Thirteen simple lifestyle scores and risk of cancer, cardiovascular disease, diabetes, and mortality: Prospective cohort study in the UK Biobank

Non‐communicable diseases (NCDs), also known as chronic diseases, are the leading causes of death worldwide. NCDs accounted for 74% of all deaths (42 million) globally in 2019, with over 70% attributed to the major types of NCDs, including cancer, cardiovascular disease (CVD), and type 2 diabetes (T2D).1 At least 70% of premature CVD, T2D, and up to 40% of cancer cases were estimated to be caused by lifestyle‐related risk factors,2, 3, 4, 5 which highlights the pivotal role of promoting multiple lifestyle habits in NCD prevention.

Various recommendations proposed by organizations and evidence from prior studies guide individuals in adopting healthy lifestyle practices for NCD prevention. To measure adherence to these recommended habits, corresponding composite lifestyle scores have been developed as valid tools to evaluate the overall impact of lifestyles on health outcomes. Studies have repeatedly reported a strong inverse relationship between the number of healthy behaviors and the risk of incident events for which these scores were developed.6, 7 However, the variety of lifestyle guidelines and scores with different components and calculation methods has led to confusion. Understanding their relative predictive performance could inform better risk‐stratification tools and public health strategies. Furthermore, although some lifestyle scores were initially developed to assess associations with single disease entities, it is widely recognized that cancer, CVD, and T2D share common lifestyle‐related risk factors.8, 9, 10

Accordingly, we hypothesized that lifestyle scores originally developed to predict a single disease may also be eligible for risk assessment of other major NCDs. The objective of this study was to compare and benchmark 13 previously established lifestyle scores in terms of their associations with incident cancer, CVD, T2D, and a composite of these NCDs in the UK Biobank. We further sought to identify the common and distinct features of the top‐performing scores.

In this large‐scale prospective cohort study, we examined and compared the associations between 13 simple lifestyle scores and incidence and mortality of major NCDs. In general, a healthier combined lifestyle was associated with a lower incidence of overall NCDs, cancer, CVD, and T2D, as well as with reduced all‐cause, NCD‐related, and cancer‐related mortality, regardless of the specific NCD for which the lifestyle scores were developed. The strongest relationship with lifestyle scores was observed in the analysis of incident T2D, followed by CVD and cancer. The top three lifestyle scores for predicting incident NCDs were CDRI, ELIH, and HLS, while for NCD mortality, the leading scores were CDRI, HB, and HLS.

The assessed lifestyle scores were not only associated with the diseases they were developed for but also proved useful in predicting other NCDs and overall NCDs, supporting our hypothesis of extended utility for multiple NCDs. The ACS and WCRF/AICR scores were developed to assess how adherence to cancer prevention guidelines was linked to cancer‐related outcomes.16, 17 Our results were in line with Greenlee et al. who reported that participants with higher points on ACS also had a 34% lower risk of developing CVD and a 40% lower risk of CVD mortality compared with those with the lowest score.18 Similar to our results, the Diabetes Prevention Program Outcomes Study found a strong association of the WCRF/AICR score with T2D incidence.19 ELIH and ELIR were constructed to reflect long‐term insulin exposure and overall insulin resistance by a lifestyle with insulinemic potential.20 Previous work also showed that the ELIH was associated with colorectal and digestive system cancers, and ELIR with liver cancer, colorectal cancer, and CVD.21, 22, 23, 24 Our findings regarding the ELIH and ELIR scores were consistent; yet, we additionally found that higher scores on ELIH were strongly associated with lower CVD incidence. These associations suggest that different health risk factors have overlapping effects across multiple NCDs. Thus, despite uncertainty about the mechanism, our results highlight that cancer, CVD, and T2D can be prevented altogether by adopting a common healthy lifestyle. These results will be of value to policymakers, researchers, and the general public. In the current era of exploding health information, our results should encourage cancer, CVD, and T2D‐related organizations to jointly create a unified lifestyle guideline to promote primary prevention and early detection of multiple NCDs.

The lifestyle scores were not equally good in our benchmark analysis. The associations of each score differed by outcome. For cancer‐related outcomes, the ELIH, ELIR, MEDLIFE, and WCRF/AICR show relatively weak correlations, and all of them do not include smoking in their scores. In contrast, HLS, HB, and LIS scores included smoking and showed the strongest associations with cancer. This discrepancy may be because smoking has the greatest impact on overall cancer risk compared to other lifestyle factors.25 The definition of smoking in the lifestyle scores also seemed to affect the associations. The HLS not only includes current smoking but also takes into account previous heavy smoking, and showed the strongest associations with cancer outcomes.26 Similarly, the HLI, which includes former long‐term smoking as a risk factor, was also strongly associated with cancer risk in our study. Although the CDRI and LRLS also consider the risk of past smoking, they did not perform as well as the top‐performing scores, possibly because they defined all past smokers as the same category regardless of the amount of smoking and duration. The LIS was constructed by assigning weights to lifestyle factors based on their association with a panel of circulating biomarkers of inflammation.27 Based on the underlying biological pathways of chronic inflammation strongly associated with cancer development,28 the performance of the LIS suggests that inflammation‐related lifestyle factors are also effective predictors of cancer risk.

For CVD, the MEDLIFE score exhibited the worst performance. MEDLIFE incorporates unique lifestyle habits such as napping, sleep duration, and the Mediterranean diet, among others, but excludes smoking and BMI. Given that both are well‐established contributors to CVD risk,29 omitting these key factors might substantially weaken the predictive validity of MEDLIFE. The ELIR also showed a relatively weak association with CVD, which was created to identify a lifestyle pattern predictive of insulin resistance, while the ELIH designed to reflect hyperinsulinemia demonstrated a strong association. Our findings are consistent with previous studies that reported hyperinsulinemia and insulin resistance as risk factors for CVD,30, 31 although the role of hyperinsulinemia and insulin resistance as a cardiovascular risk factors remains controversial. Notably, our study revealed ELIH as a stronger predictor for CVD risk than ELIR. CDRI showed particularly strong associations with both incidence and mortality of CVD. CDRI was created according to recommendations on chronic disease prevention from the National Research Council and the US Department of Health and Human Service.32 CDRI includes smoking and BMI but does not include physical activity. Further analysis is required to determine which specific component contributes most significantly to its stronger association with CVD. Our findings showed the LIS with a good performance for CVD risk, aligning with studies indicating that inflammation increases the risk of CVD.33

For T2D, the results were weakest for HB and MEDLIFE, both of which do not include BMI. The important role of high BMI in T2D risk may explain their limited predictive performance. Although multiple studies from various countries showed that central obesity indices were considered to have a stronger association with T2D than BMI,34, 35 our results on the HLI with different anthropometric measures suggest that BMI is a useful and easy‐to‐assess component of a lifestyle score to predict the risk of T2D. Both the ELIH and ELIR were strongly associated with T2D risk, although these two indices do not include smoking status, possibly because obesity exerts a more dominant influence on T2D development than smoking.36, 37 The LIS in the study was also strongly related to T2D, supporting evidence linking chronic low‐grade inflammation to T2D pathogenesis.38 Among the best‐performing indices (CDRI, HLS, WCRF/AICR, and ACS), all included BMI. When synthesizing the patterns observed across all three NCDs, the HLS emerged as the strongest predictor of NCDs. This is likely due to its inclusion of the five most important modifiable risk factors and its consideration of the cumulative impact of past heavy smoking behavior.

Collectively, a healthier lifestyle was most strongly associated with risk reduction of T2D and less strongly with total cancer. Cancer may be less dependent on behavior than T2D. A multinational cohort study reported a strong inverse association between adherence to a healthy lifestyle and risk of T2D, with a 33% risk reduction, whereas the association with cancer risk was weaker, showing only an 11% reduction.39 Furthermore, the lower effectiveness may be due to certain cancer types having a weaker association with lifestyle factors, which attenuates the overall impact on total cancer incidence. This assumption is supported by our subgroup analyses on lifestyle‐related cancers. Because of the heterogeneity of various cancers, the associations of lifestyle scores and the reduction of risk depend on the specific type of cancer. Given that there is no conclusive evidence on which cancers are affected by lifestyles and to what extent they are affected, further research is needed.

In line with our results in the other subgroup analysis, numerous studies have revealed that women and younger individuals were more likely to benefit from healthy lifestyle behaviors. The inverse association between specific factors comprising lifestyle scores and NCDs supported our results. Wang et al. showed various dietary patterns influenced more on women or people younger than 65 years in terms of developing major NCD.40 A review from Bolego et al. summarized that smoking posed a stronger risk factor for coronary heart disease in women who were current smokers compared to men.41 Moreover, drinking three bottles of wine per week was associated with an absolute risk of cancer of 1.9% for males, but 3.6% for females over their lifetime.42 However, there remains insufficient evidence to derive definitive conclusions. In a 6‐year prospective study, Yoo et al. indicated that a higher level of alcohol drinking would result in more alcohol‐related cancers and total cancer in individuals ≥65 years.43

This study is the first to directly compare existing simple lifestyle scores side‐by‐side within the same population regarding their ability to predict incidence and mortality of major NCDs beyond their original field of application. Using NCDs as a composite endpoint extends knowledge about the preventive potential of the assessed lifestyle scores and enables the selection of the scores best suited for widespread use. Further strengths of this study are its large sample size, the prospective design, and the long‐term follow‐up period available for the UK Biobank cohort.

Several limitations of this study should be acknowledged. First, for certain lifestyle scores, we lacked adequate information in the UK Biobank to replicate them exactly as originally developed. Wherever possible, variables were replaced with alternatives available. Still, such modifications in the construction of a lifestyle score may affect the strength of the associations. Second, for lifestyle scores identified by self‐reported 24‐h dietary assessment, random error caused by day‐to‐day variability and systematic error related to recall bias or misreporting of food intake would impact the results. To improve accuracy, we relied on at least two times of 24‐h dietary assessments and excluded implausible energy intake to improve representativity for typical dietary intake, and to account for seasonal variations in dietary intake. At the same time, online 24‐h dietary assessment had some benefits. It reduced recall bias to some extent by asking about yesterday's diet, and online inquiry may make the participants more honest. Third, differences in participant characteristics by response pattern could limit generalizability. In UK Biobank, participants who responded to the online questionnaire were more likely to be female, older, white, less poor, and more educated than those who did not respond. Participants who responded multiple times were more likely to be white, older, and more educated than those who completed only one response.36 Fourth, early symptoms of T2D are often overlooked, leading to failure to obtain a timely diagnosis, so the time variable can easily be calculated longer than the real time from baseline to T2D diagnosis. To minimize this bias, we used both hospital inpatient admission data and primary care data to identify incident T2D cases. Lastly, although we made efforts to divide the population into four equally sized groups based on each score, the non‐continuous nature of the scores made it challenging to ensure even distribution, which may affect the fairness of the head‐to‐head comparison.

Our results suggest that the use of simple lifestyle scores designed for a single disease group can be extended for predicting multiple NCDs and mortality. The inclusion of BMI and smoking in lifestyle scores is crucial. We found that the HLS and CDRI were the overall best performing scores to predict NCD risk and mortality, which might be useful information for future investigations and assessments. These findings highlight the value of collaborative efforts among organizations related to cancer, CVD, and T2D to develop common lifestyle recommendations to reduce the burden of NCDs.

Jie Ding: Methodology; software; investigation; conceptualization; writing – original draft; writing – review and editing; formal analysis; data curation; visualization. Ben Schöttker: Resources; writing – review and editing. Hermann Brenner: Methodology; writing – review and editing; supervision. Michael Hoffmeister: Conceptualization; methodology; supervision; writing – review and editing.

The UK Biobank was established by the Wellcome Trust, Medical Research Council, Department of Health, Scottish government, and Northwest Regional Development Agency. It has also received funding from the Welsh assembly government, British Heart Foundation, Cancer Research UK, and Diabetes UK. UK Biobank is supported by the National Health Service (NHS). This project was supported by a grant from the German Ministry for Education and Research (Grant No. 01KD2104A). JD was supported by the Chinese Scholarship Council (CSC, No. 202006180030). The sponsors had no role in the design of the study or acquisition of data, nor any other role in the publication of this analysis.

All authors declare no potential conflicts of interest.

This work uses data provided by patients and collected by the NHS as part of their care and support. We would like to thank all participants of the UK Biobank cohort as well as the staff of the UK Biobank assessment centers for their contribution to the studies this research is based on. During the preparation of this work, the authors used ChatGPT 4.0 (OpenAI, 2024) for minor language editing. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication. Open Access funding enabled and organized by Projekt DEAL.

Ding J, Schöttker B, Brenner H, Hoffmeister M. Thirteen simple lifestyle scores and risk of cancer, cardiovascular disease, diabetes, and mortality: Prospective cohort study in the UK Biobank. Int J Cancer. 2025;157(12):2495‐2505. doi: 10.1002/ijc.70064

This work has been conducted using the UK Biobank Resource under Application Number “66591”. The UK Biobank is an open access resource and bona fide researchers can apply to use the UK Biobank dataset by registering and applying at https://www.ukbiobank.ac.uk/↗. Further information is available from the corresponding author upon request.