Role of capillary ketone testing in the diagnosis and monitoring of diabetic ketoacidosis

Diabetes has become a major worldwide epidemic being the 4th leading contributor towards mortality in developed countries.1 Pakistan occupies 7th place on the diabetes prevalence list.2 Pakistan has an estimated prevalence of 12% in the adult population, with significant differences across the country in prevalence rates according to geography and demography.3, 4 A rise in prevalence results in a growing number of hospital admissions with diabetes related emergencies.5, 6 In any clinical setting, it is important to differentiate the hyperglycaemic hyperosmolar state of type 2 diabetes (which may be accompanied by ketosis) from diabetic ketoacidosis which predominantly occurs in type 1 diabetes, both of which are potentially life-threatening emergencies. Although urinary ketone testing was the more commonly employed investigation, capillary ketone testing which checks the presence of beta-hydroxybutyrate, 1 of the 3 products of fatty acid metabolism in blood (acetone, acetoacetate and beta-hydroxybutyrate), is a recently introduced investigation which has shown a lot of promise in revolutionising the diagnosis of diabetic ketoacidosis and its subsequent management.7, 8 Several studies have strengthened the case for using capillary ketone testing and established good correlations with the clinical condition of patients.9 In this study, we aimed to determine whether capillary ketone testing can provide an easy to use test to distinguish those patients with diabetic ketoacidosis who definitely need hospital admission from those patients with non-ketotic hyperglycaemia who may not require admission, in a resource constrained country. Thus, the aim of the study was to investigate the diagnostic predictability of capillary ketone concentration as a marker of diabetic ketoacidosis. This is a retrospective study looking into the data of 109 consecutive cases of patients who presented to the department of diabetes and endocrinology, Postgraduate medical institute, Hayatabad medical complex, Peshawar, in the province of Khyber Pukhtoonkhwa, Pakistan, via the accident and emergency and outpatient department with symptoms and signs suggestive of diabetic ketoacidosis. This is a tertiary care hospital catering to the health needs of residents of Peshawar and surrounding areas (Figure 1). The department subjected all patients who presented with symptoms of hyperglycaemia and clinical signs of diabetic ketoacidosis to capillary ketone testing. A precision Xtra-Meter manufactured by Abbott Australia (Unit D, 31-33 Sirius Road, Lane Cove, New South Wales 2066, Australia) was employed to this end. Those testing positive for blood ketones were included in the study. Using the manufacturer recommended cut-off value of 0.8 mmol/L, patients testing positive underwent arterial blood gas sampling. Those fulfilling the parameters laid down by the consensus criteria6 including blood glucose >13.9 mmol/L, pH of <7.3 and HCO3 of <18 mmol/L were deemed to have diabetic ketoacidosis. The remainder of ketone positive individuals served as controls. The ketone meters were checked every 2 months by personnel from Abbott (Australia) to ensure they were working within the recommended parameters. Anthropometric and metabolic data are expressed as geometric means with standard deviation (SD) or 95% confidence interval (95%CI). Comparisons between 2 groups used Mann-Whitney testing. Kernel density estimation available at Free Statistics Software, Office for Research Development and Education, version 1.1.23-r7, URL (http://www.wessa.net/) was used to plot the frequency distributions for all participants. The department receives patients from the entire province, adjoining Punjab and neighbouring Afghanistan. Data for 109 patients could be included in the study. Forty-five patients formed the diabetic ketoacidosis group and 64 patients served as controls. Patients with type 1 diabetes formed the majority in the diabetic ketoacidosis group (30/45), while patients with type 2 diabetes formed the majority in the non-ketoacidotic hyperglycaemia category (43/64). The major precipitating factor in the category suffering from diabetic ketoacidosis group was poor treatment adherence or symptoms related to poor control (26/45), while for the non-ketoacidotic hyperglycaemia category it was intercurrent infections (22/43). Importantly, failure to initiate insulin at the initial diagnosis of diabetes (ie ketosis was missed) led to the development of diabetic ketoacidosis, in 4 patients. The main symptoms at presentation were excessive thirst, polydipsia and weight loss. Five deaths were noted in all. Three deaths were in the diabetic ketoacidosis category with only one because of severe acidosis and the other two were the result of comorbidities superimposed on debilitated states of the patients, stroke in one and lower respiratory tract infection in the other. Two deaths were observed in the non-diabetic ketoacidosis category and both were attributed to coexistent septicaemia. The mean of capillary ketones measured in patients with diabetic ketoacidosis with a prior diagnosis of type 1 diabetes was 4.6 mmol/L (geometric mean) (95%CI 4.0-5.1) similar to those with a prior diagnosis of type 2 diabetes also at 4.6 mmol/L (95%CI 3.5-5.6) as distinct from the means of capillary ketones measured in patients in non-ketoacidotic hyperglycaemia group, 0.87 mmol/L (95%CI 0.3-1.5) in type 1 diabetes and 0.74 mmolL (95%CI 0.4-1.1) in type 2 diabetes (difference was statistically significant at the P<.0001 level as calculated on Mann-Whitney test). The kernel density plots for capillary ketone level for patients with and without diabetic ketoacidosis are shown in Figures 2 and 3. In relation to the diagnosis of diabetic ketoacidosis using a cut-off for blood ketones of 0.8 mmol/L, the test was found to have a sensitivity of 100% and specificity of 68.8%. Raising the cut-off to 1.5 mmol/L yielded a higher specificity of 82.8% but the sensitivity dropped to 95.2%. Thus, although a higher cut-off improved specificity in diagnosing patients with diabetic ketoacidosis, a lower cut-off ensured that no patients with this life threatening condition were missed. The positive predictive value of the test using the 0.8 mMol/L cut-off was found to be 67%, while the negative predictive value of was 100% for the same cut-off limit. Our study establishes the role of capillary ketone testing in diagnosing diabetic ketoacidosis using a cut-off of 0.8 mmol/L with a negative predictive value of 100%. Its application in early triage of patients with hyperglycaemia and subsequently monitoring of resolution of diabetic ketoacidosis, indicates potential benefit in improving patient safety. Our study clearly shows the predominance of type 1 diabetes patients presenting as diabetic ketoacidosis. This has already been evidenced both in Pakistan and worldwide.10, 11 The number of patients previously thought to have type 2 diabetes presenting in diabetic ketoacidosis was quite significant, suggesting that a proportion of patients were previously misdiagnosed. A literature search revealed this to be a not uncommon finding.12 However, an underlying precipitating factor is nearly always present in such cases.13 In our study, the major precipitant was poor compliance with patients stopping insulin for 3 months in one case. This has also been found in other studies.14 Late diagnosis and under treatment are significant contributors towards prolonging morbidity and mortality especially in a resource challenged country like Pakistan where implementation of international standards is difficult. In our study, patients presented from different districts of the province, in some cases from the adjoining Punjab and neighbouring country Afghanistan. This lends further evidence that a test that can be administered with good reliability and reproducibility is needed to reduce the risk of patients with life threatening diabetic ketoacidosis not being triaged appropriately, particularly when they present to community based clinics, sometimes far from specialist centres. Introduction of capillary ketone testing and its reliability in differentiating diabetic ketoacidosis from non-ketoacidotic hyperglycaemia is major step towards providing immediate care to patients in low-income countries. Our results show a clear separation between capillary ketone levels in diabetic ketoacidosis compared with non-diabetic ketoacidosis (Figures 2 and 3). This study provides further evidence to support the introduction of capillary ketone testing as a first line investigation in patients presenting with hyperglycaemia to diagnose diabetic ketoacidosis. In conclusion, capillary ketone testing is an ideal tool for diagnosing diabetic ketoacidosis in a resource limited country like Pakistan. Application of this technology will save lives. No author has any conflict of interest in relation to the findings presented here.