4.4 Screening, Extracting & Synthesizing Healthcare Design Literature

Screening, Extracting & Synthesizing Healthcare Design Literature

Retrieving hundreds of citations from academic databases is only the initial phase of the research process. Raw search results represent an unrefined, heterogeneous collection of papers varying widely in scientific rigor, clinical relevance, and architectural applicability. An evidence-based design team cannot simply hand a stack of 150 journal printouts to an architect or hospital board and consider the evidence integrated.

To move from Step 2 (Find the Evidence) to Step 3 (Critically Interpret the Evidence), the team must execute a systematic, disciplined methodology: screening the literature against explicit criteria, extracting critical data points into standardized matrices, and synthesizing heterogeneous findings into coherent design directives.

---

The Multi-Stage Literature Screening Process

To prevent selection bias—the unconscious tendency of designers to select only those studies that support their preexisting design preferences—the team must establish predefined inclusion and exclusion criteria before examining search results.

Defining Inclusion and Exclusion Criteria

Inclusion and exclusion criteria serve as the quality filters that determine which papers enter the final evidence base.

The Two-Stage Screening Funnel

Once criteria are established, the team screens citations through a two-stage funnel:

┌─────────────────────────────────────────────────────────────┐
│         INITIAL DATABASE SEARCH RESULTS (e.g., N = 480)     │
└──────────────────────────────┬──────────────────────────────┘
                               │
                               ▼
┌─────────────────────────────────────────────────────────────┐
│ STAGE 1: TITLE & ABSTRACT SCREENING                         │
│ • Rapidly assess titles and abstracts against criteria       │
│ • Remove duplicates and obvious non-healthcare papers       │
│ • Outcome: Exclude clearly irrelevant studies (e.g., n=395) │
└──────────────────────────────┬──────────────────────────────┘
                               │
                               ▼
┌─────────────────────────────────────────────────────────────┐
│ STAGE 2: FULL-TEXT CRITICAL REVIEW (e.g., N = 85)           │
│ • Retrieve and read complete full-text articles              │
│ • Evaluate methodological rigor, sample size & controls     │
│ • Outcome: Exclude papers failing detailed criteria (n=63)  │
└──────────────────────────────┬──────────────────────────────┘
                               │
                               ▼
┌─────────────────────────────────────────────────────────────┐
│ FINAL SYNTHESIS EVIDENCE BASE (e.g., N = 22)                │
│ • Populate Data Extraction Matrix                           │
│ • Formulate Evidence-to-Design Translation Directives       │
└─────────────────────────────────────────────────────────────┘

---

Evaluating and Recording Evidence

EDAC study materials give three criteria for evaluating existing evidence: relevance, validity, and reliability. When studies conflict, the reliability and validity of each study are especially useful for deciding how much weight to give it.

Record keeping is part of the method, not an afterthought:

• The most efficient way to keep track of evidence is a searchable electronic database based on standard library catalog rules (not scattered folders).
• For each item, record basic information that helps organize and use the collection: author, title, subject, abstract, and link or location in the database.
• Meticulous record keeping documents the links between project goals, desired outcomes, and the supporting evidence—so the rationale survives staff turnover and value engineering.

---

The Data Extraction Matrix (Evidence Table)

A Data Extraction Matrix (also called an Evidence Table) is the essential analytical tool used to deconstruct, standardize, and compare findings across diverse studies. Because healthcare architecture synthesizes papers from medicine, nursing, acoustics, and psychology—all of which employ different reporting formats—the extraction matrix establishes a uniform structure.

Standard Components of an EBD Extraction Matrix

1. Study Identifier & Citation: Author(s), publication year, journal name, and country of study.
2. Research Design & Evidence Level: Study typology (e.g., randomized controlled trial, prospective cohort study, quasi-experimental pre/post with control, cross-sectional observational survey) and its formal hierarchy level.
3. Facility Setting & Sample Population: Type of hospital unit (e.g., 30-bed surgical ICU, 450-bed academic medical center), sample size (N), patient demographics, and staff characteristics.
4. Independent Environmental Variable(s): The precise physical design intervention examined, including technical specifications (e.g., acoustic ceiling tiles with NRC 0.90, dynamic LED lighting delivering 300 Equivalent Melanopic Lux, single-patient room with private toilet).
5. Dependent Clinical/Operational Variables: The specific outcome metrics measured (e.g., hospital-acquired C. difficile infection rates, unassisted patient fall rates, average decibels Leq, nursing travel steps per shift).
6. Key Empirical Findings & Statistical Significance: Quantitative results, including effect sizes, percentage changes, and statistical significance markers (p-values, confidence intervals, odds ratios).
7. Methodological Limitations & Confounders: Identified study weaknesses (e.g., lack of a concurrent control group, small sample size, simultaneous change in nurse staffing ratios or EHR systems).
8. Design Translation & Takeaway: Specific, actionable implications for the current architectural project.

Worked Example: Data Extraction Matrix (Illustrative Rows)

The rows below are hypothetical examples that show how to fill in the matrix; they do not summarize specific published studies.

---

Resolving Contradictory or Equivocal Research Findings

A critical competency is the ability to navigate situations where published research yields contradictory, equivocal, or conflicting conclusions. Healthcare design researchers frequently encounter studies that disagree.

Classic Example: Nursing Station Configuration

• Study A finds that decentralized nurse stations reduce nurse walking distance by 30% and increase bedside charting time.
• Study B finds that decentralized nurse stations increase nurse feelings of isolation, decrease multidisciplinary communication, and have no measurable impact on patient fall rates.

When faced with conflicting evidence, an evidence-based designer does not discard the research or pick the study that flatters their bias. Instead, the team executes a four-step conflict resolution protocol:

┌─────────────────────────────────────────────────────────────┐
│         STEP 1: EVALUATE METHODOLOGICAL RIGOR               │
│ • Compare study designs, sample sizes & statistical power    │
│ • A multi-site prospective cohort outweighs an informal POE │
└──────────────────────────────┬──────────────────────────────┘
                               │
                               ▼
┌─────────────────────────────────────────────────────────────┐
│         STEP 2: IDENTIFY CONFOUNDING VARIABLES              │
│ • Were clinical workflows, staffing ratios, or technology    │
│   implemented differently across the study environments?    │
└──────────────────────────────┬──────────────────────────────┘
                               │
                               ▼
┌─────────────────────────────────────────────────────────────┐
│         STEP 3: ANALYZE ENVIRONMENTAL DOSAGE & FIDELITY     │
│ • What were the precise physical parameters of the layout?   │
│ • Did "decentralized" include sightlines & supply alcoves?  │
└──────────────────────────────┬──────────────────────────────┘
                               │
                               ▼
┌─────────────────────────────────────────────────────────────┐
│         STEP 4: EXAMINE CONTEXTUAL OPERATIONAL FIT          │
│ • Which study setting matches the client's care model?      │
│ • Synthesize a hybrid solution addressing identified gaps   │
└─────────────────────────────────────────────────────────────┘

1. Evaluate Methodological Rigor and Hierarchy Level

Studies with stronger designs carry greater evidential weight, and EDAC study materials point to reliability and validity as key tools for weighing conflicting results. A multi-site quasi-experimental study with a concurrent control group and automated sensor tracking possesses substantially higher internal validity than a single-site retrospective survey relying on self-reported nurse recall.

2. Identify Confounding Variables and Operational Co-Interventions

Conflicting findings frequently arise because one hospital altered its operational protocols while changing its physical space. For example, if Study B implemented decentralized nurse stations while simultaneously cutting nurse staffing, the negative nurse responses may reflect staff shortages rather than the physical station layout.

3. Analyze Environmental Intervention Dosage and Architectural Fidelity

Researchers often use identical words to describe radically different physical designs:

• In Study A, the decentralized nurse alcoves featured direct glass sightlines into the patient bed, integrated supply/medication drawers, and a centralized team collaboration hub at the unit entrance.
• In Study B, the decentralized stations were enclosed pods recessed behind solid drywall columns with zero visibility to neighboring nurses and required nurses to walk 150 feet to a distant centralized medication room. The contradiction was not about decentralization itself; it was about the presence or absence of visual connectivity and decentralized supply support.

4. Synthesize a Contextual Design Solution

Resolving the conflict produces a superior, nuanced design directive: implement hybrid nursing stations—pairing decentralized bedside charting alcoves with unobstructed bed sightlines, supported by a central collaborative workspace for interprofessional rounds.

---

Communicating Synthesis: The Evidence-to-Design Translation Matrix

The culmination of Step 2 and Step 3 is converting synthesized scientific findings into a format that design teams, contractors, and healthcare leadership can act upon. The primary tool for this translation is the Evidence-to-Design Translation Matrix.

┌──────────────────┐       ┌──────────────────┐       ┌──────────────────┐
│ RESEARCH EVIDENCE│  ──>  │ DESIGN STRATEGY  │  ──>  │ MEASURABLE TARGET│
│     SUMMARY      │       │ (Spatial Action) │       │   (POE Metric)   │
└──────────────────┘       └──────────────────┘       └──────────────────┘

Practical Application of the Translation Matrix (Illustrative)

Targets such as "reduce falls by X%" should come from the team's baseline data and the strength of the evidence, not from a single study's headline number.

[!TIP]

EXAM TIP: The Nature of Evidence Synthesis

Remember that synthesizing research literature is not simply calculating a mathematical average of study results. Nor does it mean selecting only positive studies while ignoring negative ones.

True evidence synthesis involves critically appraising the methodological quality of each study, reconciling conflicting operational contexts, understanding the physical nuances of the intervention, and extracting the generalizable design principles that directly advance the project's strategic goals.