We searched for clinical trials published from January 2000 to January 2017 that studied the effect of text messaging on screening for the four cancers of interest. We included all languages of publication. Types of screening methods were limited to mammography for breast cancer; Pap test and HPV test for cervical cancer; colonoscopy, FOBT, and sigmoidoscopy for colorectal cancer; and low-dose CT for lung cancer. Research subjects of all ages were considered. The outcome measures were absolute screening rates or relative screening rates in the text message group compared with a control group.

Studies were identified by searching several electronic databases and additionally searching references of relevant papers. Papers published in a language other than English were translated for review. The search was performed in PubMed, EMBASE, the Cochrane Library, Clinicaltrials.gov, Inspec, HSRProj (Health Services Research Projects in Progress), and NIH Reporter (Textbox 1). Attempts to identify gray literature were made by searching the BIOSIS Citation Index and Proquest Dissertations & Theses Global database, as well as by a manual search.

Citations were independently assessed by 2 investigators (CU and JL). Discrepancies were resolved by consensus and, when needed, with input from a third investigator (PSL). For studies requiring further clarification, the corresponding authors were contacted. The quality of the studies was evaluated using the Cochrane risk of bias tool. Extracted information included the following: (1) characteristics of the participants (including age and sex); (2) type of intervention (including test, duration of follow-up, and number of text messages sent); (3) outcome measure (including screening rates, odds ratio [OR], hazard ratio [HR], and confidence interval [CI] for unadjusted and adjusted results); (4) secondary endpoints; and (5) study limitations.

Our search yielded 2238 citations. After removing duplicates and screening abstracts, 31 papers underwent full review, 22 of which did not meet our prespecified criteria. A total of nine papers were included in the systematic review (Figure 1). These studies were based in England, Spain, Malaysia, Israel, and the United States. Five studies analyzed rates of breast cancer screening by mammography, one analyzed rates of cervical cancer screening by Pap test, and three studies measured rates of colorectal cancer screening (two by FOBT alone and one by a combination of colonoscopy, flexible sigmoidoscopy, or FOBT). No studies were found for lung cancer screening. Together, the included studies enrolled 77,099 participants, most of whom were women aged between 20 and 75 years (Tables 1 and 2). The length of follow-up from intervention delivery to outcome measurement ranged from 2 days to 6 months. Seven studies used a single or set of text messages delivered in 1 day [15-21]. The remaining trials delivered text messages over differing periods: one delivered several text messages over 7 days, and the other delivered individual texts at 1-month intervals until screening occurred (up to three texts maximum; [22,23]). Absolute screening rates were reported in all studies, and we compared the rates in intervention and control groups using the chi-square test (Table 3). Six studies also reported relative risk estimates using OR or HR (Table 4; [17-21,23]).

In this systematic review of text messaging interventions and evidence-based cancer screening, we found that SMS can moderately increase screening rates for breast and cervical cancers and may improve colorectal cancer screening to a lesser degree. Across all studies, text messaging interventions led to increases in absolute screening rates of 0.6% to 15% and relative screening rates of 4% to 63%. Given the small number of studies and heterogeneity in design, we did not attempt to quantify the overall effect of SMS on cancer screening using meta-analysis.

Of the six studies that examined text messaging in breast and cervical cancer screening, increases in absolute screening rates ranged from 4.5% to 15%, and the three studies that reported relative screening rates found improvements of 20% to 63%. Although the smallest reported change in absolute screening rate was not statistically significant (Arcas et al, 4.5%), both the overall direction and the magnitude of absolute effect for SMS seem consistent for breast and cervical cancers. However, as only one cervical cancer screening study met our inclusion criteria, confirmatory investigations are clearly needed.

For colorectal cancer screening, the three included studies found much smaller effects on absolute screening rates, with increases ranging from 0.6% to 3.3%. Hirst et al [21] found a 0.6% difference in absolute screening and a nonsignificant 4% increase in relative screening with FOBT, but this almost certainly underestimates the true impact of SMS because only 24.81% (1026/4134) of the intervention group actually received a text message reminder. Hagoel et al [20] conducted the largest study included in this review, and the results suggest that how the message is phrased has important implications for how it is received: messages posed as a question were more effective than those phrased as a statement, and the most effective messages were questions accompanied by social context. Although the 1.2% increase in absolute screening rate is small, it should be noted that participants in the study had to first order and then complete the FOBT kit, and it has been shown that directly mailing kits to patients significantly improves screening rates [24]. The study of Alaskan Natives by Muller et al [23] was the only colorectal cancer screening study to include colonoscopy and flexible sigmoidoscopy as well as FOBT, and the 3.3% increase in absolute screening rate is more impressive after factoring in that every member of the health care system already receives telephone and letter reminders as well as in-office physician referrals for screening. Whether SMS interventions are more or less effective for colonoscopy than FOBT is an important question, because colonoscopy has become the predominant screening test in the United States and remains the requisite follow-up test for all abnormal FOBTs.

There are at least two reasons why the effectiveness of SMS may differ for colorectal cancer screening and breast or cervical cancer screening. The first may be a difference in the complexity or acceptability of the screening test. Although mammography and the Pap test simply require patients to attend a clinic appointment, patients must be willing to collect one or more stool specimens to complete a FOBT. Colonoscopy is even more complicated, as it requires patients to restrict their diet for several days, drink a large volume of fluids for bowel preparation, and finally undergo an invasive examination. A few trials have studied the impact of text messaging on the quality of bowel preparation, but none has measured the effect of bowel preparation instructions via SMS on screening rates themselves [25,26]. If the problem lies with a higher barrier to overcome for colorectal versus breast and cervical cancer screening, then the strategies used in some included studies may help to maximize the effect of SMS. Providing culturally tailored messages, as both Lee et al and Muller et al did, may be more impactful than generic messages. Similarly, Hagoel et al demonstrated the additional value of using interrogative and contextualized messages. The language of text messaging will continue to be an important area of research, especially for complex screening procedures such as colonoscopy that may require instructions as well as reminders.

A second reason may be the gender difference in the study populations, as the cervical and breast cancer study participants were all women. Indeed, perhaps the most striking finding from the Muller et al study was that only women benefited from the intervention. Hagoel et al also found that women who received the text messages had a statistically significant 21% greater odds of FOBT uptake than men (OR 1.21; 95% CI 1.13-1.28 [20]). However, Hirst et al found no significant difference in FOBT uptake between men and women. It is unclear whether true gender differences exist in how text messages are received or how it motivates behavioral change. Additional studies on the influence of SMS on colorectal cancer screening will help to better quantify its effect for each of the various screening tests, as well as any differences that may exist with respect to gender.

There are many clear advantages to using text messaging to increase cancer screening. According to the United Nations, in 2013, nearly 6 billion of the world’s 7 billion people possessed mobile phones, compared with the 4.5 billion who had access to toilets or latrines [27]. These statistics have clear implications for the potential of using mobile technology to reach underserved communities, including those living in rural and underdeveloped areas [28,29]. A 2016 survey by the National Center for Health Statistics found that 49% (18,074/36,885) of all US adults live in households with only wireless telephones, and those living in (63%, 23,274/36,885) or near (54%, 19,918/36,885) poverty were more likely to live in wireless-only households than those with higher income (48.2%, 17,779/36,885) [30]. In addition, all minority racial groups were more likely to live in wireless-only households, with the greatest difference being between whites and those of Hispanic or Latino descent (45.0%, 16,598/36,885 vs 63.7%, 23,496/36,885). For populations with low health literacy, the short and simple format of text messages may also improve comprehension and reduce a formidable barrier to access. Although the studies included in this review did not specifically report results in participants with low health literacy, text messaging has been shown to improve medication adherence in this population [31].

Beyond the prevalence of mobile phone usage and its reach in low-income settings, another benefit of utilizing text messages for patient outreach is the ability to directly send discreet information in real time, which can then be accessed at the patient’s convenience. This delivery method does not require patients to be available at a particular time or place to receive information, as a telephone call would. Texting also does not rely on an updated or even stable address, as a letter would. However, SMS does require a stable phone number and also may raise concerns about confidentiality if multiple family members share a mobile phone. The ability of SMS to connect with remote and socioeconomically disadvantaged populations is supported by several of the included studies. The mammography study by Vidal et al [18] showed that women who lived in areas with less reliable postal service benefited more from the SMS intervention. Hagoel et al found higher rates of FOBT uptake in patients of low versus high socioeconomic status (OR 1.19; 95% CI 1.10-1.30 [20]). Finally, Kerrison et al [17] found that text messaging was particularly effective at increasing screening rates for patients from the lowest socioeconomic category, who saw a 13.6% absolute increase in screening from the control to intervention group. Vidal et al also demonstrated that adding text messages to an invitation letter was cost-effective for breast cancer screening, and similar conclusions have been reached about the cost-effectiveness of text messaging to improve other health care outcomes [32,33].

Text messaging has been found to be effective for promoting a variety of beneficial health behaviors. Whittaker et al [34] conducted a meta-analysis of text messaging interventions for smoking cessation and found a 67% higher likelihood of successful cessation at 6 months compared with controls. Arambepola et al [35] analyzed 13 trials of diabetic patients and found a statistically significant 0.53% decrease in hemoglobin A1c as well as a nonsignificant 0.25 kg/m² reduction in body mass index in patients who received an SMS intervention. Two meta-analyses have shown that text messaging also improves attendance at primary care and hospital outpatient clinic appointments by 14% to 48% compared with controls [36,37]. Together, these studies consistently convey text messaging’s efficacy as an intervention for health behavioral change, which is further supported by our analysis of cancer screening.

Some studies did not meet our inclusion and exclusion criteria but merit mentioning. Oakley-Girvan et al [38] found that text messaging decreases time to follow-up after abnormal mammograms in a Spanish-speaking Latina population, suggesting that text messaging may be useful to increase rates of both regular screening and follow-up for management of abnormal results. This study also suggests the utility of text messaging for communicating in patients’ preferred language, potentially overcoming communication barriers between physicians and patients. For colorectal cancer, a multifaceted intervention that included text messaging in addition to a mailed letter, instructions on using the stool test, paid return envelopes, and phone messages increased rates of FOBT screening for colorectal cancer from 37% (84/225) to 82% (185/225) [39]. However, we excluded this study because the effect of text messaging could not be isolated.

Notably, our search did not find any studies that assessed the effect of text messaging on screening for lung cancer, a leading cause of cancer death in the United States that affects both men and women. This is not entirely surprising, given the relatively new guidelines that support screening for the subset of the population with a substantial smoking history. However, it highlights a knowledge gap in the relationship between an effective behavioral intervention and a screening test that has been proven to save lives in heavy smokers, and further studies on the subject are warranted.

Several trials are currently under way that may further quantify the relationship between text messaging and cancer screening (Table 5). In addition, as seven of the nine studies we found were conducted in Europe and Asia, more domestic research is needed to ensure that the results are applicable to the US population.

Although we developed a comprehensive search strategy to query the relevant literature, it is possible that some studies were not identified. There was a paucity of data that met our inclusion criteria, and we found only one study on cervical cancer and three on colorectal cancer. In addition, there was substantial heterogeneity in study design, content of the text message intervention, as well as time to follow-up in the included studies. For these reasons, we did not perform a meta-analysis.

A systematic review of the literature suggests that text messaging interventions can increase screening rates for breast and cervical cancers to a moderate degree and for colorectal cancer to a smaller degree. Implementation of text messaging for cancer screening may be an effective method to increase screening rates and thereby decrease cancer-related mortality, even in resource-poor and non-English-speaking populations. However, additional research is needed to determine whether these results apply to all cancer screening tests at the population level.