Supplementary Online Content - JAMA€¦ · Supplementary Online Content . Layton JB, Kim Y, Alexander GC, Emery SL. Association between direct-to-consumer advertising and testosterone
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Supplementary Online Content
Layton JB, Kim Y, Alexander GC, Emery SL. Association between direct-to-consumer advertising and testosterone testing and initiation in the United States, 2009-2013. JAMA. doi:10.1001/jama.2016.21041
eTable 1. Yearly Characteristics of the 75 Largest US Designated Market Areas (DMA) eFigure 1. Geographic Distribution of Cumulative Testosterone Advertisement Exposures From Nielsen Television Ratings by Year in the 75 Largest US Designated Market Areas (DMAs) eTable 2. QIC Goodness of Fit Estimates for Models Assessing Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation Among US Men Living in the 75 Largest Designated Market Areas eFigure 2. Mean Testosterone Testing and Initiation Rates and Interquartile Ranges Compared to Mean Model-Predicted Rates Among Adult Men in the 75 Largest Designated Market Areas in the US, January 2009 – December 2013 eTable 3. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation by Year Among US Men Living in the 75 Largest Designated Market Areas eTable 4. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation With Non-Linear Trends Among US Men Living in the 75 Largest Designated Market Areas, 2009-2013 eTable 5. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation Among US Men Living in the 75 Largest Designated Market Areas by Varying Lag Times eTable 6. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation Among US Men Living in the 75 Largest Designated Market Areas, Restricted to Years 2009-2012 and Compared to the Full Study Period
This supplementary material has been provided by the authors to give readers additional information about their work.
Mean testosterone testing rate, (SD) 18.6 (6.2) 18.3 (6.4) 22.7 (8.5) 29.4 (9.4) 30.0 (8.8) Mean testosterone initiation rate, (SD) 6.4 (3.4) 5.6 (2.6) 7.4 (3.20) 8.8 (4.4) 9.0 (4.2) Mean initiation without recent test, (SD)
3.0 (2.7) 2.1 (1.3) 2.5 (1.4) 2.5 (1.5) 2.7 (1.5)
Data represents information derived for 75 included DMAs assessed in January of each year; Age and race distributions from the US Census American Communities Survey; Income and physician density from the Health Resources and Services Administration’s Area Resource File; Enrollees, testosterone testing and initiation data from MarketScan Commercial Insurance Databases. Adult enrollees defined as adult men enrolled in the MarketScan database residing in the DMA on the 15th of the month with 6 months previous enrollment Rates calculated as number of testosterone tests or initiators per 10,000 men with at least 6 months baseline enrollment in the MarketScan database on the 15th of the month
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eFigure 1. Geographic Distribution of Cumulative Testosterone Advertisement Exposures From Nielsen Television Ratings by Year in the 75 Largest US Designated Market Areas (DMAs)
Exposure Range Number of DMAs Median Mean (SD) 3.5-4.3 9 3.9 3.9 (0.3) 4.5-4.9 12 4.7 4.7 (0.2) 5.0-5.4 11 5.2 5.2 (0.2) 5.5-5.7 10 5.6 5.6 (0.1) 5.8-6.1 10 6.0 5.9 (0.1) 6.3-6.6 10 6.4 6.4 (0.1) 6.7-8.3 13 7.2 7.3 (0.5)
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
Monthly advertisement exposures were summed across the entire year within DMAs, and DMAs were stratified into 7 strata of cumulative advertising exposures containing approximately equal number of DMAs. The borders between exposure strata were defined empirically, thus there are gaps between strata where no DMA had measures in that range. DMAs in white were not included in our analysis.
Source: Nielsen television ratings, 2008-2013
Note: no advertisements were aired in 2008
Exposure Range Number of DMAs Median Mean (SD) 62.9-70.6 11 70.0 68.3 (3.0) 70.9-73.7 10 72.4 72.3 (1.1) 73.8-75.6 11 74.6 74.7 (0.7) 75.7-76.4 10 76.2 76.1 (0.2) 76.6-78.7 10 78.0 77.7 (0.8) 79.0-80.9 12 80.0 80.0 (0.7) 81.0-64.9 11 82.4 82.5 (1.2)
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eTable 2. QIC Goodness of Fit Estimates for Models Assessing Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation Among US Men Living in the 75 Largest Designated Market Areas Outcome No independent
variables Covariates alone Covariates
and DTCA New testosterone tests -186,796.4889 -970,644.4297 -980,898.8325 New initiation -100,190.8113 -443,412.6609 -446,615.8535 New initiation without recent tests -80,296.2809 -140,416.0290 -140,831.5391 Poisson regression models with Generalized Estimating Equations adjusted for income, race/ethnicity, age, physician density, overall time trend, and seasonality. Source: Truven Health Analytics MarketScan databases, Nielsen television ratings, 2009-2013
Interpretation of QIC Goodness of Fit Statistics
The decreasing QIC with the addition of covariates and the primary testosterone variable does show better predictive ability of the model with the addition of our independent variables. Particularly, the decrease in QIC upon addition of DTCA to the covariates suggests improved model fit due to the DTCA variable.
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eFigure 2. Mean Testosterone Testing and Initiation Rates and Interquartile Ranges Compared to Mean Model-Predicted Rates Among Adult Men in the 75 largest Designated Market Areas in the US, January 2009 – December 2013
New A) total testosterone testing and B) initiation and C) initiation without recent serum testosterone testing
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eTable 3. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation by Year Among US Men Living in the 75 Largest Designated Market Areas
Outcome Year Association of one-unit increase in advertising
exposures with testosterone
Tests for comparison of yearly associations to 2013
RR (95% CI) RRR (95% CI) p-value New testing 2009 0.996 (0.976, 1.016) 1.005 (0.985, 1.026) 0.63 2010 1.010 (0.995, 1.024) 1.019 (1.004, 1.034) 0.01 2011 1.017 (1.014, 1.020) 1.026 (1.021, 1.031) <0.001 2012 1.008 (1.002, 1.015) 1.017 (1.009, 1.025) <0.001 2013 0.991 (0.987, 0.995) -- -- -- New initiation 2009 0.997 (0.967, 1.027) 1.003 (0.973, 1.034) 0.85 2010 1.025 (1.010, 1.041) 1.031 (1.015, 1.048) <0.001 2011 1.016 (1.011, 1.022) 1.022 (1.014, 1.030) <0.001 2012 1.004 (0.995, 1.013) 1.010 (0.999, 1.021) 0.06 2013 0.994 (0.989, 1.000) -- -- -- Initiation without recent tests 2009 0.955 (0.909, 1.003) 1.006 (0.957, 1.058) 0.81 2010 1.022 (0.992, 1.053) 1.033 (1.001, 1.066) 0.04 2011 1.016 (1.007, 1.026) 1.027 (1.014, 1.041) <0.001 2012 1.005 (0.992, 1.017) 1.016 (1.000, 1.032) 0.05 2013 0.989 (0.979, 0.998) -- -- -- The Rate Ratio (RR) and 95% Confidence Intervals (CI) were derived from Poisson GEE regression models adjusted for income, race/ethnicity, age, physician density, overall time trend, and seasonality. Testosterone advertising was included as a continuous variable, and the RRs estimate the association between a one-unit increase in ad exposure and the testosterone rate; for example, in 2010, one exposure to a testosterone advertisement was associated with a 2.5% higher subsequent initiation rate (RR=1.025, 95% CI: 1.010-1.041), corresponding to a mean absolute increase of 0.18 initiations (95% CI: 0.07-0.29) per 10,000 men per ad. 5 ad exposure to 5 ads in a month was associated with a 13.1% higher subsequent initiation rate (RR=1.131, 95% CI: 1.049-1.220), and a mean absolute increase of 0.94 initiations (95% CI: 0.35-0.1.57) per 10,000 men. The Ratio of Rate Ratios (RRR) compares the RR in the year of interest to the 2013 RR estimate, testing for any difference between the two year estimates Source: Truven Health Analytics MarketScan databases, Nielsen television ratings, 2009-2013
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eTable 4. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation With Non-Linear Trends Among US Men Living in the 75 Largest Designated Market Areas, 2009-2013
DTCA Term
New testing New initiation Initiation without tests
RR (95% CI) RR (95% CI) RR (95% CI) Linear 1.0164 (1.0122, 1.0207) 1.0153 (1.0101, 1.0206) 1.0168 (1.0077, 1.0259) Quadratic 0.9990 (0.9987, 0.9994) 0.9992 (0.9988, 0.9997) 0.9992 (0.9984, 0.9999) The Rate Ratio (RR) and 95% Confidence Intervals (CI) were derived from Poisson GEE regression models adjusted for income, race/ethnicity, age, physician density, overall time trend, and seasonality. Testosterone advertising was included as a continuous variable, and the RRs estimate the association between a one-unit increase in ad exposure and the testosterone rate. Source: Truven Health Analytics MarketScan databases, Nielsen television ratings, 2009-2013
Interpretation of Non-linear Trends
The positive rate ratio resulting from the linear term indicates that every one-unit increase in advertising exposure, (e.g. an increase from 1 to 2 advertisement views per month) is associated with increased measures of testosterone testing and initiation. However, the rate ratio below one resulting from the quadratic term indicates that with each one-unit increase in advertising exposure, the magnitude of the positive association is reduced. For example the association of a monthly increase in advertising views from 5 to 6 (RR=1.012, 95% CI: 1.009-1.014) is not as strong as a monthly increase from 1 to 2 advertisement exposures (RR=1.015, 95% CI: 1.012-1.019).
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eTable 5. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation Among US Men Living in the 75 Largest Designated Market Areas by Varying Lag Times Lag time New testing New initiation Initiation
without tests RR (95% CI) RR (95% CI) RR (95% CI) One month 1.000 (0.997, 1.002) 0.999 (0.996, 1.002) 1.003 (0.999, 1.007) Three months 0.998 (0.995, 1.000) 0.999 (0.996, 1.003) 0.998 (0.992, 1.004) The Rate Ratio (RR) and 95% Confidence Intervals (CI) were derived from Poisson GEE regression models adjusted for income, race/ethnicity, age, physician density, overall time trend, and seasonality. Testosterone advertising was included as a continuous variable, and the RRs estimate the association between a one-unit increase in ad exposure and the testosterone rate; for example, in the one month lag analysis, one exposure to a testosterone advertisement was associated with a 0.0% lower subsequent initiation rate (RR=0.999, 95% CI: 0.996-1.002), corresponding to a mean absolute increase of -0.01 initiations (95% CI: -0.03-0.01) per 10,000 men per ad. Source: Truven Health Analytics MarketScan databases, Nielsen television ratings, 2009-2013
Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 03/01/2021
eTable 6. Association of Total Monthly Testosterone Direct-to-Consumer Advertising With Testosterone Testing and Initiation Among US Men Living in the 75 Largest Designated Market Areas, Restricted to Years 2009-2012 and Compared to the Full Study Period
Outcome Association of one-unit exposure to advertising
with testosterone, 2009-2012
Association of one-unit exposure to advertising
with testosterone, 2009-2013
Comparison of the 2009-2012 estimate to the 2009-2013
The Rate Ratio (RR) and 95% Confidence Intervals (CI) were derived from Poisson GEE regression models adjusted for income, race/ethnicity, age, physician density, overall time trend, and seasonality. Testosterone advertising was included as a continuous variable, and the RRs estimate the association between a one-unit increase in ad exposure and the testosterone rate. The Ratio of Rate Ratios (RRR) compares the RR in the year of interest to the 2013 RR estimate, testing for any difference between the two year estimates Source: Truven Health Analytics MarketScan databases, Nielsen television ratings, 2009-2013