Problematic use of the internet in low- and middle-income countries before and during the COVID-19 pandemic: A scoping review.

People from low- and middle-income countries (LMICs) represent large portions of the world population, often occupy less favorable living conditions, and typically suffer greater health risks, yet frequently receive little research and global health attention. The present study reviews emerging evidence on problematic use of the internet (PUI) in LMICs prior/during the COVID-19 pandemic. Analyzed studies mainly focused on general properties of PUI in university students, problematic gaming in youth, or problematic use of social media in adults, registering higher prevalence estimates, as compared to earlier reports. Research mainly focused on initially affected regions and COVID-exposed populations. Overall, unfavorable circumstances including poor social support, family relationships and lifestyle tendencies/habits may present potential risk for PUI in LMICs, likely exacerbated during the pandemic.

Current Opinion in Behavioral Sciences 2022, 0:xx–yy

This review comes from a themed issue on VSI: Internet Addiction (2022)

Edited by

https://doi.org/10.1016/j.cobeha.2022.101208

2352-1546/© 2022 Published by Elsevier Ltd.

Introduction

The largest [1] and fastest growing [2] portion of the world population currently comprises 84.3% of all people and resides in low- and middle-income countries (LMICs) [3]. In comparison to high-income countries, people in LMICs typically occupy less favorable living conditions and live in societies with lower levels of wealth, health, and education [4]. As a result, they are more likely to experience mental health problems during a global health crisis, yet they receive relatively few global health resources [5•]. The risk for mental health concerns and increased use of the internet during the COVID-19 pandemic may be more pronounced in vulnerable populations and manifested as excessive, maladaptive or problematic use of the internet (PUI). Disease-related anxieties and fears, economic insecurities, and financial losses, as well the desire to reduce emotional distress during the pandemic, may all contribute to increased risk for PUI in vulnerable populations, regardless of the country or world region 6••, 7.

To date, comparatively little is known about the mental health of people in LMICs as most psychological research has been conducted on narrow populations from countries with established research infrastructures and abundant resources, often referred to as Western, educated, industrialized, rich and democratic (WEIRD) countries and populations [8]. This potentially generates an imbalanced global perspective that lacks sufficient insight into the circumstances of the less developed countries.

The present article aims to contribute to fill this knowledge gap and reviews recent data on PUI in LMICs during the period that preceded or coincided with the COVID-19 pandemic, summarizing studies published between 2018 and 2021. Specifically, we aim to provide a broad overview on PUI-related areas of investigation, frequently employed measures and explored populations, countries or regions in LMICs for the appointed periods.

The findings presented in this review stem from original research articles and are considered with respect to more comprehensive articles (reviews and meta-analyses) on more general topics (like mental health, PUI, COVID-19, LMICs and regions), thus providing more complete coverage.

Methods

A broader collection of related studies was retrieved with a search strategy (see Fig. 1) that was designed to include articles in accordance with the following criteria:

Fig. 1

Conceptual framework of the search strategy and criteria for selection of relevant articles on problematic use of the internet in low- and middle-income countries in the period preceding/coinciding with the COVID-19 pandemic (2018-2021). The search strategy included original research studies, published two years prior and two years into the pandemic. The search criteria included a combination of terms or phrases pertaining to the topics of interest: problematic use, internet activity, and low- or middle-income countries. The search procedure was conducted via two academic databases, covering literature in the matching areas of interest from both the biomedical and the psychological domain (PubMed and APA PsycInfo). The list of low-income, lower-middle and upper-middle income countries is based upon the latest criteria and classifications by the World Bank. (Data source: World Bank Data Help Desk; URL: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups).

The search was conducted via specialized academic databases, covering literature in the matching areas of interest from both biomedical and psychological domains (PubMed and APA PsycInfo).

The period of publication spanned between 2018 and 2021, covering studies from two years prior and two years into the COVID-19 pandemic.

The studies of interest originated from low-income, lower-middle income and upper-middle income countries (in accordance with the latest classifications by the World Bank).

The searched keywords were terms and phrases that pertain to the topics of interest: problematic use, internet activity, and low-income or middle-income countries. The search was performed by the title of the original research articles.

The broader collection of retrieved studies was then reduced to the most relevant studies (see Fig. 2), after exclusion of articles in accordance with the following criteria:

Fig. 2

A flow diagram depicting selection of relevant studies on problematic use of the internet in low- and middle-income countries in the period preceding or coinciding with the COVID-19 pandemic (2018-2021) The diagram was is informed by the standards for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. (Data source: PRISMA, 2020. URL: http://www.prisma-statement.org/) 2 Reports on problematic gambling in LMICs were excluded from the final review since they predominantly explored onsite, rather than online gambling.

Duplicates, or articles with similar reports (regarding used samples and methods) in different academic outlets.

Studies on topics that were outside the specific scope of interest.

Studies from countries that were outside the target list.

Studies conducted outside the target period and/or studies published in languages other than EN.

Studies with insufficient data regarding the study period and the methodologies used.

The organization of work throughout the selection process was conducted in two phases. In Phase 1, the initial selection was performed by the first author (BG) and supervised by the last author (ZD) on the basis of search criteria that were previously agreed upon by all authors (Fig. 1). In Phase 2, the pre-finalized selection, informed by international standards for review studies and meta-analyses (Fig. 2), was reviewed separately by the remainder of the authors (MNP, JJ, CMDS & GH). The individual evaluations sought to promote unbiased feedback and objective reporting of the results. Four additional studies were identified in this process, and included in the final selection as relevant for the current review (Fig. 2). Ultimately, sixty-nine studies were reviewed, and findings were organized according to: most frequently researched topics (PUI in general, problematic gaming or problematic use of social media), investigated populations, frequently employed measures, reported prevalence estimates, potential risk factors (see Table 1 for a summary of studies and findings), and geographical regions (see Table 2 for the global distribution of studies). Reports on problematic gambling in LMICs were excluded from the final review since they predominantly explored onsite, rather than online, gambling.

Table 1

Summary of reviewed studies and reported findings on problematic use of the internet in low- and middle-income countries in the period preceding or coinciding with the pandemic.

PUI		PERIOD2	GENERAL	GAMING	SOCIAL MEDIA
POPULATION1	Children	BEFORE2 studies	Cao et al. 2020 [56]Cai et al. 2021a [20]	No matches found	No matches found
		DURING4 studies	Dong et al. 2020 [19]	Chen et al. 2021 [13•]	Fung et al. 2021 [12•]Chen et al. 2021 [13•]
	Adolescents	BEFORE13 studies	Cam & Ustuner, 2020 [57]; Chi et al. 2020 [58]Popadić et al. 2020 [59]; Iqbal et al. 2021 [60]Nguyen et al. 2021 [61]; Karki et al. 2021 [62]; Kaya & Dalgiç, 2021 [63]	Maftei & Enea, 2020 [64]; Yu et al. 2020 [40] Areshtanab et al. 2021 [23]; Yu et al. 2021a[65] Yu et al. 2021b [66]; Yu et al. 2022 [24]	No matches found
		DURING5 studies	Dong et al. 2020 [19]; Li et al. 2021a [38]Rakhmawati et al. 2021 [10];Sarıalioğlu et al. 2020[67]	Cuong et al. 2021 [22]	No matches found
	Youngadults	BEFORE17 studies	Arafa et al. 2019a [68]; Jahan et al. 2019 [69]; Akhter et al. 2020 [70]Asrese & Muche, 2020 [71]; Hassan et al. 2020 [72]; Mboya et al. 2020 [73]Salama, 2020 [74]; Sharma et al. 2020 [75]; Wang et al. 2020 [76]Al Shawi et al. 2021 [77]; Khazaie et al. 2021 [32]; Mohanty et al. 2021 [11]Özarıcı & Cangöl Sögüt, 2021 [78]; Shan et al. 2021[79]; Zenebe et al. 2021 [80]	Yu et al. 2019 [81]	Basu et al. 2021 [30]
		DURING11 studies	Cai et al. 2021b [31]; Condori-Meza et al. 2021[34]; Fernandes et al. 2021 [44]Hosen et al. 2021 [51]; Sayeed et al. 2021 [42]; Shehata & Abdeldaim et al. 2021 [33]Xie et al. 2021 [48]; Zhao et al. 2021 [50]	No matches found	Sayeed et al. 2020 [43];Fernandes et al. 2021 [44] Larnyo et al. 2021 [46];Sujarwoto et al. 2021 [41]
	Adults	BEFORE3 studies	Arafa et al. 2019b [82]Singh et al. 2019 [35]	Shao et al., 2021 [83]	No matches found
		DURING16 studies	Islam et al. 2020 [53]; Jovic et al. 2020 [84]Siste et al. 2020 [52]; Sun et al. 2020 [85]Abir et al. 2021 [54]; Huang et al. 2021 [49]Li et al. 2021b [86•]; Zhou et al. 2021 [39]	No matches found	Lee et al. 2020 [36]; Ni et al. 2020 [37]Zhang et al. 2020 [26];Guelmami et al. 2021 [55]Lugito et al. 2021 [28];Luo et al. 2021 [47]Mahmood et al. 2021 [27]; Rizwan et al. 2021[45]
MEASURES	Timespentonline3	BEFORE3 studies	Arafa et al. 2019b [82]; Karki et al. 2021 [62]; Popadić et al. 2020 [59]	No matches found	No matches found
		DURING13 studies	Zhou et al. 2021 [39]; Ni et al. 2020 [37]; Jovic et al. 2020 [84]Huang et al. 2021 [49]; Hosen et al. 2021 [51]; Islam et al. 2020 [53]Abir et al. 2021 [54]	No matches found	Lee et al. 2020 [36]; Lugito et al. 2020 [28]Ni et al. 2020 [37]; Luo et al. 2021 [47]Zhang et al., 2020 [26]; Rizwan et al. 2021 [45]
	Most frequently used scales	BEFORE20 studies	IAT (Young, 1998a)[14]Singh et al. 2019 [35]; Asrese & Muche, 2020 [71]; Cam & Ustuner, 2020 [57]Hassan et al. 2020 [72]; Mboya et al., 2020 [73]; Salama, 2020 [74]Sun et al. 2020 [85]; Wang et al. 2020 [76]; Cai et al, 2021a [20]Kaya & Dalgiç, 2021 [63]; Khazaie et al., 2021 [32]; Mohanty et al., 2021 [11]Al Shawi et al., 2021 [77]; Zenebe et al., 2021 [80]	DSM-5 checklist (APA, 2013) [21]Yu et al. 2019 [81]; Yu et al. 2020 [40]Shao et al. 2021 [83];Yu et al. 2021a [65]Yu et al. 2021b [66]; Yu et al., 2022 [24]	BSMAS (Andreassen, 2016) [25]No matches found
		DURING12 studies	IAT (Young, 1998a) [14]Dong et al. 2020 [19]; Sayeed et al. 2020 [43]; Cai et al. 2021b [31]Condori-Meza et al. 2021 [34]; Guelmami et al. 2021[55]; Li et al. 2021a [38]Li et al. 2021b 86•, 86•; Shehata & Abdeldaim et al. 2021 [33]; Zhao et al. 2021[50]	DSM-5 checklist (APA, 2013) [21]No matches found	BSMAS (Andreassen, 2016) [25]Fung et al. 2021 [12•]Luo et al.,2021[47]Sujarwoto et al. 2021 [41]
	Other used scales	BEFORE8 studies	YDQ (Young, 1998b) [87]CIUS (Meerkerk et al. 2009)[16]Cao et al. 2020 [56]; Chi et al. 2020 [58] Arafa et al. 2019a [68]GPIUS2 (Caplan, 2010) [88]CIAS (Chen et al., 2003)[13•]Sharma et al. 2020 [75]Shan et al. 2021[79]	IGD-20 (Pontes et al. 2014)[89]Maftei & Enea, 2020[64]Areshtanab et al. 2021 [23]IGCS (King & Delfabbro, 2016) [90]Yu et al. 2019[81]	SMUQ (Xanidis & Brignell, 2016) [91]Basu et al. 2021[30]
		DURING5 studies	YDQ (Young, 1998b) [87]CIUS (Meerkerk et al., 2009) [16]Xie et al., 2021 [48]Fernandes et al., 2021 [44]GPIUS2 (Caplan, 2010) [88]CIAS (Chen et al., 2003)[13]Sayeed et al., 2020 [43]No matches found	IGD-20 (Pontes et al., 2014)[89]Cuong et al., 2021[22]	SMUQ (Xanidis & Brignell, 2016) [91]Fernandes et al., 2021 [44]
FINDINGS	Prevalence estimates4	BEFORE	IAT score ≥ 50Prevalence estimates: 33.04% (mean); 29.40 (median)14.0% Singh et al. 2019 [35]20.0% Cai et al. 2021a [20]21.1% Cam & Ustuner, 2020 [57]23.0% Al Shawi et al., 2021 [77]27.1% Hassan et al. 2020 [72]29.4% Zenebe et al. 2021 [80]31.0% Mboya et al. 2020 [73]34.8% Asrese & Muche, 2020 [71]46.8% Sun et al. 2020 [85]47.5% Salama, 2020 [74]68.8% Khazaie et al. 2021 [32]	DSM-5 ≥ 5Prevalence:18.38% (mean); 3.5 (median)11.7% Yu et al. 2019 [81]13.1% Yu et al. 2022 [24]13.5% Yu et al. 2021a [65]13.6% Yu et al. 2020 [40]40.0% Yu et al. 2021b [66]	BSMAS score ≥ 24Prevalences: N/A6.8% Luo et al., 2021)[47]
		DURING	IAT score ≥ 50Prevalence estimates:36.79% (mean); 33.60 (median)14.7% Condori-Meza et al. 2021[34]23.3% Cai et al. 2021b [31]28.4% Zhao et al. 2021 [50]31.2% Li et al. 2021a [38]36.0% Dong et al. 2020[19]36.7% Li et al. 2021b 86•, 86•43.8% Sayeed et al. 2020[43]80.2% Shehata & Abdeldaim et al. 2021 [33]	DSM-5 ≥ 5Prevalence estimates: N/ANo matches found	BSMAS score ≥ 24Prevalence estimates: N/ANo matches found
	Potential risk factors5	BEFORE /DURINGCOVID-19	Demographic characteristicsPredominantly males (Dong et al. 2020 [19]; Condori-Meza et al. 2021 [34]; Kaya & Dalgiç, 2021 [63]; Sayeed et al. 2021 [42]; Shan et al. 2021 [79]; Sharma et al. 2020 [75]) at younger age (Arafa et al. 2019b [82]; Islam et al., 2020 [53]; Kaya & Dalgiç, 2021 [63]) with lower level of education and/or poor academic performance (Asrese & Muche, 2020 [71]; Chi et al. 2020 [58]; Shehata & Abdeldaim et al. 2021 [33]).Personality features / Coping stylesLow self-esteem (Arafa et al. 2019b [82]; Asrese & Muche, 2020 [71]; Cam & Ustuner, 2020 [57]), negative coping styles (Shan et al. 2021 [79]) with boredom, loneliness and depression, especially during the pandemic, due to increased isolation and decreased social interaction (Dong et al 2020 [19]; Sayeed et al. 2021 [42]).Parenting strategies / Social surroundingsInadequate mediation strategies by mothers for safe internet practices of children (Iqbal et al. 2021 [60]); poor relationships with family and friends (Asrese & Muche, 2020 [71]) expressed as detachment and isolation during the pandemic (Li et al. 2021b 86•, 86•; Zhou et al. 2021 [39]).Lfestyle tendencies / habitsLack of physical activity and avoidance of household chores (Islam et al. 2020 [53]; Hosen et al. 2021 [51]; Sharma et al. 2020 [75]); increased use of alcohol, cigarettes, online games or networking sites with prolonged exposure to mis/dis/information and distressing content (Guelmami et al., 2021 [55]); decreased sleep quality and/or quantity (Jahan et al. 2019 [69]; Singh et al. 2019 [35]; Wang et al. 2020 [76]; Shehata & Abdeldaim et al., 2021 [33]; Hosen et al., 2021 [51]).	Demographic characteristicsPredominantly males (Shao et al. 2021 [83]), adolescents (Maftei & Enea, 2020 [64]; Areshtanab et al. 2021 [23]; Cuong et al. 2021 [22]; Yu et al. 2022 [24]).Personality features/ Coping stylesLow self-control (Yu et al. 2019 [81]), high impulsivity (Yu et al. 2021b [66]), pronounced loneliness and depression (Yu et al. 2020 [40]), poor social support and relationship adaptation (Yu et al. 2022 [24]).Parenting strategies / Social surroundingsLack of (or undisciplined) parental supervision with overly permissive mediation strategies for internet use of children (Maftei & Enea, 2020 [64]; Cuong et al. 2021 [22]); inadequate parenting style of mothers (Areshtanab et al. 2021 [23]) and/or dysfunctional parent-child relationships,, lower parental education, lack of social support.Lifestyle tendencies / habitsInsomnia (Yu et al. 2020 [40]).	Demographic characteristicsFenales (Zhang et al. 2020 [26]), males (Luo et al. [36] 2021) or both (Rizwan et al. 2021 [45]).Personality features / Coping stylesEmotional distress and hyper-arousal (Lee et al. 2020 [36]; Larnyo et al. 2021 [46]; Lugito et al. 2021 [28]; Luo et al. 2021 [47]), depression (Fung et al. 2021 [12•]; Sayeed et al. 2020 [43]), perceived threat (Mahmood et al. 2021 [27]), percieved weight stigma (Fung et al. 2021 [12•]).Parenting strategies / Social surroundings Dysfunctional romantic and/or domestic relationships (Sayeed et al. 2020 [43])Lifestyle tendencies / habitsIncreased consumption of news on internet (Ni et al. 2020 [37]); prolonged exposure to social media; poor sleep quality with sleep disturbances (Basu et al. 2021 [30]).

DSM-5 = Diagnostic Statistical Manual (5th revision); IAT = Internet Addiction Test; CIUS = Compulsive Internet Use Scale; GPIUS2 = Generalized Problematic Internet Use Scale 2; BSMAS = Bergen Social Media Addiction Scale; IGD = Internet Gaming Disorder Test; SMUQ = Social Media Use Questionnaire; YDQ = Young Diagnostic Questionnaire; CIAS = Chinese Internet Addiction Scale; IGCS = Internet Gaming Cognition Scale. Referencing styles: Alphabetical and numerical. Referencing order: Where applicable, the studies are ordered by the year of publication (primary criteria), the alphabetic order of the first author’s name (secondary criteria), and the sequential order in the bibliography (tertiary criteria).

1Presented categories reflect a combination of age and the manner in which the cohorts were defined in each study. In general, children are youth attending elementary school (aged approximately 7-10 years), adolescents are youth attending middle or high school (aged approximately 11-17 years), young adults attending universities or colleges (aged approximately 18-25 years), while other adults are aged approximately 26 years or higher.

2The before/during period refers to the years that preceded (2018-2019) or coincided (2020-2021) with the COVID-19 pandemic.

3Time spent online was measured as an average number of hours per day for the corresponding PUI activity.

4The reported prevalence rates were measured with frequently used scales, while the conventional cutoff scores pertain to frequently used criteria. In studies relying on different criteria, the prevalence rates for the conventional cutoff scores among healthy individuals were extracted from the information provided in the articles.

5The summaries highlight frequently reported risk factors across reviewed studies for each of the PUI types

Table 2

Cumulative number of original research articles on problematic use of the internet in low- and middle-income countries across different world regions, in the period preceding or coinciding with the COVID-19 pandemic.

GEOGRAPHIC DISTRIBUTION OF RESEARCH ON PUI
World regions1	Studies conducted before COVID-19 (2018-2019)	Sum	Studies conducted during COVID-19 (2020-2021)	Sum
East Asia & Pacific	Yu et al., 2019 [81] Cao et al. 2020 [56]Yu et al., 2020 [40] Wang et al. 2020 [76]Cai et al, 2021a [20]Shao et al. 2021[83]Kaya & Dalgiç, 2021 [63] Nguyen et al. 2021 [61]Shan et al. 2021[79] Yu et al. 2021a [65]Yu et al. 2021b [66]Yu et al. 2022 [24]	12	Dong et al. 2020 [19] Lee et al. 2020 [36]Ni et al. 2020 [37] Siste et al. 2020 [52]Sun et al. 2020 [85] Zhang et al. 2020 [26]Cai et al. 2021b [31] Chen et al. 2021 [13•]Cuong et al. 2021[22] Fernandes et al. 2021 [44]Fung et al. 2021 [12•] Huang et al. 2021 [49]Larnyo et al. 2021 [46] Li et al. 2021a [38]Li et al. 2021b 86•, 86• Lugito et al. 2021 [28]Luo et al. 2021 [47] Rakhmawati et al. 2021 [10]Sujarwoto et al. 2021 [41] Zhao et al. 2021 [50]Zhou et al. 2021 [39]	21
South Asia	Jahan et al. 2019 [69] Singh et al, 2019 [35]Hassan et al. 2020 [72] Sharma et al. 2020 [75]Basu et al. 2021 [30] Iqbal et al. 2021 [60]Kaya & Dalgiç, 2021[63] Mohanty et al. 2021 [11]	8	Islam et al. 2020 [53] Sayeed et al. 2020 [43]Abir et al. 2021[54] Fernandes et al. 2021 [44]Hosen et al. 2021 [51] Mahmood et al. 2021 [27]Rizwan et al. 2021 [45] Sayeed et al. 2021 [42]	8
Europe & Central Asia	Cam & Ustuner, 2020 [57] Maftei & Enea, 2020 [64]Popadić et al. 2020 [59] Kaya & Dalgiç, 2021 [63]Özarıcı & Cangöl Sögüt, 2021 [78]	5	Jovic et al. 2020 [84]Sarıalioğlu et al. 2020 [67]	2
Middle East & North Africa	Arafa et al. 2019a [68] Arafa et al. 2019b [82]Salama, 2020 [74] Areshtanab et al. 2021 [23]Al Shawi et al. 2021 [77] Khazaie et al., 2021[32]	6	Guelmami et al. 2021 [55]Shehata & Abdeldaim et al. 2021 [33]	2
Sub-SaharanAfrica	Asrese & Muche, 2020 [71]Mboya et al. 2020 [73]Zenebe et al. 2021 [80]	3	No matches found	0
LatinAmerica & Caribbean	No matches found	0	Condori-Meza et al. 2021[34]Fernandes et al. 2021 [44]	2

1 The list of world regions is based upon the latest classifications by the World Bank. (Data source: World Bank Data Help Desk; URL: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups)

Results and Discussion

PUI is a relatively recent phenomenon, and many LMICs still lack resources or policies to properly understand or address PUI [9]. The need for a broader outlook and more general understanding of PUI in LMICs is reflected in the fact that most studies focused on exploring the general properties and correlates of PUI (n = 46). A smaller number of studies explored specific characteristics of problematic use of social media (n = 14) and problematic gaming (n = 9) in LMICs (see Table 1: “Gaming” and “Social media” columns, 1-8 rows).

With one notable exception that provided qualitative evidence [10], the remainder of the reviewed studies were quantitative, reporting findings that were based on survey methodologies and statistical analyses. Also, three longitudinal studies 11, 12•, 13• presented exceptions to the overwhelming body of cross-sectional research. The sample sizes varied considerably across studies, ranging between 200 and 20,000 participants, with an average size of around 2,000 and a median size of approximately 750 participants per study. The most frequently represented populations also differed across research topics, depending on whether studies explored PUI in general, problematic gaming or problematic use of social media. For more information regarding the study topics and types, methodologies, populations and findings, please see the following sections of this manuscript.

An overview of PUI in LMICs

Generalized PUI was mainly assessed using convenience samples, with half of the studies (23 of 46 publications) surveying young adults attending universities or colleges (participants aged appro-ximately 18-25 years). Approximately half of the studies investigating generalized PUI (22 of 46 studies) utilized the Internet Addiction Test (IAT) [14], a 20-item survey with 0-5 point Likert type responses and 0-100 score range. The IAT was used to quantify self-reported preoccupation and compulsive use of the internet, as well as behavioral problems, emotional changes and diminished functionality due to internet use. The measure has been reported to have relatively “high internal consistency reliability within homogenous samples (α =.90—.93), test–retest reliability (ρ =.83), and a relatively simple factor structure of between one and two dimensions” [15•]. However, lately, the IAT has been subject to academic criticism regarding its psychometric properties. Some of the identified issues pertain to potentially redundant or outdated items, an unstable factor structure, arbitrary cutoff scores, and possible lack of universal validity [15•], so research may shift towards newer scales with better psychometric properties, like the Compulsive Internet Use Scale [16]. However, this trend is still not evident in the latest research on PUI across LMICs. A considerable number of studies relied on IAT, while others relied on the average number of daily hours spent on the internet as a rough estimation of PUI. Only a small group of studies relied on more targeted instruments (see Table 1 for the lists of assessment instruments that were used most frequently).

A frequently used cutoff score (≥ 50) for the IAT was considered for PUI in the present review (even though cutoff scores often differed across studies and the prevalence rates varied accordingly). Wherever applicable, the prevalence rate for the conventional cutoff score in healthy (control) individuals was extracted from the original report, to calculate an average prevalence estimate for PUI among the general population in LMICs. The final average rates (34.6%) and median prevalence estimates (31.0%) were retrieved on the basis of reports from 19 studies. The average prevalence rate in particular was considerably higher than earlier estimates, obtained from large samples with 89,281 participants [17] and 693,306 participants [18••] in 31 nations (6.0% and 7.0% accordingly). Such a discrepancy may reflect contextual factors, like the time period and region. Namely, earlier meta-analyses relied on studies that were published in earlier time periods, considerably before the onset of the COVID-19 pandemic (1996-2012 and 1996-2018, respectively). On the other hand, the present review scopes evidence for the period shortly preceding and coinciding with the COVID-19 pandemic (2018-2021), which is marked by a global expansion of internet use. Regarding the regional analysis, earlier studies have indicated that the prevalence estimates are likely higher in Eastern regions (10.9% and 8.9%, respectively) 17, 18•• and societies with disadvantaged living conditions or dissatisfied populations [17]. Considerable 17, 18•• differences in prevalence estimates between the present and the two referenced studies may also be technical in nature and attributable to the frequently used conventional cutoff score (IAT ≥ 50) being more inclusive than a stricter one (IAT ≥ 60) [18••]. In addition, several articles in the present review utilized the IAT to assess generalized PUI in children and adolescents 19, 20, despite the IAT having been developed for assessing PUI in young and healthy adults. Younger and more vulnerable populations may be more susceptible to PUI behaviors, and this may in part explain the higher scores.

In this regard, research on problematic gaming has explored almost exclusively effects on younger populations, comprised of youth attending elementary school (aged approximately 7-10 years) or middle or high school (aged approximately 11-17 years). Eight (of 10) studies focused on problematic gaming, and prevalence rates were frequently estimated using a 9-item checklist by the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [21]. Overall, prevalence estimates of problematic gaming across five studies, ranged between 11.7% and 40.0%, while the average prevalence was estimated at 18.4%. This value is higher than an earlier estimate of 2.47% from a comprehensive meta-analysis [18••]. Moreover, studies on problematic gaming have largely focused on parent-child relationships, examining roles of parenting styles or internet mediation strategies on gaming behaviors in children and adolescents. Poor family relations and poor parental education, dysfunctional families, lack of parental supervision and overly permissive maternal mediation strategies for internet use of children were recurring determinants associated with problematic gaming 22, 23, 24.

Problematic use of social media has been explored in different populations (mainly adults and young adults), in multiple ways (mainly via quantity of social media use and the Bergen Social Media Addiction Scale [25]), and in different contexts (mainly the COVID-19 pandemic). Hence, it is difficult to identify common patterns and draw general conclusions (see Table 1). Nonetheless, the use of social media may have been beneficial during the COVID-19 pandemic, possibly serving as a corrective force that enabled more efficient health communication with safe and timely delivery of information that was provided by close and reliable sources [26]. Protective behaviors and self-efficacy of people may have increased as a result [27], while feelings of impending threat, anxiety and depression decreased in some instances [28]. However, a larger body of research conducted during the same period describes the opposite (positive) relationship between the increased use of social media (usually more than 2-3 hours/day) and associated concerns among youth 12•, 13• and adults (see the next section for more details).

PUI in LMICs during the COVID-19 pandemic

Research conducted during the COVID-19 pandemic mainly stems from initially affected regions, with most studies (21 of 35) conducted in East Asia. In fact, the intensity of research of PUI in East-Asian countries nearly doubled in the years coinciding with the pandemic (2020-2021), as compared to the years that preceded the pandemic (2018-2019). The was not the case with the rest of the world regions (see Table 2). China was a regional leader in research on the subject, exploring multiple PUI behaviors in different contextual settings and populations during the pandemic. Overall, prevalence estimates of PUI types in Eastern countries were higher than those previously reported. There is recent evidence to suggest that the prevalence estimates in Southeast Asia are higher than in other jurisdictions, but the findings stem from a single meta-analysis performed on non-representative populations [29]. Hence, the present review may provide a more nuanced and better understanding of the situation in regions that were initially affected by the pandemic.

In addition to citizens from affected regions, other populations exposed early to the virus also received considerable scholarly attention. These included medical and nursing students 11, 30, 31, 32, 33, 34, medical residents, doctors and nurses, among others 35, 36, 37. However, the list of comorbidities frequently associated with PUI during the pandemic appears similar for medical and general populations. The problems ranged from amplified levels of stress and pronounced traumatic experiences including depression 19, 36, 37, 38, 39, 40, 41, 42, 43, anxiety 12•, 31, 37, 44, 45, 46, 47 or post-traumatic stress disorder [48] (in which case, the link with PUI was established due to increased exposure to distressing content and disinformation on the internet), to problems associated with instant gratification and stimulation like substance use [49] and attention-deficit/hyperactivity [50] disorders.

Across different research topics and contexts, findings suggest that PUI behaviors link to various potential risk factors, broadly categorized as demographic characteristics, personality features, coping styles, parenting strategies, social surroundings and lifestyle tendencies/habits (see Table 1, section “Findings”). Importantly, poor lifestyle tendencies/habits, living conditions and negative coping styles appear implicated across different types of PUI and LMICs. For instance, poor quantity and quality of sleep (characterized by insufficient sleep hours or disorganized sleeping patterns with inadequate or irregular sleeping periods, and manifested as daytime sleepiness or even insomnia) was repeatedly described as a possible cause or a consequence of PUI during the pandemic 35, 51, 52. Lack of physical activities (e.g., exercise and outdoor recreation) [53] and physical discomfort (e.g., headaches, back pains, finger numbness) were also associated with PUI [54]. Prolonged exposure to inaccurate or distressing content on the internet was also associated with PUI 30, 55. Regarding negative coping styles, feelings of boredom, isolation and loneliness, coupled with a lack of social or emotional support from family and friends during long periods of quarantine and lockdown, were often associated with general and the specific forms of PUI 24, 44, 55.

Limitations

In line with journal aims, the present review focused on recent studies (conducted in the period around the COVID-19 pandemic) and aimed to present findings in a condensed format (offering a snapshot of PUI in LMICs). To achieve this end, the authors performed targeted searches by article titles in bibliographic databases with matching areas of interest. Future studies could benefit from expanded searches covering longer periods (e.g., last five or ten years of research), and extending across different article fields (e.g., keywords, title, abstract, body of text or combinations thereof), as well as additional academic databases (e.g., Web of Science or Scopus). In essence, the present review scopes the existing evidence and synthesizes recent findings, thus serving as a useful precursor for future reviews that could systematically assess the quality and quantity of accumulated knowledge and propose viable solutions.

Conclusions

The present study provides evidence on PUI in LMICs shortly prior to, and during, the COVID-19 pandemic. The articles reviewed mainly focused on the generalized PUI in university students, problematic gaming among children and adolescents, or problematic use of social media in adults, with most reporting higher-than-average prevalence estimates, as compared to earlier studies. Research covering PUI during the COVID-19 pandemic nearly doubled in the initially affected geographical regions and populations that were first exposed to the novel coronavirus. Overall, unfavorable conditions associated with poor lifestyle tendencies/habits, social support and family relationships may represent risk factors for PUI in LMICs prior to and during the pandemic.

This manuscript reviews a modest body of knowledge from less represented countries, thus contributing to a more comprehensive and balanced view of PUI across different geopolitical, social and cultural contexts. The summary of findings may inform and inspire future research and policy strategies across concerned regions, countries or populations, to mitigate PUI.

References

Papers of particular interest, published within the period of review, have been highlighted as:

•of special interest

•• of outstanding interest