Table 4 Summary of study characteristics

Ang et al., 2024

United States

Evaluate the accuracy of CGM devices compared with point-of-care blood glucose testing

Cohort study

CGM devices

In one of adult ICU

59 postoperative patients with hyperglycaemia and requiring intravenous insulin infusion

Post-intervention: 99.7% of the paired CGM glucose levels and point-of-care blood glucose testing fell within the Zone A and Zone B of the Clarke Error Grid which indicated a high accuracy CGM measurements for postoperative patients in ICUs

90% of time spent within the glucose targeted range by using the CGM devices

Armstrong, 2023

United States

Assess the impacts of standardised reporting system via the electronic health information record system on the development of HAPIs

Cohort study

Standardised reporting system via the electronic health information record system

In cardiothoracic ICU and neurologic ICU

Total 619 patients were analysed for HAPIs for 2 years

Pre-intervention period: from May 2018 to April 2019: total 1235 HAPIs were identified and from April 2019 to May 2020, total 1031 HAPIs were identified.

Post-intervention period: From May 2020 to April 2021, there was total 631 HAPIs which was reduced by 38.8%. From May 2021 to April 2022, there was total 423 HAPIs identified which was reduced by 33%.

Behrendt et al., 2014.

United States

Hypothesis: CBPM would improve efficiency of patients’ repositioning, reducing HAPIs

Non-randomised experimental study

Pressure Ulcer Systems-CBPM

Medical ICU in a tertiary-care hospital.

422 patients (CBPM n = 213; control n = 209).

Significant reduction in development of Stage II pressure ulcers: CBPM group n = 2 patients (0.9%); control group n = 10 patients (4.8%); p = 0.02.

Chapuis et al., 2010.

France

Assess the impact of an ADDS on the incidence of medication errors related to picking, preparation, and administration.

Cohort study

Medication Administration Systems-ADDS

Two MICUs in a 2,000-bed university hospital. Both units (8 and 10 beds) had comparable activities and shared the same staff

68 nurses were observed.

1,476 medications were picked, prepared and administered.

No difference in % Total Opportunities for Error (TOE) identified between control and study units prior to ADDS implementation (19.3% TOE and 20.4% TOE respectively)).A Significant difference was observed in %TOE post ADDS implementation (18.6% and13.5% TOE, respectively; p < 0.05).

%TOE significantly decreased in the study unit pre and post ADDS (20.4% TOE pre-ADDS (Phase I) to 13.5% TOE post-ADDS (Phase II), p < 0.01).

Preparation dose errors decreased from 3.8–0.5% Detailed Opportunities for Error (DOE) (p = 0.017) in the study unit. No reduction in picking or administration errors.

Storage errors reduced post-ADDS (study unit pre n = 51, 27.7%, post n = 2, 0.7%; control unit pre n = 65, 34.9%, post n = 27, 14.4%; p < 0.01).

Most errors (n = 244, 84%) caused no harm. ADDS implementation did not change the % of medication errors causing harm (Control = 0.6% DOE, study group = 0.7% DOE).

Curtis et al., 2020

Australia

Examine the impact of a consolidated electronic checklist on risk screening rates for falls, pressure ulcers and substance use.

Cohort study

Electronic health information record system

Four EDs in a regional health service, between November 2016 and February 2019.

A total of 33,561 ED presentations were analysed for the pre group and 35,807 for the post group

The proportion of patients who had all three screens completed increased from 1.3–5.5% (p < 0.001). Substance use screening increased from 1.7–12.4% (p < 0.001). Pressure ulcer risk screening increased from 38.6–41.7% (p < 0.001). When only patients aged 65 years and above were examined, the completion rate of pressure ulcer risk screening increased from 46.6% (pre) to 53.1% (post) (p < 0.001).

In contrast, falls screening decreased from 38.0–32.6% (p < 0.001).

Feral-Pierssens et al.,, 2022

Canada

Assess the safety of a redirection process by triage nurses using CDSSs for low-acuity patients

Cohort study

CDSSs

A level 1 academic trauma centre

642 low-acuity patients redirected to nearby clinics

Post-implementation, among a total of 642 redirected low-acuity patients, there were 2.8% of the patients (n = 18) and 4.8% of the patients (n = 31) returned back to the ED unexpectedly within 48 h and within 7 days, respectively.

There were no hospital admissions or deaths identified within 7 days among those redirected low-acuity patients.

Kahn et al., 2014.

United States

Examine ICU care delivery and outcomes following nurse-led EHR use

Cohort study

Electronic health information record system

8 subspecialty ICUs in an Academic Medical Centre of a University Hospital

13,227 patients were included in the study. 4,339 (32.8%) in preintervention period, 8,938 (67.6%) in postintervention period.

Post EHR intervention, daily sedation interruptions increased (IRR, 1.57; 95% CI, 1.45–1.71; p < 0.001), daily spontaneous breathing trials increased (IRR, 1.24; 95% CI, 1.20–1.29, p < 0.001), mean ICU length of stay reduced (pre = 4.1 ± 5.4 days, post = 3.9 ± 5.0 days; p = 0.005) and hospital length of stay reduced (pre = 11.9 ± 12.5 days, post = 10.8 ± 11.2 days; p < 0.001).

no difference found in Catheter-associated urinary tract infection (1.58 before, 1.77 after, IRR 1.12; 95%CI 1.20–1.29; p = 0.63), central catheter-associated bloodstream infection (0.72 before, 0.77 after, IRR 1.06, 95%CI, 0.58–1.94; p = 0.84), ventilator-associated pneumonia rates (3.24 before, 2.67 after, IRR 0.82 (95%CI, 0.57–1.19, p = 0.30), or hospital mortality (0.96 95%CI (0.84–1.09) p = 0.54).

Legambi et al., 2021

United States

Assess the impacts of an electronic behavioural activity rating scale (BARS) on risk assessments rates for agitation

Cohort study

Electronic health information record system

Beltimore Emergency department

Total 780 patients with behavioural and medical health presentations

Post-BARS implementation: of total 780 patients with behavioural and medical health presentations, nearly 65.77% patients (n = 513) had BARS documented every 2 h.

Agitation was also detected and documented for 206 patients (n = 26.41%) which indicated their BARS score 5 or 6 out of 7.

Among those agitated patients, about 68% (n = 140) of agitated patients’ behaviours were reduced by nonrestraint interventions, including medications, de-escalation techniques and diversional activities.

Total 18 episodes of restraint were used post-BARS implementation comparing to 20 episodes of restraint use pre-BARS implementation.

Although there was no statistical significance regarding the incidence of restraint use post-BARS implementation, 75% of reduction was documented for patients who stayed with restraint more than one day in EDs post-BARS implementation (n = 8 patients pre-BARS; n = 2 patients post -BARS).

Levesque et al., 2015.

France

Evaluate the effects of ICIS on the outcome of critically ill patients.

Cohort study

Electronic health information record system

15-bed Liver ICU of a University Hospital

1,397adult patients (BEFORE, n = 662 and AFTER n = 735)

Implementation of ICIS decreased the ICU length of stay (pre = 8.5 ± 15.2 days, post = 6.8 ± 12.9 days; p = 0.048).

No significant change to length of hospital stay (pre = 27.7± 34.6 days, post = 28.6±33.3 days; p = 0.79), ICU readmission rate (pre = 4.4%, post = 4.2%; p = 0.86), or mortality rate (pre = 11.2%, post-= 9.6%; p = 0.35). However, observed mortality was significantly lower than predicted by SAPS II post ICIS (SMR 0.75; p < 0.001).

Lowenstein et al., 2023

United States

Examine the impacts of an electronic clinical opioid withdrawal scale (COWS) on risk assessments rates for opioid misuse

Quasi-experimental study

Electronic health information record system and CDSSs

5 EDs including 3 intervention EDs and 2 control EDs under the same health systems

In the intervention group: total presentations were 2462. There were 1258 presentations pre-intervention period and 1204 post-intervention period.

In the control group: total presentations were 731. There were 459 presentations pre-intervention period and 272 post-intervention period.

In the intervention EDs, the completion rates of COWS have been increased by 21.5% from 26% (n = 332) in the pre-implementation periods to 48% (n = 577) in the post-implantation periods in the intervention EDs (95% CI: 17.7 to 25.3).

However, there were no statistically significant changes in the control EDs (9.6% (n = 44) COWS completion rates pre-implementation; 14.3% (n = 39) COWS completion rates post-implementation; 95% CI: -0.5 to 10).

Mann et al., 2011.

United States

To determine the safety and efficacy of Clinical Decision Support Systems (CDSSs) to control serum glucose concentration in a burns intensive care unit

Randomised controlled trial

CDSSs

16-bed regional adult burn centre ICU responsible for the care of both military and civilian burn patients.

22 patients enrolled, but data reported on 18 patients as some did not complete the study.

Mean blood glucose levels in CDSS group were significantly lower than those in the paper protocol group (CDSS = 113 ± 10.2 mg/dL, paper = 119 ± 14 mg/dL; p = 0.02).

Time in BGL target range was significantly longer in the CDSS cohort (CDSS = 47 ± 17% time, paper protocol = 41 ± 16.6% time; p < 0.05).

Time over target range was not significantly reduced in the CDSS group (CDSS group: 49 ± 17.8% vs. Paper group: 54 ± 17.1%; p = 0.08); and time less than 80 mg/dl was similar between groups (CDSS: 4.5 ± 2.8% vs. Paper protocol: 4.8 ± 3.3%; p = 0.8).

A total of four events of hypoglycaemia (< 40 mg/dl) occurred, two events in each study arm. No adverse clinical events were noted for any episode of low blood glucose level.

McLeod et al., 2020.

Canada

To determine the interrater agreement of triage score pre- and post-implementation of eCTAS.

Determine the triage time and accuracy pre- and post-implementation of eCTAS.

Cohort study

CDSSs

7 hospital EDs across Ontario, Canada.

A total of 1,491 individual patient triage assessments (752 pre-eCTAS, 739 post-implementation) were audited

Improvements in accuracy were observed across all triage categories post-eCTAS implementation. eCTAS significantly reduced the number of patients over-triaged (pre = 12.0%, post = 5.1%; 95% CI 4.0 to 9.7,) and under-triaged (pre = 12.6%, post = 2.2%; 95% CI 7.9 to 13.2), and this was consistent across all participating sites.

Interrater agreement was higher post eCTAS. Aggregate unweighted κ pre-eCTAS = 0.63 (95% CI 0.58 to 0.68), post-eCTAS = 0.89 (95% CI 0.86 to 0.92); quadratic-weighted \(\:\kappa\:\) pre-eCTAS = 0.79; post-eCTAS = 0.93.

Triage time was captured for 3,808 patients pre-eCTAS and for 3,489 post-eCTAS. Median triage time increased post eCTAS implementation (pre-eCTAS = 312 s, post e-CTAS = 347 s; 95% CI 29 to 40 s).

Meyfroidt et al., 2011.

Belgium

Assess the impact of a computer-generated blood glucose alert, generated by a Patient Data Management System and superimposed on a paper-based guideline, on tight glycaemic control in the ICU.

Cohort study

CDSSs

56-bed, predominantly surgical ICU of a 1900-bed tertiary University Hospital.

Pre-alert cohort n = 729 adults admitted to ICU between 31/1/2007 and 31/7/2007, and alert cohort n = 644 adults admitted to ICU between 31/8/2007 and 6/2/2008.

CDSS significantly reduced mean blood glucose value per patient (pre-alert = 112 (105–122) mg/dl, post alert 110 (104–119) mg/dl; p = 0.002), and mean Glycaemic Penalty Index (GPI) (pre-alert = 20 (14–28), post-alert = 19 (13–26); p = 0.029).

HGI also significantly reduced, pre-alert = 10 (5–17) mg/dl, post-alert = 9 (4–15); p = 0.004).

The percentage of patients who experienced an episode of hypoglycaemia significantly declined from 6.5% (n = 47) pre-alert system to 4.0% (n = 26) post-alert system (\(\:\rho\:\) =0.043). The introduction of the alert did not result in a reduction in the HoGI (0.5 mg/dl in both groups).

Meer et al., 2012.

Switzerland

Investigate the safety of computer-assisted telephone triage for walk-in patients with non-life-threatening medical conditions in an ED

Cohort study

Telehealth

Interdisciplinary Adult Emergency Centre of a University Hospital.

208 patients

The unweighted κ was 0.092 and the weighted κ was 0.115 between hospital physicians versus call centre nurses. The unweighted κ was 0.080 and weighted κ was 0.159 between primary care physicians and call centre nurses.

Ruesch et al., 2012.

United States

Examine the impact of a nurse-implemented tele-ICU staffing model on patient complications and outcomes.

Cohort study

Telehealth

Adult Critical Care Unit

1308 patients

Overall ICU length of stay significantly decreased on a per day basis from 4.1 to 3.5 days (p≤0.05).

Severity-adjusted mortality decreased the actual mortality compared with predicted mortality, indicating 22 lives saved.

The incidence of VAP decreased by 13% related to a change in the median VAP from 2.99 in 2008 to 2.6 in 2009.

Staff compliance with VAP bundle significantly increased, from 87.2–93.3% (p = 0.02).

Compliance with patient deep vein thrombosis and peptic ulcer disease bundles demonstrated continuous improvement of 1%and 0.5%, respectively. These results were not statistically significant.

Zhang et al., 2024

China

Explore the impacts of the electronic health information record quality control system on the real-time data collection and quality control for nursing assessments and medication administrations

Cohort study

CDSSs working as quality control purpose

in one of the ICUs in China

Total 600 patients’ cases were analysed

Post-intervention, the results demonstrated the significant improvements in the percentages of inaccurate vital signs documentations (decreasing from 9% pre-implementation to 1.33% post implementation, p < 0.001).

The incidence of incomplete mediation administrations was reduced by 1.66% dropping from 3.33% pre-implementation to 1.67% post-implementation (p < 0.001).

The prevalence of missed nursing assessments dropped down from 8% pre-implementation to 1.33% post-implementation (p < 0.001).

Zikos et al., 2014.

Greece

Investigate the effect of an electronic trauma documentation system on ED length of stay

Cohort study

Electronic health information record system

Emergency department of a university hospital with a capacity of 950 beds

Control group paper-based documentation (n = 99) (Year 1), intervention group electronic documentation (n = 101) (Year 2).

Time between admission and completion of planned care was significantly lower in the intervention group (100 ± 92 min) than the control group (149 ± 29 min) (p < 0.001).

A similar effect was found on the total ED length of stay (intervention group = 127 ± 93 min, control group = 206 ± 41 min in the control group; p < 0.001). Time between completion of care and discharge from the ED also significantly reduced (intervention 26 ± 10 min, control 57 ± 23 min; p < 0.001)