A caregiver change is an identity problem, not a new score

When a veterinary clinic asks a client to maintain an at-home observation log—whether monitoring recovery after orthopaedic surgery, tracking mobility during an osteoarthritis trial, or observing respiratory patterns in a congestive heart failure patient—the recording task rarely stays confined to a single individual. In most households, care is distributed. Two partners working alternate shifts, an older child handling afternoon exercise, a visiting grandparent, or a hired professional pet sitter may each interact with the patient across a single 24-hour cycle. When these different individuals contribute to the same clinical journal, the diary frequently suffers from a critical structural failure: observer identity, observation timing, environmental context, and firsthand sensory observations collapse into ambiguous, secondhand summaries.

The core clinical question facing pet owners and veterinary teams is straightforward: How can pet owners and a veterinary team distinguish not observed, not recorded, and a real change when different caregivers contribute to a requested home-observation diary? Solving this problem does not require designing an unvalidated quality-of-life scoring instrument, imposing an arbitrary numeric pain threshold, or borrowing clinical trial validation rules that treat family members like trained clinical research associates. It requires recognizing that a multi-caregiver observation log fails first as an identity and attribution system, not as a missing diagnostic algorithm.

To maintain clinical utility across handovers, a pet observation log must preserve raw observation identity across seven foundational fields: the specific observer, the calendar date and exact time, the physical context, the concrete firsthand sensory observation, an explicit mark when a scheduled check did not occur along with the reason, medication administration strictly as reported by the person who administered or witnessed it, and predefined clinic contact triggers. Above all, pet owners and veterinary staff must establish one non-negotiable rule: a blank cell is missing documentation, not clinical stability. An empty box does not mean the pet was normal, does not mean the condition worsened, and does not provide permission to adjust medical therapy. It merely indicates that no verifiable data was recorded.

TrialEndpoints is a publication covering clinical outcome assessment methods, not a clinical research organization (CRO), statistician, instrument developer, or source of trial results. In human clinical research, observer-reported measures are defined as reports of observable signs, events, or behaviors without medical judgment, as discussed in the TrialEndpoints analysis of fit-for-purpose clinical outcome assessment evidence in FDA PFDD guidance. Those human PFDD concepts are background only. They do not validate veterinary diaries, do not establish clinical thresholds for companion animals, and do not require the same observer throughout as if this were a human trial protocol.

Adjacent clinical tasks are covered in dedicated VetMedGuide resources: pet owners navigating end-of-life decisions and palliative thresholds should consult our guide on quality-of-life scales for dogs and cats; veterinary teams structuring internal telephone follow-ups should review our protocol for post-op callback boards; home monitoring for diabetic patients is detailed in our guide to feline diabetes glucose curves and home tracking; and subtle hypoglycemia recognition is analyzed in our guide on ferret insulinoma monitoring. Additional operational workflows are addressed in our guides to first veterinary oncology visit checklists, front-desk triage in CSR phone triage protocols, staging canine dementia in canine cognitive dysfunction DISHAA stages, and multimodal arthritis strategies in canine osteoarthritis treatment. This article focuses strictly on the operational problem of data handover when caregivers change.

What VICH GL9 and CVM guidance clarify about firsthand notes

The United States Food and Drug Administration (FDA) Center for Veterinary Medicine (CVM) Guidance for Industry #85 (VICH GL9), titled Good Clinical Practice (final, May 2001, docket FDA-1999-D-0754, HTML landing content current as of 15 August 2018), is marked Contains Non-binding Recommendations. FDA's landing page states that the objective is guidance on the design and conduct of all clinical studies of veterinary products in the target species, directed at individuals and organizations involved in designing, conducting, monitoring, recording, auditing, analyzing, and reporting those studies, and that the guidance should be followed when developing clinical study data intended to be submitted to regulatory authorities. FDA also describes Good Clinical Practice as an international scientific quality standard for those studies. Those identity rules can structure a household handover worksheet. They do not convert pet owners into study investigators, do not convert a clinic-requested diary into a GCP trial, and do not create a legal duty for an ordinary home diary.

Section 8.3.1 of GFI #85 states that raw data—whether handwritten or electronic—should be attributable, original, accurate, contemporaneous, and legible. In the context of a shared home diary, each of these principles addresses a specific vulnerability:

• Attributable: GFI #85 says attributable means the raw data can be traced by signature or initials and date to the individual who observed and recorded the data. If more than one individual observes or records the raw data, that fact should be reflected in the entries. When a household log contains unattributed checks, the clinic cannot tell who actually watched.

• Original and Accurate: GFI #85 says original and accurate means the raw data are the firsthand observations. A later rewrite from memory or a text message is not that firsthand record.

• Contemporaneous: GFI #85 says contemporaneous means the raw data are recorded at the time of observation. A later summary written after a handover is not that contemporaneous note.

• Legible: GFI #85 says legible means the raw data are readable and recorded in a permanent medium, for example ink for written records or electronic records that are unalterable.

GFI #85 Glossary Section 1.24 defines raw data as any original worksheets, calibration data, records, memoranda and notes of first-hand observations and activities of a study that are necessary for the reconstruction and evaluation of the study. The same definition states that facsimile transmissions and transcribed data are not considered raw data. This principle was reaffirmed in CVM Guidance for Industry #287 (Raw Data for Safety and Effectiveness Studies, final June 2025, landing-page content current as of 4 June 2025), which restates that, for GCP studies, transcribed data are not considered raw data and that CVM considers raw data the first permanent recording of an observation. When applied to household management, this definition reveals why common habits undermine clinical evaluations: if a pet sitter sends a text message stating 'the dog seemed slightly stiff,' and an owner later transcribes 'mild stiffness' into the clinic chart from memory, that transcribed entry is no longer raw data. It is a secondary interpretation subject to transcription drift.

GFI #85 Section 8.3.4 states that if a portion of the raw data needs to be copied or transcribed for legibility, an authenticated copy should be made; the reason should be explained in a dated memorandum or dated notation on the transcribed record, signed by the individual or individuals making the copy or transcription; and the original, the copy or transcript, and the memorandum should be kept together. In a home-care setting, if this identity logic is being used at all, a later rewrite of another caregiver's row should keep the original first-hand note with a dated reason for the copy. Section 8.3.5 states that a handwritten correction should be made by drawing one straight line through the original entry so the original remains legible, initialed and dated by the individual making the correction at the time the correction is made, with a reason for the change. That is study-documentation language borrowed as identity fields. It is not a clinic transcription SOP and not a legal duty on an ordinary household diary.

Not observed, not recorded, and a real change are three distinct states

The primary failure mode in shared observation diaries is the conflation of four completely different operational situations into an identical blank box or an ambiguous note. To interpret home monitoring accurately, veterinary teams and pet owners must formally differentiate among these states:

1. Not Observed (Unperformed Check): The scheduled clinical assessment was never conducted. For instance, the veterinary discharge instructions requested a resting respiratory rate count at 14:00, but both caregivers were unexpectedly delayed at work and no one was present in the home. Marking the cell explicitly as 'Not Observed (Caregivers away)' records the reality of the schedule gap without fabricating data.

2. Not Recorded (Unwritten Observation): The caregiver actually observed the animal at the scheduled time, but did not write the entry at the time of observation. For example, a morning mobility look occurred at 07:30, but the note was written at 20:30 from memory. That delayed write-up should be labeled as not recorded at the time, because it is not contemporaneous raw data under GFI #85 Section 8.3.1.

3. Real Change (Verified Firsthand Observation): The caregiver conducted a planned look at the scheduled time and wrote a firsthand description of what was seen—such as two episodes of dry retching, or reluctance to bear weight on the left pelvic limb. The entry describes the specific physical action witnessed. It is not a diagnosis, not a score, and not permission to change treatment.

4. Secondhand Transcription: An entry logged by a person who did not witness the behavior, based on a text message, voicemail, or conversation from another caregiver. GFI #85 glossary 1.24 states that facsimile transmissions and transcribed data are not considered raw data. The row should be labeled as a secondhand report, with the original message kept, rather than copied as if it were a firsthand observation.

The diagram below illustrates how an observation entry should be routed through a multi-caregiver household handover, ensuring that unperformed checks and delayed notes never masquerade as pristine clinical data:

graph TD
    Start["Scheduled Observation Timepoint"] --> CheckOccur{"Did caregiver observe the pet at the planned time?"}
    
    CheckOccur -- "No" --> MarkNotObs["Mark 'NOT OBSERVED'<br/>Record reason: absent, asleep, schedule conflict<br/>DO NOT assume normal or worsening"]
    
    CheckOccur -- "Yes" --> LogContemp{"Was observation written down contemporaneously?"}
    
    LogContemp -- "No" --> MarkNotRec["Mark 'NOT RECORDED AT TIME'<br/>Note delayed retrospective entry with exact write-up time<br/>Flag as memory reconstruction"]
    
    LogContemp -- "Yes" --> IsFirsthand{"Is the person logging the firsthand observer?"}
    
    IsFirsthand -- "No" --> MarkTranscribed["Mark 'SECONDHAND REPORT'<br/>Name source e.g. SMS from pet sitter<br/>Preserve original message"]
    
    IsFirsthand -- "Yes" --> EvaluateChange{"Did observable behavior differ from baseline?"}
    
    EvaluateChange -- "No" --> LogStable["Log concrete sensory details<br/>Name observer, date, context, specific signs seen"]
    
    EvaluateChange -- "Yes" --> LogRealChange["Log 'REAL CHANGE'<br/>Describe concrete physical deviation<br/>Check against clinic escalation red flags"]
    
    LogRealChange --> CheckRedFlag{"Does deviation match clinic red-flag criteria?"}
    CheckRedFlag -- "Yes" --> CallClinic["Contact veterinary hospital immediately<br/>Reference specific logged observation"]
    CheckRedFlag -- "No" --> MonitorNext["Continue monitoring at next scheduled timepoint"]

A new clinical sign is an open question, not a collapsed diagnosis

When different family members record changes in an animal, there is a natural human tendency to bypass objective physical description and jump directly to diagnostic speculation. A caregiver who notices an older canine patient hesitating before standing might record 'hip dysplasia flaring up,' while another caregiver noticing decreased food consumption might write 'nausea from medication.' In both instances, the raw observation has been prematurely collapsed into an unsupported diagnostic conclusion.

The clinical danger of this diagnostic collapse is highlighted in the 2026 AAHA Oncology Guidelines for Dogs and Cats (Section 9: 'Post-Treatment Monitoring and Follow-Up Care,' published by Christensen et al., JAAHA 2026;62(1):1-37, PMID 41448590). Section 9 emphasizes that post-treatment monitoring serves as an open communication channel between clients and clinicians to detect resolution, improvement, progression, or recurrence of original clinical signs, as well as new complications arising from therapy.

Crucially, the AAHA guidelines present the foundational New Clinical Signs in Cancer Patients clinical framework: because many veterinary patients receiving chronic or post-procedural care are geriatric, new clinical conditions and symptoms are remarkably common. Whenever a patient presents with a newly observed clinical sign following medical intervention, the veterinary team must evaluate three distinct etiologies:

1. Primary Disease Progression or Recurrence: The new sign may reflect tumor progression or recurrence. That possibility is a clinical question for the veterinary team, not a diagnosis a caregiver should write into a diary cell.

2. Treatment Toxicity or Complication: The sign may reflect treatment toxicity or a complication of initial therapy. A named observation is a communication input, not a conclusion that treatment caused the change.

3. A New, Unrelated Medical Condition: If the new clinical sign is not one that would be expected as a result of tumor progression or treatment toxicity, AAHA states it is more likely that the pet has a new and potentially undiagnosed problem not related to the tumor or treatment. A diary row should keep that question open rather than collapsing it into the cancer diagnosis.

Section 9 of the AAHA guidelines explicitly observes that clients frequently attribute any newly noticed change in their pet's demeanor, appetite, or mobility directly to their existing cancer diagnosis, even when an unrelated etiology is substantially more probable. When home caregivers write interpretive diagnoses rather than raw physical observations into the log, they introduce severe confirmation bias into the veterinary consultation. If an owner writes 'cancer progressing' instead of 'dog vomited undigested kibble 30 minutes after eating,' the clinician may be steered away from investigating dietary indiscretion or concurrent gastritis.

Section 9 states that the frequency of evaluations should be based on the patient's condition, diagnosis, time since therapy, and client preferences and limitations. It reports that it is common for pets to be evaluated every 2 to 4 weeks for 2 to 3 months after they complete radiation therapy, usually by the treating oncologist, then every 2 to 3 months for the first year, then every 3 to 6 months during the second year. Those visit-interval examples are clinic-schedule language. They are not a required home-diary cadence, not a completeness threshold, and not a scoring instrument.

Why impressions drift: Lessons from veterinary caregiver placebo research

The absolute necessity of recording concrete, firsthand physical behaviors rather than global caregiver impressions is supported by landmark veterinary clinical research on observer subjectivity. In a prospective, randomized, double-blinded, placebo-controlled multicenter clinical trial published in the Journal of the American Veterinary Medical Association, Conzemius and Evans (JAVMA 2012;241(10):1314-1319, PMID 23113523) investigated the caregiver placebo effect among dog owners and examining veterinarians evaluating dogs with lameness secondary to osteoarthritis.

The study evaluated 58 dogs enrolled in the placebo control arm of an FDA-approved clinical investigation over a 42-day treatment period. The researchers compared subjective observational assessments completed by owners and veterinarians against objective kinetic measurements captured using force-platform gait analysis. The peer-reviewed findings in the PubMed abstract demonstrated an extraordinary divergence between subjective perception and physical reality:

• Owner Caregiver Placebo Effect: A caregiver placebo effect occurred in 39.7% of owner evaluations of lameness, as reported in the PubMed abstract. That figure is the study's caregiver-placebo result in that placebo arm. It is not a statement that owners reported substantial improvement, and it is not a typical household diary-error rate.

• Veterinarian Placebo Effect: Examining veterinarians exhibited an even higher subjective placebo response: a caregiver placebo effect occurred in 44.8% of veterinary lameness evaluations conducted at a walking gait, in 44.8% of examinations at a trotting gait, and in 43.1% of veterinary assessments of joint pain on physical palpation.

• Temporal Enhancement: The magnitude of the subjective caregiver placebo effect increased significantly over time as the study progressed, reflecting deepening observer expectation bias.

• Objective Kinetic Invariance: Throughout the 42-day trial, mean ground reaction forces remained unchanged during placebo treatment. Individually, of 58 dogs, 5 had ground reaction forces that worsened by 5% or more over 42 days, 7 improved by 5% or more, and 46 remained unchanged.

Conzemius and Evans concluded that a caregiver placebo effect was common in owner and veterinarian evaluation of osteoarthritis treatment response, that force-platform gait analysis was an unbiased outcome measure in that study, and that a caregiver placebo effect should be considered when interpreting owner and veterinary reports of patient response to treatment. Those findings are a reason to keep who watched and what was seen from collapsing into a later rewrite such as 'he seems better.' They are not a diary-error rate, and force-platform gait analysis is not a required household tool.

However, critical evidentiary boundaries must be strictly maintained: the 39.7% owner placebo rate and 44.8% veterinarian rate represent trial-specific outcomes in a blinded canine osteoarthritis placebo cohort. They do not establish a universal diary error rate, do not quantify household completeness percentages, and must never be cited as an excuse to dismiss genuine client observations. Rather, their operational lesson for multi-caregiver handovers is profound: when different family members record vague summary impressions such as 'he seemed much happier today' or 'she moved a bit better,' those impressions are heavily contaminated by emotional hope, fatigue, and expectation bias. Keeping the diary restricted to concrete, observable physical parameters—such as the exact number of stairs climbed, seconds required to rise from recumbency, or distance walked before sitting—protects the record from subjective impression drift.

Human outcome assessment concepts remain background context

In human therapeutic development, regulatory bodies have established formal taxonomies to standardize clinical outcome measures. Under the Patient-Focused Drug Development (PFDD) framework mandated by the 21st Century Cures Act and PDUFA VI, the FDA published its comprehensive Patient-Focused Drug Development Glossary (content current as of 5 March 2026). The glossary formally defines an Observer-Reported Outcome (ObsRO) as a measurement based on a report of observable signs, events, or behaviors related to a patient's health condition by someone other than the patient or a healthcare professional—generally a parent, caregiver, or someone who observes the patient in daily life.

Crucially, the FDA glossary specifies that an ObsRO measure is particularly vital when patients cannot self-report (such as nonverbal pediatric patients or cognitively impaired adults), and that an ObsRO measurement must not include medical judgment or clinical interpretation. For example, the glossary cites a parent report of a child's vomiting episodes. An ObsRO measure does not include medical judgement or interpretation. That distinction appears again in FDA PFDD Guidance 3 (Selecting, Developing, or Modifying Fit-for-Purpose Clinical Outcome Assessments, final October 2025, landing-page content current as of 23 October 2025), which states that this human COA guidance is generally applicable to PRO, ObsRO, ClinRO, and PerfO measures, and that a COA is considered fit-for-purpose when the level of validation associated with a medical product development tool is sufficient to support its context of use, as discussed in the TrialEndpoints review of fit-for-purpose clinical outcome assessment development.

While these human measurement-type statements help explain why a non-clinician observer should record observable signs, events, or behaviors rather than medical judgment, human PFDD guidelines do not validate veterinary diaries and cannot be extrapolated to animal health. A household handover worksheet does not require the same observer throughout, as if this were a human trial protocol, and it does not transfer a human ObsRO into dogs or cats. The job is to keep multi-caregiver entries attributable, firsthand, and explicit about what was not observed or not recorded.

Caregiver-change handover worksheet and hypothetical scenarios

To establish an effective observation log that preserves data identity across multiple household members, veterinary clinics and pet owners should adopt an eight-field handover standard. Each entry must capture:

1. Observer Identity: The full first name or initials of the individual who personally conducted the observation (e.g., 'Marcus (Owner)' or 'Elena (Pet Sitter)').

2. Date and Exact Timestamp: Calendar date (YYYY-MM-DD) and specific 24-hour time (e.g., '2026-09-10 at 07:45') when the evaluation occurred.

3. Environmental Context: The specific physical setting and baseline conditions (e.g., 'Rising from dog bed on laminate flooring' or 'During 10-minute leash walk on level turf in cool morning air').

4. Firsthand Sensory Observation: Concrete physical actions, counts, or behaviors witnessed directly by the observer, avoiding subjective adjectives like 'fine,' 'better,' or 'unhappy' (e.g., 'Coughed 4 times after drinking water; no nasal discharge visible'). Write counts as seen. Do not convert a count into a diagnosis or a homemade cutoff.

5. Not Observed Checkmark: An explicit checkbox or indicator showing that a scheduled evaluation was completely unperformed.

6. Reason for Missing Look: A brief explanation accounting for the unperformed check (e.g., 'Caregiver held at work meeting; pet unattended').

7. Medication Administration Status: Medication information only as reported by the person who offered or watched the dose: the labeled product as written on the clinic instructions, the time offered, and whether swallowing was witnessed. Example wording: 'Labeled morning tablet offered at 08:00; swallowing witnessed' or 'Dose left in bowl; ingestion unwitnessed.' Do not infer that a dose 'must have been given,' and do not invent a milligram amount at home.

8. Veterinary Escalation Triggers: Veterinary-contact instruction copied from the clinic's written call-now list, if one was provided. Leave unknown thresholds unknown. Do not invent numeric cutoffs, scoring limits, or permission to change treatment at home.

The following original decision table presents five labeled fictional caregiver-change rows. Names, dates, and details are invented and are not a real patient, trial case-report form, or scale result. The columns keep observer, date, context, the observation cell, classification as not observed, not recorded, a real change, or a later transcription, what that cell is not, and the next household or clinic question.

As demonstrated in the worked table above, each hypothetical scenario maintains a clear demarcation between what was directly witnessed and what remains unknown. In Scenario 1, the explicit 'Not Observed' designation prevents the clinic from assuming the dog was comfortable simply because no limp was reported. In Scenario 2, flagging the retrospective write-up protects the clinician from relying on a memory that may have drifted under expectations of recovery. In Scenario 3, the teenager's firsthand description provides the exact behavioral facts needed for clinic triage without attempting to diagnose joint pathology. In Scenario 4, noting the secondhand nature of the text message prevents attribution collapse. And in Scenario 5, documenting the recovered tablet avoids dangerous medication overdosing.

Practical implementation checklist for clinic teams and households

To translate these attribution principles into sustainable daily practice, veterinary teams and pet-owning households can implement four operational checkpoints:

• Establish a Central, Physical Worksheet Location: While mobile applications and digital spreadsheets are popular, they frequently fail across multi-generational or multi-caregiver households because different family members lack app access, shared logins, or battery charge. Placing a physical, single-sheet printed handover worksheet on a clipboard on the kitchen counter—immediately adjacent to the patient's medication containers—ensures that every caregiver sees the log, reads previous entries, and records notes at the exact moment care is provided.

• Enforce Witnessed Medication Logging: Households must establish a strict protocol separating 'food offered' from 'medication consumed.' Many companion animals, particularly cats and convalescing dogs, skillfully spit out tablets or leave medication-laced treats behind. Caregivers must log a dose as administered only when complete swallowing has been directly witnessed. If a dose is left unattended in a dish, it must be recorded as 'offered unwitnessed' until the bowl is inspected.

• Conduct Structured Clinic Intake Reviews: When a client presents an observation diary at a recheck appointment, veterinary technicians should scan specifically for blank cells before the veterinarian examines the patient, asking whether those looks were not made, not written down, or actually different. Resolving those ambiguities before clinical examination keeps a missing row from being treated as worsening, as a completed observation, or as permission to change treatment.

• Differentiate Monitoring Logs from Emergency Triage: Clients should be reminded that an observation diary is a record, not an emergency channel. If a patient shows acute respiratory distress, repeated uncontrolled vomiting, sudden collapse, or signs of severe pain, caregivers should use the clinic or emergency contact the veterinary team provided rather than waiting to write the event at the next scheduled timepoint. Copy those contact details onto the worksheet if the clinic supplied them; do not invent a coverage roster at home.

Sources

Primary standards, regulatory guidance documents, and peer-reviewed scientific literature supporting this article:

• U.S. Food and Drug Administration, Center for Veterinary Medicine: CVM GFI #85 (VICH GL9) Good Clinical Practice (May 2001). Final guidance for industry on designing, conducting, and documenting veterinary clinical trials in target species. Sections 8.3.1 (attributable, original, accurate, contemporaneous, legible raw data), 8.3.2 (diligently recorded observations), 8.3.4 (authenticated copies and transcription rules), 8.3.5 (handwritten correction procedures), and Glossary 1.24 (raw data definition excluding transcribed data). Docket FDA-1999-D-0754; HTML landing content current as of 15 August 2018.

• U.S. Food and Drug Administration, Center for Veterinary Medicine: CVM GFI #287 Raw Data for Safety and Effectiveness Studies (June 2025). Final guidance for industry on submitting copies of raw data in support of new animal drug applications. Reaffirms that raw data represents the first permanent recording of an observation and that transcribed data is not raw data. Landing page content current as of 4 June 2025.

• American Animal Hospital Association: 2026 AAHA Oncology Guidelines for Dogs and Cats, Section 9: Post-Treatment Monitoring and Follow-Up Care. Professional clinical guidelines for small animal post-treatment monitoring. Outlines the three potential etiologies for new clinical signs in recovering patients (tumor progression, treatment toxicity, or unrelated comorbidities), discusses client attribution bias, and details clinical follow-up visit intervals.

• Journal of the American Animal Hospital Association: AAHA Oncology Guidelines for Dogs and Cats (JAAHA 2026;62(1):1-37). Christensen et al. Comprehensive peer-reviewed publication of the 2026 AAHA Oncology Guidelines covering diagnosis, staging, multimodal treatment, supportive care, and long-term follow-up protocols. PMID 41448590.

• Journal of the American Veterinary Medical Association: Caregiver placebo effect for dogs with lameness from osteoarthritis (Conzemius and Evans, JAVMA 2012). Conzemius MG, Evans RB. JAVMA 2012;241(10):1314-1319. Prospective, randomized, double-blinded, placebo-controlled clinical trial in 58 dogs evaluating caregiver placebo effect in owners (39.7%) and veterinarians (44.8% at walk, 44.8% at trot, 43.1% joint palpation) versus objective force-platform kinetic gait analysis (mean ground reaction force unchanged; 46 unchanged, 7 improved ≥5%, 5 worsened ≥5%). PMID 23113523.

• U.S. Food and Drug Administration: FDA Patient-Focused Drug Development Glossary. Official regulatory terminology for patient-focused drug development under the 21st Century Cures Act. Defines Observer-Reported Outcomes (ObsRO) as reports of observable signs or behaviors by a non-clinician observer without medical judgment. Content current as of 5 March 2026.

• U.S. Food and Drug Administration: Patient-Focused Drug Development: Selecting, Developing, or Modifying Fit-for-Purpose Clinical Outcome Assessments (October 2025). Final regulatory guidance for industry outlining principles for establishing fit-for-purpose evidence for clinical outcome assessments (including ObsROs) within specific contexts of use. Content current as of 23 October 2025.