Applied Health Services, Implementation and Managerial Decision-Making

Introduction

Qualitative research is central to health-care management because many service problems are not reducible to numerical indicators. Waiting-time data may show that access is poor, but interviews can reveal that patients cannot navigate referral rules, staff avoid a pathway they consider unsafe, or managers have created targets that unintentionally distort practice. Similarly, an implementation plan may be technically sound yet fail because it conflicts with professional identity, local routines, power relationships, or patient priorities.

This chapter examines how qualitative methods support needs assessment, program evaluation, implementation research, patient-experience studies, workforce research, quality improvement, and policy analysis. Its central concern is translation: how to move from credible accounts of experience and context to decisions about service design, leadership, resource allocation, implementation, and accountability.

Qualitative evidence does not replace epidemiological, financial, operational, or clinical evidence. Rather, it explains how systems work, why interventions succeed or fail, whose interests are represented, and what changes are feasible and acceptable. At postgraduate level, the researcher or manager must therefore assess not only what participants say, but also how evidence was generated, whose voice is absent, how findings relate to context, and what level of confidence is justified.

Learning objectives

By the end of this chapter, readers should be able to:

1. Explain the contribution of qualitative methods to applied health-services research.
2. Design qualitative needs assessments, evaluations, implementation studies, workforce studies, and patient-experience inquiries.
3. Select appropriate sampling, data-collection, and analytic strategies.
4. Use implementation and evaluation frameworks critically rather than mechanically.
5. Translate qualitative findings into defensible managerial decisions and actionable recommendations.
6. Address reflexivity, ethics, equity, governance, and stakeholder power.
7. Judge the quality, transferability, and practical significance of qualitative evidence.

1. The applied logic of qualitative health research

Applied qualitative research asks questions about meaning, process, context, interaction, and change. Common questions include:

• How do patients experience access to a service?
• Why do clinicians adopt, adapt, or reject a new intervention?
• How does a policy operate in everyday practice?
• What informal work enables a service to function?
• Which barriers and facilitators are modifiable?
• How do organizational decisions affect different population groups?

A useful distinction is between descriptive, interpretive, and critical purposes. Descriptive studies document experiences and processes. Interpretive studies examine how people understand and construct those experiences. Critical studies investigate power, inequality, institutional assumptions, and whose knowledge is treated as authoritative. A single project may combine all three purposes, but the research team should state its primary orientation.

Qualitative evidence is especially valuable where:

• outcomes are difficult to define or measure;
• services depend on relationships and trust;
• implementation varies across settings;
• populations are marginalized or underrepresented in routine data;
• organizational culture and professional norms shape behavior;
• an intervention has unexpected consequences; or
• managers need to understand mechanisms rather than only associations.

The unit of analysis may be an individual, interaction, team, organization, service pathway, policy process, or community. Confusing these levels produces weak conclusions. For example, an interview with one nurse can illuminate a team culture, but cannot by itself establish that all nurses share that view.

Important terminology

Needs assessment is a systematic inquiry into health needs, service gaps, priorities, assets, and barriers in a defined population or system.

Program evaluation is the systematic assessment of a program’s design, implementation, processes, outcomes, value, or consequences.

Implementation research studies how evidence-based practices, policies, or innovations are taken up, delivered, sustained, and adapted in real-world settings.

Patient experience concerns what patients and carers encounter, perceive, feel, and value during interactions with health services. It is distinct from satisfaction, which is a broader evaluative judgment.

Workforce study examines the experiences, practices, identities, conditions, and relationships of health-care workers.

Quality improvement (QI) uses structured, iterative activity to improve local processes and outcomes. QI and research may overlap, but they have different purposes, governance requirements, and claims to generalizability.

Policy analysis examines how policies are formulated, interpreted, implemented, experienced, and contested.

2. From service problem to researchable decision

A managerial problem is usually expressed in operational language: “clinic utilization is low,” “staff turnover is high,” or “the digital referral system is not being used.” A qualitative design begins by reformulating the problem into questions about actors, processes, context, and consequences.

A strong applied question specifies the population, setting, phenomenon, and intended decision. For example: “How do adults with multiple long-term conditions experience referral between primary and specialist care, and what service changes could reduce avoidable duplication?” This is more actionable than “What do patients think about referrals?”

The decision-use pathway should be agreed at the outset:

1. What decision is pending?
2. Who will make or influence it?
3. What evidence would change the decision?
4. What constraints cannot be changed?
5. How will findings be communicated and acted upon?

This prevents data collection becoming an interesting but disconnected description of experience.

3. Designing studies for applied health services

Sampling for relevance and variation

Qualitative sampling is normally purposive rather than statistically representative. Participants are selected because they can illuminate the research question. Strategies include:

• Criterion sampling: all participants meet specified conditions, such as having used a service within six months.
• Maximum-variation sampling: participants differ on characteristics likely to shape experience, such as age, ethnicity, disability, geography, or professional role.
• Typical-case sampling: participants or sites reflect ordinary practice.
• Critical-case sampling: cases are selected because they are especially revealing, such as a service with unusually high uptake.
• Theoretical sampling: later cases are selected to develop or test emerging analytic categories.
• Stakeholder sampling: patients, carers, professionals, managers, commissioners, and community organizations are included to compare perspectives.

Sample adequacy is not determined by a universal numerical rule. It depends on the scope of the question, diversity of the population, quality of data, analytic strategy, and whether additional data are contributing meaningful conceptual insight. Researchers should document the logic of stopping, rather than claiming automatically that “saturation” has occurred.

Data-collection methods

Semi-structured interviews are suitable for personal experience, sensitive topics, professional reasoning, and perceived barriers. An interview guide should contain open questions, prompts, and probes, not a rigid questionnaire. Asking “Can you describe what happened?” often produces richer data than “Was the process satisfactory?”

Focus groups examine shared language, disagreement, norms, and collective priorities. They are useful for service redesign but may suppress dissent where status differences are strong. Separate groups may be needed for patients and professionals, or for junior and senior staff.

Observation reveals work as performed rather than work as described. Shadowing, non-participant observation, and observation of meetings can identify workarounds, interruptions, tacit coordination, and differences between formal policy and actual practice.

Document analysis includes policies, complaints, clinical templates, meeting minutes, training materials, procurement documents, and social-media or public consultation records. Documents should be treated as social products with authorship, purpose, omissions, and institutional assumptions.

Participatory and co-design methods involve patients, carers, staff, or communities in defining questions, interpreting findings, and designing solutions. These methods can redistribute expertise, but participation must be meaningful rather than symbolic.

Digital and remote methods include online interviews, patient portals, discussion forums, and digital ethnography. Researchers must consider unequal digital access, privacy, platform governance, and whether online interaction is comparable to face-to-face interaction.

Reflexivity and positionality

The researcher is part of the knowledge-production process. Professional status, institutional affiliation, language, gender, ethnicity, clinical role, and assumptions can affect access, disclosure, interpretation, and reporting. A reflexive account should identify relevant relationships and decisions, not merely state that bias exists.

In managerial research, the problem is particularly acute when a manager interviews subordinates. Participants may fear consequences, and apparent agreement may reflect organizational hierarchy. Independent interviewers, confidential recruitment, clear separation from performance management, and careful reporting are important safeguards.

4. Analysis and quality of evidence

Applied analysis usually proceeds through familiarization, coding, category development, comparison, interpretation, and synthesis. Coding may be inductive, deductive, or abductive. Inductive coding develops concepts from the data; deductive coding applies a prior framework; abductive analysis moves iteratively between empirical findings and theory to explain surprising patterns.

A practical analytic sequence is:

1. Read transcripts, field notes, and documents repeatedly.
2. Record initial observations, contradictions, and reflexive notes.
3. Apply descriptive and process codes.
4. Compare data within and across participant groups and sites.
5. Develop categories that explain relationships, conditions, and consequences.
6. Search for negative cases and rival explanations.
7. Construct themes or a process model.
8. Test interpretations against the dataset and stakeholders.
9. Translate findings into implications, options, and risks.

A theme is not simply a topic mentioned frequently. It is a patterned and meaningful interpretation relevant to the research question. For example, “communication” is a topic; “patients experience repeated requests for the same information as evidence that the service does not know them” is an interpretive theme.

Quality is multidimensional. Common criteria include credibility, dependability, confirmability, transferability, authenticity, and practical utility. Techniques may include triangulation, reflexive journaling, peer debriefing, negative-case analysis, member reflection, audit trails, thick description, and transparent reporting. None is a ritual guarantee. Returning a transcript to a participant, for example, may improve factual accuracy but does not automatically validate the researcher’s interpretation.

Triangulation can involve different data sources, researchers, theories, or methods. Convergence is not always the goal: disagreement may reveal differences in role, power, or exposure to the service. A patient and a manager may offer genuinely different but valid accounts of the same pathway.

5. Needs assessment and service design

A qualitative needs assessment should distinguish between:

• felt need: what people say they need;
• expressed need: need demonstrated through service use or requests;
• normative need: need judged against professional or policy standards; and
• comparative need: inequity identified by comparing populations or areas.

Qualitative research contributes by showing how needs are prioritized, negotiated, and experienced. It can identify assets as well as deficits: trusted community workers, informal carers, peer networks, culturally specific knowledge, and existing organizational capabilities.

A service-design process may combine journey mapping, interviews, observation, and co-design workshops. The team should map critical touchpoints, emotional and practical burdens, information transitions, points of exclusion, and opportunities for prevention. Recommendations should distinguish between changes to policy, workflow, staffing, environment, communication, and technology.

Avoid treating patient preference as a simple demand that must always be met. Preferences are shaped by information, previous treatment, cultural expectations, affordability, and available alternatives. The managerial task is to make these conditions visible and assess what can be changed ethically and feasibly.

6. Program evaluation

Qualitative evaluation can examine four linked questions:

1. Design: Is the program based on a plausible account of the problem?
2. Implementation: What was delivered, to whom, and under what conditions?
3. Mechanism: How did participants respond, and why?
4. Consequences: What intended and unintended effects occurred?

Process evaluation is particularly important when outcome results are disappointing. A program may have failed because it was not delivered, because the target group could not access it, because the mechanism was implausible, or because the context changed. Interviews, observations, and documents help distinguish these explanations.

Realist evaluation asks, in effect, “What works, for whom, in what circumstances, and why?” Its analytic language of context–mechanism–outcome configurations is useful for management. For example, reminder messages may improve attendance when patients trust the service and can reschedule easily, but have little effect where transport is unreliable and appointments are inflexible.

Qualitative evaluation should be integrated with quantitative outcome data where appropriate. If attendance increases, interviews may explain whether this resulted from improved access, fear of penalties, or changed recording practices. Mixed-methods integration can occur at design, data-collection, analysis, or interpretation stages; merely placing qualitative quotations beside statistics is not integration.

7. Implementation research

Implementation concerns the gap between what is known and what is routinely done. Relevant constructs include acceptability, appropriateness, feasibility, adoption, fidelity, adaptation, reach, cost, equity, and sustainability.

Several frameworks are useful:

• CFIR organizes determinants across intervention characteristics, outer setting, inner setting, individual characteristics, and implementation processes.
• The Consolidated Framework for Implementation Research is best used as a sensitizing structure, not a checklist that prevents unexpected themes.
• Normalization Process Theory examines how practices become embedded through coherence, cognitive participation, collective action, and reflexive monitoring.
• RE-AIM draws attention to reach, effectiveness, adoption, implementation, and maintenance.
• Proctor’s implementation outcomes distinguish outcomes such as acceptability, feasibility, fidelity, penetration, and sustainability from clinical outcomes.

Implementation studies should examine adaptation carefully. Fidelity is not always uniform delivery. Some adaptations preserve the intervention’s core function while improving fit; others remove essential components. Researchers should distinguish the intervention’s core mechanism from its adaptable form.

A useful implementation inquiry compares intended and actual pathways: who was expected to do what, what they actually did, what conditions shaped the difference, and what consequences followed. Frontline workarounds should not automatically be treated as non-compliance; they may indicate that the formal design is incompatible with real work.

8. Patient experience, workforce, quality improvement, and policy

Patient experience

Patient-experience research should examine respect, communication, involvement in decisions, continuity, coordination, dignity, responsiveness, and emotional safety. Complaints are not merely negative feedback; they are situated accounts of institutional encounters. Researchers should analyze both the event and the institutional response to it.

Equity requires attention to people who do not reach the service, cannot use standard feedback mechanisms, have limited literacy, or fear authority. Interpreters, accessible formats, community partnerships, and flexible recruitment are methodological necessities, not optional additions.

Workforce studies

Workforce research should move beyond morale surveys. Interviews and observation can examine workload, emotional labor, moral distress, supervision, discrimination, professional boundaries, psychological safety, and informal coordination. A staff member’s account may reveal system design problems rather than individual resilience deficits. Confidentiality is essential where findings could affect employment or professional reputation.

Quality improvement

Qualitative methods are useful before, during, and after QI cycles. They can identify failure points, explain variation, assess acceptability, and reveal unintended consequences. A small number of focused interviews may be appropriate for rapid-cycle learning, but speed does not justify poor consent, insecure data handling, or unsupported general claims.

Policy analysis

Policy is not only a formal document. It is also interpretation, implementation, negotiation, and street-level practice. Policy analysis can examine problem representation, assumptions, affected groups, administrative discretion, accountability, and consequences. Comparing policy language with frontline accounts often reveals gaps between universal aims and unequal effects.

9. Translating findings into managerial decisions

The central translational product should not be a long list of quotations. It should be a defensible chain from evidence to action:

Finding → interpretation → implication → option → decision → indicator → review point

For example:

Recommendations should be prioritized by expected benefit, equity, feasibility, acceptability, cost, risk, and reversibility. Qualitative evidence is particularly valuable for identifying implementation risks and distributional effects, but managers should not claim that a qualitative study proves cost-effectiveness or population prevalence.

A decision brief should state: the problem, affected groups, evidence base, level of confidence, options considered, recommended action, assumptions, risks, responsible owner, timeline, and monitoring plan. It should also identify what evidence would trigger revision.

10. Illustrative case: improving integrated diabetes care

The following is a fictional illustrative case.

Background

A regional health system introduced an integrated diabetes pathway linking primary care, specialist clinics, pharmacy, and community education. After twelve months, referral completion remained low and patients reported repeated assessments.

Problem

Routine data showed that 62% of referrals were recorded as completed, but the system could not explain why completion varied by neighborhood or whether patients experienced the pathway as integrated.

Analysis

The research team conducted 28 patient interviews, 18 staff interviews, observations of referral meetings, and analysis of referral protocols and complaint records. Maximum variation was used to include older adults, working-age adults, people using interpreters, and patients with complex conditions.

Three themes emerged. First, patients experienced “integration” as continuity of responsibility, yet no professional clearly owned the pathway. Second, staff used parallel spreadsheets because the electronic referral system did not show pending actions. Third, patients in poorer neighborhoods faced transport and appointment-timing barriers that were invisible in completion statistics. A negative case was important: one clinic achieved high completion because a diabetes nurse actively tracked referrals and contacted patients.

Approach and solution

The team recommended a pilot with four linked actions: assign pathway ownership; create a visible pending-referral queue; offer evening appointments and transport advice; and establish a community liaison role. The recommendations were not presented as proven solutions. They were treated as testable service hypotheses.

A six-month implementation study combined observation, brief interviews, and routine monitoring. Staff reported that the queue improved visibility but initially increased perceived workload. The pathway owner therefore received protected time. Community liaison improved engagement among patients who had previously missed appointments, while evening clinics were less useful than expected for people with caring responsibilities.

Outcome

Referral completion improved, but the most important result was organizational learning: managers discovered that the original indicator measured administrative closure rather than completed patient care. The revised dashboard distinguished referral sent, appointment attended, care plan completed, and patient-reported continuity.

Learning points

1. A performance indicator may conceal the process that managers need to understand.
2. Qualitative evidence can identify mechanisms behind variation.
3. Solutions should be treated as context-sensitive and testable.
4. Equity analysis must examine who benefits and who remains excluded.
5. Implementation requires ownership, resources, feedback, and adaptation.

11. Ethics, governance, and responsible use of data

Applied studies involve ethical issues beyond ordinary informed consent. Participants may disclose unsafe care, discrimination, abuse, illegal practice, or serious distress. Protocols should specify limits of confidentiality, safeguarding procedures, data retention, and escalation routes. Researchers must avoid promising absolute confidentiality when disclosure obligations exist.

Organizational access can create coercion. Invitations should make clear that participation is voluntary and unrelated to treatment, employment, appraisal, or service eligibility. Reporting should avoid identifying individuals or small teams through combinations of role, location, and quotation.

Data governance includes lawful collection, secure storage, controlled access, de-identification, retention limits, and transparent secondary use. Audio, transcripts, field notes, photographs, and digital traces may contain identifiable information even after names are removed.

Artificial intelligence can assist transcription, translation, coding, search, and data organization, but it introduces risks of hallucinated text, loss of culturally specific meaning, confidentiality breaches, automation bias, and opaque analytic decisions. Human researchers remain responsible for verification, interpretation, consent, and governance. Sensitive data should not be uploaded to an unapproved system. Any AI assistance should be documented, audited, and checked against original data.

12. Challenges, limitations, and best practice

Qualitative research can be limited by restricted access, socially desirable accounts, incomplete participation, researcher influence, uneven data quality, and the difficulty of transferring findings across settings. These are not reasons to reject qualitative evidence; they are reasons to make claims proportionate to design and data.

Best practice includes:

• aligning the question, paradigm, methods, and intended decision;
• involving affected communities in design and interpretation;
• sampling for relevant variation and documenting exclusions;
• combining accounts with observation and documentary evidence where useful;
• preserving contradictory and negative cases;
• maintaining a transparent analytic audit trail;
• separating evidence, interpretation, and recommendation;
• reporting uncertainty and contextual limits;
• integrating equity and power throughout the study; and
• returning findings in forms that decision-makers and communities can use.

The most useful applied qualitative report is neither an advocacy document nor a detached description. It is a transparent account of how people experience a system, how the system produces those experiences, what can realistically be changed, and what should be monitored after change.

Chapter Summary

Qualitative methods make health-care systems intelligible as lived, organizational, and political processes. They support needs assessment by identifying unmet and differently distributed needs; evaluation by explaining delivery, mechanisms, and consequences; implementation research by examining adoption, adaptation, and sustainability; and patient, workforce, QI, and policy studies by revealing experience, context, power, and unintended effects.

Strong applied studies begin with a decision-relevant question, use purposeful and equitable sampling, collect data suited to the phenomenon, and analyze patterns through reflexive and transparent procedures. Frameworks such as CFIR, Normalization Process Theory, RE-AIM, realist evaluation, and implementation-outcome models can structure inquiry, but should not replace empirical openness.

Managerial translation requires a clear chain from finding to interpretation, option, decision, indicator, and review. Recommendations should be feasible, equity-sensitive, proportionate to evidence, and explicit about risks and uncertainty. Ethical governance, confidentiality, reflexivity, and responsible technology use are essential to trustworthy applied research.

Key Terms

Needs assessment: Systematic inquiry into needs, gaps, priorities, barriers, and assets.

Program evaluation: Assessment of a program’s design, delivery, outcomes, value, or consequences.

Implementation research: Study of how interventions or policies are adopted, delivered, adapted, and sustained.

Patient experience: What patients and carers encounter, perceive, feel, and value in care.

Workforce study: Qualitative inquiry into workers’ practices, conditions, identities, and relationships.

Quality improvement: Iterative activity intended to improve local health-care processes and outcomes.

Policy analysis: Examination of policy formulation, interpretation, implementation, and effects.

Purposive sampling: Selection of participants for their relevance to the research question.

Maximum-variation sampling: Sampling designed to include important differences among participants or sites.

Theoretical sampling: Selection of later data sources to develop emerging concepts or theory.

Reflexivity: Critical examination of how the researcher and research process shape knowledge.

Triangulation: Comparison of perspectives, methods, researchers, theories, or data sources.

Negative case: Evidence that challenges an emerging interpretation or pattern.

Process evaluation: Examination of how a program was delivered and experienced.

Mechanism: The process through which an intervention produces a response or outcome.

Context: Social, organizational, institutional, material, and policy conditions surrounding action.

Normalization Process Theory: Framework explaining how practices become embedded in routine work.

Implementation fidelity: The extent to which essential components are delivered as intended.

Transferability: The reasoned applicability of findings to another context.

Decision-use pathway: The chain linking research findings to a managerial decision and subsequent review.

Review Questions

Short Answer Questions

1. What distinguishes implementation research from program evaluation?
2. Why is purposive sampling appropriate in applied qualitative research?
3. Define reflexivity and explain its importance in managerial research.
4. What is the difference between a topic and an interpretive theme?
5. Identify two ethical risks when managers study their own staff.

Long Answer Questions

1. Explain how qualitative methods can contribute to a health-services needs assessment, from question formulation to service-design recommendations.
2. Compare process evaluation with outcome evaluation and explain why both may be needed.
3. Discuss how CFIR, Normalization Process Theory, and RE-AIM can guide implementation research, including their limitations.
4. Explain how qualitative evidence can be integrated with routine performance data without overstating what either form of evidence can establish.
5. Describe the ethical, governance, and equity considerations involved in a workforce qualitative study.

Analytical / Application Questions

1. A hospital’s data show high discharge completion, yet patients report confusion after leaving. Design a qualitative inquiry and show how its findings could change the discharge process.
2. A digital referral tool is used inconsistently across clinics. Develop an implementation-focused study that distinguishes barriers, adaptations, and unacceptable deviations from the intended design.
3. Using the illustrative diabetes-care case, construct a decision-use pathway from one qualitative finding to a management action, implementation indicator, and review point.

Multiple Choice Questions with Answers

Question 1

Which qualitative method is most appropriate for examining how workarounds occur during routine clinical practice?

A. Non-participant observation

B. A population prevalence survey

C. A randomized controlled trial

D. A financial audit

Correct Answer: A. Non-participant observation

Question 2

What is the main purpose of maximum-variation sampling?

A. To achieve statistical representativeness

B. To include participants with relevant differences in experience

C. To select only the most typical participant

D. To eliminate the need for reflexivity

Correct Answer: B. To include participants with relevant differences in experience

Question 3

Which statement best describes a qualitative theme?

A. Any topic mentioned by at least one participant

B. A direct quotation selected for emphasis

C. A patterned and meaningful interpretation relevant to the question

D. A numerical summary of coded responses

Correct Answer: C. A patterned and meaningful interpretation relevant to the question

Question 4

In implementation research, acceptability refers primarily to:

A. The financial price of an intervention

B. The extent to which stakeholders consider an intervention agreeable or satisfactory

C. The number of patients receiving treatment

D. Whether an intervention has been randomized

Correct Answer: B. The extent to which stakeholders consider an intervention agreeable or satisfactory

Question 5

A process evaluation is especially useful when:

A. A program’s outcome is known but delivery and mechanisms are unclear

B. No program or intervention exists

C. Only national prevalence estimates are required

D. The researcher wants to avoid studying context

Correct Answer: A. A program’s outcome is known but delivery and mechanisms are unclear

Question 6

Which practice most directly addresses the risk that a manager’s staff may feel pressured to participate?

A. Recruiting only high-performing staff

B. Linking participation to annual appraisal

C. Using independent recruitment and separating research from performance management

D. Reporting individual quotations with job titles

Correct Answer: C. Using independent recruitment and separating research from performance management

Question 7

What does a negative case contribute to qualitative analysis?

A. It automatically invalidates the entire study

B. It tests and may refine an emerging interpretation

C. It proves that the largest group is correct

D. It replaces the need for data collection

Correct Answer: B. It tests and may refine an emerging interpretation

Question 8

Which sequence best represents a decision-use pathway?

A. Finding → interpretation → option → decision → indicator → review

B. Quotation → publication → recruitment → diagnosis → budget

C. Policy → sample → transcript → randomization → outcome

D. Indicator → finding → consent → sampling → intervention

Correct Answer: A. Finding → interpretation → option → decision → indicator → review

Question 9

Why should adaptation be distinguished from loss of fidelity in implementation research?

A. All adaptations are failures

B. Adaptation may preserve the core mechanism while improving contextual fit

C. Fidelity means that every setting must use identical procedures

D. Adaptation can only occur before implementation begins

Correct Answer: B. Adaptation may preserve the core mechanism while improving contextual fit

Question 10

Which is a responsible use of artificial intelligence in qualitative health research?

A. Uploading identifiable interviews to an unapproved public tool

B. Accepting automatically generated codes without checking source data

C. Using approved tools with human verification, governance, and documented assistance

D. Allowing an AI system to make final safeguarding decisions

Correct Answer: C. Using approved tools with human verification, governance, and documented assistance