Conceptual and Theoretical Frameworks

Overview

A framework is an organised representation of how a researcher understands a problem. It identifies important concepts, expected relationships, contextual influences and assumptions. A good framework connects the literature review to the research question, methods and interpretation. It is not merely a diagram added after the protocol has already been designed.

Terminology varies across disciplines. A theory offers a general explanation of phenomena and relationships. A theoretical framework applies one or more established theories to a study. A conceptual framework is the study-specific organisation of concepts and proposed relationships, informed by theory, evidence and context. A logic model links resources, activities, outputs and outcomes for a programme. A causal diagram represents assumed causal directions under formal graphical rules [1–3].

These tools can overlap, but their purposes should be stated.

Why this matters

Without an explicit framework, study decisions may become a list of disconnected variables. Researchers may measure what is readily available, omit important pathways, adjust for the wrong variables or interpret findings without considering context.

A framework improves coherence by showing why particular exposures, outcomes, mechanisms and contextual factors are included. It also reveals uncertainty. An arrow or relationship in a framework is an assumption or hypothesis, not a proven fact.

Distinguishing common framework types

Theory

A theory contains general concepts and propositions intended to explain or predict phenomena. Examples in health research include theories of behaviour, implementation, stress and social organisation. A theory is broader than one dataset.

Theoretical framework

A theoretical framework selects concepts and propositions from an established theory and applies them to the study. The researcher should explain why the theory fits the problem and where it may not capture the local context.

Conceptual framework

A conceptual framework is tailored to the study. It can combine theory, empirical evidence, clinical knowledge and stakeholder perspectives. It may include variables, processes, context and anticipated relationships. Its value lies in making the study’s reasoning traceable [1].

Logic model

A logic model commonly represents:

inputs → activities → outputs → short-term outcomes → longer-term outcomes

It is particularly useful for intervention and programme evaluation. It does not by itself establish causal identification or specify all confounding pathways.

Causal diagram

A directed acyclic graph uses nodes and arrows to encode causal assumptions and support decisions such as covariate adjustment. It follows stricter rules than a general conceptual diagram. The next lecture develops this method.

Framework types and principal uses

illustrative

Table 1. Choosing a framework form

Form	Main question	Typical product
Theory	What general explanation is proposed?	Concepts and propositions
Theoretical framework	How does an established theory guide this study?	Selected theoretical constructs and relationships
Conceptual framework	How does this study organise the problem?	Study-specific concepts, pathways and context
Logic model	How should a programme produce outcomes?	Inputs, activities, outputs and outcomes
Causal diagram	Which causal assumptions govern effect estimation?	Directed nodes, paths and adjustment implications

Sources for a framework

A framework may be informed by:

• established theory;
• systematic or structured literature synthesis;
• clinical and service knowledge;
• prior qualitative research;
• patient and public perspectives;
• policy and organisational context;
• implementation experience; and
• causal assumptions.

No single source is automatically sufficient. A theory developed in another context may omit local service constraints. Local expert opinion may overlook evidence from other settings. Combining sources should be transparent rather than presenting a newly assembled framework as an established theory.

Steps in developing a conceptual framework

1. Define the purpose and outcome

State whether the framework is intended to explain occurrence, guide an intervention, organise qualitative inquiry, select variables, support causal analysis or plan implementation.

2. Identify central concepts

Start with the research question and evidence synthesis. Include concepts needed to represent the outcome, exposure or intervention, mechanisms, context and competing explanations.

3. Define each concept

A label such as “access” is insufficient. Specify its meaning, dimensions and boundaries in this study.

4. Propose relationships and time order

Describe which factors are expected to precede, influence, mediate, modify or result from others. Distinguish evidence-supported relationships from plausible hypotheses.

5. Represent levels

Health outcomes may arise across patient, family, clinician, clinic, community and policy levels. A framework should not force system-level processes into individual variables.

6. Map concepts to methods

For each element, identify whether it will be measured quantitatively, explored qualitatively, represented through routine data, held as contextual interpretation or left unmeasured.

7. Review with stakeholders

Ask whether important experiences, mechanisms or unintended consequences are missing. Review should include people with relevant lived and operational knowledge.

8. Revise and document

Frameworks can evolve as knowledge develops. Record the version used for protocol decisions and distinguish prospective from post hoc changes.

PROTOCOL CHECKLIST

Framework development record

• purpose and intended use;
• theories, evidence and stakeholder sources;
• definitions of all central concepts;
• assumed directions and time order;
• individual, service and structural levels;
• mapped variables, instruments or qualitative domains;
• unmeasured but important elements;
• implications for design and analysis; and
• date, version and rationale for revisions.

Building the postpartum follow-up framework

The team proposes five domains.

Service coordination

Referral generation, appointment booking, transfer of information, reminder systems and responsibility between maternity and primary care.

Perceived need

Understanding of future diabetes risk, perceived susceptibility, expected benefit of testing and clarity of recommendations.

Practical access

Travel, cost, clinic hours, appointment availability, work and caregiving responsibilities.

Social and relational support

Family support, clinician communication, trust and assistance navigating care.

Outcome and downstream consequences

Test completion, timeliness, abnormal result detection, clinical follow-up and participant experience.

The framework assumes that appointment booking can reduce coordination burden; reminder receipt may reinforce perceived need; practical access can limit attendance even when motivation is high; and clinic capacity shapes both intervention fidelity and outcome.

Study-specific framework matrix

illustrative

Table 2. From conceptual domain to method

Domain	Proposed pathway	Quantitative representation	Qualitative exploration
Service coordination	Booking reduces work required to arrange care	booking status, referral completeness, reminder receipt	experiences of transfer between services
Perceived need	Risk understanding influences priority given to testing	risk-perception items	how women interpret postpartum risk
Practical access	Constraints impede attendance	travel time, cost, appointment availability	competing responsibilities and service fit
Social support	Practical and emotional support enables attendance	childcare/support indicators	negotiation of support within households
Clinic context	Capacity affects booking and test availability	clinic, staffing and appointment indicators	staff workflow and implementation barriers

Mapping the framework to measurement

A framework becomes operational when each important concept is linked to a measurement strategy or an explicit decision not to measure it. For every node or domain, ask whether the study needs a direct observation, an established instrument, several indicators, a routine-data proxy, qualitative exploration or contextual description.

The mapping should reveal gaps. If clinic capacity is central to the proposed explanation but only patient questionnaires are collected, the framework and methods are misaligned. If trust is represented by one locally written item, the team should acknowledge the narrow measurement rather than imply comprehensive assessment.

A useful measurement map includes the construct definition, level, source, time point, instrument or coding rule, expected direction and role in the analysis. It should also identify constructs that cannot be measured well. Unmeasured concepts remain part of the interpretation and may indicate residual confounding or incomplete explanation.

Frameworks can guide pilot work. Cognitive interviews may test whether questionnaire items represent the intended concepts. Service mapping may show that referral responsibility differs from the assumed pathway. Early findings can justify revision, but the version used for confirmatory decisions should be preserved.

Frameworks for quantitative research

A quantitative framework can guide:

• variable selection;
• operational definitions;
• timing of measurement;
• hypothesised directions;
• confounder and mediator distinction;
• interaction questions;
• outcome hierarchy; and
• interpretation of unexpected associations.

Not every framework element must enter one regression model. Some elements may define sampling, stratification, process evaluation or sensitivity analyses.

Frameworks for qualitative research

A framework can provide sensitising concepts and organise an interview guide, but it should not force participant accounts into predetermined categories. Researchers should remain open to concepts and relationships that challenge the initial framework. Reflexivity includes examining how the selected theory shapes attention and interpretation.

Frameworks for mixed-methods research

A common framework can show where quantitative and qualitative strands contribute different evidence. Quantitative data may estimate pathway frequency, while interviews explain how and why the pathway operates. Integration should occur at the level of the research question and interpretation, not merely by placing two datasets in the same project.

Logic models for intervention planning

A logic model for discharge booking might specify:

• inputs: trained staff, scheduling access and linked information;
• activities: eligibility check, appointment booking, written instructions and reminders;
• outputs: proportion receiving a booked appointment and reminder;
• proximal outcomes: reduced coordination burden and increased attendance;
• longer outcomes: timely detection and ongoing preventive care.

The logic model should include assumptions, external factors and possible harms, such as appointments that are booked but inaccessible or increased workload that displaces other services.

Revising a framework after data collection

Unexpected findings may reveal missing concepts or incorrect pathways. Revision is legitimate, but the original and revised frameworks should be distinguished. The original framework documents prospective reasoning; the revised framework may explain new hypotheses generated by the study. Presenting a post hoc framework as if it had guided the original design hides learning and overstates confirmation.

When the study is iterative, specify planned points for framework review, the evidence considered and the authority for changes. Keep dated versions so readers can distinguish planned explanation from interpretation developed after observing findings. This is especially important in mixed-methods and implementation research, where emerging qualitative findings may refine subsequent data collection.

Appraising a framework

APPRAISAL CHECKLIST

Is the framework methodologically useful?

• Is its purpose explicit?
• Are concepts defined rather than merely named?
• Are relationships and time order understandable?
• Does it represent relevant individual, service and structural levels?
• Is it grounded in theory, evidence and context?
• Do measurements and qualitative domains map to the framework?
• Are important omissions and uncertainties acknowledged?
• Does the framework influence design and interpretation?

Ethical and equity considerations

Frameworks determine whose explanations are made visible. A model focusing only on knowledge and motivation may blame individuals while excluding cost, service design, discrimination or caregiving burden. Stakeholder involvement can identify missing pathways and inappropriate assumptions.

Interface with statistical analysis

The conceptual framework should precede model specification. It can identify candidate confounders, mediators and modifiers, but a formal causal diagram is preferable when the goal is selecting an adjustment set for a causal effect. Prediction models require a separate framework based on prediction time, candidate information and intended use.

The analysis plan should state which framework relationships are:

• directly tested;
• used to select adjustment variables;
• explored as mechanisms;
• examined as effect modification;
• addressed qualitatively; or
• retained as contextual assumptions.

Using implementation frameworks without framework overloadMMed extension

Implementation research offers determinant frameworks, process models and evaluation frameworks. The Consolidated Framework for Implementation Research is one influential determinant framework that organises intervention, inner-setting, outer-setting, individual and process factors [4]. A study should select a framework because it addresses the implementation question, then adapt and operationalise it transparently. Combining several frameworks without explaining their distinct functions can create an unmanageable list rather than a coherent design [5].

Worked example

WORKED EXAMPLE

From evidence synthesis to a study framework

Evidence synthesis: Completion is associated with referral processes, perceived need, practical barriers and service integration. Reminder-only interventions have variable effects.

Framework purpose: Explain where women are lost in the pathway and guide development of a booking-and-navigation intervention.

Central outcome: laboratory-confirmed testing completion by day 112.

Proposed pathway: discharge booking reduces the administrative burden of arranging care; clear counselling and reminders support perceived need; navigation addresses practical barriers; clinic capacity affects whether the intervention is delivered and appointments remain accessible.

Measurements: booking status, referral completeness, reminder receipt, travel burden, appointment availability, risk-perception indicators and outcome records.

Qualitative domains: interpretation of risk, coordination experiences, household responsibilities, trust and clinic workflow.

Design implication: a mixed-methods feasibility study can assess pathway operation before a definitive effectiveness trial.

Analysis implication: clinic context and baseline differences require explicit consideration; post-booking processes should not be treated automatically as baseline confounders.

Suggested protocol or reporting wording

The revision process should also record which new findings prompted each change.

Common mistakes

Knowledge check

How does a conceptual framework differ from a theory?

Answer and reasoning

A theory offers a more general explanation through defined concepts and propositions. A conceptual framework is tailored to the study and may combine theory, empirical evidence and context to organise the specific problem and methods.

Is a logic model the same as a causal directed acyclic graph?

Answer and reasoning

No. A logic model describes how programme resources and activities are expected to produce outputs and outcomes. A causal DAG follows formal graphical rules and can support identification of causal paths and adjustment sets. A logic model may inform a DAG but does not replace it.

Should every concept in a framework be included in one statistical model?

Answer and reasoning

No. Some concepts guide sampling, measurement, qualitative inquiry, process evaluation or interpretation. Model inclusion depends on the specific estimand and causal role, not merely presence in the framework.

Why involve patients and service staff when developing a framework?

Answer and reasoning

They can reveal pathways, language, constraints and unintended consequences that are absent from published theory or researchers’ assumptions. Their input improves relevance but should still be integrated transparently with other evidence.

Key take-home messages

SUMMARY

Key take-home messages

• Frameworks organise assumptions and connect evidence to study design.
• Theory, conceptual frameworks, logic models and causal diagrams have related but distinct purposes.
• Define concepts, time order, levels and pathways before mapping measurements.
• A framework should influence methods and interpretation, not merely decorate the protocol.
• Include service and structural context to avoid over-individualising health problems.

References

1. Kulesa J, Induru S, Hubbard E, Bhansali P. The conceptual framework: a practical guide. Hosp Pediatr. 2024;14(11):e503–e508. https://doi.org/10.1542/hpeds.2024-007794
2. Battista RN, Contandriopoulos AP, Champagne F, Williams JI, Pineault R, Boyle P. An integrative framework for health-related research. J Clin Epidemiol. 1989;42(12):1155–1160. https://doi.org/10.1016/0895-4356(89)90113-3
3. Jabareen Y. Building a conceptual framework: philosophy, definitions, and procedure. Int J Qual Methods. 2009;8(4):49–62. https://doi.org/10.1177/160940690900800406
4. Damschroder LJ, Aron DC, Keith RE, Kirsh SR, Alexander JA, Lowery JC. Fostering implementation of health services research findings into practice: a consolidated framework for advancing implementation science. Implement Sci. 2009;4:50. https://doi.org/10.1186/1748-5908-4-50
5. Nilsen P. Making sense of implementation theories, models and frameworks. Implement Sci. 2015;10:53. https://doi.org/10.1186/s13012-015-0242-0