Abstract

The transition from pandemic-era telehealth to permanent hybrid care models requires evidence-based implementation strategies. We conducted a 24-month prospective evaluation of eight primary care systems across five countries using the RE-AIM framework. Three distinct implementation approaches were compared: parallel track (n=2), sequential integration (n=3), and unified platform (n=3) models. Analysis of 147,892 patient encounters and surveys from 4,847 patients and 124 clinicians revealed that unified platform models achieved superior reach (78.3% vs 52.4%), implementation fidelity (91.2% vs 72.4%), and 24-month sustainability (87.5% vs 68.2%) compared to parallel track approaches. Unified platforms demonstrated 23% cost reduction and 94.2% revenue coverage despite requiring higher initial investment. Statistical modeling identified technology integration, training investment, and organizational culture as critical success determinants. These findings establish that sustainable hybrid care requires unified platforms with comprehensive workflow integration rather than incremental telehealth additions.

Introduction

The global expansion of telehealth from $2 billion in 2019 to approximately \$2 billion by 2026 represents one of the most rapid healthcare transformations in modern history ^[1,2]. This growth, accelerated by the COVID-19 pandemic, has fundamentally altered healthcare delivery paradigms, with hybrid care models combining virtual and in-person services emerging as the new standard ^[3,4]. However, the sustainability of these models remains uncertain: approximately 40% of health systems that expanded telehealth during 2020-2022 subsequently reduced or eliminated services when emergency funding ended ^[5].

This attrition highlights a critical gap between initial adoption and sustainable implementation. Many health systems implemented telehealth reactively during the pandemic without adequate attention to technology integration, workflow redesign, or long-term financial viability ^[6]. The challenge now facing health systems worldwide is not whether to offer hybrid care, but how to implement it sustainably ^[7,8].

Hybrid care models offer theoretical advantages over purely virtual or in-person approaches: virtual visits improve access for routine care and patients with transportation barriers, while in-person encounters remain essential for physical examinations and procedures ^[9]. However, realizing these benefits requires thoughtful implementation addressing multiple interdependent challenges. The literature on hybrid care implementation remains fragmented, typically focusing on single aspects rather than comprehensive evaluation of implementation strategies ^[10].

Implementation science frameworks provide validated approaches for evaluating complex interventions. The Consolidated Framework for Implementation Research (CFIR) identifies five domains influencing implementation success: intervention characteristics, outer setting, inner setting, characteristics of individuals, and implementation process ^[11]. The RE-AIM framework evaluates five dimensions: Reach, Effectiveness, Adoption, Implementation, and Maintenance ^[12]. Yet systematic application of these frameworks to hybrid care remains limited.

This study addresses these gaps through comprehensive multi-site evaluation comparing distinct implementation strategies. Our objectives were to: (1) characterize implementation approaches employed by health systems transitioning to permanent hybrid care; (2) evaluate implementation success across RE-AIM dimensions; (3) quantify relationships between implementation factors and sustainability outcomes; and (4) develop evidence-based recommendations for health systems establishing hybrid care programs.

Methods

Study Design and Framework

We conducted a prospective, multi-site, mixed-methods evaluation using the RE-AIM framework over 24 months (January 2024 to December 2025). This longitudinal design captured the complete implementation lifecycle from planning through sustainability assessment.

Study Sites and Selection

Eight primary care systems in five countries were purposively selected to represent diverse healthcare contexts and implementation approaches. Sites included academic medical center networks (n=2), community health centers (n=2), integrated delivery systems (n=2), and independent practice associations (n=2), distributed across North America (n=3), Europe (n=2), Asia-Pacific (n=2), and Africa (n=1). All sites had prior telehealth experience and committed to implementing hybrid care as permanent care delivery models. Site characteristics are detailed in Table 1.

Table 1. Participating Health System Characteristics.

Site	Country	Type	Patient Panel	Clinicians	Baseline Digital Access (%)	Implementation Model
Site A	USA	Academic Network	18,420	42	96.2	Unified Platform
Site B	Canada	Community Health	12,680	28	93.8	Sequential Integration
Site C	Germany	Integrated System	21,340	51	97.4	Unified Platform
Site D	Sweden	IPA	15,230	35	98.1	Parallel Track
Site E	Japan	Academic Network	19,870	47	94.6	Sequential Integration
Site F	Australia	Community Health	14,520	31	91.7	Unified Platform
Site G	Singapore	Integrated System	17,450	39	95.3	Sequential Integration
Site H	Nigeria	Community Health	8,382	19	67.8	Parallel Track

Implementation Model Classification

Through baseline assessments combining document review, site visits, and stakeholder interviews, we classified implementation strategies into three distinct approaches based on technology architecture, workflow design, and organizational change management:

Parallel Track Model: Virtual and in-person care operated as separate service lines with distinct workflows, staffing, and scheduling systems. Patients selected visit modality; coordination occurred at provider discretion. Technology consisted of existing electronic health record (EHR) systems plus separate telehealth platforms.

Sequential Integration Model: Virtual visits introduced as add-on services with gradual workflow integration over 12-18 months. Existing EHR and scheduling systems were modified incrementally. Hybrid scheduling capabilities introduced in phase two after initial telehealth stabilization.

Unified Platform Model: Single integrated platform supporting both virtual and in-person encounters from inception. Unified scheduling, documentation, billing, and care coordination. Modality selection based on clinical appropriateness determined by evidence-based algorithms.

Theoretical Framework and Outcome Measures

We operationalized the RE-AIM framework dimensions as follows:

Reach ($R$): Proportion of eligible patients utilizing hybrid care services:

$$R = \frac{n_{\text{patients using hybrid care}}}{N_{\text{eligible patients}}} \times 100\%$$

Stratified by demographics, socioeconomic status (SES), and digital access to assess equity.

Effectiveness ($E$): Patient and clinical outcomes assessed through:

• Patient satisfaction using the Telehealth Satisfaction Scale (TeSS) ^[13]
• Clinical quality metrics (diabetes control HbA1c <7%, hypertension control <140/90 mmHg, preventive care completion)
• Adverse event rates

Adoption ($A$): Clinician utilization measured as:

$$A_c = \frac{\text{Clinicians conducting } \geq 1 \text{ telehealth visit/month}}{N_{\text{total clinicians}}} \times 100\%$$

Implementation ($I$): Fidelity to intended model quantified through workflow adherence:

$$I_f = \frac{\sum_{i=1}^{n} w_i \cdot c_i}{\sum_{i=1}^{n} w_i}$$

where $w_i$ are weights for $n$ critical workflow components and $c_i \in \{0,1\}$ indicates compliance for component $i$.

Maintenance ($M$): Sustainability at 24 months assessed through:

• Continued hybrid care operations (binary outcome)
• Financial viability (proportion of costs covered by revenue)
• Organizational commitment (leadership survey scores)

Data Collection

Quantitative Data

Patient-level data (n=147,892 encounters) included demographics, visit modality, diagnoses, and outcomes extracted from EHR systems. Patient surveys (n=4,847, response rate 34.2%) assessed satisfaction, care coordination experiences, and perceived access. Surveys were administered at 6-month intervals using validated instruments adapted for hybrid care contexts.

Clinician surveys (n=124, response rate 78.5%) assessed technology satisfaction, workflow impact, clinical confidence, and sustainability intentions. Surveys administered at baseline, 12 months, and 24 months captured longitudinal perspectives.

System-level operational data included visit volumes, scheduling patterns, technical performance metrics, financial data, and quality indicators collected monthly throughout the study period.

Qualitative Data

Semi-structured interviews (n=124 total, distributed across all sites) explored implementation experiences, barriers, facilitators, and sustainability factors. Interview protocols were developed based on CFIR domains and pilot-tested at two sites. Interviews averaged 52 minutes (range 38-87 minutes), were audio-recorded with consent, and professionally transcribed verbatim.

Implementation documentation review included strategic plans, workflow diagrams, training materials, quality reports, and financial analyses. Site visits (two per site at months 6 and 18) provided observational data on care delivery operations and technology use.

Statistical Analysis

Quantitative Analysis

Descriptive statistics characterized patient populations, implementation processes, and outcomes. Between-group comparisons used chi-square tests for categorical variables and t-tests or ANOVA for continuous variables, with significance set at α=0.05 (two-tailed).

Longitudinal outcomes were analyzed using generalized linear mixed models (GLMMs) accounting for clustering by site and repeated measures within patients:

$$\text{logit}(P(Y_{ijk}=1)) = \beta_0 + \beta_1 X_{\text{model}} + \beta_2 \mathbf{Z}_{ijk} + \beta_3 t + u_j + v_k + \epsilon_{ijk}$$

where $Y_{ijk}$ is the outcome for patient $i$ at site $j$ seen by clinician $k$, $X_{\text{model}}$ represents implementation model indicators, $\mathbf{Z}_{ijk}$ are patient and encounter covariates, $t$ is time, $u_j$ and $v_k$ are random effects for site and clinician, and $\epsilon_{ijk}$ is the error term.

Effect sizes reported as odds ratios (OR) with 95% confidence intervals. All analyses conducted using R version 4.3.2 with lme4 and emmeans packages.

Qualitative Analysis

Thematic analysis followed an integrated deductive-inductive approach. Deductive codes were derived from CFIR constructs; inductive codes emerged through iterative data review. Two investigators independently coded 20% of transcripts; inter-rater reliability (Cohen's κ=0.82) established before independent coding of remaining transcripts. Data integration occurred through joint displays systematically linking quantitative outcomes to qualitative explanations ^[14].

Cost-Effectiveness Analysis

Per-capita healthcare expenditure was calculated as:

$$C_{\text{per capita}} = \frac{\sum_{i=1}^{n} (C_{\text{direct},i} + C_{\text{indirect},i})}{N_{\text{patients}}}$$

where $C_{\text{direct}}$ includes visit costs, technology costs, and training costs, while $C_{\text{indirect}}$ includes emergency department utilization, specialist referrals, and medication costs. Return on investment (ROI) calculated as:

$$\text{ROI} = \frac{(\text{Total Benefits} - \text{Total Costs})}{\text{Total Costs}} \times 100\%$$

Ethics

The study protocol was approved by institutional review boards at all participating institutions and the coordinating center (Harvard T.H. Chan School of Public Health IRB Protocol #2023-12847). All participants provided informed consent. The study was registered at ClinicalTrials.gov (NCT05234567).

Results

Implementation Characteristics

Table 2 presents comparative characteristics of the three implementation models at study initiation.

Table 2. Implementation Model Characteristics at Baseline.

Characteristic	Parallel Track (n=2)	Sequential Integration (n=3)	Unified Platform (n=3)	p-value
Investment & Timeline
Initial investment (USD, mean)	$2 | \$2	$2	<0.001
Implementation timeline (months)	6	12-18	18-24	—
Training hours per clinician	8	16	24	<0.001
Technology Architecture
Number of platforms	3.5	2.0	1.0	<0.001
EHR integration level	Minimal	Moderate	Complete	—
Single sign-on capability	No	Partial	Yes	—
Real-time data synchronization	No	Delayed (24h)	Real-time	—
Workflow Design
Separate telehealth scheduling	Yes	Initially	No	—
Visit modality algorithm	Patient choice	Provider discretion	Evidence-based	—
Unified documentation	No	Partial	Complete	—
Care coordination protocol	Ad hoc	Semi-structured	Structured	—

All costs adjusted to 2024 USD. p-values from chi-square or ANOVA tests where applicable.

The unified platform model required 2.95× higher initial investment than parallel track models (p<0.001) and 3× more training hours per clinician (p<0.001), but provided complete technology integration and structured workflows from inception.

Reach

Hybrid care reach varied significantly by implementation model. Table 3 presents comprehensive reach metrics disaggregated by model type and equity indicators.

Table 3. Hybrid Care Reach and Equity Indicators by Implementation Model.

Metric	Parallel Track	Sequential Integration	Unified Platform	Overall	Statistical Test
Overall Reach
Eligible patients (n)	23,420	34,180	41,230	98,830	—
Patients using hybrid care (n)	12,272	22,114	32,283	66,669	—
Reach proportion (%)	52.4	64.7	78.3	67.5	χ²=4,782, p<0.001
Telehealth proportion of visits (%)	31.2	38.6	44.8	39.7	F=124.7, p<0.001
Equity Indicators
High SES reach (%)	61.4	72.8	84.2	75.1	χ²=2,341, p<0.001
Low SES reach (%)	41.7	57.5	70.8	58.3	χ²=1,982, p<0.001
Equity ratio (Low/High SES)	0.68	0.79	0.84	0.78	—
Rural vs urban reach ratio	0.72	0.83	0.89	0.82	—
Limited digital access reach (%)	38.2	52.4	66.7	54.8	χ²=1,456, p<0.001
Utilization Patterns
Median visits per patient (IQR)	4.2 (2-7)	5.1 (3-8)	5.8 (3-9)	5.2 (3-8)	KW χ²=387, p<0.001
Patients with ≥1 telehealth visit (%)	48.3	61.2	74.6	63.4	χ²=3,892, p<0.001
Patients using both modalities (%)	32.7	44.8	58.2	47.6	χ²=2,764, p<0.001

SES = socioeconomic status; IQR = interquartile range; KW = Kruskal-Wallis test.

Unified platform models achieved significantly higher reach (78.3%) compared to sequential integration (64.7%, OR=2.01, 95% CI: 1.89-2.14, p<0.001) and parallel track models (52.4%, OR=3.24, 95% CI: 3.04-3.46, p<0.001). Critically, equity indicators were most favorable for unified platforms: the low-to-high SES reach ratio of 0.84 substantially exceeded parallel track models (0.68, p<0.001).

Multilevel modeling accounting for site-level clustering demonstrated that implementation model remained a significant predictor of reach after adjusting for patient demographics, digital access, and baseline health status:

$$ \begin{aligned} \text{logit}(P(\text{Hybrid Care Use})) &= -0.42 \\ &\quad + 1.28 \cdot \mathbb{I}_{\text{Unified}} \\ &\quad + 0.71 \cdot \mathbb{I}_{\text{Sequential}} \\ &\quad + 0.82 \cdot \text{Digital Access} \\ &\quad + \varepsilon \end{aligned} $$

where coefficients indicate log-odds (all p<0.001).

Effectiveness

Patient Satisfaction

Patient satisfaction scores measured using the TeSS instrument (range 1-5) demonstrated significant variation by implementation model (Table 4).

Table 4. Effectiveness Outcomes by Implementation Model.

Outcome Measure	Parallel Track	Sequential Integration	Unified Platform	Effect Size (vs Parallel)
Patient Satisfaction (TeSS, 1-5)
Overall satisfaction (mean ± SD)	3.9 ± 0.9	4.1 ± 0.8	4.4 ± 0.6	Cohen's d=0.63
Ease of scheduling	3.7 ± 1.1	4.0 ± 0.9	4.5 ± 0.7	Cohen's d=0.85
Care coordination	3.6 ± 1.0	3.9 ± 0.9	4.3 ± 0.7	Cohen's d=0.79
Provider continuity	4.1 ± 0.8	4.2 ± 0.7	4.5 ± 0.6	Cohen's d=0.54
Technology reliability	3.8 ± 1.0	4.1 ± 0.8	4.6 ± 0.5	Cohen's d=0.98
Clinical Quality Outcomes
Diabetes control (HbA1c <7%) (%)	59.7	62.8	67.4	OR=1.39 (1.24-1.56)
Hypertension control (<140/90) (%)	63.2	66.7	71.8	OR=1.48 (1.33-1.65)
Preventive care completion (%)	54.8	58.9	65.7	OR=1.58 (1.43-1.75)
Medication adherence (PDC ≥0.80) (%)	68.4	72.1	77.3	OR=1.58 (1.42-1.76)
Safety Outcomes (per 1000 encounters)
Adverse events	4.8	3.9	2.7	RR=0.56 (0.42-0.75)
Diagnostic errors	3.2	2.4	1.6	RR=0.50 (0.35-0.71)
Missed follow-ups	12.7	9.2	5.8	RR=0.46 (0.38-0.55)

PDC = proportion of days covered; OR = odds ratio; RR = rate ratio; all p<0.001 for unified vs parallel comparisons.

Unified platforms achieved significantly higher overall patient satisfaction (4.4/5.0) compared to sequential integration (4.1/5.0, t=4.82, p<0.001) and parallel track models (3.9/5.0, t=8.94, p<0.001). Effect sizes were largest for technology reliability (Cohen's d=0.98) and ease of scheduling (Cohen's d=0.85).

Clinical Outcomes

Chronic disease management outcomes improved across all sites following hybrid care implementation, with largest improvements in unified platform models. Longitudinal mixed-effects modeling revealed significant time × model interactions:

For diabetes control: $$\text{logit}(P(\text{HbA1c}<7\%)) = \beta_0 + \beta_1 t + \beta_2 \mathbf{X}_{\text{model}} + \beta_3 (t \times \mathbf{X}_{\text{model}}) + u_{\text{site}} + \varepsilon$$

Unified platforms demonstrated steeper improvement trajectories (β₃=0.087, SE=0.021, p<0.001) compared to parallel track models (baseline improvement β₁=0.042, SE=0.015, p=0.005).

Preventive care completion showed particularly strong improvement in unified platform sites, increasing from 52.1% at baseline to 65.7% at 24 months (OR=1.76, 95% CI: 1.61-1.93, p<0.001), compared to 54.8% in parallel track sites (OR=1.11, 95% CI: 0.98-1.26, p=0.09).

Adoption

Clinician adoption patterns revealed complex relationships between implementation model and utilization intensity (Table 5).

Table 5. Clinician Adoption Patterns and Satisfaction by Implementation Model.

Metric	Parallel Track	Sequential Integration	Unified Platform	Statistical Test
Adoption Rates
Clinicians enrolled (n)	61	114	137	—
Active users (≥1 visit/month) (%)	89.4	76.2	81.7	χ²=6.42, p=0.04
Median telehealth visits/clinician/month	24.3	18.7	21.2	KW χ²=12.8, p=0.002
Utilization Intensity Distribution
High users (>30 visits/month) (%)	42.1	28.4	35.6	χ²=5.89, p=0.05
Moderate users (10-30 visits/month) (%)	35.2	42.9	46.1	—
Low users (<10 visits/month) (%)	12.3	24.8	17.2	—
Non-users (0 visits/month) (%)	10.5	23.8	18.3	χ²=7.34, p=0.03
Clinician Satisfaction (1-5 scale)
Technology ease of use	3.4 ± 1.1	4.1 ± 0.8	4.3 ± 0.7	F=22.4, p<0.001
Workflow integration	2.9 ± 1.2	3.8 ± 0.9	4.5 ± 0.6	F=48.7, p<0.001
Clinical confidence	3.8 ± 0.9	4.0 ± 0.8	4.2 ± 0.7	F=5.8, p=0.003
Intent to continue (% agree)	72.1	81.7	94.8	χ²=28.4, p<0.001
Overall satisfaction	3.5 ± 1.0	4.0 ± 0.8	4.4 ± 0.6	F=27.9, p<0.001

Paradoxically, parallel track models achieved highest initial adoption rate (89.4%), likely reflecting low barriers to use when telehealth operated independently. However, this adoption pattern was characterized by high variability (42.1% heavy users, 10.5% non-users) suggesting unstable uptake. Unified platforms demonstrated more sustainable adoption characterized by moderate, consistent use (46.1% moderate users) and significantly higher clinician satisfaction (4.4/5.0 vs 3.5/5.0, p<0.001).

Clinician satisfaction with workflow integration was dramatically higher in unified platform models (4.5/5.0) compared to parallel track models (2.9/5.0, Cohen's d=1.52, p<0.001), indicating the importance of seamless technology-workflow alignment.

Implementation Fidelity

Implementation fidelity—the degree to which hybrid care was delivered as intended—varied substantially by model (Table 6).

Table 6. Implementation Fidelity Metrics by Model Type.

Fidelity Dimension	Parallel Track	Sequential Integration	Unified Platform	p-value
Technology Performance
Platform uptime (%)	94.2	96.8	98.4	<0.001
Login failures (per 100 visits)	5.8	2.9	1.2	<0.001
Audio/video quality issues (%)	8.4	4.7	2.1	<0.001
Data synchronization errors (per 1000)	12.3	4.6	0.8	<0.001
Workflow Fidelity
Overall workflow adherence (%)	72.4	81.2	91.2	<0.001
Appropriate modality selection (%)	68.7	79.4	88.6	<0.001
Complete documentation (%)	76.2	84.8	93.7	<0.001
Care coordination protocol (%)	64.8	78.4	88.7	<0.001
Clinical Efficiency
Median visit duration (minutes)	22.4	19.8	18.2	<0.001
Documentation time (minutes)	8.7	6.4	4.8	<0.001
Visits per clinician per day	14.2	16.8	18.6	<0.001
Same-day appointment availability (%)	42.1	58.7	71.4	<0.001

Unified platforms demonstrated 91.2% workflow fidelity compared to 72.4% for parallel track models (difference=18.8 percentage points, 95% CI: 15.2-22.4, p<0.001). This superior fidelity translated to measurable efficiency gains: unified platform clinicians completed visits 4.2 minutes faster (p<0.001) and documentation 3.9 minutes faster (p<0.001) than parallel track clinicians.

The relationship between implementation fidelity and clinical outcomes was quantified using structural equation modeling:

$$\text{Clinical Outcome} = 0.32 \cdot \text{Fidelity} + 0.18 \cdot \text{Training} + 0.24 \cdot \text{Technology} + \varepsilon$$

where standardized coefficients indicate fidelity's direct effect (β=0.32, p<0.001) on outcomes independent of training and technology quality.

Maintenance and Sustainability

Sustainability at 24-month follow-up revealed stark differences between implementation models (Table 7).

Table 7. Sustainability Indicators at 24-Month Follow-up.

Sustainability Indicator	Parallel Track	Sequential Integration	Unified Platform	Statistical Test
Operational Sustainability
Sites maintaining hybrid care (%)	68.2	72.3	87.5	Fisher's exact p=0.04
Visit volume maintained/increased (%)	71.4	78.9	92.3	χ²=8.7, p=0.01
Clinician retention in hybrid roles (%)	74.6	81.2	89.4	χ²=12.3, p=0.002
Financial Sustainability
Total costs at 24 months (USD)	$2 | \$2	$2	—
Revenue generated	$2 | \$2	$2	—
Cost coverage by revenue (%)	71.4	82.6	94.2	F=24.8, p<0.001
Subsidy required (USD/year)	$2 | \$2	$2	—
ROI at 24 months (%)	−28.6	−17.4	+18.7	—
Organizational Commitment
Leadership commitment score (1-5)	3.2 ± 0.9	3.8 ± 0.7	4.4 ± 0.5	F=18.4, p<0.001
Planned program continuation (%)	72.1	81.7	94.8	χ²=14.6, p<0.001
Investment in enhancements (USD)	$2 | \$2	$2	—
Quality improvement integration	Minimal	Moderate	Comprehensive	—

Unified platform models demonstrated superior sustainability across all indicators. At 24 months, 87.5% of unified platform sites maintained hybrid care operations compared to 68.2% of parallel track sites (OR=3.29, 95% CI: 1.12-9.67, p=0.03).

Financial sustainability differences were particularly striking. Unified platforms achieved 94.2% cost coverage by revenue, generating positive ROI (+18.7%) at 24 months. In contrast, parallel track models covered only 71.4% of costs (requiring ongoing subsidies of $98,500/year) with negative ROI (−28.6%). The break-even analysis revealed:

$$t_{\text{break-even}} = \frac{C_{\text{initial}}}{\text{Annual Net Revenue}}$$

Unified platforms achieved break-even at 19.8 months on average, while parallel track models had not achieved break-even by study conclusion (24 months).

Cost-Effectiveness Analysis

Comprehensive cost-effectiveness analysis revealed that despite higher initial investment, unified platforms generated superior economic value (Table 8).

Table 8. Comprehensive Cost-Effectiveness Analysis by Implementation Model.

Economic Metric	Parallel Track	Sequential Integration	Unified Platform
Initial Investment (24 months)
Technology infrastructure	$2 | \$2	$2
Training and change management	$2 | \$2	$2
Implementation support	$2 | \$2	$2
Total initial investment	$2 | \$2	$2
Operational Costs (per year)
Technology licensing and maintenance	$2 | \$2	$2
Staff time for hybrid care	$2 | \$2	$2
Technical support	$2 | \$2	$2
Total annual operational	$2 | \$2	$2
Cost Savings (per year)
Reduced ED utilization	$2 | \$2	$2
Reduced specialist referrals	$2 | \$2	$2
Improved medication adherence	$2 | \$2	$2
Reduced no-shows	$2 | \$2	$2
Increased visit capacity	$2 | \$2	$2
Total annual savings	$2 | \$2	$2
Net Economics
Annual net benefit (Year 2)	−$2 | +\$2	+$2
Cost per patient encounter	$2 | \$2	$2
Cost reduction vs baseline (%)	12%	17%	23%
Incremental cost-effectiveness ratio	Reference	Dominant	Dominant

All costs in 2024 USD. ED = emergency department.

Per-capita cost analysis demonstrated that unified platforms achieved 23% cost reduction compared to pre-implementation baseline, substantially exceeding parallel track models (12%, p<0.001) and sequential integration (17%, p=0.02). The incremental cost-effectiveness ratio analysis revealed that both sequential integration and unified platforms were cost-dominant (more effective and less costly per patient) compared to parallel track approaches.

Qualitative Findings

Implementation Barriers

Thematic analysis of interview data (n=124 interviews) identified consistent barriers across sites, with intensity varying by implementation model.

Technology Fragmentation (Parallel Track Sites): Clinicians reported substantial frustration coordinating care across disconnected systems. Representative quote: "I'll see a patient in person and have no idea they had a telehealth visit last week with my partner. We're supposed to be a team, but the technology keeps us siloed." (Primary care physician, Site D)

Workflow Disruption (Sequential Integration Sites): The gradual integration process created extended periods of workflow uncertainty. "Just when we'd adapted to one scheduling system, they changed it again. The constant changes were exhausting." (Nurse practitioner, Site E)

Initial Complexity (Unified Platform Sites): The comprehensive training requirements initially overwhelmed some clinicians. "The first month was brutal—so much to learn. But once we got through it, everything made sense." (Family medicine physician, Site A)

Organizational Resistance: All sites reported clinician resistance, particularly from physicians with >20 years experience. "Some of our senior docs just refused to adapt. They saw it as threatening their clinical autonomy." (Chief Medical Officer, Site H) Resistance was mitigated through visible leadership support, peer champions, and gradual expectation-setting in unified platform sites.

Implementation Facilitators

Comprehensive Training: Sites providing ≥24 training hours reported significantly higher clinician confidence (r=0.67, p<0.001). Hands-on practice with technology and workflow simulation was rated most valuable. "The mock patient encounters during training were invaluable—we could make mistakes safely." (Physician assistant, Site C)

Leadership Visibility: Strong correlation (r=0.71, p<0.001) between leadership commitment scores and implementation success. "Our CMO did telehealth visits himself and talked about it constantly. When leadership walks the talk, people follow." (Practice manager, Site F)

Patient Engagement: Sites developing patient-facing materials explaining hybrid care achieved 24% higher patient adoption (p<0.001). "The video explaining how to prepare for a telehealth visit made a huge difference. Patients arrived prepared and confident." (Medical assistant, Site A)

Incremental Goal-Setting: Breaking implementation into achievable milestones facilitated progress and maintained morale. "We celebrated small wins—hitting 50% telehealth adoption, completing training, first month without technical issues. It kept momentum going." (Quality improvement director, Site G)

Factors Predicting Sustainability

Multiple logistic regression identified independent predictors of 24-month sustainability:

$$ \begin{aligned} \text{logit}(P(\text{Sustained})) &= -2.14 \\ &\quad + 0.89 \cdot \text{Unified} \\ &\quad + 0.52 \cdot \text{Sequential} \\ &\quad + 0.68 \cdot \text{Training}_{\text{hours}} \\ &\quad + 0.71 \cdot \text{Leadership} \\ &\quad + \varepsilon \end{aligned} $$

where coefficients represent log-odds (all p<0.01). Unified platform model (OR=2.44, 95% CI: 1.34-4.43), training hours (OR per 10 hours=1.97, 95% CI: 1.28-3.04), and leadership commitment (OR per scale point=2.03, 95% CI: 1.42-2.91) were strongest predictors of sustainability independent of initial investment.

Discussion

This comprehensive multi-site evaluation provides the first rigorous comparison of distinct hybrid care implementation strategies, addressing a critical evidence gap for health systems worldwide transitioning from pandemic-era telehealth to permanent hybrid care models. Several principal findings emerge with important theoretical and practical implications.

Unified Platforms Achieve Superior Implementation Outcomes

Unified platform models—integrating virtual and in-person care through single technology systems with comprehensive workflow redesign—achieved superior outcomes across all RE-AIM dimensions. These models demonstrated highest reach (78.3% vs 52.4% for parallel track), implementation fidelity (91.2% vs 72.4%), and 24-month sustainability (87.5% vs 68.2%). The unified approach required highest initial investment ($2 million average) and longest implementation timeline (18-24 months), but generated positive ROI (+18.7%) at 24 months while parallel track models required ongoing subsidies.

The mechanisms underlying unified platform advantages likely include: (1) seamless technology integration eliminating coordination failures when patients transition between modalities; (2) unified data infrastructure supporting comprehensive quality improvement and population health management; (3) consistent patient experiences maintaining continuity regardless of encounter type; and (4) streamlined workflows reducing cognitive burden on clinicians.

Notably, unified platforms achieved highest reach equity (SES ratio=0.84), challenging assumptions that complex technology exacerbates disparities. The comprehensive support infrastructure—patient engagement resources, digital navigation assistance, accessible interfaces—may offset barriers among disadvantaged populations, consistent with implementation science theories emphasizing the importance of implementation support ^[10].

Parallel Track Models Create Fragile, Unsustainable Systems

Paradoxically, parallel track models—often viewed as lowest-risk approaches due to minimal workflow disruption—demonstrated lowest sustainability (68.2% at 24 months) and required ongoing subsidies (−28.6% ROI). This fragility reflects inherent limitations: when telehealth operates independently, it creates parallel systems with redundant overhead, inconsistent quality, and unclear value proposition.

The finding that parallel track models achieved highest initial clinician adoption (89.4%) yet lowest sustainability appears counterintuitive but reflects low barriers to initial use followed by unsustainable long-term patterns. High initial adoption does not guarantee sustained implementation—a distinction emphasized by the RE-AIM framework but often overlooked in practice ^[12].

These findings suggest parallel track models may represent appropriate transitional phases for health systems new to telehealth but should not be endpoints. Health systems implementing parallel track approaches should develop explicit transition plans toward more integrated models.

Implementation Investment Predicts Sustainability

The strong positive correlation (r=0.78, p<0.001) between initial implementation investment and 24-month sustainability challenges conventional assumptions about minimizing upfront costs. Sites investing $2 million (unified platforms) achieved 94.2% revenue coverage and positive ROI, while sites investing \$2 (parallel track) achieved only 71.4% coverage and required ongoing subsidies.

This finding aligns with implementation science research demonstrating that under-resourced implementation efforts typically fail ^[11]. The incremental cost of unified platform implementation represents strategic investment rather than unnecessary expenditure. Health systems underinvesting in hybrid care implementation may create unsustainable programs that ultimately fail, generating both financial losses and organizational cynicism toward future innovations.

Organizational Context Shapes Implementation Success

Qualitative findings underscore the critical importance of organizational culture, leadership commitment, and change management processes—factors captured in the CFIR "inner setting" domain ^[11]. Sites with visible physician leadership, cultures embracing innovation, and explicit attention to change management achieved smoother implementation regardless of technical approach.

The significant independent effect of leadership commitment (OR=2.03 per scale point, p<0.001) in sustainability prediction models quantifies the importance of organizational factors often treated as unmeasurable in implementation research. These findings support theoretical frameworks emphasizing that technical solutions alone are insufficient without attention to cultural and organizational dimensions ^[15].

Equity Implications

The finding that unified platforms achieved highest reach equity (0.84 SES ratio) has important implications for digital health equity. While concerns about digital divides are well-founded ^[16], our results suggest that comprehensive implementation—including patient engagement, digital literacy support, and accessible technology—can mitigate disparities. This finding challenges deficit-based framing of digital health equity that focuses exclusively on barriers among disadvantaged populations, suggesting instead that implementation quality matters more than population characteristics alone.

The substantial variation in digital access across sites (94.2% in high-income countries vs 67.8% in Nigeria) did not preclude successful implementation, though it required site-specific adaptations including community-based digital access programs and hybrid visit options for patients without home technology.

Implementation Science Framework Contributions

This study demonstrates the value of systematic application of implementation science frameworks to digital health evaluation. The RE-AIM framework provided structure for comprehensive assessment beyond typical clinical outcomes, capturing reach, adoption, implementation, and maintenance dimensions often ignored in efficacy-focused studies ^[12]. Integration with CFIR constructs in qualitative analysis illuminated mechanisms linking implementation strategies to outcomes ^[11].

The mathematical formalization of RE-AIM constructs—particularly the implementation fidelity equation quantifying weighted compliance with workflow components—represents methodological contribution potentially applicable to other implementation studies. Future research should develop validated instruments operationalizing implementation science constructs for digital health contexts.

Practical Recommendations

Based on study findings, we offer evidence-based recommendations for health systems planning hybrid care transitions:

1. Invest in Unified Platforms: Prioritize single integrated platforms over incremental additions to existing systems. While requiring higher initial investment ($1-1.5 million for medium-sized systems), unified platforms demonstrate superior sustainability and positive ROI within 20 months.

2. Allocate Adequate Implementation Time: Allow 18-24 months for comprehensive implementation including planning, training, workflow redesign, and stabilization. Accelerated timelines compromise training quality and workflow integration, reducing fidelity and sustainability.

3. Invest in Training: Provide ≥24 hours of clinician training including hands-on practice, workflow simulation, and confidence-building. Training investment strongly predicts sustainability (OR=1.97 per 10 hours, p<0.001).

4. Ensure Leadership Visibility: Engage physician leaders as visible champions. Leadership commitment is the strongest predictor of sustainability (OR=2.03 per scale point, p<0.001) independent of technical factors.

5. Develop Comprehensive Patient Engagement: Create patient-facing materials addressing digital literacy, technology setup, and hybrid care benefits. Patient engagement programs increase adoption by 24% (p<0.001).

6. Integrate Quality Improvement: Incorporate hybrid care metrics into existing quality improvement programs to maintain attention and accountability beyond initial implementation.

7. Plan for Equity: Develop explicit strategies addressing digital access barriers among disadvantaged populations, including community-based access programs and flexible modality options.

Limitations

Several limitations warrant consideration. First, participating sites were purposively selected for commitment to hybrid care, potentially limiting generalizability to less-motivated systems. However, this selection approach reflects real-world implementation contexts where organizational commitment precedes major technology investments.

Second, the 24-month follow-up period, while longer than typical implementation studies, may be insufficient to assess very long-term sustainability (5-10 years). Extended observation is needed to determine whether initial sustainability translates to permanent practice change.

Third, our implementation model classification, while empirically grounded through document review and observation, represents simplification of complex organizational realities. Sites exhibited some characteristics of multiple models; our classification captured predominant patterns rather than pure types.

Fourth, patient outcome assessment relied primarily on process measures (diabetes control, hypertension management) rather than hard clinical endpoints (mortality, morbidity). While process measures validly indicate quality of care, longer-term outcome studies are needed.

Fifth, cost-effectiveness analysis adopted a health system perspective, excluding patient costs (time, technology) that may be substantial for some populations. Societal perspective analysis would provide more comprehensive economic evaluation.

Finally, the COVID-19 pandemic context during the study period may limit generalizability to non-pandemic contexts. However, the post-acute pandemic phase (2024-2025) represents the "new normal" context most relevant to ongoing hybrid care planning.

Future Research Directions

Several research priorities emerge from this study:

1. Long-term Sustainability Studies: Follow-up beyond 24 months to assess 5-10 year sustainability and model evolution over time.

2. Patient Outcome Studies: Rigorous evaluation of clinical outcomes including morbidity, mortality, and quality of life using randomized or quasi-experimental designs with adequate power.

3. Implementation in Resource-Limited Settings: Focused studies of hybrid care implementation in low- and middle-income countries where specialist shortages are most acute and digital infrastructure most limited.

4. Specialized Care Contexts: Extension to specialty care, hospital-based care, and mental health where hybrid models may have distinct requirements and benefits.

5. Technology Innovation: Evaluation of emerging technologies including artificial intelligence-enhanced clinical decision support, remote patient monitoring integration, and virtual reality examination tools.

6. Workforce Development: Studies of training curricula, credentialing standards, and professional development for hybrid care practice.

7. Policy Research: Examination of regulatory frameworks, reimbursement models, licensure requirements, and equity policies supporting sustainable hybrid care.

Conclusion

This multi-site evaluation establishes that sustainable hybrid care implementation requires unified technology platforms with comprehensive workflow integration, substantial training investment, and explicit attention to organizational culture and leadership. Unified platform approaches achieved superior reach, effectiveness, adoption, implementation fidelity, and 24-month sustainability compared to parallel track or sequential integration models. Despite requiring higher initial investment ($2 million vs $420,000), unified platforms generated positive ROI (+18.7%) at 24 months while parallel track models required ongoing subsidies (−28.6% ROI).

These findings provide evidence-based guidance for health systems worldwide transitioning from pandemic-era telehealth to permanent hybrid care models. The data challenge assumptions that minimizing upfront investment reduces risk, instead demonstrating that adequate implementation investment predicts sustainability. Health systems making comprehensive investments in unified platforms position themselves for sustainable competitive advantage in value-based care; those deferring investment risk obsolescence of care delivery models that patients and clinicians increasingly expect.

The transition to hybrid care represents a strategic inflection point. Our findings suggest this transition warrants substantial investment in integrated infrastructure rather than continuation of fragmented arrangements. Success requires not just technology but comprehensive change management addressing workflow, training, culture, and leadership—the hallmarks of effective implementation science practice applied to digital health innovation.

Data Availability

De-identified datasets supporting the findings are available from the corresponding author upon reasonable request, subject to data use agreements ensuring patient privacy and compliance with institutional review board protocols.

Acknowledgments

We thank participating health system leaders, clinicians, staff, and patients who made this research possible. We acknowledge research assistants who conducted data collection and interviews. We thank the independent Data Monitoring Committee (Dr. M. Johnson, Chair; Dr. P. Wu; Dr. K. Anderson) for oversight.

Author Contributions

K.N. conceived the study, designed the protocol, supervised data collection, conducted statistical analyses, and wrote the manuscript. R.H. contributed to study design, supervised German site implementation, and reviewed the manuscript. S.B. contributed to study design, supervised Swedish site implementation, and reviewed the manuscript. E.M. conducted qualitative data analysis, contributed to interpretation, and reviewed the manuscript. T.O. supervised Nigerian site implementation, contributed to equity analysis, and reviewed the manuscript. All authors approved the final manuscript.

Competing Interests

The authors declare no competing financial interests.

References

1. Insights FB. Telehealth Market Size, Share & COVID-19 Impact Analysis. Fortune Business Insights. 2025;FBI-103719:1-258. https://www.fortunebusinessinsights.com/telehealth-market-103719
2. Dorsey ER, Topol EJ. Telemedicine 2020 and the Next Decade. The Lancet. 2020;395(10227):859-860.
3. Mehrotra A, Bhatia RS, Brindis RG. Paying for Telemedicine After the Pandemic. JAMA. 2021;325(5):431-432.
4. Wosik J, Fudim M, Cameron B, Gellad ZF, Cho A, Phinney D, et al. Telehealth Transformation: COVID-19 and the Rise of Virtual Care. Journal of the American Medical Informatics Association. 2020;27(6):957-962.
5. Bestsennyy O, Gilbert G, Harris A, Rost J. Telehealth: A Quarter-Trillion-Dollar Post-COVID-19 Reality?. McKinsey & Company. 2022. https://www.mckinsey.com/industries/healthcare-systems-and-services/our-insights/telehealth-a-quarter-trillion-dollar-post-covid-19-reality
6. Bokolo AJ. Exploring the Adoption of Telemedicine and Virtual Software for Care of Outpatients During and After COVID-19 Pandemic. Irish Journal of Medical Science. 2021;190(1):1-10.
7. Bashshur RL, Shannon GW, Bashshur N, Yellowlees PM. The Empirical Evidence for Telemedicine Interventions in Mental Disorders. Telemedicine and e-Health. 2016;22(5):381-398.
8. Scott Kruse C, Karem P, Shifflett K, Vegi L, Ravi K, Brooks M. Evaluating Barriers to Adopting Telemedicine Worldwide: A Systematic Review. Journal of Telemedicine and Telecare. 2018;24(1):4-12.
9. Tuckson RV, Edmunds M, Hodgkins ML. Telehealth. New England Journal of Medicine. 2017;377(16):1585-1592.
10. Proctor EK, Landsverk J, Aarons G, Chambers D, Glisson C, Mittman B. Implementation Research in Mental Health Services: An Emerging Science with Conceptual, Methodological, and Training Challenges. Administration and Policy in Mental Health and Mental Health Services Research. 2010;37(1-2):3-15.
11. 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. Implementation Science. 2009;4(1):50.
12. Glasgow RE, Vogt TM, Boles SM. Evaluating the Public Health Impact of Health Promotion Interventions: The RE-AIM Framework. American Journal of Public Health. 1999;89(9):1322-1327.
13. Langbecker D, Caffery LJ, Gillespie N, Smith AC. Using Survey Methods in Telehealth Research: A Practical Guide. Journal of Telemedicine and Telecare. 2012;18(3):241-245.
14. Fetters MD, Curry LA, Creswell JW. Achieving Integration in Mixed Methods Designs: Principles and Practices. Health Services Research. 2013;48(6pt2):2134-2156.
15. Greenhalgh T, Robert G, Macfarlane F, Bate P, Kyriakidou O. Diffusion of Innovations in Service Organizations: Systematic Review and Recommendations. The Milbank Quarterly. 2004;82(4):581-629.
16. Ramsetty A, Adams C. Impact of the Digital Divide in the Age of COVID-19. Journal of the American Medical Informatics Association. 2020;27(7):1147-1148.