Inpatient diabetes care requires adequate support, not just HbA_1c screening

Routine admission screening of patients is desirable, but financial and personnel support for diabetes services is essential

Hyperglycaemia is common among hospitalised patients and is associated with increased complication rates, longer admissions, more frequent re‐admissions, and higher mortality.1,2,3,4 Its detection can also be an opportunity for identifying undiagnosed diabetes. Trials have confirmed that improving inpatient glycaemic control reduces complication rates,2 and guidelines recommend that the blood glucose levels of all patients be checked on admission; if hyperglycaemia is detected, glycated haemoglobin (HbA_1c) should also be measured.2,5 HbA_1c assessment can help differentiate stress‐induced hyperglycaemia from undiagnosed diabetes, and also aids evaluation of treatment adequacy in people with known diabetes. Nevertheless, the identification of undiagnosed diabetes in hospitalised patients is sub‐optimal.

In this issue of the MJA, Cheung and colleagues report their cluster randomised trial of an intervention in which HbA_1c was automatically measured in all adult patients admitted to hospital from emergency departments (EDs) with blood glucose levels of 14 mmol/L or more, and the hospital diabetes service was notified of their hyperglycaemia.6 The primary outcome of the trial was documentation of both a new diagnosis of diabetes and a corresponding follow‐up plan. Secondary outcomes included new diabetes diagnoses (with or without plans), length of stay, re‐admission rate, and mortality. While the results for all outcomes were similar in the intervention and control arms, there is still much to be learned from this study.

The findings should not be misinterpreted as arguing against routine glucose screening in the ED. Routine assessment of blood glucose in patients being admitted to hospital is an important component of acute care. However, the goal is not only to detect new cases of diabetes but to screen for clinically significant hyperglycaemia or hypoglycaemia, as each has implications for acute management and monitoring of the patient.

The specific intervention in the study by Cheung and colleagues was automatic HbA_1c measurement and notification of hospital diabetes services of patients with marked hyperglycaemia. Diabetes services in intervention hospitals were not provided with extra resources or guidance on how to manage the additional workload. We do not know whether the notifications led to more inpatient reviews by specialist services or to improved inpatient glycaemic control. As Cheung and colleagues note, it is quite possible that hospital diabetes services were already working at full capacity, so that the lack of improvement in primary or secondary outcomes may be more related to a failure of systems of care than to shortcomings of the screening protocol.

Interestingly, the primary outcome (new diabetes diagnosis and follow‐up care plan) was achieved less frequently in the intervention hospitals (16%) than in the control hospitals (26%), with an unadjusted odds ratio of 0.55 (P = 0.001). The two groups did not differ with respect to new diabetes diagnosis rate (that is, with or without plans: 31% in both arms), suggesting that documentation of follow‐up was better at control sites. The only hospital with a dedicated inpatient diabetes team had been randomised to the control arm of the study; perhaps the improved care infrastructure at this site contributed to better implementation of follow‐up. Another possibility is that an automatic notification system without adequate financial and personnel support is potentially harmful; admitting teams are perhaps less likely to pay attention to hyperglycaemia if they perceive diabetes management to be the responsibility of another unit.

Further potential systems problems are highlighted by Cheung and colleagues. First, laboratories were only able to assess HbA_1c in 53% of patients in the intervention hospitals. Insufficient blood may have been available in ethylenediaminetetraacetic acid (EDTA) tubes for many patients. Another consideration not noted by the authors is that at some hospitals the EDTA tube is routinely sent to the haematology laboratory for a full blood examination, and it can be difficult for the biochemistry department to gain access to it for measuring HbA_1c. Second, poor documentation of follow‐up plans in the discharge summary may reflect recognised problems with the clinical handover process.7 Clinical notes are typically written by overburdened junior clinicians who sometimes have only been peripherally involved in the care of the patient.

A recent trial in Melbourne compared the outcomes of early identification by a specialist inpatient diabetes team of inpatients with known diabetes or a random blood glucose level exceeding 11.1 mmol/L with the outcomes of a traditional referral‐based system.8 The specialist team was able to provide daily consultations. The intervention improved inpatient glycaemic control and reduced the rate of hospital‐acquired infections, showing that, with adequate systems support and resources, early identification and notification of patients with hyperglycaemia can lead to meaningful changes in patient outcomes.

Improving the detection of previously unrecognised diabetes and care for people with diabetes in hospital requires innovative and comprehensive systems‐based solutions that take a range of questions into account, including fragmentation of care, communication between health professionals, and adequate resourcing of clinical services. In isolation, automatic HbA_1c assessment and notification are unlikely to have a positive impact.