Does cyber dating abuse victimization predict next-day alcohol and cannabis use among college students?
Abstract
Background and Objectives
Cyber dating abuse (CDA) is prevalent on college campuses, with 43% of college students experiencing CDA each year. Yet, the potential impacts of CDA victimization on college students' health outcomes remain poorly understood. Informed by the self-medication hypothesis and longitudinal data linking dating abuse to substance use outcomes, the present study tested the hypothesis that CDA victimization positively associates with college students' next-day alcohol use (number of drinks consumed, odds of any drinking) and odds of cannabis use (yes/no). We also explored whether CDA's association with next-day substance use varied as a function of gender.
Methods
Daily data on alcohol use, cannabis use, and CDA victimization were collected across 60 consecutive days from 236 undergraduates who were in dating relationships (73.73% cisgender women).
Results
Multilevel modeling revealed gender differences. Among men, CDA victimization predicted next-day cannabis use (aOR = 7.34, p < .001), but negatively related to the number of drinks consumed the following day (B = −2.63, p < .001). Among women, CDA victimization did not relate to next-day cannabis or alcohol use. Regardless of gender, CDA victimization was unrelated to the odds of any drinking.
Discussion and Conclusions
College men are more likely to use cannabis the day after experiencing CDA victimization relative to other days. Targeting CDA and related aftermath may benefit college-based substance use intervention.
Scientific Significance
This is the first study to identify that cannabis use occurs subsequent to CDA victimization, which can inform college-based CDA and cannabis use prevention programming.
Intimate partner violence (IPV; i.e., psychological, physical, sexual, and cyber abuse) peaks between ages 18–25, a time in which many young adults are in college.1 The emergence and ubiquity of technology, particularly among young adults, provides new contexts through which individuals can monitor, harass, humiliate, and otherwise abuse their partners (i.e., cyber dating abuse [CDA]).2 CDA victimization is conceptually distinct from other forms of IPV and can include overt (e.g., sending a partner harassing or threatening messages) and covert behaviors (e.g., monitoring a partner's GPS location without their consent).3 CDA can be perpetrated 24/7 and does not require face-to-face contact, thereby posing a consistent threat to victims. Between 40% and 43% of undergraduates experience CDA victimization each year, with comparable prevalence across genders.4, 5 Yet, the potential impact of CDA on undergraduates’ health remains understudied. To advance this research, the present study leveraged data from a 60-day diary study to examine the association between CDA victimization and next-day alcohol and cannabis use among college students.
The field's understanding of CDA lags behind decades of research linking psychological, physical, and sexual IPV to alcohol and cannabis use. Longitudinal and event-level data informed by the self-medication hypothesis (i.e., using substances to alleviate painful affective states)6 revealed that IPV victimization is associated with greater odds of subsequent cannabis7 and alcohol use.8, 9 By contrast, investigators only recently started examining associations between CDA victimization and substance use. Cross-sectional data revealed that CDA victimization is positively associated with problem drinking, alcohol use, and cannabis use among young adults.10, 11 Longitudinal data examining CDA victimization in relation to substance use is scant and limited to adolescent populations. For instance, adolescents' CDA victimization was positively associated with alcohol and cannabis use cross-sectionally, and with cannabis use (but not alcohol use) 1 year later.12 While informative, these limited data preclude conclusions regarding CDA victimization's prospective and proximal association with substance use among college students. No study examined whether experiencing CDA victimization associated with greater likelihood of next-day alcohol or cannabis use, thereby obfuscating the potential acute impacts of CDA on undergraduates.
While there are many potential mechanisms that could explain why CDA victimization may precede and increase the risk of substance use (e.g., coping with the emotional aftermath of CDA,17 seeking out socialization experiences to enhance positive affect following CDA14-16), it is necessary to first determine whether a proximal association between CDA and substance use exists. Data collected from a nationally representative sample of men and women revealed that, of all substances, alcohol and cannabis had the most robust associations with IPV victimization.13 If undergraduates are more likely to use substances in the aftermath of CDA, campus-based prevention and intervention programs that target IPV and/or substance use may be adapted to account for CDA-related substance use (e.g., psychoeducation on the impact of CDA).
Purpose and hypotheses
- 1.
CDA victimization would positively relate to next-day alcohol use (i.e., the number of alcoholic drinks consumed and odds of drinking).
- 2.
CDA victimization would positively relate to odds of next-day cannabis use.
No data exist to determine whether associations between CDA victimization and next-day alcohol/cannabis use vary across genders. As such, an exploratory aim of the present study was to examine whether the association between CDA victimization and next-day alcohol and cannabis use varies across men and women.
METHOD
Participants
Undergraduates (N = 236) from a large, public, southeastern university participated in the study. Cannabis is not legal in the state in which the study was conducted. The present study's sample was part of a large study on alcohol use and CDA.18 To be eligible, participants had to (1) be at least 18 years old, (2) be in a dating relationship for at least 1 month, (3) have a dating partner who was at least 18 years old, (4) have consumed alcohol at least once in the past month, and (5) have at least two face-to-face or technology-facilitated contact days with their partner each week. Participant demographic characteristics are presented in Table 1. Most participants identified as cisgender women (73.73%), white (83.89%), and heterosexual (78.81%). Participants' average age was 20.53 (SD = 3.31) years. The average relationship length was 18.83 (SD = 18.04; Mdn = 12.00) months.
| Characteristics | % |
|---|---|
| Sex assigned at birth | |
| Female | 68.22 |
| Male | 23.30 |
| Chose not to disclose | 8.47 |
| Gender | |
| Cisgender woman | 73.73 |
| Cisgender man | 25.42 |
| Other | 0.42 |
| Chose not to disclose | 0.42 |
| Race | |
| White | 83.89 |
| Black and/or African American | 5.51 |
| Hispanic or Latina/o | 4.24 |
| Asian | 2.97 |
| Other | 3.38 |
| Multiracial | 1.69 |
| American Indian/Native American or Alaska Native | 0.42 |
| Indian/Middle Eastern | 0.42 |
| Chose not to disclose | 0.85 |
| Sexual orientation | |
| Straight or heterosexual | 78.81 |
| Bisexual | 13.56 |
| Gay or Lesbian | 2.54 |
| Queer | 1.27 |
| Asexual | 0.42 |
| Chose not to disclose | 3.39 |
| Relationship status | |
| Dating | 92.80 |
| Engaged | 4.24 |
| Married | 2.97 |
| Characteristics | M (SD) |
|---|---|
| Age (years) | 20.53 (3.31) |
| Average relationship length (months) | 18.83 (18.04) |
Procedure
Participants were recruited via advertisements posted in key areas (e.g., flyers, campus listservs, and a university research participation website). Students interested in participating completed a brief eligibility screener via Qualtrics.com. Eligible participants received an email inviting them to participate in the study. Participants who consented to participate completed a baseline assessment on Qualtrics.com, for which they had the option to receive partial course credit or $20.00 monetary compensation.
The day following the baseline assessment, participants were emailed a link to their daily survey. Each daily survey was emailed to participants at 6:00 AM for 60 consecutive days to inquire about each participant's experiences on the previous day. Participants received reminder emails at 12:00 PM and 5:00 PM. Participants were instructed to answer all daily surveys regarding their baseline partner, regardless of whether participants ended their relationships because aggression often continues after relationships end.19 Participants earned $1.00 for each daily survey completed and $5.00 bonuses each week they completed all seven surveys for a total possible compensation of $120.00. Study procedures were approved by the Institutional Review Board of the University of Tennessee.
Measures
Baseline assessments
Demographics
A demographics questionnaire assessed participants' age, gender identity, sexual identity, relationship status/length, and racial/ethnic identity.
Alcohol use and problems
The 10-item Alcohol Use Disorders Identification Test (AUDIT)20, 21 assessed participants' self-reported past-year intensity and frequency of alcohol use, negative consequences of alcohol use, and symptoms of alcohol tolerance and dependence. Scores were summed, with higher scores representing greater alcohol use and problems. The AUDIT evidenced adequate reliability within the present sample (α = .78).
Drug use and problems
The 14-item Drug Use Disorders Identification Test (DUDIT)22 assessed participants’ self-reported past-year intensity and frequency of illicit drug use, symptoms of drug tolerance and dependence, and negative consequences of drug use (i.e., cannabis, cocaine, hallucinogens, stimulants, sedatives/hypnotics/anxiolytics, opiates, and inhalant use). The DUDIT demonstrated adequate reliability within the present sample (α = .77).
Past-year CDA victimization
The 18-item Controlling Partner Inventory (CPI)34 victimization subscale measured past-year CDA victimization (e.g., a partner checking and monitoring their behaviors, harassing and threatening them, and using cameras/spyware to keep tabs on them). Item responses ranged from 0 (never) to 4 (4 or more times) and were summed such that higher scores represent more frequent victimization. The CPI demonstrated good internal reliability within the present sample (α = .80).
Daily measures
Daily CDA victimization
Participants were asked if they experienced CDA victimization during the previous day using six items adapted from validated CDA measures (i.e., the Controlling Partner Inventory23; the Electronic Aggression Scale24; the Facebook Mate-Retention Tactics Inventory25). Items asked participants if their partner used any of several CDA behaviors the prior day. Each of the six items assessed several CDA behaviors including abuse (e.g., “sent mean, hurtful, or threatening emails, text messages, or online messages [e.g., on a social networking site or blog]”), harassment (e.g., “sent excessive number of text messages, online messages, emails, or phone calls to partner”), humiliation (e.g., “posted embarrassing, insulting, or inappropriate material [e.g., posts, photos, or videos] about a partner online [e.g., on a social networking site or blog]”), monitoring (e.g., “snooped through partner's sent/received call histories, text messages, or online messages”), coercing (e.g., “told partner to block, unfriend, or unfollow someone on social media”), and threatening (e.g., “threatened partner that something bad would happen if they interacted with a certain person online”). All items included gender-neutral wording. Days on which CDA victimization occurred were coded “1”, and days on which CDA victimization did not occur were coded “0.” Daily CDA was lagged to create a variable that represented next-day CDA victimization.
Next-day alcohol use
Each day, participants were asked how many standard alcoholic drinks they consumed. Participants were provided with a definition of a standard drink.26 For the present study, “number of drinks” refers to the total number of standard alcoholic drinks an individual consumed on a given day (i.e., a continuous value). We also computed an “any drinking” variable that was coded “0” if no alcohol use occurred, and “1” if any drinking occurred on a given day (i.e., a dichotomous variable). Daily alcohol use variables were lagged to create variables that represented next-day alcohol use.
Next-day cannabis use
Each day, participants were asked if they used cannabis during the previous day. Responses were coded such that “1” represents cannabis use and “0” represents no cannabis use on a given day. The daily cannabis use variable was lagged to create a next-day cannabis use variable.
Relationship status
Each day, participants were asked if they were still dating the person they were dating at the beginning of the study.
Data analytic strategy
Descriptive analyses were completed using SPSS Version 29.0. Hypotheses were tested with multilevel modeling using HLM Version 8.0. Each alcohol use outcome variable (i.e., number of drinks consumed, any drinking) and the cannabis use were examined in separate models. We used multilevel modeling with fixed slopes specifying a logit link function and full maximum likelihood estimation for analyses due to the multilevel nature of these models (i.e., individuals' repeated reports). A Bernoulli sampling distribution was used for the cannabis use and any drinking models, given the dichotomous nature of these outcomes. All participants and available data were used in analyses because the multilevel modeling procedure estimates model parameters according to Bayesian rules.
Two-level models were employed. We tested hypotheses by estimating the within-person effects of daily CDA victimization (uncentered, Level 1 variable), gender (uncentered, Level 2 variable; women = 1, men = 0), and their cross-level interaction on repeated measures of next-day alcohol (or cannabis) use controlling for (1) baseline CDA victimization, (2) next-day CDA victimization (uncentered, Level 1 variable) and (3) AUDIT scores (grand-mean centered, Level 2 variable in the alcohol outcome models) or DUDIT scores (grand-mean centered, Level 2 variable included in the cannabis outcome model). This approach allowed us to better estimate the within-person effects of CDA victimization on next-day alcohol and cannabis use controlling for past-year CDA victimization, and baseline alcohol and drug use and problems, respectively. Similarly, controlling for the proportion of variance in the outcomes attributed to CDA victimization that occurred the next day (i.e., when the outcomes of interest occurred) yielded more conservative estimates of the relation between CDA victimization and next-day alcohol/cannabis use. Only one participant identified as nonbinary. Thus, gender differences were tested by comparing participants whose gender identities were women and men. If significant interactions emerged, the interaction was explicated by testing CDA victimization's association with the outcome variable at each level of gender. If the interaction term was not significant, it was removed from the model to interpret the main effects.
RESULTS
Descriptive statistics
Compliance
Participants completed 71.67% of all surveys sent. On average, each participant completed 42.57 daily surveys (SD = 20.09). The number of daily surveys completed was unrelated to participants' baseline AUDIT scores (r = −0.11, p = .12), DUDIT scores (r = −0.09. p = .20), number of days on which an individual used cannabis during the study (r = 0.10, p = .13), or the number of days a participant experienced CDA victimization (r = 0.08, p = .25). Number of surveys completed negatively correlated with the percentage of completed daily surveys that included cannabis use (r = −0.17, p = .01), the number of drinking days an individual had throughout the study (r = 0.40, p < .001), and the average number of drinks an individual consumed on a drinking day (r = 0.17, p = .01). Men (M = 40.80, SD = 20.11) and women (M = 43.09, SD = 20.12) completed a comparable number of surveys, t(233) = 1.05, p = .29.
Descriptive statistics
Of the completed daily surveys (N = 10,149), there were 78 CDA victimization events that occurred on 72 days across 42 participants (i.e., 17.79% of all participants experienced CDA victimization). The total number of days on which CDA victimization was experienced by women (M = 0.36, SD = 1.2) and men (M = 0.18, SD = 0.48) was comparable, t(229.27) = 1.50, p = .13. There were 14 days of CDA victimization among 12 men (range = 1–3 days), and 58 days of CDA victimization among 29 women (range = 1–11 days). A minority (14.41%) of participants broke up at least once during the study, and CDA continued among 5.88% of participants who broke up.
There were 518 days of cannabis use across 72 participants (i.e., 30.51% of participants used cannabis) and 1497 days of alcohol use across 197 participants (i.e., 83.47% of participants used alcohol). Across all surveys, 168 days included both alcohol and cannabis use (i.e., 11.36% of all alcohol use days included cannabis use; 48.0% of all cannabis use days also included alcohol use). There were no significant differences between women (M = 2.55, SD = 7.66) and men (M = 1.45, SD = 3.63) in terms of the number of cannabis use days throughout the study. Women (M = 6.74, SD = 8.28) and men (M = 5.60, SD = 8.00) had a comparable number of drinking days throughout the study, t(233) = 0.99, p = .32, though men consumed more drinks on a drinking day (M = 3.54, SD = 3.51) than did women (M = 2.48, SD = 1.78), t(92.35) = −2.48, p = .01. Women and men had comparable AUDIT (Mwomen = 6.60, SD = 4.32; Mmen = 6.83, SD = 4.66; t[207] = −0.32, p = .75), DUDIT (Mwomen = 2.14, SD = 3.23; Mmen = 1.70, SD = 2.70; t[210] = 0.89, p = .33), and CPI scores (Mwomen = 5.44, SD = 7.16; Mmen = 5.65, SD = 7.63; t[213] = 0.19, p = .85). Most (61.4%) participants reported experiencing at least one CDA victimization event in the past year.
Hypothesis testing
Cannabis use model
See Table 2 for final model parameters. The association between CDA victimization and odds of next-day cannabis use varied as a function of gender. For men, CDA victimization positively associated with odds of next-day cannabis use (aOR = 7.34, p < .001) whereas for women, CDA victimization was unrelated to odds of next-day cannabis use (aOR = 0.78, p = .63). DUDIT scores positively related to next-day cannabis use odds (aORmen = 0.36, p < .001; aORwomen = 1.42, p < .001) whereas next-day CDA victimization was unrelated to next-day cannabis use odds (aORmen = −0.04, p = .92; aORwomen = 0.97, p = .95). Past-year CDA victimization was unrelated to cannabis use.
| Odds of cannabis use | |||||
|---|---|---|---|---|---|
| B | SE | OR | 95% CI | p | |
| Level 2 (between-person effects) | |||||
| Gender | 0.34 | 0.24 | 1.40 | (0.88, 2.23) | .16 |
| DUDIT | 0.36 | 0.09 | 1.43 | (1.19, 1.71) | <.001 |
| Past-year CDA victimization | 0.02 | 0.01 | 1.02 | (1.00, 1.05) | .05 |
| Level 1 (within-person effects) | |||||
| Next-day CDA victimization | −0.04 | 0.43 | 0.96 | (0.41, 2.22) | .92 |
| CDA victimization | 1.99 | 0.46 | 7.34 | (2.99, 18.01) | <.001 |
| Gender x CDA victimization | −2.24 | 0.69 | 0.11 | (0.03, 0.41) | .00 |
| Odds of alcohol use | |||||
|---|---|---|---|---|---|
| B | SE | OR | 95% CI | p | |
| Level 2 (between-person effects) | |||||
| Gender | −0.23 | 0.17 | 0.79 | (0.57, 1.10) | .17 |
| AUDIT | 0.09 | 0.02 | 1.09 | (1.05, 1.14) | <.001 |
| Past-year CDA victimization | −0.02 | 0.01 | 0.98 | (0.97, 1.00) | .08 |
| Level 1 (within-person effects) | |||||
| Next-day CDA victimization | 0.39 | 0.42 | 1.47 | (0.65, 3.77) | .35 |
| CDA victimization | −0.31 | 0.35 | 0.74 | (0.37, 1.46) | .38 |
- Note: DUDIT = Drug Use Disorders Identification Test total scores for past-year drug use and problems. CDA = cyber dating abuse. AUDIT = Alcohol Disorders Identification Test total scores for past-year alcohol use and problems. For the alcohol use model, the interaction term was not significant (i.e., p > .05), so the main effects model is shown. Bolded values indicate statistical significance (i.e., p < .05).
Alcohol use model
See Table 2 for final model parameters. The association between CDA victimization and next-day alcohol use (i.e., odds of any alcohol use) did not vary as a function of gender. Examining the main effects model revealed that CDA victimization was unrelated to next-day alcohol use (aOR = −0.31, p = .38). AUDIT scores positively associated with next-day alcohol use odds (aOR = 0.09, p < .001), whereas next-day CDA victimization was unrelated to next-day alcohol use odds (aOR = 0.39, p = .35). Past-year CDA victimization was unrelated to alcohol use.
Number of alcoholic drinks model
See Table 3 for final model parameters. The association between CDA victimization and the number of drinks consumed the next day varied as a function of gender. For women, CDA victimization was unrelated to the number of drinks consumed the following day (B = 0.33, p = .56). For men, CDA victimization negatively associated with the number of drinks consumed the following day (B = −2.63, p < .001). AUDIT scores were positively associated with number of drinks consumed the following day (B = 0.20, p < .001), whereas next-day CDA victimization was unrelated to the number of drinks consumed the following day (B = 1.09, p = .07). Past-year CDA victimization was unrelated to number of drinks.
| B | SE | p | |
|---|---|---|---|
| Level 2 (between-person effect) | |||
| Gender | −1.26 | 0.59 | .03 |
| AUDIT | 0.20 | 0.03 | <.001 |
| Past-year CDA victimization | −0.01 | 0.60 | .75 |
| Level 1 (within-person effect) | |||
| Next-day CDA victimization | 1.10 | 0.60 | .07 |
| CDA victimization | −2.63 | 0.02 | <.001 |
| Gender × CDA victimization | 2.96 | 0.57 | <.001 |
- Note: AUDIT = Alcohol Disorders Identification Test total scores for past-year alcohol use and problems. CDA = cyber dating abuse. Bolded values indicate statistical significance (i.e., p < .05).
DISCUSSION
This study examined whether undergraduates' CDA victimization associated with next-day alcohol and cannabis use. In partial support of hypotheses, college men, regardless of their prior substance use patterns, were more likely to use cannabis following a day on which they experienced CDA victimization relative to a day on which they did not experience CDA victimization. Unexpectedly, men consumed fewer drinks than they usually do on the days after they experienced CDA victimization. Women's next-day alcohol and cannabis use remained unchanged regardless of whether they experienced CDA victimization on a previous day. These results are the first to identify a proximal, prospective association between CDA victimization and cannabis use.
Our results suggest that college men used cannabis in the aftermath of CDA victimization, which is consistent with prior research limited to college women.7 While we did not assess motives for using substances on a given day, it is possible that men used cannabis to cope with CDA, given that college students often cite stress reduction and relaxation as reasons for using cannabis.27 Alternatively, it is possible that men in our study used cannabis the day after CDA victimization for reasons unrelated to coping (e.g., using cannabis to bond with a partner or friend). Additional research is needed to understand event-level cannabis use motives in the context of CDA victimization, particularly among men.
Unexpectedly, men drank fewer drinks than they normally do the day after CDA victimization. Examining the content of CDA experiences may provide clarity in future studies. For instance, partners may have criticized men's alcohol use via text/harassing phone calls; as a result, men may have drunk less the following day. The present study did not assess men's emotional reactions to CDA concurrently or over time, which has implications for men's use of potentially harmful coping strategies (e.g., substance use) and abuse-related help-seeking behavior.28, 29 This line of research would benefit from assessing men's emotional and behavioral reactions to CDA to elucidate mechanisms linking CDA victimization to subsequent drinking.
That CDA victimization did not associate with women's next-day cannabis use is inconsistent with the finding7 that college women's physical and sexual IPV victimization positively related to their odds of next-day cannabis use. However, Shorey et al.7 also found that college women's IPV victimization was unrelated to their next-day alcohol use. It may be that college women's CDA is a less salient antecedent to their substance use than in-person IPV, though additional research is needed to evaluate this supposition. It is also plausible that women did not experience levels of CDA victimization necessary to potentiate substance use above and beyond women's typical motivations for alcohol/cannabis use. Importantly, the present results do not imply that college women experiencing CDA victimization are not adversely impacted by such experiences. By contrast, numerous studies linked women's CDA victimization to mental health concerns (e.g., posttraumatic stress disorder and depression symptoms)30, 31 and fear of their partner above and beyond in-person IPV victimization.33 It is plausible that the impacts of CDA victimization accumulate over time, as is seen with other forms of IPV.9, 13, 29, 32 It is also plausible that CDA victimization has more immediate, same-day effects rather than next-day effects, which future research should explore.
LIMITATIONS
Results are based on a sample of white, heterosexual, cisgender college students, most of whom were women (73.73%). Equal distributions of men and women may result in greater power to detect gender differences. Results may not generalize to more racially- and gender-diverse populations, undergraduates who are more vulnerable to IPV (e.g., those who drink heavier), or those who experience frequent and repeated IPV. Although most participants reported experiencing past-year CDA victimization, CDA victimization was infrequent across the 60-day period. Some CDA types were not investigated in this study (e.g., sexual CDA). We focused on substance use subsequent to CDA victimization; however, IPV is often bidirectional, which underscores the need to examine substance use in the aftermath of CDA perpetration. Our data cannot determine causality, and recall of behaviors may be impacted by heavier drinking and cannabis use. We did not estimate the daily quantity of cannabis used given psychometric concerns,33 which limits interpretability. Lastly, we did not assess substance use motives or emotional distress during data collection, which limits the interpretability of results.
RESEARCH AND CLINICAL IMPLICATIONS
Future event-level research designed to capture emotional reactions and potential motivations for substance use (or non-use) in the aftermath of CDA would inform intervention efforts. Little is known about the proximal consequences of CDA victimization within and between subpopulations of undergraduates. Future investigations of CDA victimization and prospective impacts should consider designs that capture day-level associations and cumulative effects of CDA victimization on substance use over time (e.g., longitudinal EMA burst designs). Characterizing the CDA type, frequency, and severity would clarify whether CDA characteristics are differentially associated with substance use.
Clinicians working with undergraduates who experienced CDA should screen for substance use, which may point to potential points of intervention and support. College campuses should offer services specifically tailored to men's victimization experiences and related substance use. Notably, no college-based interventions demonstrated efficacy in reducing IPV victimization and substance use concurrently. Despite gender differences that emerged in our results, we caution clinicians against neglecting the possibility of substance use occurring in the aftermath of CDA across genders. As noted by others,13, 32 the impact of IPV may accumulate over years to potentiate greater substance use and related problems. Thus, substance use may not be a proximal correlate of women's CDA victimization, but helping women adaptively cope may curtail future substance use.
ACKNOWLEDGMENTS
This work was supported by a Visionary Grant from the American Psychological Foundation (APF) and by grant F31AA026489 from the National Institute on Alcohol Abuse and Alcoholism (NIAAA) awarded to the first author while at the University of Tennessee, Knoxville, TN. The content is solely the responsibility of the authors and does not necessarily represent the official views of the APF, NIAAA, or the National Institutes of Health.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
