Women Living Alone Have an Increased Risk to Develop Diabetes, Which Is Explained Mainly by Lifestyle Factors

Screening was performed in 1995–2000 in 6,917 women aged 50–64 years, living in a defined area of Southern Sweden, and identified through a population register comprising all 10,766 inhabitants. Informed consent was obtained, and the ethics committee at Malmö/Lund approved the study.

The baseline physical examination included body weight, BMI, waist-to-hip ratio, and the average of two blood pressure recordings in the seated position after 15 min of rest. Blood glucose and serum levels of triglycerides and total cholesterol were measured in capillary whole blood (Cholestech LDX; Cholestech, Hayward, CA). Hormonal assays of serum fasting insulin and leptin were performed using an enzyme-linked immunosorbent assay (DRG, Marburg, Germany). Insulin resistance was expressed through the homeostasis assessment model of insulin resistance (HOMA-IR) (13).

A questionnaire was used at baseline and at follow-up regarding medical history, perimenopausal status, pharmacological treatment, family history of diabetes, physical activity, dietary, alcohol and smoking habits, marital status, household, education, occupation, and mental health.

At baseline, 3,593 subjects with one or more features of the metabolic syndrome were identified, and a 75-g oral glucose tolerance test (OGTT) was performed in the fasted state. The positive cutoff values were BMI ≥30 kg/m², waist-to-hip ratio ≥0.9, systolic blood pressure ≥160 and/or diastolic blood pressure ≥95 mmHg, capillary blood glucose level ≥8.0 mmol/l, or serum triglyceride level ≥2.3 mmol/l; family history of diabetes (parents/siblings) and treatment for hypertension or hyperlipidemia were also considered. A control group without positive screening was also examined with an OGTT. The procedure efficiently identified subjects with IGT or diabetes (14). In all, 626 women had IGT and received lifestyle advice regarding exercise, dietary, smoking, and alcohol habits individually for 1 h and after the examination were invited to a 2.5-year follow-up with a new OGTT and measurement of BMI, blood pressure, fasting insulin, and leptin. From the initial group, 461 returned for the follow-up examination (12). No in-between examinations were performed. Fasting insulin and leptin were randomly analyzed in 153 of the 461 women who had IGT and returned for the follow-up study. No differences in baseline biological data were seen among the 461 participating women and the nonparticipants or among the random group of women from whom hormones were analyzed and remaining women with IGT.

Household status included women living alone, with a partner, with children, with a partner and children, or with other adults. Marital status meant married, unmarried, divorced, or widowed. Education was categorized into comprehensive school (9 years), upper secondary school (12 years), or university degree. Occupational status referred to full-time or part-time occupation, being a housewife, retired, unemployed, long-term sick-listed, or having a disability pension.

Subjective mental health was measured by the validated Gothenburg Quality of Life instrument (15), which refers to the World Health Organization definition of health. Mental well-being was assessed in seven steps with five questions on mood, energy, endurance, self-esteem, and sleep. The mean sum score of estimations from “very bad” (1) to “excellent, could not be better” (7) was calculated. Ten mental symptoms regarding restlessness, difficulty to relax, nervousness, impaired concentration, sleeping disturbance, irritability, exhaustion, general fatigue, crying easily, and depression were answered with “yes” or “no” depending on whether the woman had any of these symptoms during the last 3 months. The total number of “yes” answers was used in analyses.

Self-reported dietary habits were based on four questions regarding fat, fibers, fruits/vegetables, and sweets/carbohydrates, with three sublevels from unhealthy to healthy, and on a food questionnaire, which ensured that the basic questions discriminated between unhealthy and healthy diets. Women defined as having healthy dietary habits were those indicating a healthy pattern on at least three of four basic questions with no indications of an unhealthy pattern. The subjects summarized their leisure time exercise during the last year into hardly any activity, <30 min/week, 30–60 min/week, >60–120 min/week, and >120 min/week of jogging and equivalent activities. Weekly consumption of wine, beer, and spirits was converted into grams of alcohol and divided into no consumption, ≤83 g/week, and ≥84 g/week. Smoking was categorized by lifetime consumption of pack-years. One pack-year corresponded to 20 cigarettes/day/year. Subjects were divided into never smokers (<1 pack-year) and past and current smokers (≥1 pack-year).

Calculations were performed using SPSS 12.0. Values are given as means ± SD. The χ² test was used to analyze differences in proportions for categorical variables, and the independent-samples t test or paired-samples t test was used for continuous variables. Multiple logistic regression analyses were used to evaluate the influence of household status on the risk of developing diabetes and to reverting to normal glucose tolerance (NGT). Women living alone were compared with the remaining group of women with any other household condition. Multivariate adjustments in different blocks were entered stepwise. The fit of the models was analyzed by calculation of Hosmer-Lemeshow goodness-of-fit statistics, and statistics for the overall models was analyzed by using Cox and Snell R² and Nagelkerke R². P values <0.05 were considered statistically significant.

Of 461 women with IGT at baseline, 175 (38%) had NGT at follow-up, 9 (2%) had impaired fasting glucose, 222 (48%) had IGT, and 55 (12%) progressed to diabetes. Sixty-six women (14.3%) lived alone, 327 (71.1%) lived with a partner without children, 13 (2.7%) lived with children only, 53 (11.5%) lived together with a partner and children, and 2 (0.4%) lived with other adults. There were no differences in household status between women who developed diabetes and those who changed to NGT when each household status was simultaneously analyzed. However, more women who lived alone progressed to diabetes compared with those with other household conditions combined (P = 0.029) (Table 1). Corresponding analyses for each other household status, compared with those for the remaining women, showed no differences in progression to diabetes (P = 0.06–0.84). Among women living alone, 32 (48.5%) were divorced, 15 (22.7%) were unmarried, 12 (18.2%) were widowed, and 7 (10.6%) were married but not living with their husband. There was no difference in rates of women progressing to diabetes according to their marital status (P = 0.59). At follow-up, 17 previously cohabiting women were living alone, whereas 6 had changed from living alone to living with a partner (n = 5) or parent (n = 1).

The 66 women living alone and the 395 remaining women did not differ in education (P = 0.49), occupation (P = 0.53), subjective mental well-being (P = 0.56), or mental symptoms (P = 0.39), and no differences were seen in age at baseline or follow-up or in follow-up time (Table 2). Fasting glucose did not differ at baseline (P = 0.25) but was higher at follow-up in women living alone compared with the remaining women (P = 0.023). No differences were seen at baseline or at follow-up in 2-h blood glucose level (P = 0.99 and P = 0.17) or in systolic (P = 0.72 and 0.71) or diastolic blood pressure (P = 0.90 and 0.34). Among women living alone, systolic (P = 0.046) and diastolic (P < 0.001) blood pressure had become lower at follow-up. In the remaining women, systolic (P < 0.001) and diastolic (P < 0.001) blood pressure, as well as fasting (P = 0.017) and 2-h (P < 0.001) blood glucose level, had decreased (data not shown). There were no differences in fasting insulin or leptin levels or in HOMA-IR at baseline or follow-up between the two groups (Table 2). However, leptin had increased at follow-up, both among women living alone (P = 0.003) and among the other women (P < 0.001). BMI was unchanged among women living alone, whereas it was lower at follow-up among the remaining women (P = 0.046).

No differences were seen in exercise at baseline, and nearly equivalent rates of women in both groups had changed their weekly exercise habits at follow-up (Table 3). At baseline, there was no difference in diet. At follow-up, fewer women living alone had a diet regarded as healthy (P < 0.001), and only 1.5% had improved their dietary habits and 86.4% had an unchanged poor diet. Corresponding figures among the other women were 18.3 and 61.7% (P = 0.001). More women living alone had high intake of saturated fat and carbohydrates (data not shown). Alcohol consumption did not differ at baseline, but more women who lived alone were abstainers at follow-up (P = 0.045). The amount of weekly alcohol intake was lower at baseline (P = 0.025) and follow-up (P = 0.013) for women living alone, and they reduced their consumption (P = 0.036), whereas habits remained unchanged among the other women. Fewer women living alone had never smoked and more were current smokers at baseline (P = 0.002) and follow-up (0.006). Women living alone and who smoked consumed more cigarettes per day at baseline (P = 0.039) and had not changed the daily amount at follow-up, in contrast to the reduction seen in cohabiting smoking women (P = 0.030).

Women living alone had a 2.47-fold increased risk of developing diabetes after adjustment for age (95% CI 1.06–5.71) and a 2.68-fold increased risk (1.02–7.05) after adjustments for age, follow-up time, BMI, 2-h blood glucose at baseline, serum triglyceride levels, total cholesterol level, blood pressure, family history of diabetes, and perimenopausal status (Table 4). The risk increased to 3.26 (1.19–8.96) when education and occupation were added to the analysis. The risk increased if education and occupation were added simultaneously as well as separately. Further adjustments for subjective mental health, physical exercise, dietary habits, and alcohol consumption showed remaining significant odds ratios (ORs), decreasing from 3.18 (1.16–8.73) to 3.03 (1.02–8.99). When smoking status was added, the OR became nonsignificant (2.07 [0.62–6.88]). Also, the risk was attenuated to null when smoking was included without adjustments for exercise, diet and alcohol (2.29 [0.78–6.75]). P values for smoking were at nonsignificant levels in these models, both with (P = 0.15) and without (P = 0.13) simultaneous adjustments for the three other lifestyle variables. Smoking was not related to diabetes incidence (P = 0.28) after adjustments for age and household only. The Hosmer-Lemeshow values ranged from 0.134 to 0.848, the Cox and Snell R² ranged from 0.210 to 0.229 (data not shown), and the Nagelkerke R² ranged from 0.314 to 0.340 in models with P < 0.05.

This prospective study showed that household conditions influenced the transition from IGT to diabetes. Middle-aged and older women with IGT and living alone had a higher risk of developing diabetes. Negative health effects from living alone may potentially be mediated through social and mental conditions. There is a known socioeconomic gradient, often attributed to a low level of education, in health outcomes such as cardiovascular disease (16), although it has not been fully described in diabetes. Low-paid labor is more common among women (17), making women without a supporting partner especially susceptible for socioeconomic strain. Work stress and a low sense of coherence in middle-aged women have been associated with type 2 diabetes (18).

In this study, adjustments for education and occupation did not eliminate the risk in women living alone but instead augmented the risk. This may be explained by that fact that a relatively high proportion of women living alone and who had a university degree and worked developed diabetes. A limitation was the rather finite definition of occupational status as being capable of working and having work outside home or not.

Our findings showed that low subjective mental health explained some of the differences between women living alone and the other women. This finding is consistent with other reports showing that diabetes is associated with lower well-being (19) and that depression is more common in patients with type 2 diabetes (7). We did not separate replies to questions on depression in this analysis but have previously shown that poor self-rated health is associated with treatment for depression (20), indicating that the Gothenburg Quality of Life instrument (21) used in the Women’s Health in the Lund Area study is reliable. We have also shown that women who lived alone had a higher risk for severe mental symptoms (21). The present data are in line with previous results showing that living alone was associated with neglect of self-care in insulin-dependent diabetes (22).

The importance of different psychosocial conditions in women living alone seems to explain the increased risk of progression to diabetes only to some degree. Instead, this risk was mostly associated with smoking, dietary, and alcohol habits at baseline and modifications during the follow-up period. There were more current smokers among women living alone. Also, these women consumed more cigarettes, and the importance of smoking for the outcome of diabetes in women living alone was the most consistent finding of the lifestyle variables. Smoking has in other studies been linked to type 2 diabetes, probably due to increased insulin resistance (23). However, smoking in this study was not independently associated with diabetes and cannot be regarded as important as the status of living alone. Thus, living alone seems to be a better predictor of diabetes.

A well-balanced diet and a decrease in body weight prevent progression to diabetes in patients with IGT (2,3). BMI did not differ at baseline or follow-up, but cohabiting women reduced their BMIs slightly during the follow-up period. No differences were seen in physical exercise at baseline, and both groups had altered their weekly exercise habits to the same degree at follow-up. Still, exercise seemed to have a slight impact on the risk estimates for the increased risk among women living alone. Weight reduction and regular physical activity improve glucose tolerance by increased insulin release and improved insulin sensitivity (24). However, we could not find any difference in insulin levels or insulin resistance, which might be explained by the small number of samples or short follow-up time. Serum leptin levels increased in all women, but without any differences between the two groups. The increase was somewhat bigger in women living alone, which could be a sign of augmented insulin resistance, despite unchanged insulin levels that usually correlate positively with leptin (25). Leptin has been shown to be positively associated with many metabolic variables and insulin resistance, but the levels have been similar in subjects with and without diabetes (26), which may support the results in the present study.

Among women living alone, there were more alcohol abstainers and lower consumption among users, which seemed to have a negative impact on glucose tolerance. Moderate alcohol consumption has been shown to decrease the risk of diabetes (27), possibly related to long-term effects on glucose metabolism of liver transformations (28).

The relatively small total numbers of subjects and large difference in sample size between the two studied groups implies low power to detect real differences. However, because this was an observational study, these were the subjects who were available to us. The analyses of goodness-of-fit and statistics for the overall models showed satisfactory fit of the presented models.

In summary, middle-aged and older women with IGT and living alone had a higher risk of progression to diabetes, mostly due to their smoking, dietary, and alcohol habits. Leisure time exercise and other psychosocial conditions had less influence on the outcome.

This study was supported by Region Skane Sweden.