AI Synthesis: Qualitative analysis of food delivery riders

According to the Frontiers in Public Health review by Xu et al. (2026), platform-based food delivery in China creates intertwined occupational risks that require contextual, multi-source evidence synthesis. This post is for qualitative researchers and UX or occupational-health teams who need to translate the Frontiers (2026) narrative review into rigorous, actionable qualitative analysis. The primary keyword for this guide is "qualitative analysis of food delivery riders" and the payoff is concrete: methods to extract themes, codework, and cross-segment evidence from interviews, platform texts, and the review’s 47-source evidence base so you can prioritize interventions and measure impact.

Key Takeaways

According to the Frontiers in Public Health review by Xu et al. (2026) Frontiers in Public Health, occupational injuries and health risks among Chinese food delivery riders are multi-dimensional and driven by platform rules, road environments, long hours, and limited protections.

The review searched literature from January 1, 2015 to March 31, 2026 and included 47 eligible studies, as reported by Xu et al. (2026).
A Guangzhou study cited in the review reported a median weekly working time of 63 hours and that 70.1% of riders worked at least 55 hours per week (He et al., 2023).
The review found prevalences of occupational stress (30.1%), depressive symptoms (27.5%), insomnia (34.7%), and cumulative fatigue (40.8%) in the Guangzhou sample (He et al., 2023).
A cross-sectional survey of 1, 050 riders in Shenyang reported a 31.33% prevalence of somatic symptoms including low-back and knee pain (Bai et al., 2025).

What happened and how the Frontiers review was done

The Frontiers in Public Health narrative review by Xu et al. (2026) synthesized studies about platform-based food delivery work in China, searching PubMed, Web of Science Core Collection, and CNKI for publications from January 1, 2015 to March 31, 2026.

Xu et al. (2026) screened 1, 023 initial records and, after exclusions, included 47 eligible studies; the authors noted that most evidence was cross-sectional and relied heavily on self-reported measures.

Xu et al. (2026) conclude that "risky riding should not be explained simply as a lack of personal safety awareness, " and that risks should be framed as occupational and organizational in origin.

Findings Snapshot

Date or Study	Metric	Value	Implication (from Xu et al., 2026)
Search period (Xu et al., 2026)	Literature window	Jan 1, 2015; Mar 31, 2026	Narrative synthesis, 47 eligible studies; limits causal inference
He et al. (2023) cited in Xu et al. (2026)	Median weekly working time	63 hours (median); 70.1% ≥55 h/week	Long hours linked to stress, fatigue, and injury risk
He et al. (2023) cited in Xu et al. (2026)	Prevalence: stress / depression / insomnia / fatigue	30.1% / 27.5% / 34.7% / 40.8%	High mental-health and sleep burdens among riders
Bai et al. (2025) cited in Xu et al. (2026)	Sample size and somatic symptom prevalence	n = 1, 050; 31.33% reported somatic symptoms	Chronic physical complaints common in delivery work
Xu et al. (2026)	Number of included studies	47 studies	Evidence is heterogeneous; calls for longitudinal/intervention research

Implications for qualitative analysis of food delivery riders

For qualitative researchers, the Frontiers in Public Health review by Xu et al. (2026) implies that data collection should capture platform rules, temporal pressure, and contextual road environments as core analytic domains.

Xu et al. (2026) identify gaps including the predominance of cross-sectional designs and the heavy reliance on self-report, so qualitative work should collect time-stamped diaries, experience-sampling interviews, and platform-text artifacts to triangulate claims.

Xu et al. (2026) recommend studying heterogeneity across full-time, crowdsourced, and part-time riders; qualitative sampling should therefore stratify by employment model, city traffic density, and vehicle type.

How Evidano Helps

Problem: multi-source evidence is hard to synthesize

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents.

According to the Frontiers review (Xu et al., 2026), evidence on riders spans interviews, surveys, platform rules, and observational studies; Evidano ingests those document types to create a single coded corpus.

Solution mapping: Problem: scattered texts and time-stamped data → Evidano feature: document ingestion, transcript alignment, and unified thematic coding (see Evidano features).

Problem: need to triangulate self-report with platform/temporal data

The Frontiers review (Xu et al., 2026) highlights reliance on self-report; Evidano supports multi-source linkage so qualitative notes, platform logs, and survey free-text can be cross-analyzed.

Solution mapping: Problem: triangulation and cross-segment frequency analysis → Evidano feature: cross-segment thematic and frequency analysis and AI chat over your documents.

Problem: scalable transcription and privacy concerns

Xu et al. (2026) call for richer longitudinal and intervention data; Evidano provides encrypted transcription with PII redaction and custom dictionaries to process large numbers of interviews securely (see Speech-to-Text and Data Security).

Solution mapping: Problem: scale and security → Evidano feature: secure, automated transcription with custom dictionaries and privacy controls.

FAQ: qualitative analysis of food delivery riders

How should qualitative researchers sample riders after reading the Frontiers (2026) review?

Answer: Stratify sampling by employment type, hours worked, and city context, as recommended by Xu et al. (2026).

Supporting detail: Xu et al. (2026) note differences among dedicated, crowdsourced, and part-time riders and call for subgroup analyses; sample targets should therefore include riders with high platform dependence, riders working ≥55 hours per week, and riders operating in dense urban cores.

What qualitative methods best test the "platform rules → time pressure → injury" pathway?

Answer: Combine semi-structured interviews with experience sampling and platform-log artifact analysis, following the gaps identified by Xu et al. (2026).

Supporting detail: Xu et al. (2026) point to the need for longitudinal and mixed-method work; collect repeated micro-interviews during shifts, capture screenshots of app assignment rules, and pair narratives with time-stamped order and online-time records.

Can AI help code and surface causal-looking patterns in qualitative data about riders?

Answer: Yes, AI-enabled qualitative platforms can accelerate coding, extract co-occurrence patterns, and surface temporal associations for human validation, consistent with the Frontiers review’s call for integrated data sources (Xu et al., 2026).

Supporting detail: Xu et al. (2026) recommend combining platform order data, online working time, trajectory and wearable data; AI-assisted thematic and cross-segment analyses make that hybrid synthesis feasible at scale.

What ethical safeguards should qualitative teams apply when analyzing rider data?

Answer: Use informed consent, clear opt-out options, and data minimization for platform logs and health screening, as recommended by Xu et al. (2026).

Supporting detail: Xu et al. (2026) emphasize transparency and privacy safeguards for health monitoring; teams should anonymize identifiers, store data encrypted, and separate operational surveillance from health monitoring.

Conclusion & Next Steps

The Frontiers in Public Health narrative review by Xu et al. (2026) shows that occupational injuries and health risks for Chinese food delivery riders are shaped by platform rules, road environments, long hours, and protection gaps, and that evidence is dominated by cross-sectional studies.

Qualitative researchers should operationalize the review’s recommendations by collecting multi-source, time-stamped, and subgroup-stratified data and by using AI-enabled synthesis to test plausible pathways such as "platform rules → time pressure → fatigue → risky riding → injury" (Xu et al., 2026).

To convert multi-format interviews, platform artifacts, and surveys into prioritized findings and visualizations, try a platform designed for qualitative synthesis such as Evidano; see Evidano features for capabilities and Evidano data security for privacy assurances.

Ready to operationalize these methods on your study data? Try Evidano for free.

Topics

qualitative analysis of food delivery riders
AI-enabled qualitative research
food delivery rider health China

Keep reading

Browse all articles