Care Team Design

Module 5: Organizational Design and Team Coordination Depth: Application | Target: ~2,000 words

Thesis: Care teams are coordination systems — their effectiveness depends on role clarity, communication protocols, shared accountability, and the organizational support that enables all three.

The Care Team as a Designed System

A care team is not a collection of providers who share patients. It is a coordination system with defined roles, communication protocols, shared mental models, and collective accountability for a defined population. The distinction matters because teams that are merely assembled — providers who happen to treat the same patients — produce coordination by accident. Teams that are designed produce coordination by architecture.

The difference is structural, not attitudinal. An assembled team relies on individual initiative for information sharing: the nurse calls the physician when something seems wrong, the care coordinator follows up when they remember, the social worker intervenes when a problem is visible. A designed team has protocols that make coordination automatic: standing orders that define who acts under what conditions, huddle structures that synchronize information daily, escalation pathways that specify who decides what and when. The assembled team coordinates when individuals choose to. The designed team coordinates because the system requires it.

Gittell’s research on relational coordination (2002, 2009) provides the theoretical foundation. Gittell studied surgical teams, airline operations, and primary care practices and found that performance depended not on the quality of individual providers but on the quality of relationships connecting them — specifically, the frequency, timeliness, accuracy, and problem-solving orientation of their communication, undergirded by shared goals, shared knowledge, and mutual respect. Relational coordination is a property of the connections between roles, not of the individuals filling them. It can be measured, and it predicts outcomes: surgical teams with higher relational coordination scores had shorter lengths of stay, lower readmission rates, and better functional outcomes (Gittell et al. 2000). The mechanism is coordination cost reduction — when roles are connected by high-quality relationships and structured protocols, less information is lost, fewer tasks fall through gaps, and problems are detected earlier.

Wagner’s Chronic Care Model (1998) operationalizes this principle for primary care. The CCM identifies productive interactions between informed, activated patients and prepared, proactive practice teams as the mechanism that improves chronic disease outcomes. “Prepared, proactive practice team” is not a slogan — it is a design specification. It means a team with defined roles for each component of chronic disease management, decision support embedded in workflows, clinical information systems that surface actionable data, and self-management support protocols that extend the team’s reach between visits. The CCM’s evidence base across diabetes, heart failure, depression, and asthma consistently shows that outcomes improve when these team design elements are present and degrade when the “team” is actually a physician working alone with fragmented support.

The Teamlet Model

Bodenheimer and colleagues at the UCSF Center for Excellence in Primary Care developed the teamlet model as the minimum effective coordination unit for primary care. The teamlet consists of three roles: the clinician (physician or NP), a medical assistant or nurse who functions as the health coach, and a care coordinator who manages the panel between visits.

This is not an arbitrary configuration. Each role addresses a distinct coordination requirement:

The clinician provides clinical judgment — the diagnostic reasoning, treatment decisions, and clinical authority that licensure reserves for this role. In a well-designed teamlet, the clinician is freed from tasks that do not require clinical judgment (see Workforce M3, the delegation ladder) and focuses on the 35-40% of care delivery work that genuinely demands their training.

The health coach (MA or nurse) handles pre-visit preparation, in-visit documentation, medication reconciliation, patient education, and chronic disease self-management support. Bodenheimer and Laing (2007) demonstrated that trained MAs functioning as health coaches could deliver diabetes self-management support with outcomes comparable to nurse-led programs, at roughly one-third the labor cost. The role works because the tasks are protocol-driven: the MA follows standing orders, uses standardized assessment tools (PHQ-9, HbA1c tracking), and escalates to the clinician when findings exceed defined thresholds.

The care coordinator manages the panel population — tracking care gaps, managing referrals, following up on hospital discharges, addressing social determinant barriers. This role operates between visits, ensuring that patients who do not present are not lost. The care coordinator’s effectiveness depends entirely on role design: clear panel assignment, defined authority to act (schedule appointments, authorize transportation, contact specialists), and access to a shared tracking system (the registry).

Role clarity within the teamlet determines throughput and quality. When the MA knows exactly which pre-visit tasks are their responsibility, the clinician walks into a room with vitals, medication list, and care gaps already surfaced. When the care coordinator has standing authority to schedule follow-up appointments, post-discharge patients do not languish waiting for a physician order to book. When escalation criteria are defined — HbA1c above 9.0 triggers a physician review, PHQ-9 above 15 triggers a behavioral health referral — the team processes routine cases efficiently and concentrates clinical attention where it matters. Without this clarity, every case requires ad hoc negotiation about who does what, consuming the coordination bandwidth that should be spent on patient care.

Huddle Mechanics

The daily huddle is the team’s coordination mechanism — the synchronization point where shared mental models are built and maintained (connecting directly to the team cognition principles in HF M7). A well-designed huddle converts distributed information into shared awareness. A poorly designed huddle is a performative attendance check that wastes fifteen minutes every morning.

What makes huddles effective:

Structured agenda. The huddle follows a fixed protocol, not an open discussion. A primary care teamlet huddle reviews the day’s schedule patient by patient, identifying: care gaps to address during today’s visit, pre-visit tasks not yet completed, patients requiring extra time or specific resources, and follow-up items from previous visits. The structure ensures that every patient encounter has been pre-coordinated before the clinic day begins.

Role-specific preparation. Each team member arrives having done their part. The care coordinator has pulled the registry and flagged patients overdue for labs or screenings. The MA has reviewed the schedule for medication refill needs and equipment requirements. The clinician has reviewed complex cases flagged by the team. If huddle preparation is not assigned to specific roles, the huddle devolves into real-time chart review — slow, incomplete, and reactive.

Actionable items with owners. Every huddle item produces an action assigned to a person with a timeframe. “Mrs. Rodriguez needs her HbA1c drawn” becomes “MA will order the lab and hand the requisition to the patient during rooming.” Items without owners are items that will not happen.

Time-boxed. Effective huddles run 10-15 minutes for a teamlet managing a half-day clinic session. If the huddle regularly exceeds 20 minutes, the structure is insufficient — the team is problem-solving in the huddle rather than using it for coordination and deferring problem-solving to the appropriate setting (case conference, one-on-one consult, escalation).

What makes huddles performative: No structured agenda — the huddle is an open-ended check-in. No preparation — team members learn the day’s challenges in real time. No follow-through mechanism — items are discussed but not assigned, tracked, or verified. Attendance without engagement — the huddle is mandatory but no one has changed their behavior based on it in months. Leadership absence — the clinician skips huddles when running late, signaling that coordination is optional.

Handoff Reliability Within Teams

Intra-team handoffs — shift changes, task delegation within the care day, coverage transitions — require the same structured communication discipline as inter-team handoffs. The cognitive mechanisms are identical: working memory limits constrain information transfer, salience bias filters what gets communicated, and receiver context shapes what gets retained (HF M5 on handoff degradation applies fully here).

SBAR (Situation, Background, Assessment, Recommendation) provides the communication structure. Closed-loop communication provides the verification: the sender transmits, the receiver confirms understanding, the sender verifies the confirmation. These are not bureaucratic rituals — they are error prevention mechanisms validated by the TeamSTEPPS evidence base (AHRQ) and the CRM tradition from aviation. The reason they matter within teams, not just between them, is that familiarity breeds assumption. Team members who work together daily develop shortcuts — a nod instead of a verbal confirmation, a sticky note instead of a documented handoff, an assumption that “she always checks the labs” instead of an explicit request. These shortcuts work until they do not, and when they fail, the failure is invisible until the error reaches the patient.

The design requirement is protocol consistency regardless of relationship. The MA who hands off a task to the covering MA at lunch uses the same structured format as the nurse handing off to the night shift. The clinician who delegates a follow-up call to the care coordinator specifies what information to convey, what responses require escalation, and how to document the outcome — every time, not just when the patient is complex.

Team Stability

Hackman’s research on team effectiveness (2002) established that stable team composition is a prerequisite for high performance. Teams need time to develop the shared mental models and transactive memory (Wegner 1987, explored in HF M7) that enable implicit coordination. Frequent membership changes destroy this cognitive infrastructure and force the team to rebuild it with each new configuration.

The mechanism is transactive memory decay. In a stable teamlet, the MA knows that this physician prefers to see the medication list before the chief complaint. The care coordinator knows which community resources the social worker trusts and which she considers unreliable. The physician knows that this MA will catch a blood pressure discrepancy without being asked. None of this is documented. It lives in the team’s transactive memory — the distributed system of who knows what, who can do what, and who will notice what. When a team member is replaced, this entire network must be rebuilt for every connection that person held.

Healthcare teams face a structural challenge here. Shift-based staffing, float pool assignments, locum providers, and high turnover in MA and care coordinator roles mean that many teams never achieve the stability Hackman identified as necessary. The result is teams that are perpetually in the forming stage — individually competent members who have not developed the coordination infrastructure to function as a team. They compensate through explicit communication, which consumes more time and cognitive bandwidth than the implicit coordination that stable teams achieve. The coordination tax is real: Lewis (2003) found that teams with underdeveloped transactive memory performed worse than teams with less individually knowledgeable members who had worked together long enough to build the directory.

The implication for team design is that roster stability must be treated as a design parameter, not an HR outcome. Assigning the same MA to the same clinician, keeping teamlet composition consistent across scheduling cycles, and minimizing float pool disruption to established teams are not convenience preferences — they are coordination investments. Every roster change carries a hidden cost measured in coordination failures, repeated information requests, and degraded anticipatory behavior that no orientation checklist can replace.

Healthcare Example: Community Health Center Diabetes Panel

A federally qualified health center serving a predominantly Medicaid population implements team-based care for a panel of 2,400 patients with diabetes. The team is designed, not assembled.

Team composition: Physician (0.5 FTE), nurse practitioner (0.5 FTE), RN care manager (1.0 FTE), medical assistant (1.0 FTE), community health worker (0.5 FTE). Together, the physician and NP provide 1.0 FTE of clinical coverage. The RN care manager owns the registry — tracking HbA1c values, overdue screenings, medication adherence, and hospital utilization for all 2,400 patients. The MA handles pre-visit preparation, in-visit support, and post-visit task completion. The community health worker addresses social determinant barriers: transportation, food access, health literacy, and appointment adherence for the highest-risk patients.

The design process:

Panel assignment. Every patient is assigned to a specific clinician (physician or NP) and appears on the team registry. No patient is unattached. The RN care manager can identify, at any moment, every patient on the panel, their last HbA1c, their next scheduled visit, and their care gap status.

Role-specific standing orders. The RN care manager has standing orders to: order HbA1c labs for patients overdue by more than 30 days, schedule foot exams and retinal screenings per protocol, initiate outreach calls for patients who no-show, and refer patients with HbA1c above 9.0 for a clinician visit within two weeks. The MA has standing orders to reconcile medications at every visit, administer point-of-care HbA1c testing, and complete a social determinant screening annually. The community health worker has standing protocols for home visits, transportation arrangement, and connection to food assistance programs. These standing orders mean that routine care management does not require clinician authorization — it happens automatically, reserving clinician time for clinical judgment.

Daily huddle protocol. Each morning, the team reviews the day’s scheduled patients using a structured template: care gaps flagged by the registry, pre-visit tasks assigned (lab orders, screening tools prepared), high-risk patients requiring extra time, and follow-up items from previous days. The huddle runs 12 minutes. The RN care manager facilitates, ensuring that the registry data drives the conversation.

Weekly case conference. One hour, structured around the RN care manager’s registry report: patients with HbA1c above 9.0 who have not improved after two medication adjustments, patients with three or more no-shows, patients recently discharged from the hospital, and patients flagged by the community health worker for social determinant barriers affecting adherence. The clinician provides treatment plan adjustments. The team collectively problem-solves cases where standard protocols are insufficient.

Shared dashboard. A panel-level dashboard visible to all team members displays: percentage of patients with HbA1c below 8.0 (the team’s quality target), percentage with completed annual screenings, ED visit rate per 1,000 panel patients, and 30-day post-discharge follow-up completion rate. The dashboard measures the team, not individuals — reinforcing shared accountability.

12-month results: HbA1c control (below 8.0) improved from 52% to 68% of the panel. ED visits among panel patients decreased 22%. Patient satisfaction scores increased 15 points (on a 100-point scale). The RN care manager’s registry-driven outreach closed 340 care gaps that would not have been identified through visit-based care alone. The community health worker’s interventions reduced no-show rates from 24% to 16% among the highest-risk quartile.

These results are not attributable to any single role. They are the product of team design — role clarity that eliminated redundancy and gaps, standing orders that automated routine decisions, a huddle that synchronized the team daily, a case conference that concentrated clinical attention on the patients who needed it most, and a shared dashboard that made collective performance visible and accountable.

What Breaks Team Design

Role creep. Tasks migrate to whoever is available rather than whoever is designated. The RN care manager starts handling prior authorizations because “they know the patients.” The MA starts scheduling follow-ups because the care coordinator is overwhelmed. Each migration seems efficient in the moment but degrades the role architecture, overloads some roles, and undercuts the coordination logic the team was designed around.

Unclear escalation. When the MA discovers a blood pressure of 195/110 during pre-visit preparation, is this an immediate clinician interruption, a huddle item, or a notation in the chart? If the escalation protocol does not define the threshold and the pathway, the MA either interrupts constantly (destroying clinician workflow) or waits too long (creating clinical risk). Escalation ambiguity is the most common operational failure in teamlet-based care.

Leadership vacuums. Every team needs someone who owns the coordination function — not the clinical decisions, but the operational rhythm. In the teamlet model, this is typically the RN care manager or the lead MA. When this role is vacant, unfilled, or assigned to someone without the authority to hold team members accountable to the protocols, the huddle becomes optional, the registry falls behind, and the team reverts to individual practice.

Individual metrics instead of team metrics. When the physician is measured on their panel’s quality scores but the MA and care coordinator are measured on activity volume (calls made, forms completed), the incentive structure fractures the team. The physician wants care gaps closed; the MA wants tasks checked off. These are not the same thing. Team-based measurement — panel-level outcomes attributed to the team — aligns incentives with the coordination design. Individual measurement creates optimization at the role level that may suboptimize the team.

Warning Signs

Huddles consistently run over 20 minutes or are skipped entirely. Both indicate structural failure — the former suggests insufficient preparation or absent protocols, the latter signals that coordination has been deprioritized.

Team members cannot name the panel size or current quality metrics. If the MA does not know the team’s HbA1c control rate, shared accountability has not been established. The dashboard exists but is not driving behavior.

The same care gap appears visit after visit without closure. This indicates that the registry-to-action loop is broken — care gaps are identified but not assigned, or assigned but not tracked to completion.

High turnover in the care coordinator or MA role with stable clinician tenure. This pattern signals that the non-clinician roles were designed with insufficient authority, unclear scope, or workload that exceeds the position’s capacity. The clinician persists because the role has structural power; the support roles churn because the design is defective.

The clinician routinely performs tasks assigned to other team members. When the physician reconciles medications because “it’s faster than explaining it to the MA,” the role architecture has failed. Speed in the moment creates dependency that degrades team capacity over time.

Product Owner Lens

What is the workforce problem? Care teams assembled without deliberate design produce coordination by accident — inconsistent, personality-dependent, and fragile. Effective team-based care requires designed coordination: defined roles, structured protocols, and shared accountability that does not depend on individual heroics.

What system mechanism explains it? Gittell’s relational coordination: team performance is a property of the connections between roles, not the individuals filling them. Hackman’s team stability research: shared mental models and transactive memory require stable composition to develop. Wagner’s Chronic Care Model: chronic disease outcomes depend on prepared, proactive teams with defined roles and decision support.

What intervention levers exist? Panel assignment to ensure every patient has a team. Role-specific standing orders to automate routine coordination. Huddle protocols with structured agendas and preparation requirements. Team-level metrics that align incentives with collective outcomes. Roster stability policies that protect team composition from unnecessary disruption.

What should software surface? Panel-level dashboards showing team quality metrics (HbA1c control, screening completion, ED utilization, follow-up rates) attributed to the team, not individual providers. Registry views that flag care gaps by urgency and assign them to specific roles based on standing order logic. Huddle preparation screens that auto-populate the day’s schedule with registry-derived flags. Team roster stability tracking — how long the current team configuration has been intact, and whether recent changes correlate with quality metric movement.

What metric reveals degradation earliest? Care gap closure rate — the percentage of identified care gaps that are resolved within the protocol-defined timeframe. When this rate drops, the coordination loop between registry identification, role assignment, and task completion is breaking down. It degrades before HbA1c control rates or ED utilization shift, providing a 60-90 day leading indicator of team coordination failure.

Integration Points

HF Module 7 (Team Cognition). Care team design is the organizational implementation of the cognitive principles HF M7 describes. Shared mental models — the overlapping representations of task, team, equipment, and situation that Cannon-Bowers and Salas (1993) identified as the foundation of team coordination — do not emerge from organizational charts. They emerge from daily huddles that synchronize the team’s understanding of today’s work, from stable roster assignments that allow members to learn each other’s capabilities, and from structured communication protocols that make implicit knowledge explicit. Transactive memory — the team’s distributed directory of who knows what (Wegner 1987) — is what a well-designed teamlet builds over months of working together: the MA who knows which patients will need extra time, the care coordinator who knows which community resources actually deliver, the clinician who knows which specialist returns calls promptly. When team design ignores stability and communication structure, it prevents the cognitive infrastructure that HF M7 identifies as the actual mechanism of team performance from ever developing.

Workforce Module 3 (Role Design). Role design (M3) provides the building blocks; team design assembles them into a functioning coordination system. M3’s delegation ladder determines which tasks belong at which credential level. Team design determines how those tasks flow between roles during a care episode — who prepares what before the visit, who documents what during it, who follows up afterward, and what triggers escalation between roles. A team composed of well-designed individual roles can still fail if the interfaces between those roles are undefined. The teamlet model succeeds not because each role is well-designed in isolation (though that matters — see M3’s analysis of care coordinator design), but because the connections between roles are specified: the MA’s pre-visit preparation feeds the clinician’s decision-making, the clinician’s orders feed the care coordinator’s follow-up, the care coordinator’s registry data feeds tomorrow’s huddle. Role design without team design produces capable individuals working in parallel. Team design without role design produces coordinated confusion. Both are required.

Summary

A care team is a coordination system, not a staffing assignment. Its effectiveness depends on deliberate design: roles with clear scope and matched authority, communication protocols that synchronize the team daily, shared accountability measured at the team level, and organizational support for the roster stability that allows coordination capital to accumulate.

The teamlet model — clinician, health coach, care coordinator — provides the minimum effective unit for primary care, with each role addressing a distinct coordination requirement. Huddles provide the daily synchronization mechanism, but only when they are structured, prepared for, and action-generating. Intra-team handoffs require the same structured communication as inter-team transitions, because familiarity breeds assumption and assumption breeds error.

What breaks team design is predictable: role creep that degrades the architecture, unclear escalation that creates risk or inefficiency, leadership vacuums that let coordination protocols atrophy, and individual metrics that fracture shared accountability. The community health center example demonstrates that when these elements are designed correctly — panel assignment, standing orders, daily huddles, weekly case conferences, shared dashboards — a modestly resourced team can produce population-level quality improvements that individual-practice models cannot match.