Identified studies and quality assessment
Of 924 studies identified, 41 studies were included in this review (Fig. 1). Of those, 22 studies were deemed to be of poor methodological quality, and 19 of moderate quality (See Additional file 1: Table S2). Three cohort and 38 cross-sectional studies were included, with 41,871 participants, 55% of which were female. Included studies were conducted in Europe (n = 9), Asia (n = 30), and America (n = 2) (See Additional file 1: Table S3). There was wide variability in the definitions of PSU (See Additional file 1: Table S4), and the criteria used ranged from a single criterion such as psychological withdrawal phenomena (n = 2), to measurement of tolerance, withdrawal, loss of control, preoccupation, neglect of other activities and evidence of harm, which form the criteria for behavioural addictions (n = 19) (See Additional file 1: Table S5).
Types of PSU usage
Communication was the most frequent type of smartphone usage by those with PSU, reported by 14 studies [28,29,30,31,32,33,34,35,36,37,38,39]. Problematic users reported that social networking was the most important or preferred activity on smartphones [34, 40]. ‘Addicted’ or ‘problem user’ groups were linked with particular phone uses: TV watching [35] and social networking [37, 38]. Lee and Lee [41] found that use of phones to gain peer acceptance was the most significant type of use related to PSU.
Both substance use and other behavioural addictions were associated with PSU. Internet addiction [33, 42,43,44], Facebook addiction [31], compulsive buying [43], increased alcohol use [42], and cigarette smoking [42] were also found to be positively associated with PSU.
Sociodemographic characteristics associated with PSU
Across 14 studies, age was correlated with PSU [28,29,30,31,32,33, 40,41,42, 45,46,47,48,49], and 17 to 19 year-olds were the most frequent sufferers of PSU. Females were reported as more prone to PSU by 13 studies [31,32,33,34, 40, 41, 48,49,50,51,52,53,54]; however 4 studies reported the opposite [35, 46, 55, 56]. PSU in males was correlated with use of media applications and games, while in females it was correlated with communication and social networking applications [28]. PSU was also positively associated with monthly cost of living [28], family income [36], and a higher economic status [42].
Estimating the prevalence of PSU
Prevalence was assessed using 24 different questionnaires, with the most common being the Smartphone Addiction Scale, Short Version (SAS-SV; n = 7) and the Smartphone Addiction Proneness Scale (SAPS; n = 5), for further details (See Additional file 1: Table S4).
The majority of studies (n = 31) found a prevalence between 10 and 30%, and the median was 23.3% (interquartile range 14–31%, Fig. 2).
PSU associated with mental health outcomes
PSU has been consistently associated with measures of poor mental health, in particular relating to depression, anxiety, stress, poor sleep quality, and day to day functional impairment demonstrated by poor educational attainment. Of the studies included, 20 investigated the relationship between PSU and mental health amongst CYP. This is summarized in a qualitative synthesis (See Additional file 1: Table S6).
Depression
Eight studies [28, 36, 48, 57,58,59,60,61] reported a significant association between PSU and depression across 10,099 participants. Dichotomous data from four studies was extracted using standard cut-offs for the clinical diagnosis of depression. In those with PSU the odds ratio (OR) of depression was 3.17 (95% CI, 2.30, to 4.37; p < 0.001; I
2 = 78%; Fig. 3). Given the consistency of the study findings, we have upgraded this to a GRADE of moderate quality.
Anxiety
Seven studies [28, 30, 36, 57, 59,60,61] investigated the relationship between PSU and anxiety in CYP. Of seven studies across 9359 participants, six found a significant positive association between PSU and anxiety; one study reported a negative association [60]. The pooled OR for anxiety amongst CYP with PSU was 2.60 (95% CI 1.39, to 4.85; p < 0.001; I
2 = 78%). The large heterogeneity due to Tavakolizadeh et al. [60], is explained by geography, and the Iranian protests of 2011–2012. After accounting for the heterogeneity, the OR for anxiety amongst CYP with PSU was 3.05 (95% CI 2.64, to 3.53; p < 0.001; I
2 = 0%;. 3) The GRADE of evidence was categorised as low quality.
Stress
Five studies [36, 39, 40, 44, 62] investigated perceived stress across 3618 participants. Four studies found a significant association between PSU and perceived stress amongst CYP, whilst Tahtsidou et al. [44] found no significant relationship. A subgroup analysis was introduced due to PSU prevalence. Most heterogeneity was accounted for by Venkatesh et al. [39], a study reporting a PSU prevalence of 71.9% – this study was subsequently excluded. The pooled OR for typical PSU prevalence and perceived stress amongst CYP was 1.86 (95%CI 1.24, to 2.77; p = 0.002; I
2 = 65%; Fig. 3). The GRADE of the evidence was categorised as low quality.
Sleep
There were seven studies [28, 35, 57, 61, 63,64,65] which investigated poor sleep across 4194 CYP. Six studies reported a significant positive association between PSU and poor sleep, while Demirci et al. [58] reported no significant association. The pooled OR for the extracted data on PSU and subsequent poor sleep was 2.60 (95%CI, 1.39, to 4.85, p = 0.003, I
2 = 78%; Fig. 3). The GRADE of evidence was categorised as low quality after accounting for both the narrative analysis and the pooled analysis.
Educational attainment
Six studies [41, 48, 50, 60, 66, 67] explored PSU and educational attainment across 6655 CYP. Four studies reported a significant association between PSU and poor educational attainment, whilst one [60] found no significant relationship.
Variations in measures of educational attainment were used; it was therefore not appropriate to pool the results of the studies. However, they are summarised to demonstrate the consistency of reported associations between PSU and poor educational attainment (See Additional file 2: Figure S1).
Suicide
One study reported an increased odds of suicidal ideation amongst those with PSU [62]; however, this was assessed through a single screening question and caution should be taken with this finding.
Psychological factors associated with PSU
A range of different personality and emotional factors were investigated in relation to PSU. Somewhat paradoxically, traits associated with greater risk-taking (such as low self control, impulsivity, emotional instability, and openness) and traits associated with avoidance of risk taking (such as perfectionism and conscientiousness), were more common amongst problematic smartphone users [51, 52, 61, 66]. An insecure attachment style, loneliness [45, 56, 65], and low self esteem [49] were all associated with PSU.