DISCUSSION
Hypoglycaemia, severe hypoglycaemia and IAH are important clinical problems. Hypoglycaemia is the major limiting factor in achieving optimal glycaemic control2, while IAH is one of the most important risk factors for severe hypoglycaemia, increasing the risk up to six-fold5,9,14. Severe hypoglycaemia can lead to permanent cognitive impairment, cardiac arrhythmias, seizures and coma3,5,6,15,16. Other consequences of IAH include fear of hypoglycaemia, which leads to increased anxiety, decreased satisfaction and poor treatment compliance, reduced work productivity, strain in interpersonal relationships and decreased quality of life5. To compound this problem, IAH is common in people with T1D. In prior studies, depending on the study population and method of assessment, the frequency of IAH can range from 19.5% to 62.5%7,10,14,17. Therefore, assessment for both symptomatic and asymptomatic hypoglycaemia should be part of routine diabetes care in patients at risk for hypoglycaemia3,16. However, objective methods to screen for IAH are not in routine clinical use.
Based on Gold score >= 4 and Clarke score >= 4, the frequency of IAH in our study population was 28.3% and 24.6% respectively, which was similar to previous studies. In a study of 140 T1D subjects in Edinburgh by Geddes et al, using a Gold score of >=4 and Clarke score of >=4, the frequency of IAH was 24% and 26% respectively7. In a subsequent study by the same author of 518 subjects, 19.5% of subjects had IAH, based on a Gold score of >=414. A study by Choudhary et al found a similar IAH frequency of 22.1%, based on a Gold score of >=415. That said, these studies were all done in Western populations – as far as we are aware, the Gold and Clarke scores have only been studied in one prior local study, which had looked at patients with Type 2 Diabetes (rather than T1D)18.
Despite fairly similar frequency of IAH, the frequency of severe hypoglycaemia was lower in our study (32.4% and 41.4% in subjects with Gold score >= 4 and Clarke score >= 4 respectively) compared to the aforementioned studies by Geddes et al and Choudhary et al, in which the frequency of severe hypoglycaemia was 50.5 – 57.1% (in subjects with Gold score >= 4) and 57% (in subjects with Clarke score >= 4)7,14,15. Increasing age and longer duration of diabetes are well-recognized risk factors for severe hypoglycaemia1,10,19–21. In our study, the subjects had a younger age (median age 35 years) and shorter duration of diabetes (14 years) compared to the subjects in these studies (mean age 45 – 52 years, mean duration of diabetes 23 – 34 years)5,7,14,15, which might account for the observation of lower frequency of severe hypoglycaemia.
The Gold and Clarke scores are simple clinical tools which can be used to screen for IAH. The underlying basis of these scores is that T1D patients who believe they have reduced awareness of hypoglycaemia are generally correct8. Based on the literature, both Gold and Clarke scores have been validated to be associated with increased frequency of severe hypoglycaemia over the preceding one year and increased frequency of biochemical hypoglycaemia over a prospective 4 week period of SMBG5,7,8,14,15. The Clarke score also correlated with hypoglycaemic events on CGM, defined by glucose < 3.9 mmol/L and < 3 mmol/L22, which is not surprising as recurrent hypoglycaemic events leads to blunted counterregulatory hormonal responses23,24. Both Gold and Clarke scores were associated with increasing age and longer duration of diabetes – aforementioned risk factors for severe hypoglycaemia5,7,14,15, and correlated well with each other (r2 = 0.868)7,14.
In our study, IAH as defined by Clarke score >= 4 was associated with increased frequency of hypoglycaemia during the prospective 4 week period of SMBG and increased frequency of severe hypoglycaemia over the preceding one year, which was consistent with the literature. However, no significant association between the Gold score and these outcomes was found in our study. There was also no observed association between both scores and age, duration of diabetes or glycaemic control. A positive, albeit weaker correlation (r2 = 0.415) between the Gold and Clarke scores was observed.
There are several possibilities to explain the discrepancies between our study findings compared to the literature. Firstly, there is an inherent difference between the Gold and Clarke scores. The Clarke score involves eight questions while the Gold score requires just a single question. Therefore, the Gold score might be more susceptible to day-to-day variation, lower specificity and lower test-retest and inter-rater reliability25. When added to the fairly small sample size in our study, this might account for the lack of association between the Gold score and hypoglycaemic events, as well as the modest correlation with the Clarke score. Relevant to our findings, Hatle et al had found that the Clarke score was superior to the Gold score in identifying subjects at risk of severe hypoglycaemia and clinically significant hypoglycaemia25.
The HYPO score is a composite hypoglycaemia score designed by Ryan et al based on the frequency, severity and degree of unawareness of hypoglycaemia11. Its primary utility is to provide an objective discriminator of the severity of problems that occur with hypoglycaemia, in subjects who were being considered for islet transplantation11. Based on a cohort of T1D patients in Edmonton, a score of < 423 indicated that hypoglycaemia was unlikely to be a major clinical concern, a score of 423 – 1046 indicated moderate problems with hypoglycaemia while a score of >= 1047 (representing 90th percentile) indicated severe problems with hypoglycaemia11. For comparison, in our study, the median HYPO score was 6 while the 90th percentile score was 80. However, the drawback of the HYPO score is that it is labour and resource intensive, requiring subjects to perform 4 weeks of SMBG followed by data collection and calculation of the scores, and is therefore less suitable for routine clinical use. We observed an association between the HYPO and Clarke score in our study – as far as we are aware, this is the first study to directly compare the utility of the HYPO and Clarke score.
On the basis of the relatively high rates of IAH, serious consequences of severe hypoglycaemia and negligible cost of test administration, we propose the routine use of the Clarke score in all patients with T1D to identify IAH7. As described, the Clarke score is associated with the important clinical outcome of severe hypoglycaemia as well as diabetes self-management practices, which justifies its use despite its more time-consuming format. If there are limitations with its use, for instance due to time constraints in a busy outpatient clinic, we suggest that physicians obtain the Gold score despite its lower specificity25. As previously demonstrated by Pedersen et al, the distribution of severe hypoglycaemia is highly skewed, with 5% of subjects accounting for 54% of all episodes of severe hypoglycaemia17. Therefore, the benefit-risk ratio remains in favour of using these scores, which can allow the timely diagnosis of this subset of patients at high risk of severe hypoglycaemia and introduction of specific interventions to prevent serious complications.
For patients with IAH, the previous main recommendation was strict avoidance of hypoglycaemia by relaxation of glycaemic targets until restoration of glycaemic awareness3,9,26. However, this process might take several weeks to three months, lead to deterioration in glycaemic control and is difficult to achieve and sustain in clinical practice9,26. Thus, specific interventions can be taken to restore glycaemic awareness. These can broadly be divided into three groups: education, use of technology such as continuous glucose monitoring (CGM) and continuous subcutaneous insulin infusion (CSII), and pharmacological methods such as switching to insulin analogs9,24. For patients who are refractory to all other measures, islet cell transplantation can be considered24.
Our study confirms the central role of education in the potential reduction of IAH, which was previously demonstrated by the Dose Adjustment For Normal Eating-Hypoglycaemia Awareness Restoration Training (DAFNE-HART), HyPOS and HAATT trials27–30. These studies showed that in subjects with IAH, a psychobehavioural education program led to improved hypoglycaemic awareness with reductions in severe hypoglycaemia27–30.
In addition, our study clarified the content and delivery of education which is most efficacious. In our study, lower rates of IAH (based on the Clarke score) were seen in patients who had received group education, dietary education, education on self-adjustment of insulin doses and education on blood glucose targets. This is consistent with the findings by Yeoh et al that group education can improve hypoglycaemia awareness and reduce severe hypoglycaemia in up to 45% of people with IAH9. Thus, we suggest that T1D patients who have high Gold and/or Clarke scores be referred for structured diabetes education, in a group format if possible, to decrease the risk of IAH and its complications.
Our study is not without its limitations. This was a single centre study and the sample size is fairly small, which might account for the lack of association between the scores and established risk factors for IAH, as well as the modest correlation between the Gold and Clarke scores. In this study, capillary glucose testing was used, with about one-third of patients performing at least 3 blood glucose tests daily. At the time of this study, flash glucose monitoring was not available in Singapore, while continuous glucose monitoring was costly and not widely available. We acknowledge that episodes of hypoglycaemia may be missed during SMBG due to the limited frequency of testing. However, even today, capillary blood glucose testing (rather than continuous or flash glucose monitoring) remains the mainstay of home glucose monitoring for most of our patients. The continued utility of SMBG even today was well illustrated by the HypoCOMPaSS study, which showed that when subjects with IAH were provided with hypoglycaemia-focused structured education, glucose monitoring and optimised insulin replacement, there was no difference in hypoglycaemic awareness and treatment satisfaction between the different methods of glucose monitoring (SMBG vs real-time CGM)31.
Thus, as our study reflects the “real world” situation in actual clinical practice, we believe it remains relevant and provides important information for the management of T1D patients. In addition, the retrospective nature of the study in the form of a questionnaire precludes drawing conclusions about causation, for which an interventional study is required. Despite these limitations, we believe that based on this study, important recommendations which impact clinical practice can be drawn.
In conclusion, this is the first available study assessing IAH and its complications in a cohort of Singapore T1D subjects. Our study demonstrates that IAH is common and is associated with increased risk of serious complications, particularly severe hypoglycaemia. Simple questionnaires such as the Gold and Clarke scores should be routinely administered to identify patients with IAH who might benefit from specific interventions, particularly structured diabetes education, to decrease IAH and its complications.