Protocol P1836-2025 RE-AIM × IMARA × WHO 6 Cluster RCT · NI Margin 7.5% Baseline + Midline · v5.5

Study Design & Framework

Design Cluster RCT (non-inferiority; NI margin = 7.5%)
Facilities 50 (KEPH Level 2–4) across Nairobi, Kiambu, Kitui
Conditions HIV, Hypertension, Diabetes — all combinations
Intervention IMARA 5-domain HIV-NCD integration model
Analysis GLMM (binomial logit), facility random intercept, bobyqa
Status ✓ ARMS LIVE · Baseline + Midline Facility Data

Key Findings — Baseline Snapshot

All results at BASELINE. Findings update automatically with sidebar filters. Facility midline data available in 'Facility Data' tab.

Condition Distribution by Arm

Clinical Outcomes vs Benchmark (by Arm)

Age Distribution by Arm

Participants by County

Sex Distribution by Arm

Table 1. Baseline Characteristics of EIS Study Participants

RE-AIM: Reach. Arm-stratified table provides baseline balance check for the CRT.

Table 2. Clinical Outcomes at Baseline by Condition Group

VL suppression: <200 copies/mL (UNAIDS 2023). BP control: <140/90 mmHg (JNC-8/WHO). Glycaemic: HbA1c<7.5% → FBS<7.2 → RBS<11.1 mmol/L. NI margin = 7.5%.

Table 3. Service Integration and Delivery Indicators

RE-AIM: Reach + Implementation. Integration indicators apply to comorbid patients (≥2 conditions). Acceptability scale: Unacceptable → Acceptable → Highly Acceptable.

Table 4. Patient Satisfaction with Healthcare Services

RE-AIM: Effectiveness (patient-reported). Ceiling effects expected in sub-Saharan Africa facility surveys. Aninanya et al. 2021 framework.

Satisfaction Ratings — Likert Profile

Stacked bar chart of satisfaction levels by domain (Part 10 of patient CRF). Five domains: Communication, Attentiveness, Privacy, Information provided, Concerns addressed. Agree/Strongly Agree colours indicate satisfaction; orange/red indicate dissatisfaction.

Table 5. Out-of-Pocket Health Expenditure by Condition Group

RE-AIM: Maintenance | HF Pillar (Health Financing). CHE = catastrophic health expenditure (OOP >10% monthly income). KES = Kenyan Shillings (~1 USD ≈ 130 KES). SHA = Social Health Authority (replaced NHIF 2024).

Table 6. HIV-Related Stigma Profile Among PLHIV

Agree/Strongly Agree = stigma present. Bazant et al. 2014 SBM-R framework: stigma reduces provider quality perception and patient engagement. PLHIV only (filter applied).

HIV Stigma — Likert Profile

Stacked bar chart per stigma domain (Part 12 of patient CRF). Four domains: Perceived (others' assumptions), Internalized (shame/hesitancy), Anticipated (fear of disclosure), Experienced (discrimination). Green = Agree/Strongly Agree (stigma present). PLHIV only. Consistent with Turan et al. 2012 SHAIP pattern.

Model Configuration

GLMM (binomial logit) · facility random intercept · bobyqa. Arm + County always included. Un-check covariates to simplify.

Fast mode uses nAGQ=0 (3-5× faster). Turn off for full adaptive quadrature.

ICC (facility clustering):

Model Output — Adjusted Odds Ratios

Forest Plot — Adjusted Odds Ratios

Figures match the Baseline Analysis Report. Use Arm/County filters (sidebar) to subset. All figures are interactive — hover, click legend to toggle arms.

Figure 1. Distribution of Chronic Disease Conditions

HIV-only = 44.1%; comorbid patients = 27.4% of total. Consistent with Kenya's dual burden of disease.

Figure 2. Clinical Outcome Prevalence at Baseline by Arm

VL suppression above benchmark; BP and glycaemic control below 50% target. Dashed = Kenya NCD 50% benchmark.

Figure 3. Transport Cost per Visit by Condition Group

Kitui patients face highest transport burden; comorbid patients disproportionately affected. Links to CHE.

Figure 4. Patient Satisfaction by Domain and Arm

High overall satisfaction; privacy scored lowest. Ceiling effects likely. See Aninanya et al. 2021.

Figure 5. HIV-Related Stigma Profile (PLHIV, by Arm)

Perceived > anticipated > internalized > experienced stigma — typical East Africa pattern (Turan et al. 2012).

Figure 6. Median OOP Cost by Component per Scheduled Visit

Medication dominates for NCD patients; HIV patients benefit from subsidised ART. Linked to SHA coverage.

Figure 7. Financial Coping Mechanisms Used by Patients

31% use harmful coping (borrowing, selling assets, skipping meals). Integration expected to reduce burden.

Figure 8. VL Suppression by Arm & County

Arm × County baseline VL suppression. Dashed = UNAIDS 95% target. County as key effect modifier.

Facility Roster — Baseline & Midline

50 facilities across 3 counties. Each facility has a Baseline and Midterm record. Filter by time point.

Total Clinical Workforce by Arm & Level

Support Partner Presence by Arm

Table F1. Facility Readiness & Governance Indicators

WHO Building Block: Service Delivery + Governance. Readiness indicators: guidelines available, MDT meetings, integration meetings, focal persons. Comparison: Baseline vs Midline.

Guideline Availability by Arm (Baseline vs Midline)

Integration Meeting & MDT Activity by Arm

Table F2. Staffing Summary by Cadre, Arm, and Time Point

WHO Building Block: Health Workforce. Staff trained in HIV vs NCD vs combined services. Baseline vs Midline change.

Staff Trained in Combined HIV+NCD by Arm

Staff Attrition (Left) by Arm — Baseline Period

Table F3. Essential Medicine Availability by Arm and Time Point

WHO Building Block: Health Products & Medicines. Products: TLD (HIV), Metformin (DM), Amlodipine (HTN), Gliclazide (DM), Insulin (DM). 'Yes' = in stock at time of data collection.

Medicine Availability Score (0–5) by Arm

Equipment Availability by Arm

Table F4. Health Information Systems & Digital Health Adoption

WHO Building Block: Health Information. EMR = Electronic Medical Records (OpenMRS, KenyaEMR, etc.). Digital health: appointment reminders, telemedicine.

EMR Coverage by Service Type & Arm

Digital Health Tool Adoption by Arm

Table F5. Monthly Patient Volumes by Condition — Baseline Year

Monthly data from facility registers for the baseline assessment period.

Total HIV Patients per Month by Arm

Intraclass Correlation Coefficients (Facility-level)

ICC quantifies facility-level clustering. ICC > 5% validates the cluster RCT design. High ICC for integrated care confirms facility-level intervention is appropriate.

Sensitivity Analyses

Download Reports

Output files match the EIS_Study_Baseline_Analysis_Report format. Files reflect current sidebar filter selection.

All tables include Arm stratification (Intervention vs Control). v5.3 includes facility tables.

⬇ EIS_Baseline_Tables.docx
⬇ EIS_Baseline_Models.docx
⬇ EIS_Facility_Summary.docx
⬇ Cleaned Patient Dataset (.csv)

Export Notes

  • Tables Word doc: Tables 1–6 (patient-level, arm-stratified).
  • Models doc: All 7 GLMM model outputs with aOR, 95% CI, p-value.
  • Facility doc: Tables F1–F5 (readiness, staffing, meds, HIS, volumes).
  • Arm + County are fixed covariates in all models (geography-adjusted).
  • ICC table and sensitivity analyses are included in models doc.
  • Filtered data (arm/county/condition) is used for all patient exports.
  • CSV export contains all original + derived patient-level variables.