Text-Message Reminders to Improve Sunscreen Use: Title and subTitle BreakA Randomized, Controlled Trial Using Electronic MonitoringText-Message Reminders to Improve Sunscreen Use FREE

The use of sunscreen is a preventive health behavior aimed at reducing the risk of developing skin cancers.^{1 - 4} The rates of skin cancer have increased steadily during the past 4 decades, and skin cancer now accounts for 1 in 3 cancer cases worldwide.⁵ In the United States, more than 1 million new cases of skin cancer are diagnosed each year, exceeding all other types of cancers combined.⁵ Exposure to UV light is strongly associated with the development of most skin cancers, especially basal cell carcinomas, squamous cell carcinomas, and some types of melanomas.^{6 - 8} Many of these cases could be prevented if the population took consistent measures to avoid direct sunlight by wearing protective clothing and applying sunscreen.^{3 ,9 - 13}

During the past 30 years, steadfast patient education efforts have increased public awareness of the harmful effects of excessive sun exposure as well as the benefits of using sunscreen.^{14 - 16} Nevertheless, only about 20% of adults in the United States report regular use of sunscreens, and many do not use sufficient amounts of sunscreen, according to the National Health and Nutrition Examination Survey.¹⁷ This persistent discrepancy between many individuals' awareness of evidence-based recommendations and their subsequent lack of adherence is not well understood.

Although adherence to oral medications used in the treatment of chronic diseases averages 50%,^{18 - 19} adherence to topical treatments averages only 20% to 30%.^{20 - 24} The frequently cited barriers to regular sunscreen use include forgetting, inconvenience of application, and a desire to be tanned. Individuals who exhibit high adherence to sun-protective behaviors, including sunscreen use and wearing sun-protective clothing, are more likely to have been counseled by a physician about sun safety.^{11 ,25 - 27} Given the enormous number of people at risk for skin cancer, improving adherence to sunscreen application demands a cost-effective solution.

During the past decade, the increasing availability of affordable consumer-grade technologies has made the cellular telephone virtually ubiquitous. As of 2008, more than 80% of the US population owned a cellular telephone, and approximately 28 billion text messages were sent every month.²⁸ Appropriate leveraging of these novel communication tools may enable more effective delivery of health care education and increase the feasibility of public health interventions on a population-wide scale.

In this study, we used cellular telephone text messaging as a reminder tool to affect participants' preventive health behavior. By using an electronic monitoring device, we were able to assess the effect of text-message reminders on participants' adherence to daily sunscreen application. The long-term goal of this research is to implement effective strategies that will increase adherence to daily sunscreen application in the general population. The primary end point of the study was daily adherence to sunscreen application; secondary end points included weekly adherence rates, subgroup analysis examining whether certain demographic factors were predictive of adherence, and participants' attitudes toward the reminders.

We developed an electronic adherence monitor for topical agents that was adaptable to tubes of varying sizes (Figure 1). Each time the cap on the tube was removed, the electronic monitor sent an SMS (Short Message Service) message in real time to a secured, central data repository, where the dates and times of openings were recorded. The electronic monitors underwent a 2-month technology testing and evaluation process, during which any device malfunctions were addressed before study initiation.

We conducted a randomized, controlled trial of the effect of an electronic text-message reminder system on adherence to sunscreen application. This study was approved by the Institutional Review Board at Massachusetts General Hospital and Brigham and Women's Hospital. Participants were recruited from the general population in the Boston area. Specifically, advertisements were placed in the major local newspapers as well as the Web site Craigslist. The inclusion criteria required participants to be 18 years or older, to own a cellular telephone with text-message features, and to demonstrate the ability to retrieve text-messages. Seventy participants (Figure 2) were compensated nominally for their involvement in the study.

All 70 participants were asked to apply sunscreen daily for 6 weeks in the autumn of 2007. Autumn was defined as September 1 through November 30, according to the standard definition for the meteorological autumn for the northern hemisphere.²⁹ All participants had an initial visit that lasted approximately 80 minutes in which the study staff explained about and showed the participants how to dispense the study sunscreen. Furthermore, the participants were asked to demonstrate their ability to dispense sunscreen using the study device. All participants were also given written instructions regarding the proper frequency of sunscreen application and how much to apply (Table 1). Participants' adherence was captured in real time using transmitting electronic monitors. A 1:1 simple, nonstratified randomization scheme was generated using GraphPad randomization software (San Diego, California). Randomization sequences were concealed within numbered envelopes until interventions were assigned. Half the participants received daily text-message reminders via cellular telephones, and the other half did not receive reminders (control participants). With a sample size of 35 participants in each arm, the study had 80% power to detect a 15% difference in adherence between the reminder and control groups, assuming a standard deviation of 0.25.

The text-message reminders consisted of 2 components: a “hook” text detailing daily local weather information and a “prompt” text reminding users to apply sunscreen. For example, a reminder message would read, “Wed. Partly Cloudy. High 70, Low 55. Your skin would appreciate some sunscreen today.” The reminders were sent daily between 6:30 AM and 7 AM, and a variety of “prompt” messages were used to help maintain participant interest. At the end of the study, we surveyed participants in the reminder group regarding their attitudes toward text-message reminders.

We also collected weather information in the Boston area during the study period and categorized the daily weather patterns as “sunny,” “cloudy,” and “rainy.” If 2 weather patterns were present within the same day (sunny and then cloudy), we recorded the weather pattern that persisted for greater than 50% of the time between 7 AM and 7 PM.

The primary end point of the study was adherence to sunscreen application measured by the number of days participants applied sunscreen during the 6-week study period. The adherence rate for a given participant was defined as the number of days that the participant applied sunscreen divided by the total number of days of the study. Secondary end points were (1) weekly adherence between the 2 study groups, (2) subgroup analysis examining whether certain demographic factors were predictive of adherence, and (3) participants' attitudes toward the reminders. All statistical analyses were performed based on intention-to-treat analysis using SAS statistical software, version 9.1 (SAS Inc, Cary, North Carolina). The normality of the primary end point was examined separately in each group. We deemed the data to be normally distributed in the reminder and control groups based on the Wilk-Shapiro test for normality. A 2-tailed t test was performed to examine the difference in the mean number of days participants were adherent between the 2 groups.

Demographic factors of sex, race, educational level, and age were analyzed in a subgroup analysis to determine whether they were predictive of adherence. Within the no-reminder and reminder arms, participants were dichotomized according to their number of “adherent” days along the medians of each arm (no-reminder arm at 11 days and reminder arm at 21 days). Those at or above the median were defined as having higher adherence, and those below the median were defined as having lower adherence. Sex was dichotomized into men and women; race was categorized as white, black, Asian, and Hispanic; educational level was dichotomized as college education or below college education. In addition, χ² tests were performed relating adherence to sex, race, or educational level. Age was used as a continuous variable in a t test examining the relationship between adherence and age.

The baseline demographic characteristics of the study population are shown in Table 2. The mean age of participants was 33.6 years (age range, 18-72 years), 49 were women (70%), 34 identified themselves as white (49%), and 56 had 1 or more years of college education (80%). Data on sun-protective behaviors were collected from all participants at baseline, including frequency of staying in the shade, frequency of wearing broad-brimmed hats, frequency of wearing long-sleeved shirts, frequency of sunscreen application, SPF (sun protection factor) of sunscreen used, and number of sunburns in the past year. There were no statistically significant differences in baseline demographic characteristics or sun-protective behaviors between the 2 study groups.

At the end of the 42-day (6-week) study period, the control group had a mean adherence of 12.6 days (95% confidence interval [CI], 9.7-15.5) of sunscreen application, which corresponded to a 30.0% (23.1%-36.9%) daily adherence rate. In comparison, the group that received daily reminder messages had a mean adherence of 23.6 days (95% CI, 20.2-26.9) and a daily adherence rate of 56.1% (48.1%-64.1%). A 2-sample t test with equal variances assessing the difference in daily adherence between these 2 groups yielded a P value of less than .001.

The weekly adherence rates in the reminder and control groups are shown in Figure 3. During week 1, the mean adherence rates were similar between the 2 groups: 50% in the control group and 58% in the reminder group (P = .21). However, the mean adherence rates between the 2 groups began to differ significantly in week 2 (P = .01), and this difference persisted throughout the remainder of the study (P < .001 in weeks 3-6).

Adherence data were also stratified based on the 3 predominant types of daily weather patterns—sunny, cloudy, and rainy. After adjusting for the daily weather conditions, we compared the percentage of participants who were adherent to daily sunscreen application between the 2 study arms (Figure 4). During the 8 rainy days of the study period, a mean of 13.3 of the participants in the reminder group (38%) and 3.2 in the control group (9%) were adherent to daily sunscreen application (P < .001). On the 31 cloudy days, a mean of 19.3 of the reminder participants (55%) and 10.5 of the control participants (30%) were adherent (P < .001). Finally, during the 17 sunny days, a mean of 23.5 of the reminder participants (67%) and 14.4 of the control participants (41%) were adherent to sunscreen application (P < .001).

We also investigated whether weather conditions affected the adherence rate after adjusting for the reminder intervention. A Mantel-Haenszel χ² test demonstrated that the weather conditions (rainy, cloudy, and sunny) were significantly associated with the adherence rate after adjusting for the reminder intervention (P < .001).

Subgroup analysis revealed that, in the control group, no significant relationship was found between adherence and sex (P = .52), race (P = .17), educational level (P = .77), or age (P = .64). Similarly, in the reminder group, no significant relationships were identified between adherence and sex (P = .80), race (P = .76), educational level (P = .58), or age (P = .24).

In a poststudy survey, the participants from both comparison arms were asked to recall their frequency of sunscreen application based on a 5-point scale (0, never used sunscreen; 1, forgot to apply ≥3 times per week; 2, forgot to apply 1-2 times per week; 3, forgot to apply 1-2 times per month; and 4, always remembered). The mean (SD) frequency of sunscreen application recalled by participants in the reminder group was 2.42 (0.95); for the control group, the mean (SD) was 2.14 (1.17) (P = .44 by the Wilcoxon rank-sum test).

Participants in the reminder group were also asked to report their opinions about the extent to which the text-message reminder system improved their adherence to sunscreen application. On a scale of 0 to 10 (0, not useful at all; 10, most useful), the mean (SD) rating for the usefulness of the text-message reminders was 8.31 (1.99). Among participants in the reminder group, 24 (69%) reported that they would like to continue using the text-message reminders as a way of improving adherence, and 31 (89%) reported that they would recommend the text-message reminder system to others.

Although sunscreen use is recommended as a preventive health behavior against sunburns and the development of certain types of skin cancers,^{10 - 11 ,30} data from a nationally administered questionnaire estimated that only 20% of the US adult population uses sunscreen regularly.¹⁷ To our knowledge, our study is the first to electronically monitor daily sunscreen adherence. We demonstrated a mean adherence rate of 30% in the control group, which is slightly higher than the 20% reported sunscreen adherence rate in the general US population and could be attributed to the participants' awareness of being monitored during the study.¹⁷

We used cellular text messaging as a reminder strategy to improve adherence to sunscreen application. Participants who received text-message reminders were nearly twice as adherent to a regimen of daily sunscreen application compared with control participants who did not receive reminders. These differences in the adherence rates between the reminder and control groups remained significant after adjusting for daily weather patterns.

A closer examination of the weekly adherence data showed that the adherence rate between the 2 groups started to differ significantly by week 2 of the study, with this difference persisting throughout the remainder of the study. In the control group, the adherence rate continued to decrease throughout the study, with stabilization at approximately 20% adherence at the end of the study, whereas the adherence rate remained stable in the reminder group. Because participants knew that their adherence was monitored and were therefore more likely to be motivated than the general population to comply with sunscreen application, the rapid decline in the adherence rate in the control group is striking and suggests that true adherence in the wider population is likely even lower.

Although adherence rates in the control group decreased during the study, the reminder group maintained a relatively stable level of adherence. The data showed that a simple daily reminder could maintain adherence to sunscreen application during a 6-week study period, thereby encouraging sun-protective behavior on a short-term basis.

Results of the poststudy survey asking participants to recall their adherence rate revealed that the participants in both comparison arms overestimated their actual sunscreen adherence. The overestimation is especially large for participants in the control group, for whom the mean subjective estimation of adherence was approximately 70%, whereas the electronic monitors recorded a mean adherence rate of only 30%. This data corroborate with existing literature that patient recall about medication adherence is often unreliable.¹⁸

Several studies examining the use of text messaging in health care have found that text-message reminders improve outpatient clinic attendance,^{31 - 35} encourage weight loss,³⁶ provide support to diabetic patients,³⁷ and serve as a useful system for critical value notification.³⁸ The results of this study corroborate and extend findings from prior studies that text messages can promote behavioral changes. Cellular text messaging may be an effective reminder tool to promote preventive health behaviors or help individuals adhere to medication regimens for several reasons. In the United States, wireless network coverage is geographically widespread, and many wireless plans have become affordable for the general population. The portability of cellular telephones renders their owners readily accessible through a variety of communications media, and text messaging is a simple, low-cost, and immediate method of delivering reminders without the need for computers. More important, text messaging may be especially attractive to younger individuals, an important target population for development of positive preventive health habits.

It is relevant to note that, despite the ubiquity of cellular telephones, certain socioeconomic groups have a significantly lower rate of cellular telephone ownership compared with the rest of the US population, according to national survey data.³⁹ Specifically, those who do not own a cellular telephone tend to be older, less educated, having lower income, or unemployed.³⁹

The results of this study need to be interpreted in the context of the study design. The long-term benefit of this technology has not yet been demonstrated. This study lasted 6 weeks, and a longer follow-up period would be required to comment on the long-term effects of text-message reminders. However, it has been suggested that 21 to 42 days may be an adequate time frame to develop a new habit.^{40 - 41} Our study used text-message reminders to target forgetfulness, one of the primary reasons for nonadherence to sunscreen use. However, other reasons, such as concern about adverse effects or not understanding the benefits of using sunscreen, may also contribute to nonadherence. The participants were provided with sunscreen in this study, whereas individuals in the community would need to purchase sunscreen to practice this sun-protective behavior. In addition, although electronic monitors instantaneously transmitted information on the dates and times the sunscreen bottle was opened, they were not able to assess how much and where the sunscreen was applied. It is also theoretically possible that some participants might open and close the cap without actually applying the sunscreen. Finally, seasonality may play a part in the observed differences. We chose to conduct this study in the autumn and avoided seasons with excessively high or low sun exposure, such as the summer or winter. It is important to note that adherence to sunscreen application may be greater in the summer when people are more likely to engage in water-related and other outdoor recreational activities. Similarly, sunscreen adherence may be lower in the winter when some people are more likely to stay indoors. Therefore, had the trial been conducted in the summer or winter, the adherence rate for the control group might have been either higher or lower, respectively, than what was observed in this study.

Several technical lessons from this study are applicable to the future development of more effective monitoring and reminder methods to increase adherence. The strengths of the adherence monitors used in this study include objective and instantaneous capturing of adherence data and immediate data upload onto a secure server for viewing. These important features can be further explored in future studies. For example, reminders could be linked to adherence patterns such that the frequency or type of reminders that participants receive is adjusted in response to fluctuations in their individual adherence patterns. The monitor technology could also be improved in several ways to measure adherence more effectively. First, the size of the electronic monitor could be reduced to minimize the “monitoring effect.” Second, an electronic weight-sensing mechanism could be engineered in the device to measure dispensing of the medication.

The suggestions offered by study participants to improve reminder strategies could be grouped into 2 categories: customizing delivery time of the reminder text messages and personalizing the “hook” messages. Future efforts to use text-message reminders as a platform to increase adherence need to focus on designing ultra-compact adherence monitors as well as personalizing the delivery times and the “hook” messages. Moreover, the “prompt” component of the text messages will need to contain relevant instructions about how to take oral medication or apply topical agents. For example, text-message reminders could be sent to students during summer breaks or to people vacationing in sunny regions. The messages may be customized according to individual user preferences, such that the users would be able to choose the time of delivery and the content of the “hook” message (eg, weather report, stock quotes, sports statistics) that accompanies the reminder texts.

Despite continuing educational efforts, a wide gap persists between patients' understanding of the harmful effects of excessive sun exposure and their regular application of sunscreens. The short-term results of our study suggest that cellular telephone text-message reminders are a low-cost, scalable, and effective method of bridging this knowledge-action gap. Introduction of a program that incorporates text-message reminders to a large population may be an innovative preventive health measure against the development of skin cancer

Correspondence: April W. Armstrong, MD, Teledermatology Program, Department of Dermatology, University of California–Davis Health System, 3301 C St, Ste 1400, Sacramento, CA 95816.

Accepted for Publication: June 23, 2009.

Author Contributions: Drs Armstrong, Watson, and Frangos had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Armstrong, Watson, Makredes, Frangos, Kimball, and Kvedar. Acquisition of data: Armstrong, Makredes, Frangos, and Kimball. Analysis and interpretation of data: Armstrong, Watson, Makredes, Kimball, and Kvedar. Drafting of the manuscript: Armstrong and Watson. Critical revision of the manuscript for important intellectual content: Armstrong, Watson, Makredes, Frangos, Kimball, and Kvedar. Statistical analysis: Armstrong and Kimball. Obtained funding: Watson and Kvedar. Administrative, technical, or material support: Makredes, Frangos, and Kimball. Study supervision: Watson, Kimball, and Kvedar.

Financial Disclosure: None reported.

Funding/Support: This study was supported by an institutional grant from Information Systems Council of Massachusetts General Hospital and Brigham and Women's Hospital. Study sunscreen was provided by Neutrogena. Dr Makredes was a Doris Duke Clinical Research Fellow and Dr Armstrong was a Stiefel/Connetics Research Fellow for Improving Access to Dermatologic Care at the time of the study.

Role of the Sponsors: The sponsors played no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.