No observed reduction of non-attendance rate in telehealth models of care

A patient failing to attend a scheduled appointment is an undesirable occurrence for a health service provider. Valuable health resources are underutilised and efforts to reduce clinic waiting lists are hindered. In the search for solutions to this problem, it is often reported that telehealth models of care can reduce rates of non-attendance.^1,2 This claim is easily accepted because it is intuitive that, by reducing the burden of travel, attending is easier and a patient is more likely to make their appointment.

In the 2017–18 financial year, Queensland Health delivered 52 743 specialist out-patient appointments using telehealth models of care. We considered this a sufficiently large volume to compare rates of attendance of these appointments with traditional (in person) events and to determine whether rates of non-attendance differed between these two delivery methods.

An extract of all specialist out-patient appointments reported in Queensland Health’s corporate patient administration systems between 1 July 2017 and 30 June 2018 was obtained (n = 2 921 702). Variables collected included hospital location, patient residential address (suburb and postcode only), a unique identifier and clinic descriptions; in addition, how the service was delivered and whether the patient attended were captured for each event. Services that were delivered through methods other than in person or telehealth (i.e. telephone) were considered out of scope and removed from the study. Appointments that were rescheduled or cancelled were omitted to focus the study on whether an appointment was attended or not. Patients reporting a residential address outside of Queensland were also removed because the transient nature of this group can skew inquiries of non-attendance. A comparison of patient demographic information by delivery method is presented in Table 1.

Table 1.  Comparison of patient demographic information (n = 2 921 702)
FTA, failed to attend; MMM, modified Monash model

Attendance averages by delivery method for the 2017–18 financial year, including a comparison of in-person and telehealth attendance rates for the 10 largest telehealth specialties are shown in Fig. 1. A similar comparison, grouped by the remoteness index of the patient’s residential address, is shown in Fig. 2. In Queensland during the 2017–18 financial year, the probability of non-attendance by a patient with a scheduled telehealth appointment was 15% higher than for a patient with an equivalent in-person appointment ( = 113.56, P < 0.001, relative risk = 1.15).

Fig. 1.  Comparison of in-person and telehealth attendance rates for the top 10 specialties by telehealth activity (2017–18).

Fig. 2.  Comparison of in-person and telehealth attendance rates by remoteness index, where 1 is a major city and 7 is very remote (2017–18). MMM, modified Monash model.

Reports of a reduction in non-attendance attributed to telehealth models of care^1,2 are unsubstantiated by the dataset used in this study. Previous investigations of non-attendance converged on patient behaviour, such as forgetting or being confused with appointment details, as the most common explanation.^3–14 Telehealth models of care reduce the distance a patient is required to travel, but they do not remove these potential points of failure, as the results of this study appear to confirm.

Interventions demonstrating success in reducing non-attendance have typically involved improving interactions with patients, such as appointment reminder services, simplified cancellation or rescheduling processes and reducing the practice of booking appointments far in advance.^15–17 Evidence that reducing travel requirements improves attendance is limited to isolated studies.^18,19 In these examples, additional factors, such as increased promotion, communication and supplementing telehealth clinics with the provision of services in more convenient, non-clinical settings, are likely to have been greater contributors to the observations. The surprising results of a recent study in which free rideshare trips were provided to a largely urban-based population in order to make attendance at a primary health clinic easier were that participants were no less likely to increase attendance than those who arranged the journey themselves.²⁰ One explanation for this is that travel represented a relatively minor contributor to non-attendance among the participants of that study.

When a patient is required to attend a medical appointment, a period of time beforehand is required to travel to the facility providing the treatment. In the case of patients living in rural and remote locations, this time could amount to many hours, if not several days. When a patient receives a health service using a telehealth model of care at their local facility as an alternative to travelling further to the hospital where their specialist is located, the time needed for the patient to prepare for the appointment is reduced. Service providers have time that would have otherwise been lost during travel to confirm appointment details with patients and avoid situations where this reminder occurs when they are simply too far away to make their appointment on time. This may provide an opportunity for telehealth models of care to contribute to efforts in reducing rates of non-attendance. Where communication with patients is insufficient or unclear, telehealth models of care cannot compensate and are no less likely to contribute to rates of non-attendance by patients.

Several patterns were noted in this study, including: (1) the mean age of non-attendances for both in-person and telehealth appointments was lower than the mean age of those attending appointments; and (2) variation of attendance rates between different specialities is greater than the difference in attendance between in-person and telehealth appointments for the same speciality. Further investigation of these observations may be useful in better understanding non-attendance and developing strategies to reduce their effects.

Large variations in travel requirements to receive health services exist in Queensland. Despite this, no correlation between travel reduction and reductions in non-attendance was observed in this study. Non-attendance may be a poor measure of travel burden, as plausible as it may seem. This finding supports most available evidence that human factors, including forgetfulness, confusion or anxiety towards a scheduled appointment, and not travel requirements, contribute most significantly to rates of non-attendance. It is likely that telehealth models of care have a role to play in reducing non-attendance but, in the absence of clear communication, timely and appropriate appointment reminders and the ability to easily cancel or reschedule appointments, telehealth appointments will remain as susceptible to non-attendance as traditional appointments are.

The authors declare no conflicts of interest.