This is a mixed methods implementation study with a prospective cohort design for the quantitative component and a qualitative exploratory component. The quantitative component captures all emergency referrals recorded through the digital platform during a defined 6-month pilot period using preimplementation historical referral data from participating institutions for pretest-posttest comparison. The qualitative component examines health care provider and patient experiences through surveys, semistructured interviews, and focus group discussions. Quantitative and qualitative findings will be integrated using a convergent mixed methods framework.

The research will be carried out at various levels of the Pakistani public health care system in KP. Primary care is provided by tehsil (subdistrict-level) hospitals and basic health units, secondary care is provided by district headquarters hospitals, and tertiary care is provided by Medical Teaching Institutions (MTIs). The 3 large tertiary care MTIs involved in this study are LRH, KTH, and HMC, all in the provincial capital, Peshawar.

The KP MTI Referral Application was developed through a user-centered design approach informed by continuous stakeholder engagement and iterative refinement. Structured requirement-gathering sessions were conducted with neurosurgeons, emergency physicians, medical officers, nurses, bed managers, hospital administrators, and IT personnel from participating institutions to ensure alignment with existing referral workflows and operational requirements.

A beta version of the platform was deployed in a controlled testing environment at LRH. Beta testing involved volunteer health care providers, including physicians and bed managers, who participated in simulated trauma referral scenarios. Testing evaluated successful referral submission, transmission of clinical information and images, notification delivery, specialist response receipt, referral tracking functionality, and overall usability. End-to-end workflow testing confirmed that referrals could be submitted successfully, notifications were delivered appropriately, specialist responses were received and tracked, and referral progress could be monitored in real time.

Feedback obtained during beta testing informed iterative platform refinements. Major modifications included simplification of the referral form interface, optimization of image-sharing workflows, and enhancement of the notification system through automated reminder alerts repeated every 10 minutes until a specialist response was documented. Validation testing was subsequently conducted against predefined functional and clinical requirements before pilot deployment. Formal training materials and implementation support resources were also developed for frontline users prior to rollout.

The intervention is a digital emergency referral platform, which encompasses a mobile app and web-based application—the KP MTI Referral Application. The platform offers a standardized electronic referral form, bed availability in real time, unique referral identifiers, and triage by urgency (emergency, urgent, and routine; Table 1). The platform has a 2-way communication feature that enables receiving specialists to give clinical management advice to referring health care professionals without the latter necessarily transferring the patient. Automated notifications are set based on the urgency category, with time-sensitive alerts on emergency and urgent referrals and regular alerts on routine referrals. This urgency-stratified method does not cause alarm fatigue linked to universal reminders. Figure 1 illustrates the referral process workflow. Figures 2 and 3 show mobile app interfaces.

The research will engage health care professionals such as emergency physicians, consultants, medical officers, nurses, bed managers, and administrative personnel who have access to referral dashboards to monitor and coordinate. The patients or their legal attendants who undergo the referral process within the study period will also be invited to participate.

In the quantitative component, all the emergency referral records created via the digital platform within the 6-month study period will be considered. Records will be omitted when they are not complete, marked as duplicates, or created in the process of system testing. In the qualitative component, health care providers will be eligible if they have made at least one referral through the digital platform. Patients or attendants will be included if they have been referred to a hospital via the platform. Those who do not give consent will be excluded.

The quantitative part will use a census-based cohort design: all the referral records created via the platform over the 6-month pilot will be considered. According to the administrative referral logs of LRH, KTH, and HMC, it is estimated that 5000 to 7000 emergency referrals will be made. As the sample size of a census-based implementation study cannot be calculated using formal power, the rationale behind the sample size is the expected volumes of institutional referrals. For the qualitative component, health care providers will be selected using stratified purposive sampling to ensure representation across facility levels (primary, secondary, and tertiary), clinical roles (physicians, nurses, and administrators), and geographic locations (urban and rural). Recruitment will continue until thematic saturation is reached, with an anticipated sample of 20 to 30 individual interviews and 4 to 6 focus group discussions involving 6 to 8 participants each. Patients and attendants will be recruited using convenience sampling at the time of referral completion.

Primary outcomes are change in the proportion of unnecessary referrals (defined as referrals where the receiving specialist determines that the patient could have been managed at the referring facility without transfer) and change in referral initiation–to-response time. Secondary outcomes are change in bed use, number of unanswered referral requests, referral acceptance and decline rates, number of cases managed through remote advice without transfer, patient mortality among emergency referrals where feasible, referral-related costs where feasible, patient and health care professional satisfaction, and platform usability (Table 2).

Quantitative data will be collected automatically by the digital platform, recording referral time stamps, referral reason, urgency category, facility identifiers, bed availability status, and referral outcome. Preimplementation historical data will be extracted from paper registers at participating institutions. Health care professional surveys will be administered at 2 time points: within 2 weeks of deployment and at the end of the 6-month period. Patient surveys will be administered at the point of referral completion. Semistructured interviews will be conducted at midimplementation and at end of study. Focus group discussions will be held at approximately 3 months after deployment. Data quality will be checked through double-entry verification of a random 10% sample and monthly data monitoring committee reviews.

The structured survey instrument (Multimedia Appendix 1) comprises separate versions for health care providers and patients or attendants, covering usability, perceived efficiency, communication quality, referral safety, workload, and overall acceptability. Each domain contains 4 to 6 Likert-scale items (1=“strongly disagree”; 5=“strongly agree”). Domain scores are calculated as the mean of item responses.

The semistructured interview and focus group discussion topic guide (Multimedia Appendix 2) covers 7 domains: platform usability, fit with clinical workflow, barriers to adoption, facilitators, perceived impact on referral decisions, bidirectional communication, and recommendations for scale-up.

Quantitative analysis will use SPSS (version 28; IBM Corp). Descriptive statistics, normality testing, pretest-posttest paired comparisons (2-tailed t test or Wilcoxon signed-rank test), chi-square or Fisher exact tests, and multivariable logistic regression will be conducted. A P value of less than .05 will be the significance threshold. Missing data will be handled using complete-case analysis (<10% missing) or multiple imputation (>10% missing). Historical referral data from participating institutions covering the 6 months immediately preceding platform deployment (December 2025-May 2026) will serve as the preimplementation comparison period.

Qualitative analysis will follow the six-step reflexive thematic analysis framework by Braun and Clarke [15]: (1) familiarization with the data, (2) generation of initial codes, (3) search for themes, (4) review of themes, (5) definition and naming of themes, and (6) production of the report. Interviews and focus group discussions will be audio recorded and transcribed verbatim prior to analysis. Two researchers will independently code the transcripts; coding disagreements will be resolved through discussion and consensus. Reflexive memos and an audit trail will be maintained throughout the analytical process. Recruitment will continue until thematic saturation is achieved.

Quantitative and qualitative datasets will first be analyzed independently, and results will subsequently be merged during the interpretation phase using a convergent mixed methods approach. Joint display matrices will be used to facilitate integration. Merged findings will be examined for convergence, complementarity, and divergence between the quantitative and qualitative strands.

Ethics approval has been obtained from the institutional review board of LRH, Peshawar (application reference: 399/LRH/MTI). All participants will provide written informed consent. All data will be deidentified prior to analysis. Data will be stored on encrypted institutional servers accessible only to named study personnel. No financial compensation will be provided to participants.

This study received no external funding. As of manuscript submission, participant recruitment has not yet commenced because pilot deployment is scheduled to begin in June 2026. System development, stakeholder engagement, user-centered design, beta testing, and validation activities have been completed. Data collection is expected to continue through November 2026, quantitative and qualitative data analysis is planned for December 2026 through February 2027, and study findings are expected to be submitted for peer-reviewed publication in spring 2027. System development, user-centered iterative design, stakeholder engagement, beta testing, and validation have been completed. The pilot implementation phase is scheduled to commence in June 2026 across a phased sequence of participating facilities beginning with primary and secondary care hospitals and expanding to the 3 tertiary MTIs by July 2026. Data collection will continue for a 6-month period, with an anticipated sample of 5000 to 7000 emergency referral records. Qualitative data collection will begin at approximately 2 months following platform deployment and will continue until November 2026. Quantitative and qualitative data analysis is planned for December 2026 to February 2027.

The study timeline is outlined in Table 3.

This protocol describes the design, development, and planned evaluation of a neurosurgery-led digital emergency referral system in KP, Pakistan. It is anticipated that the study will show that a structured mobile- and web-based referral platform may decrease the rate of unnecessary emergency transfers, decrease the time spent on initiating a referral and responding to it, and enhance bed occupancy rates in tertiary care facilities.

This research is based on the accumulating evidence on digital referral systems in LMICs, such as the Mobile Obstetric Referral Emergency System in Liberia [8], the Saudi Medical Appointments and Referrals Centre in Saudi Arabia [9], web-based neurosurgical referral systems in higher-income countries [10], and evidence from Indonesia [11] and systematic reviews on e-referral solutions [12].

The strengths are a multicenter design with all levels of health care, high expected volume of samples, and the combination of quantitative and qualitative approaches. Limitations are the lack of randomization, the digital literacy and connectivity differences between facilities, the possible incompleteness of preimplementation paper records, and operational pressures that can influence study compliance.

Future directions could include provincial scale-up, interoperability with the National Database and Registration Authority in Pakistan, artificial intelligence–assisted triage, and economic assessment.

The results of the study will be shared with the KP Health Department and the MTI administrations through peer-reviewed publications, conference presentations, and formal reports. Open access publication will be given priority.

This protocol describes a stringent mixed methods implementation study of a neurosurgery-based digital emergency referral platform in KP, Pakistan. The results can be used to offer a transferable evidence base to justify the use of digital referral systems in KP and similar LMIC contexts.