Residency Advisor
The belief that “patient portals will cut our phone volume in half” is fantasy. The data shows something very different: portals rarely reduce total communication load; they redistribute it—and often increase it.
If you are post‑residency and stepping into an attending role, this matters. Your job satisfaction, panel size, and compensation are going to be increasingly tied to how your practice handles this shift from phone calls to portal messages.
Let’s dissect what the numbers actually show.
1. What the Data Shows About Calls vs Portals
Most health systems rolled out portals with a simple narrative: self‑service plus messaging equals fewer calls. The reality has been more nuanced and, frankly, more brutal.
Across multi‑site systems I have seen data like this pattern repeat:
- Phone calls drop modestly (10–25% in best‑case settings).
- Asynchronous messages explode (3–10× growth over a few years).
- Total “touches” per patient episode often increase.
Here is a simplified but representative snapshot from a large primary care network that went from low portal use to mature adoption over three years.
| Metric | Year 0 (pre‑portal) | Year 3 (mature portal) | % Change |
|---|---|---|---|
| Office phone calls / 1000 pts/mo | 320 | 250 | -21.9% |
| Portal messages / 1000 pts/mo | 15 | 180 | +1100% |
| Total communication events | 335 | 430 | +28.4% |
So yes, calls went down. But the total volume of interaction with the practice went up by almost 30%.
In another hospital-owned cardiology group, new secure messaging policies were implemented with fairly strict triage. The pattern:
- Calls: -12% over 24 months
- Portal messages: +450%
- Total contacts: +18%
The clearest trend from multi‑organization benchmarking:
- Stand‑alone effect of a portal on call volume is modest.
- Net effect is more communication, not less.
To visualize this shift from calls to mixed‑modal communication:
| Category | Phone Calls / 1000 pts | Portal Messages / 1000 pts |
|---|---|---|
| Year 0 | 320 | 15 |
| Year 1 | 300 | 60 |
| Year 2 | 275 | 120 |
| Year 3 | 250 | 180 |
The naive question “Do portals reduce phone calls?” is the wrong one. The more accurate question is: “Do portals shift synchronous demand into asynchronous channels, and what does that do to total workload and revenue?”
Short answer: Yes, they shift demand. And often they expand it.
2. Why Phone Calls Do Not Drop As Much As Advertised
If you expect a clean substitution effect—one portal message replacing one phone call—you will be disappointed. The data does not support that.
Several measurable dynamics interfere with simple substitution.
2.1. New Demand: “If You Build It, They Will Message”
Before portals, a patient might:
- Delay asking a question until the next visit.
- Forget about it entirely.
- Decide it is not worth the trouble of calling, waiting on hold, going through a phone tree.
Portals lower the friction to almost zero. Two taps and a question gets sent. That friction drop is powerful.
I have seen clinics where, after portal rollout, unique patients contacting the practice per month increased from ~28% to ~40% of the panel. That is a 12‑point absolute jump—hundreds of extra touchpoints for a single PCP with a 1,800‑patient panel.
If we oversimplify and quantify:
- Pre‑portal: 28% of panel contacts clinic monthly
- Post‑portal: 40% of panel contacts clinic monthly
For a 1,800‑patient panel, that is:
- 504 patients contacting the office per month → 720 patients per month
- +216 distinct patients contacting, each likely generating ≥ 1 message or call
That is not “replacing calls with messages.” That is more demand.
2.2. Channel Preference and Demographics
Younger, tech‑comfortable patients move quickly to portals. Older, high‑complexity patients often do not. But guess who drives most of your call volume? The older, sicker cohort.
Many practices see:
- 60–70% of messages coming from <65 age group
- 60–70% of phone calls coming from ≥65 age group
- Minimal reduction in calls from the highest‑risk, highest‑touch patients
So you add messaging load without meaningfully shrinking your high‑complexity call volume. The math is ugly but honest.
2.3. Triaging Actually Increases Total Work
A call usually involves:
- Reception or call center → nurse → possibly physician.
A portal environment often adds more micro‑hand‑offs:
- Patient message → inbox pool → triage nurse → clarification message → final answer from physician → sometimes follow‑up call anyway.
One confusing portal message can easily become 3–5 message exchanges. If the original question had gone through a nurse phone call, it might have been 1 contact.
When I have reviewed communication data at message‑level:
- Mean messages per initial portal thread often land around 2.3–2.8.
- 15–25% of threads generate at least one follow‑up within 72 hours.
- Roughly 10–15% still result in a phone call or visit scheduling.
So that “reduced call” was not replaced by a single neat message. It turned into a small conversation that may still involve a phone call.
3. Time Cost: How Portals Change Physician and Staff Workload
Your life as an attending will be increasingly defined by this inbox. Let’s quantify that.
3.1. Per‑Message Time and Daily Totals
Multiple time‑motion studies and EHR log analyses converge on a fairly tight range:
- Simple message (lab normal, brief FYI, scheduling clarification): 30–60 seconds
- Moderate message (med adjustment, brief symptom review): 2–3 minutes
- Complex message (multiple issues, polypharmacy, chronic disease management): 4–6+ minutes
Now layer in typical volume. Reasonable estimates for primary care and many specialties:
- 30–70 messages per physician per weekday in mature portal environments
- Often spiking to 80–100 on Mondays or post‑holiday periods
Use conservative numbers:
- 50 messages/day
- Average 2 minutes per message (which is low if you count reading + thinking + EHR clicks + documentation)
That is roughly 100 minutes per day—1 hour 40 minutes of pure portal time. Five days per week, that is over 8 hours: essentially one full workday a week spent in the inbox.
Now compare pre‑portal call‑related physician time. In many practices before portal growth, attending‑level engagement with phone calls (not staff time, physician time) was in the ballpark of:
- 20–40 minutes/day (orders, callbacks, advice, refills)
You can see the gap. What was once half an hour of loosely structured call follow‑up can easily become 90+ minutes of tightly tracked portal activity.
3.2. Staff vs Physician Burden
Portals do not just hit physicians.
From staffing analyses:
- Call center FTEs: sometimes reduced slightly (5–15%) in high‑adoption clinics.
- Nursing inbox FTEs: often increased (or at least not reduced), since triage simply moves channels.
- MA/admin staff: increased touchpoints for scheduling, refill routing, routing misdirected messages.
In a mid‑size multi‑specialty clinic I analyzed:
- Pre‑portal: 1.2 FTE nurses per 3 physicians to handle calls and refills
- Three years post‑portal: 1.6 FTE nurses per 3 physicians (33% increase), despite a minor drop in call volume
The calls were lower. The asynchronous churn more than filled the gap.
4. Revenue, Compensation, and the Myth of “Free Care”
Most of the early portal era was clinically generous and financially naive: large amounts of uncompensated physician work delivered through messaging.
This is now changing, and you need to pay attention.
4.1. Billing for Portal Messages
CPT codes 99421–99423 (online digital E/M services for established patients) and their equivalents have become more relevant as systems attempt to recapture uncompensated work.
Reported adoption patterns from large systems:
- Only 5–20% of portal threads meet both time and complexity thresholds for billing.
- Of those, many organizations historically billed only a fraction due to fear of backlash.
- Early adopters saw message volume drop 5–15% after transparent messaging‑billing policies, but not collapse.
Typical cutoffs:
- 99421: cumulative 5–10 minutes in a 7‑day period
- 99422: 11–20 minutes
- 99423: 21+ minutes
Now look back at the message time estimates. A single 2‑minute message does not trigger this. But a 6‑message back‑and‑forth with chart review might.
Yet the math still bites:
- 50 messages/day ≈ 250/week
- If 10% are billable threads and you actually bill them, that is roughly 25 billable events/week.
- Spread across 5–10 minute codes, the revenue is meaningful but does not fully offset the entire messaging workload, especially when you count nursing triage time.
4.2. Effect on RVU‑Based Compensation
For RVU‑comp models, portals are a double bind:
- Phone calls never generated RVUs unless tied to billable encounters.
- Portal visits can generate some RVUs but require tight documentation, timing, and correct use of codes.
- Meanwhile, portals cannibalize some in‑person or telehealth visits that would have been higher‑RVU.
So you may find:
- Slightly reduced volume of short in‑person visits.
- Increased asynchronous work that is partially, not fully, billable.
- Physician workdays that feel busier with lower direct visit counts.
That is not sustainable unless compensation models catch up. A few progressive systems are now:
- Adding inbox volume metrics into panel‑size and compensation formulas.
- Setting message‑volume thresholds tied to “stable panel size.”
- Adjusting staffing ratios per 1,000 patients based on message density, not just visit volume.
Most are not there yet.
5. When Do Patient Portals Actually Reduce Phone Calls?
There are scenarios where portals meaningfully cut phone volume. The secret: they have to be used as a self‑service tool first, messaging tool second.
The data is clearer when you separate:
- Transactional requests
- Clinical questions
- Administrative noise
5.1. Transactional Requests: The Real Win
Things like:
- Appointment scheduling and rescheduling
- Medication refill requests
- Lab result viewing
- Simple forms and documentation
In systems that aggressively pushed portal use for these functions, I have seen:
- Refill‑related calls drop by 40–60%
- Lab result “call backs” (patients asking about results) drop by 50–70% when results and comments are clearly visible in the portal
- Scheduling calls shift partially to online requests, though complex scheduling still produces calls
Here is what that looks like structurally in one large internal medicine clinic (normalized per 1,000 patients per month):
| Call Type | Pre‑Optimization | Post‑Optimization | % Change |
|---|---|---|---|
| Refill‑related calls | 90 | 40 | -55.6% |
| Lab result inquiry calls | 60 | 18 | -70.0% |
| Scheduling calls | 110 | 85 | -22.7% |
| Symptom/clinical calls | 60 | 58 | -3.3% |
Notice the pattern: transactional calls plummet. Clinical calls do not.
So if your system:
- Forces refills through a portal request.
- Releases results quickly with a short commentary from you.
- Enables simple scheduling or at least scheduling requests online.
You will see fewer purely administrative phone calls. The total communication load might still rise, but the phone tree misery drops.
5.2. Clinical Questions: Trade Phone for Messages, Not Less Total
For clinical questions (“my blood pressure is running 150s,” “I have new ankle swelling,” “this new med is making me dizzy”), we see:
- Shift from phone calls to portal threads.
- Smaller total reduction in phone volume.
- Increase in total touches per issue (messages + occasional calls + follow‑up visits).
A study‑style summary from aggregated data across several ambulatory specialties:
| Category | Clinical Phone Calls / 1000 pts | Clinical Portal Threads / 1000 pts |
|---|---|---|
| Year 0 | 80 | 5 |
| Year 1 | 75 | 35 |
| Year 2 | 70 | 60 |
| Year 3 | 65 | 85 |
Total clinical contacts increased, even as calls fell a bit.
So your day shifts from:
- “Fewer bursts of synchronous interruptions on the phone”
To - “A constant stream of short asynchronous tasks that are impossible to batch perfectly”
Whether that feels better or worse is personal. But it is not less work.
6. Practical Strategies If You Are the Attending Living With This
You cannot stop portals. You can demand data‑driven design around them.
Here is what I tell attendings looking at this from a cold numbers angle.
6.1. Get Your Own Data
Do not accept generic assurances. Ask for:
- Your portal messages per clinic day, trend over the last year.
- Messages per 1,000 panel patients, compared to peers.
- Mean and 90th percentile messages/day. Mondays and post‑holiday spikes matter.
- Distribution by type: refills, results, clinical questions, admin.
If your leadership cannot produce these numbers, that is a red flag. They are flying blind while asking you to absorb the consequences.
6.2. Push for Message Triage Protocols
The best‑performing groups statistically have:
- Standardized routing: nurses handle first pass on most symptom questions; MAs handle refills within protocol; only escalated or complex cases hit your inbox directly.
- Clear message categories in the portal UI: separate “refill,” “billing,” “symptom question,” “test result question.”
- Automated replies that set expectations for response time and appropriate use.
Well‑structured triage can reduce your direct portal workload by 20–40% without increasing call volume if done correctly and supported by staffing.
6.3. Align Compensation and Staffing With Actual Message Volume
You should be asking:
- How is portal volume factored into my panel size expectations?
- Is there any inbox‑linked component in my comp formula or RVU credit?
- How many nursing FTEs per 1,000 patients are we staffing for, given our message density?
In data‑mature organizations, I have seen thresholds like:
- Above 60 messages per 1,000 patients per month → triggers review of panel size or support staff needs.
- Above certain message workloads → documented as justification for 0.2–0.5 additional RN FTE per group.
If your system is not tracking this, they are essentially getting free labor from you and your staff. That is not sustainable.
FAQ (Exactly 5 Questions)
1. Do patient portals ever meaningfully reduce total workload for physicians?
Rarely, and only when tightly designed as self‑service tools that offload staff work more than physician work. Even then, what you typically see is a shift: fewer phone interruptions and more structured electronic tasks. Total physician workload often stays the same or increases unless visit volume is explicitly adjusted down or compensation structures are changed to recognize inbox time.
2. Is there any specialty where portals clearly reduce phone calls without increasing total work?
Specialties with high transactional load and relatively predictable care patterns—like dermatology follow‑up, straightforward endocrinology follow‑up (stable diabetes or thyroid), or low‑acuity pediatrics—can see clearer gains. However, even in those fields, complex patients and anxious families will still call. Portals help most where questions are simple, repeatable, and easily protocolized.
3. Does billing for portal messages significantly reduce message volume?
Data from systems that implemented transparent portal billing policies shows modest reductions, typically in the 5–15% range for total message volume. Patients do not abandon portals; they become slightly more selective about what they send. High‑value questions remain. Low‑value, casual check‑ins and “just curious” threads decrease somewhat. It trims the extremes, not the core.
4. Are portals worse than phone calls for clinician burnout?
They are different. Phone calls create acute interruption overload. Portals create chronic low‑level cognitive load and “never done” inbox pressure. Burnout data suggests that unmanaged inbox volume is a major driver of emotional exhaustion, especially when it spills into evenings and weekends. Whether that feels worse than phone‑heavy days depends on your personality, but the numbers are clear that unbounded portal work is corrosive over time.
5. If I am negotiating a new attending contract, what should I ask for related to portals and calls?
You should request concrete metrics and guarantees: expected panel size, historical message and call volumes per FTE in your department, staffing ratios for nurses and MAs, and how inbox work is factored into compensation. Ask explicitly whether portal messages can generate RVU credit or fixed stipends, and how after‑hours inbox expectations are handled. You want clarity on workload and a documented plan for adjusting panel size or support if portal volume grows beyond agreed thresholds.
Key points: Patient portals rarely cut your phone calls in half; the data shows modest call reductions and large increases in total communication. The burden shifts toward asynchronous messaging that often lacks adequate compensation. If you want portals to work for you instead of against you, you need hard data on your communication patterns, disciplined triage, and explicit alignment between inbox workload, staffing, and pay.
