We used an explanatory sequential mixed methods design. Survey findings informed the interview guide and purposive sampling, and integration was conducted through a prespecified comparison framework and meta-inferences based on convergence, complementarity, and discrepancies. Integrated findings are reported using narrative weaving that links quantitative results with qualitative themes and illustrative quotes. The reporting of the quantitative component follows the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guideline, and the qualitative component was designed, conducted, and reported with reference to the COREQ (Consolidated Criteria for Reporting Qualitative Research). From October 2022 to May 2024, consecutive patients with PNs were recruited via a thoracic surgery e-consultation clinic on a national telemedicine platform [14]. In total, 701 participants completed the structured survey. Eligibility criteria were newly identified or undersurveillance PNs, age ≥18 years, and ability to complete the questionnaire. We excluded individuals who had already undergone resection or had a definitive pathological diagnosis.

To complement and contextualize the quantitative findings, we conducted semistructured interviews with a purposive, maximum-variation sample of 10 patients and 10 physicians. Patient participants varied by age (26‐73 y), sex, and educational level. Physician participants were thoracic surgeons spanning junior, mid-career, and senior ranks.

This study was approved by the Research Ethics Committee of Guangdong Provincial People’s Hospital (approval number: 2025-KY-122‐01). All participants provided written informed consent under institutional ethics approval. All questionnaire and interview data were deidentified before analysis, and access was restricted to the research team. The research team collected questionnaire responses and qualitative interview materials; we did not collect, download, store, or process participants’ DICOM files or other raw radiology image files as part of the study procedures. Participants received no compensation for participation.

In China, CMI is commonly obtained or shared as DICOM files viewable in a browser, QR codes appended to radiology reports, or links on hospital websites. For QR code or web link access, the code or link typically directs patients to an institution-hosted or vendor-hosted web viewer. Depending on the implementation, the viewer may be DICOM-based (including zero-footprint viewers) or may rely on proprietary web formats; therefore, QR or link access should not be assumed to be DICOMweb-compliant. Some patients receive only a textual report and/or film, with no CMI access. We defined use of CMI as the self-reported ability to view radiological images on a mobile device. This measure does not differentiate between passive access and more active or intensive use, nor does it quantify frequency or depth of use. Sociodemographic and clinical variables included age, sex, education (primary, middle school, high school, junior college, undergraduate, and graduate), and surveillance status. Care-seeking and cost measures comprised the number of hospitals used (internet and offline), the total number of physicians consulted, and annual out-of-pocket expenditures (care-related fees, transportation, lodging). Self-efficacy and disease perceptions included perceived disease severity, level of disease awareness, and change in anxiety level.

We developed separate interview guides for patients and physicians (Multimedia Appendix 1). The guides covered CMI awareness and use, preferences of CMI format, perceived clinical benefits and limitations, workflow usability and pain points, psychological responses, and priorities for system improvement. Trained qualitative researchers conducted in-person interviews (20‐40 min). With permission, sessions were audio-recorded and accompanied by field notes. Recordings were transcribed verbatim, deidentified, and stored on password-protected devices. Texts were managed and queried in NVivo (version 11; QSR International Pty Ltd). Bilingual researchers translated quotations for manuscript presentation with dual review; discrepancies were resolved by revisiting the Chinese original. Additionally, interviews and analysis proceeded iteratively. During concurrent coding, we tracked the emergence of new codes and concluded interviewing when successive interviews no longer generated new major themes and the coding framework had stabilized.

After 1:1 propensity score matching, baseline balance was achieved, the main directional differences observed prematching generally persisted (Table S1 in Multimedia Appendix 2) and the covariate balance before and after propensity score matching persisted in Multimedia Appendix 3 . In the matched cohort, as shown in Figure 1A, the distribution for number of internet hospitals (median 1, IQR 0‐1 vs median 1, IQR 0‐1; P=.01) used differed significantly between users and nonusers, whereas, as shown in Figure 1B, the number of offline hospitals (median 2, IQR 1‐2 vs median 1, IQR 1‐2; P=.19) did not differ. The number of physicians consulted (median 3, IQR 1‐3 vs median 2, IQR 1‐3; P=.02) is also displayed in Figure 1C. CMI users were significantly more likely to spend 0‐2500 Chinese Yuan (CNY; US $1=6.87 CNY) on treatment and diagnostic costs (229/291, 78.7% vs 201/291, 69.1%; P=.004) and on travel and accommodation expenses (249/291, 85.6% vs 229/291, 78.7%; P=.03), suggesting they incurred lower expenses compared to nonusers. As presented in Figure 1F-H, regarding disease perceptions, users reported higher levels of disease awareness (overall P=.009), whereas perceived disease severity (overall P=.97) and anxiety change (overall P=.81) did not differ (Figure 1).

Among CMI users, in Figure 2A, the DICOM group differed from the QR or link group in distributions of the number of offline hospitals (median 2, IQR 1‐2 vs median 2, IQR 1‐2; P=.04) and the number of physicians consulted (median 3, IQR 2‐4 vs median 3, IQR 1‐3; P=.02) as depicted in Figure 2B, whereas as shown in Figure 2C, the number of internet hospitals (median 1, IQR 0‐2 vs median 1, IQR 0‐2; P=.99) did not differ. The DICOM group had no difference in treatment and diagnostic costs (95/132, 72.0% vs 209/261, 80.1%; P=.09) and travel and accommodation expenses (108/132, 82.6% vs 223/261, 85.4%; P=.52) with the majority spending 0‐2500 CNY, but higher disease knowledge (overall P=.004) as presented in Figure 2G. The groups did not differ in perceived severity (overall P=.17) or anxiety change (overall P=.19; Figure 2).

Across both strands, we observed a consistent possession-to-use gap: most participants obtained CMI, but only a subset reported using it. The interview data indicate that this gap reflects not only access but also day-to-day usability barriers and system limitations, including burdensome authentication steps, inconsistent workflows across hospitals, and short retention periods that make access links unreliable. After matching, the observed patterns in health care use and self-reported costs aligned with the convenience and efficiency benefits described in interviews, while psychological responses to viewing images varied in direction. A joint display linking key quantitative results with qualitative themes and illustrative quotes is provided in Table S2 in Multimedia Appendix 4.

This mixed methods study shows that most hospitals can provide CMI in varied formats, yet a substantial possession-to-use gap persists. Compared with nonusers, CMI users engaged more with internet hospitals and physicians, reported lower care-related and travel or lodging costs, and self-reported higher disease knowledge. Qualitative findings pinpointed system usability, retention, timeliness, and interoperability as principal limitations. Overall, our results align with recent evidence on the accessibility and cost advantages of internet hospitals [6] and the association between interoperability and reduced duplicate imaging [5], offering practical directions for improving workflows.

Older age was independently associated with lower CMI use, whereas education was not significant after adjustment. This aligns with reviews showing that older adults face barriers with digital health tools—account/identity verification complexity, dense information, jargon, input burden, and screen design can all hinder adoption and sustained use [15,16]. Our interviews echoed concrete obstacles, such as multiplatform authentication, link expiry, and loading instability. For this population, the most effective improvements are those that reduce usability barriers rather than simply expanding access. Examples include single sign-on with unified entry points, clinic-side preloading combined with direct viewer access, caregiver proxy accounts, and concise instructional micro-guides, all of which have been shown to enhance eHealth participation among older adults [17-19].

Quantitatively, users interacted with more internet hospitals and physicians, while the number of offline hospitals did not change, suggesting that CMI may improve the efficiency and quality of care pathways rather than simply reducing in-person visits. Lower patient-reported costs for users are consistent with the reduced service frequency and improved accessibility observed with internet hospitals [6]. From a continuity-of-information perspective, integrated longitudinal viewers and HIE platforms reduce repeat imaging and patient burden [5], aligning with our cost-direction findings. Although users reported greater disease knowledge, perceived severity and anxiety did not differ, echoing prior work that immediate online access can increase knowledge but may sustain “scanxiety” in the absence of explanatory support [20]. Qualitative findings further contextualize these patterns by illustrating how CMI facilitates more efficient remote consultations and transforms surveillance into a continuous trajectory. These insights help interpret the observed cost reductions as a direct result of fewer unnecessary repeat scans and improved clinical efficiency during cross-institutional visits. The qualitative theme of limited retention and poor interoperability may also help explain the possession-to-use gap observed in our quantitative results, that even when patients have access to CMI, short retention periods and cross-institution barriers can reduce the practical usefulness of CMI and discourage actual use. Finally, the qualitative theme of bidirectional psychological effects explains why increased knowledge may not lower anxiety, as patients reported feeling more in control while simultaneously feeling confused by complex terminology in the absence of professional translation.

Stratification by access modality indicated advantages for DICOM over QR or link access in engagement and knowledge—plausibly due to DICOM’s interactivity, longer-term retention, and standardization, which facilitate time-series comparison and remote consultation. Technically, mature DICOMweb workflows support browser-based import and zero-footprint viewing suited to both lightweight patient access and professional clinical review [9]. At a systems level, better regional informatics and interoperability correlate with improved health service performance [21].

Our interviews highlighted short retention periods, inconsistent rules, and fragile links and QR codes as major obstacles, which undermine in-clinic retrieval. Consistent with research on patient portals, stable and renewable access points with clear audit trails and proactive expiry reminders are foundational for practical use, especially among older adults [22]. Clinically, immediate simultaneous access to prior images during the visit is critical. Rather than adding new entry points, default patient-centered retrieval organized by patient, chronological timeline, and imaging series, together with online comparison, should be standard capabilities. Workflows based on DICOMweb and zero footprint viewers already enable standardized retrieval and reliable online comparison without local installation [8], and access to prior images reliably reduces repeat testing and delays [23].

Our findings suggest that interoperability and durable access in CMI are not only technical issues but also governance issues that require consistent rules for minimum retention periods, renewable access, audit trails, and accountability across institutions [24]. China’s national health informatization agenda has emphasized standardization and interoperability maturity certification, which provides an actionable foundation to operationalize imaging-specific retention and exchange requirements at scale [25]. Building on recent analyses that identify standards, interoperability, legislation, and compliance as the core foundations of China’s digital health governance for the Healthy China 2030 initiative, the clinical management pathway for pulmonary nodule surveillance can be strengthened by aligning platform requirements with these governance elements [24,26]. Finally, equity-oriented implementation is needed to ensure that national digital health strategies translate into broadly accessible longitudinal imaging services rather than widening existing digital divides [26].

This cross-sectional design precludes causal inference. Visit counts, costs, and disease perceptions were self-reported and thus subject to recall and reporting bias. Recruitment through online and hospital settings may have overrepresented digitally literate individuals, limiting external validity. Consequently, the 57.6% usage rate likely represents an upper bound for this population, and the lack of regional stratification may mask even steeper barriers in underconnected or rural areas. Therefore, our findings are most applicable to digitally connected patients with PNs who engage in cross-institutional care and remote consultation and should be extrapolated cautiously to offline-only pathways and underserved settings. Although propensity score matching mitigated confounding, residual confounding cannot be excluded. The measure of CMI use does not distinguish passive access from more active use, and we did not capture frequency or depth of use. The qualitative sample was modest and did not include participant checking. Future studies could improve construct validity by incorporating intensity indicators ideally complemented by objective usage logs where feasible. Future multisite sampling, longitudinal, or quasi-experimental studies are also warranted to validate these findings and assess long-term adoption trajectories.

In PN management, CMI is widely available yet remains underused. Actual use was associated with greater engagement in online care, lower related costs, and higher self-reported disease knowledge. Qualitative findings identified 3 actionable constraints—usability thresholds, unstable data retention, and inconsistent clinical viewing functions. Addressing these limitations requires ensuring renewable, auditable, and stable access, together with default support for longitudinal, in-browser comparison.