Effectiveness and psychological mechanism of group-based IMPROVE therapy for treatment seeking college students with problematic internet use in China: study protocol for a prospective, single-blind, randomised clinical trial

Abstract

Introduction

Problematic internet use (PIU) is a prevalent and heterogeneous mental health problem among university students, associated with significant psychological distress and functional impairment. Although various psychosocial interventions have demonstrated efficacy, existing randomised controlled trials (RCTs) are frequently limited by small sample sizes, single-component designs and insufficient investigation of underlying psychological mechanisms. This study aims to evaluate the effectiveness and psychological mechanisms of IMPROVE (Identify, Monitor, Plan, Real-world activities, Observe emotion, Vulnerability and Emotion regulation), a manualised, multimodal group-based psychotherapy integrating cognitive-behavioural strategies with motivational elements, mindfulness training and socially interactive exercise for the treatment of PIU among university students.

Methods and analysis

This single-blind RCT will recruit 600 university students with moderate-to-severe PIU (Internet Addiction Test (IAT) ≥50) from multiple campuses. Participants will be randomised (1:1) to either an 8-week IMPROVE intervention or an enhanced waitlist control condition. The primary outcome is change in PIU severity (IAT score) from baseline to post-intervention. Secondary outcomes include remission rates, internet-related behaviours, craving, functional impairment, comorbid psychopathology and well-being. Candidate psychological mediators—maladaptive cognitions, self-efficacy, emotion regulation, mind wandering, anhedonia, resilience and perceived stress—will be assessed longitudinally to support formal mediation analyses. Prespecified moderation analyses will examine treatment heterogeneity across PIU subtypes and baseline characteristics.

Ethics and dissemination

The study was approved by the Ethics Committee of Sir Run Run Shaw Hospital, Zhejiang University School of Medicine (approval no. 2026–0152). Written informed consent will be obtained from all participants prior to enrolment. Study findings will be disseminated through peer-reviewed publications and conference presentations.

Trial registration number

NCT07434778 (https://clinicaltrials.gov/study/NCT07434778).

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The planned sample of 600 participants will support prespecified moderation analyses examining heterogeneity in treatment effects across problematic internet use (PIU) subtypes and baseline characteristics.

• The trial features a manualised, multimodal intervention integrating cognitive-behavioural strategies, mindfulness and socially interactive exercise, which systematically targets the multifactorial psychological components of behavioural addiction.

• The prospective, repeated assessment of candidate mediators will enable longitudinal mediation analyses examining whether early changes in these psychological processes statistically account for subsequent improvement in PIU severity.

• Due to the inherent practical constraints of delivering psychotherapeutic interventions, the study employs a single-blind design where only outcome assessors and data analysts are blinded, which cannot rule out participant or therapist expectation biases.

• The use of an enhanced waitlist control design ethically ensures eventual treatment access for all enrolled participants, but this crossover precludes controlled between-group comparisons for long-term efficacy beyond the 3-month follow-up.

Introduction

Problematic internet use (PIU) has emerged as an important public mental health concern in the digital era.¹ Although no universally accepted diagnostic definition exists, PIU is generally conceptualised as a maladaptive pattern of internet use associated with psychological distress, social dysfunction and academic or occupational impairment.^{2 3} Growing evidence supports a spectrum perspective, whereby problematic online behaviours are regarded as related but distinct conditions rather than manifestations of a single homogeneous disorder.^4–6 Within this spectrum, individuals may predominantly engage in online gaming, social media use, short-form video consumption or other internet-mediated activities. Although these behaviours differ in their content and context, they share overlapping addictive features and psychological vulnerabilities. This heterogeneity complicates the development of interventions that are sufficiently standardised while remaining responsive to different patterns of problematic use.

University students are particularly vulnerable to PIU.^{7 8} Emerging adulthood is characterised by increasing autonomy, substantial academic and social demands and near-universal access to digital technologies. Approximately two-fifths of university students worldwide suffer from PIU, with robust associations with academic impairment, sleep disturbance, emotional dysregulation, depression and anxiety.^{9 10} This developmental period may also represent an important opportunity for intervention because maladaptive behavioural patterns remain relatively modifiable. University settings additionally provide a practical context for delivering scalable, group-based interventions.

Psychological and behavioural interventions currently constitute the principal treatment approaches for PIU.^11–15 Cognitive-behavioural therapy (CBT) is the most extensively studied modality and aims to modify maladaptive cognitions and behavioural patterns that maintain PIU.¹⁶ Evidence also supports the potential value of complementary approaches. Exercise-based interventions may reduce PIU severity through effects on stress regulation, reward processing and executive control.¹⁷ Mindfulness-based interventions and motivational interviewing may further strengthen self-regulation, insight and treatment engagement.¹⁸ These findings suggest that several intervention modalities may be beneficial and that their effects may operate through partly distinct psychological pathways.

Multimodal interventions have therefore emerged as a promising approach to PIU treatment. From a theoretical perspective, the heterogeneous nature of PIU suggests that no single mechanism is likely to account for the development and maintenance of problematic use across all individuals and contexts. Integrative frameworks such as the Interaction of Person–Affect–Cognition–Execution model conceptualise PIU as arising from dynamic interactions among individual predispositions, affective and stress-related processes, maladaptive cognitions and impairments in executive and behavioural control.^19–21 Within this view, interventions that simultaneously target multiple psychological processes may be better aligned with the multifactorial nature of PIU than single-component approaches.

From a practical perspective, engagement and attrition may limit the effectiveness of CBT-based interventions, particularly among individuals with low motivation or substantial emotional distress. Incorporating motivational, experiential and behavioural components may improve adherence and acceptability.²² Evidence from depression research further suggests that exercise may produce acute psychobiological effects, including short-term improvements in affective states and neuroplasticity, which may augment the effects of CBT.^{23 24} Consistent with this rationale, network meta-analyses indicate that combined or multicomponent interventions tend to rank among the more effective approaches, although the certainty of the available evidence remains limited by methodological weaknesses.^{14 15}

Despite the potential advantages of multimodal interventions, two important gaps limit their translation into standardised and precision-oriented practice. First, existing randomised controlled trials (RCTs) are commonly limited by small sample sizes.¹² This is particularly problematic given the pronounced heterogeneity of PIU. Inadequately powered studies cannot reliably determine whether treatment effects differ across PIU subtypes, such as problematic gaming, social media use or short-form video use, thereby limiting the generalisability and clinical interpretability of their findings.

Second, research on mechanisms of change remains limited. Most RCTs prioritise symptom reduction as the primary endpoint but do not examine whether treatment-related improvements are statistically accounted for by changes in specific psychological processes. This limitation constrains efforts to refine intervention components, identify potentially active ingredients and develop more mechanism-informed treatment strategies.

To address these limitations, we developed IMPROVE—Identify, Monitor, Plan, Real-world activities, Observe emotion, Vulnerability and Emotion regulation—a manualised, group-based, integrative psychotherapy programme for university students with PIU. IMPROVE comprises three core therapeutic components. First, CBT-oriented modules incorporating motivational elements aim to modify maladaptive internet-related cognitions and strengthen self-efficacy, thereby enhancing participants’ perceived control over their online behaviour. Second, mindfulness-based modules are designed to cultivate emotional awareness and non-reactivity, improve emotion regulation and reduce maladaptive attentional processes such as mind wandering. Third, socially interactive physical activity modules aim to increase engagement in rewarding offline experiences, potentially reducing anhedonia and perceived stress, both of which may contribute to excessive internet use as a maladaptive coping strategy. These candidate psychological processes were selected because they correspond to the principal components of the intervention, are potentially modifiable within a group-based format and are supported by previous theoretical and empirical research.^{11 17 18 25}

The present study describes the protocol for a single-blinded RCT comparing IMPROVE with an enhanced waitlist control condition among university students with PIU. The primary objective is to evaluate the effectiveness of IMPROVE in reducing PIU severity. Secondary objectives include examining whether early changes in maladaptive cognitions, self-efficacy, emotion regulation difficulties, mind wandering, resilience, anhedonia and perceived stress statistically account for subsequent changes in PIU outcomes.

Methods and analysis

Study design and timeline

This study is a single-blind RCT with two parallel arms, comparing the IMPROVE intervention with an enhanced waitlist control condition. Participants will be allocated in a 1:1 ratio after completion of the baseline assessment. The study procedures are shown in figure 1.

Figure 1. The study flowchart. IMPROVE, Identify, Monitor, Plan, Real-world activities, Observe emotion, Vulnerability and Emotion.

The primary objective is to evaluate the effectiveness of a multimodal, group-based psychotherapy for PIU among university students. Secondary objectives are to examine candidate psychological mechanisms of change and heterogeneity in treatment effects across PIU subtypes and baseline characteristics. The protocol is reported in accordance with the²⁶ Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2025 statement,²⁶ and the completed checklist is provided in online supplemental table S1.

The trial was prospectively registered with ClinicalTrials.gov (NCT07434778) before enrolment of the first participant. Recruitment and baseline data collection commenced in March 2026. Final 12-month follow-up assessments are expected to be completed by June 2028.

Participants and recruitment

The study is coordinated by Zhejiang University. Participants will be recruited from approximately 20–30 university sites through a combination of online and offline strategies. Each site is expected to recruit approximately 20–30 participants. Recruitment channels will include campus-wide advertisements, social media platforms, including WeChat and university forums, and referrals from university counselling centres.

Interested individuals will be directed to a brief online screening questionnaire. The questionnaire will include self-reported items corresponding to the prespecified eligibility criteria, including age, student status, recent substance use other than tobacco or alcohol, current or previous psychological treatment or counselling, severe physical or neurological conditions and any psychiatric disorder previously diagnosed by a healthcare professional. The 20-item Internet Addiction Test (IAT) will be administered to determine whether the eligibility threshold of 50 or higher is met. Participants will also complete a single item identifying the internet activity that they perceive as the most time-consuming, difficult to control or problematic during the previous 12 months. Response options will include gaming, online shopping, online pornography use, social media use, online gambling, short-form video use and other internet activities. No additional psychiatric symptom scales will be used to determine eligibility.

To be eligible for the study, participants must meet the following inclusion criteria: (1) be university students aged between 18 and 30 years; (2) report subjective suffering regarding their own internet use and score 50 or above on the IAT, indicating at least a moderate level of PIU; (3) own a personal smartphone or other electronic communication device to ensure accessibility to digital intervention components; and (4) provide written informed consent prior to enrolment.

Candidates will be excluded if they meet any of the following criteria: (1) a self-reported history of substance use (excluding tobacco and alcohol) within the past 12 months; (2) current or previous participation in any form of psychotherapy for PIU; (3) the presence of severe physical or neurological conditions (eg, organic brain diseases, history of traumatic brain injury or coma); or (4) a self-reported history of diagnosed psychiatric disorders (eg, schizophrenia, bipolar disorder). Depressive and anxiety symptoms assessed using the Patient Health Questionnaire-9 (PHQ-9) and Generalised Anxiety Disorder-7 (GAD-7) will not constitute exclusion criteria, and no prespecified threshold on either scale will be used to determine eligibility. Participants without a previous psychiatric diagnosis will therefore not be excluded solely on the basis of depressive or anxiety symptom severity. Participants whose screening responses indicate potentially high clinical risk will undergo an individual clinical review to determine whether trial participation is appropriate and whether additional clinical care is required.

Randomisation and blinding

After completion of the baseline assessment, participants will be randomly assigned in a 1:1 ratio to the IMPROVE intervention group or the enhanced waitlist control group using a centralised, computer-generated allocation sequence. No stratification factors will be used. The allocation process will be fully automated, and personnel responsible for enrolment will not have access to the allocation sequence.

The original protocol specified stratified randomisation according to sex, baseline PIU severity and primary PIU subtype. During early recruitment, the number of eligible participants within individual sites and several prespecified stratification cells was lower than anticipated, resulting in sparse strata and making the planned stratified allocation operationally impractical. The randomisation procedure was therefore amended to centralised 1:1 allocation without stratification. This amendment did not change the study arms or allocation ratio. Sex, baseline PIU severity, primary PIU subtype and other baseline characteristics will instead be examined in prespecified moderation analyses. The amended randomisation procedure received approval by the ethics committee and corresponding update of the trial registry.

Because of the nature of the psychotherapeutic intervention, participants and therapists cannot be blinded to treatment allocation. Outcome assessors will remain blinded to allocation, and data analysts will use masked group labels until completion of the primary analysis.

Intervention

Intervention group

Participants allocated to the intervention group will receive IMPROVE, a manualised, group-based, multimodal psychotherapy developed for university students with PIU. The programme integrates three therapeutic components: CBT-informed strategies incorporating motivational elements, mindfulness-based practices and socially interactive exercise. The intervention will be delivered in 8 weekly sessions.

Each session will last approximately 120 min and will be facilitated by two trained therapists, comprising a lead therapist and a co-therapist. Each group will include approximately 10–15 participants. Sessions will follow a standardised three-part structure: (1) group CBT (approx. 50 min): This component targets cognitive and behavioural processes associated with PIU. Early sessions focus on motivational enhancement, self-monitoring and identification of triggers and consequences. Later sessions address cognitive restructuring, emotion-related coping, behavioural planning and relapse prevention; (2) mindfulness training (approx. 30 min): This component targets attentional control and emotion regulation. Practices progress from focused-attention exercises, such as breath awareness and body scanning, to open-monitoring exercises involving emotions, cravings and intrusive thoughts. Participants are encouraged to adopt a non-judgemental and non-reactive stance towards internal experiences; (3) socially interactive exercise (approx. 40 min): This component consists of structured, team-based physical activities intended to increase engagement in rewarding offline experiences, promote social interaction, reduce sedentary behaviour and address stress and anhedonia.

The curriculum is organised around the acronym IMPROVE, followed by a concluding session. The content of each weekly session is presented in online supplemental table S2.

Enhanced waitlist control group

Participants allocated to the enhanced waitlist control (EWC) group will remain on the waitlist until completion of the 3-month follow-up assessment at T4. During the initial 8-week intervention period from T0 to T2, they will receive one standardised digital newsletter per week through WeChat, resulting in eight newsletters in total. Each newsletter will be delivered as a brief text-based message of approximately 100–120 Chinese characters and will require approximately 1 min to read. The weekly themes are: (1) awareness of habitual smartphone-use periods; (2) bedtime smartphone use and its potential effects on sleep; (3) situational and emotional triggers of smartphone use; (4) smartphone-related interruptions to attention; (5) establishing a brief phone-free period; (6) the relationship between emotional states and smartphone use; (7) re-engagement in offline activities; and (8) reviewing personal smartphone-use patterns and identifying a small area for future change. The complete content of the newsletters is provided in online supplemental table S3.

The newsletters will be identical for all participants and will contain general psychoeducational information and brief self-reflection prompts. They will be non-personalised and delivered in a one-way format. They will not include individualised assessment or feedback, therapist contact, group interaction, structured homework review or therapist-guided training in cognitive restructuring, motivational interviewing, mindfulness, emotion regulation, relapse prevention or socially interactive exercise. The enhanced waitlist condition is therefore intended to provide minimal informational contact and support participant retention without reproducing the active therapeutic components of IMPROVE.

Following completion of the 3-month follow-up assessment (T4), participants in the EWC group will be offered the full 8-week IMPROVE intervention.

All therapists will complete standardised training based on the IMPROVE manual before delivering the intervention. During the intervention period, weekly supervision meetings will be led by a senior clinician from the coordinating research team. Therapists will review the delivery of completed sessions, discuss implementation difficulties and deviations from the manual and receive feedback and guidance for subsequent sessions. Sessions will not be routinely audio-recorded, video-recorded or independently observed. No formal scored fidelity checklist or inter-rater reliability assessment is planned. Consistency of intervention delivery will therefore be supported through standardised therapist training, use of the intervention manual and regular clinical supervision.

Outcome measures

Measurement plans

To comprehensively evaluate the effectiveness and potential mechanisms of the IMPROVE intervention, assessments will be conducted at baseline (T0), mid-intervention at week 4 (T1), post-intervention at week 8 (T2; primary endpoint) and at 1-month (T3), 3-month (T4), 6-month (T5) and 12-month (T6) follow-ups. This assessment schedule will allow changes in clinical outcomes and candidate mediators to be tracked over time. The full measurement schedule is presented in table 1. Outcome data will be collected using standardised electronic questionnaires administered through Wenjuanxing, a Chinese online survey platform. To support participant retention and follow-up completion, reminders will be sent through WeChat before each assessment and modest compensation will be provided after completion of each follow-up assessment.

Table 1. Measurement plan.

Assessment phase	Screening	Baseline	Intervention	Post-intervention	Follow-up phase
Time point	–	T0	T1	T2	T3	T4	T5	T6
Timeline	Enrolment	Week 0	Week 4	Week 8	1-month	3-month	6-month	12-month
Administrative
Informed consent
Demographics and history	✓	✓
Inclusion/exclusion check	✓
Primary outcome
Symptom severity (IAT)	✓	✓	✓	✓	✓	✓	✓	✓
Secondary outcomes
Specific internet-use disorder symptoms (ACSID-11)		✓		✓		✓		✓
Smartphone addiction (SAS-SV)		✓		✓		✓		✓
Usage metrics (time)		✓	✓	✓	✓	✓	✓	✓
Craving intensity (QSU-Brief and VAS)		✓	✓	✓	✓	✓	✓	✓
Functional impairment (BNCS)		✓		✓		✓		✓
Depression (PHQ-9)		✓		✓		✓		✓
Anxiety (GAD-7)		✓		✓		✓		✓
Sleep quality (ISI)		✓		✓		✓		✓
ADHD symptom (ASRS V.1.1 check list)		✓		✓		✓		✓
Well-being (WHO-5)		✓		✓		✓		✓
QOL (VAS)		✓		✓		✓		✓
Physical activity (IPAQ-SF)		✓		✓		✓		✓
IGD severity (IGDS9-SF for gamers)		✓	✓	✓	✓	✓	✓	✓
Mechanisms (mediators)
Maladaptive cognitions (CAMCS for all participants and C-RIGCS for gamers)		✓	✓	✓		✓
Emotion regulation (DERS-16)		✓	✓	✓		✓
Self-efficacy (GSES)		✓	✓	✓		✓
Anhedonia (ASA-C)		✓	✓	✓		✓
Perceived stress (PSS-10)		✓	✓	✓		✓
Mind wandering (MWQ)		✓	✓	✓		✓
Resilience (CD-RISC)		✓	✓	✓		✓
Explanatory analysis
Participant satisfaction				✓
Motivation		✓
Readiness to change		✓
Blood sample		✓		✓
Structural and functional MRI		✓		✓

ACSID-11, Assessment of Criteria for Specific Internet-use Disorders-11; ADHD, attention-deficit/hyperactivity disorder; ASA-C, Anhedonia Scale for Adolescents-Chinese version; ASRS, ADHD Self-Report Scale; BNCS, Brief Negative Consequence Scale; CAMCS, Chinese Adolescents Maladaptive Cognitions Scale; CD-RISC, Connor-Davidson Resilience Scale; DERS, Difficulties in Emotion Regulation Scale; GAD-7, Generalised Anxiety Disorder-7; GSES, General Self-Efficacy Scale; IAT, Internet Addiction Test; IGDS9-SF, Internet Gaming Disorder Scale-Short Form; IPAQ-SF, International Physical Activity Questionnaire-Short Form; ISI, Insomnia Severity Index; MWQ, Mind Wandering Questionnaire; PHQ-9, Patient Health Questionnaire-9; PSS, Perceived Stress Scale; QOL, quality of life; QSU-Brief, Questionnaire on Smoking Urges-Brief; SAS-SV, Smartphone Addiction Scale-Short Version; VAS, Visual Analogue Scale.

Measurement tools

Internet-related outcomes

Internet Addiction Test

The IAT is a widely used 20-item self-report scale assessing the severity of PIU.²⁷ Items are rated on a 5-point Likert scale (1=not at all to 5=always), yielding total scores from 20 to 100. The IAT has demonstrated good validity in Chinese university populations.²⁸ A cut-off score of ≥50 indicates moderate-to-severe PIU, while scores >80 suggest severe PIU.

Smartphone Addiction Scale-Short Version

The Smartphone Addiction Scale-Short Version (SAS-SV) is a validated self-report scale containing 10 items designed to identify the level of smartphone addiction risk.²⁹ Items are rated on a 6-point Likert scale (1=strongly disagree to 6=strongly agree), covering domains such as daily-life disturbance, withdrawal and overuse. The SAS-SV will be administered at baseline (T0), post-intervention (T2), 3-month follow-up (T4) and 12-month follow-up (T6).

Internet Gaming Disorder Scale-Short Form

To specifically assess the severity of gaming disorder symptoms according to criteria of Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, the Internet Gaming Disorder Scale-Short Form (IGDS9-SF) will be administered. This 9-item unidimensional scale evaluates the core symptoms of gaming disorder (eg, preoccupation, withdrawal, tolerance, loss of control) over the past 12 months. Items are rated on a 5-point Likert scale ranging from 1 (‘Never’) to 5 (‘Very Often’), with total scores ranging from 9 to 45.³⁰ Higher scores indicate greater severity of gaming disorder. It has demonstrated excellent psychometric properties in Chinese populations. In this study, the IGDS9-SF will be administered only to participants who report recent gaming experience, and it will be assessed at every measurement point (T0–T6).

Assessment of Criteria for Specific Internet-use Disorders

The Assessment of Criteria for Specific Internet-use Disorders (ACSID-11) assesses PIU according to criteria of International Classification of Diseases, 11th Revision.³¹ In this study, only frequency scores will be used to evaluate four core diagnostic features: impaired control, increasing priority, continuation despite negative consequences and functional impairment. Assessments will occur at T0, T2, T4 and T6.

Internet usage behaviour

Internet use duration will be assessed using a single self-report item asking participants to estimate their average daily non-academic internet use during the previous week. This measure will be collected at all assessment points from baseline to the 12-month follow-up (T0–T6).

Craving symptoms

Subjective craving will be measured using a modified version of the Questionnaire on Smoking Urges-Brief (QSU-Brief) adapted for internet use.³² The 10-item scale assesses urges related to negative affect relief and reward seeking. In addition, a Visual Analogue Scale (VAS; 0=no craving to 100=extreme craving) will be used to assess current craving intensity. Both measures will be administered at all time points (T0–T6).

Functional impairment

Functional impairment will be assessed using an adapted Brief Negative Consequence Scale (BNCS).³³ This 6-item scale evaluates negative consequences of PIU across six domains: occupational, academic, interpersonal, family/intimate relationships, health and safety. Items are rated on a 4-point Likert scale (0=not at all to 3=severe) and will be administered at T0, T2, T4 and T6.

Psychopathology and well-being

Depressive symptoms

Depression will be assessed using the PHQ-9, which measures the nine criteria for major depressive disorder in Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition on a 4-point scale (0–3). The Chinese version has demonstrated excellent reliability and validity.³⁴ Assessments will occur at T0, T2, T4 and T6.

Anxiety symptoms

Generalised anxiety will be assessed using the GAD-7, a 7-item self-report scale with strong psychometric properties in Chinese samples.³⁵ It will be administered at T0, T2, T4 and T6.

Insomnia

Sleep problems will be measured using the Insomnia Severity Index (ISI), a 7-item scale assessing insomnia severity and functional impact over the past 2 weeks.³⁶ Items are rated on a 5-point Likert scale (0–4), with higher scores indicating more severe insomnia. The ISI will be administered at T0, T2, T4 and T6.

ADHD symptom assessment

The adult attention-deficit/hyperactivity disorder (ADHD) Self-Report Scale (ASRS V.1.1) symptom checklist will be employed to screen for symptoms of ADHD.³⁷ Developed in conjunction with the WHO, this 18-item instrument is designed to assess the frequency of ADHD symptoms consistent with DSM criteria. The scale comprises two components: Part A (6 items), which serves as a brief screener identifying the most predictive symptoms of the disorder and Part B (12 items), which provides additional symptom clarification. Participants rate the frequency with which they have experienced specific symptoms over the past 6 months on a 5-point Likert scale ranging from 0 (‘Never’) to 4 (‘Very Often’). Higher cumulative scores indicate greater ADHD severity. The ASRS V.1.1 is widely recognised for its strong psychometric properties, demonstrating high internal consistency and concurrent validity with clinician-administered assessments. It will be assessed at T0, T2, T4 and T6.

WHO-5 Well-Being Index

The WHO-5 is a short self-report measure of current mental well-being.³⁸ It consists of five items reflecting positive mood (eg, ‘I have felt cheerful and in good spirits’), rated on a 6-point Likert scale. It will be conducted at T0, T2, T4 and T6.

Physical activity levels

Baseline physical activity will be assessed using the International Physical Activity Questionnaire-Short Form.³⁹ This widely validated 7-item self-report instrument measures the frequency and duration of physical activity across three intensity levels—vigorous activity, moderate activity and walking—as well as time spent sitting, over the last 7 days. It will be conducted at T0, T2, T4 and T6.

Quality of life

Subjective quality of life will be assessed using a single-item VAS (QOL-VAS). This instrument consists of a 0–100 linear scale, anchored by ‘worst imaginable quality of life’ at 0 and ‘best imaginable quality of life’ at 100. Participants will be instructed to indicate their overall satisfaction with their current life situation by selecting the point on the scale that best reflected their subjective experience. It will be conducted at T0, T2, T4 and T6.

Mechanistic outcomes (mediators)

All candidate mediators will be assessed at baseline (T0), mid-intervention (T1), post-intervention (T2) and 3-month follow-up (T4) to support longitudinal mediation analyses.

Maladaptive cognitions

Maladaptive internet-related cognitions will be assessed using the Chinese Adolescents Maladaptive Cognitions Scale (CAMCS), a 12-item measure capturing three dimensions: social comfort, distraction and self-realisation.⁴⁰ For gamers, we will additionally assess the maladaptive cognition related to gaming using the Chinese version of Revised Internet Gaming Cognition Scale (C-RIGCS).⁴¹

General self-efficacy

General self-efficacy will be assessed using the 10-item General Self-Efficacy Scale (GSES), measuring perceived ability to cope with challenges and achieve goals.⁴²

Anhedonia

Anhedonia will be assessed using the Anhedonia Scale for Adolescents-Chinese version (ASA-C), a 14-item instrument sensitive to changes in reward processing and motivation.⁴³

Perceived stress

Stress levels will be measured using the 10-item Perceived Stress Scale (PSS-10), assessing the degree to which life situations are perceived as unpredictable and overwhelming.⁴⁴

Emotion regulation

Emotion regulation difficulties will be assessed using the 16-item Difficulties in Emotion Regulation Scale (DERS-16), covering multiple dimensions of emotion dysregulation.⁴⁵

Mind wandering

Attentional control will be assessed using the 5-item Mind Wandering Questionnaire (MWQ), measuring the frequency of task-unrelated thoughts.⁴⁶

Resilience

This study employs the 10-item Connor-Davidson Resilience Scale (CD-RISC 10), a widely validated psychometric instrument designed to measure stress coping ability and personal competence.⁴⁷

Participant satisfaction (feasibility and acceptability)

Participant satisfaction and acceptability will be assessed at post-intervention (T2) among participants who receive the IMPROVE intervention. A structured questionnaire adapted from the Client Satisfaction Questionnaire and supplemented with study-specific items will be used.⁴⁸ The questionnaire assesses overall programme quality, perceived helpfulness, the extent to which participants’ needs were met, willingness to recommend the programme or participate in a similar programme again, satisfaction with programme content and therapist guidance and the perceived helpfulness of each of the 8 weekly sessions and the three core intervention components. It also assesses the perceived appropriateness of the 120-minute session length, the 8-week programme duration, the amount of between-session practice and the group atmosphere.

General satisfaction, perceived helpfulness and session-specific and component-specific helpfulness are rated using 5-point response scales. Ratings of individual sessions include an additional response option indicating that the participant did not attend the session. Items assessing session length, programme duration, practice burden and group atmosphere use three-category response options. The complete questionnaire is provided in online supplemental table S4.

Demographic and other covariates

Demographics

Socio-demographic variables will include sex, age, ethnicity, level of study (undergraduate, master’s or doctoral) and marital status. Family and childhood characteristics will include only-child status, number and birth order of siblings where applicable, parental marital status and primary place of residence during childhood and adolescence, ranging from provincial capitals to rural areas. Participants will also report the perceived quality of their relationships with their mother and father and their subjective socioeconomic status.

Internet use patterns and primary addiction subtype

At baseline, participants will report the frequency with which they engage in specific online activities, including gaming, social media use, short-form video use and other internet-mediated activities, using a 5-point scale ranging from ‘never’ to ‘nearly every day’. Participants will then identify the single online activity that they perceive as the most time-consuming, difficult to control or problematic.

Internet use motivations

Internet use motivations will be assessed at baseline (T0) using a self-developed, multidimensional instrument grounded in previous literature. This 14-item scale evaluates four distinct motivational domains: (1) Social Connection and Acceptance (eg, ‘I use internet/smartphone to stay in touch with friends and family’); (2) Information Seeking and Learning (eg, ‘I often learn new knowledge and skills through internet/smartphone’); (3) Emotional Regulation and Escape (eg, ‘Internet/smartphone helps me relax when I feel stressed or anxious’); and (4) Entertainment and Pass Time (eg, ‘I engage in smartphone/internet to relieve boredom when I have nothing else to do’). Participants rate the extent to which each statement applies to them on a 5-point Likert scale ranging from 1 (‘Strongly Disagree’) to 5 (‘Strongly Agree’).

Readiness to change

We will employ a motivational ruler battery adapted from the motivational interviewing framework. This instrument consists of three distinct VAS ranging from 0 to 100. Participants are asked to rate: (1) importance: ‘How important is it for you to change your internet use behavior?’ (0=not important at all, 100=extremely important); (2) confidence: ‘If you decided to change, how confident are you that you would succeed?’ (0=not confident at all, 100=extremely confident); and (3) readiness: ‘How ready are you to start changing right now?’ (0=not ready at all, 100=fully ready).

Biological mechanisms (exploratory substudy)

Approximately 60 participants allocated to the IMPROVE intervention group will be invited to participate in an optional biological substudy. The substudy is designed to explore within-participant changes in neuroimaging and peripheral biological measures following the intervention and their associations with clinical improvement. Participation will be entirely voluntary and independent of participation in the main trial. Declining or withdrawing from the substudy will not affect access to the IMPROVE intervention or any other study procedures.

Participants who provide additional written informed consent will complete biological assessments at baseline (T0) and post-intervention (T2). Neuroimaging assessments will include structural and functional MRI, with a focus on resting-state functional connectivity and task-related neural activity associated with inhibitory control, emotion regulation and craving. Peripheral blood samples will also be collected, processed and stored in a biobank for future exploratory analyses of biomarkers potentially related to stress responses, neuroplasticity and inflammation, including neurotrophic factors and stress-related hormones.

All biological analyses will be exploratory and hypothesis-generating. Biological measures will not contribute to the primary or secondary efficacy outcomes and will not be included in the main trial analyses.

Outcome definitions

The primary outcome is the severity of PIU, assessed using the IAT. The primary estimand is defined as the between-group difference in change in IAT scores from baseline (T0) to the post-intervention endpoint (T2).

The secondary outcomes are as follows:

• Sustained and long-term efficacy: Between-group differences in IAT scores at the 1-month (T3) and 3-month (T4) follow-ups. Due to the waitlist crossover design, outcomes at the 6-month (T5) and 12-month (T6) follow-ups will be evaluated descriptively based on within-group symptom trajectories of the intervention arm relative to baseline.

• Clinical remission rate: Defined as the proportion of participants achieving ‘clinical remission’, operationally set as an IAT total score <50 at post-intervention (T2) and all subsequent follow-up points.

• Secondary symptom improvement: Defined as the between-group differences in change scores from baseline to the 3-month follow-up (T4) across multiple internet-related domains. These include: smartphone addiction risk (SAS-SV); PIU severity based on ICD-11 criteria (ACSID-11); IGD severity for those with problematic gaming (IGDS9-SF); subjective craving intensity (QSU-Brief and VAS); functional impairment in daily life (adapted BNCS); and daily internet usage duration derived from subjective estimations.

• Psychopathological improvement: Defined as the between-group differences in the reduction of comorbid psychiatric symptoms from baseline to the 3-month follow-up (T4). This encompasses depressive symptoms (PHQ-9), anxiety symptoms (GAD-7), insomnia severity (ISI), ADHD severity (ASRS), as well as the enhancement of general well-being (WHO-5), QOL and physical activity.

• Feasibility and acceptability: Feasibility and acceptability will be characterised using participants’ post-intervention ratings of overall programme quality, perceived helpfulness, needs fulfilment, willingness to recommend or participate again, satisfaction with programme content and therapist guidance, the helpfulness of individual sessions and intervention components, session and programme duration, between-session practice burden and group atmosphere.

Psychological mechanisms: Changes in the candidate psychological mediators from baseline (T0) to mid-intervention (T1) will be examined as proximal indicators of therapeutic change. These mediators include maladaptive internet-related cognitions (CAMCS), general self-efficacy (GSES), anhedonia (ASA-C), perceived stress (PSS-10), emotion regulation difficulties (DERS-16), mind wandering (MWQ) and resilience (CD-RISC 10). Among participants reporting recent gaming, gaming-related maladaptive cognitions will additionally be assessed using the C-RIGCS. Longitudinal mediation analyses will estimate indirect effects and examine whether early changes in these variables statistically account for subsequent changes in PIU severity. The findings will be interpreted as evidence consistent with the hypothesised mechanisms of change rather than as definitive evidence of causal mediation.

Sample size determination

The sample size was determined a priori to ensure robust statistical power for both the primary efficacy endpoint and independent exploratory analyses of specific PIU subtypes. Our unpublished pilot survey of 185 help-seeking university students revealed a prevalence distribution of 47.6% for short-form video use, 23.2% for online gaming and 18.4% for social media use. Based on recent meta-analyses indicating large effect sizes for multimodal interventions (Standardized Mean Difference >1.0),^{14 15} we anticipated a large effect size (Cohen’s d=0.8) for the primary outcome. A minimum of 34 evaluable participants per arm would be required to detect an effect size of Cohen’s d=0.8 with 90% power at a two-sided significance level of 0.05. To retain approximately 34 participants per arm within a PIU subtype expected to account for 20% of the sample, approximately 170 participants per arm would be required before allowing for attrition. Assuming an attrition rate of 30%, this number increases to approximately 243 participants per arm. We therefore set the final recruitment target at 300 participants per arm (total N=600) to provide additional allowance for uncertainty in subtype prevalence and follow-up attrition.

Statistical analysis

Descriptive and preliminary analyses will be conducted using SPSS V.27.0 (IBM, Armonk, New York, USA). Longitudinal, multilevel and mediation analyses will be conducted using R software (R Foundation for Statistical Computing, Vienna, Austria). All tests will be two-sided, with a nominal significance level of p<0.05 unless otherwise specified.

The primary analyses will follow the intention-to-treat principle and include all randomised participants irrespective of intervention attendance or protocol adherence. Per-protocol analyses will be conducted as sensitivity analyses and will include participants who attend at least 60% of the intervention sessions and complete the post-intervention assessment.

Continuous variables will be summarised using means and SDs or medians and IQRs, as appropriate. Categorical variables will be summarised using frequencies and percentages.

Primary outcome analysis

To appropriately account for the longitudinal and hierarchical data structure, linear mixed-effects models (LMMs) will be fitted using the lme4 package in R. The repeated outcome measurements are nested within individuals, who are in turn nested within participating study centres. Fixed effects will include group (IMPROVE vs EWC), time (categorical) and the group×time interaction, which constitutes the primary test of intervention efficacy. Random intercepts will be specified for both participants and study centres to account for within-subject correlation and centre-level clustering. Baseline IAT score as well as other baseline characteristics will be included as a covariate to improve precision and adjust for residual baseline imbalance. The primary contrast of interest will be the estimated marginal mean difference in change from baseline between groups at T2. Effect sizes will be reported as Cohen’s d, calculated from model-based estimated marginal means and pooled SDs, along with 95% CIs.

Based on the planned recruitment structure, approximately 20–30 university sites are expected to participate, with each site recruiting approximately 20–30 participants. Following 1:1 randomisation, each site is expected to contribute approximately 10–15 participants to the intervention arm, corresponding to one IMPROVE therapy group delivered by one trained therapist dyad. Under this implementation plan, study centre, therapist dyad and therapy group will be fully aligned and their effects will not be statistically separable. Therefore, the centre-level random intercept in the primary model will account for clustering attributable to the study site, therapist dyad and intervention group and a separate therapy-group random effect will not be included.

If recruitment circumstances require more than one intervention group to be delivered at any participating site, therapy group will no longer be fully confounded with centre. In that situation, the mixed-effects model will be expanded to include a random intercept for therapy group nested within centre. This modification will be specified and documented before the analysis of trial outcomes.

Secondary outcome analysis

Secondary continuous outcomes will be analysed using the same LMM framework as described for the primary outcome. For binary outcomes (eg, remission status), generalised linear mixed models with a logit link function will be employed. Results will be reported as ORs with 95% CIs. To address multiplicity arising from multiple secondary outcomes, the Benjamini-Hochberg procedure will be applied to control the false discovery rate. Participant satisfaction and acceptability outcomes, assessed post-intervention, will be summarised descriptively.

Heterogeneity and moderation analysis

To examine treatment effect heterogeneity and identify subgroups with differential responses, prespecified moderation analyses will be conducted by extending the mixed-effects models to include three-way interaction terms (group×moderator×time). Moderators will include socio-demographic variables (eg, gender, education level), clinical characteristics (baseline PIU severity: moderate (IAT 50–79) vs severe (IAT ≥80); primary PIU subtype: gaming, social media, short-form video or other) and continuous variables such as baseline motivation, physical activity level and readiness to change and coexisting mental health conditions (such as anxiety, depression and ADHD severity). All moderators are prespecified to minimise the risk of data-driven inference.

Mechanism analysis (longitudinal mediation)

To examine the potential psychological pathways through which the IMPROVE intervention may influence PIU severity, longitudinal mediation analyses will be conducted using structural equation modelling implemented in the lavaan package in R. Group assignment (IMPROVE vs EWC) will serve as the independent variable (X). Changes in the hypothesised mediators from baseline (T0) to mid-intervention (T1)—including maladaptive internet-related cognitions (CAMCS), anhedonia (ASA-C), emotion regulation difficulties (DERS-16), self-efficacy (GSES), perceived stress (PSS-10), mind wandering (MWQ) and resilience (CD-RISC 10)—will serve as mediators. Among participants reporting recent gaming, C-RIGCS scores will additionally be examined as a gaming-specific mediator and change in IGDS9-SF score will be examined as an additional outcome. Indirect effects will be estimated using bias-corrected bootstrap resampling with 2000 iterations and reported with 95% CIs. This temporal sequence will allow us to examine whether early changes in the candidate mediators statistically account for subsequent improvements in PIU severity. However, because the mediators are not randomised, the estimated indirect effects will be interpreted as evidence consistent with the hypothesised mechanisms of change rather than as definitive evidence of causal mediation.

Exploratory analyses of treatment outcomes

Exploratory analyses will examine whether baseline characteristics, including socio-demographic factors, motivation and readiness to change and primary PIU subtype, are associated with treatment efficacy, clinical remission and study attrition. For treatment efficacy, each baseline characteristic will be examined as a potential moderator by adding a group×time×predictor interaction term to the mixed-effects model. Clinical remission status and study attrition will be analysed using multivariable logistic regression models. Associations will be reported with effect estimates and 95% CIs and interpreted cautiously given the exploratory nature of these analyses.

Exploratory prediction analysis (dropout and response)

Exploratory analyses will be performed to identify predictors of attrition and treatment response. First, to identify risk factors for dropout, baseline demographic and clinical characteristics will be entered into a multivariate logistic regression model to predict study non-completion. Second, to identify predictors of treatment efficacy within the intervention group, linear regression models (or machine learning approaches such as Lasso regression, depending on data dimensionality) will be employed to examine the association between baseline profiles (eg, motivation subtypes, comorbidity) and the magnitude of symptom reduction at T2. These analyses aim to inform future precision medicine approaches for PIU.

Exploratory biological analyses

Analyses of the biological substudy will be restricted to participants assigned to the IMPROVE intervention group who provide additional consent and complete the relevant assessments. Changes in neuroimaging and peripheral biological measures from T0 to T2 will be examined, together with their associations with changes in clinical outcomes. Baseline characteristics of substudy participants and non-participants within the intervention group will be compared descriptively to characterise potential selection bias. These analyses will be considered exploratory.

Handling of missing data

Missing data will generally be handled using full information maximum likelihood estimation under the missing at random assumption, which uses all available data to produce unbiased estimates. Sensitivity analyses using multiple imputation (creating 10 imputed datasets) will be performed.

Safety and adverse events

IMPROVE is a non-pharmacological, group-based psychological intervention and is considered to present minimal risk. Adverse events are expected to be uncommon and generally mild; however, participants will be monitored for potential adverse events, including temporary exacerbation of psychological distress.

Participants whose questionnaire responses or clinical presentation indicate potentially high clinical risk, including severe psychological distress or suicidal ideation, will be reviewed individually by a qualified clinician from the research team. Where clinically indicated, participants will be advised or referred to receive individual psychiatric or psychological assessment and treatment. Decisions regarding enrolment or continued participation will be made on a case-by-case basis according to clinical need and safety considerations. Participation in, withdrawal from, or exclusion from the trial will not restrict access to appropriate clinical care.

Participants may withdraw at any time without penalty. Participants allocated to the enhanced waitlist control group will retain access to usual care and will be offered IMPROVE after completion of the waiting period. Given the minimal-risk nature of the intervention, no formal data monitoring committee, interim efficacy analyses or prespecified stopping boundaries are planned.

Participants who experience study-related harm will receive prompt assessment and appropriate treatment. Reasonable treatment costs for harms considered by the investigators to be related to study participation will be covered in accordance with institutional policy. No additional financial compensation beyond such treatment costs is planned.

Ethics and dissemination

The study was approved by the Ethics Committee of Sir Run Run Shaw Hospital, Zhejiang University School of Medicine on 12 February 2026 (approval no. 2026–0152). The amendment to the randomisation procedure described in Section 2.3 was reviewed and approved by the Ethics Committee of Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, and the revised procedure has been incorporated into the study protocol. The trial will be conducted in accordance with the Declaration of Helsinki.

Before enrolment, eligible participants will receive a comprehensive explanation of the study objectives, procedures, potential risks and benefits, confidentiality protections and their rights as research participants. Written informed consent will be obtained by trained research staff before any study procedures are undertaken. The original Chinese participant information sheet and consent form, together with their English translations and the optional biological substudy addendum, are provided in online supplemental file 4. Participants invited to the biological substudy will provide additional written informed consent specifically covering neuroimaging, peripheral blood collection and biological sample storage.

Participants will be informed that participation is voluntary and that they may withdraw at any time without academic, clinical or other adverse consequences. Important protocol amendments will be submitted to the ethics committee for approval and updated in the trial registry before implementation, where required.

All participant data will be de-identified and assigned a unique study identifier. The linkage file containing directly identifying information will be stored separately on a secure, password-protected server and will be accessible only to the principal investigator and study coordinator. The principal investigator and lead statistician will have access to the final trial dataset. Data entry and management will be conducted using a secure electronic database with prespecified range and logical consistency checks.

De-identified individual participant data underlying the primary and secondary trial publications, together with the data dictionary, study protocol, statistical analysis plan and analytical code, will be made available on reasonable request to the principal investigator. Data will be available from publication of the primary trial results for 5 years. Requests must include a scientifically sound research proposal and statistical analysis plan and will be reviewed by the principal investigator and study team. Access will be subject to participant consent, applicable ethical and data-protection requirements and execution of a data-use agreement. Approved data may be used only for the analyses specified in the accepted proposal.

Findings will be disseminated through peer-reviewed publications and presentations at academic conferences. Authorship will be determined in accordance with International Committee of Medical Journal Editors criteria. The funder has imposed no restrictions on publication.

Patient and public involvement

Patient and public involvement was incorporated at the formative and feasibility stages of the study. Prior to finalising the intervention protocol, informal consultations and pilot surveys were conducted with help-seeking university students experiencing PIU. In addition, participant perspectives will continue to be incorporated through post-intervention satisfaction assessments, which will be used to inform future refinement and dissemination of the intervention.

Discussion

This protocol describes a multicentre RCT evaluating the IMPROVE programme for university students with PIU. The trial was designed in response to several limitations of the existing intervention literature, including relatively small samples, limited consideration of heterogeneity across PIU patterns and infrequent prospective assessment of candidate mechanisms of change. IMPROVE combines CBT-informed strategies incorporating motivational elements, mindfulness training and socially interactive exercise within a standardised group-based format. The intervention therefore addresses several psychological processes implicated in PIU, including maladaptive cognitions, self-efficacy, emotion regulation, attentional control, anhedonia and perceived stress.

Several features of the trial are relevant to the interpretation of the study. The planned sample size permits prespecified examination of heterogeneity in treatment effects across PIU subtypes and baseline characteristics. Repeated assessment of candidate mediators at baseline, mid-intervention, post-intervention and follow-up allows the temporal relationships between early psychological changes and subsequent changes in PIU severity to be examined using longitudinal mediation models. Because the mediators are not randomised, these analyses are intended to evaluate whether the observed data are consistent with the hypothesised mechanisms of change rather than to establish causal mediation. The multicentre group-based design also allows the intervention to be implemented across different university settings using a common treatment manual and therapist training procedures.