An Evaluation of Methods of Assessing Impaired Awareness of Hypoglycemia in Type 1 Diabetes

A total of 140 participants were recruited; 80 completed the study. Those who completed the study were significantly older than those who did not (n = 60) (mean ± SD age 47.6 ± 12.7 vs. 41.1 ± 12.6 years, respectively, P = 0.04). No differences in duration of diabetes (P = 0.7) or in glycemic control (P = 0.35) were observed between these two groups. All completed a questionnaire to assess awareness of hypoglycemia using each of the methods presented by Gold et al. (7), Clarke et al. (8), and Pedersen-Bjergaard et al. (9). The participants were then asked to perform capillary blood glucose measurements (using their own blood glucose meters) four times daily, prospectively over a 4-week period. When any blood glucose value <3 mmol/l (54 mg/dl) was recorded, the subjects were asked to complete a validated symptom questionnaire, the Edinburgh Hypoglycemia Score (10), to document the nature (autonomic, neuroglycopenic, or malaise) and the intensity of the hypoglycemic symptoms that were experienced. Completed diaries and information sheets (n = 80) were returned at the conclusion of the monitoring period.

The Gold method (7) poses the question “do you know when your hypos are commencing?” The respondent then completes a 7-point Likert scale, with 1 representing “always aware” and 7 representing “never aware”. A score of ≥4 implies impaired awareness of hypoglycemia.

The Clarke method (8) comprises eight questions characterizing the participant's exposure to episodes of moderate and severe hypoglycemia. It also examines the glycemic threshold for, and symptomatic responses to, hypoglycemia. A score of four or more implies impaired awareness of hypoglycemia.

The Pedersen-Bjergaard method (9) requires the patient to respond to the question “can you feel when you are low?” requiring the selection of one response from “always,” ”usually,” “sometimes,” or “never.” Only patients who answer “always” are considered to have normal symptomatic awareness of hypoglycemia; the others are designated as having impaired or absent awareness.

Differences between groups (normal awareness vs. IAH) were analyzed using the two-sample t test/Mann-Whitney U test or the χ²/Fisher's exact test. To assess the linear relationship between two variables, a Spearman rank correlation coefficient was calculated. All analyses were performed using SPSS, version 12.0, for Microsoft Windows.

The prevalences of IAH as identified by the Gold, Clarke, and Pedersen-Bjergaard methods were 24, 26, and 62.5%, respectively. A strong association, using Spearman's test, was found between the Gold and Clarke methods for identifying impaired awareness (r_s = 0.868, P = 0.001). If the Pedersen-Bjergaard method was revised to include “always and usually” representing normal awareness and “occasionally and never” representing IAH in response to the question “can you feel when you are low?” the percentage of IAH fell substantially to 15.4%. A poorer correlation was also demonstrated between this revised method and the other methods of assessment (Gold r_s = 0.531, Clarke r_s = 0.536).

Those patients with IAH identified by the Gold method (P = 0.001) and the Clarke method (P = 0.007) were significantly older than those with normal awareness. No such age difference was observed using the Pedersen-Bjergaard method (P = 0.10). The duration of diabetes was significantly longer in the IAH group of patients using all three methods, but no statistical difference was observed in A1C between the two groups, subdivided by state of awareness.

Data regarding the frequency of biochemical hypoglycemia can be found in Table 1. The patients designated as having IAH using the Gold and Clarke methods reported a significantly higher number of episodes of biochemical hypoglycemia over the 4-week monitoring period than those considered to have normal awareness. No statistical differences were observed between the two subgroups using the Pedersen-Bjergaard method (P = 0.06). During this period, the reported intensity of autonomic symptoms was lower during biochemical hypoglycemia in those in whom IAH had been identified using the Clarke and Gold methods compared with patients designated as having normal awareness. No symptomatic differences were observed between the groups identified using the Pedersen-Bjergaard method (P = 0.22). Using all three methods, no statistical difference was observed between the groups in self-reported neuroglycopenic symptoms. The mean incidence of severe hypoglycemia in the year preceding the study was statistically different between those identified as having IAH compared with those with normal awareness using all three methods.

In the present study, the three methods currently available to assess symptomatic awareness of hypoglycemia were evaluated for their concordance in identifying impaired awareness of hypoglycemia. In the present randomly selected cohort of adults with type 1 diabetes, equivalent prevalences of impaired awareness (24 and 26%, respectively), with a strong correlation (r_s = 0.868), were obtained with two of the methods (Gold and Clarke). This is consistent with previous population surveys, which have suggested that, based on clinical history, ∼25% of unselected adults with type 1 diabetes have some form of this acquired syndrome (4,11,12). A much higher (62.5%) prevalence was observed using the method of Pedersen-Bjergaard. Differences between the methods were also apparent with respect to patients considered to be at high risk of impaired awareness. With the Clarke and Gold methods, the patients identified as having IAH were older, had a longer duration of diabetes, had experienced more episodes of severe hypoglycemia during the preceding year, and recorded frequent mild biochemical hypoglycemia during the monitoring period. Those with IAH according to the Gold and Clarke methods had significantly lower autonomic and nonsignificantly higher neuroglycopenic symptom scores during hypoglycemia compared with those with intact awareness, which are recognized characteristics of this syndrome (7). The Pedersen-Bjergaard method appears to overestimate the prevalence of IAH and identified only a long duration of diabetes and a history of previous severe hypoglycemia as characteristics relevant to those who had impaired symptomatic awareness.

When methods that utilize questionnaires are used to ascertain awareness of hypoglycemia, some overlap may occur. No currently available method can be considered to be fully reliable and valid. However, the Pedersen-Bjergaard method to identify patients with impaired awareness of hypoglycemia offers too simplified an approach to this complex clinical condition and appears to be insensitive and undiscriminating, thus overestimating its prevalence. It cannot therefore be endorsed for routine clinical use.

In conclusion, for clinical and research use, the Clarke and Gold methods should be used preferentially, either separately or in combination, to identify people with type 1 diabetes who have impaired awareness of hypoglycemia.

J.G. was supported by a research grant from the Chief Scientist Office of Health of the Scottish Executive.