JMIR Publications

ABSTRACT

Background:

Patient-pathway visualisation increasingly supports clinical decision-making, yet it depends fundamentally on the clinical coding systems, principally the International Classification of Diseases (ICD) and the Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) that structure the underlying data. Despite widespread adoption, no prior systematic review has examined how the choice of coding system influences the effectiveness and clinical utility of pathway visualisations

Objective:

To systematically synthesise how standardised clinical coding systems are combined with visualisation techniques to represent patient pathways; to determine which coding–visualisation combinations have been clinically validated; and to identify the principal interoperability and evidence gaps bearing on implementation.

Methods:

Following the PRISMA 2020 statement, we searched three bibliographic databases (PubMed, Science Direct, ACM Digital Library), with Google Scholar as a supplementary source, for studies published January 2014 to December 2025. Eligible studies combined a standardised coding system with patient-pathway representation; visualisation was the analytic focus. Of 43,539 records screened, 40 studies met the inclusion criteria and passed a 10-item quality checklist (minimum 7/10). We extracted coding system(s), visualisation method(s), analytic approach, and evaluation strategy, and cross-tabulated coding system against visualisation type and validation status.

Results:

ICD-based coding dominated, appearing in 36 of 40 studies (90%); timeline, flow (including Sankey), and node-link visualisations were most common, typically for single-disease pathways. SNOMED CT appeared in only 4 studies (10%), in every case alongside ICD rather than independently; no study clinically compared ICD with SNOMED CT. Multiple coding systems were combined in 11 studies, but for data coverage rather than unified representation, and none was clinically validated. Notably, only 35 of 40 studies presented a substantive visualisation where 5 represented coded pathways with no visual layer, itself evidence of the visualisation gap. Formal user-centered or clinician evaluation was rare (≈3 of 40, ≈8%), while 9 studies reported no evaluation metrics at all. We found no randomised comparisons, no real-time deployments, and no study quantifying the ICD-9-to-ICD-10 transition’s effect on visualisation.

Conclusions:

The evidence base is dominated by ICD-coded, single-disease pathway visualisations and offers only limited, largely invalidated support for SNOMED CT or multi-coding approaches. The field’s central unmet need is not more visualisation techniques but rigorous clinical validation and a solution to cross-terminology integration. Priority directions include comparative effectiveness studies of ICD versus SNOMED CT, standardised user-centered evaluation, and multi-site validation; developers should prioritise validated ICD-10 approaches with local pilot testing. Clinical Trial: This review was not registered