Digital Health Operations Case Studies
Healthcare5 Case StudiesUpdated Sep 2026
These five cases look at digital health operations from the side of the people who run it: the nurse reading a portal message at the end of a shift, the scheduler balancing a telehealth queue, the quality lead trying to explain why two similar clinics report different results. The scenarios are composites drawn from common patterns in outpatient, inpatient, and virtual services. Each one asks what the patient needed, what evidence was available, how the team coordinated, and which local protocols applied. The organizations and people are fictional. Work through each situation before reading the actions.
Case Study 1
Rising symptoms reported through a home monitoring app
Kestrel Ridge Medical Group, a three-clinic outpatient practice with a shared patient portal and after-hours nurse line.
The situation
Ronald, a retired surveyor in his late sixties, joined the practice's home monitoring program after a stay in the hospital for a long-standing lung condition. Each morning he records his weight, a breathlessness score, ankle swelling, and appetite in the patient app. For five days the entries moved in the same direction. His weight rose, his breathlessness score climbed from two to four, and he reported eating less. The app generated amber flags and routed them to a shared nurse inbox. Two flags were acknowledged with a standard reply asking him to keep monitoring. The third flag arrived late on a Friday, when one nurse was covering the inbox alongside a clinic backlog.
The challenge
The inbox nurse, Priya, had to decide whether the trend met the threshold for same-day clinician review under the practice's escalation protocol, or whether it could wait for the routine call on Monday. The app had sent the flags as individual alerts, so the pattern was visible only to someone who opened every earlier entry. Priya also had no direct view of the patient's recent appointment notes, which sat in a separate system.
Analysis
Practitioners working in Digital Health Operations usually start by separating what the patient is reporting from what the system is counting. Ronald's needs were plain: he needed someone to read the whole series and decide whether his breathlessness was changing faster than his plan anticipated. The evidence sat in five consecutive entries, and a single amber flag on one of them looked minor. Trend data in a home program is only as useful as the reading applied to it. A rising score with reduced appetite and new swelling calls for a clinician's assessment, not a reply template. No written plan covered that combination, so the nurse had to interpret it alone.
Under the practice's escalation protocol, an amber flag that combines a worsening trend with fluid or appetite changes should reach a covering clinician the same working day. The protocol also requires a recorded reason whenever a flag is closed without contact, which the inbox template does not prompt. Coordination broke down in two places. The flag was not linked to the care plan, so the covering clinician could not see the agreed thresholds, and the evening handover did not mention the open item. A stronger workflow would route trend-based flags to a named clinician and show the last five entries beside each alert.
- Early in the weekScore climbs and the first amber flag is generated
- Mid-weekTwo flags acknowledged with a standard reply to keep monitoring
- Friday, lateThird flag waits in the inbox behind a clinic backlog
- Friday, end of shiftNurse phones the covering clinician and shares the trend
- Next morningPhone review agrees a revised schedule and records reasons
Actions taken
- Add a rule that any flag combining two or more symptom trends routes to a named covering clinician, with the last five entries displayed beside the alert.
- Require a coded reason, chosen from a short list, before any amber flag can be closed without patient contact, and report closures without reasons weekly.
- Link each home monitoring enrollment to a written escalation plan that lists agreed thresholds, so covering staff see them in the patient record during handover.
- Run a monthly audit of trend flags for the first two quarters, comparing time to clinician review against the protocol target, and share results at the practice meeting.
Results
Priya rang the covering clinician before the end of her shift, and Ronald was reviewed by phone the next morning. The covering clinician agreed a revised review schedule with him and recorded the reasoning in his chart. The workflow changes went live by the end of the quarter. In the next quarter, trend flags reached a clinician in a median of about five hours, down from just over two working days. Closures without a recorded reason fell from 31 to 4. Nurses said the displayed entries cut the number of confirmation calls they needed to make.
Questions for discussion
- Which change would have shortened the delay most in this case, the routing rule, the reason code, or the visible entry history, and why?
- How should a practice decide how many trend flags a single nurse can review in one shift before the inbox itself becomes a safety risk?
- What would a covering clinician need to see in a home monitoring record to judge a trend without calling the patient first?
Case Study 2
Pending result missing from an electronic handover
Ashford Bend Regional Hospital, a 380-bed acute care hospital running one shared electronic record across its inpatient units.
The situation
Ruth, a staff nurse on a medical ward, handed over a woman in her seventies admitted for observation after a fall at home. A blood sample had been taken during the afternoon, and the result was still pending at the end of the day shift. Ruth completed the electronic handover template, but the pending results section was free text, and she listed only the mobility plan and fluid chart. The sample was not mentioned. Hassan, the night nurse, received the handover at the bedside and then took charge of a ward with two other patients who were getting worse. Overnight, the result posted to the ordering doctor's inbox.
The challenge
The result needed review that night, and the ordering doctor was covering three wards without a second clinician. The message sat unopened in an inbox that the day team had not been asked to monitor after their shift ended. Hassan did not know the sample existed, and the template had no field that kept an outstanding item visible once the handover was signed.
Analysis
The patient's need was simple to state and easy to lose in transit. She needed a clinician to read a result that was not yet in her notes and to decide what it meant for her observation plan. The evidence in this case was the pending sample itself, plus the nursing observations recorded that afternoon. Neither was wrong. The difficulty was that a pending result is a promise to act, and a handover that leaves it out quietly drops the commitment. Pending items are tasks, not background. Clinical evidence should be visible to whoever holds the patient at that moment, not only to whoever ordered it.
The hospital's handover protocol calls for a structured summary, with each pending investigation named, owned, and given a review time. All three of these elements were missed here. The sample had no owner after the day nurse left the ward, and the review time was never set. The protocol also requires closed-loop acknowledgment of abnormal results, but the acknowledgment step sat in the doctor's inbox, not on the ward record, so nursing had no prompt to check it. Coordination failed at the seam between two systems. Safeguards that depend on someone remembering to look are weaker than ones that put the open item in front of the next person automatically.
Actions taken
- Replace the free-text pending section with a structured field that pulls outstanding samples and results from the laboratory system automatically.
- Add a closed-loop acknowledgment step so each result must be marked seen, actioned, or handed over before the handover record can be signed.
- Set an escalation rule that routes any unacknowledged result to the on-call clinician after a fixed interval, and show the timer on the ward dashboard.
- Audit handover records weekly for pending items, measuring the share of outstanding results listed at handover and the time to acknowledgment for three months.
Results
In the audit period after the change, the share of handovers that listed every outstanding result rose from 64 percent to 96 percent. The median time from result posting to acknowledgment fell from nearly four hours to under forty minutes. Early on, the escalation rule caught one result posted overnight and unacknowledged, and the on-call clinician reviewed it. The decision was documented in the record before the morning round began. The team now treats the outstanding list as the first item on every handover.
Questions for discussion
- Which of the three missing elements, owner, review time, or visibility, would have prevented this gap on its own, and why?
- How should a ward balance a structured pending-results field against the risk of nurses completing it without checking the laboratory record?
- What evidence would convince nursing leadership that acknowledgment rates reflect real review rather than prompts clicked through quickly?
Case Study 3
Telehealth triage queue during a seasonal demand surge
Northgate Virtual Care, a telehealth service that supports three partner practices and a central triage line.
The situation
Northgate Virtual Care receives about 1,200 video and telephone visits a week for three partner practices. In the third week of the season, referrals rose by roughly 40 percent, and the callback queue grew from a median of 35 minutes to over two hours. The service had 14 triage nurses and six clinicians on the roster, and the video platform allowed only 18 concurrent sessions. Routine follow-up calls and urgent symptom calls shared the same queue. Samira, the operations manager, had authority to move staff between shifts but not to add clinicians without a budget change that would take weeks to approve.
The challenge
The team needed to protect urgent symptom calls without turning routine patients away for long periods, and they had to decide within the same day. Options included triage tiering, shifting routine work to asynchronous messages, opening a second video pool by borrowing from a partner, or reducing scheduled slots. Each option had a cost to someone, and the clinical lead, Dmitri, wanted evidence before changing the rules.
Analysis
Patients in a queue do not experience capacity as a number. A person with a worsening cough and a person needing a routine review are both waiting, but only one may deteriorate while waiting. Triage has to sort by clinical urgency, which means the symptom information captured at the front of the queue matters more than the order of arrival. The evidence here came from the intake form, which asked about symptoms in free text for about half the callers. That made the urgent cases harder to find quickly. Good triage treats intake data as clinical evidence and checks that every nurse applies the sorting rule the same way.
The escalation protocol requires any call that meets red-flag criteria to reach a clinician immediately, whatever its place in the queue. The rule was sound, but the transfer request went to a shared clinician pool that was itself stretched, so urgent transfers sometimes waited behind routine work. Weighing the four options against the protocol shows clear trade-offs. Tiering protects urgent calls directly. Asynchronous messages move routine demand off the phone. A borrowed video pool adds capacity without new rules. Reduced slots lower demand but narrow access. Only the first two protect urgent work without a budget change.
- Measure queue Track urgent and routine waits separately every hour
- Re-tier intake Sort calls by symptom urgency before they join the queue
- Rebalance rosters Move clinician time toward peak hours using borrowed video capacity
- Review outcomes Check missed urgent calls and feed findings into the next surge plan
Actions taken
- Introduce a three-tier triage rule at intake, with urgent symptom calls routed to the front of the queue and routine reviews moved to scheduled callbacks within the day.
- Move prescription reviews and other routine follow-ups to an asynchronous message channel with a published response window, so phone capacity is reserved for symptom calls.
- Agree a same-day procedure with the partner practices for borrowing clinician video time, with a named coordinator who decides within an hour whether a transfer is urgent.
- Review the intake form so that symptom questions are structured rather than free text, and audit a sample of calls weekly to confirm the sorting rule is applied the same way.
Results
The tiering rule and the messaging channel went live in the second week of the surge. Urgent calls reached a clinician in a median of 11 minutes, down from 47 minutes, and no urgent transfer waited more than 20 minutes during the peak. Routine callbacks slipped to a median of six hours, compared with about two before, and two partner practices asked for clearer messaging. The service did not add clinicians. The next surge plan now includes a standing borrowing agreement and a weekly audit of urgent calls.
Questions for discussion
- How would you decide how many routine callbacks a service can afford to delay before the delay itself becomes a safety concern?
- Which of the four options would you test first in a smaller service, and what would count as early evidence that it was working?
- What should a borrowing agreement with a partner practice specify so that its clinicians are not drawn away from their own urgent work?
Case Study 4
Uneven referral follow-up across three outpatient clinics
Calloway Ridge Referral Service, a shared office that routes primary care referrals into three outpatient clinics.
The situation
Calloway Ridge Referral Service routes about 3,400 referrals a month from primary care into three outpatient clinics. Over one reporting period, the share of referrals seen within target ranged from 61 percent at North to 88 percent at Riverside, although all three clinics used the same form and scheduling rules. Teresa, the quality lead, pulled anonymized records for 600 referrals per clinic, identified only by reference code and age band. Clinicians at North believed their case mix was heavier, but age and urgency profiles were broadly similar. Two of the returned referral letters to primary care were incomplete at North, and the scheduling team there used a different waiting list order.
The challenge
The team had to decide whether the gap reflected differences in patient need, differences in clinical judgment, or differences in process. The indicator could not separate those explanations on its own, and the local protocol for referral triage did not require a recorded reason when a referral was held. The improvement group had one quarter to show a change and no additional scheduling staff.
Analysis
A first appointment has to come soon enough that the reason for referral still holds. A late visit can mean the evidence that prompted the referral has changed before a specialist sees it. The anonymized records showed that North's late first appointments clustered among referrals with incomplete letters. Without the referrer's stated concern and urgency, the triage nurse had less to work from, and some of those referrals were placed in routine slots. Quality review in this area should start with the content of the referral letter as well as the timing of the appointment.
The referral triage protocol ranks referrals by the urgency the referrer selects and requires that any hold be recorded with a reason. At North the waiting list was sorted by date received, which the scheduling team had adopted during a period of staff turnover and never reverted. Letter completeness was checked by no one before triage, so missing details were discovered only at the first appointment. Coordination between the referral office and the clinics had no shared review meeting, so sites saw only their own figures. The likely causes are therefore a sorting rule that contradicts the protocol, an unchecked intake step, and no shared feedback loop across sites.
Actions taken
- Align the North waiting list with the protocol by sorting on the urgency selected at referral, then date received, and confirm the change with the scheduling team in writing.
- Add a completeness check at intake that holds any referral letter missing a stated concern or urgency for review before it enters the triage queue.
- Hold a monthly shared review across the three clinics using anonymized indicators, with each site presenting its late first appointments and the reasons recorded for holds.
- Run a monitored cycle: track the share of referrals seen within target and the share with complete letters each month, and review progress against a fixed threshold at the end of the quarter.
Results
After the sorting rule was corrected, North's share of referrals seen within target rose from 61 percent to 79 percent over the quarter. The completeness check held 7 percent of incoming letters for review in its first month, and most holds were resolved within two working days. The monthly review showed that Riverside's timeliness depended partly on urgent slots other sites did not have. Central's figures did not move, and the team has not yet found a cause. Next quarter's targets include a fixed threshold for completeness and a clearer reason code for holds.
Questions for discussion
- Which of the three possible explanations, case mix, clinical judgment, or process, would you test first, and what data would you need to rule out the others?
- How much should a monthly shared review rely on anonymized indicators before someone asks whether a site's lower figures reflect local conditions?
- What would make a fixed completeness threshold useful rather than a target that clinics learn to meet on paper?
Case Study 5
Shared portal access and a patient's withdrawn consent
Tamarack Family Health Partnership, a primary care group of four sites with a patient portal that supports proxy access.
The situation
Ingrid, a retired teacher in her eighties, has early memory changes that her family physician, Ruben, is monitoring. Her daughter, Leah, has proxy access to the patient portal, which the practice set up when Ingrid was managing her own affairs. At a recent visit, Ingrid asked that one assessment result be held back from her daughter until she had talked with her herself. The portal's default setting releases every result to a proxy, and the note asking for a hold was added to the record but not to the proxy settings. Leah then called patient services asking for all results, citing worry about her mother's safety.
The challenge
Chloe, the patient services coordinator, could not release or withhold results herself, and the on-call clinician was not available until the afternoon. The team had to decide whether the proxy's concern was a reason to share, whether Ingrid's earlier request still held, and whether her current capacity needed a formal check before any decision. Each option left someone exposed to a different risk.
Analysis
Ingrid's needs are split between two things she is entitled to: her wish to hear difficult news on her own terms, and her family's role in supporting her care. The clinical evidence is the assessment result itself, which the family physician has not yet discussed with her in full. Memory changes can fluctuate, and a decision about information sharing needs an assessment of her ability to understand and weigh that information on the day it is made. The record does not show that check. Until it is done, the practice cannot assume the earlier request is fully informed. Good practice keeps the clinical question separate from the confidentiality question.
The practice's proxy policy allows a proxy to see results unless the patient has asked for a restriction that is recorded in the proxy settings. Here the restriction existed in the clinical note but was never transferred to the portal configuration, which is a digital operations gap rather than a clinical decision. The policy sends withholding requests to the lead clinician and escalates proxy disputes within one working day. Chloe should have logged the call as a potential disclosure dispute and passed it on. Instead, the call was closed with a promise to call back, which left no owner and no time limit.
Actions taken
- Transfer every restriction recorded in a clinical note into the portal proxy settings within the same working day, with a checklist item for the clinician who signs the note.
- Create a patient services script for proxy calls that asks whether a restriction exists, logs the call as a disclosure question, and names the clinician who will respond.
- Set a response limit of one working day for any proxy access dispute, with the lead clinician owning the decision and the governance lead copied on every case.
- Arrange a documented capacity conversation with Ingrid, with her consent to include Leah in part of it, before any further results are released to the proxy.
Results
Chloe logged the call as a disclosure question and passed it to the lead clinician the same afternoon. The clinician spoke with Ingrid within two working days, and a capacity assessment was arranged with Ingrid's agreement. Ingrid chose to share the result with Leah after that conversation, in a meeting with the clinician present. The practice changed its proxy checklist and script, and in the following quarter all restrictions recorded in notes reached the portal on the same day. One proxy access dispute was handled within the one-day limit, and no other results were released in error.
Questions for discussion
- When a patient's earlier request and a proxy's current concern point in different directions, what should a front-line coordinator be permitted to decide alone?
- How would you design proxy settings so that a restriction in the clinical note cannot be silently lost, without turning every note edit into an administrative task?
- What would you want documented after a capacity conversation so that the next person who receives a proxy call does not have to reopen the question?
Learning Resources
Healthcare CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.