Please see Additional file 1 for translation for the abstract into the six official working languages of the United Nations.

Non-attendance at outpatient clinics can lead to poor health outcomes and high costs. There are healthcare- and patient-related factors that can lead to appointments being missed. Healthcare-related factors include poor communication, duration of time between appointments and poor facilities in waiting areas, such as lack of confidentiality [1–3]. Patient-related factors include health beliefs, difficulty with transportation and forgetfulness [4–6]. Any form of reminder may, therefore, reduce the rate of missed appointments.

Mobile technology is frequently used as a tool to improve adherence and treatment outcomes in healthcare settings. It has been used to increase knowledge [7], improve health outcomes [8–11], encourage behaviour change [12] and improve collaboration between healthcare workers in resource-limited settings [13]. A recent Cochrane review considered the impact of text messaging on outpatient attendance [14]. Four randomised control studies were included, all set in high- and middle-income countries: two were set in the UK [15],[16], one in Malaysia [17] and the other in China [18]. The review found that text message reminders improved attendance compared with no reminders (relative risk [RR] = 1.10, 95% confidence interval [CI] 1.03-1.17) and was no better or worse than a phone call. It is likely that factors leading to attendance in clinics will be different in high-income countries as opposed to low- and middle-income countries due to differences in health services provided and general infrastructure. Leong et al. studied mobile phone reminders compared with phone calls in primary care in Malaysia, which found that for every nine text message reminders sent, there was one additional attendance [17]. The text messaging reminder system costs less than half of the mobile phone reminder per attendance. Chen et al. also found similar attendance rates for text messages and phone calls, although text messages were significantly less expensive [18]. None of the studies investigated the reason for non-attendance and all studies stated the importance of high levels of mobile phone ownership.

One randomised control trial (RCT) has been published investigating the impact of mobile phone reminders on outpatient attendance in Kenya; no other RCTs have been identified from low-income countries. Odeny et al. investigated the use of text messaging to improve attendance at post-operative clinic visits after adult male circumcision for HIV prevention [19]. Patients were more likely to return for follow-up after text message reminders than those who did not receive reminders (RR = 1.09, 95% CI 1.00–1.20; p = 0.04). The study also identified that the higher cost of travel to the clinic led to reduced attendance.

Crankshaw et al. and Kunutsor et al. found that mobile phone ownership is between 60-80% in low-income countries, with patients being willing to be contacted verbally or by text message [20],[21]. Therefore, phone reminders may be feasible in these settings. In addition, staff working in resource-poor settings broadly support the use of mobile phone technology [22],[23]. Although there is some evidence of the use of mobile phones to improve health (mobile Health or mHealth) in resource-poor settings, there are few evaluations of the impact of mobile technology on health, with systematic reviewers calling for further evidence on this [10],[14],[24],[25].

These studies suggest that mobile technology may be a useful tool to improve attendance at outpatient service clinics. However, in resource-limited settings, the cost of text messaging may be a barrier to implementation. Text messages in Swaziland, as in the rest of the world, are relatively cheap. In Swaziland, a single text message costs 0.80SZL (£0.06, 1 GBP = 13.99 SZL, 30/4/2013). Although this is a cheap intervention, funding within Swaziland’s health system, as in many developing countries, is limited and often intermittent [26]. Health funding is often found through external donor sources, although these are also not reliable as they are usually associated with larger programmes and may not cover costs of mobile technology. During the course of this study, hospital departments frequently ran out and were not able to replenish ‘airtime’ (mobile phone credit) to conduct follow-up. This outlines the benefit of an intervention that will continue to work regardless whether there is regular funding.

A no-cost alternative to text messaging is using a missed call to act as a reminder. This practice is common in the general population in Africa, South Asia and South America as it is free for both the caller and the receiver. It has different names in different countries, including ‘beeping’ , ‘missed calling’ , ‘flashing’ , ‘pranking’ , ‘lost calling’ and ‘fishing’ [27],[28]. In Swaziland, this practice is called ‘buzzing’ and this is the term that will be used throughout this paper.

Swaziland is facing a severe public health crisis with the highest estimated national prevalence of human immunodeficiency virus (HIV) at 26% among adults aged 15–49, and tuberculosis (TB) incidence rates at 1,287 TB cases/100,000/year, with HIV/TB co-infection adding a particular challenge [29],[30]. This study was based in the regional hospital for the rural region of Lubombo, Swaziland. Diagnosis and management of HIV in the region is conducted by the regional hospital and community-based clinics spread throughout the region. It is coordinated and supported by the Communities Department based at the regional hospital, in conjunction with a number of non-government organisations (NGOs).

Although access to treatment for HIV is improving, there are still considerable challenges in the delivery of services, in particular in relation to follow-up of HIV patients. Within the regional hospital, around 20% of patients who test for HIV fail to collect results and are lost to follow-up. Within the regional hospital, the majority of patients are tested within the HIV testing and counselling department, although some are tested within the maternal and child health clinic, the TB clinic or as an inpatient.

In an attempt to increase the number of patients returning to collect blood results after initial diagnosis with HIV, a mobile phone reminder system using buzzing technology was implemented. A specialised phone which sends automated buzzes to patients the day before appointments was supplied to the HIV testing and counselling department. The aim of this study was to measure the feasibility and effectiveness of buzzing patients as a mobile phone reminder on follow-up attendance to collect blood results (CD4) in the HIV testing and counselling department of a rural hospital in Swaziland.

When analysed using an intention to treat analysis, including all cases with or without mobile phones, 82.97% (346/417) collected CD4 results before the intervention, and 82.43% (183/222) collected the results after the intervention, was implemented (p = 0.863).

We found no significant difference in the rate of attendance of patients for their CD4 results after the introduction of no-cost mobile phone reminders. One hypothesis was that forgetfulness was the main reason for non-attendance, which is in line with the literature on the use of mobile phone reminders for appointments [6],[31–36], and therefore phone reminders would help with this. However, if the assumption is that forgetfulness is a key factor in non-attendance, the short timeframe in which patients are expected to come back for appointments may have limited the measurable benefit of a reminder system. The average number of days between the two appointments is four to six days and, therefore, the reason for non-attendance may not be failing to remember to attend.

Patients who collected CD4 results tended to be older, male, live closer to the hospital and be eligible for initiation of ART. Younger age has previously been suggested as a risk factor for non-attendance [37]. Generally, male sex is associated with poorer attendance [37] and, therefore, the finding in this study may be related to cultural beliefs. Patients who live further from the hospital have been found not to attend their appointments, in line with existing literature [4],[19],[33]. This is particularly likely to impact patients in rural Swaziland as some patients were required to travel distances over 100 km to reach the clinic. Patients who were eligible for initiation of ART were also more likely to attend their appointments. This may be because patients were symptomatic and therefore concerned to find out their results. Additional studies may be required to further understand reasons for non-attendance in this setting as information on patients’ reasons for non-attendance was not collected during this study and is thus a limitation of the study. A further limitation of the study is that it does not investigate patients’ prior knowledge, experience and expectations of healthcare, and HIV counselling and testing in particular, which would be helpful to understand reasons for non-attendance.

As was reported in the cited literature from resource-limited settings [20],[21], mobile phone ownership is high in Swaziland, with over 70% of patients owning a mobile phone. In addition, all patients who owned a mobile phone consented to enter the study and receive the reminder buzz. This shows that being sent a reminder message was acceptable to all patients. A large number of returning patients reported to staff at the HIV testing and counselling department that they were grateful for the reminder. Staff at the HIV testing and counselling service found the technology easy to use and reported that they would be happy to continue using the service at the completion of the study. This suggests that the use of buzzing is feasible and acceptable in this setting.

There were a number of differences in the before and after groups, which are largely due to the operational nature of this research being conducted in a real-world setting against a background of a rapidly changing national emergency. The number of patients testing positive before the implementation of the intervention was higher than after the intervention was in place. This is likely to be due to changes in the way services were being delivered at the hospital, with an emphasis on decentralisation of HIV testing and counselling, and integrating testing into general services such as outpatients, the TB department, and the maternal and child health department. During the course of this study, a new HIV counsellor was recruited to work within the outpatient department and this is likely to have decreased the number of patients attending the stand-alone HIV testing and counselling department. There were more patients with unknown addresses after the intervention compared with before, 16% (26) versus 1% (3), preventing the calculation of their distance from hospital in these patients. This was due to the adoption of a new data management system for HIV services in the hospital soon after the implementation of the intervention. There were a number of concerns about data completeness soon after the adoption of the data system, which were rectified quickly but led to gaps in data collection.

The mean number of days taken to collect CD4 was higher in the group that received the intervention compared with the before group. There is no clear explanation of why this may be, but it may be due to chance as the study is not powered to detect a significant difference in the secondary outcome. As this is a before-and-after study, this difference may also be due to external factors such as the changes in the economic climate, which has been deteriorating within Swaziland and has led to public sector and transport strikes perhaps providing a barrier to access and leading to difficulties paying for travel.

The sample size required to power the study to detect a 9% difference was not reached due to the limited numbers of newly-diagnosed HIV positive individuals. With 297 patients in the before group, 162 in the after group, an alpha of 0.05 and a power of 0.8, the study was powered to detect a 10% increase in attendance from 80% to 90%, not quite the 9% difference that was aimed for.

This study failed to detect any difference in attendance after providing mobile phone reminders using buzzing. However, it does suggest that using mobile phone technology in rural Swaziland may be feasible and acceptable. The main barrier to using mobile phone technology can be cost and, therefore, using a free service such as buzzing may be an affordable alternative to phone calls or text messaging. Further study in chronic care settings and as part of adherence packages is planned to investigate if buzzing can improve attendance and adherence in resource-limited settings.