Diabetes in British Nursing and Residential Homes: A pragmatic screening study

Newcastle upon Tyne has a population of ∼260,000, of whom 3.5% have diabetes. There are 68 care homes in the city, comprising residential care homes offering 24-h social support and nursing homes offering additional nursing care. Residential and nursing homes for the elderly mentally infirm (EMI) provide specialist care for older adults with dementia. We recruited residents from all four types of care home.

After consent/assent was obtained, all volunteers gave 200 μl of capillary whole blood, for glucose estimation, using a glucose analyzer (glucose dehydrogenase method; Hemocue, Derbyshire, U.K.). Both fasting glucose and 2-h PPG were estimated. World Health Organization diagnostic criteria were followed: for impaired fasting glycemia (fasting glucose ≥100 mg/dl) and for diabetes (fasting glucose ≥110 mg/dl or random [in this case 2-h PPG] ≥200 mg/dl) (5). Recorded weight and blood pressure were documented. Subjects, carers, and family doctors were told the results as appropriate.

A total of 1,630 residents were recruited, of whom 186 had previously diagnosed diabetes. Data on 111 subjects from six care homes screened during the pilot phase were not included in the final analysis, and 58 subjects refused to take part. Thus, at least one glucose measurement was obtained on 1,275 subjects: 86 had fasting glucose only, 212 had PPG only, and 977 had both.

Statistical methods used included ANOVA, χ²goodness-of-fit test, and CIs, which were estimated using a desktop computer (Stata 6; StataCorp, College Station, TX). The local ethics committee approved the study.

Results of screening are presented in Table 1 by care-home type. Diabetes (P = 0.005) but not IFG was more common in EMI homes. Comparing residential care alone, diabetes was more common in EMI homes (P = 0.001). For nursing homes, IFG was more common in EMI homes (P = 0.02). Differences remained after adjustment for age.

Of l,275 subjects with at least one screening test, 105 had undiagnosed diabetes (prevalence of 8.2%). Thus, with 186 having previously diagnosed diabetes, 291 subjects with diabetes were found among 1,461 subjects, giving a diabetes prevalence of 19.9%.

Using the results of both fasting glucose and PPG measurements as gold standard in 977 subjects with both results, fasting glucose alone had a sensitivity of 71% and a negative predictive value of 97%. PPG had a sensitivity of 43% and a negative predictive value of 95%. Subjects with diabetes were 3.7 kg (95% CI 0.4–7.0, P = 0.03) heavier than nondiabetic subjects. Mean weight was 56.5 kg for those with diagnostic PPG alone and 68.5 kg for those with diagnostic fasting glucose and PPG (difference of 12 kg; 95% CI 0.4–24.0, P = 0.04). The mean weight for those with raised fasting glucose alone was 63.1 kg.

Screening for diabetes among care-home residents using PPG and fasting capillary glucose achieves excellent population coverage with a simple bedside test. Undiagnosed diabetes was common, especially in those with dementia, where it was seen in up to 13%.

We acknowledge potential limitations in this study. Failure to fast could cause false-positive results. However, investigators arrived early at the homes, before subjects were out of bed, to ensure that fasting results are valid. Consumption of breakfast was encouraged but the carbohydrate load may not have raised PPG to diagnostic levels, resulting in underestimation of diabetes prevalence. Finally, although plasma glucose is now advocated for the diagnosis of diabetes (6), measuring capillary glucose optimized population coverage and diagnostic thresholds were those of the World Health Organization and American Diabetes Association at the time.

American Diabetes Association guidelines advocate fasting glucose for the routine diagnosis of diabetes (6), but PPG (using the threshold for casual glucose of 200 mg) increased pick up. There is a physiological basis for use of PPG particularly in the elderly (7). This is especially relevant for leaner diabetic subjects with possible failure of insulin release in response to a dietary load, whom we found less likely to have diagnostic fasting glucose alone levels. An OGTT would have been a gold standard, but poor adherence among frail elders has been observed (2), and screening may not have been completed with the added work and cost involved in using an OGTT.

The interpretation of IFG found in this population is not clear, as the link to cardiovascular risk and progression to diabetes may be less robust in the elderly (8). We therefore believe that more research is required on the natural history of abnormal glucose tolerance in the elderly so that appropriate clinical advice might be offered.

Finally, we found worryingly high rates of undiagnosed diabetes among care-home residents with dementia. Epidemiological evidence supports an association of diabetes with cognitive impairment (9,10), and the risk of diabetes with newer antipsychotic agents may also be relevant, as they are frequently prescribed in this group (11).

In conclusion, ∼20% of care-home residents were identified with diabetes (compared with 3.5% from the district diabetes register). For diabetes screening in this population, we recommend that fasting glucose be augmented by PPG estimation, particularly in the leaner elderly population. Targeted screening of elderly residents with dementia is also likely to identify the highest rates of undiagnosed diabetes.