Moderate weight change following diabetes diagnosis and 10 year incidence of cardiovascular disease and mortality

Adults with type 2 diabetes are at high risk of developing cardiovascular disease (CVD) [1]. CVD is the most common and most costly diabetes complication, accounting for nearly half of total treatment costs globally [2, 3]. Weight loss has been shown to improve CVD risk factors such as HbA_1c, blood lipids and BP among people with type 2 diabetes [4, 5, 6] and may also lead to diabetes remission [7]. However, while improvements in these risk factors have been demonstrated in short-term studies [4, 8], few studies have assessed CVD events.

The existing literature on change in weight and CVD among adults with diabetes is limited, and studies have shown inconclusive and inconsistent associations of weight loss and CVD risk [9, 10, 11, 12]. In the Action for Health in Diabetes (Look AHEAD) randomised trial among obese adults with clinically diagnosed diabetes of 7 years’ duration, the behavioural intervention assessed in the trial did not appear to reduce CVD incidence [6], although participants who lost at least 10% of their body weight in 1 year had a 21% lower 10 year risk of CVD compared with those with stable weight or weight gain [10]. However, weight loss observed among trial participants may not be representative of what would be achievable in the general population in the absence of an intensive intervention programme. Furthermore, as population-based screening for type 2 diabetes has become increasingly common, understanding the impact of weight changes early in the diabetes disease trajectory is needed to inform clinical practice at the time of diagnosis, when patients may be more receptive to care [13].

Considering the challenge of controlling CVD risk factors among individuals with established diabetes and the risks associated with intensification of therapy [14, 15], the year following diagnosis may be a critical time to establish healthy behaviours that lead to long-term comorbidity reduction. Weight loss during this period may result in a legacy effect, whereby improvements in risk factors early in the course of the disease yield long-term CVD benefits. In the UK Prospective Diabetes Study (UKPDS), intensive glucose control following diabetes diagnosis was associated with lower risk of CVD 10 years after the trial, even though differences in glucose control were not maintained after the end of the trial [16]. The Diabetes Care in General Practice trial of people with a new diagnosis of diabetes showed no association between 13 year CVD incidence and weight loss achieved through a 6 year intensive intervention [17]. However, this study had a relatively small sample size (N = 444) and participants had to survive until the end of the 6 year intervention period in order to be followed up for CVD, which might have resulted in a healthier study population relative to the target population [18]. The Retrospective Study of Cardiovascular Events Related to the Use of Glucose-Lowering Drug Treatment in Primary Care (ROSE) cohort study of individuals in Swedish primary care (N = 8486) showed that weight loss in the 18 months following diabetes diagnosis had no protective association with CVD events during a median follow-up time of 4.6 years; however, weight gain was associated with 63% higher hazard of CVD [12]. Given the mixed results of studies of weight loss and CVD and the limited generalisability of results from intensive weight loss trials, additional population-based research is needed to characterise the impacts of weight loss after diabetes diagnosis on long-term CVD incidence.

Using data from a population-based study of screening for type 2 diabetes, the Anglo–Danish–Dutch Study of Intensive Treatment in People with Screen-Detected Diabetes in Primary Care (ADDITION)–Cambridge [19], we assessed the association of weight change in the year following diabetes diagnosis and 10 year CVD and mortality incidence, and the impact of weight change on CVD risk factors (BP, blood lipid and HbA_1c levels) at 1 year and 5 years following diagnosis.

ADDITION–Cambridge (ISRCTN86769081) is a pragmatic cluster-randomised trial comparing multifactorial intervention vs routine care among people with screen-detected diabetes from 49 general practices (GPs) in eastern England [19]. The present study is an observational analysis of the trial cohort. A validated risk score was used to identify eligible individuals aged 40–69 years who were at high risk of diabetes, using the electronic records of participating GPs [20]. From 2002 to 2006, 33,539 high-risk individuals were invited to attend a stepwise screening programme; 24,654 (74%) took part [21]. During screening, 867 participants were diagnosed with type 2 diabetes using the 1999 WHO criteria [22] and all consented to enrol in the study. The GPs were cluster-randomised to intensive treatment (n = 26) or routine care (n = 23). In the routine care group, GPs were advised to follow current UK guidelines for diabetes management [23, 24, 25]. Intensive treatment included more frequent consultations, provision of educational materials and GP-based academic-detailing sessions encouraging earlier use of medication to improve control of risk factors. The intervention did not include behavioural treatment. Written informed consent was obtained from all participants and ethics approval was obtained from the research ethics committees in Cambridge (ref. 01/063), Huntingdonshire (ref. 00/609), Peterborough and Fenland (ref. P01/95), West Essex (ref. 1511–0103), North and Mid Essex (ref. MH395 MREC02/5/54), West Suffolk (ref. 03/002), and Hertfordshire and Bedfordshire (ref. EC03623) Local Research Ethics Committees and the Eastern Multi-Centre Research Ethics Committee (ref. 02/5/54).

Of the 99 first incident CVD events, 36 were revascularisations, 24 were strokes, 21 were CVD deaths, 17 were myocardial infarctions and one was an amputation. Of the 95 incident deaths during the study period, 31 were related to CVD, 45 to cancer and 19 to another cause. Mean follow-up time was 9.8 years.

Missing weight information at baseline or 1 year was associated with SES and smoking: unskilled workers were more likely to have missing weight compared with professional workers, and non-smokers were less likely to have missing weight compared with current smokers (ESM Table 3).

In this study of 725 individuals with screen-detected type 2 diabetes, loss of ≥5% body weight was associated with a 48% lower 10 year hazard of CVD incidence vs maintaining weight, after adjustment for age, sex, baseline BMI, SES, smoking, medication use and trial arm. Associations were independent of changes in weight between 1 and 5 years in the study and were consistent among adults aged ≥65 years at the time of diabetes diagnosis. Loss of ≥10% body weight was associated with improvements in HbA_1c, total cholesterol and triacylglycerols at 5 years following diagnosis, which might have contributed to the observed association of weight loss and CVD events. There was no apparent association between weight loss and all-cause mortality. Weight gain was also non-statistically significantly associated with lower hazard of CVD, but possibly higher hazard of all-cause mortality.

While extreme weight loss among people with type 2 diabetes achieved through bariatric surgery has been shown to reduce the risk of CVD [36], studies of weight loss in non-surgery populations have shown inconsistent associations with mortality and incidence of CVD [9, 37]. A Scottish observational study of weight change within 2 years of diabetes diagnosis and 5 year incidence of CVD showed no benefits of weight loss [11]. However, the magnitude of weight lost in this cohort was relatively low (mean weight change −0.7% ± 6.7%) compared with the −3.7% ± 5.7% mean weight change in the current study. The ROSE study also showed no association between reductions in BMI in the 18 months following diabetes diagnosis and 5 year CVD risk [12]. While the magnitude of weight loss in the ROSE study was comparable to that in ADDITION–Cambridge (mean BMI decrease among those who lost weight was 2.5 kg/m², or 7.6% of baseline BMI), those who had lost weight had a higher BMI across the study period compared with those with unchanged weight. This was not the case in our study, where those who lost the most weight had a lower BMI across the study period compared with those who had maintained or gained weight (ESM Table 2). Additionally, the ROSE study had a relatively short follow-up (median 5 years) and did not control for antihypertensive medication use or smoking [38]. Furthermore, unintentional weight loss due to underlying disease might have masked any protective impact of weight loss on CVD.

One year weight loss of ≥10% body weight was associated with a 21% lower 10 year hazard of CVD in the Look AHEAD trial. We observed similarly strong, protective associations with ≥5% weight loss. This may be due to the fact that we targeted weight loss following diagnosis of diabetes via screening, whereas in Look AHEAD weight loss occurred, on average, 7 years after clinical diabetes diagnosis. Weight loss in the early stages of diabetes may result in a legacy effect to reduce long-term incidence of CVD, as was observed with short-term intensive glucose control in the UKPDS [16]. Weight loss was associated with improvements in HbA_1c, lipid levels and BP at 1 year in the study. Few of these associations persisted at 5 years; however, it is possible that improvement in risk factors early on contributed to the observed lower incidence of CVD among those who lost the most weight. Furthermore, changes in BP, blood lipids and blood glucose only partially explain the mechanisms by which healthy behaviours lower CVD risk [39]. Increases in physical activity among those who lost weight might have reduced the risk of CVD through inflammatory mechanisms independently of the risk factors measured in our study.

Unintentional weight loss due to sarcopenia or cancer increases the short-term risk of mortality [40] and may be a source of unmeasured confounding in our study. To address potential confounding, we excluded deaths occurring in the year following the assessment of weight change, to reduce the likelihood that weight loss in our study was caused by disease. The protective association between weight loss and CVD risk observed in this study remained evident after restriction to participants aged ≥65 years at the time of diagnosis. We cannot definitively determine that weight loss in this study was intentional and because of healthy lifestyle changes rather than age-related or cancer-related wasting; however, the fact that we excluded deaths occurring within 1 year following the initial weight loss, and that our results were robust among older participants, supports our interpretation that weight loss was unlikely to be the result of underlying disease. Unmeasured socioeconomic or baseline health differences may also modify associations with weight loss; it is possible that weight loss may be achievable and helpful for some people but unachievable or unhealthy for others. We cannot rule out the possibility that those who lost weight might have been more willing and able to healthily lose weight compared with those who did not lose weight, or the potential for other unmeasured confounding. Therefore, our results may only be applicable to people who are willing and able to lose weight.

Compared with maintaining weight, weight gain >2% had a suggestive protective association with CVD and a suggestive adverse association with mortality. These results were not statistically significant and should be interpreted with caution, as only 75 study participants gained weight. Considering the existing literature, there is no substantiating evidence that weight gain is protective against CVD among people with type 2 diabetes. In the ROSE study, weight gain was associated with a higher hazard of 5 year all-cause mortality compared with maintaining weight [12]. Smoking cessation is associated with reduced CVD risk and weight gain; however, it is unlikely that it would have introduced confounding, as we controlled for cigarette smoking and only 17 study participants had given up smoking during the 1 year period when weight was measured. There may be heterogeneity in the associations of weight gain and CVD by baseline BMI; we adjusted for baseline BMI but we did not have a sufficient sample size to perform stratified analyses. Among those who gained weight, we might have under-ascertained CVD incidence because of the competing risk of mortality; however, as we censored participants at the time of death, person-time following a competing event was not included in the risk set. In a sensitivity analysis we estimated subdistribution HRs [35] to incorporate person-time of individuals who experienced a competing event, and the results were unchanged. Regardless, the observed associations of weight gain and CVD had poor precision and may be due to chance.

The ADDITION–Cambridge cohort was predominantly white, which limits generalisability to other populations. The majority of participants were overweight or obese at diagnosis, and although we adjusted for baseline BMI, the results would not necessarily apply to people with a normal weight or those who are underweight. We had approximately 15% missing information on weight at baseline or 1 year follow-up. Multiple imputation of missing information on weight showed generally similar associations, indicating results were robust to missing data under the assumption that missingness occurred at random, conditional on the covariates included in the imputation model. The relationship between weight change and CVD risk might have been confounded by health differences between those who maintained, gained or lost weight and differences in medication use across the study period. However, we adjusted for CVD risk factors including smoking, BMI, SES and cardio-protective medication use. While we adjusted for glucose-lowering medication, we were not able to separately account for metformin and sulfonylureas, as only 28 study participants were using sulfonylureas at 1 year. Sulfonylurea use was most common among participants who gained weight. Metformin use comprised the majority of glucose-lowering drug use in this cohort, reducing our concerns of bias related to differential effects of glucose-lowering drugs on weight and CVD [41]. Exclusion of participants who reported sulfonylurea use at 1 year did not change any associations (not shown). We assessed a number of risk factors at two time points, which might have increased the probability of a type I error. However, other studies have reported similar associations between weight loss and cardiovascular risk factors [42].

Of all high-risk individuals who were invited to participate in screening, 74% attended screening and all 867 individuals diagnosed with diabetes consented to enrol in the study, resulting in a study population which affords generalisability to the broader primary care population of eastern England. Due to the screening-based nature of this study, we were able to capture weight change during the period immediately following diabetes diagnosis, to assess whether moderate, achievable weight loss during this period can reduce long-term CVD risk. Adjustment for confounders including cigarette smoking and cardio-protective medication use improves the internal validity of our study relative to other studies of weight loss and CVD which did not adjust for these [11, 12].

Loss of ≥5% body weight during the year following diabetes diagnosis was associated with a lower hazard of CVD events at 10 years but was not associated with all-cause mortality. Weight loss ≥5% was also associated with improvements in HbA_1c, diastolic BP and lipids at 1 year, while weight loss ≥10% was associated with improvements at 5 years. There is growing recognition of the impact of weight loss on diabetes remission, and our research supports the hypothesis that moderate weight loss in the year following diabetes diagnosis may yield substantial long-term CVD reduction and may be an achievable target for individuals with a new diagnosis of type 2 diabetes.