These results highlight the challenges that LHWs experience in extending the services, usually provided by health professionals at healthcare facilities, to the homes of patients [5, 6]. The high proportion of LHWs’ time that was categorised as ‘non-contact time’ should not be seen as indicating poor productivity on the part of LHWs. Rather, these data reflect the difficulties of finding patients’ homes in settings with no formal street addresses as well as the likelihood of finding patients at home during working hours.
Our findings are similar to those reported by Mallidou et al.  and Qian et al.  in their studies of non-professional carers in health care facilities. In these two studies, only 52% and 31% of the observed time respectively were spent in direct contact-care with patients. Non-contact time, whether travelling to a patient’s home, waiting for patients or undertaking administrative duties, is an inevitable part of LHWs’ daily activities in most settings. In community settings, assigning patients to LHWs based on their home’s proximity to the LHW’s home may reduce walking time. However, there are cases where patients express a preference for a particular LHW who does not live close by. An example from Site 3 in this study illustrates this challenge: here the walking time to a ‘patient-preference home’ was 70 minutes, resulting in approximately one quarter of the LHW’s day being spent on one patient. These detailed data on how LHWs organize and structure their time could contribute to the development of wider explanations of how LHW programmes impact on health outcomes, including TB and ART outcomes, and the factors affecting the implementation of LHW programmes at scale [5, 28]. The challenges encountered by LHWs may also contribute to poor retention rates for this cadre and may, in some settings, also impact on recruitment.
This study found that LHWs conduct an average of six visits per day, close to the 5.6 visits per day in urban settings that has been recommended by the National Department of Health, as part of the re-engineering of primary health care in South Africa . However, a ‘once size fits all’ approach for case-loads across different neighbourhoods may not be useful as this does not take geographical and logistical differences into account. This is illustrated by the differences between the three sites (Table 5): the walking time in high density communities such as informal settlements (Sites 1 and 2) was substantially lower than that in formal, lower density urban areas (Site 3), where homes are further apart. Local estimates of travelling, waiting and visit times should be used to determine the ideal number of patients to assign to each LHW.
Our results emphasize the need for careful and considered human resources management for LHWs [2, 28]. Appropriate job descriptions and performance assessment criteria are important tools for increasing the effectiveness of LHW programmes , but need to be based on the realities of LHW’s work. Our findings support Vale’s  view that evidence of LHWs’ experiences and working conditions is needed to counter unrealistic employment conditions and expectations from their employers. Data from T&M studies such as this one should prompt agencies that employ and support LHWs to consider strategies to optimize the delivery of LHW services. The following approaches could mitigate some of the challenges recorded in this study:
Match LHWs and patients based on smaller geographical areas so as to reduce travel time for LHWs
Establish systems to identify and prioritize ‘at-risk’ patients who may benefit from more intensive LHW support, so as to keep the case-load manageable for each LHW
Explore the use of mobile phones to improve communication between LHWs and patients, including around the scheduling of support visits, so as to reduce the number of unnecessary home visits that LHWs make 
Explore the use of mobile phone-based GIS systems to mark the location of patients’ home so that they can easily be found by the health services in the future 
This study has several limitations. The small scale of the research means that the findings should be generalized with caution to other settings and programmes. Also, the proportion of patients found at home in Sites 2 and 3 might be higher than that is seen typically, as the LHWs needed to obtain patients’ consent in advance for the researcher to visit. The true proportion of patients found at home in relation to visits done may therefore be lower than reported here. Finally, it is possible that the LHWs changed their behaviours and practices as a consequence of being observed. For example, LHWs may have wanted to demonstrate to the researchers their challenging working conditions and therefore chosen, on the observation day, to visit problematic patients who lived furthest from the clinic. However, all of the patients visited had been assigned to the LHWs for treatment support, and the data is therefore likely to reflect how LHWs spend their time on a day-to-day basis. Also, while independent observations might shape the behaviours of those observed, this approach counters the tendency of participants to overreport activities that they view as more desirable [15, 16].