ResearchPod Summary
Patients treated for high-grade cervical dysplasia (such as CIN 2-3 or HSIL) remain at a significantly elevated risk of developing cervical cancer compared to the general population. To manage this risk, clinical guidelines mandate a period of close surveillance, typically involving multiple co-tests (HPV testing combined with cytology) in the years following treatment. This study sought to determine the real-world adherence rates to these surveillance guidelines by analyzing longitudinal data from 3,146 patients treated for high-grade dysplasia between 2010 and 2019 across two diverse U.S. health care systems: an academic medical center (Massachusetts General Brigham) and a public safety-net system (Parkland Health).
The researchers found that adherence to recommended surveillance is alarmingly low. Across both study sites, only 45.5% of patients completed the required two surveillance co-tests within the 30-month window following their treatment. Adherence rates varied by site, with 55.3% completion at the academic center versus 40.6% at the safety-net system.
Crucially, among the patients who did manage to complete the two-test surveillance protocol, approximately one-third received at least one abnormal result, confirming that this population remains at high risk for persistent or recurrent disease. While the overall incidence of cervical cancer in the cohort was low (0.5%), the findings underscore that many patients are falling out of the necessary follow-up care, leaving them vulnerable to undetected progression.
These results demonstrate a significant gap between clinical guidelines and actual patient care. Because patients treated for high-grade dysplasia are at a persistent, twofold to fivefold increased risk of cervical cancer, returning to routine, less-frequent screening is insufficient. The low rate of guideline-concordant surveillance suggests that current health care delivery models are failing to keep these high-risk patients engaged. The study highlights an urgent need for systematic improvements—such as better patient education, targeted outreach, and improved continuity of care—to ensure that these patients receive the long-term monitoring required to prevent future cancer diagnoses.
Sam: Across two major integrated health systems, only 45.5% of patients treated for high-grade cervical dysplasia completed the recommended two-test surveillance protocol within 30 months. That comes from a retrospective analysis by Victoria Wang and colleagues in Obstetrics and Gynecology.
Alex: That is a large gap. These patients sit inside integrated systems, Mass General Brigham on one side and Parkland on the other. So what is actually going wrong?
Sam: The paper points to a disconnect between the surgical procedure and long-term care. The LEEP gets coded, billed and closed out as a discrete event. The surveillance is a multi-year obligation, with co-tests at roughly 12 and 24 months. The system treats the procedure as the endpoint rather than the start of a monitoring period.
Alex: So is the failure that patients are lost, or that nobody flags them for these specific tests?
Sam: Probably a combination, and a retrospective design can't cleanly separate the two. The interpretation is that there is no automated mechanism carrying patients from treatment into long-term monitoring. The system is passive. It waits for the patient to come back rather than managing the timeline.
Alex: The paper also reports that among those who did complete surveillance, about 31% had at least one subsequent abnormal result.
Sam: That is the second load-bearing number. It indicates that recurrence risk in this population is real and not theoretical. But I'd read it carefully. It describes the people who returned, so it doesn't tell you the rate among those who didn't. Missed surveillance matters because a meaningful share of returners had something to catch, but the paper can't say what happened to the other half.
Alex: Which raises the confounding question. Could socioeconomic factors explain the 45.5% better than system design?
Sam: As I read it, the authors adjusted for variables like income and education. The gap also appears in both an academic system and a safety-net system. That argues against a purely patient-level explanation, though an observational design can't rule out residual confounding. The authors' interpretation is that the missing piece is standardized, cross-system infrastructure.
Were there timing differences between the people who completed the protocol and those who didn't?
AI-generated third-party summary by ResearchPod. Not official content or an endorsement by the paper authors or affiliated organizations.
Sam: There were. The completers stayed in the system for a median of over sixty-four months, while those who dropped off were gone in under forty. The median time to the first co-test was a little over seven months, and the protocol then calls for a twelve-month interval to the second.
Alex: A referee might say that comparison is partly mechanical. You can't complete a two-year protocol if you leave the system at month thirty.
Sam: Fair point, and it is closely tied to the paper's main limitation. The authors acknowledge they can't account for patients who leave through insurance changes or geographic mobility. So some of the 45.5% likely reflects care received elsewhere, which means true adherence is probably somewhat higher. The authors still read the remaining gap as a continuity problem.
Alex: But even if some patients are being seen elsewhere, the system isn't facilitating that handoff. It is just losing track of them.
Sam: That is the argument. Continuity falls on the patient. The discussion also raises logistical barriers, like parking and childcare, that coverage of the tests themselves doesn't address. In effect, resources go to the diagnostic procedure and not to the infrastructure that keeps someone in a multi-year cycle.
Alex: And the proposed remedy?
Sam: Registry-based approaches that do more than record data. They would trigger alerts to both clinicians and patients, so follow-up doesn't depend on someone remembering a twelve-month window. It's a recommendation, not something this study tested. The evidence supports the size of the problem and the argument that passive systems produce it. Whether automated tracking closes the gap is a separate question for prospective work.
Alex: So the study establishes a large surveillance gap in two very different settings, and that gap is the part the data support best. If you want the figures and the method choices we skipped, you can generate a deep dive of this paper. The paper has the rest either way.
Sam: Thanks for listening.