Abstract
Background
Digital technologies in healthcare are seen as mechanisms to improve and optimise management of health conditions.
Aim
To assess the impact of digitisation on clinical outcomes and medication errors for patients with diabetes.
Methods
This repeated cross‐sectional study used data collected from the Queensland Inpatient Diabetes Survey (QuIDS), which was conducted in 2019 and 2021 at digital and paper‐based hospitals in Queensland. Relevant data were collected from patients with diabetes admitted to participating hospitals during a single day of the study week. Outcomes and error rates of patients who were admitted to digital versus paper‐based hospitals were compared. Regression determined the factors that contributed to ‘good diabetes days’ and ‘no hypoglycaemic days’. Prescribing and management errors were compared.
Results
Data on 1942 patient admissions (6977 patient bed days) were collected. Of these, 1076 patient admissions (55%) were at a digital hospital, while 866 patient admissions (45%) were at a paper‐based hospital. Using regression, it was found that being admitted to a digital hospital increased the odds of a ‘good diabetes day’ by 45% (P < 0.001), but there was no change in ‘no hypoglycaemic days’ (P = 0.183). There was a reduction in the proportion of patients with at least one error type across all error categories amongst those admitted to a digital hospital (P < 0.001).
Conclusion
Admission to a digital hospital improves the odds of experiencing a good diabetes day but does not change the occurrence of hypoglycaemia. There are potential quality and safety considerations for those hospitals still delivering paper‐based diabetes care.
Keywords: diabetes mellitus, clinical outcomes, medication error, digital health, electronic health record
Introduction
The prevalence of diabetes mellitus continues to rise globally, with an estimated prevalence of 537 million people worldwide in 2021. 1 Among hospitalised patients, diabetes and its complications are attributed to an increased risk of extended length of stay and poor patient outcomes and represent a significant cost to health‐related expenditure. 2 In the International Diabetes Federation Diabetes Atlas, it was estimated that the direct health expenditure due to diabetes is close to US$1 trillion globally. 1
Evidence demonstrates that among hospitalised patients, both hyperglycaemia and hypoglycaemia are associated with higher rates of morbidity, mortality and increased length of stay. Avoidance of hypoglycaemia and excessive hyperglycaemia during hospital admission is an important goal of care. 3 Despite this, rates of hypoglycaemia and hyperglycaemia remain high. 4
Digital technologies in healthcare are seen as a mechanism to improve and optimise management of health conditions. A 2022 systematic review found that computerised physician order entry (CPOE) and clinical decision support systems (CDSS) improve mean patient‐day‐weighted glucose levels, although the impact on rates of hypoglycaemia, hyperglycaemia and medication errors was inconclusive. 5 Most of the studies in this review were of single hospital sites which were usually large academic hospitals and examined only one of either prescribing or administration errors in digital hospital settings. 5
The primary aim of this study was to determine whether patients at digital or paper‐based hospitals had different clinical outcomes during their inpatient stay.
Methods
Study design
The Queensland Inpatient Diabetes Survey (QuIDS) is an Australian study that surveys an entire jurisdiction with the aim of assessing the quality of care for patients with diabetes in Queensland hospitals. 4 This survey is administered to hospitals across Queensland regardless of their size or the level of care they provide and represents a ‘whole system’ assessment of diabetes management. This repeated cross‐sectional study used data collected from two QuIDS across two time periods. The first QuIDS was conducted at participating hospitals during 18–29 March 2019, while the second QuIDS was conducted during 11–21 May 2021. Each site selected a single day during the study week to undertake the survey.
Setting
The QuIDS were administered across public and private hospitals in Queensland, Australia. There are 106 hospitals in Queensland which serve a population of 5 million people. 6 All acute hospitals in Queensland were invited to participate in QuIDS, with invitations sent to all members of the Queensland Diabetes Clinical Network, which consists of approximately 300 clinicians. Involvement was voluntary and relied on clinicians undertaking their own data collection at each site without any additional resources.
For paper‐based hospitals, state‐wide standardised blood glucose and insulin prescription charts are utilised for blood glucose and insulin management. These standardised charts also contain basic decision support to assist with initiating insulin therapy, dosing of sliding‐scale insulin and adjustment of doses for patients with hypoglycaemia. Charts also contain information to assist staff with identification and response to hyper‐ or hypoglycaemia.
Digital hospitals were defined as those hospitals that at least had electronic health records (EHRs) but could also include CPOE or remote blood glucose monitoring. The CPOE system contained inbuilt decision support for supplemental insulin dosing, management of hypoglycaemia and links to the same protocols as present on the paper insulin‐glucose charts. The definition of what constituted a digital hospital was deliberately chosen to be broad (i.e. could include only EHR without CPOE or remote blood glucose monitoring) in order to assess the effects of digital systems across the entire state of Queensland. All digital technology (EHR and CPOE) is standardised across the entire state and does not vary in content from site to site. The presence of this digital system within a given institution was self‐reported by each hospital's survey lead as part of QuIDS and checked by the investigators.
Participants
Data were collected about patients admitted to participating hospitals with diabetes during the study day at eligible hospital sites. Participants were 18 years of age or older and admitted to medical, surgical, cancer, gynaecology, mental health wards or intensive care/high dependency unit, but paediatric, maternity, emergency department or short‐stay (<24 h) ward admissions were excluded.
Data
Audit data were collected on a single day chosen by the hospital site within the specified weeks of data collection. Glycaemic control data were also collected for up to five preceding consecutive inpatient days. The metrics of glycaemic control included the rates (per 100‐patient days) of ‘good diabetes days’ and ‘no hypoglycaemic days’. 4 , 7
‘Good diabetes days’ were defined as a day on which the patient had appropriate blood glucose monitoring, no hypoglycaemia and no more than one blood glucose level measurement outside the Australian Diabetes Society target range of 5–10 mmol/L. 3 , 4 , 7 Patients for whom tight glycaemic control was not appropriate (e.g. frail elderly, palliative care patients) were not expected to have ‘good diabetes days’ and so were excluded from this analysis. It was up to data collectors to use clinical judgement to determine which patients would not be suitable for tight glycaemic control.
Participants had ‘no hypoglycaemic days’ if all blood glucose levels were ≥4.0 mmol/L on any given day. All patients in the dataset were included for this outcome.
Medication errors associated with diabetes therapies included either glucose management errors (i.e. failure to appropriately adjust treatment after either hyper‐ or hypoglycaemia) or prescribing errors (an unclear or unsafe order for a diabetes therapy that might lead to an incorrect dose being administered). 4 Insulin errors included either a management or prescribing error related to the use of insulin. Participants were defined as having had an error if they had had one or more of that particular error within the 5 days prior to the audit (i.e. the same period of time as the glycaemic control data).
Other data collected as part of the survey included patient demographics, diabetes treatments and complications, specialty of ward and treating consultant and involvement of diabetes specialist or specialist inpatient diabetes team. Data collection forms had in‐built decision support to help clarify questions. Most data collection was undertaken by diabetes clinicians such as endocrinologists and diabetes educators. Each site was given the opportunity to attend an information session (virtually) prior to the data collection to clarify any uncertainty. Further information on the QuIDS can be found in a paper published on the results of the 2019 survey. 4
Outcomes
The primary outcome of the analysis was the rates of ‘good diabetes days’ at digital versus paper‐based hospitals. 3 Secondary outcomes included the rates of ‘no hypoglycaemic days’ and the proportion of patients that had one or more of a glucose management error, prescribing error or insulin error within the preceding 5 days.
Statistical methods
Data were analysed using R (version 4.1.2). A P‐value was considered significant if it was less than 0.05. Means were used to summarise continuous variables. Counts and percentages were used to summarise categorical variables.
Univariable and multivariable analysis was conducted to test for the association between glycaemic control (‘good diabetes days’ and ‘no hypoglycaemic days’ – dependent variables) and digital and paper‐based hospitals, as well as a range of potential confounding variables.
The relationship between being a patient admitted to a digital hospital and the rate of ‘good diabetes days’ was modelled using quasi‐binominal regression for each outcome (expressed as a proportion). Quasi‐binominal is similar to ordinary binominal logistic regression, but it allows for values between 0 and 1 to be used for the dependent variable, not just 0 and 1. It also incorporates an estimated over‐dispersion factor.
For the hypoglycaemia outcome, the original data were a count variable. For ease of regression analysis, the data were converted to a binary variable (‘no hypoglycaemic days’ – yes/no – for each day of blood glucose monitoring data available), with modelling undertaken using binomial regression.
Additional models of good diabetes days, no hypoglycaemic days and medication error rates were undertaken grouping those hospitals who were paper‐based and with only EHR into the ‘paper‐based hospital’ category and leaving those hospitals that had combined EHR and CPOE as the ‘digital hospitals’, given that it was thought likely that EHR without CPOE would be less likely to be associated with better glycaemia outcomes. Additionally, models were fit to subsets of the data stratified by type of diabetes, to determine if different factors were significant when only patients with type 1 or type 2 diabetes mellitus were considered.
Patients with missing data for any of the included variables were excluded from the models.
Error rates were compared between digital and paper‐based hospitals using Pearson chi‐squared test.
Dependent variables
The two dependent variables were ‘good diabetes days’ and ‘no hypoglycaemic days’.
Independent variables
The primary independent variable was admission to a digital or paper‐based hospital. Potential confounders (independent variables) incorporated in the models included age, gender (male, female, indeterminate), self‐identified Aboriginal or Torres Strait Islander (yes, no), type of diabetes (type 1, type 2, other), CKD 3 or higher (yes, no), insulin prescribed while inpatient (yes, no), patient on sulphonyl urea prior to admission (yes, no), the specialty of the ward the patient was in (medical, non‐medical) and involvement of specialist diabetes credentialed diabetes educator (CDE) or diabetes consultant (yes, no).
Information on the clinical capacity of each of the hospital sites was also collected using the Clinical Services Capability Framework. 8 This Queensland Health‐wide framework describes the level of complex care that can be provided at each facility, ranging from level 1 to 6. These levels are defined as follows: 1 – low complex ambulatory care services; 2 – low complex inpatient and ambulatory services (small rural or remote hospital); 3 – low to moderate complex inpatient and ambulatory care services (small rural or remote hospital); 4 – moderate complex inpatient and ambulatory care services (smaller outer metropolitan or small regional hospitals); 5 – moderate to high complex inpatient and ambulatory care services (large regional or outer metropolitan hospitals); and 6 – high complex inpatient and ambulatory care services (tertiary or quaternary referral hospitals).
Medication error rates
Errors were grouped into four categories: prescribing errors, glucose management errors, medication errors and insulin errors. 4 As each medication order could have multiple types of prescribing or management errors, the groups of medication and management errors were not mutually exclusive. Medication error rates were expressed as a proportion of the number of patients with at least one error for up to 5 days of diabetes management prior to and including the day of data collection.
Ethics
This study was considered an audit of practice, and an exemption from formal hospital human research ethics review was granted for all sites by the chair of the Royal Brisbane and Women's Hospital Human Research Ethics Committee (Ex/2022/QRBH/81946).
Patient and public involvement
It was not appropriate to involve patients or the public in the design, conduct, reporting or dissemination plans of our project.
Results
Descriptive
Screening for eligibility occurred for 4265 patients in 2019 and 5398 in 2021. This represents 64.1% and 71.6% of all eligible acute overnight hospital beds in Queensland respectively. Diabetes was identified in 1003 (24.0%) in 2019 and 1277 (23.7%) in 2021.
QuIDS were completed for 850 in 2019 (84.7%, 850/1003) and 1092 in 2021 (85.5%, 1092/1277). Of these patients across both time periods, 1076 (55%) were from a digital hospital, while 866 (45%) were from paper‐based hospitals (Table 1).
Table 1.
Comparison of demographics between digital and paper‐based hospitals
| Number of patients per group | Paper‐based hospital, n = 866 | Digital hospital, n = 1076 | P * |
|---|---|---|---|
| Bed days | 3023 | 3954 | — |
| Year | 0.2 | ||
| 2019 | 393 (45%) | 457 (42%) | |
| 2021 | 473 (55%) | 619 (58%) | |
| Age group | 0.3 | ||
| 18–30 | 18 (2.1%) | 31 (2.9%) | |
| 31–40 | 19 (2.2%) | 34 (3.2%) | |
| 41–50 | 62 (7.3%) | 90 (8.5%) | |
| 51–60 | 124 (15%) | 172 (16%) | |
| 61–70 | 200 (24%) | 239 (23%) | |
| 70+ | 423 (50%) | 487 (46%) | |
| Unknown† | 20 | 23 | |
| Gender | 0.9 | ||
| Female | 365 (42%) | 461 (43%) | |
| Male | 499 (58%) | 612 (57%) | |
| Indeterminate | 1 (0.1%) | 2 (0.2%) | |
| Unknown† | 1 | 1 | |
| Aboriginal or Torres Strait Islander | 75 (8.8%) | 155 (15%) | <0.001 |
| Known | 876 | 1066 | |
| Unknown† | 10 | 10 | |
| Macrovascular complications present | 402 (47%) | 548 (51%) | 0.070 |
| Known | 857 | 1071 | |
| Unknown† | 9 | 5 | |
| Microvascular complications present | 214 (25%) | 277 (26%) | 0.6 |
| Known | 859 | 1066 | |
| Unknown† | 7 | 10 | |
| Diabetes type | 0.075 | ||
| Type 1 | 51 (5.9%) | 53 (5.0%) | |
| Type 2 | 806 (93%) | 994 (93%) | |
| Other | 7 (0.8%) | 21 (2.0%) | |
| Unknown† | 2 | 8 | |
| Insulin administered while inpatient | 406 (51%) | 546 (54%) | 0.2 |
| Known | 796 | 1004 | |
| Unknown† | 70 | 72 | |
| Involvement of diabetes specialist or team | 208 (25%) | 332 (31%) | 0.004 |
| Known | 829 | 1066 | |
| Unknown† | 37 | 10 | |
| Ward type | 0.12 | ||
| Medical | 576 (68%) | 753 (71%) | |
| Non‐medical | 277 (32%) | 308 (29%) | |
| Unknown† | 13 | 15 | |
| Specialty of treating consultant | 0.7 | ||
| Medical | 577 (70%) | 664 (71%) | |
| Non‐medical | 247 (30%) | 270 (29%) | |
| Unknown† | 42 | 142 | |
| Receiving sulphonyl ureas | 145 (17%) | 135 (13%) | 0.011 |
| CKD Stage 3 or higher | 235 (27%) | 306 (28%) | 0.6 |
| Clinical service capability framework level | <0.001 | ||
| 2 | 32 (4.0%) | 0 (0%) | |
| 3 | 45 (5.6%) | 7 (0.7%) | |
| 4 | 215 (27%) | 75 (7%) | |
| 5 | 232 (29%) | 384 (36%) | |
| 6 | 285 (35%) | 610 (57%) | |
| Unknown† | 57 | 0 |
Two‐sample tests for equality of proportions; Welch two‐sample test; chi‐squared test.
Unknown values represent those where data was missing or not applicable. Unknown values were excluded from proportions and statistical tests.
In 2019, 20 paper‐based hospitals and seven digital hospitals participated in QuIDS. In 2021, 23 paper‐based hospitals and 12 digital hospitals participated in the survey. Of the seven digital hospitals that participated in QuIDS in 2019, four (57%) had EHR and CPOE with insulin order sets (three (43%) also had remote blood glucose monitoring), while three (43%) had only EHR. Of the 12 digital hospitals that participated in 2021, 10 (83%) had EHR, CPOE and insulin order sets, while two (17%) had only EHR, with only one (8%) reported to have remote blood glucose monitoring.
Table 2 summarises the outcomes of the two glycaemic outcomes of interest for this project, ‘good diabetes days’ and ‘hypoglycaemic days’.
Table 2.
Glycaemic outcomes compared between digital and paper‐based hospitals
| Variable | Paper‐based hospital | Digital hospital | P * |
|---|---|---|---|
| Good diabetes day per 100 patient days† | 46.4 | 50.5 | 0.013 |
| Hypoglycaemic days per 100 patient days‡ | 4.5 | 5.8 | 0.090 |
Welch two sample t test.
Unknown for 215 patients in paper‐based hospitals and 243 in digital hospitals.
Unknown for 90 patients in paper‐based hospitals and 102 patients in digital hospitals.
Model of good diabetes days and no hypoglycaemic days
Two regression models were estimated with ‘good diabetes days’ and ‘no hypoglycaemic days’ as the two dependent variable and independent variables of: digital hospital, type of diabetes, CKD3 or higher, insulin prescribed while inpatient, sulphonyl urea, involvement of a specialist diabetes CDE or diabetes consultant, specialty of ward and clinical service capability framework level. The results of the models can be seen in Table 3.
Table 3.
Differences in patients admitted to paper‐based and digital hospitals using regression models of ‘good diabetes days’ and ‘no hypoglycaemic days’
| Model | OR | SE | 95% CI | P |
|---|---|---|---|---|
| Quasi‐binominal regression model of good diabetes days | ||||
| Paper‐based hospital | 1.00 | — | — | |
| Digital hospital | 1.45 | 0.110 | 1.17–1.80 | <0.001 |
| Binominal regression model of no hypoglycaemic days | ||||
| Paper‐based hospital | 1.00 | — | — | |
| Digital hospital | 0.77 | 0.193 | 0.53–1.12 | 0.183 |
CI, confidence interval; OR, odds ratio; SE, standard error.
For ‘good diabetes days’ the data of 1384 patients were included in the final model. The variable ‘digital hospital’ was significant (P < 0.001), odds ratio (OR) of 1.45 (95% confidence interval (CI) 1.17–1.80) indicating that if all other variables were held constant, the odds of a patient‐day being a ‘good diabetes days’ would be 45% higher in a digital hospital.
For the ‘no hypoglycaemic days’ outcome, the data of 1626 patients were included in the final model. After adjustments for potential confounders, there was no statistically significant difference in rates of hypoglycaemia between digital and paper‐based hospitals (OR 0.77 (95% CI 0.53–1.12), P = 0.183).
Supplementary items 1–6 in Appendix S1 provide a breakdown of all variables in the regression models for ‘good diabetes days’ and ‘no hypoglycaemic days’ as well as these two outcomes for patients type 1 and type 2 diabetes respectively.
Models using reclassified definitions of the ‘digital hospitals’ (those with EHR and CPOE) and ‘paper‐based hospitals’ (those with no digital capability or those with EHR only) suggested the odds of good diabetes days to be higher (OR 1.86 (95% CI 1.38–2.54), P < 0.001) compared to the base model, while other outcomes, including no hypoglycaemic days, and medication error rates remained similar (supplementary item 7 in Appendix S1).
Medication error rates
The proportions of patients who had at least one error type are described in Table 4. In digital hospitals, there were statistically significantly fewer patients with errors across all error categories.
Table 4.
Number of patients with at least one of each error type between digital and paper‐based hospitals
| Number of patients with at least one of the error types | Paper‐based hospital | Digital hospital | P * |
|---|---|---|---|
| Prescription errors | 166/635 (26%) | 121/764 (16%) | <0.001 |
| Glucose management errors | 144/589 (24%) | 126/721 (17%) | 0.002 |
| Medication errors | 240/639 (38%) | 207/769 (27%) | <0.001 |
| Insulin errors | 164/357 (46%) | 153/490 (31%) | <0.001 |
Pearson chi‐squared test.
Discussion
While a previous systematic review 5 demonstrated changes in glucometric outcomes of patients in pre‐post studies of digital systems, our study adds to this literature as, for the first time, improvements in glycaemic control in digital hospitals have been demonstrated simultaneously, across multiple hospitals of different sizes, geographical locations and acuities, irrespective of patient demographics, ward type, treating consultant and involvement of diabetes specialists.
For patients admitted to a digital hospital, the odds of having a good diabetes day were almost 50% higher compared to those admitted to a paper‐based hospital after adjusting for key covariates. This significant difference is likely related to improved prescribing practices and clinical decision support found within digital systems. This is supported by the significant reduction in all prescribing and management errors between digital and paper‐based hospitals. This was enabled by a single standardised EHR across an entire jurisdiction.
These results align with the systematic review by Sly et al., 5 which found that all studies that included multicomponent system intervention (likely reflective of the digital hospitals included in both QuIDS) improved the percentage of blood glucose levels within the target range. 5 However, the impact of multicomponent system interventions on hypoglycaemic events was less clear. 5 Our results indicate that there was no difference in hypoglycaemic events between patients admitted to digital or paper‐based hospitals. It should be noted that the rates of hypoglycaemic events were much lower than good diabetes days; therefore, the analysis of the differences in hypoglycaemic events may be less powered to detect change. However, a previous systematic review also found that electronic prescribing had no impact on hypoglycaemia. 9 Further research is required to determine what factors, specifically within digital systems, are related to good diabetes days and why they improve.
There were fewer prescribing and management errors in the digital hospitals compared to the paper‐based hospitals. For prescribing errors, this was somewhat expected as some of the errors would not be possible in the digital hospital (such as inappropriate unit or drug name abbreviations). This is consistent with what has been found in the literature, with a recent single‐site longitudinal study demonstrating improved insulin prescribing between paper‐based and digital prescribing. 10 It was reassuring to observe that management errors also decreased. While it is unclear which of the features of the digital hospital led to this improvement, literature around diabetes management indicates that CPOE and CDSS have positive impacts on medication errors, 5 this is found more broadly with other medicines too. 11 , 12 In the additional analysis where digital hospitals were reclassified, we found that patients admitted to hospitals with CPOE and EHRs had a higher rate of good diabetes days than those with just EHRs or no digital prescribing capability, further reinforcing that CPOE had a positive impact on diabetes management.
While admission to a digital hospital appears to be associated with comparatively better diabetes management, further research is required to determine which aspects of the digital systems contribute to these improvements. Across the digital hospitals included in this study, each was at various stages of implementation of digital systems across the two survey periods. Additionally, some sites have additional locally developed interventions such as diabetes management dashboards or insulin prescribing protocols that may contribute to differences in diabetes management across sites. Workforce differences between sites, such as clinical pharmacy ward services or CDEs, may also contribute to differences between sites. Additionally, although digital systems appear to perform better than paper systems, the overall rates of good diabetes days in digital hospitals remain low by international comparisons. 13 Other interventions, including specialist inpatient diabetes teams, which are rare in Queensland, have been shown to improve glycaemic outcomes and complications and could be implemented in addition to digital systems. 14 , 15
The strengths of this study include the diversity of the hospitals involved (from large metropolitan quaternary facilities to small regional hospitals) representing a ‘whole system’ digital transformation across Queensland, with data from over two thirds of acute hospital beds in the state being used in this analysis. This study also had a large sample size with almost 7000 bed days across the two study periods. These findings extrapolated across all patient bed days demonstrate the benefits of digital transformation on diabetes care.
Limitations
This study was not interventional, and although we adjusted for potentially important confounders, unmeasured variables could have confounded the identified association, such as differing levels of diabetes expertise in each hospital, FTE of the health workforce, and patient socioeconomic status. Therefore, it makes it difficult to attribute causation, although our results are consistent with previously published literature. 5 Additionally, using post hoc analysis, we did explore some other patient‐related variables including age, Aboriginal or Torres Strait Islander identification, gender and the presence of microvascular or macrovascular complications. These are described in supplementary item 8 in Appendix S1 and gave similar results to what is presented in the main paper.
Despite digital hospitals comprising only 31% (19/62) of the facilities that participated in QuIDS across the two time periods, they accounted for 55% of the patient cohort in this study. This is likely due to the greater accessibility of digital systems, which allow charts to be audited even if the patient is off the ward in a procedure or clinic. Despite this disproportionate representation of digital hospital patients, the modelling used would have taken this into account.
For the variable ‘good diabetes days’ data collectors were asked to use their clinical judgement to determine if a patient was suitable for a ‘good diabetes day’. Common exclusion reasons included patients who were palliative or frail elderly. While examples were provided, the decision was ultimately based on clinical judgement, making ‘good diabetes day’ a potentially imprecise measure of patients who should have been audited for a ‘good diabetes day’.
Further, QuIDS were administered by different people at each facility. This may have led to different interpretations of the survey tool and could have resulted in variability in reporting. To limit this impact, data collection forms had in‐built decision support to help clarify questions; most data collection was undertaken by diabetes clinicians; each site was given the opportunity to attend an information session prior to the data collection to clarify any uncertainty.
Conclusion
This study identified a difference in glycaemic outcomes between patients admitted to a digital hospital compared to a paper‐based hospital, including improved rates of good diabetes days. There are now potential quality and safety considerations for those hospitals still delivering paper‐based diabetes care. Further investigation on the mechanisms behind this improvement, any impact on hospital or patient outcomes and ways to optimise systems to support good diabetes management is required.
Code availability statement
The code that supports the findings of this study is available from the corresponding author upon reasonable request.
Supporting information
Appendix S1. Supporting Information.
Acknowledgements
The authors thank all those who collected data for the QuIDS in both 2019 and 2021 and the Digital Health Collaborative Research Centres Program. Open access publishing facilitated by The University of Queensland, as part of the Wiley ‐ The University of Queensland agreement via the Council of Australian University Librarians.
Funding: This publication is supported by Digital Health CRC Limited (‘DHCRC’). DHCRC is funded under the Commonwealth's Cooperative Research Centres (CRC) Programme.
Conflict of interest: None.
Data availability statement
Due to the strict ethical approvals provided to this study, data is not available unless approved by Queensland Health. Data that support the findings of this study may be available from the corresponding author upon reasonable request if approved by Queensland Health.
References
- 1. International Diabetes Federation . IDF Diabetes Atlas. Belgium: International Diabetes Federation; 2021. [Google Scholar]
- 2. Australian Institute of Health and Welfare . Diabetes: Australian Facts. Canberra: Australian Institute of Health and Welfare; 2023. [Google Scholar]
- 3. Australian Diabetes Society . Guidelines for Routine Glucose Control in Hospital. Sydney, Australia: Australian DIabetes Society; 2012. [Google Scholar]
- 4. Donovan P, Eccles‐Smith J, Hinton N, Cutmore C, Porter K, Abel J et al. The Queensland Inpatient Diabetes Survey (QuIDS) 2019: the bedside audit of practice. Med J Aust 2021; 215: 119–124. [DOI] [PubMed] [Google Scholar]
- 5. Sly B, Russell AW, Sullivan C. Digital interventions to improve safety and quality of inpatient diabetes management: a systematic review. Int J Med Inform 2022; 157: 104596. [DOI] [PubMed] [Google Scholar]
- 6. Queensland Health . Queensland reporting hospitals. 2022. [cited 2023 April 22]. Available from URL: http://www.performance.health.qld.gov.au/Hospital/Index/99999
- 7. Centre HaSCI . National Diabetes Inpatient Audit (NaDIA). London: NHS; 2011. [cited 2023 April 16]. Available from URL: https://digital.nhs.uk/data-and-information/publications/statistical/national-diabetes-inpatient-audit/national-diabetes-inpatient-audit-nadia-2011. [Google Scholar]
- 8. Queensland Health . Clinical Services Capability Framework: Fundamental of the Framework. Brisbane: Queensland Health; 2015. [Google Scholar]
- 9. Roumeliotis N, Sniderman J, Adams‐Webber T, Addo N, Anand V, Rochon P et al. Effect of electronic prescribing strategies on medication error and harm in hospital: a systematic review and meta‐analysis. J Gen Intern Med 2019; 34: 2210–2223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Setra A, Jani Y. A longitudinal assessment of the quality of insulin prescribing with different prescribing systems. Pharmacy 2021; 9: 53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Eden R, Burton‐Jones A, Scott I, Staib A, Sullivan C. Effects of eHealth on hospital practice: synthesis of the current literature. Aust Health Rev 2018; 42: 568–578. [DOI] [PubMed] [Google Scholar]
- 12. Keasberry J, Scott IA, Sullivan C, Staib A, Ashby R. Going digital: a narrative overview of the clinical and organisational impacts of eHealth technologies in hospital practice. Aust Health Rev 2017; 41: 646–664. [DOI] [PubMed] [Google Scholar]
- 13. NHS Digital . National Diabetes Inpatient Audit England and Wales, 2017. London: NHS; 2018. [Google Scholar]
- 14. Bansal V, Mottalib A, Pawar TK, Abbasakoor N, Chuang E, Chaudhry A et al. Inpatient diabetes management by specialized diabetes team versus primary service team in non‐critical care units: impact on 30‐day readmission rate and hospital cost. BMJ Open Diabetes Res Care 2018; 6: e000460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Kyi M, Colman PG, Wraight PR, Reid J, Gorelik A, Galligan A et al. Early intervention for diabetes in medical and surgical inpatients decreases hyperglycemia and hospital‐acquired infections: a cluster randomized trial. Diabetes Care 2019; 42: 832–840. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1. Supporting Information.
Data Availability Statement
Due to the strict ethical approvals provided to this study, data is not available unless approved by Queensland Health. Data that support the findings of this study may be available from the corresponding author upon reasonable request if approved by Queensland Health.