World Sleep Industry ENCYCLOPEDIA

OSA patient recruitment

6.5.2

Definition, purpose, and service boundaries

OSA participant recruitment is the end-to-end process of identifying, approaching, prescreening, verifying diagnostic evidence, and enrolling people with the obstructive sleep apnea characteristics required by an ethically approved research question. Populations may include suspected but undiagnosed people, previously diagnosed untreated people, those using positive airway pressure or another therapy, defined severity or phenotype groups, and appropriate controls. The scope depends on whether the study concerns screening, diagnosis, mechanism, medicine, device, behavior, surgery, or outcomes. Within WSIC 6.5.2, “Subject and Sample Management,” recruitment does not replace clinical referral, diagnosis, or treatment.

A positive risk questionnaire, an OSA code in a record, a consumer device alert, a positive prior test, protocol eligibility, and randomization are different states. Recruitment staff may describe a study and collect minimum necessary data using approved scripts, but snoring, body habitus, sleepiness, or a watch reading cannot establish a diagnosis, and staff cannot promise enrollment, allocation, or benefit. When research screening identifies a reason for medical evaluation, it should activate a predefined referral route rather than converting a candidate list into a clinical diagnosis list.

The recruitment plan should define population, age, geography, language, diagnostic evidence, treatment state, channels, stage-specific criteria, testing arrangements, risk escalation, compensation, data sources, contact attempts, and stopping rules. Sponsors, investigators, sleep centers, testing services, recruitment vendors, remote staff, and device providers need a responsibility matrix. Outsourcing may support outreach, scheduling, and technical operations, but diagnostic confirmation, complex comorbidity decisions, consent, final eligibility, and participant protection remain with authorized teams.

Target population, recruitability, and representation

The research question should become necessary, verifiable criteria. The target may be newly diagnosed, previously diagnosed, treatment-naive, treatment-stable, residually symptomatic, a defined anatomic or physiological phenotype, or a broad OSA population; the primary outcome may require particular severity, symptoms, or comorbidity. Children and adults differ in presentation, diagnostic rules, guardian permission, and intervention risk, so an adult questionnaire, test threshold, and consent process cannot simply be reused for children. A prior diagnosis also calls for review of its report, date, method, scoring rule, and subsequent treatment or weight changes.

Feasibility funnels should separately estimate lawful contacts, positive risk screens, completed sleep tests, interpretable results, eligible candidates, consent, enrollment, and randomization. They should account for waiting time, testing capacity, night work, transport, caregiving, home setup, connectivity, language, cost, and current therapy. Counts of record codes, people who snore, people with obesity, or device users are not enrollable counts. If research testing must establish the diagnosis, resources for technical failure, repeat testing, return of results, and clinical referral belong in the plan rather than counting only advertisement clicks.

Classic snoring, witnessed apneas, and sleepiness may signal risk, but reliance on that image alone can miss atypical presentations, people without a bed partner, women, older adults, people without obesity, varied racial and ethnic groups, and people with limited resources. Channel and prescreening rules should be assessed in the intended-use population rather than selecting only the easiest phenotype to recognize. Every exclusion needs a safety or scientific justification. If broad cardiopulmonary, metabolic, psychiatric, or insomnia comorbidity exclusions are used, the resulting limits on generalizability and subgroup screen failures should be reported.

Risk prescreening, diagnostic verification, and evidence consistency

Risk questionnaires, symptoms, physical characteristics, and prediction algorithms may determine who needs further evaluation or enrich testing yield, but they cannot diagnose OSA alone. Formal eligibility should use the protocol-specified clinical assessment and objective testing. Laboratory polysomnography and home sleep apnea testing differ in signals, environment, and suitable populations. Whether home testing is appropriate depends on an uncomplicated adult population, technical feasibility, and comorbidity, not on convenience or cost alone.

Report verification should preserve test type, device, sensors, valid recording duration, sleep-time or recording-time denominator, event definition, desaturation rule, arousal information, position, sleep stage, technical quality, and interpreter. Similar indices such as AHI, RDI, and REI are not interchangeable by number alone; a change in scoring version or hypopnea definition can change classification. A single negative, indeterminate, or technically inadequate home study should not automatically define a no-OSA control when clinical suspicion remains. The protocol and qualified clinicians determine any further testing.

The eligibility window should define how old a prior study may be and what triggers reassessment, such as therapy, substantial weight change, pregnancy, surgery, or a new cardiopulmonary or neurological condition. Central reading can improve consistency but requires prespecified transfer, de-identification, scoring version, masking, review, and discrepancy adjudication. If a device algorithm or automated scoring enriches recruitment, its intended use should be validated in the target population and the algorithm version and human review retained. Better yield does not mean the enriched sample represents all people with OSA.

Differential assessment, comorbidity, and safety routing

Screening should distinguish obstructive events, central events, sleep-related hypoventilation, primary snoring, upper-airway resistance patterns, and other causes of hypoxemia or sleepiness. Heart failure, chronic lung disease, neuromuscular disease, stroke, opioid use, severe insomnia, pregnancy, and childhood may alter test selection, phenotype, or risk. A comorbidity does not automatically negate OSA, but the study needs to determine whether its intervention is suitable, which condition needs priority attention, and whether the data can answer the original question.

Daytime sleepiness should be considered with insufficient sleep, shift work, sedating medication, alcohol, narcolepsy, depression, and other causes. A recent sleepiness-related crash or near miss, involuntary sleep at work, or another acute safety issue should pause ordinary recruitment and activate timely local clinical and safety routing with confirmed handoff rather than a leaflet alone. Recruitment staff should not decide driving fitness or prescribe treatment. Applicable reporting and occupational obligations are handled under local rules by authorized personnel.

Interventional screening must also consider product- or procedure-specific respiratory, cardiovascular, cognitive, fall, anesthesia, dental, surgical, pregnancy, reproductive, and device risks. Capacity, acute mental-health crisis, substance use, or severely unstable disease warrant proportionate review and escalation, but a stable diagnosis label should not automatically exclude someone. Screening intensity should match research risk; eligibility from a different PAP, oral-appliance, medicine, or surgery study should not be copied unchanged.

Existing treatment, study intervention, and informed consent

Screening should reconcile positive-airway-pressure device and mode, prescription, mask, objective use, residual events, and recent changes, as well as oral appliances, positional measures, surgery, weight interventions, oxygen, medication, and other therapy. A study may require untreated, stable treatment, a defined adherence state, or treatment failure, but definitions and evidence must be prespecified. Candidates must not stop PAP, oxygen, or another therapy on their own to qualify. Withdrawal, washout, sham treatment, or delayed therapy requires scientific rationale, clinical authorization, monitoring, withdrawal criteria, and a fallback.

Consent should clearly describe home or laboratory testing, overnight sensors, randomization, masking, active or sham devices, pressure or device adjustments, remote uploads, specimens, images, daytime tests, follow-up, possible discomfort, and alternatives. Candidates should understand that a study device may differ from usual care and that participation does not guarantee active therapy or improvement. If a partner provides witnessed-breathing, sleep, or device information, that person's role and privacy need separate explanation; participant consent does not automatically cover a partner's personal data.

Payment and reimbursement may address time, overnight inconvenience, transport, parking, lodging, installation, connectivity, and caregiving costs and should receive applicable review. Credit for completed portions should accrue reasonably rather than making withdrawal costly through a large final payment. Multilingual, accessible, in-home technical support and flexible scheduling can reduce rural, shift-work, disability, and digital-access barriers. Convenience must not become undue influence on people with financial hardship, and declining research should not impair clinical testing or care.

Outreach channels, remote processes, and authenticity

Recruitment materials should accurately state that the activity is research, its major procedures, time burden, basic criteria, location or remote requirements, and contact route. Snoring should not be equated with disease, and fear of sudden death, guaranteed cure, or a free device should not be used as inducement. Records, primary care, dental, cardiovascular or metabolic clinics, communities, and driving or occupational settings can be channels only with lawful contact authority, suitable clinical boundaries, and nonstigmatizing language. Uncertainty about general-population screening should not be hidden behind a marketed risk test.

Remote prescreening and home testing should verify identity, address, current location, emergency contact, device fit, understanding of instructions, sensor placement, return logistics, technical support, and repeat procedures after failure. Uploaded raw signals, video, audio, precise time, location, and health data require minimum-necessary protection; nonenrolled people also need access, retention, and deletion rules. Lack of a smartphone, reliable network, fixed address, or bed partner should not automatically exclude someone. Telephone, paper, onsite, or loaned-device alternatives should be offered when scientifically feasible.

High compensation or device studies may attract duplicates, impersonation, borrowed reports, manipulated device use, or fabricated symptoms. Authenticity checks can combine contact details, identity, report provenance, device serial numbers, and cross-stage consistency, with anomalies reviewed by people. Shared addresses, name changes, minoritized languages, or a limited digital footprint are not fraud by themselves. Verification methods, retention, and appeal after a false flag should be designed in advance; a vendor risk score cannot replace investigator eligibility review.

Eligibility handoff, quality measures, and vendor governance

Final eligibility should use a versioned checklist recording each evidence source, diagnostic report, treatment state, decision maker, date, exception, and medical escalation. Applicable consent, baseline, and safety requirements precede enrollment. Recruitment teams should hand off contact preferences, accessibility needs, pending tests, hazardous-sleepiness alerts, current care, and promised arrangements to the site under the minimum-necessary principle. Screen failures should receive the predefined limited results and referral information; research automated scoring should not be issued as a formal diagnostic report.

Quality measures should separately report impressions, inquiries, permission to contact, risk prescreens, completed tests, valid reports, eligibility, consent, enrollment, and randomization, with elapsed time, technical failures, repeats, lost contacts, withdrawals, channel yield, referrals completed, authenticity events, and subgroup differences. The cheapest candidate, highest questionnaire-positive rate, or highest AHI is not necessarily the best channel; chasing only severe classic cases distorts the target population. Protocol, scoring, device, or eligibility clarifications require assessment of existing candidates, equity, and selection bias, with versions retained.

Sponsors should assess the competence, script and version control, device management, scoring consistency, privacy and security, subcontracting, risk escalation, complaint, audit, and shutdown processes of recruitment vendors, testing organizations, and central readers. Contracts should prohibit sale of leads, unauthorized remarketing, enrollment-linked promises, and silent changes to algorithms or scoring rules. Closeout should reconcile candidate records, contact authority, raw studies and reports, screen status, referrals, compensation, devices, return of data, and deletion evidence. An “eligible list” does not prove that investigators verified diagnosis or protected participants.