The rapid scale-up of antiretroviral therapy (ART) for HIV-positive individuals has transformed HIV from a terminal to a chronic illness, with annual deaths declining from 2.3 to 1.6 million globally between 2005 and 2012.1 However, substantial challenges remain. Poor adherence to ART medications has been associated with treatment failure, progression of HIV to AIDS, development of resistant strains of HIV, and death.2–8 Although regimen improvements may put less demand on perfect adherence,9–11 successful treatment requires sustained lifetime adherence, with a goal of maintaining adherence above 95%. Several reviews suggest that behavioral interventions can improve ART adherence, although the intervention effect is rarely durable.12–16 These reviews underscore the fact that relatively few interventions have been rigorously tested, especially outside highly developed countries.
The use of mobile phone technologies has emerged as a potentially powerful strategy for ART adherence promotion.17–22 A recent meta-analysis of evaluations of text message interventions showed that such interventions increased ART adherence; a few improved biological outcomes.23 Of note, although 4 of the 8 trials in the analysis were conducted in low-resource settings, none took place in Asia.
China has Asia's second-largest HIV epidemic, with an estimated 780,000 individuals infected.1 New infections have numbered roughly 50,000 annually,24,25 although updates suggest an increase in new cases in recent years.26–28 Western and southern border areas have been affected disproportionately, mainly because of high rates of heroin use.25,29–31 China's government scaled up free provision of ART beginning in 200332,33 and has reported 140,000 individuals receiving ART by 2012.25 Drug resistance is emerging as a problem, indicating widespread suboptimal adherence;34–36 however, few adherence interventions have been studied in China. Two exceptions include a nurse-delivered counseling intervention conducted in Beijing, which showed positive effects of counseling and reminders,37 and our own past work in the heavily impacted border province of Yunnan, the “Adherence for Life” (AFL) study. AFL used electronic drug monitoring (EDM) as an information and counseling tool in a predominantly heroin-using population and found that EDM-informed counseling (EDM feedback) significantly improved ART adherence.38
We hypothesized that adherence information and education are likely to be most effective when delivered in real time and in direct response to lapses when they occur. We therefore developed and tested a real-time web-enabled adherence support intervention using wireless technology for ART adherence monitoring. This real-time adherence messaging and counseling intervention included the use of a wireless medication container/communicator (Wisepill Technologies, Capetown, South Africa), which records the date and time of each container opening and communicates the data immediately through general packet radio service to a central server.39 Piloting showed this technology to be feasible and acceptable in China, with reliable monitoring of adherence over time.40 To date, however, no studies have reported on its utility as an ART adherence support.40–44 The “China Adherence through Technology Study” assessed the effect of this real-time feedback using triggered cell phone reminders coupled with Wisepill-generated data-enhanced counseling.
Subjects and Study Design
The province of Guangxi, China, like neighboring Vietnam and other nations of Southeast Asia, has been greatly affected by the regional heroin use epidemic and associated HIV transmission. Home to numerous ethnic minorities, Guangxi has an estimated 80,000–100,000 HIV-positive individuals,45 with new infections averaging 10,000–15,000 annually (data from late 2011).46 This study was conducted at the Guangxi Center for Disease Control and Prevention (GX-CDC) ART clinic in Nanning, a city of 7 million residents.47 The clinic is staffed with 4 physicians, 2 nurses, and 3 HIV counselors and treats over 1000 patients.
The randomized controlled trial enrolled HIV-positive adult patients on HIV treatment at the GX-CDC clinic. Most clinic patients followed a twice-daily ART regimen consisting of nevirapine or efavirenz, plus lamivudine with stavudine or lamivudine with zidovudine, although some were on a once-daily regimen of lopinavir/ritonavir plus tenofovir or abacavir. As part of usual care, all patients on ART met with adherence counselors who were available for support at the request of a clinician or patient.
Patients were eligible if they were receiving or initiating ART, aged 18 years or above, owned a mobile phone, and deemed at risk for poor adherence by clinicians or themselves. To operationalize this latter criterion, clinic staff referred to the study coordinator all patients they felt might face adherence challenges for any reason, including substance abuse, alcohol dependency, previous treatment failure, and mental health problems. Treatment-experienced patients and those initiating ART were eligible given time constraints and use of a randomization procedure designed to address the greatest source of potential bias (adherence level), an approach used previously with success.38 Posters were displayed at the clinic encouraging any patient who believes at risk of poor adherence to consider participation. Subjects provided written informed consent before enrollment. All subjects received 150 yuan (approximately US $25) monthly as reimbursement for lost work time and travel costs associated with study participation.
On enrollment, an on-site study coordinator gave each subject an electronic adherence monitoring container for use with his/her ART medications. In consultation with the study coordinator, subjects selected one or more ART medications to be monitored within the device. Selection was based on fit and subject preference for refilling frequency. All subjects underwent baseline adherence monitoring using the device for 3 months and then were stratified into optimal or suboptimal adherence groups (defined as ≥95%, <95% average adherence) before starting the intervention period. Optimal adherence was defined as taking the dose within a ±1-hour period around the specified dose time, based on previous work indicating that this measure best predicted clinical outcomes.48 Subjects were randomized within each stratum in a 1:1 ratio to intervention and control groups. Randomization was performed on site, through a block randomization procedure in which the site coordinator pulled an unmarked allocation envelope, the inside of which had a single paper stamped with either “intervention” or “control,” from a larger envelope (labeled “optimal” or “suboptimal” as appropriate given the subject's adherence category) that originally held 10 such allocation envelopes, 5 for each arm. When each large envelope was empty, it was replaced with another one, similarly containing 10 allocation envelopes.
All subjects were seen monthly, and all received electronic adherence monitoring throughout the study. Intervention subjects received adherence counseling as clinically indicated, and an SMS mobile phone reminder sent whenever the Wisepill system failed to detect a device opening by 30 minutes past a scheduled dose time. The text messages were personalized, with subjects selecting from a list of 10 options developed jointly by clinicians and patients, including “carry on, carry on!” and “be healthy, have a happy family.” To prevent disclosure of HIV, text reminders did not refer to HIV, ART, or other disease-related topics. When seen monthly in clinic, subjects with prior-month adherence <95% received a behaviorally targeted counseling session with a counselor guided by a detailed day-to-day adherence performance report with a visual display of doses taken and summary of doses taken on time, off time, and missed in the previous month. These sessions had no predetermined length; practice sessions indicated that 15–20 minutes were sufficient for a meaningful discussion in most cases. Control subjects received usual care adherence counseling as clinically indicated at each visit or if they self-reported suboptimal adherence (<95%). Given the nature of the intervention, it was impossible to blind subjects or clinicians to subjects' randomization arm.
Adherence counselors were trained to use supportive counseling methods. Counselors received specific training for intervention subjects: how to review the report with the subject, explore reasons for missed or off-time doses, inquire about possible challenges, and strategize about approaches to overcome them. The approach was based on the AFL counseling model, in which counseling sessions were designed to foster a personalized discussion of each subject's unique experiences characterized by lack of judgment of poor adherence.38
CD4-cell count and HIV plasma RNA tests were conducted at 2 points: (1) during or shortly before the preintervention interval (month 0 to month 3) and (2) postintervention, defined as months 6–7 postrandomization (study months 9–10). CD4-cell counts were measured by FACSCalibur flow cytometry (Becton-Dickinson, CA). HIV plasma RNA tests were performed with an Organon Teknica NucliSens machine (Boxtel, the Netherlands). The lower limit of the viral load assay was 50 copies per milliliter.
Institutional review boards at Boston University Medical Center and the Guangxi Provincial CDC in Nanning, China, approved the protocol. The study is registered on ClinicalTrials.Gov (NCT01722552).
Date and time records for each device opening were used to construct detailed records of adherence over the study period. Adherence was defined as proportion of doses taken on time, in accordance with our past research (and that of others) showing that on-time adherence predicted undetectable viral load (UDVL) most significantly.48–50 Accordingly, adherence was defined as [number of doses taken ± 1 hour of dose time]/[total number of prescribed doses]. Doses taken outside the ±1-hour window were considered nonadherent. Other end points included the proportion of all scheduled doses taken, mean CD4-cell count and changes in CD4-cell count, and UDVL.
To assess efficacy of the intervention, the primary outcome was the difference between intervention and control subjects in the proportion achieving optimal (≥95%) on-time adherence postintervention (specifically, the last 30 days of the 6-month intervention period). In addition, we compared proportions with optimal adherence over the entire 6-month intervention period and mean adherence (both in the last intervention month and over the entire intervention period) between arms and within adherence groups. The secondary outcomes were postintervention differences in CD4-cell count and UDVL and change in CD4-cell count from baseline to month 9 between arms.
The primary analysis was by intention to treat (ITT); a secondary per protocol analysis was also conducted. The ITT analysis included data for all randomized subjects, with postintervention adherence measured by the last 30 days of available data; adherence over the 6-month intervention period was measured using all available postintervention data. All baseline and postintervention CD4-cell count and HIV viral load data were used in clinical outcome analyses. We also conducted bivariate and multivariate regression analyses to assess potential bias on intervention effect of variables imbalanced at randomization. Because the ITT and per protocol results were very similar, here we present the results of the ITT analysis.
To assess the impact of the triggered cell phone reminders specifically, as opposed to the combined effect of reminders and enhanced counseling, we compared the mean monthly number of “late doses,” defined as doses not taken by 30 minutes after scheduled time, when messages were triggered for intervention subjects but not for controls, during the preintervention and intervention periods between arms. To explore the impact of reminders on adherence behavior, we also compared the proportion of all doses taken between 30 and 60 minutes past dose time. Our hypothesis was that individuals who were more than 30 minutes late for a dose would be more likely to take their dose in the next 30–60 minutes if they received a reminder.
Our sample size was designed to detect a 25 percentage-point difference in proportion achieving optimal adherence postintervention. This difference was based conservatively on the previous AFL study, in which proportions achieving optimal adherence were 84% vs. 39% in intervention subjects vs. controls in the last intervention month. The target sample size was 120, assuming a minimum of 80% power at a 2-sided alpha of P = 0.05, and allowing for 20% attrition. The study was not powered to detect differences in clinical outcomes. We used Cochran–Mantel–Haenszel χ2 tests for categorical variables and Student t tests for continuous variables, with findings expressed as risk ratios (RRs) and 95% confidence intervals (CIs) for categorical variables and mean and SD values for continuous variables. All inferences were based on a type 1 error equal to P = 0.05. We used SAS version 9.4 (SAS, Cary, NC).
Subjects were enrolled between December 2012 and April 2013. Of 166 patients eligible to participate, 120 were enrolled and 119 were randomized (63 intervention and 56 controls). Refusals to participate were because of lack of time for monthly visits (20, 43.5%), fear of using the device around other people (18, 39.1%), belief that the device was inconvenient to carry (16, 34.8%), living far from the clinic, making clinic visits inconvenient (11, 23.9%), and lack of concern about adherence (5, 10.9%). Of the 120 enrolled, 1 dropped out before randomization, 3 more dropped out postrandomization, 1 intervention subject, and 2 controls (Fig. 1). Of these three, 1 subject completed 6 months of the intervention, 1 completed 5 months, and the third completed 3 months. A total of 116 completed the 6-month intervention period: 62 intervention subjects and 54 controls.
Randomized subjects were primarily male (63.9%) and mean age was 38 years (Table 1). About one-half were married. Most (58.0%) had a middle school education; just over one-half (55.5%) were employed, with mean monthly income of approximately 3000 yuan (US $500). Preintervention CD4-cell counts were 389 vs. 363 cells per microliter in intervention and control subjects, respectively. Fewer intervention than control subjects had UDVL at baseline (74.6% vs. 98.2%, P < 0.001). Mean time on ART was just over 30 months, and similar in both arms, with most on twice-daily regimens (62% in intervention vs. 79% in controls, P = 0.049). Only 10 (8.4% overall) subjects were treatment naive, defined as less than 1 month on ART; all 10 were in the intervention arm. Sexual transmission was the principal infection route. Among men, 43% of intervention subjects vs. 18% of controls reported infection through unprotected sex with another man (data not in table; P = 0.020).
Effect of Real-time Feedback on Adherence
At randomization, the proportion with ≥95% on-time adherence during the 3 months before randomization was 40/63 (63.5%) and 33/56 (58.9%) in intervention vs. control subjects (P = 0.611) (Table 2). In month 9, 6 months after the start of the intervention, a higher proportion of intervention subjects had ≥95% on-time adherence [55/63 (87.3%)] than controls, among whom adherence fell slightly [29/56 (51.8%)] (RR for optimal adherence in month 9, intervention vs. control, 1.69; CI: 1.29 to 2.21, P < 0.001). Analysis of adherence during the entire intervention period found that the proportion of subjects who achieved ≥95% on-time adherence over months 4–9 was similar: 52/63 (82.5%) and 29/56 (51.8%) for intervention vs. control subjects, respectively (RR: 1.59; CI: 1.21 to 2.10, P < 0.001). Secondary analyses found no significant effect of variables imbalanced at randomization on any of the adherence outcomes.
The beneficial effect of the intervention remained significant when stratified by whether subjects had optimal vs. suboptimal adherence at baseline (during the prerandomization period). Among suboptimal adherers, optimal adherence in month 9 was 18/23 (78.3%) vs. 7/21 (33.3%), respectively (RR: 2.35; CI: 1.24 to 4.46, P = 0.003). Among optimal adherers at baseline, the proportions were 37/40 (92.5%) vs. 22/35 (62.9%) (RR: 1.47; CI: 1.12 to 1.93, P = 0.002), respectively (Table 2).
Mean adherence rates also improved in intervention subjects (Table 2). Adherence was similar at randomization (month 3), 91.6% and 91.5% in intervention and control subjects, respectively (P = 0.970), but was higher in intervention subjects than controls in month 9: 96.2% vs. 89.1% (P = 0.003), respectively. Similarly, mean adherence during the entire 6-month intervention period was significantly higher in intervention vs. control arm: 96.3% vs. 88.9% (P < 0.001).
In the stratified analysis among those with suboptimal adherence at baseline, mean adherence in intervention vs. control subjects was 93.3% vs. 84.7% (P = 0.039), respectively. Among previously optimal adherers, rates were 97.8% vs. 91.7% (P = 0.028), respectively. Mean monthly adherence rates were higher in the intervention in every month, in both adherence groups (Fig. 2).
Effect of Real-time Feedback on Markers of HIV Disease Progression
Compared with controls, intervention subjects had similar postintervention CD4-cell counts and rates of UDVL. At baseline, mean CD4-cell counts were 389 and 363 cells per microliter, in intervention and control subjects, respectively (P = 0.408). These counts improved to 445 vs. 391 cells per microliter by month 9 in both groups (P = 0.080). The mean change in CD4-cell count between baseline and month 9 trended higher but was not significantly different in intervention subjects vs. controls, an average gain of 52 vs. 28 cells per microliter (P = 0.297). The proportion of intervention subjects who achieved UDVL increased significantly (P = 0.004) between baseline and month 9, but proportions were similar between intervention vs. control subjects at month 9, 93.6% vs. 98.2%, respectively (P = 0.218).
Analysis of Reminder Messages
During the preintervention period, the mean monthly number of delayed doses (not taken by 30 minutes after scheduled dose time) was 3.3 vs. 3.5 in intervention vs. control arms (P = 0.825). Among intervention subjects, this number declined to 2.4 during the intervention period; the number increased among controls to 4.6 (P = 0.036). Among those subjects who had delayed doses (N = 100, both preintervention and intervention periods), before the intervention, intervention subjects took 46.6% of delayed doses within the next 30 minutes (30–60 minutes after dose time, “on time” according to the adherence measure), compared with 56.9% among controls (P < 0.001) (Fig. 3). During the intervention period, this proportion rose substantially in the intervention group but dropped slightly in controls. In intervention subjects, the increase was 30 percentage points, with 76.1% of delayed doses taken on time, contrasted with a 2 percentage-point drop to 54.6% of delayed doses taken on time among control subjects (P < 0.001).
The use of triggered cell phone reminders and enhanced counseling based on objective adherence data from the Wisepill monitor significantly improved ART adherence in this population of HIV-infected patients. Although scheduled reminders delivered to mobile phones have been shown to improve adherence,23 this is the first study to demonstrate the impact on adherence of triggered reminders sent only when patient behavior suggests less-than-perfect adherence. This finding adds to the growing evidence regarding the potential of wireless technologies generally as an adherence tool, while highlighting the unique benefit of “smart messages”—reminders that communicate in real time with patients based on pill-taking actions, allowing them to quickly adjust their behavior to improve adherence.
This result builds on and confirms both our previous work and that of others indicating that effective ART adherence interventions should be individualized.14,38 Several features of this intervention were personalized: patients selected their own reminder messages, a reminder was sent only when prompting appeared necessary given the patient's medication-taking behavior, and each counseling session was informed by the patient's individual adherence data.
As in AFL and other studies, the intervention was most helpful among the subjects whose adherence was the lowest at randomization13 and thus had the greatest potential for improvement. Nonetheless, we also observed a benefit in those with optimal adherence at randomization. Among these subjects, the apparent benefit was to prevent adherence from declining over time, a common occurrence in chronic disease management,51–53 including in HIV.54 This suggests that even patients with high adherence may benefit from real-time adherence monitoring and support.
Our findings also address the concern as to whether reminders could paradoxically train patients to take their medicines only in response to a reminder, which might leave patients vulnerable in the event of device failures or loss of cellular connectivity. Our finding that the proportion of doses taken before the 30-minute mark increased markedly during the intervention period mitigates this concern. This suggests that the intervention improved participants' dose-taking self-management behavior in advance of a reminder. This result should be contrasted with the experience described by Pop-Eleches et al22 in their work on scheduled cell phone reminders on ART adherence, in which the finding was that daily messages were no more effective than no-reminders, whereas weekly messages were significantly beneficial for supporting adherence. At the same time, it is worth highlighting that these previous reminder interventions have all used prescheduled messages, in contrast to ours, which delivered reminders based on actual behavior. One might speculate that consistent daily messages became routine, and ignorable, or possibly even an irritant to subjects (ie, SPAM). If so, then triggered reminders as in this study may be more effective, as excellent adherence leads to relief from reminders, which may have a motivating effect.
Adherence interventions based on detailed theory-driven behavioral counseling methods can be difficult to implement and to scale up for delivery to large numbers of ART users. In contrast, the wireless monitoring and text reminder intervention is relatively simple and can be used as a tool by providers and adherence counselors already in the field. The potential for broad scalability may make it feasible to target patients specifically known to be poorly adherent or those who develop drug resistance.
The ultimate goal of any ART adherence intervention is to improve HIV viral suppression to prevent disease progression, drug resistance, and HIV transmission. This study was not designed to detect meaningful differences in HIV RNA suppression or CD4-cell count response, and although we would expect adherence changes to ultimately effect biological outcomes, we were unable to show this. Although we observed a large increase in proportion with viral suppression in the intervention arm, postintervention proportions of UDVL were similar, due in part to the disproportionately high level of suppression among controls at baseline. The chief explanation for this combination of results is that our relatively treatment-experienced subject population turned out to be doing well in terms of adherence, CD4-cell counts, and levels of viral suppression. Given the relevance of clinical markers, we recommend that future research use larger sample sizes and target patients particularly at risk for biological failure, such as those initiating therapy for the first time or beginning a second regimen after an initial regimen failure.
We acknowledge several study limitations. First, subjects and clinicians were not blinded, and thus some bias may have affected counseling provided to control subjects. However, blinding was not possible in an intervention of this kind because it is impossible to conceal reminders. Although clinicians and patients were unblinded, all analyses were conducted without knowledge of intervention assignment. Second, the study had a relatively short duration of follow-up. Six months may be too short a time to know whether subjects may become habituated to the intervention so that it loses potency over time; changes in UDVL and CD4-cell count can be delayed by up to 2 years from suboptimal adherence.55,56 Third, the study design did not permit a rigorous analysis of the individual contributions of cell phone reminders vs. enhanced counseling. That said, our analysis of the dose taking relative to delivery of a reminder suggests that triggered reminders were highly efficacious, which by design reduced the counseling sessions required by poor adherence. Fourth, the study was not designed to measure an impact on biological end points. To do so, a larger cohort, ideally with low rates of UDVL at baseline, will need to be enrolled and followed for a longer period.
Despite these limitations, this study highlights the potential of real-time feedback in the search for effective adherence promotion strategies. We conclude that “smart reminders” that are sent to patients only when their behavior suggests a need for reminding is a promising approach in the management of HIV and other chronic diseases. We recommend further assessment and adaptation in other patient settings.
The authors thank the late Richard Denisco and Dr. Tang Zhirong, Deirdre Pierotti and Sherley Brice at Boston University, former program manager Bram Brooks, Katherine Semrau of Harvard Medical School and Brigham and Women's Hospital, and Mark Harrold and Evan Hecht. The authors appreciate the support of experts at the US CDC-GAP office and the National Center for AIDS/STD Control and Prevention (NCAIDS), Chinese Center for Disease Control and Prevention, in Beijing.
1. UNAIDS. Global Report: UNAIDS Report on the Global AIDS Epidemic 2013. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS; 2013.
2. Garcia de Olalla P, Knobel H, Carmona A, et al.. Impact of adherence and highly active antiretroviral therapy on survival in HIV-infected patients. J Acquir Immune Defic Syndr. 2002;30:105–110.
3. Bangsberg DR, Perry S, Charlebois ED, et al.. Non-adherence to highly active antiretroviral therapy predicts progression to AIDS. AIDS. 2001;15:1181–1183.
4. Hogg RS, Heath K, Bangsberg D, et al.. Intermittent use of triple-combination therapy is predictive of mortality at baseline and after 1 year of follow-up. AIDS. 2002;16:1051–1058.
5. Nachega JB, Hislop M, Dowdy DW, et al.. Adherence to highly active antiretroviral therapy assessed by pharmacy claims predicts survival in HIV-infected South African adults. J Acquir Immune Defic Syndr. 2006;43:78–84.
6. Colson P, Ravaux I, Yahi N, et al.. Transmission of HIV-1 variants resistant to the three classes of antiretroviral agents: implications for HIV therapy in primary infection. AIDS. 2002;16:507–509.
7. Yerly S, Kaiser L, Race E, et al.. Transmission of antiretroviral-drug-resistant HIV-1 variants. Lancet. 1999;354:729–733.
8. Siegrist CA, Yerly S, Kaiser L, et al.. Mother to child transmission of zidovudine-resistant HIV-1. Lancet. 1994;344:1771–1772.
9. Bangsberg DR. Less than 95% adherence to nonnucleoside reverse-transcriptase inhibitor therapy can lead to viral suppression. Clin Infect Dis. 2006;43:939–941.
10. Kobin AB, Sheth NU. Levels of adherence required for virologic suppression among newer antiretroviral medications. Ann Pharmacother. 2011;45:372–379.
11. Viswanathan S, Detels R, Mehta SH, et al.. Level of adherence and HIV RNA suppression in the current era of highly active antiretroviral therapy (HAART). AIDS Behav. 2015;19:601–611.
12. Amico KR, Harman JJ, Johnson BT. Efficacy of antiretroviral therapy adherence interventions: a research synthesis of trials, 1996 to 2004. J Acquir Immune Defic Syndr. 2006;41:285–297.
13. Simoni JM, Pearson CR, Pantalone DW, et al.. Efficacy of interventions in improving highly active antiretroviral therapy adherence and HIV-1 RNA viral load. A meta-analytic review of randomized controlled trials. J Acquir Immune Defic Syndr. 2006;43(suppl 1):S23–S35.
14. Simoni JM, Amico KR, Smith L, et al.. Antiretroviral adherence interventions: translating research findings to the real world clinic. Curr HIV/AIDS Rep. 2010;7:44–51.
15. Barnighausen T, Chaiyachati K, Chimbindi N, et al.. Interventions to increase antiretroviral adherence in sub-Saharan Africa: a systematic review of evaluation studies. Lancet Infect Dis. 2011;11:942–951.
16. Chaiyachati KH, Ogbuoji O, Price M, et al.. Interventions to improve adherence to antiretroviral therapy: a rapid systematic review. AIDS. 2014;28(suppl 2):S187–S204.
17. Mbuagbaw L, Ongolo-Zogo P, Thabane L. Investigating community ownership of a text message programme to improve adherence to antiretroviral therapy and provider-client communication: a mixed methods research protocol. BMJ Open. 2013;3:e002816.
18. Mbuagbaw L, van der Kop ML, Lester RT, et al.. Mobile phone text messages for improving adherence to antiretroviral therapy (ART): an individual patient data meta-analysis of randomised trials. BMJ Open. 2013;3:e003950.
19. Lester RT, Ritvo P, Mills EJ, et al.. Effects of a mobile phone short message service on antiretroviral treatment adherence in Kenya (WelTel Kenya1): a randomised trial. Lancet. 2010;376:1838–1845.
20. Curioso WH, Zuistberg DA, Cabello R, et al.. “It's time for your life”: how should we remind patients to take medicatines using short text messages? AMIA Annu Symp Proc. 2009:129–133.
21. Gatwood J, Balkrishnan R, Erickson SR, et al.. Addressing medication nonadherence by mobile phone: development and delivery of tailored messages. Res Social Adm Pharm. 2014;10:809–823.
22. Pop-Eleches C, Thirumurthy H, Habyarimana JP, et al.. Mobile phone technologies improve adherence to antiretroviral treatment in a resource-limited setting: a randomized controlled trial of text message reminders. AIDS. 25:825–834.
23. Finitsis DJ, Pellowski JA, Johnson BT. Text message intervention designs to promote adherence to antiretroviral therapy (ART): a meta-analysis of randomized controlled trials. PLoS One. 2014;9:e88166.
24. Ministry of Health of the People's Republic of China. 2011 nian Zhongguo Aizibing Yiqing Guji (2011 China AIDS Epidemic Appraisal and Estimates). Beijing, China; 2011.
25. Ministry of Health of the People's Republic of China. 2012 China AIDS Response Progress Report. Beijing, China; 2012.
26. Mascolini M. New HIV Rate in China Climbs over 10% in First 10 Months of 2012. International AIDS Society (webpage). December 3, 2012.
27. Tatlow DK. H.I.V. Infections among young chinese rising. New York Times, Sinosphere: Dispatches from China. New York City, NY; December 3, 2013.
28. China CDC. 2014 nian 5 yue quanguo aizibing xingbing yiqing ji zhuyao fangzhi gongzuo jinzhang (Update on thh AIDS/STD epidemic in China and main response in control and prevention in May, 2014). Beijing, China; 2014.
29. Wu Z, Rou K, Cui H. The HIV/AIDS epidemic in China: history, current strategies and future challenges. AIDS Educ Prev. 2004;16(3 suppl A):7–17.
30. Cui Y, Liau A, Wu ZY. An overview of the history of epidemic of and response to HIV/AIDS in China: achievements and challenges. Chin Med J (Engl). 2009;122:2251–2257.
31. He N, Detels R. The HIV epidemic in China: history, response, and challenge. Cell Res. 2005;15:825–832.
32. Zhu H, Napravnik S, Eron JJ, et al.. Decreasing excess mortality of HIV-infected patients initiating antiretroviral therapy: comparison with mortality in general population in China, 2003-2009. J Acquir Immune Defic Syndr. 2013;63:e150–e157.
33. Zhang F, Dou Z, Ma Y, et al.. Effect of earlier initiation of antiretroviral treatment and increased treatment coverage on HIV-related mortality in China: a national observational cohort study. Lancet Infect Dis. 2011;11:516–524.
34. Wu Z, Sullivan SG, Wang Y, et al.. Evolution of China's response to HIV/AIDS. Lancet. 2007;369:679–690.
35. Xiao Y, Kristensen S, Sun J, et al.. Expansion of HIV/AIDS in China: lessons from Yunnan Province. Soc Sci Med. 2007;64:665–675.
36. Xing H, Ruan Y, Li J, et al.. HIV drug resistance and its impact on antiretroviral therapy in Chinese HIV-infected patients. PLoS One. 2013;8:e54917.
37. Simoni JM, Chen WT, Huh D, et al.. A preliminary randomized controlled trial of a nurse-delivered medication adherence intervention among HIV-positive outpatients initiating antiretroviral therapy in Beijing, China. AIDS Behav. 2011;15:919–929.
38. Sabin LL, Desilva MB, Hamer DH, et al.. Using electronic drug monitor feedback to improve adherence to antiretroviral therapy among HIV-positive patients in China. AIDS Behav. 2010;14:580–589.
40. Bachman DeSilva M, Gifford A, Xu K, et al.. Feasibility and acceptability of a real-time adherence monitoring device among HIV-positive patients in China. Presented at: 8th International Conference on HIV Treatment and Prevention Adherence; June 2–4, 2013; Miami, FL.
41. Haberer JE, Kahane J, Kigozi I, et al.. Real-time adherence monitoring for HIV antiretroviral therapy. AIDS Behav. 2010;14:1340–1346.
42. Haberer JE, Kiwanuka J, Nansera D, et al.. Realtime adherence monitoring of antiretroviral therapy among HIV-infected adults and children in rural Uganda. AIDS. 2013;27:2166–2168.
43. Haberer J. Adherence monitoring: state of the science and future Innovations. Presented at: 9th International Conference on HIV Treatment and Prevention Adherence; June 8–10, 2014; Miami, FL.
44. Vervloet M, van Dijk L, Santen-Reestman J, et al.. Improving medication adherence in diabetes type 2 patients through real time medication monitoring: a randomised controlled trial to evaluate the effect of monitoring patients' medication use combined with short message service (SMS) reminders. BMC Health Serv Res. 2011;11:5.
45. China Ministry of Health, UNAIDS, WHO. 2011 China AIDS Epidemic Situation and Estimates (in Chinese). Beijing, China; 2011.
46. Wang Y, Tang Z, Zhu Q, et al.. Analysis of features of AIDS transmission in guangxi 2009-2011 (in Chinese). South China J Preventative Med. 2013;1:6–11.
47. China National Bureau of Statistics (web). Available at: http://www.citypopulation.de/php/china-admin.php?adm2id=4501
. Accessed July 28, 2014.
48. Gill CJ, Sabin LL, Hamer DH, et al.. Importance of dose timing to achieving undetectable viral loads. AIDS Behav. 2010;14:785–793.
49. Parienti JJ, Barrail-Tran A, Duval X, et al.. Adherence profiles and therapeutic responses of treatment-naive HIV-infected patients starting boosted atazanavir-based therapy in the ANRS 134-COPHAR 3 trial. Antimicrob Agents Chemother. 2013;57:2265–2271.
50. Gill CJ, Sabin LL, Hamer DH, et al.. Importance of dose timing to achieving undetectable viral loads. AIDS Behav. 2009;14:785–793.
51. Howard AA, Arnsten JH, Lo Y, et al.. A prospective study of adherence and viral load in a large multi-center cohort of HIV-infected women. AIDS. 2002;16:2175–2182.
52. Pearson CR, Simoni JM, Hoff P, et al.. Assessing antiretroviral adherence via electronic drug monitoring and self-report: an examination of key methodological issues. AIDS Behav. 2007;11:161–173.
53. Etard JF, Laniece I, Fall MB, et al.. A 84-month follow up of adherence to HAART in a cohort of adult Senegalese patients. Trop Med Int Health. 2007;12:1191–1198.
54. Remien RH, Stirratt MJ, Dolezal C, et al.. Couple-focused support to improve HIV medication adherence: a randomized controlled trial. AIDS. 2005;19:807–814.
55. Wools-Kaloustian K, Kimaiyo S, Diero L, et al.. Viability and effectiveness of large-scale HIV treatment initiatives in sub-Saharan Africa: experience from western Kenya. AIDS. 2006;20:41–48.
56. Nachega JB, Hislop M, Dowdy DW, et al.. Adherence to nonnucleoside reverse transcriptase inhibitor-based HIV therapy and virologic outcomes. Ann Intern Med. 2007;146:564–573.