The 21-day median diagnosis gap is a systems failure - and it has a specific, fixable architecture.
The median time from first suspicious clinical finding to confirmed cancer diagnosis in the United States is 21 days for colorectal cancer. For lung cancer, the median is 35 days. For pancreatic cancer - the cancer where time matters most urgently, given its rapid progression - the median is 29 days. These are not edge cases. They are the typical experience for hundreds of thousands of patients diagnosed each year.
Each of those days carries a probability cost. For cancers that double in volume on timescales of weeks to months, a three-week diagnostic delay is not merely an inconvenience. It is a staging risk. An analysis published in BMJ in 2020 estimated that for colorectal cancer, a four-week diagnostic delay was associated with a 6% increase in relative mortality risk. For lung cancer, the estimate was 14% per four weeks. These are population-level estimates with wide confidence intervals, but the direction of effect is consistent across cancer types and study designs: time in the diagnostic gap has a cost.
The question is not whether to reduce diagnostic delays - that argument is settled. The question is where the delays are actually occurring and which interventions address the root causes rather than optimizing around them.
Diagnostic delay is not a monolithic phenomenon. It is a compound of several distinct delays, each with different causes and different solutions. Understanding the anatomy of the delay is prerequisite to reducing it.
Data assembly delay. The first and most pervasive bottleneck is the time required to bring together all the information relevant to a diagnostic decision. Laboratory results from one system, imaging reports from another, pathology from a third, prior medical history from a legacy EHR that has not been migrated. In health systems without unified data integration, assembling the complete picture for a single patient can take days of manual effort across multiple information systems. Pegasi's FHIR-based ingestion layer eliminates this delay by continuously pulling from all connected data sources, producing a real-time consolidated patient view rather than a weekend-and-manual-compilation exercise.
Specialist routing delay. After a primary care physician or emergency physician encounters a suspicious finding, the path to an oncologist is not always direct. In many health systems, an internal medicine consultation precedes the oncology referral. The referral requires scheduling. The scheduling requires insurance authorization. Authorization requires documentation. In our analysis of 180,000 cases across partner health systems, the specialist routing step added a median of 8.3 days to the diagnostic timeline. Pegasi's alert system notifies the appropriate oncology team member directly at the time of signal detection, bypassing the sequential referral chain where clinically appropriate.
Report turnaround delay. NGS genomic panels typically return in 10-14 days. Specialized pathology consultations can take 7-10 days. During this window, the diagnostic workup is effectively paused. While no AI platform can accelerate laboratory processing times, the handoff from report receipt to clinical action is a second delay that commonly occurs. Reports return and sit unreviewed for days because no alert system notifies the treating team that results are available. Pegasi addresses this through continuous monitoring of incoming lab results, generating a clinical alert within seconds of genomic panel data appearing in the FHIR feed.
Scheduling delay. Once the diagnostic picture is clear enough to warrant a follow-up procedure - a colonoscopy, a bronchoscopy, a stereotactic biopsy - the scheduling system becomes the bottleneck. Many health systems have backlogs measured in weeks for these procedures. Pegasi's platform integrates with clinical scheduling systems to flag high-priority diagnostic workup requests, surfacing them for scheduling staff as needing expedited placement rather than joining a standard queue.
Pancreatic ductal adenocarcinoma has the worst five-year survival rate of any major solid tumor: approximately 12% overall, compared to 92% for Stage I colorectal cancer. The primary reason is late detection - 85% of pancreatic cancer patients present with advanced disease, when surgical resection is no longer feasible.
The window for curative resection in pancreatic cancer is narrow. Tumors that are 2 cm or smaller and confined to the pancreas have a 5-year survival rate after surgery of approximately 30-40%. Tumors that have involved the major mesenteric vessels have a surgical resectability rate below 10%. The difference between those two scenarios is often 6-8 weeks of disease progression. In no other major cancer type does the diagnostic delay arithmetic have such direct and predictable consequences for surgical eligibility.
Our pilot at MD Anderson is targeting pancreatic cancer specifically in high-risk populations: patients with new-onset diabetes in their 50s and 60s (a recognized pancreatic cancer risk signal), patients with BRCA2 or PALB2 germline variants, and patients with known intraductal papillary mucinous neoplasms (IPMNs) that require periodic imaging surveillance. For each of these populations, the early warning signals exist - they are just rarely being aggregated and acted upon rapidly. That is the problem Pegasi is built to solve in this context.
At Houston Methodist, our most mature deployment, we have prospectively tracked the time from suspicious finding to confirmed diagnosis for all oncology patients since Pegasi went live 14 months ago. The comparison group is the prior 12 months at the same institution.
The median time from suspicious imaging finding to oncology team notification decreased from 4.2 days to 0.8 days - a reduction attributable to the automated alert routing eliminating the manual specialist referral chain. The median time from NGS panel result receipt to documented oncologist review decreased from 3.1 days to less than 6 hours. The median time from confirmed diagnosis to treatment planning appointment decreased by 11 days, which we attribute to the Pegasi team helping Houston Methodist's scheduling system identify and clear bottlenecks in their oncology scheduling queue.
The combined effect was a 14.3-day reduction in the median time from initial suspicious finding to treatment initiation. Across 340 flagged cases in year one, that represents a substantial aggregate reduction in the diagnostic gap. We cannot claim with certainty how that translates to clinical outcomes - establishing causal attribution in a health system with many concurrent variables requires a prospectively designed trial. But the directionality is consistent with the published literature on delay reduction and outcomes, and it supports continued investment in diagnostic delay as a primary quality metric for oncology programs.
Technical solutions to diagnostic delay are necessary but not sufficient. The organizations that achieve the most dramatic delay reductions share a common characteristic: they have appointed clinical leadership who own the diagnostic delay metric as a performance accountability. Without that organizational accountability, alert systems get ignored, scheduling priority adjustments revert to default, and the path of least resistance wins.
At Pegasi, we require a named clinical champion at each partner institution as a condition of deployment - typically a medical director of oncology or a chief quality officer. That champion receives a monthly dashboard showing median delays by stage of the diagnostic workflow, alert response rates by care team, and case-level outcomes for patients where delay reduction was documented. The dashboard does not stay in IT. It goes to the clinical quality committee. That accountability structure is what converts a diagnostic alert system into a measurable improvement in patient care.