
A screening test is only as good as what happens next. Between a positive result and a confirmed diagnosis, data passes through many hands and many formats. This guide covers the 7 gaps where it most often breaks, and how test companies and screening programmes can close them.
Early detection has moved quickly. Blood-based tests can look for signals of several cancers at once, stool and cell-free DNA tests are offered alongside traditional screening, and health systems, employers and insurers are building screening into their programmes. Every one of these tests rests on the same assumption: that a positive result leads to timely diagnostic follow-up, and that someone knows what that follow-up found.
In practice, that assumption often breaks. The follow-up happens at a different provider, the result comes back as a letter or a PDF, the person is recorded under different details, and the outcome never makes it back to the team that ran the test. For a programme, that means people lost to follow-up. For a test company, it means real-world performance that cannot be shown. For both, it means decisions made on incomplete evidence.
This guide is written for early detection and screening test companies, screening programmes run by health systems, employers and insurers, and the clinical, laboratory and medical affairs teams around them. It is organised in three parts: getting from result to follow-up, turning follow-up into outcomes, and measuring and using what you learn.
Figure 1. What happened to 711 patients with an abnormal stool-based screening test in US community health centres. Nine in ten were referred for colonoscopy, but just over half completed a pre-procedure visit and fewer than half completed a colonoscopy within a year. Each step down is a hand-off where follow-up was not tracked to completion.
Figure 1 shows how quickly people fall away after a positive screen. In that study of 711 patients, 90 per cent were referred for colonoscopy, but only 52 per cent completed a pre-procedure visit and 43 per cent completed a colonoscopy within a year.1 The 7 gaps below explain why, and what to do about each.
When a test result leaves the lab, responsibility often becomes unclear. The test company reports to the ordering clinician, the clinician refers to a specialist, the specialist's team books an appointment, and the person is expected to attend. If any link fails, nobody is necessarily watching. A review of health systems taking part in a US Centers for Disease Control and Prevention (CDC) colorectal cancer programme found that follow-up colonoscopy completion varied from 24.2 to 75.5 per cent, averaging 47.9 per cent, and that few health systems could track the time to completion.2
Every positive result needs a named owner and a tracked status: result communicated, referral made, appointment booked, diagnostic test done, outcome known. Programmes that keep a register of open positives, and actively chase those without a recorded outcome, close far more of them. For a test company, that may mean agreeing with ordering clinicians, at the time of ordering, how follow-up will be confirmed.
The diagnostic follow-up is usually done somewhere else: a colonoscopy at a hospital, imaging at a radiology centre, a biopsy read by a pathology lab. The results come back, if they come back at all, as letters, scanned reports and PDFs, each in its provider's own format and sometimes its own language.
A folder of reports is not an outcome. To know what happened, each report needs to be matched to the right person and the right positive result, and the key facts pulled out in a consistent form. Plan for this from the start: agree which reports you need for each type of positive result, how they will reach you, and how they will be structured when they arrive.
The screening provider, the referring clinic, the hospital that performs the diagnostic test and the cancer centre that treats the person may each know them by different identifiers, different spellings and different contact details. Linking the follow-up report to the original screening result is often the hardest step, especially across borders or between public and private providers.
Capture enough identifying detail at the time of screening (with consent) to support matching later, match on several details rather than name alone, and send uncertain matches to a person to confirm. Where follow-up data will be used for evidence, plan pseudonymised linkage, so records can be joined without exposing who people are.
The facts that matter most are rarely in a single field. Whether cancer was diagnosed, where it started, what type it is, how advanced it was and when it was diagnosed are written in pathology reports, imaging reports and clinical notes. Turning them into data is called abstraction, and it needs rules.
Figure 2. An illustrative pathology report and the outcome fields abstracted from it. Site, histology, behaviour and grade can be read from this report. Stage is not yet known and must be followed up after surgery, and the date of diagnosis needs an agreed rule (biopsy date, report date or another date) applied the same way every time.
Figure 2 shows that abstraction is more than reading. Some facts are present, some are pending, and some depend on rules. Cancer registries use the International Classification of Diseases for Oncology, third edition (ICD-O-3), to code site, histology, behaviour and grade, usually from a pathology report,3 and their minimum datasets also record the basis of diagnosis.4 Agree your outcome fields and rules up front, code them to recognised standards, and check consistency. Reliability can be high when rules are clear: in one study, independent abstractors agreed on 86 to 100 per cent of items.5 It cannot be assumed, so measure it.
How long follow-up takes matters for patients and for evidence. In a study of 70,124 patients with a positive stool test, follow-up colonoscopy at 10 months or later was associated with a higher risk of colorectal cancer and of advanced-stage disease, compared with follow-up within 8 to 30 days.6 For newer tests, time to diagnostic resolution (the time from a positive result to cancer being confirmed or ruled out) has become a key measure. In the PATHFINDER multi-cancer early detection study, the median was 79 days overall: 57 days for true-positive results and 162 days for false-positive results.7
Figure 3. Median time from a positive multi-cancer early detection result to diagnostic resolution in the PATHFINDER study. True-positive results were resolved in a median of 57 days, false-positive results took a median of 162 days, and the overall median was 79 days. False positives take longest to resolve because cancer has to be ruled out.
Figure 3 shows why dates matter as much as diagnoses. Measuring time well depends on capturing the right dates consistently: the date of the screening result, the date it was communicated, the referral date, the date of each diagnostic test and the date of resolution. These usually sit in different documents from different providers, so define each date precisely and abstract it the same way every time.
Positive results get attention because they demand action. Negative results are where a test's real-world sensitivity is revealed. An interval cancer is commonly defined as a cancer diagnosed after a screening test in which no cancer was detected, and before the next recommended screen.8 Analysis of interval cancers is critical in determining the sensitivity of screening and is an objective measure of programme quality.9
Finding interval cancers means following people with negative results over time, usually by linking screening records to cancer registries or hospital records. Programme sensitivity is then calculated as screen-detected cancers divided by the total of screen-detected and interval cancers.10 Without this, a test or programme can look more sensitive than it is. Plan the linkage and the follow-up period at the start, rather than trying to reconstruct it years later.
Follow-up data is valuable well beyond care: for performance studies, real-world evidence, regulatory submissions, reimbursement discussions and improving the test itself. Each of these uses needs a proper basis. Decide at the time of screening what people are consenting to, how follow-up data will be used, and who will see it.
Pseudonymise data used for analysis, keep identifiable data with the provider or programme that holds the care relationship, and check local rules on secondary use, cross-border transfer and in-country processing, which vary widely across Asia, Europe and elsewhere. A test company working with providers in several countries may need a different data flow for each, and designing this early is much easier than retrofitting it.
Table 1 brings the 7 gaps together for test companies and screening programmes.
Table 1. A checklist of the 7 data gaps that break early detection follow-up, and how to close each one.
The thread running through all 7 gaps: a screening result is the start of a story, not the end. The value of the test depends on knowing, reliably and on time, how that story ended.
Teams usually come to us with one of two needs. Some are designing a screening programme or launching a test in new markets and want follow-up and evidence built in from the start. Others already have positive results, follow-up reports and records arriving from many providers, and need them turned into outcomes they can report and rely on. We support both.
JondaX has processed more than 500,000 files and 26 million data points at 99.9 per cent field-level accuracy. Clinical reviewer corrections feed back into future mapping, so abstraction becomes more consistent as more of your records are processed. The simplest first step is a sample: send a pseudonymised set of follow-up reports, and we will show you the outcome fields we can abstract and how consistently.
Often because no one owns the next step and no system tracks it. In one study of 711 patients with an abnormal stool test, 90 per cent were referred for colonoscopy but only 43 per cent completed one within a year.1 Across health systems in a US national programme, follow-up colonoscopy completion averaged 47.9 per cent, and few systems could track how long follow-up took.2
Outcome abstraction is the process of pulling structured facts, such as whether cancer was diagnosed, the primary site, histology, stage and date of diagnosis, from pathology reports, imaging reports and clinical notes. It turns follow-up records into data that can show how a screening test or programme performs.
It is the time from a positive screening result to a confirmed answer: cancer diagnosed, or cancer ruled out. In the PATHFINDER multi-cancer early detection study, the median was 79 days overall, 57 days for true-positive results and 162 days for false-positive results.7
An interval cancer is a cancer diagnosed after a screening test that did not detect cancer, and before the next recommended screen.8 Interval cancers are a key measure of how sensitive a screening programme really is, so programmes need to follow people with negative results as well as positive ones.
Cancer registries principally use the International Classification of Diseases for Oncology, third edition (ICD-O-3), which codes the site (topography), histology (morphology), behaviour and grade of a tumour, usually from a pathology report.3
Yes. Jonda Health Services offers outcome abstraction for screening and early detection programmes, with JondaX structuring pathology reports, imaging reports, clinical notes and lab results from many providers, and clinical review built in for anything flagged.
This article is for general information and is not medical advice.
About the author. Suhina Singh is the Founder & CEO of Jonda Health, a health data infrastructure company that turns fragmented clinical data, digital and non-digital, into structured, standardised information that clinicians and systems can actually use. Jonda Health is ISO 27001 certified and works with diagnostics and life sciences companies, health programmes, laboratories, hospitals and digital health teams. Get in touch at hello@jonda.health.