National diabetes awareness programs often emphasize the importance of symptoms in undiagnosed diabetes. Here we describe the frequency of diabetes symptoms in a multiethnic community in New Zealand.
Randomly selected nondiabetic European, Maori, and Pacific residents were asked if any of six cardinal symptoms of hyperglycemia (thirst, weight loss, polyuria, boils, tiredness, and blurred vision) were present. Subjects were screened using random venous blood sampling as previously described (6). Those with a random glucose ≥6.5 mmol/l within 2 h of a meal, or ≥6.0 2 h after a meal, and a random 20% of others were invited to attend a 75-g, 2-h oral glucose tolerance test (OGTT) (1999 World Health Organization criteria for diabetes, impaired glucose tolerance [IGT], and impaired fasting glucose [IFG]). Those diagnosed elsewhere, or with no OGTT but a random glucose of ≥11.1 mmol/l, have been included in the undiagnosed diabetes group. Odds ratios (ORs) and 95% CIs are shown.
Of the 2,423 people invited to participate, 1,585 (65.4%) were interviewed. Overall (and after removing those with new diabetes), Maori and Pacific individuals were significantly more likely to have more than two symptoms (Europeans 4.2% [3.8%] vs. 14.3% [13.5%] and 9.4% [9.0%], both P < 0.001, respectively). Each of the individual symptoms was present significantly less frequently among Europeans.
Of the 786 invited to the 75-g OGTT, 534 (67.9%) attended. Symptoms were uncommon among those with new diabetes, but still 2.6-fold (1.4–4.8) as likely when compared with IGT/IFG or normal subjects (e.g., more than two symptoms present in 16.9% vs. 6.5% and 7.5%, respectively, P < 0.001). This is a similar frequency to that found in the Australian Diabetes Screening Study (2) but less than in the Hoorn study (3). IGT/IFG and normal subjects had very similar symptom frequencies. A logistic regression for each symptom revealed that after adjusting for ethnicity, age, sex, and BMI, only thirst was significantly associated with hyperglycemia, and then only at ≥17.0 mmol/l vs. <5.0 mmol/l, OR = 9.2 (3.2–26.6). In a similar analysis, more than two symptoms were 3.5-fold (1.2–10.1) more common among those with a random glucose ≥11.0 mmol/l. When diabetes replaced random glucose in each regression, thirst (2.0 [1.0–4.0]), polyuria (1.8 [1.0–3.2]), tiredness (1.9 [1.0–3.4]), and having two symptoms (1.9 [1.1–3.4]) were significant entrants. Obesity was not associated with excess symptoms.
We conclude that few individuals with undiagnosed diabetes had symptoms, and that symptoms were only marginally increased in undiagnosed diabetes. Furthermore, symptoms were not increased in IGT/IFG. Ethnic differences in symptoms exist, confounding their interpretation in some settings. Our data suggest that unless a patient presents acutely, using prompted symptomatology to decide who should be screened for dysglycemia using blood testing is futile.
Article Information
We are grateful to Servier for funding a research fellow (C.F.T), Auckland Medical Research Foundation, Maurice and Phyllis Paykel Trust, Hector Trust, SmithKline Beecham, and the members of the Diabetes Projects Trust.
