Text messaging became one of the most widely deployed public health tools of the COVID-19 pandemic, used for everything from checking on sick patients at home to nudging people toward vaccination appointments. Some of these programs produced striking results: an automated text monitoring system at one large health system was linked to a roughly two-thirds reduction in mortality among outpatients with COVID-19. Others fell flat, with carefully crafted nudge messages producing no measurable change in behavior. The gap between those outcomes tells a useful story about when simple text messages actually change health outcomes and when they amount to digital noise.
Monitoring Sick Patients at Home
The most consequential use of COVID-related texting was probably the simplest in concept. In March 2020, Penn Medicine launched a program called COVID Watch that sent automated text messages to outpatients diagnosed with COVID-19 twice a day for two weeks. The messages asked about symptoms, and when a patient reported new or worsening shortness of breath, an alert went to a clinical team that called the patient within an hour. That call determined whether the person needed to go to the emergency department, get a telemedicine visit, or keep monitoring at home.
In a study comparing about 3,500 patients enrolled in COVID Watch to roughly 4,400 patients receiving usual care, the text-monitored group had significantly lower mortality. At 30 days, the odds of death were about 68% lower for patients in the program, translating to about 1.8 fewer deaths per 1,000 patients. At 60 days, the gap widened to about 2.5 fewer deaths per 1,000 patients. Patients in the program also showed up at emergency departments sooner, by an average of about two days, along with having more telemedicine encounters and hospitalizations. The mechanism was straightforward: catching deterioration early and routing people to care before they became critically ill.1PubMed Central. Comparative Effectiveness of an Automated Text Messaging Service for Monitoring COVID-19 at Home
That said, it is worth flagging that a separate study of a similar remote monitoring program found that while patients who engaged with the system had better outcomes, the difference was likely driven by baseline differences between the groups rather than the intervention alone.2PubMed Central. Associations between remote patient monitoring programme responsiveness and clinical outcomes for patients with COVID-19 In other words, the people who responded to texts may have already been healthier, more engaged, or better connected to the health system. Penn Medicine’s study used propensity score weighting to account for that, which makes its findings more robust, but the broader lesson is that remote monitoring programs can look more effective than they are if the comparison group differs in ways researchers do not fully control for.
Texting People to Get Vaccinated
Vaccination nudges were among the most heavily studied text-based interventions during the pandemic. The results were genuinely mixed, which matters because the headline-grabbing successes sometimes obscure how context-dependent these programs were.
The most prominent success came from a pair of large randomized trials conducted within one health system. Patients who had just become eligible for the COVID-19 vaccine received text reminders designed to make vaccination feel salient and easy. The first reminder, sent a day after eligibility notification, boosted vaccine appointment rates by about 84% and vaccination rates by about 26%. A second reminder sent eight days later produced smaller but still positive effects.3PubMed Central. Behavioural nudges increase COVID-19 vaccinations These are the numbers that got attention, published in one of the world’s most prominent scientific journals.
But other trials told a different story. A randomized trial involving health system employees found that text reminders increased vaccination adherence by about five percentage points within two weeks, a meaningful short-term bump. By four weeks, though, the difference between the nudge group and the control group had vanished. The texts sped up vaccination but did not ultimately change how many people got vaccinated.4JAMA Network Open. Effect of Text Message Reminders and Vaccine Reservations on Adherence to a Health System COVID-19 Vaccination Policy
A third trial tested several variations of text messages, including standard texts, messages featuring clinician endorsement, scarcity framing (“supplies are limited”), and endowment framing (“a vaccine has been reserved for you”). None of these produced a significant increase in vaccination rates compared to an outbound phone call alone. Just over 3% of patients in each group completed a vaccine dose within a month.5JAMA Network Open. Effect of Text Messaging and Behavioral Interventions on COVID-19 Vaccination Uptake And a study in Kenya found that SMS nudges, even ones offering financial incentives, had no detectable effect on vaccination completion. While 91% of recipients remembered getting the message, most reported not being particularly worried about the virus in the first place.6Journal of Development Economics. Can SMS interventions increase vaccination? Evidence from the COVID-19 pandemic in Kenya
The pattern suggests that text nudges work best when the recipient is already motivated and the barrier is logistical, like forgetting to book an appointment. When the barrier is motivational, whether driven by low perceived risk or active hesitancy, a text message alone rarely moves the needle.
What the Message Says Matters, Sometimes
Behavioral scientists spent considerable effort testing whether the framing of COVID-related messages could shift attitudes or behavior. The findings are genuinely complicated and depend heavily on who is receiving the message.
A study of vaccine-hesitant adults tested different ways of presenting the same vaccine information. Among people with high vaccine concerns, messages framed around personal freedom (“protecting your freedom to do the things you enjoy”) produced a preference shift of about 34 percentage points toward vaccination compared to government-framed messages. Messages framed around protecting others also showed positive effects, and those effects held across hesitancy levels rather than being concentrated among the already resistant.7JAMA Network Open. COVID-19 Vaccine Framing and Acceptance Among Adults Who Are Vaccine Hesitant Among people with low vaccine concerns, the framing did not matter much, likely because those people were going to get vaccinated regardless of how the message was worded.
A survey experiment in South India found that loss-framed messages, those emphasizing what you stand to lose by not vaccinating, were modestly more persuasive than gain-framed messages, those emphasizing what you gain by vaccinating.8PubMed Central. Message framing and COVID-19 vaccine acceptance among millennials in South India But a UK study found no significant interaction between message framing and vaccination intention at all. Instead, what predicted intention was how afraid people were of COVID-19, their gender (men had higher stated intention), and whether they had already received a booster dose.9PubMed. Understanding vaccine hesitancy: The role of fear and message framing in COVID-19 vaccination intention in adults in the UK
The honest takeaway is that message framing can matter for specific subpopulations, particularly people who are hesitant but persuadable, and it barely registers for people who have already made up their minds in either direction. The idea that there is some magic wording that dramatically shifts vaccine uptake across a whole population has not held up well.
Government Emergency Alerts
Governments around the world used mass text messages and wireless emergency alerts during COVID-19, and the public reception varied. A study in South Korea examined wireless emergency alert messages and found that “warning” messages, those emphasizing danger and risk, were associated with a significant reduction in public foot traffic. “Guidance” messages, those offering practical instructions about what to do, showed no significant relationship with foot traffic.10Scientific Reports. Effectiveness of wireless emergency alerts for social distancing against COVID-19 in Korea In that same analysis, news articles about the pandemic actually had a stronger association with reduced foot traffic than the government alerts themselves.
A separate study surveying recipients of government-driven smartphone text messages found that participants reported high compliance with preventive measures and positive feelings toward the messages overall. Most perceived the texts as playing an important role in keeping them on track with behaviors like mask wearing and distancing.11CIN: Computers, Informatics, Nursing. Effects of Government-Driven Smartphone Text Messages on Public Compliance With COVID-19 Preventative Measures Whether that reflects genuine behavior change or just self-reported enthusiasm is an open question, since compliance surveys tend to overstate adherence. But at a minimum, most people did not resent receiving the messages.
Triage, Telehealth, and Keeping ERs Clear
One of the earliest practical deployments of text-based tools was pre-hospital triage. Health systems recognized quickly that most COVID-19 cases were mild and did not need in-person emergency care, but anxious patients were flooding emergency departments anyway, straining resources and creating transmission risk. Mass General Brigham (formerly Partners HealthCare) rolled out a digitally automated triage system designed to direct patients to the right level of care before they showed up in person.12PubMed Central. Digital triage: Novel strategies for population health management in response to the COVID-19 pandemic
The University of California, San Francisco built a patient portal-based tool that sorted symptomatic patients into four categories: emergent, urgent, nonurgent, or self-care. Asymptomatic patients were screened for exposure history and provided relevant guidance. The system connected patients to the right care pathway directly, either through self-scheduling or a telephone hotline.13Journal of the American Medical Informatics Association. Rapid design and implementation of an integrated patient self-triage and self-scheduling tool for COVID-19 These tools were not pure text messaging in the SMS sense, but they shared the same underlying principle: use simple digital prompts to sort people efficiently without requiring a phone call or an in-person visit.
Keeping Patients Connected When Care Stopped
COVID-19 disrupted routine healthcare dramatically. Elective surgeries were postponed, clinics shut down, and patients were left in a communication vacuum. One orthopedic practice used text messages to keep patients informed during surgical delays, sending periodic updates and embedded links. Engagement rates averaged 90%, and among those who completed a follow-up survey, the response was overwhelmingly positive: about 98% liked the texts, roughly 96% felt more connected to their care team, and about 92% said the updates helped them avoid calling the office.14PubMed Central. Evaluating the Utility of Using Text Messages to Communicate With Patients During the COVID-19 Pandemic
An interesting wrinkle emerged around frustration. Patients who were in more pain from their delayed procedures were significantly more frustrated with the wait, unsurprisingly, and those more frustrated patients wished they had received even more text messages. They also rated the content as less helpful. The lesson is that for patients under real duress, frequency and content depth matter more. A generic check-in feels inadequate when you are the one hurting.
Text Programs in Low-Resource Settings
In countries with limited healthcare infrastructure, text messaging offered a way to follow up with large numbers of COVID-19 cases that would have been impossible through in-person visits. Rwanda’s home-based care program used a two-way SMS platform called WelTel to monitor patients and contacts remotely. Case managers described being able to follow up with hundreds of patients daily from a single location, a scale that door-to-door visits could never match. The two-way messaging, where patients could respond and request help, was highlighted as a key advantage over one-way broadcast systems.15PubMed Central. SMS-based digital health intervention in Rwanda’s home-based care program for remote management of COVID-19 cases and contacts
But the Kenya vaccination study mentioned earlier is a useful counterpoint. In settings where perceived risk is low, even well-designed SMS interventions with financial incentives attached may not change behavior.16Journal of Development Economics. Can SMS interventions increase vaccination? Evidence from the COVID-19 pandemic in Kenya Text messaging can overcome logistical barriers efficiently, but it cannot manufacture motivation. That distinction played out consistently across high-income and low-income settings alike.
The Language Gap
For text-based health communication to work, people need to be able to read and understand it. That sounds obvious, but a study of U.S. state government COVID-19 vaccine websites found that only 56% of jurisdictions provided professionally translated information about vaccine safety and efficacy in at least one non-English language. Only half provided translated instructions on how to register for or obtain the vaccine. The researchers estimated that roughly 26 million Americans may not have had accurate vaccine safety information available from their state in their primary language, and about 29 million may have lacked translated information on how to actually get the vaccine.17Health Equity. Analysis of Language Translations of State Governments’ Coronavirus Disease 2019 Vaccine Websites
While this study focused on websites rather than SMS messages specifically, the same translation infrastructure, or lack of it, affects text-based programs. An automated text check-in system is useless to a patient who cannot read the message. And machine translation of health content introduces its own risks, since medical terminology translated poorly can range from confusing to actively dangerous. This was one of the pandemic’s quieter failures: the assumption that digital tools are inherently accessible, when in practice they replicate and sometimes amplify existing disparities.
Scams That Rode the Same Rails
The same channels that public health agencies used to reach patients were also exploited by criminals. A survey of 54 studies and numerous reports from government agencies and security firms documented a sharp increase in phishing attacks during the pandemic, many of which used COVID-19 themes to impersonate health organizations, vaccine registration systems, and contact tracing programs.18PubMed Central. The COVID-19 scamdemic: A survey of phishing attacks and their countermeasures during COVID-19 Scammers sent text messages mimicking public health alerts, offering fake vaccine appointments, or claiming recipients needed to “verify” personal information to receive benefits.
The practical consequence was a trust problem. When legitimate health systems send texts that ask patients to click links and enter information, and scammers send nearly identical-looking texts designed to steal data, the recipient has no reliable way to tell the difference. Some patients likely ignored genuine health outreach because they had been warned about phishing, and others fell for scams because they expected their health system to reach them by text. This erosion of trust is a cost that rarely shows up in studies evaluating the effectiveness of text-based health interventions, but it looms over the entire enterprise.
Privacy, HIPAA, and What You Can Actually Say in a Text
Even before COVID-19, public health agencies faced legal uncertainty about sending health information via text message. The HIPAA Security Rule was written with enough flexibility to accommodate new technologies, but in practice that flexibility produced confusion rather than clarity. Agencies that wanted to text patients about their health had two basic options: strip messages of anything that could count as personal health information, or conduct a formal risk analysis and satisfy a set of regulatory requirements to justify including limited health details.19PubMed Central. Text messaging to communicate with public health audiences: how the HIPAA Security Rule affects practice
This constraint shaped what pandemic text programs could actually say. A message like “Your COVID-19 test result is positive” contains personal health information and triggers HIPAA requirements. A message like “Please check your patient portal for an update” does not, but it also requires the patient to have portal access and the motivation to log in. Many programs threaded this needle by keeping texts vague and directing patients to call or log in for specifics. The tradeoff was that vagueness reduced the immediacy that made texting attractive in the first place. During the pandemic, some of these rules were temporarily relaxed under emergency waivers, but as the emergency declarations have ended, the underlying tension remains for any health system that wants to use texting for ongoing patient communication.
Mining Social Media Text for Outbreak Signals
A parallel use of “COVID text” during the pandemic involved not sending messages to people but analyzing the vast quantities of text people were already producing on social media. Researchers used machine learning to classify tweets and other social media posts as either indicating a possible COVID-19 case or not, then mapped positive signals geographically to identify potential clusters. In one application focused on Pakistan, this approach identified regions with likely high transmission rates before formal surveillance systems caught up.20Computers, Materials and Continua. Automatic Surveillance of Pandemics Using Big Data and Text Mining
More broadly, researchers have highlighted the potential for social media text mining to serve as an early warning system for new viral variants by tracking discussions about symptoms and outbreaks.21PubMed. Pandemic transition: A review of social media text mining for pandemic transition in the post-vaccination era The appeal is obvious: people tweet about feeling sick before they get tested, and clusters of such reports can signal an emerging outbreak days before official case counts reflect it. The limitations are equally obvious. Social media users are not a representative sample of the population, people describe symptoms inaccurately, and bots and misinformation generate enormous noise. As a complement to formal surveillance rather than a replacement for it, though, text mining offered genuine value during the pandemic and continues to be developed for future use.
What Carried Over After the Emergency
Some pandemic-era text programs have outlived the emergency that spawned them. In Ireland, a nationwide text message support program for people with chronic respiratory disease launched during COVID-19, targeting a population that was especially vulnerable both to the virus and to the disruption of their routine care. Of the nearly 2,000 people invited, about 36% enrolled. Among those who provided feedback, 98.5% found the messages easy to understand, about 92% said they felt more supported, and 88% reported the texts helped them manage their health.22International Journal of Environmental Research and Public Health. Lung Support Service: Implementation of a Nationwide Text Message Support Program for People with Chronic Respiratory Disease during the COVID-19 Pandemic Programs like this represent the kind of thing the pandemic normalized: using texts not just for acute crisis response but for ongoing chronic disease management.
Clinical trial logistics also adapted. One group of researchers documented how shifting from hybrid to fully remote trial operations during the pandemic improved study task adherence by an average of 20 to 30% across visits, with study completion rates reaching 82%.23PubMed Central. From hybrid to fully remote clinical trial amidst the COVID-19 pandemic Text-based reminders and check-ins were central to keeping participants engaged when in-person visits were impossible. And hospitals found that enterprise text messages to employee smartphones were an efficient channel for rapidly distributing changing clinical guidelines, a practice that could serve just as well during routine operations as during a crisis.24PubMed Central. Text Messaging Real-Time COVID-19 Clinical Guidance to Hospital Employees
The Mental Health Dimension
The pandemic also drove a surge in demand for text-based crisis support. A national crisis chat hotline in the Netherlands saw total chats increase by 48% during the COVID-19 period compared to before. Suicide-related chats rose in absolute number, and the number of severe cases requiring urgent police intervention increased during the lockdown months of April and May 2020 compared to the same period the year before. Anxiety-related issues were more common in 2020 than 2019, rising from about 16.5% to about 19.4% of chats, while depression-related issues actually fell in proportion.25PubMed Central. Suicide-related calls to a national crisis chat hotline service during the COVID-19 pandemic and lockdown
The drop in depression as a proportion of chats does not mean fewer people were depressed. It more likely reflects that the surge in new contacts, many driven by acute situational anxiety from lockdowns and uncertainty, diluted the share of contacts related to longer-term depressive conditions. The overall volume was up sharply, and the most urgent cases increased. For people in crisis who would not call a phone line, text-based chat services were sometimes the only lifeline available during periods of isolation, and demand consistently outstripped what services were built to handle.