.003
Page 11 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Jones, S., & Fox, S. (2009). Generations Online in 2009 (Pew Internet & American Life Project, Issue. P. R.
Center. http://pewrsr.ch/1iPqUn7
Merianos, A. L., Jandarov, R. A., Klein, J. D., & Mahabee-Gittens, E. M. (2019, Nov). Characteristics of
Daily E-Cigarette Use and Acquisition Means Among a National Sample of Adolescents. Am J
Health Promot, 33(8), 1115-1122. https://doi.org/10.1177/0890117119854051
University of Michigan. (2021). Monitoring the Future. Retrieved May 1 from
http://www.monitoringthefuture.org/
US Centers for Disease Control and Prevention. (2019a). National Youth Tobacco Survey (NYTS).
Retrieved October 3 from
https://www.cdc.gov/tobacco/data_statistics/surveys/nyts/data/index.html
US Centers for Disease Control and Prevention. (2019b). Youth Tobacco Survey (YTS). Retrieved October
3 from https://www.cdc.gov/tobacco/data_statistics/surveys/yts/index.htm
US Centers for Disease Control and Prevention. (2020). Youth Risk Behavior Surveillance System (YRBSS).
Retrieved October 4 from https://www.cdc.gov/healthyyouth/data/yrbs/index.htm
US Food and Drug Administration. (2019a). Tobacco 21. Retrieved October 5 from
https://www.fda.gov/tobacco-products/retail-sales-tobacco-products/tobacco-21
US Food and Drug Administration. (2019b, September 11). Trump Administration Combating Epidemic of
Youth E-Cigarette Use with Plan to Clear Market of Unauthorized, Non-Tobacco-Flavored E-
Cigarette Products https://www.fda.gov/news-events/press-announcements/trump-
administration-combating-epidemic-youth-e-cigarette-use-plan-clear-market-unauthorized-non
Wang, T. W., Neff, L. J., Park-Lee, E., Ren, C., Cullen, K. A., & King, B. A. (2020, Sep 18). E-cigarette Use
Among Middle and High School Students - United States, 2020. MMWR Morb Mortal Wkly Rep,
69(37), 1310-1312. https://doi.org/10.15585/mmwr.mm6937e1
Zhan, Y., Etter, J. F., Leischow, S., & Zeng, D. (2019, Jan 1). Electronic cigarette usage patterns: a case
study combining survey and social media data. J Am Med Inform Assoc, 26(1), 9-18.
https://doi.org/10.1093/jamia/ocy140
Page 12 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Trends in Vaping-Related Online Searches and Survey-Measured Prevalence of High School E-Cigarette Use.
The dotted and dashed lines show the change in normalized search interest over time for the vaping-related
Google web search terms ‘vape’, ‘vapes’, and ‘vape pen’. A normalized search interest value of 1.0
represents the peak popularity for that term. A value of 0.5 means the search term was half as popular than
it was at 1.0. The solid line shows the normalized prevalence of current (past 30 day) e-cigarette use among
US high school students from NYTS. A value of 1.0 means the peak prevalence of current e-cigarette use
among this cohort (27.5% in 2019). A value of 0.5 means current e-cigarette use was half as prevalent as
at 1.0.
Page 13 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review Only
Table 1. Zero-Normalized Cross-Correlations between NYTS Prevalence Estimates and
Vaping-Related Google Trends Search Interests
Cross-Correlation Coefficients
Number of
Months Search
Interest Lags
NYTS
Prevalence By
‘vapes’ Search
Term
‘vape’ Search Term
‘vape pen’ Search
Term
3
0.960*
0.931*
0.966*
2
0.967*
0.934*
0.970*
1
0.979*
0.938*
0.966*
0 (no lag)
0.972*
0.929*
0.958*
-1
0.967*
0.923*
0.945*
-2
0.969*
0.925*
0.940*
-3
0.963*
0.916*
0.915*
Bold font indicates maximum correlation
* p<0.001
Page 14 of 14
URL: http:/mc.manuscriptcentral.com/lsum eincert@gmail.com
Substance Use and Misuse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
