Apr 17, 2026
Development and Validation of a Multi-Dimensional Caregiver Taxonomy for Older Adult Caregivers: Study Protocol
- Ahd Shahin1,
- Eman Zaid Almaazmi1,
- Lama AlKhuja2,
- Bassam Al Khameri1,
- Khalifa Baqer3,
- Kabir Girohtra1,
- Srishti Mehrotra1,
- Alali Mohamed2,
- Abu Omayer2,
- Nabil Zary2
- 1Community Care Department, Dubai Health;
- 2Institute of Learning, Mohammed Bin Rashid University of Medicine and Health Sciences,Dubai Health;
- 3Operations Sector, Dubai Health
- NeuroInk
Protocol Citation: Ahd Shahin, Eman Zaid Almaazmi, Lama AlKhuja, Bassam Al Khameri, Khalifa Baqer, Kabir Girohtra, Srishti Mehrotra, Alali Mohamed, Abu Omayer, Nabil Zary 2026. Development and Validation of a Multi-Dimensional Caregiver Taxonomy for Older Adult Caregivers: Study Protocol. protocols.io https://dx.doi.org/10.17504/protocols.io.dm6gp7kb1gzp/v1
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: April 16, 2026
Last Modified: April 17, 2026
Protocol Integer ID: 315185
Keywords: caregiver, taxonomy, classification, typology, older adults, elder care, health workforce, Delphi, content validity, cognitive interviewing, mixed methods, cross-cultural adaptation, domestic workers, study protocol caregivers of older adult, study protocol caregiver, older adult caregiver, discrete caregiver type, care function, older adult, care intensity, intervention customization, multidimensional taxonomy, taxonomy development, formal professional, methods protocol spans 21 month, systematic literature synthesis, methods protocol span, professional status
Disclaimer
DISCLAIMER – FOR INFORMATIONAL PURPOSES ONLY; USE AT YOUR OWN RISK
The protocol content here is for informational purposes only and does not constitute legal, medical, clinical, or safety advice, or otherwise; content added to protocols.io is not peer reviewed and may not have undergone a formal approval of any kind. Information presented in this protocol should not substitute for independent professional judgment, advice, diagnosis, or treatment. Any action you take or refrain from taking using or relying upon the information presented here is strictly at your own risk. You agree that neither the Company nor any of the authors, contributors, administrators, or anyone else associated with protocols.io, can be held responsible for your use of the information contained in or linked to this protocol or any of our Sites/Apps and Services.
Abstract
Caregivers of older adults constitute a heterogeneous population that includes informal family members, formal professionals, volunteers, and hybrid arrangements across diverse economic and cultural contexts. No standardized, multidimensional taxonomy exists to classify this heterogeneity, limiting cross-study comparisons and intervention targeting.
This protocol systematically develops and validates a multidimensional taxonomy that operationalizes six core dimensions (D1–D6: relationship to care recipient, formality and professional status, payment status, role prominence and decision authority, care functions provided, care intensity and regularity) and eight discrete caregiver types (T1–T8), suitable for research stratification, intervention customization, and policy development.
The mixed-methods protocol spans 21 months (M0–M21) and uses systematic literature synthesis, testing of dimension independence, expert consensus (Modified Delphi with n=25 international experts), cognitive interviewing, and algorithm validation. Paper I implements taxonomy development (Phases 1–2); Paper II implements validation and cultural calibration (Phase 3).
Pre-specified content-validity thresholds: I-CVI ≥ 0.78 per item; S-CVI/Ave ≥ 0.90 for core dimensions; S-CVI/Ave ≥ 0.80 for extension dimensions; S-CVI/Ave ≥ 0.85 for types; vignette inter-rater κ ≥ 0.70.
Materials
OSF Supplementary Materials Repository
Appendices (conceptual framework, dimension operationalizations, instruments, algorithm pseudocode, figures, reporting checklists)
Before start
Institutional Review Board: Research Ethics Committee (application prepared; submission pending). Protocol
follows the principles of the Declaration of Helsinki, UAE research regulations, and the GDPR for European expert panel participants.
Informed consent: Written informed consent obtained for all expert panelists and cognitive-interview participants; explicit agreement to participate across all Delphi rounds; anonymity maintained via unique ID codes.
Data management: De-identified data; encrypted storage (AES-256 at rest, TLS 1.3 in transit); access restricted to the core research team; 5-year retention post-publication.
Pre-registration: All consensus thresholds, decision rules, and analysis procedures are pre-registered prior to expert recruitment to prevent post-hoc threshold adjustment. This protocols.io entry is part of that pre-registration.
Risk level: Minimal risk — expert consultation and cognitive interviewing only; no medical intervention.
Scope Definition and Target Population
Define the taxonomy scope and the inclusion/exclusion criteria for the target caregiver population.
The target population is caregivers (primary or secondary) of community-dwelling older adults (age ≥ 60 years) providing unpaid or paid care.
Inclusion categories:
(a) informal family caregivers: spouse, adult child, sibling, or other relative providing unpaid care;
(b) informal non-kin caregivers: friends, neighbors, faith-community members;
(c) formally employed caregivers: salaried or hourly workers in healthcare, home care, or social services; (d) volunteers: unpaid, formally affiliated with health systems or NGOs;
(e) hybrid arrangements: paid family caregivers, volunteers with professional oversight.
Exclusion: institutional caregivers employed in hospitals, nursing homes, or permanent residential facilities.
Geographic scope: high-income, middle-income, and low-resource settings; Western, Islamic, Asian, African, and Latin American contexts. The breadth is intentional to enhance cross-cultural
relevance and international applicability.
Success criterion: scope document reviewed and approved by the Research Ethics Committee and the research advisory board.
Systematic Literature Synthesis
Synthesize existing caregiver classification approaches and the empirical evidence on dimensions of caregiver heterogeneity.
Databases: PubMed (1990–2026), CINAHL, PsycINFO.
Hand-searches: Journal of Aging and Health, The Gerontologist.
Search terms combine the MeSH headings “Caregivers,” “Family Nursing,” and “Community Health Nursing” with “taxonomy,” “typology,” “classification,” “heterogeneity,” “burden,” and “outcomes.”
Inclusion: peer-reviewed studies, systematic reviews, and policy documents with ≥ 30 participants.
Exclusion: non-English language, editorials, and case studies.
Per-study data extraction:
(a) dimensions used for caregiver classification;
(b) empirical findings linking dimensions to burden, health, or intervention response;
(c) sample characteristics and
context.
Output: preliminary extraction table listing all identified dimensions from the literature, with an
evidence-synthesis narrative.
Success criterion: extraction table completed; ≥ 50 publications reviewed; preliminary dimension list documented.
Dimension Independence Testing
Confirm that the six core dimensions (D1–D6) are empirically independent and non-redundant.
Core dimensions (D1–D6):
- D1: Relationship to Care Recipient (spouse, adult child, other relative, non-kin, professional)
- D2: Formality and Professional Status (informal/volunteer vs. formal employment)
- D3: Payment Status (unpaid vs. paid; full-time vs. part-time)
- D4: Role Prominence and Decision Authority (primary vs. secondary; primary decisions vs. supportive)
- D5: Care Functions Provided (personal care, medical care, financial/legal, supervision, emotional support, household management)
- D6: Care Intensity and Regularity (hours/week, daily vs. episodic, primary vs. backup)
Optional extension dimensions (CD1–CD7) per Paper 1 MA2 canonical operationalisation:
- CD1: Care Setting and Co-residence
- CD2: Geographic Proximity
- CD3: Care Recipient Condition / Diagnosis
- CD4: Caregiver Employment Status and Work Impact
- CD5: Financial and Structural Context
- CD6: Caregiver Health Status
- CD7: Social Support and Service Environment
Naming convention note. Within the Research Program, the “CD” prefix appears in two distinct namespaces: In Papers 1 and 2 (this protocol), CD1–CD7 denotes Contextual (extension) Dimensions of the caregiver taxonomy;
In Papers 3 onwards, CD1–CD7 denote the seven Competency Domains of the caregiver competency framework. Both usages are retained for citation continuity and are disambiguated by context.
Independence testing: Spearman rank correlation coefficients and variance inflation factors (VIFs) for all dimension pairs, using literature data and expert panel input.
Decision rule: dimensions with R > 0.85 or VIF > 3.0 indicate confounding and are consolidated.
Success criterion: all core-dimension pairs R < 0.85 and VIF < 3.0; optional dimensions assessed for inclusion.
Type Development and Precedence Hierarchy
Define eight discrete caregiver types (T1–T8) using a precedence hierarchy and decision-tree logic.
Precedence hierarchy (priority order for type assignment; when a caregiver meets multiple type criteria, the highest-priority type — the lowest number — is assigned):
- T1: Registered Nurse, Institutional (D6 = “licensed” AND institutional AND formal)
- T2: Certified Nursing Assistant, Institutional (D6 = “certified” AND institutional AND formal)
- T3: Home Health Aide, Formal Employment (home-based AND formal/semi-formal AND professional)
- T4: Personal Care Assistant, Informal (home-based/mixed AND informal AND professional/non-kin)
- T5: Family Caregiver, Primary (family AND informal AND ≥ 4 h/day)
- T6: Family Caregiver, Secondary (family AND informal AND < 4 h/day)
- T7: Community Health Worker (community-based AND semi-formal AND health education/oversight)
- T8: Migrant Domestic Worker, Caregiving Functions (home-based AND informal, AND professional AND migrant status)
Type profiles document dimensional characteristics, demographic features, expected burden signatures, typical conditions, intervention needs, and research applicability for each type.
A complete Type Definition Matrix is maintained as Multimedia Appendix 15 (available in the OSF project repository).
Success criterion: all eight type definitions documented with rationale; precedence hierarchy validated through cognitive interviews.
Classification Algorithm Development
Develop a formal decision-tree algorithm that translates dimensional data into type assignments with
confidence scoring.
Algorithm specification: pseudocode plus reference implementations in R, Python, and SAS.
Input: dimension data (D1–D6).
Output: (a) type assignment (T1–T8); (b) confidence score (0.00–1.00); (c) a flag for UNCLASSIFIABLE cases.
Confidence-score interpretation:
- ≥ 0.85: highly confident; suitable for analysis
- 0.70–0.85: moderately confident; note in analysis
- 0.60–0.70: low confidence; provisional
- < 0.60 or UNCLASSIFIABLE: exclude or flag for resolution
Missing-data handling: count missing core dimensions; determine whether the missing dimensions prevent type assignment; reduce confidence proportionally when non-definitional dimensions are missing; document assumptions.
Sensitivity analysis: test the algorithm on ten archetypal caregiver profiles with known classifications; require identical results across all test cases.
Implementation validation: train coding staff on the algorithm; verify reproducibility across coders.
Success criterion: fully specified algorithm implemented in at least two software environments; 10-profile test suite classified with 100 % accuracy.
Survey Instrument Development
Operationalize the taxonomy into a 40–60-item self-report survey suitable for paper, online, and interview
administration.
Item development: plain language (Grade 6–8 reading level); culturally neutral; measurable for each dimension.
Example items:
- D1 (Relationship): “What is your relationship to the person you care for?” (eight options: spouse, adult child, parent, sibling, other family, friend/neighbor, professional, volunteer)
- D3 (Payment): “Are you paid for care?” (unpaid, irregular payment, part-time paid, full-time paid)
- D6 (Intensity): “How many hours per week do you spend caring?” (< 5, 5–15, 16–35, > 35)
- D5 (Functions): “Which activities do you perform?” — checklist: personal care, medications, household, financial, emotional support, supervision
Branching logic: respondents complete only items relevant to their category — e.g., paid caregivers skip
unpaid-specific items.
Pilot testing: cognitive interviews with 5–10 caregivers to ensure comprehension before validation studies.
Finalization: target 40–60 items total across all dimensions.
Success criterion: survey completed and pilot-tested; branching logic functional in paper and digital formats; comprehension ≥ 85 %.
Expert Consensus Validation (Modified Delphi)
Achieve international expert consensus on taxonomy elements using a three-round Modified Delphi.
Expert panel: n = 25 international experts stratified across six WHO regions, nine disciplines, and twelve
countries. Minimum 40 % from low- and middle-income countries; minimum two from
non-Western geographies.
Round 1 (Weeks 1–4, relative). Individual expert ratings of all dimensions (D1–D6, CD1–CD7) and types (T1–T8), and the algorithm, on 4-point Likert scales for relevance, clarity, comprehensiveness, and distinctiveness, with free-text feedback.
Round 2 (Weeks 5–8, relative). Aggregated results (bar charts, qualitative themes); modified items and ten to twelve clinical vignettes presented; experts re-rate and reclassify vignettes.
Round 3 (Weeks 9–10, relative; conducted only if S-CVI/Ave < 0.90 for core dimensions after Round 2).
Pre-specified consensus thresholds:
- I-CVI (item-level): ≥ 0.78 per item
- S-CVI/Ave (core dimensions D1–D6): ≥ 0.90
- S-CVI/Ave (extension dimensions CD1–CD7): ≥ 0.80
- S-CVI/Ave (types T1–T8): ≥ 0.85
- Algorithm: S-CVI/Ave ≥ 0.85; vignette inter-rater κ ≥ 0.70
Success criterion: core dimensions reach S-CVI/Ave ≥ 0.90; all thresholds and decision rules pre-registered at protocols.io prior to expert recruitment.
Cognitive Interviewing
Validate response-process validity and ensure the comprehensibility of the taxonomy across diverse caregiver populations.
Sample: 24–30 caregivers stratified across types (minimum three per type), contexts (informal family, paid formal, volunteer, hybrid), and geographies (minimum two countries).
Interview protocol (20–30 minutes):
- Part 1: Background and caregiving context.
- Part 2: Self-classification into one of eight types.
- Part 3: Structured probing per COSMIN standards: paraphrasing, distinctiveness, confidence; explore term clarity.
- Part 4: Open feedback on comprehension and suggested improvements.
Analysis: transcript analysis to identify comprehension gaps, term ambiguities, and type-boundary confusion.
Success criteria: ≥ 85 % of caregivers correctly self-identify their type; researcher-caregiver classification agreement κ ≥ 0.70.
Empirical Type Validation
Empirically validate that caregivers can be reliably sorted into eight types using dimensional criteria.
Method: comparison of caregiver self-classification with algorithmic type assignment.
Sample: n = 40 caregivers, five per operational type across T1–T8.
Data collection: self-reported type assignment alongside completion of the dimensional survey items.
Agreement analysis: Cohen’s κ for self-classification vs. algorithm classification.
Target threshold: κ ≥ 0.80 (substantial agreement).
Contingency: if κ < 0.80, conduct additional cognitive interviews with misclassified caregivers to identify
whether type definitions require refinement before deployment.
Success criterion: κ ≥ 0.80 achieved; if κ < 0.80, misclassification patterns are documented, and type revisions are completed prior to Papers 4b and 5b advancing.
Cultural Calibration
Adapt and validate the taxonomy for cross-cultural applicability; begin with an English–Arabic translation and UAE cultural equivalence testing.
Translation protocol: ISPOR 12-step guidelines. Forward translation, back-translation, expert-panel review,
cognitive debriefing (n = 10), pilot testing (n = 30).
Language pairs: English ‱ Arabic (primary).
Future expansion of languages: Hindi, Urdu, Tagalog, Malayalam, Tamil, Bengali.
Cultural-equivalence assessment: semantic equivalence (items translate without loss of meaning), conceptual equivalence (dimensions are culturally applicable), and operational equivalence (scoring is
interpretable across cultures).
Cognitive interviews: ensure translated items are comprehensible to Arabic-speaking caregivers; identify cultural adaptations needed.
Documentation: translation decision log; rationale for any deviations from literal translation.
Success criterion: Arabic translation pilot-tested with ≥ 30 caregivers; comprehension ≥ 85 %; cognitive-interview findings documented.
Post-Deployment Research Platform
Establish infrastructure enabling ongoing taxonomy validation and refinement across diverse healthcare settings.
Open-access repository: publication of the final taxonomy, algorithm code (R, Python, SAS), survey instrument, and decision trees.
Implementation guidance: training manual for coding staff, FAQ, and troubleshooting guides for edge cases.
Proof-of-concept deployment: pilot implementation in two to three healthcare settings, public sector, private agencies, and community programs.
Feedback mechanism: structured reporting of type-assignment challenges, algorithmic failures, and cultural variations; quarterly review.
Versioning protocol: document all modifications and rationale; maintain backward compatibility where possible.
Success criterion: platform operational; ≥ 2 pilot sites recruited; feedback mechanism established.
Expected Outcomes
Primary outcomes:
(a) Six core dimensions validated as independent.
(b) eight discrete types defined with S-CVI/Ave ≥ 0.90;
(c) classification algorithm with confidence scoring;
(d) survey instrument validated through cognitive interviews at κ ≥ 0.70.
Secondary outcomes:
(a) Modified Delphi international consensus;
(b) type assignability empirically tested at κ ≥ 0.80;
(c) cultural adaptation completed for English–Arabic.
Decision rules:
If I-CVI < 0.78 for any item, revise and re-test;
If S-CVI/Ave < 0.90 for core dimensions, conditional Delphi Round 4;
If κ < 0.70 for type classification, refine type boundaries before any onward RCT deployment.
Acknowledgements
This work is supported by the Institute of Learning (IoL) internal research resources at Mohammed Bin Rashid University of Medicine and Health Sciences (MBRU). No external sponsors. No commercial interests in the taxonomy or its derivatives.