Data were collected from a public health survey distributed in 2016 to pupils in the 9th grade of primary school and 2nd grade of secondary school. The survey distribution covered all 33 municipalities in Skåne, a region in southern Sweden. The overall response rate was 77% (9143/11,868) among 9th grade pupils and 73.4% (7949/10,832) among 2nd grade pupils, resulting in a total of 17,092 responses. The main purpose of the survey was to investigate health and various social factors among Swedish adolescents. Previous school surveys in Skåne were primarily focused on alcohol, drug, and tobacco use. In contrast, the public health school survey of 2016 included a broad spectrum of public health questions regarding demographics and family characteristics (section A); general self-perceived health (section B); accidents and injuries (section C); leisure-time activities and habits (section D); dietary habits (section E); alcohol (section F); tobacco smoking and snuff use (section G); narcotic drugs (section H); sex and life together (section I); school context (section J); security and exposure (section K); gambling (section L); and general health, life satisfaction, and beliefs concerning the future (section M).

The survey was provided by the regional council of Scania County (Region Skåne) in cooperation with the municipal association of Skåne, and it was answered anonymously in classroom settings. Participation was voluntary; all questions were described as optional, and all measures were based on self-reports (see Multimedia Appendix 1).

The R statistical programming language (version 4.0.4) [20], along with several functions from the tidyverse package [21], was used for intermediate data processing and statistical analysis. We opted for a fully Bayesian approach, and all Bayesian models were specified using the R package brms [22]. The brms package interfaces R with the Stan probabilistic programming language [23], which is a state-of-the-art language for specifying and estimating Bayesian models. Bayesian binomial regression models were used to examine whether the frequency of the associated factors outlined above differed between adolescents reporting excessive smartphone use and those who did not. All models used weakly informative priors centered around zero, which should provide moderate regularization while still having minimal impact on obtained estimates [24]. Finally, the R package emmeans [25] was used for postprocessing results.

We present group differences as estimated median absolute percentage differences along with the associated odds ratio (OR), reported with 95% highest density intervals (HDIs). In contrast to a frequentist CI, the 95% HDI may be interpreted such that it has a 95% probability of actually containing the values inside it [26]. Furthermore, the region of practical equivalence (ROPE) approach was used to determine whether an estimated difference was of practical and/or clinical importance [26]. Specifically, we considered an estimated difference of at least 5% (in either direction) as the minimal difference for “practical equivalence.” If the 95% HDI was not beyond this cutoff value, we deemed the results as uncertain in terms of practical and clinical importance.

Information about gender was missing for 86 respondents, bringing the total sample size available for group-based analysis to 17,006. Furthermore, there were varying levels of missing data for smartphone use as well as for the associated factors, as indicated in the tables below. Excessive smartphone use was more prevalent among girls (approximately 60%) than among boys (approximately 35%) in both grades (see Table 1 for details). Although results varied across gender and grade in terms of robustness and size of the estimated differences, overall, we found that excessive smartphone use was associated with a higher frequency of multiple suspected associated factors such as ever having tried smoking, alcohol, and other substances; poor sleep; and often feeling low and often feeling anxious. Several of these findings were both robust, with differences exceeding the ROPE with 95% probability by a large margin, and substantial, with some estimated differences reaching as high as 15%. In addition, for several other variables where the differences, with 95% probability, did not exceed the ROPE, the differences nonetheless robustly exceeded zero.

A total of 33.6% (499/1484) of the boys in the 9th grade of primary school who reported excessive smartphone use were categorized as self-reporting poor sleep compared to 25.1% (674/2682) of those who did not report excessive smartphone use, with an estimated difference of 8.5% (95% HDI 6.1%, 10.9%) and an associated OR of 1.51 (95% HDI 1.33, 1.68). Furthermore, participants who reported excessive smartphone use had a higher frequency of having tried smoking (575/1435, 40.1%) and alcohol (939/1453, 64.6%) than those who did not report excessive smartphone use (smoking: 685/2609, 26.3%; alcohol: 1378/2646, 52.1%), with an estimated difference of 13.8% (95% HDI 11.3%, 16.5%) and OR of 1.88 (95% HDI 1.66, 2.1) for smoking and an estimated difference of 12.5% (95% HDI 9.9%, 15.1%) and OR of 1.68 (95% HDI 1.5, 1.87) for alcohol use. Furthermore, boys who reported excessive smartphone use had higher frequencies of often feeling low, often feeling anxious, ASD, and having tried other substances, as well as a lower frequency of being satisfied with their own health, although these differences did not reliably exceed the ROPE.

In summary, excessive smartphone use among boys in the 9th grade of primary school was robustly associated with a higher frequency of poor sleep and having tried smoking and alcohol. Details are presented in Table 2 and Figure 1.

Of the girls who reported excessive smartphone use, 27.4% (678/2475) reported often feeling low and 22.9% (565/2469) reported often feeling anxious, as compared to 18% (303/1684) and 14% (236/1682), respectively, of the girls who did not report excessive smartphone use. The estimated difference and OR for often feeling low were 9.4% (95% HDI 7.2%, 11.5%) and 1.72 (95% HDI 1.5, 1.94), respectively, and the corresponding values for often feeling anxious were 8.9% (95% HDI 6.9%, 10.9%) and 1.82 (95% HDI 1.57, 2.08), respectively. In addition, 41.9% (1052/2509) of those reporting excessive smartphone use were classified as having poor sleep, compared to 27.8% (473/1702) of those not who did not report excessive smartphone use, with an estimated difference of 14.1% (95% HDI 11.7%, 16.5%) and an associated OR of 1.88 (95% HDI 1.67, 2.08).

Girls who reported excessive smartphone use also reported higher frequencies of having tried smoking (985/2478, 39.7%) and alcohol (1568/2494, 62.9%) compared to those not who did not (381/1689, 22.6% for smoking and 768/1703, 45.1%, for alcohol use), with an estimated difference of 17.2% (95% HDI 14.9%, 19.6%) and OR of 2.27 (95% HDI 2, 2.53) for smoking, and an estimated difference of 17.8% (95% HDI 15.2%, 20.3%) and OR of 2.06 (95% HDI 1.85, 2.29) for alcohol use. Moreover, girls with excessive smartphone use had lower frequencies of being satisfied with their own health as well as loneliness, and a higher frequency of having tried other substances, but these differences did not reliably exceed the ROPE.

In summary, excessive smartphone use among girls in the 9th grade of primary school was robustly associated with a higher frequency of often feeling low, often feeling anxious, poor sleep, and having tried smoking and alcohol. Details are presented in Table 3 and Figure 1.

Boys who reported excessive smartphone use had higher frequencies of poor sleep (636/1336, 47.6% vs 858/2253, 38.1%), having tried smoking (851/1292, 65.9% vs 1120/2195, 51%), and having tried other substances (280/1267, 22.1% vs 300/2181, 13.8%) compared to those who did not report excessive smartphone use, with estimated differences of 9.5% (95% HDI 6.7%, 12.4%) and OR 1.48 (95% HDI 1.31, 1.65) for poor sleep, 14.9% (95% HDI 12.1%, 17.7%) and OR 1.85 (95% HDI 1.64, 2.08) for having tried smoking, and 8.3% (95% HDI 6.1%, 10.6%) and OR 1.78 (95% HDI 1.52, 2.06) for having tried other substances. Boys who reported excessive smartphone use also had higher frequencies of often feeling low, often feeling anxious, ASD, and having tried alcohol, although these differences did not reliably exceed the ROPE.

In summary, excessive smartphone use among boys in the 2nd grade of secondary school was robustly associated with a higher frequency of poor sleep and having tried smoking and other substances. Details are presented in Table 4 and Figure 1.

Girls who reported excessive smartphone use had had higher frequencies of often feeling low (702/2198, 31.9% vs 332/1499, 22.1%) and often feeling anxious (560/2211, 25.3% vs 269/1493, 18%), with an estimated difference of 9.8% (95% HDI 7.3%, 12.1%) and OR of 1.65 (95% HDI 1.44, 1.86) for often feeling low and an estimated difference of 7.3% (95% HDI 5.1%, 9.5%) and OR of 1.55 (95% HDI 1.34, 1.76) often feeling anxious. In addition, 48.6% (1081/2224) of girls who reported excessive smartphone use were classified as having poor sleep, compared to 38.7% (584/1508) of those who did not report, with an estimated difference of 9.9% (95% HDI 7.3%, 12.7%) and an associated OR of 1.5 (95% HDI 1.33, 1.67).

Furthermore, girls who reported excessive smartphone use had higher frequencies of having tried smoking (1299/2181, 59.6% vs 640/1492, 42.9%) and alcohol (1903/2201, 86.5% vs 1136/1498, 75.8%), with an estimated difference of 16.7% (95% HDI 14%, 19.4%) and OR of 1.96 (95% HDI 1.74, 2.18) for smoking and an estimated difference of 10.6% (95% HDI 8.5%, 12.8%) and OR of 2.04 (95% HDI 1.75, 2.33) for alcohol. Finally, although the differences did not reliably exceed the ROPE, girls who reported excessive smartphone use had a relatively higher frequency of having tried other substances, as well as lower frequencies of ADHD, ASD, being satisfied with one’s own health, and loneliness.

In summary, excessive smartphone use among girls in the 2nd grade of secondary school was robustly associated with a higher frequency of often feeling low, often feeling anxious, poor sleep, and having tried smoking and alcohol. Details are presented in Table 5 and Figure 1.

Using a large and representative sample of Swedish adolescent pupils, we found that excessive smartphone use was more prevalent among girls (approximately 60% of all respondents) than among boys (approximately 35% of all respondents). Furthermore, excessive smartphone use was robustly associated with a substantially higher prevalence of poor sleep and, with slight differences between grades and gender, with higher frequencies of having tried smoking, alcohol, and other substances. Among girls, both in the 9th grade of primary school and 2nd grade of secondary school, we found that excessive smartphone use was robustly associated with a higher frequency of often feeling low and feeling anxious. Several other factors differed reliably from zero between the groups, although these differences did not, with 95% probability, exceed the ROPE. Our study adds to the knowledge of excessive smartphone use by investigating the corresponding male and female characteristics and possible associated factors among adolescents in an ordinary Swedish school setting.

Excessive smartphone users of both male and female genders in the 9th grade showed a disproportionate high prevalence of having used cigarettes and alcohol. A similar observation was made for smartphone users of the 2nd grade of secondary school, but in this grade, boys also had a higher probability of experience with illicit drugs. Similar results can be found in the literature; for example, Marmet et al [12] investigated the coexistence of behavioral and substance addiction among adult men and found that individuals with smartphone addiction were more likely to also be addicted to alcohol, tobacco, and illicit drugs. Behavioral and substance addiction have previously been reported as heavily related, and a sharing of a common personality trait has been hypothesized. Our findings warrant for additional research on excessive smartphone use in adolescents in order to implement prevention plans to hinder the development of other forms of addiction [3,27].

The relationship between ADHD and excessive smartphone use has been previously established [1,2,28,29]. The mechanism is thought to act through the lack of social interactions with others, a key characteristic in patients with ADHD, who concordantly feel a stronger need to be assured by and connected to others. Another suggested mechanism is the tendency to be easily bored typically exhibited by individuals with ADHD, resulting in a search for constant stimulation [30-32]. Children with ASD spend significantly more time using screen-based media than any leisure activity, and the correlation between internet addiction and ASD has already been established [15,16,33-37]. In ASD, the mechanism of internet overuse is considered to be due to their autistic traits: restricted, repetitive patterns of behavior, interests, or activities [38]. Some studies prove that children with ASD can learn via smartphone use, especially when the content is responsive to their interests, which makes smartphone use a valuable experience.

In this study, we found a relationship between ADHD and excessive use of smartphone with the strongest probability among boys in the 2nd grade of high school, but it did not exceed the ROPE in any of the groups. We also found increased probability of excessive smartphone use in individuals with ASD, which was the strongest among boys in the 9th grade of primary school but the probability did not exceed the ROPE in any of the groups. One possible explanation is that ADHD and ASD were self-reported; even though the questionnaire was filled in anonymously, one still cannot rule out the tendency to underreport stigmatizing diagnoses as ADHD and ASD.

In none of the groups, loneliness was associated with excessive smartphone use. Previous research suggests a reversed relationship in which close relationships serve as a protective factor against smartphone addiction, when investigating a population comprising both boys and girls [39]. Perhaps our finding could be considered in correspondence with findings that girls, unlike boys, usually use their phones for social reasons such as social media or texting; hence, they may not express a feeling of loneliness [39]. The act of ignoring others in favor of smartphone use at a social setting, also called phone snubbing (or phubbing), has become increasingly common. This is associated with poorer relationship satisfaction and lower family well-being and can be supported by other psychological effects in relation to the increased use of electronic devices, such as feeling low or anxious [40-42]—a finding we were able to verify in our study.

In both age groups, we found that girls who reported excessive smartphone use had a higher probability of often feeling low and often feeling anxious. This finding is in line with previous research findings stating excessive smartphone use is significantly associated with depression and anxiety [43,44].

Elhai et al [10] performed a systematic review on problematic smartphone use and reported that both anxiety and depression are related to problematic smartphone use. The female gender is usually described as a risk factor for problematic smartphone use [10], but whether the female gender also increases the negative consequences thereof, such as psychological complaints, is a question for future studies to answer.

Furthermore, the reporting of less than 7 hours of sleep per night (labeled as “poor sleep” in this study) was reported in both sexes and in both grades. Standard sleep recommendations for teenagers (14-17 years) propose 8 to 10 hours of sleep on a daily basis [45,46]. The importance of sleep during adolescence is a key factor for many neurobiological processes, and sleep contributes to physical and mental health [47,48]. Over the past 20 years, sleep patterns among adolescents have changed, and a link to the increasing amount of time adolescents of today spend on the internet has been suggested [47,49,50]. Royant-Parola et al [51] found that smartphone use, in particular, is associated with poor sleep and negative daily functioning, as well as negative mood. The use of screens such as smartphones and sleep patterns have been previously studied, and suggested proposed mechanisms include (1) displacement of time spent sleeping by time spent using screens, (2) psychological stimulation from screen media content, and (3) alerting and circadian effects of exposure to light from screens [52]. Many adolescents use their smartphones just before bedtime, often leaving their phones in bed and repeatedly and frequently checking for notifications. This behavior is thought to increase smartphone use over time, engaging the person in social reassurance from friends and partners and increasing the possibility for excessive smartphone use [53]. Billieux et al [54,55] described this type of behavior is associated with depression and anxiety. This is also in line with our findings, since participants with the highest probability of poor sleep (ie, girls in both age groups) also had the highest probability of feeling low and feeling anxious.

This study has some limitations. One of the limitations is the cross-sectional design of the study, which does not allow for conclusions to be drawn regarding causation since such this would require a longitudinal investigation. Moreover, all the measures used for this study were based on self-report, which implies a risk for recall bias that could influence the findings. One could also argue for the use of more objective measures, such as electronic registration of smartphone use, as well as more objective indicators of psychological health (eg, cortisol profiles and actual diagnoses).

This study also has considerable strengths. These include the large, representative sample size along with the high response rate, which reduces the risk of selection bias. The survey also included many variables that are not to be found in registers and can only be captured in questionnaires or interviews. Another strength is the Bayesian approach to statistical analysis, which facilitates genuine probabilistic statements about our findings. Furthermore, using the ROPE procedure as a guide to determine the effects that may be of clinical and practical importance offer further robustness to our findings. Future research exploring excessive smartphone use during adolescence should use longitudinal design for an in-depth understanding of the topic.

Although results varied across gender and grade in terms of robustness and size of the estimated difference, overall, we found that excessive smartphone use was associated with a higher frequency of multiple suspected associated factors, including ever having tried smoking, alcohol, and other substances; poor sleep; and often feeling low and often feeling anxious. Moreover, our findings suggest that girls with excessive smartphone use are more prone to experience psychological health concerns than boys—a discrepancy that warrants further investigation. The current study brings light to some features and distinctions of a relevant potentially problematic behavior among adolescents of today.