This study used data from the population-based longitudinal study “Aging in Leganés” which began in 1993. The methodological characteristics of this study have been described elsewhere.14,15 In brief, Leganés is a city located in a metropolitan area 8km from Madrid (Spain). The census population in 1991 was 171,400 inhabitants, of whom 13,679 (8%) were over 64 years of age. “Aging in Leganés” was launched in 1993, starting with a gender- and age-stratified random sample obtained from the Leganés census of the population aged 65 years and over. In the baseline cohort 1283 persons completed the study.

Data were collected using a standardized questionnaire (a face-to-face home interview) that included sociodemographic variables, health status and history of chronic diseases.14,15

The present study uses data from the 1993 baseline, as well as the 1995, 1997 and 1999 waves to know the prevalence and incidence of diabetes and the sociodemographic and clinical variables selected for this study. Mortality at 31/12/2011 was obtained from National Death Index.

All individuals with available diabetes data were included in the analysis (1277 at baseline, 939 in 1995, 743 in 1997 and 519 in 1999) (see Flow chart for the study).

The local ethics committee approved the protocol for the Leganés study, and all participants signed an informed consent form.

History of diabetes was collected by self-reporting using two questions to the participants at baseline and in follow-up waves: “Do you have diabetes?” and “Did you have ever seen a physician in the last year for this problem?” People who answered “yes” to both the questions were considered to have diabetes and were asked to state the year when they became aware of their condition. Diabetes history was measured in years of diabetes at the time of incorporation of the individual in the baseline cohort (1993) and classified in three categories: no diabetes, less than 10 years of diabetes and 10 or more years of diabetes.

Mortality data were obtained from the National Death Index (NDI), a computerized database with information on the vital status of all residents in Spain. To compensate for the delay in updating mortality in the NDI, data were collected until June 2012. The cut-off date for the mortality analysis was 31/12/1999 (n=330; 25.8%) for medium-term mortality (6 years) and 31/12/2011 (n=994; 77.8%) for long-term mortality (18 years).

Age (years), sex, body mass index, heart disease, comorbidity, self-rated health and years of diabetes, obtained from the 1993 baseline Leganés questionnaire, were used as co-variables in the analysis.

Body Mass Index (BMI) in kg/m2 based on measures of weight and height.

Self-rated health (response to the question: “How would you describe your health at present?”: fair, poor or very poor versus good or very good).

Heart disease includes angina, myocardial infarction and/or heart failure.

Comorbidity was defined as having 4 or more chronic conditions of a list of 9 (heart disease, stroke, peripheral vascular disease (PVD), chronic respiratory disease (CRD), joint or bone pain, cancer, Parkinson's, genital or urinary pathology and gastrointestinal pathology). Chronic conditions were assessed by self-report of having been diagnosed by a medical doctor.

Due to the sampling design, prevalence and incidence figures are weighted using the age and sex distribution of the Leganés population in 1995. The data were weighted by assigning a specific weight (W) to each individual in the sample, calculated as W=N1/N:n1/n where N is the total population in Leganés, and n is the number of people in the cohort; N1 and n1 are the respective numbers in each age and sex group.

Fort the study of continuous variables, the goodness of fit to the normal distribution was analyzed using the Kolmogorov–Smirnov test; all cases fit the Gaussian curve, therefore the continuous variables were expressed as means±standard deviation, and categorical variables as absolute and relative frequencies (n and %). A bivariate analysis was made for sex with distribution of variables in the baseline using χ2 test. The prevalence of diabetes was calculated by age group (65–74 and ≥75 years) and sex for each of the four waves (1993, 1995, 1997 and 1999). To study the change in the prevalence of diabetes over time, the χ2 tendency test was used.

For the calculation of incidence, all non-diabetic individuals in the previous waves were used: 978 persons in 1995, 783 in 1997, and 566 in 1999. The incidence density (incident cases/person-time periods) was calculated for biannual periods and expressed as the number of cases of diabetes per 100 persons by 2-year time periods. In total, 1965 persons/time periods were used.

Mortality analyses were conducted with the Kaplan–Meier analysis and adjusted by the Cox regressions. To evaluate the association between diabetes and general mortality, Cox regression was used at two moments: 31/12/1999 (6 years of follow-up) and 31/12/2011 (18 years of follow-up). Median survival was calculated, and the main results were summarized with hazard ratios (HRs) and their 95% CIs. Person follow-up time was calculated from the time people were incorporated into the study in 1993. Individuals were censored if they died or dropped out of the study.

Diabetes, as a principal variable, was used as a dummy variable (fictitious variable): (0) no diabetes, (1) less than 10 years of diabetes at baseline (1993), and (2) 10 or more years of diabetes at baseline (1993). Four multivariate Cox models were built: (1) crude diabetes data, (2) diabetes adjusted for age and sex, (3) additionally adjusted for heart disease and (4) additionally adjusted for comorbidity. Results with p less than 0.05 were considered statistically significant. SPSS version 19.0 and Epidat 3.1 were used for the analysis.

Of the 1283 participating individuals in the baseline cohort (1993), diabetes self-reported data were available in 1277 individuals who participated in this wave, of which 50.4% were men and 49.6% were women. The flow chart for the study shows the number of participants in each of the waves, as well as the number of those who died or were lost to follow-up in each wave. The distribution of the variables studied in the baseline cohort is presented in Table 1. Women had a higher body mass index than men, as well as poorer self-rated health, a higher frequency and longer history of diabetes, and more comorbidity (p≤0.001).

The prevalence of diabetes in the study population (weighted data) increased progressively over time, from 10.3% (95% CI: 8.6–11.9) in 1993 to 16.1% in 1999 (95% CI: 13.2–18.9), p≤0.001. In the 1993 and 1995 waves, women had a significantly higher prevalence than men (p≤0.05) (Table 2).

The mean duration of diabetes in the 142 participants who had the disease at study baseline (1993) was 7.8 years (SD±6.8; range 0–30 years).

The total biannual incidence density of diabetes was 2.6 (95% CI: 1.9–3.3) cases per 100 persons at 2 years of follow-up, 2.8 (95% CI: 1.7–3.8) in men and 2.5 (95% CI: 1.6–3.4) in women. In general, the incidence density of diabetes decreased with increasing age, although this difference was not statistically significant. Differences between sexes were not statistically significant (Table 3).

Our data show that survival time for older persons with 10 or more years history of diabetes is 2.7 years less than for non-diabetic individuals. The Kaplan–Meier medians in the medium-term models did not reach statistical significance. Median survival in the long term was 10.8 years (95% CI: 10.1–11.5) for non-diabetic persons; 9.0 (95% CI: 6.8–11.1) for persons with less than 10 years’ history of diabetes from baseline (1993); and 8.1 years (95% CI: 5.5–10.6) for those with 10 or more years’ history of diabetes from baseline (1993).

As can be seen in Table 4, the hazard ratio for medium- and long-term mortality in persons with 10 or more years’ history of diabetes was 2.0 (95% CI: 1.2–3.3) and 1.7 (95% CI: 1.1–2.5), respectively, after adjusting for age and sex, heart disease and comorbidity. Accordingly, it can be said that medium-term mortality is twice as high in persons with 10 or more years’ history of diabetes at baseline than in non-diabetic individuals, and is 70% higher in the long term.

The hazard ratio for medium- and long-term mortality in diabetic individuals with less than 10 years’ history of diabetes at baseline was 1.3 (95% CI: 0.9–1.9) and 1.5 (95% CI: 1.2–1.9), respectively. Thus, it can be said that in the medium term (6 years), those with less than 10 years’ history at baseline do not have higher mortality than non-diabetic persons, whereas their long-term mortality is almost 50% higher (Fig. 1).

Figure 1.

Medium- and long-term survival of diabetic patients and non-diabetic patients adjusted for age and sex, heart disease and comorbidity.

(0.19MB).

Because diabetes has a silent course, without a sudden or exact date of onset, it is difficult to design studies of disease incidence.10

In Spain, the few studies that have estimated the incidence of diabetes are not consistent because they did not follow a standardized methodology.24 Moreover, most previous studies in Spain have provided diabetes incidence data in people under age 65 years, and therefore are not comparable with our data.10,25 A recent study in Spain in persons aged 18–65 years25 found an annual diabetes incidence of 19.1 cases per 1000 person-years, slightly higher than that in our study. However, the annual incidence of diabetes in European studies conducted in the general population varies between 1.2 and 4.1 cases/1000 people,26,27 which is lower than the figures published for the Spanish population.

Our results for diabetes incidence in persons aged 65 years and over show an incidence rate equivalent to 13.1 cases per 1000 person-years (2.6 cases per 100 persons-2 years of follow-up), which is between the 24 per 1000 person-years reported in a study made in US elderly12 and 9.4 cases per 1000 person-years for the population 65 years and over in Spain.28 In these two studies the incidence rates increased with increasing age. In our study the incidence is higher in the 65–74 years age group [3.2 (95% CI: 2.3–4.2)] than in those aged 75 years and over [1.7 (95% CI: 0.8–2.6)].

Diabetes has been associated with an average 10 years of life lost for individuals diagnosed during middle age.29 Individuals with diabetes have an excess mortality compared with people without diabetes.11 However, it is not clear how diabetes affects mortality when the diagnosis is made at an older age.13 Our data show that the survival time for people in the older population with 10 or more years history of diabetes is 2.7 years less than that for non-diabetic individuals.

Mortality in the medium term for persons with 10 or more years history of diabetes is almost double that of people without diabetes, while long-term mortality is almost 70% higher. Those with less than 10 years history of diabetes at baseline do not have higher mortality than non-diabetic individuals in the medium term, but long-term mortality is almost 50% higher. In a systematic review of observational studies11 reporting mortality in patients over age 60 diagnosed with type 2 diabetes, those with diabetes were found to have 40% higher mortality than non-diabetic persons, which is lower than found in our study.

Thus, our results show a clear association between diabetes and increased all-cause mortality, although this effect is statistically significant only in patients who have had diabetes for at least 10 years.

The main strength of our study is the long follow-up time (18 years). Added to this is the fact that, to date, few studies of diabetes incidence and mortality have been made in Mediterranean populations.

One of the limitations of this study is that we used self-reported data to assess diabetes. Several authors point out that self-reported data could underestimate diabetes prevalence.10,12,20 Although this underestimation is possible, other authors have confirmed that self-reported diabetes is a reasonably reliable surrogate for diagnosed diabetes, based on their studies of the accuracy of patients’ self-reports compared with medical records.30,31

In health care systems where general practitioners have a complete vision of their patients’ medical status and in which patients have easy and frequent access to medical services, patients are more likely to be well informed about their diseases, especially in the case of chronic diseases like diabetes. This is the case in Spain,32 where every community-dwelling person is registered in the patient list of a general practice, and access to other health care facilities, including outpatient clinics, is only possible through referral by a general practitioner.

Furthermore, in our study, individuals were classified as having diabetes not only based on their awareness of their condition, but also because they had seen a physician in the last year for this problem. The second question on medical consultations in the last year increases the validity of the self-report. In 1993 people in our sample averaged 8.3 consultations with general practitioners, similar to the current rate of general practitioner consultations in Spain.33

Our sample was drawn from a metropolitan area of Madrid, therefore this study may not be comparable to other non-Mediterranean population studies and cannot strictly be generalized to the whole older Spanish population.

Together with Sherwin R and Jastreboff AM,34 we can say that “today's message is that a diabetes tsunami is imminent, and the only way to minimize the damage is to create an early warning system and improve interventions to protect those in its path”. Effective public health interventions to prevent diabetes are needed, together with the resources required to manage the increasingly large number of people who are living longer with this debilitating disease.17 In accordance with Rose, we think that the “population strategy” with regard to prevention should always be the discovery and control of the causes of the incidence of chronic diseases.35 In the case of diabetes, important factors for prevention could be reducing obesity, increasing physical activity and maintaining a Mediterranean diet.