Table 4 Summary of findings for outcomes of interest in the systematic review, with overall risk of bias assessment
Nocturnal digital surveillance | ||||
|---|---|---|---|---|
Population: Persons ≥ 50 years of age Settings: living at home or in care settings in OECD countries or equivalent Intervention: digital sensors or monitors used at night for non-physiological surveillance Comparison: Care as usual | ||||
Outcomes | Number of studies | Relative change due to intervention | Overall risk of bias a | Comments |
Health-related outcomes: injury, unexplained absence, other adverse medical outcome for receivers of care | 2 | 1: No difference in rates of injury (0.02, p = 0.828; Holmes et al. 2007) [7] 2: No difference in reduction of dangerous events (X2 = 1.72; p = 0.79 for whole period; X2 = 3.58; p = 0.058 for when system turned on; Rowe et al. 2009) [8] | High | 192 accidents and 9 dangerous events, respectively, were observed during the two studies. |
Welfare outcomes: quality of life, perceived safety or security, and other related welfare outcomes for receivers of care | 2 | 1: Improved affect (mean change in 5-point affect scale: 0.29, p = 0.034; Holmes et al. 2007) [7] 2: 95 % reported some improvement in quality of life (Sivertsen and Løe, 2019) [10] | High | Improved affect (in Holmes et al. 2007) [7] was directly predicted by increased direct care due to increased surveillance. Perceived QoL was reported by caregivers for patients (in Sivertsen and Løe, 2019) [10] based on a single survey question. |
Social care provision outcomes: burden informal caregivers and social care staff and organisations | 4 | 1,2: No evidence of a difference in staff burden (-0.02, p = 0.96; Holmes et al. 2007; no statistic for Røhne et al. 2016) [11] 3: No evidence of a difference in caregiver worry (0.72; p-value not reported), sleep time (-3.91, p = 0.20) or quality (0.05, p-value not reported; all from Rowe et al. 2010) [8] 4: 75 % reported improvement in work effectiveness and processes (Sivertsen and Løe, 2019) [10] | High | Staff burden (in Holmes et al. 2007) [7] was a monthly average of daily scores on a seven-item, 3- or 5-point Likert scale. Caregiver worry measured on a 10-point Likert scale; objective sleep by actigraphy; sleep quality on a 5-point Likert scale (all from Rowe et al, 2010) [8] Improvement in work effectiveness and processes was reported (in Sivertsen and Løe, 2019) [10] based on a single survey question. |
Social care economic outcomes: e.g. costs for visiting and transportation, search costs when missing. | 1 | 15 % reduction in social care costs (Røhne et al. 2016) [11] | High | Cost reductions were associated with fewer required visits and transportation; avoided costs due to ability to remain in the home instead of care institutions were not assessed. |
aHigh = There is a high risk that bias was introduced during the study that affects the certainty of the outcome and lowers confidence in the result. This assessment is equivalent to “High” for the Risk of Bias tool, or Serious/Critical in the ROBINS-I tool. Low = There is a low risk that bias introduced during the study will affects the certainty of the outcome and lowers confidence in the result. This assessment is equivalent to “Low” for the Risk of Bias tool, or Low/Moderate in the ROBINS-I tool. | ||||