Missing Follow-Up: Can It Distort Medical Study Results?

What did pandemic researchers learn about missing follow-up?

A perspective published in Frontiers in Public Health in July 2024 describes staff experiences running a cohort involving more than 600 mother-infant pairs. The project scheduled household assessments across three years. Staff reported inconsistent healthcare records, inaccurate contact information and difficulty retaining participants. These problems complicated efforts to understand health after pregnancy. [Original research perspective](https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2024.1417712/full).

The report drew on discussions within a 15-person project team. Its observations identify operational weaknesses; they do not establish how much missing follow-up changed any clinical finding. The distinction matters because describing a barrier does not demonstrate the effectiveness of a proposed solution. [Report methods and findings](https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2024.1417712/full).

How can missing participants change a study's conclusions?

Imagine a treatment study in which people experiencing troublesome side effects become less likely to attend assessments. If researchers analyze only participants who remain, the treatment could appear more favorable than it really is. This is an illustrative example, not a finding from the Gujarat project. Cochrane's guidance explains that assessing missing-data bias requires examining why outcomes are unavailable and whether missingness could depend on the outcome itself. [Cochrane guidance](https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-08).

Missing observations also reduce the information available to estimate benefits and harms. Statistical methods can estimate unobserved outcomes, but their conclusions depend on assumptions. Researchers therefore use sensitivity analyses: they repeat calculations under alternative, plausible assumptions to see whether the main conclusion changes. A finding that disappears under reasonable alternatives deserves more caution than one that remains consistent. Neither a large starting sample nor the label 'intention-to-treat' automatically resolves missing-data problems. [CONSORT explanation](https://www.bmj.com/content/389/bmj-2024-081124), [Cochrane guidance](https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-08).

How can researchers improve follow-up and make results more trustworthy?

Follow-up does not always require travel to a central research hospital. FDA guidance describes decentralized trial activities such as telehealth appointments, home visits and assessments by local healthcare professionals. These approaches offer ways to bring research closer to participants. Their suitability depends on the procedure and the information required: a remote conversation cannot automatically replace a physical examination or laboratory measurement. The practical implication is to design convenient follow-up around the study's clinical needs. [FDA guidance](https://www.fda.gov/regulatory-information/search-fda-guidance-documents/conducting-clinical-trials-decentralized-elements).

Readers also need a clear account of what happened after enrollment. The CONSORT 2025 reporting checklist asks randomized trials to show participant numbers through allocation, treatment and analysis; explain losses and exclusions; and describe how missing data were handled. These requirements help distinguish an impressive recruitment total from the number actually contributing to a result. For patients and clinicians assessing a treatment headline, the useful questions are how many outcomes were missing, why they were missing and whether the conclusion survives alternative assumptions. [CONSORT reporting statement](https://www.bmj.com/content/389/bmj-2024-081123), [CONSORT explanation](https://www.bmj.com/content/389/bmj-2024-081124).