The 2001 and 2014 data collections were representative of public and private, 2- and 4-year US colleges with enrolment of at least 500 students. The 2001 study stratified a sampling frame of 2755 schools listed in the Peterson’s guide to 2- and 4-year colleges by the presence or absence of an HC and enrolment size (500–1000, 1001–2000, 2001–4000, 4001–8000, 8001–16 000 and >16 000 students).⁶ The 2014 study stratified a sampling frame of 2753 schools from the Integrated Postsecondary Education Data System (IPEDS) by enrolment size (using the aforementioned cut points).⁷ Of the 910 schools contacted, 736 institutions responded in the 2001 study (response rate 81%) and of the 885 schools contacted in the 2014 study, 482 responded (response rate 55%).

The 2001 study mailed surveys to 910 schools in October of 2001. If the school was part of the American College Health Association (ACHA), the ACHA contact person completed the survey. If the school did not have an ACHA contact, the survey was sent to the contact person listed in the Peterson’s guide to 2- and 4-year colleges. In the 2014 study, schools were mailed a letter asking that the individual most knowledgeable about health services on campus complete the survey electronically through SurveyMonkey between July 2014 and May 2015. Both surveys sent reminder emails and letters to schools that had not yet completed the survey and obtained approval from an institutional review board at the CDC. For additional methodological details, see Koumans et al.⁶ and Habel et al.⁷

Schools sampled in the two studies had different probabilities of selection between their respective strata. Each study created weights based on the probability of selection, and adjusted these weights for non-response. Due to differences in sampling and school characteristics, raking adjustment was performed to ensure comparability of selected variables between the two samples.⁸ Variables used to create raked weights were enrolment size, setting, funding, institution type and region. Auxiliary information used in the raking procedure was sourced from the 2014 IPEDS.

After applying the raking procedure, the estimated proportion of schools with an HC and the estimated prevalence of health service characteristics among these schools were calculated. In the 2001 and 2014 studies, 65% (weighted n = 1814) and 76% (weighted n = 1912) of schools reported an HC. We then compared estimates between the pooled responses from HCs in 2001 to the pooled responses from HCs in 2014 by calculating absolute differences and Z-test statistics. Data were analysed using RStudio version 1.0.44 (RStudio, Boston, MA, USA) and R survey package version 3.31–2 (R Foundation for Statistical Computing, Vienna, Austria), which was used to perform the raking procedure. Two-sided P < 0.05 was considered significant.

Compared with 2001, there was a statistically significant decrease in the percentage of schools offering their own health insurance plans in 2014 (65.5% vs 49.4%; P < 0.001, Z = 4.161). However, a statistically significant increase was observed between the two studies regarding the proportion of plans covering both testing of symptomatic students for STIs (78.2% vs 92.5%; P = 0.003, Z = 3.738) and screening of asymptomatic students (56.4% vs 87.9%; P < 0.001, Z = 6.931; see Table 1).

Although the percentage of schools offering any STI education remained relatively constant between 2001 and 2014 (99.7% vs 98.7%; Z = 1.397), the 2014 study observed statistically significant decreases in specific modes of STI education materials and practices. These included the use of flyers, pamphlets, newsletters, posters, lectures, health fairs, one-on- one education in HCs, orientation presentation and written orientation material (all P < 0.001; Table 1).

The percentage of schools where the HC diagnosed and treated STIs remained stable over time (70.8% vs 72.9%; Z = 0.565), but some changes were noted. In 2014, significant increases were found in the percentage of schools with an HC providing contraceptive services (56.1% vs 65.0%; P = 0.027, Z = 2.245), HIV testing (81.5% vs 92.3%; P = 0.002, Z = 3.419) and gonorrhoea testing (90.7% vs 95.8%; P = 0.045, Z = 2.153). Chlamydia testing among female students remained stable across the two studies (76.4% vs 72.2%; P = 0.347, Z = 0.933; Table 1).

Overall, condom availability remained relatively stable across the two studies (74.1% vs 79.3%; Z = 1.455). Although most schools made condoms free in open display (64.1% vs 71.6%; Z = 1.873), statistically significant declines in distribution method were noted in 2014 with regard to condoms being free upon request, being available for a nominal fee and in vending machines (all P < 0.001; Table 1).

Schools sampled in 2014 were not exactly the same as those sampled in 2001. The 2014 response rate (55%) was also much lower than that in 2001 (81%), perhaps as a result of the 2014 study being unfunded and not having dedicated staff to conduct sufficient follow-up. The wording of one survey question, as well as the survey delivery mode, varied between studies, which may account for response variations. The 2014 study included an age range (women under the age of 25 years) when asking about chlamydia screening, whereas the 2001 study did not. In addition, comparisons on method of STI education were among modes assessed in both of the surveys; however, the 2014 survey included newer, additional modes (watch parties, testing events and awareness campaigns), which were not included for the purposes of the present analysis. The 2014 study asked about the availability of emergency contraceptives and long-acting reversible contraceptives, whereas the 2001 study did not, which made comparing specific offerings of contraception not possible. In the 2014 study, schools were sent the questionnaire via email; the 2001 study sent the questionnaires in the mail. Unmeasured covariates in analyses may have also confounded our results. In addition, some components of overall sexual health, such as sexual violence prevention and services, were not addressed in either of the surveys, making analysis and comparison not possible. Finally, some comparisons were not feasible because the service did not exist in 2001 (e.g. human papillomavirus vaccine), but baseline findings can be found elsewhere.⁷

If colleges aspire to align their programs with NPS priorities, increase sexual health awareness among young adults and normalise STI testing, they may want to consider increasing the promotion of sexual health on campus and access to services in their HCs. To better evaluate and promote STI testing and education, regular assessments of HC services and sexual health best practices for colleges may also be beneficial. Further research exploring increasing access to chlamydia testing on campuses, and better understanding how schools can best protect students’ confidentiality, would also help inform the next steps for colleges and may improve service delivery.