Remote monitoring to reduce readmissions for sepsis ... increased readmissions
Over 65? Maybe silence notifications on this "health" app
Monitoring patients remotely after they’ve survived critical illness is supposed to keep them out of the hospital.
According to a new study out of UPMC in JAMA Network Open, for the elderly after sepsis it’s more like
Since 2019, the U.S. Centers for Medicare and Medicaid (CMS) has expanded a reimbursement infrastructure for remote care (tagged as “hospital at home” and “virtual care,” e.g.) through new billing codes for remote monitoring, video communication, care management, and home-based telehealth.
These diverse programs expanded during the Covid era. Though the pandemic’s proximate pressures driving telehealth decreased, CMS has continued to pin its hopes for efficiencies in care quality and cost on remote patient monitoring.
Sepsis patients, who have up to a 40% one-year mortality rate and high rates of readmission, would seem to be ideal candidates for post-discharge remote patient monitoring.
The enthusiasm may have run ahead of the evidence, however.
The So Ya Had Sepsis Didja? Trial
First, let’s be clear: that was not the real name of the trial. It wasn’t formally named.
That’s just how I (respectfully and fondly) imagine the kind, friendly Pittsburgh-based care teams opening the conversations with the 1,286 patients recently discharged from one of 19 UPMC hospitals after surviving an episode of sepsis. (Only a third had been admitted to ICUs.)
Almost 900 of them were randomized to the intervention, and the 529 who opted in were subsequently pinged with over 10,000 twice-weekly questionnaires delivered to a customized app on their smartphones. They reported their symptoms and could request a call from a nurse. (There was no physiologic telemetry like heart rate monitoring, pulse oximetry, etc.)
Intervention-arm patients were stratified into two intensities of questionnaire (infection symptoms only, or with heart/lung symptoms included) and also two levels of response teams: one had 4 staff nurses; the other included two nurse practitioners who were authorized to diagnose and treat medical conditions they recognized. All that splitting produced four midsized intervention groups.
About 400 patients received usual care, which did not include the questionnaires or ongoing contact with response teams.
Results
Overall, there was an impressive consistency in median days spent at home after discharge across all intervention groups: 90 days. Usual care? Also 90 days.
A relatively narrow range of readmission rates occurred among almost all the groups:
Usual care: 37.8%
Shorter questionnaires, staff nurses responding: 39.7%
NPs responding to either short or long questionnaires: 36-37% readmitted
The outlier, by gestalt, was when staff nurses responded to patients completing the more detailed questionnaires including heart/lung symptoms: 44% of those patients were readmitted.
Sounds Like Yinz Need Come On Back In, Hon
Notably, though, patients older than 65 who were enrolled in remote monitoring were significantly more likely to be readmitted, and had fewer days at home: 39-48% were readmitted among the intervention groups, compared to only 30% for usual care, with an odds ratio around 0.6 for days at home with the intervention (unfavorable and statistically significant).
The staff nurses were again more likely than the NP teams to recommend readmission for the elderly, although all were more likely to readmit than patient- or PCP-led behaviors (i.e., usual care).
Remote monitoring was even more likely to call back patients who had been discharged to skilled nursing facilities: ~42% in the intervention groups vs. 25% who were not enrolled.
Did any of this help?
The intervention did not clearly influence mortality at 90 days, which was 6.5% with usual care and ranged from 5.4% to 8.8% in the intervention arms among all patients.
The study was not powered to detect differences in mortality, so any more granular analysis is speculative.
However, the data suggests monitoring may have had different effects for older vs. younger patients.
Among those <65 years old, mortality was numerically slightly lower in the intervention group than with usual care: no signal of harm.
For those older than 65, there was a concerning numerical increase in mortality among patients assigned to remote monitoring: 6.6% with usual care (~identical to that with usual care in the overall population), but rising to about 10% in the intervention groups. (See table e14 in the supplemental appendix. Again, it was not powered for mortality, and this could easily be due to chance.)
We’ll Just Keep An Eye On Ya N’at
As hospital-based physicians are reminded daily, there is an increasing population in the U.S. of semi-ambulatory patients whose physiology is so abnormal at baseline that many would meet common criteria for hospitalization on an ordinary day.
When dehydration or an ordinary respiratory or GI illness intrudes, causing their pulse to rise from its usual 109 to 125, say, or they turn their oxygen up to 5 liters from 2 to ‘get more air’—and they then push a button in an app to request a call back for assistance—who’s going to tell that elderly and chronically ill person (who was just recently discharged from the hospital with sepsis) to wait it out, based solely on a phone conversation?
Not most staff nurses. NPs might assuage their own anxiety by ordering a chest film, e.g., and avoid advising immediate readmission. Physician behavior was not tested in this study, but “just go to the E.R.” has been recited by enough half-asleep on-call docs (including newsletter authors) that it might as well be a pre-recorded mantra.
Once the patient arrives at the E.D., the psychological bias toward admission for a patient who was funneled there through a hospital-sanctioned remote monitoring program would presumably also be high.
More Care is Not Always Better Care
If remote monitoring increases readmissions that are just in time to treat emerging illnesses and avert death, then bring on the monitoring.
But that’s not what happened here.
Not only was mortality not improved, it was nominally higher for elderly patients in all the intervention groups. That raises the possibility that they were harmed by the additional inpatient care they received (or the difference could be simply due to chance).
In the closest analogue to this trial, ENCOMPASS, remote monitoring paired with a more robust response framework also led to more readmissions; however, mortality was significantly lower among readmitted patients (a good result). It was conducted at seven North Carolina hospitals by the (ethically challenged) Atrium Health.
Similar programs targeted toward COPD and heart failure have for the most part failed to find a benefit in reducing readmissions or quality of life.
The unsuccessful programs have generally shared a lack of infrastructure providing for meaningful action based on changes in patient status. Building that infrastructure — the “hospital at home” model — faces large barriers, which reduce to the fact that homes aren’t, and can’t be, hospitals.
None of this is to say that remote monitoring can’t help; it probably could.
Emerging home-based monitoring technologies (e.g., wearables) will spawn a vast new body of research on how and whether monitoring can prevent not just hospital readmission, but the need for readmission, by enabling more effective post-discharge management of chronic illnesses.
For now, though, products face large barriers of their own, primarily integration with the technologically primitive and highly regulated and innovation-averse U.S. health system, which is dominated by proprietary platforms (e.g., Epic) that have thus far appeared resistant to enabling interoperability with potentially disruptive technologies.
References
MS, S. Y. M. (2026). Remote Monitoring Approaches to Reduce Readmissions After Infection and Sepsis: A Randomized Clinical Trial. JAMA Network Open, 9(6), e2616641–e2616641. https://doi.org/10.1001/jamanetworkopen.2026.16641
Taylor, S. P., Eaton, T., Rios, A., Boyd, D., Tapp, H., McWilliams, A., Chou, S.-H., Halpern, S., Angus, D. C., McCurdy, L., Ganesan, A., Nguyen, H., Connor, C. D, & Kowalkowski, M. (2025). Proactive Telehealth-Based Sepsis Transition and Recovery Support, Hospital Readmission, and Mortality. JAMA Internal Medicine, 185(10), 1238–1238. https://doi.org/10.1001/jamainternmed.2025.3699
Ong, M. K., Romano, P. S., Edgington, S., Aronow, H. U., Auerbach, A. D., Black, J. T., Marco, T. De, Escarce, J. J., Evangelista, L. S., Hanna, B., Ganiats, T. G., Greenberg, B. H., Greenfield, S., Kaplan, S. H., Kimchi, A., Liu, H., Lombardo, D., Mangione, C. M., Sadeghi, B., … Fonarow, G. C. (2016). Effectiveness of Remote Patient Monitoring After Discharge of Hospitalized Patients With Heart Failure: The Better Effectiveness After Transition–Heart Failure (BEAT-HF) Randomized Clinical Trial. JAMA Internal Medicine, 176(3), 310–318. https://doi.org/10.1001/jamainternmed.2015.7712
Asch, D. A., Troxel, A. B., Goldberg, L. R., Tanna, M. S., Mehta, S. J., Norton, L. A., Zhu, J., Iannotte, L. G., Klaiman, T., Lin, Y., Russell, L. B., & Volpp, K. G. (2022). Remote Monitoring and Behavioral Economics in Managing Heart Failure in Patients Discharged From the Hospital. JAMA Internal Medicine, 182(6), 643. https://doi.org/10.1001/jamainternmed.2022.1383
Pinnock, H., Hanley, J., McCloughan, L., Todd, A., Krishan, A., Lewis, S., Stoddart, A., van der Pol, M., MacNee, W., Sheikh, A., Pagliari, C., & McKinstry, B. (2013). Effectiveness of telemonitoring integrated into existing clinical services on hospital admission for exacerbation of chronic obstructive pulmonary disease: researcher blind, multicentre, randomised controlled trial. BMJ, 347(oct17 3), f6070–f6070. https://doi.org/10.1136/bmj.f6070