A patient-safety-first training needs assessment for nurses

SafetyPathway is built and led by Jacqueline Wastephen — MA Communications, MBA Healthcare Management, and Doctoral Candidate in Health Administration — with three years of nursing coursework and over ten years of experience in training and workforce development.

The idea for SafetyPathway was born on the floor. As a BSN student working as a nurse extern, Jacqueline observed firsthand the patient safety gaps that existing systems were not capturing — not because nurses didn't care, but because training was driven by calendar mandates, not by what staff were actually struggling with at the bedside. She worked directly with patients in tracheostomy care, traumatic brain injury (TBI), maternal and pediatric nursing, post-operative recovery, and infection prevention — and in each setting, the same pattern emerged: the skills that caused the most anxiety were rarely the ones being addressed in scheduled education.

That clinical reality is the foundation of SafetyPathway. The platform operationalises three premises from the patient safety literature:

• Preventable harm concentrates in a small set of high-acuity skills — identifying and targeting those skills proactively is more effective than broad annual competency reviews.
• Self-perceived competence and avoidance behaviour — what nurses privately call their "dreaded patient types" — are leading indicators of training gaps long before an adverse event occurs.
• Targeted, modality-matched training closes those gaps faster and more durably than blanket calendar-driven modules.

SafetyPathway exists because the most important patient safety data in any facility is often the data no one is collecting — what nurses are afraid to say out loud. This platform gives them a private, structured, specialty-specific way to say it, and gives educators the aggregate signal they need to act on it before harm reaches the bedside.

— Jacqueline Wastephen, Founder
Patient Safety & Workforce Development Specialist
wastephenjacqueline@gmail.com | safepath.company

SafetyPathway is designed from day one to operate at national scale across all 50 US states, the District of Columbia, and US territories. The instrument, scoring engine, and CPD recommender are state-aware: every assessment captures the respondent's state at intake, so findings can be aggregated and benchmarked at the unit, facility, health-system, state, regional, and national level.

This makes SafetyPathway suitable for:

• Single-facility quality-improvement programs.
• Multi-state hospital systems standardising training-needs data across markets.
• State boards of nursing and departments of health tracking workforce competency trends.
• National patient-safety organisations and academic researchers comparing state-level signal.

The platform runs on managed cloud infrastructure with row-level security, encrypted storage, and an architecture that horizontally scales with respondent volume — there is no per-state deployment, no manual provisioning, and no minimum cohort size. A nurse in rural Montana and a nurse in a Manhattan teaching hospital submit through the same instrument and contribute to the same comparable dataset.

The TNA is structured around six competence domains derived from a review of patient-safety incident registries and the WHO Patient Safety Curriculum:

• Airway & tracheostomy care
• Traumatic brain injury & neuro observation
• Maternal & newborn safety
• Pediatric escalation & deterioration
• Post-operative mobilisation & VTE prevention
• Infection prevention, sepsis, handoff

For each domain we collect three signals:

• Self-rated competence on a 4-point Likert (Novice → Proficient).
• Training intensity needed: refresher, retraining, or full training.
• Avoidance behaviour: which patient types the nurse "dreads" being assigned.

The form uses closed multi-select fields wherever possible to enable cross-respondent aggregation, with one required short-text field for context. Free text is optional and never required, to lower respondent burden and to keep PHI out of the dataset (see Privacy Policy).

Item wording was iteratively reviewed with bedside nurses and a patient-safety lead. Average completion time in pilot was ~6 minutes. The form is mobile-optimised because the majority of pilot responses came from phones during shift breaks.

Each respondent's submission yields three derived scores:

• Domain Need Score (DNS) per domain: weighted combination of (1 − competence) × intensity weight × avoidance flag. Range 0–10.
• Composite Patient Safety Risk Index (PSRI): mean of DNS across the six domains, with a 1.25× multiplier for any domain flagged "full training".
• Modality Fit: mapping from preferred training format (simulation, e-learning, bedside coaching, micro-learning) to the personalised CPD recommendations.

Scores are computed deterministically server-side. The AI layer (Lovable AI Gateway, Gemini family) is used only to rank CPD courses against the resulting need profile — it does not modify the underlying scores.

The instrument is in pilot validation. Work to date:

• Content validity: expert review by 4 senior nurses and 1 patient-safety officer; item-level CVI ≥ 0.80 retained.
• Face validity: cognitive interviews with 8 frontline nurses across med-surg, ICU, and maternity.
• Internal consistency: target Cronbach's α ≥ 0.75 per domain on the next 100-respondent cohort.
• Test–retest reliability: scheduled two-week retest study, n = 30.
• Construct validity: planned correlation of PSRI with unit-level safety incident rates in two partner facilities.

This page will be updated with statistics as each milestone closes. Raw protocols are available on request for journal reviewers.

• Participation is voluntary; respondents can submit anonymously.
• No patient identifiers are collected. Free-text fields carry an on-screen warning against PHI.
• Data is stored on encrypted managed Postgres with Row-Level Security; only named admins can read.
• De-identified aggregates may be used in publications and policy submissions (NIW, ministry briefings, conference posters). Individual records are never shared.
• Ethics review: this is a quality-improvement and educational-research instrument. For interventional studies using SafetyPathway data we obtain prior IRB/ethics committee approval at the host institution.

• AI Coach: a topic-scoped assistant for clinical refreshers (trach care, TBI red flags, sepsis bundles, handoff). It is an educational aid, not clinical advice, and explicitly refuses off-topic queries.
• CPD Recommender: a structured tool-call against the active course catalog. It re-ranks human-curated courses for the respondent's profile; it does not invent courses.
• All AI calls run through the Lovable AI Gateway. Prompts are not used to train third-party models per gateway terms.

The scoring formulae, item bank, and the seeded CPD catalog are versioned in the project repository. Each published analysis will reference an instrument version (e.g. SP-TNA v1.0) so results can be reproduced against the exact item set used.

Wastephen, J. (2026). SafetyPathway: a patient-safety-first nurse training needs assessment instrument (v1.0). Available at https://safepath.company.

Methodology questions, collaboration, or to request the full protocol and item bank:
Jacqueline Wastephen — SafetyPathway
Email: afrigraciousdove@gmail.com
Mobile: +1 (854) 844-5166