The last thing America’s colleges need during the pandemic is for their students to hug each other. But in March, shortly before being sent home, many students said goodbye with an embrace, and you can bet many of them will greet one another the same way when they are reunited in the fall.

College means closeness for students on campus, however clear the social-distancing guidelines may be. For this and other reasons, a systematic approach to unlimited testing must be part of the strategy to prevent outbreaks from shutting down our college campuses: All college students should be tested both before and right after they begin actively engaging on campus; any who develop symptoms should be tested immediately; and regular screening of asymptomatic students must follow over the course of the semester.

Despite the necessity of testing, the Centers for Disease Control and Prevention has stopped short of recommending entry testing for all students returning to institutions of higher learning, citing the lack of a systematic study to confirm the need. Yet colleges across the country must make decisions about returning to campus now. As they do so, the variations in their choices will naturally give us real-world, observational data to build the evidence base that the CDC and we are looking for. Until then, based on our best scientific assumptions, here are four recommendations for careful consideration:

First, test everyone before they engage in campus life, either before or immediately upon arrival.

Over 10 percent of people — and as many as 33 percent — tested in some regions of the U.S. have been found positive for Covid-19, and that covers only the subset of the population who actually seek a test. Data from the Diamond Princess cruise-ship outbreak suggests that over 50 percent of those who have Covid-19 (and who may spread it to others) can be asymptomatic. Only one in 3,000 young adults (18- to 29-year-olds) have been hospitalized since the Covid-19 outbreak, but they pose a real risk as “hidden carriers,” and that worries faculty and staff members.

Second, test everyone again one week after arrival.

Some Covid-19 tests may only accurately identify an infection 70 percent of the time, due to false negatives. Additionally, the trip back to campus — especially through areas with a high prevalence of Covid-19 — can expose students and faculty to the virus, and the initial test may not detect these infected individuals. Therefore, everyone should be tested a second time after a week on campus to detect those who have picked up the virus along the way, or for whom the first test may have provided a false negative result.

Third, test symptomatic and exposed individuals.

Rapid testing of symptomatic or exposed individuals with short turnaround times is essential to any campus strategy, and universities need robust contact-tracing efforts as well as quarantine and remote-learning options for these individuals.

Fourth, test asymptomatic individuals regularly.

Researchers with experience studying the behavior of young adults are doubtful that students will comply with preventative measures like social distancing, masks, and self-reporting exposure to infected individuals. That, plus the high rate of asymptomatic Covid-19-positive individuals, means that regular testing of asymptomatic individuals is essential for detecting and containing any outbreak. Educators should take a cue from the U.S. Navy, which recently announced the Sentinel Surveillance Testing program for asymptomatic service members every two weeks, starting with higher-risk populations like health-care workers, galley employees, and instructors at learning centers.

How often asymptomatic testing should be performed should be driven by epidemiological models that consider multiple factors: the reliability of symptom checking and contact tracing, the accuracy of the test method, the effectiveness of containment efforts, and the degree of transmission of the infection on and near the campus. For example, athletic teams in training on campus are undergoing weekly or even twice-a-week testing this summer. The same decisions about the rest of the university need to be guided by epidemiological science.

Current approaches to testing returning students vary widely. For instance, the University of Texas at Arlington is offering testing for symptomatic students on campus only. Given the 13–percent Covid-19–test positivity rate in Texas, administrators there should consider testing students prior to the start of school so that infected students can quarantine at home before arriving on campus. Additionally, while the prevalence of Covid-19 is high in the community, regular surveillance testing of all students on campus could help detect potentially infected individuals who are asymptomatic and keep them from infecting others. UTA, like some other colleges, posts their confirmed Covid-19 cases online (identified only as student, staff, or contractor/vendor), something that can help establish trust and transparency to help reassure an anxious community.

The state of Vermont is requiring students from lower-risk counties (defined as fewer than 400 active cases of Covid-19 per million residents in neighboring states along the East Coast) to test on Day 7 after arrival, while all others need to test on arrival to campus (Day 0) and on Day 7. Thereafter, colleges have the discretion to enforce their own policies, which we recommend should include the testing of symptomatic individuals as well as surveillance testing of asymptomatic individuals. Such state-based requirements can help keep colleges on track.

Columbia University is one of the relatively few colleges that has announced a comprehensive-testing strategy consistent with our recommendations. They will be testing all symptomatic students and those who have been in close contact with a presumed or confirmed Covid-19 case. Additionally, they encourage testing asymptomatic individuals — all students within 14 days prior to arrival, and then for those moving to campus, required testing on arrival and again 7 to 14 days later, with periodic surveillance testing thereafter.

By the fall, colleges will have several options for getting their students and staff tested, including mail-home testing kits for pre-entry, on-site testing facilities, community testing sites, and neighborhood retail stores. New testing methods such as pooled testing could address reagent shortages that are currently limiting testing availability during the recent surge, and the federal investment in testing should lead to lower-cost tests with faster turnaround.

The kind of robust testing strategy we suggest might cost $500 per student per semester. That’s less than two weeks of unemployment in most states. For 20-million college students, this would add up to $10 billion in testing to get through the fall semester. For an industry that adds $600 billion to the nation’s gross domestic product, this investment in higher education is easily warranted.

An intensive-testing strategy is by no means sufficient to keep Covid-19 at bay on our campuses this fall. As the CDC notes, testing is just one component in a comprehensive strategy. But it is a necessary part of any viable plan to bring higher education back to campus safely. It is something colleges cannot ignore.