The Efficacy of Telemedicine-Supported Discharge Within an In-Home Model of Care

Introduction

Hospital in the Home (HITH) provides care in a patient's residence for conditions requiring clinical governance, monitoring, and/or input that would otherwise require treatment in an admitted hospital setting. ¹ Transferring admitted patients early into HITH is popular in many health jurisdictions, such as Queensland, Australia, based on evidence that patients have better or equal health outcomes, increased patient and caregiver satisfaction, and reduced expenditure relating to service provision compared with patients remaining in hospital. ^2,3 Improvements in technology have enabled the provision of treatment in the home at comparable levels of care to those experienced in hospital settings with the promise of increasingly complex and varied cases supported in the home in the future. ^4,5 Enthusiasm for in-home treatment models among patients and clinicians ⁶ has not been supported by consensus among researchers that economic benefits exist, ⁷ prompting investigations into ways in home models can be delivered more efficiently while maintaining patient's safety.

Conditions commonly treated under HITH models of care include cellulitis, pulmonary embolism, urinary tract infection, respiratory infection, and venous thrombosis. Patients receive daily contact from registered nurses trained to perform Criteria Led Discharge (CLD) when appropriate for patients with few complications. ¹ Patients unsuited to CLD require the review of a physician before being discharged from the program, which traditionally involves the physician visiting the patient at home or the patient attending hospital in a scheduled clinic.

The evaluation of equipping nurses with mobile videoconferencing functionality seeks to determine if such technology can reduce the frequency in which physician or patient travel is required before discharge from HITH. Also relevant is whether a difference is observed in the readmission rates of patients who were discharged during a telemedicine consultation compared to the current practice of a being discharged by a physically present doctor.

Methods

A 6-month evaluation began in July 2015. Registered nurses, employed by HITH to provide clinical support to patients in their homes, were provided with a tablet computer (Apple iPad) with Internet connectivity (Telstra 4G Network) and videoconferencing software (Cisco Jabber for Telepresence). Four hours of training in the use of the device was provided and nurses were encouraged to initiate video calls to hospital-based physicians for treatment advice or to discuss suitability for discharge while visiting patients in their homes. Instances where the nurse initiated a video call were automatically captured using call management software (Cisco Telepresence Management Suite) and combined with patient condition and demographic details, captured within a patient administration system (Cerner).

To determine if the clinical standards of the service were maintained, rates of hospital readmission within 28 days between patients discharged from the HITH model either in person or via telemedicine were compared in two ways. First, a binary logistic regression was used to assess the extent to which the patient's discharge mode (in person or telemedicine) could predict a patient's readmission within 28 days after controlling for effects of the patient's age (years), gender, and length of admission to HITH (days). Second, a matched pairs design was used to directly compare readmission rates between patients discharged by telemedicine with patients discharged in person. Telemedicine patients were matched with in-person patients by age group (5-year age bands), gender, and length of admission to HITH (days). Length of stay was chosen as a surrogate measure for severity due to the nonstandardized and occasionally incomplete free-text descriptions of the patient's condition within the patient administration system.

Results

During the trial period, a total of 345 patients were discharged from the HITH program. Of these, 10.1% (n = 35) were discharged by a physician remotely using the telemedicine solution. Video consultations lasted an average of 4 min 31 s. Of the patients discharged in person, 16 were readmitted within 28 days (5.2%) compared with 1 readmission from patients discharged remotely (2.9%). Patients' age, gender, length of stay (days), and method of discharge (in person or telemedicine) could not predict readmission within 28 days using a binary logistic regression model. The overall model was not significant, χ²(4) = 8.11, p = 0.09. Results are summarized in Table 1.

The matched sample composed of 58 patients, with 29 patients in each of telemedicine and in-person conditions. In the matched sample, there were three readmissions from the in-person group (10.3%) and one from the telemedicine group (3.4%). A two-tailed Fisher's exact test was used to analyze the difference in readmission rates between the two groups. The odds of being readmitted following telemedicine discharge were not significantly different from the odds of readmission following in-person discharge (odds ratio: 0.34; 95% confidence interval [0.01–4.51]).

Discussion

The study identified 35 instances (10.1% of total) where nurses were confident in facilitating the discharge of a patient admitted to HITH via telemedicine with a remotely located physician. Confidence in both the nurse present with the patient and the technology was sufficient for all proposed discharges to be successfully completed. Video consultations lasted an average of 4 min 31 s. Before this solution being available, each of these discharges would have required a minimum of an hour of physician consultation and travel time. Calculating efficiencies or cost reductions beyond identifying instances where physician travel time was avoided was out of scope of this study.

Crucially, this trial found insufficient support for a difference in readmission rates between in-person and telemedicine cohorts upon analysis, both as a predictor of readmission after controlling for age, gender, and length of stay and when directly compared with the proportion of readmissions in a matched sample. This indicates that the short-term health outcomes of patients are not affected when discharged remotely using telemedicine technology as an alternative to a physician being physically present for the discharge process. However, ongoing evaluation is prudent as the option of remote discharge is made available to a wider range of patients in the HITH program.

Heterogeneity and/or comorbidity of health conditions were not able to be reliably accounted for from information contained in the patient administration system. This potentially limits the generalizability of any patient outcome evaluation. As health services globally move toward more complete digital record systems, the opportunities for timely rigorous analyses of services also expand. A greater awareness among service providers of the need for accurate reporting must be encouraged for the benefits of these systems to be realized.

Confirmation that current standards of clinical care can be maintained while delivering some services using telemedicine models has encouraged the exploration of other applications of the technology. This includes the use of a zero footprint Web real-time communication telehealth portal to allow low-acuity patients receiving treatment in their home to videoconference with HITH clinicians outside scheduled visits and without a nurse presence.

Conclusion

The results of evaluating telemedicine support for nurses providing low-acuity in-home care indicate that patients could be discharged from care remotely while maintaining existing standards of clinical care.