JPM | Free Full-Text | Preoperative Glycosylated Haemoglobin Screening to Identify Older Adult Patients with Undiagnosed Diabetes Mellitus—A Retrospective Cohort Study

Diabetes mellitus (DM) is one of the most prevalent comorbidities in surgical patients and is associated with perioperative morbidity [1,2,3,4,5]. Perioperative care providers can take measures to prevent perioperative dysglycaemia in patients with DM, e.g., by preoperative dose adjustments for diabetes medications and regular blood glucose monitoring. However, DM may not always be diagnosed at the time of surgery, as the lack of symptoms during the early course of type 2 DM can delay its diagnosis considerably [6]. Extrapolations from population-based studies indicate that over 30% of patients with DM between the ages of 20 and 79 years living in high-income countries in Europe are currently undiagnosed [7]. Since the risk of developing DM increases with age, it is likely that a substantial number of older adult patients visiting the preoperative assessment clinic have yet undiagnosed DM. Identifying these patients is relevant, because hyperglycaemia is associated with a worse in-hospital outcome in patients with undiagnosed DM, compared to those with known DM [8,9,10,11].

The diagnosis of DM is based on criteria for plasma glucose or glycosylated haemoglobin (HbA1c) [12,13,14]. Plasma glucose criteria include the fasting plasma glucose (FPG) value and the two-hour plasma glucose value during an oral glucose tolerance test (OGTT), which may not be feasible at the time of preoperative screening, as both tests require at least 8 h of prior fasting. Formed by the glycation of the haemoglobin protein, HbA1c reflects the average glycaemic control over the last three months (i.e., the average life span of red blood cells) and is a predictor of both perioperative glucose control and adverse outcome after surgery [2,15,16,17,18,19], even within the prediabetes range (39–47 mmol·mol⁻¹) [15]. Although the preoperative measurement of HbA1c in patients with DM is recommended in several perioperative guidelines [20,21,22], it is not commonly used as a screening tool for DM during preoperative assessment.

In Europe, the estimated prevalence of DM in people over 65 years of age is 27.8% [23]. We hypothesised that preoperative screening for DM by the routine measurement of HbA1c in older adult patients would identify patients with undiagnosed DM. We also aimed to evaluate the relationship between HbA1c and postoperative outcome, measured as days alive and at home within 30 days after surgery (DAH30), in patients without a history of DM.

From January to December 2019, 2015 patients aged ≥65 years underwent preoperative screening for elective non-cardiac surgery at our hospital. Duplicates were removed (n = 78) and patients with a diagnosis of DM (n = 404) were excluded, leaving 1533 patients without a history of DM. Of these, 697 had undergone HbA1c testing within 3 months before surgery (Figure 1). Demographic, clinical and surgical characteristics are shown in Table 1. Compared to patients without a recent HbA1c, the group of patients with an HbA1c within 3 months before surgery contained fewer patients with an ASA score of ≥3 and fewer patients undergoing major surgery (Table S1).

The mean preoperative HbA1c was 38.6 (±4.9) mmol·mol⁻¹. We identified 26 cases of undiagnosed DM based on HbA1c, which corresponds to a prevalence of 3.7% (95%CI 2.5–5.4%) and translates into 27 as the number of patients needed to screen in order to detect one patient with undiagnosed DM. There were 299 subjects with a preoperative HbA1c in the prediabetes range, amounting to a prevalence of 42.9% (95%CI 39.2–46.7%) (Table 2).

In all cases of DM among the general population, the International Diabetes Federation (IDF) has estimated that 45% are undiagnosed in Europe [7]. One would expect the prevalence of undiagnosed DM in an older adult surgical population to be higher than the prevalence we found in our study population. However, a considerable number of patients seen at the preoperative screening clinic may have already undergone previous blood tests, potentially reducing the prevalence of undiagnosed DM among our surgical population compared to the general population. Several studies reported on the prevalence of undiagnosed DM in patients who planned to undergo surgery based on an HbA1c ≥ 48 mmol·mol⁻¹. The majority of these studies included patients undergoing a specific type of surgery [32,33,34,35,36,37,38,39,40,41], such as cardiac [32,33,34,35], orthopaedic [36,37] and bariatric surgery [38,39]. In a prospective study conducted in a university hospital in Canada, as much as 7% of patients aged 18 years or older undergoing elective non-cardiac surgery were found to have undiagnosed DM based on their preoperative HbA1c value [27]. This difference in prevalence may be explained by their exclusion of ambulatory surgery, the relatively high number of patients classified as ASA 3–4 and geographical differences. Data from two other observational studies suggest a prevalence of undiagnosed DM similar to the prevalence in our study sample, i.e., 3.9% and 4.3%, despite the broader age range among participants in these studies [2,28]. Although the overall prevalence of DM increases considerably in each incremental age group [23], the proportion of diagnosed DM may be higher in older adults than in younger age groups.

Although the diagnosis of DM is based on either the plasma glucose or HbA1c criteria [12,13,14], the measurement of the HbA1c value is likely the most convenient option in the preoperative assessment clinic, as it does not require fasting and is not affected by factors such as timing, stress or diet. Nonetheless, it is important to note that there is considerable discordance in the diagnosis of DM when comparing the HbA1c and plasma glucose criteria, as these tests measure different aspects of glucose metabolism [33,42]. An HbA1c above the diagnostic threshold may only detect one-third of the cases of undiagnosed DM that would otherwise be identified based on FPG and OGTT testing [43]. This may at least partly explain why the prevalence of undiagnosed DM that we found was lower than we expected based on previous population estimates.

The association between preoperative HbA1c and postoperative outcome, specifically in the surgical patients without a diagnosis of DM, has been examined in several previous studies. Some studies, although not all [35,39,44], have found an association between HbA1c and various postoperative complications [9,45,46,47,48,49], length of hospital stay [50] and even mortality [46] (Table S2). It has been proposed that the risk of certain postoperative complications may be higher in patients with undiagnosed DM compared to patients with a history of DM [9,10,11]. In a retrospective cohort study that included patients undergoing peripheral arterial revascularization, subjects without a diagnosis of DM and an HbA1c above 53 mmol·mol⁻¹ were found to be at a higher risk of amputation and adverse limb events compared to subjects with known DM and a similar HbA1c [9]. In another retrospective analysis, patients with undiagnosed DM were found to have a higher one-year mortality rate after non-cardiac surgery compared to patients with known DM [10]. However, FPG was used besides HbA1c to identify patients with undiagnosed DM in this study [10]. Similar observations have been made in critically ill patients with undiagnosed DM, who seem to have higher rates of dysglycaemia and increased mortality compared to patients with a diagnosis of DM [11]. We found no predictive value of HbA1c in relation to DAH30.

Prediabetes is a heterogeneous condition characterised by glucose values or an HbA1c above the reference interval, yet below the DM diagnostic threshold. The criteria to diagnose prediabetes proposed by major international organisations are not uniform. The ADA and the National Institute for Health and Care Excellence guidelines have incorporated HbA1c as a means of diagnosing prediabetes [12,51], whereas the IDF and the World Health Organization define prediabetes based on glucose criteria only [23,52]. Prediabetes can progress to type 2 DM. However, prediabetes has no known relevance for the perioperative outcome. This raises the question of whether preoperative screening by HbA1c would contribute to public health, or if one should leave this to primary health care providers. Based on our data, in the Netherlands, the costs of detecting one patient with undiagnosed DM at the preoperative assessment clinic would be around GBP 255 (EUR 300), while it is unknown whether doing so would prevent any potential complications. However, there is some evidence that screening for (pre)diabetes could be effective in terms of long-term health benefits and cost-effectiveness [53].

The major strength of this study is its specific focus on older adult patients undergoing a diverse range of surgical interventions. However, some limitations in the study design should be noted. First, the exclusion of patients with a diagnosis of DM was solely based on the information from their electronic health records, though these records have proven to be reliable in earlier studies [54]. Second, although we implemented routine age-dependent screening of HbA1c in our preoperative assessment clinic, the majority of the patients who satisfied the criteria were not tested, which limited the sample size and may have introduced selection bias by excluding patients who were too ill to attend preoperative assessment or, conversely, were relatively healthy and were thus screened by telephone. Considering that the patients who were screened for HbA1c had a relatively lower ASA score and underwent fewer major surgical procedures compared to those who did not undergo HbA1c screening, we might have underestimated the prevalence of prediabetes and diabetes. Third, although we analysed the association between HbA1c and DAH30, the study was not powered to detect differences in this postoperative outcome. Finally, DAH30 was exclusively based on the hospitalisation data from our centre. Nonetheless, it is unlikely that post-discharge readmissions at other care facilities would have significantly influenced the study outcome, as postoperative patients will be referred to the initial clinic in case of postoperative complications within 30 days.