Remote Perioperative Symptom Monitoring via Smartphone is Feasible: Evidence from the Personalized Prediction of Persistent Postsurgical Pain (P5) study of 2,500 surgical patients

The P5 study demonstrates that remote perioperative symptom monitoring via smartphone is feasible for a large cohort of 2,500 surgical patients, achieving high participation rates despite variable compliance influenced by surgical site and demographic factors, thereby highlighting the need for careful handling of missing data to ensure model generalizability.

The Gist

Imagine you just had a major surgery. Usually, the doctor sees you once before the operation, then again a few weeks later when you're home. In between, they are flying blind, guessing how you're feeling based on your memory of the last few days.

This study is about trying to fix that "blind spot" using something we all have in our pockets: smartphones.

Here is the story of the study, explained simply with some everyday analogies.

The Big Idea: The "Digital Diary"

The researchers wanted to see if they could use a smartphone app to check in on patients three times a day (morning, afternoon, and night) for weeks before and after surgery.

Think of this like a fitness tracker for your pain and mood. Instead of just counting steps, the app asks: "How much does your incision hurt right now?" or "Are you feeling anxious?"

The goal was to see if regular people could actually stick to this routine without being forced to, and if the data they collected was reliable enough to help doctors predict who might have long-term pain after surgery.

The Experiment: 2,500 Patients, No "Rewards"

The team (from the P5 study) enrolled 2,500 adults getting major surgery. Here is the twist: They didn't pay people extra to fill out the surveys.

Usually, in experiments, if you want people to do a boring task three times a day, you give them cash for every check-in. Here, they just asked nicely. They also didn't kick anyone out of the study if they stopped answering. They wanted to see what happens in the "real world," where people get busy, tired, or forget.

The Results: Did It Work?

Yes, but with some bumps in the road.

1. The Participation Rate: About 90% of people tried it at least once. That's a huge success!
2. The Consistency:
   • Before surgery: People were pretty good at it. On average, they answered about 66% of the questions (meaning they missed about 1 out of every 3).
   • After surgery: This dropped slightly to 60%. This makes sense; if you just had heart surgery or vascular surgery, you might be too groggy, in too much pain, or just too tired to tap your phone screen.
3. The "Who" Matters:
   • Who struggled most? The study found that people who were older, had less education, were Black or African American, or relied on government insurance (Medicaid) were less likely to keep up with the daily check-ins.
   • The "Black Men" Gap: There was a specific group—Black men—who had the hardest time sticking with it. The researchers noted this is a critical issue because if we only study people who can stick with the app, our medical models will be biased and won't work for everyone.

The "Reactivity" Question: Does Checking Hurt More?

A common worry is: "If I ask someone about their pain three times a day, will they start feeling more pain just because they are thinking about it?" (Like how checking your bank account constantly makes you feel poorer).

The researchers looked for this "symptom worsening" and didn't find it. In fact, people who answered more questions actually felt slightly less anxious before surgery. It seems that checking in didn't make things worse; it might have even helped people feel a bit more connected to their care.

The Takeaway: A Feasible Tool, But We Need to Fix the Gaps

The Verdict: Using smartphones to monitor surgery patients is feasible. It works. It's not perfect, but it's a powerful new tool that can give doctors a "high-definition" view of recovery instead of a blurry snapshot.

The Catch: The study highlighted a major social problem. The "digital divide" is real. If we build AI models or medical guidelines based only on the people who are good at using apps (often younger, wealthier, or white patients), we will accidentally leave behind the people who need help the most.

The Analogy for the Future:
Imagine building a self-driving car. If you only test it on sunny days on smooth highways, it will work great. But if you don't test it in the rain or on bumpy roads, it will crash when the real world gets messy.

This study proved the "car" (the smartphone monitoring) works. But it also warned us: We need to make sure the car works for everyone, not just the people driving on the smoothest roads. To do that, doctors and researchers need to be very careful about how they handle the data from the people who didn't answer the phone.

In a Nutshell

• Can we use phones to track surgery recovery? Yes, and it's very doable.
• Did people do it? Most did, but they missed some days, especially after tough surgeries.
• Did it hurt? No, checking in didn't make pain worse.
• What's the warning? We need to make sure our technology doesn't ignore the people who are already struggling the most in the healthcare system.

Technical Summary

1. Problem Statement

• The Gap: Ecological Momentary Assessment (EMA) via smartphones offers a promising method for capturing real-time, high-frequency patient-reported outcomes (PROs) to reduce recall bias and improve ecological validity in perioperative care. However, there is a significant lack of large-scale feasibility data supporting the adoption of remote perioperative monitoring.
• The Challenge: Existing literature often inflates compliance rates by excluding participants with low adherence or failing to report non-compliance. Furthermore, it is unclear how demographic factors (e.g., race, insurance status) and clinical variables (e.g., surgical site, pain levels) impact adherence in a real-world surgical setting without direct financial incentives for daily logging.
• The Question: Is frequent smartphone-based EMA feasible for a large, diverse cohort of surgical patients undergoing major surgery, and what factors influence compliance?

2. Methodology

• Study Design: A planned secondary analysis of the Personalized Prediction of Persistent Postsurgical Pain (P5) study, a single-center, prospective cohort study.
• Participants:
  • Sample Size: 2,500 adults (aged 18–75) enrolled; 2,329 proceeded to surgery after exclusions (canceled surgeries, withdrawals).
  • Inclusion Criteria: Undergoing major surgery (>1 hour duration, planned overnight stay), access to a smartphone/tablet (own or borrowed), English proficiency.
  • Demographics: 67% female; 75.2% White/Caucasian, 18.3% Black/African heritage; mean age 53 years.
• Intervention (EMA Protocol):
  • Frequency: 15-item survey administered 3 times daily.
  • Duration: Preoperative period and the first 30 days postoperative.
  • Incentives: Participants received a flat $50 compensation for the preoperative visit and baseline surveys. No direct financial incentives were provided for daily EMA completion, and participants were not excluded from the study based on low compliance.
• Data Analysis:
  • Compliance was calculated as the percentage of completed surveys out of total prompted surveys.
  • Univariate and multivariate analyses were conducted to identify associations between compliance and demographic/clinical variables (e.g., race, insurance, education, pain catastrophizing).
  • Reactivity analysis was performed to assess if frequent monitoring worsened symptoms (anxiety/pain).

3. Key Contributions

• Scale: This is one of the largest studies (N=2,500) to date assessing the feasibility of perioperative EMA, providing robust statistical power compared to previous small-scale studies.
• Real-World Validity: By not excluding non-compliant participants and not incentivizing daily logs, the study provides a "real-world" estimate of compliance that is likely more generalizable to clinical implementation than previous literature.
• Demographic Disparities: It is the first large-scale investigation to comprehensively document how race, insurance status, and education robustly correlate with EMA adherence in a surgical context.
• Reactivity Assessment: The study provides evidence that frequent monitoring does not cause significant symptom worsening (reactivity) in the preoperative period.

4. Key Results

• Overall Feasibility:
  • 90% of participants completed at least some EMA.
  • Preoperative Compliance: Average 66% (Median = 79%). The distribution was left-skewed, indicating high compliance was common.
  • Postoperative Compliance: Average 60% (Median = 71%) in the first 30 days.
  • Response Time: Median completion time was 65 seconds (Average 89 seconds).
• Surgical Site Variations:
  • Preoperative compliance did not vary by surgical site.
  • Postoperative compliance varied significantly by site ($p < .001$). Vascular and cardiothoracic patients showed the lowest compliance (Mean 49–51%), likely due to higher acute postoperative impairment (e.g., sedation).
• Demographic & Clinical Correlates:
  • Lower Compliance was associated with:
    • Identifying as Black/African heritage.
    • Having government-subsidized insurance (Medicaid).
    • Lower education levels.
    • Higher pain catastrophizing (strongest clinical correlate).
    • Older age was associated with higher postoperative compliance.
  • Interaction Effect: Black/African heritage men exhibited the lowest compliance rates (Pre-op Median: 47.5%; Post-op Median: 33.3%).
• Reactivity (Symptom Worsening):
  • No evidence of symptom worsening due to EMA frequency.
  • Participants completing more EMAs actually showed mild improvements in preoperative anxiety, though this is not interpreted as a causal therapeutic effect.

5. Significance and Implications

• Clinical Feasibility: The study confirms that remote perioperative monitoring via smartphone is highly feasible, with compliance rates (60–66%) comparable to or exceeding many meta-analytic estimates when accounting for exclusionary practices in other studies.
• Equity and Generalizability: The findings highlight a critical equity issue: marginalized populations (Black patients, those with Medicaid, lower education) are less likely to adhere to digital monitoring.
  • Implication: Future predictive models for surgical outcomes must employ robust handling of missing data. Failure to do so risks creating algorithms that are biased against marginalized identities and fail to generalize to real-world populations.
• Implementation Strategy: The lack of direct incentives for daily logging suggests that for clinical adoption, systems must be designed to be low-burden (as evidenced by the ~65s completion time) and integrated into standard care workflows rather than relying on participant motivation alone.
• Future Directions: The authors call for randomized designs to fully assess the causal impact of EMA on symptoms and further research to address the digital divide affecting specific demographic groups.

Conclusion: Perioperative EMA is a viable tool for large-scale surgical monitoring, but successful implementation requires addressing systemic barriers to access and adherence among marginalized populations to ensure equitable clinical outcomes.