Abstract
Background
Multimorbidity has posed growing global health burdens, yet evidence on modifiable dietary risk factors remains limited. To explore the dietary factors influencing multimorbidity risk, a longitudinal study was conducted in the UK Women’s Cohort.
Methods
The cohort initiated from 1995 to 1998 has recruited 35,372 women aged 35–69 years at baseline. Daily consumption of foods and a modified Mediterranean diet was evaluated based on a validated food frequency questionnaire with 217 items. Charlson comorbidity index (CCI) was used to assess multimorbidity based on diagnostic ICD codes from Hospital Episode Statistics. Associations between diet and multimorbidity were estimated using Cox’s proportional hazards models. Subgroup analyses stratified by age at baseline and body mass index were additionally explored.
Results
During 486,656 person-years of follow-up (mean 19.7 years), 7,516 of 24,703 participants (30.4%) developed incident multimorbidity. Compared with women with low adherence to Mediterranean diet, those with moderate or high adherence had 6% (hazard ratio = 0.94, 95% confidence interval: (0.89, 0.99)) or 14% (0.86 (0.80, 0.92)) lower risk of multimorbidity in the fully-adjusted models, respectively. Increased risk of multimorbidity was observed to be associated with each additional 10 g/MJ intake of processed meat (1.60 (1.37, 1.87)), red meat (1.19 (1.12, 1.26)), and total meat (1.12 (1.08, 1.17)). Above associations remained in subgroup analyses stratified by age at baseline or body mass index (BMI), except in the subgroup aged ≥ 60 years where the association between adherence to Mediterranean diet and risk of multimorbidity was attenuated significantly. Additionally, there was no significant evidence on effect modification of age at baseline or BMI on the dietary factors (all P-interaction > 0.05).
Conclusions
Adherence to Mediterranean diet was inversely, but consumption of meat especially processed meat was positively, associated with risk of multimorbidity. These modifiable dietary factors represent promising prevention targets, warranting further investigation in diverse populations.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12937-025-01238-x.
Keywords: Multimorbidity, Mediterranean diet, Processed meat, Red meat, Longitudinal study
Introduction
Multimorbidity is widely considered as the coexistence of two or more chronic conditions in same one individual [1]. The chronic conditions usually refer to long-term health conditions that significantly affect the individual, including a physical non-communicable disease (e.g., cardiovascular disease, cancer), a mental health condition (e.g., dementia, mood disorder), or an infectious disease of long duration (e.g., HIV, hepatitis C) [2, 3]. The global disease burden is changing from premature mortality to long-term health conditions partially due to improvements in medical treatment and survival rates of chronic diseases [4]. Given the global demographic shift toward aging populations, multimorbidity has emerged as a critical public health challenge worldwide. The prevalence of multimorbidity ranges from 13–72% in the natural population, which is higher in women or elderly population (60 years and more), especially in high-income countries [3]. There were more than 50% of people with multimorbidity who aged over 65 years in high-income countries [5], which is projected to increase to 68% or more by 2035 [6]. Multimorbidity has led to huge economic burden on individuals and health-care systems due to stacking medical costs and increased complexity of medical care from the combinations of diseases [4].
Despite increasing interests in multimorbidity research, associations between dietary factors and risk of multimorbidity remain unclear. A cross-sectional study from 129,369 participants in Netherland found that dietary patterns high in meat, alcohol, and snack had higher odds of being with high morbidity scores, while patterns high in vegetable, fish and fruit were associated with lower odds [7]. Another cross-sectional study in Cyprus also showed that high adherence to a Mediterranean diet characterized by high consumption of vegetables, fruits, legumes, as well as whole grains, and low consumption of meat and meat products, was linked with lower odds of multimorbidity [8]. A multimorbidity score evaluated by the Cumulative Illness Rating Scale for Geriatrics (CIRSG-SI) was found to be negatively associated with the Mediterranean diet score (r = −0.247, P = 0.003) among 143 geriatric patients in Italy [9]. However, current evidence linking diet to multimorbidity predominantly derives from cross-sectional studies conducted in specific populations, highlighting the critical need for large-scale longitudinal research.
We utilized data from both in-hospital medical records and a large population-based cohort with 20 years of follow-up. This prospective study examined associations between specific food groups, adherence to the Mediterranean diet, and multimorbidity risk, aiming to identify modifiable dietary factors for multimorbidity prevention and management.
Methods
This study was reported according to the STROBE-nut (Strengthening the Reporting of Observational Studies in Epidemiology-nutritional epidemiology) guidelines [10], as shown in the Additional file.
Study design and population
This study is embedded within the UK Women’s Cohort Study (UKWCS), a longitudinal, population-based cohort, detailed elsewhere [11]. Initially, women aged 35–69 years, registered in the national health service (NHS) of United Kingdom, were invited to participate across England, Scotland, and Wales. In total, the UKWCS recruited 35,372 participants over the baseline data collection (1995 to 1998). At recruitment, population characteristics including lifestyle behaviors, demographic and anthropometric information, were collected during the interviews and examinations conducted. Informed consent was obtained for each participant upon entering the study. The UKWCS was approved by National Research Ethics Service Committee for Yorkshire & the Humber – Leeds East (Ref: 15/YH/0027) at initiation of the cohort for ethics in 1993; and further approved by Health Research Authority REC (Ref: 17/YH/0144).
The UKWCS was linked with electronic medical data from Hospital Episode Statistics (HES) of UK National Health Service (NHS) up to 31 March 2019. The HES is a data warehouse containing records of all patients admitted to NHS hospitals in England; therefore, those who lived outside of England were excluded from analyses [12]. Eligible participants for this study were UKWCS members who completed baseline dietary assessments. We excluded participants with prevalent multimorbidity, implausible energy intake (< 500/> 5000 kcal/day), and those with missing data on key covariates.
Of the whole participants recruited, women were excluded who lived outside of England (n = 3821), who had the modified Charlson Comorbidity Index (CCI) > 0 at baseline (n = 2629), who had implausible energy intake (n = 78), and who had incomplete data in covariates (n = 4141), leaving 24,703 subjects eligible for analyses in this study. The stepwise exclusion process of UKWCS participants for this study was summarized in a flow diagram as Fig. 1. To assess potential selection bias, we compared baseline characteristics between included and excluded participants, in particular excluded participants with prevalent multimorbidity (Supplementary Table S4). The absence of significant differences suggests minimal impact of participant exclusion on our primary findings.
Fig. 1.
Participant inclusion and exclusion flow diagram. The flowchart illustrates the sequential selection of the study population from the UK Women’s Cohort Study, detailing the number of participants excluded at each stage and the corresponding reasons
Diet assessment
A self-administered food frequency questionnaires (FFQs) with 217 food items was used to collect baseline dietary information. The FFQ was developed based on UK FFQ version in the European Prospective Investigation into Cancer and Nutrition (EPIC) study [13], and has been validated elsewhere [14, 15]. Consumed daily portions for each food item were calculated based on transformation of food intake frequencies (details in Supplementary Table S1), and multiplied by standard portion weights [16] to obtain weights of each food consumed per day (grams/day, g/d). Supplements taken by participants were not included, and food items with missing values were considered to be not taken by participants in this study.
In this study, a Mediterranean diet score was created to quantify adherence to the Mediterranean diet [17, 18]. Briefly, of ten food/nutrient components traditionally consumed in a Mediterranean diet, six (vegetables, legumes, fruits and nuts, cereals, fish, and fatty acid ratio of monounsaturated plus polyunsaturated fatty acids to saturated fatty acids, namely MUFA + PUFA: SFA) recommended by most dietary guidelines, were given a score of 1 for consumption above the median. Other three components (meat, poultry, and dairy) not highly recommended, were assigned 1 score if consumed at or below the median. For alcohol recommended to be consumed moderately, a score of 1 was given to women who had intakes of between 5 and 25 g per day. The cut-points of the ten components in this study were shown in Supplementary Table S2. The total score ranged from 0 to 10 with higher scores indicating greater dietary adherence, and was divided into three groups: low adherence (scores 0–3), moderate adherence (4–6), and high adherence (7–10).
We focused our analysis on key food groups (vegetables, fruits, total fish, and total meat) based on their high consumption frequency and potential associations with multimorbidity risk [19, 20]. In addition, given ongoing scientific debate regarding meat consumption [21], we further examined specific subtypes (processed meat, red meat, and poultry) to elucidate potential differential effects.
Ascertainment of cases
Multimorbidity was defined as the presence of two or more chronic conditions from the modified CCI list [22]. The CCI score was calculated based on a coding algorithm in Supplementary Table S3, where a weighted score was assigned to each of listed International Classification of Diseases (ICD)−10 diagnostic codes. The HES contains multiple hospitalization records for each included participant with the primary and secondary diagnostic ICD codes. Consistent with established methodology [23, 24], we excluded each participant's primary diagnosis from CCI calculations. Secondary or subsequent diagnoses matching ICD-10-AM codes (Supplementary Table S3) were assigned corresponding weighted scores. For every hospitalization record, we computed the total CCI score by summing all applicable diagnostic weights from non-primary diagnoses. Incident multimorbidity cases were defined as those with two distinct conditions or more as indicated by CCI scores. Participants were followed from study entry till first identification of multimorbidity, date of death, or until the censor date (31 March 2019) whichever came first.
Statistical analysis
The analyses were conducted in three steps. First, descriptive statistics were performed to describe baseline socio-demographic and lifestyle characteristics by three levels of adherence to a Mediterranean diet for the participants. For continuous variables, the mean and standard deviation (SD) were calculated to summarize profiles of participants, while numbers of count and percentages were used for categorical variables.
Second, survival analyses were conducted using Cox proportional hazards regression. Hazard ratios (HR) and 95% confidence intervals (95% CI) were estimated to assess associations between each dietary factor and risk of multimorbidity. Age at baseline was minimally-adjusted as a continuous variable since prevalence of multimorbidity usually increases with age. A fully-adjusted model was additionally conducted by adjusting for covariates previously identified in the literature [20], including: age at baseline; marital status (married/living as married, separated/divorced, and single/widowed); menopausal status (premenopausal, postmenopausal); ethnicity (white, Asian, black, and other); socio-economic status (SES, professional/managerial, intermediate, and routine/manual); physical activity (low, moderate, and high levels); body mass index (BMI, kg/m2, continuous); daily energy intake (MJ/d, continuous); alcohol consumption (g/d, continuous); and smoking status (never smoked, ex-smoker, and current smoker). Most covariates were self-reported at baseline. SES was derived from the United Kingdom National Statistics-Socio-Economic Classification (NS-SEC) [25]. Given the well-documented overlap between socioeconomic indicators (education, occupation, and income), we adjusted for composite SES status rather than individual components to optimize model stability [26]. Physical activity was calculated based on a series of questions about participants’ usual daily activities at baseline that were taken from the International Physical activity questionnaire (IPAQ) short form, and categorized into three levels being low, moderate, and high, according to the official guidelines for data processing and analysis [27]. All Cox models satisfied proportional hazards assumptions (global test p = 0.35), with variance inflation factors < 2.0 indicating no substantial multicollinearity among covariates.
Third, subgroup analyses were conducted to explore potential modification effect of age at baseline and BMI. The fully-adjusted models were fitted in separate age groups (< 60, ≥ 60 years), and BMI groups (normal-weight with BMI < 24.9 kg/m2, and overweight with BMI ≥ 24.9 kg/m2). In addition, age at baseline and BMI were separately added to the fully-adjusted models as interaction terms with dietary factors, where they were modelled linearly. Secondary analyses mutually adjusted for all examined food groups to assess independence of associations, while recognizing inherent collinearity in dietary data.
To account for varying consumption ranges across food groups, we performed sensitivity analyses modeling dietary exposures per 1-SD increment in energy-adjusted food consumption. Further sensitivity analysis excluding participants with survival time < 5 years was performed to check if any potential reverse causation. Additionally, participants with long-term treatment for illness at baseline were excluded as another sensitivity analysis to check if any selection bias from inclusion of potential multimorbid individuals.
For ease of interpretation, in this study HRs were estimated per 10 g of each food group per MJ of total energy consumed for energy adjustment. All analyses were performed in Stata/MP, version 17.0 (Stata Corp LP).
Results
Demographic characteristics and dietary profiles at baseline
The baseline characteristics of the study population (n = 24,703) were summarized in Table 1, stratified by Mediterranean diet adherence levels (low/moderate/high). Participants had a mean age of 51.3 years (SD = 9.0) at baseline. Over 486,656 person-years of follow-up (mean 19.7 years), we identified 7,516 incident multimorbidity cases. The distribution across diet adherence groups was: 2,033 cases (27.0%) in the low-adherence group, 4,036 (53.7%) with moderate adherence, and 1,447 (19.3%) in the high-adherence group. Most participants were the White (98.8%), married or living as married (77.1%) in the whole population.
Table 1.
Participant characteristics by Mediterranean diet adherence levels in UK Women’s Cohort Study
|
Low adherence (N = 5902, 23.9%) |
Moderate adherence (N = 13,216, 53.5%) |
High adherence (N = 5585, 22.6%) |
All participants (N = 24,703) | ||
|---|---|---|---|---|---|
| Age at baseline (years) | Mean (standard deviation) | 52.6 (9.2) | 51.3 (9.0) | 49.7 (8.5) | 51.3 (9.0) |
| Incident cases (N, %) | Count (incidence rate) | 2033 (34.4%) | 4036 (30.5%) | 1447 (25.9%) | 7516 (30.4%) |
| Follow-up time (years) | Mean (standard deviation) | 18.9 (5.6) | 19.7 (5.2) | 20.5 (4.7) | 19.7 (5.2) |
| Ethnicity (N, %) | White | 99.4% | 98.8% | 98.3% | 24,411 (98.8%) |
| Asian | 0.2% | 0.5% | 0.7% | 125 (0.5%) | |
| Black | 0.03% | 0.2% | 0.1% | 32 (0.1%) | |
| Other | 0.3% | 0.5% | 0.9% | 135 (0.6%) | |
| Educational Level (N, %) | No qualifications | 20.0% | 14.6% | 9.7% | 3369 (14.7%) |
| O-level or equivalent | 36.9% | 32.7% | 28.4% | 7480 (32.7%) | |
| A-level or equivalent | 23.0% | 25.1% | 26.4% | 5700 (24.9%) | |
| University degree | 20.1% | 27.6% | 35.6% | 6339 (27.7%) | |
| Marital status (N, %) | Married or living as married | 78.2% | 76.9% | 76.7% | 19,057 (77.1%) |
| Separated or divorced | 9.0% | 10.7% | 12.6% | 2654 (10.7%) | |
| Single or widowed | 12.8% | 12.4% | 10.7% | 2992 (12.1%) | |
| Socio-economic status (SES) (N, %) | Routine and manual | 10.8% | 8.7% | 6.6% | 2157 (8.7%) |
| Intermediate | 32.4% | 27.5% | 22.6% | 6805 (27.6%) | |
| Professional and managerial | 56.8% | 63.8% | 70.8% | 15,741 (63.7%) | |
| Physical activity | Low level | 13.9% | 10.3% | 7.1% | 2575 (10.4%) |
| (N, %) | Moderate level | 54.3% | 50.3% | 45.7% | 12,401 (50.2%) |
| High level | 31.8% | 39.4% | 47.2% | 9727 (39.4%) | |
| Body Mass Index (BMI) (kg/m2) | Mean (standard deviation) | 25.0 (4.4) | 24.3 (4.1) | 23.3 (3.5) | 24.2 (4.1) |
| Alcohol (g/d) | Mean (standard deviation) | 7.7 (11.4) | 9.3 (10.5) | 10.0 (8.8) | 9.1 (10.4) |
| Smoking status | Never smoked | 59.4% | 58.7% | 55.6% | 14,366 (58.1%) |
| (N, %) | Ex-smoker | 26.9% | 30.7% | 34.9% | 7597 (30.8%) |
| Current smoker | 13.7% | 10.6% | 9.6% | 2740 (11.1%) | |
| Menopausal status | Premenopausal | 54.9% | 49.2% | 41.4% | 12,054 (48.8%) |
| Postmenopausal | 45.1% | 50.8% | 58.6% | 12,649 (51.2%) | |
| Energy and Food groups (daily consumed) | |||||
| Energy (kcal/d) | Mean (standard deviation) | 2139 (615) | 2323 (701) | 2564 (692) | 2334 (695) |
| Vegetables (g/d) | Mean (standard deviation) | 210 (100) | 311 (162) | 438 (204) | 315 (179) |
| Vegetables_adjusted (g/MJ/d) | Mean (standard deviation) | 24 (12) | 33 (17) | 42 (18) | 33 (17) |
| Fruits (g/d) | Mean (standard deviation) | 206 (153) | 308 (218) | 421 (254) | 309 (226) |
| Fruits_adjusted (g/MJ/d) | Mean (standard deviation) | 24 (17) | 33 (22) | 40 (23) | 32 (22) |
| Total fish (g/d) | Mean (standard deviation) | 22 (16) | 28 (25) | 32 (33) | 28 (26) |
| Total fish_adjusted (g/MJ/d) | Mean (standard deviation) | 2 (2) | 3 (3) | 3 (3) | 3 (3) |
| Processed meat (g/d) | Mean (standard deviation) | 18 (14) | 13 (15) | 5 (10) | 12 (14) |
| Processed meat_adjusted (g/MJ/d) | Mean (standard deviation) | 2 (2) | 1 (1) | 0 (1) | 1 (1) |
| Red meat (g/d) | Mean (standard deviation) | 50 (36) | 34 (40) | 12 (27) | 33 (38) |
| Red meat_adjusted (g/MJ/d) | Mean (standard deviation) | 6 (4) | 3 (4) | 1 (2) | 3 (4) |
| Poultry (g/d) | Mean (standard deviation) | 22 (18) | 17 (20) | 9 (18) | 17 (20) |
| Poultry_adjusted (g/MJ/d) | Mean (standard deviation) | 3 (2) | 2 (2) | 1 (2) | 2 (2) |
| Total meat (g/d) | Mean (standard deviation) | 92 (51) | 66 (62) | 27 (47) | 63 (61) |
| Total meat_adjusted (g/MJ/d) | Mean (standard deviation) | 10 (5) | 7 (6) | 2 (4) | 7 (6) |
Compared with participants with low adherence to Mediterranean diet, those with high adherence were more likely to have high levels of education, SES, and physical activity at baseline. Participants with high adherence and moderate adherence to Mediterranean diet had a lower mean BMI of 23.3 kg/m2 and 24.3 kg/m2 respectively at baseline, compared to those with low adherence (25.0 kg/m2). We also found that the higher adherence to Mediterranean diet, the more consumption of vegetables, fruits, total fish, but the less consumption of total meat including processed meat, red meat, and poultry as shown in Table 1.
Associations between dietary factors and risk of multimorbidity in the whole participants
Risks of multimorbidity associated with consumption of main foods and adherence to Mediterranean diet in the whole participants are shown in Fig. 2. Compared with women with low adherence to Mediterranean diet, those with moderate adherence had a 11% lower risk of multimorbidity (HR = 0.89 (95%CI: 0.84, 0.94)), and those with high adherence had a 22% lower risk (0.78 (0.73, 0.83)) adjusting for age at baseline. Those associations were slightly attenuated for the moderate adherence (0.94 (0.89, 0.99)) and high adherence (0.86 (0.80, 0.92)) in the fully-adjusted models with adjusting for age, ethnicity, marital status, SES, physical activity, BMI, smoking status, alcohol consumption, total energy intake, and menopausal status.
Fig. 2.
Associations of consumed main foods and Mediterranean diet with risk of multimorbidity in the UK Women’s Cohort Study. Age was adjusted in the minimal of the figure.pausal status were adjusted in the fully-adjusted model in the right panelly-adjusted model; Additionally, ethnicity, marital status, socioeconomic status, physical activity, body mass index, smoking status, alcohol consumption, total energy intake, and menopausal status were adjusted in the fully-adjusted model in the right panel of the figure
Regarding food groups in the lower panel of Fig. 2, consumption of per 10 g of processed meat, red meat, and total meat per MJ of total energy consumed were associated with 60% (1.60 (1.37, 1.87)), 19% (1.19 (1.12, 1.26)), and 12% (1.12 (1.08, 1.17)) increased risks of multimorbidity in the fully-adjusted models, respectively. After adjustment for potential confounders, each additional 10 g/MJ of fruits was slightly associated with a lower risk of multimorbidity (0.98 (0.97, 0.99)). There was insufficient evidence of associations between consumption of vegetables (0.99 (0.98, 1.01)), total fish (1.00 (0.92, 1.09)), or poultry (1.07 (0.96, 1.20)) with risk of multimorbidity in the fully-adjusted models.
Subgroup analysis
In stratified analysis by age at baseline, the associations between adherence to Mediterranean diet and risk of multimorbidity were slightly enhanced in the subgroup aged < 60 years (0.92 (0.86, 0.99) for the moderate adherence and 0.84 (0.77, 0.91) for the high adherence), but were attenuated in the subgroup aged ≥ 60 years (0.96 (0.88, 1.05) for the moderate adherence and 0.90 (0.79, 1.01) for the high adherence) in the fully-adjusted models shown in the upper panel of Table 2. However, the p-value for the interaction effect between adherence to Mediterranean diet and age at baseline was non-significant (0.658) where the age was modelled linearly.
Table 2.
Subgroup analyses on associations between Mediterranean diet and multimorbidity risk in UK Women’s Cohort Study
| Stratified variables | Cases/subjects | Person-years | Hazard Ratio (95% Confidence Interval) | P* |
|---|---|---|---|---|
| Age subgroups | 0.658† | |||
| Age < 60 years (N = 19,889) | ||||
| Low adherence | 1224/4470 | 88,638 | 1.00 (Reference) | |
| Moderate adherence | 2598/10,626 | 217,250 | 0.92 (0.86, 0.99) | 0.026 |
| High adherence | 1020/4793 | 100,638 | 0.84 (0.77, 0.91) | < 0.001 |
| Age ≥ 60 years (N = 4814) | ||||
| Low adherence | 809/1432 | 22,893 | 1.00 (Reference) | |
| Moderate adherence | 1438/2590 | 43,211 | 0.96 (0.88, 1.05) | 0.333 |
| High adherence | 427/792 | 14,027 | 0.90 (0.79, 1.01) | 0.081 |
| Body Mass Index (BMI) subgroups | 0.884† | |||
| Normal-weight (BMI ≤ 24.9 kg/m2, N = 16,499) | ||||
| Low adherence | 1004/3442 | 66,907 | 1.00 (Reference) | |
| Moderate adherence | 2295/8808 | 177,548 | 0.92 (0.85, 0.99) | 0.030 |
| High adherence | 1006/4249 | 88,300 | 0.87 (0.79, 0.95) | 0.002 |
| Overweight (BMI > 24.9 kg/m2, N = 8204) | ||||
| Low adherence | 1029/2460 | 44,624 | 1.00 (Reference) | |
| Moderate adherence | 1741/4408 | 82,913 | 0.96 (0.89, 1.04) | 0.355 |
| High adherence | 441/1336 | 26,365 | 0.82 (0.73, 0.92) | 0.001 |
* Adjusted for age, ethnicity, marital status, socioeconomic status, physical activity, body mass index, smoking status, alcohol consumption, total energy intake, and menopausal status
†P-interaction represents the statistical significance for interaction items of Mediterranean diet score and age/BMI where age or BMI was modelled linearly in the Cox proportional regression
Further, using adherence to Mediterranean diet as a continuous variable in more finely age-stratified analyses shown in Supplementary Table S5, we observed a graded inverse association with multimorbidity risk: each 1-point increase in Mediterranean diet score corresponded to a 2% lower risk (0.98 (0.96, 1.00) in women aged < 45 years, 3% lower (45–49 years), and 4% lower (50–59 years). No significant association was observed in participants aged ≥ 60 years (0.99 (0.96, 1.02).
Other subgroup analyses stratified by age or BMI showed that the associations between consumed main foods, Mediterranean diet and risk of multimorbidity were consistent with those in the whole population, where p-values for the interaction effect between age/BMI and each dietary factor were all non-significant (≥ 0.05) shown in Tables 2 and 3.
Table 3.
Subgroup analyses on associations between food consumption and multimorbidity risk in UK Women’s Cohort Study
| Age subgroups | P for interactions with agea | BMI subgroups | P for interaction with BMIa | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Age < 60 years (N = 19,889) |
Age ≥ 60 years (N = 4814) |
BMI ≤ 24.9 kg/m2 (N = 16,499) | BMI > 24.9 kg/m2 (N = 8204) | |||||||
| HR (95%CI) | P* | HR (95%CI) | P* | HR (95%CI) | P* | HR (95%CI) | P* | |||
| Vegetables | 0.99 (0.98, 1.01) | 0.551 | 1.00 (0.97, 1.02) | 0.697 | 0.842 | 0.99 (0.98, 1.01) | 0.427 | 0.99 (0.97, 1.02) | 0.641 | 0.296 |
| Fruits | 0.98 (0.97, 1.00) | 0.009 | 0.97 (0.95, 0.98) | < 0.001 | 0.730 | 0.97 (0.96, 0.98) | < 0.001 | 0.98 (0.97, 1.00) | 0.053 | 0.120 |
| Total fish | 0.97 (0.87, 1.08) | 0.533 | 1.07 (0.93, 1.22) | 0.347 | 0.589 | 1.01 (0.91, 1.13) | 0.790 | 0.97 (0.84, 1.11) | 0.651 | 0.568 |
| Processed meat | 1.60 (1.32, 1.94) | < 0.001 | 1.67 (1.28, 2.19) | < 0.001 | 0.431 | 1.88 (1.51, 2.35) | < 0.001 | 1.39 (1.11, 1.73) | 0.003 | 0.097 |
| Red meat | 1.20 (1.12, 1.30) | < 0.001 | 1.19 (1.07, 1.31) | 0.001 | 0.493 | 1.26 (1.16, 1.37) | < 0.001 | 1.12 (1.03, 1.22) | 0.009 | 0.050 |
| Poultry | 1.11 (0.97, 1.27) | 0.133 | 1.01 (0.83, 1.23) | 0.939 | 0.382 | 1.03 (0.88, 1.20) | 0.726 | 1.10 (0.94, 1.30) | 0.238 | 0.060 |
| Total meat | 1.13 (1.08, 1.18) | < 0.001 | 1.13 (1.06, 1.22) | 0.001 | 0.714 | 1.16 (1.10, 1.22) | < 0.001 | 1.09 (1.03, 1.16) | 0.003 | 0.395 |
*Adjusted for age, ethnicity, marital status, socioeconomic status, physical activity, body mass index, smoking status, alcohol consumption, total energy intake, and menopausal status
aP-interaction represents the statistical significance for interaction items of each food group and age/BMI where age or BMI was modelled linearly in the Cox proportional regression
BMI body mass index, HR Hazards Ratio, 95%CI 95% Confidence Interval
Sensitivity analysis
Our sensitivity analyses yielded three key findings regarding the robustness of dietary associations: First, when modeling food intakes per 1-SD increment (Supplementary Table S6, columns 1–3), the directionality of all associations remained consistent with our primary 10 g/MJ models, though with attenuated effect sizes. For instance, processed meat showed an HR of 1.07 (95% CI: 1.05–1.09) per SD in the fully-adjusted model, reflecting the different scaling approaches while maintaining statistical significance.
Second, in mutually adjusted models accounting for intercorrelations between food groups (Supplementary Table S6, last two columns), all significant associations persisted with only modest attenuation. Processed meat maintained a robust association (HR = 1.56, 95% CI: 1.33–1.83) after adjustment for other dietary components, suggesting its effects are independent of overall dietary patterns.
Additionally, sensitivity analyses produced broadly consistent estimates when excluding: (1) participants with < 5 years of follow-up (n = 747) and (2) those receiving long-term medical treatment (n = 6,558). These results aligned with our primary findings for both Mediterranean diet adherence (Supplementary Table S7) and main food group consumption (Supplementary Table S8), demonstrating the robustness of our observations.
Discussion
In this large prospective cohort study of UK women, we demonstrated significant inverse associations between Mediterranean diet adherence and multimorbidity risk, with a clear dose–response relationship. These findings align with existing evidence, including a cross-sectional study of 1,140 Cypriot adults showing lower multimorbidity odds in the highest versus lowest Mediterranean diet tertile [8]. However, our longitudinal findings extend cross-sectional Mediterranean diet evidence by establishing 20-year follow-up, and demonstrating population-level generalizability (n = 24,703 vs 1,140). Additional European longitudinal evidence supports our findings. A Spanish study (n = 2,784; baseline age ≥ 65 years) similarly demonstrated that greater Mediterranean diet adherence predicted fewer chronic conditions over follow-up since 2015 [28]. While these results align with ours, our study extends the evidence by including younger age groups and longer follow-up duration. However, a British cohort study of older men (n = 2,873; aged 60–79 years at baseline) found no significant association between a Mediterranean-based Elderly Dietary Index and cardiometabolic multimorbidity [29]. This discrepancy may reflect potential sex-specific biological differences in dietary effects, and broader age range capturing earlier disease development.
The observed inverse association between Mediterranean diet adherence and multimorbidity risk may be mediated through multiple synergistic biological pathways. First, the diet's established efficacy against individual chronic conditions—including cardiovascular diseases, cancers, and cognitive decline [30]—potentially reduces disease co-occurrence collectively. Second, the diet's high content of polyphenols (from olive oil, nuts, and red wine) and omega-3 fatty acids (from fish) exerts potent anti-inflammatory effects by reducing circulating inflammatory markers like IL-6 and CRP [31]. Additionally, the Mediterranean diet improves metabolic syndrome through its rich content of fiber, omega fatty acids, and polyphenols, which collectively reduce oxidative stress, inflammation, and improve metabolic function. These bioactive components help prevent obesity, dyslipidemia, hypertension and diabetes, and their protective effects are mediated primarily through anti-inflammatory and antioxidant pathways [32]. These mechanisms collectively target fundamental aging processes—chronic inflammation, oxidative stress, and metabolic dysfunction—that drive multiple chronic conditions simultaneously. Our subsequent analysis of specific food groups (e.g., vegetables, processed meat) further supports these pathways.
In this study, we also found significantly positive associations between consumption of processed meat, red meat, and total meat with risk of multimorbidity. These results are consistent with findings from a longitudinal study using the UK Biobank dataset showing that high consumption of processed meat was associated with higher risks of multimorbidity [19]. Another cohort study conducted among 53,867 middle-aged Australian adults confirmed that high consumption of red meat was one of the leading predictors for multimorbidity of chronic conditions in women [20]. The rationale for the positive association between consumption of processed meat and red meat with risk of multimorbidity remains unclear. One potential factor may be the richness of energy, saturated fat, or protein in meat, since our previous study found that the highest quintile of daily intakes of energy and protein were positively associated with an increased risk of multimorbidity compared with the lowest quintile [24]. In addition, high consumption of red meat, especially processed red meat, is considered as an important part of a pro-inflammatory diet [33], while the dietary-inflammation was observed to be positively associated with risk of multimorbidity [34]. Furthermore, consumption of meat, in particular processed meat, may result in increased intakes of detrimental substances like saturated fat, nitrates, nitrites, or amines [35], which potentially associated with high risks of chronic diseases [36, 37]. However, more related research is needed to confirm the relationships between dietary factors and risk of multimorbidity.
Our age-stratified analyses revealed a nonlinear association, with the protective effect of Mediterranean diet adherence progressively strengthening until age 60 years before attenuating in older participants. While this specific pattern of age-modification has not been previously reported in younger adults (18–60 years), existing evidence consistently links Mediterranean dietary patterns with healthier aging and reduced risk of age-related diseases [38]. Given that advanced age is a well-established risk factor for multimorbidity [20, 39], we hypothesize that conventional age-adjustment may be insufficient to fully account for the overwhelming biological aging processes in older populations with multimorbidity. This residual confounding could potentially obscure diet-multimorbidity associations in individuals ≥ 60 years, where age-related physiological changes may dominate over modifiable risk factors.
This study was strengthened by the usage of matching data with hospital records, making it possible to assess multimorbidity based on ICD codes more precisely, and less likely to be lost to follow-up over a long-lasting follow-up. Measuring diets of participants through a detailed and validated FFQ with 217 food items, provided a more high-quality dietary dataset for exploring dietary risk factors of the multimorbidity. Moreover, a prospective study design with a long follow-up period and a large sample size provided a unique opportunity to investigate the long-term effects of diet on multimorbidity.
Nonetheless, several limitations should not be ignored. First, although prevalent multimorbid participants have been excluded, individuals with self-reported being on long-time treatment for illness at baseline who were more likely to have multimorbidity, were not excluded in Cox models which may potentially introduce selection bias. Moreover, as an observational study, reverse causality cannot be avoided entirely which might mask the real relationships. However, results were robust to exclusion of participants on long-term treatment for illness or those with a follow-up period less than 5 years in sensitivity analyses. Our study is also limited in that our study population consisted exclusively of White British women from the UKWCS, and all hospital admission data came from English NHS records. This homogeneity in ethnicity, sex, and healthcare system context may limit the generalizability of our results to other demographic groups or healthcare settings. These limitations underscore the need for multicenter studies incorporating diverse populations and healthcare systems to address these constraints.
Conclusion
Our study revealed some relationships between foods, Mediterranean diet, and risk of multimorbidity. Higher adherence to Mediterranean diet was observed to be associated with a lower risk of multimorbidity. Conversely, higher consumption of meat, especially processed meat and red meat was found to be associated with a higher risk of multimorbidity. Our findings highlight that the identified dietary factors, Mediterranean diet adherence (protective) and processed/red meat consumption (harmful), represent modifiable targets for population-level multimorbidity prevention programs targeting middle-aged populations.
Supplementary Information
Acknowledgements
The authors thank the participants of the UK Women’s Cohort Study for the provision of the data and the cohort team members who contributed to data collection. We also thank the World Cancer Research Fund (to J.E.C.) which funded the UK Women’s Cohort.
Abbreviations
- 95%CI
95% Confidence intervals
- BMI
Body mass index
- CCI
Charlson comorbidity index
- FFQs
Food frequency questionnaires
- HES
Hospital episode statistics
- HR
Hazard ratios
- ICD
International classification of diseases
- MUFA
Monounsaturated fatty acids
- NHS
National health service
- PUFA
Polyunsaturated fatty acids
- SD
Standard deviation
- SES
Socio-economic status
- SFA
Saturated fatty acids
- UKWCS
UK women’s cohort study
Authors’ contributions
Conceptualization, H.F.Z., J.H. and J.E.C; Methodology, H.F.Z., X.W., P.S., X.D. and T.G.; Software, H.F.Z. X.W., Y.Y.D. and P.S.; Validation, X.D. Y.Y.D. and H.N.Z; Formal analysis, H.F.Z. Y.Y.D. and P.S.; Resources and data curation, J.E.C.; Writing—original draft preparation, H.F.Z.; Writing—review and editing, T.G., J.H., and J.E.C.; Visualization, H.F.Z. Y.Y.D. and X.D..; Supervision, J.E.C., and J.H.; Project administration, H.F.Z., J.H., and J.E.C.; Funding acquisition, H.Z., X.W., and J.E.C. All authors have read and agreed to the published version of the manuscript.
Funding
This research was funded by the "Phoenix Introduction Plan" Project of Tangdu Hospital, Fourth Military Medical University (2025YFJH005), the General Project of Basic Research Plan of Shaanxi Province (2024-JC-YBMS-700), and the Natural Science Basic Research Program of Shaanxi Province (2023-JC-QN-0939). The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
Data availability
The datasets used and analyzed during the current study are not publicly available due to privacy/ethical restrictions, but are available from the UKWCS data access committee on reasonable request.
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki, and approved by National Research Ethics Service Committee for Yorkshire & the Humber – Leeds East (Ref: 15/YH/0027), and was updated to include linkage outcomes and related sub-studies by Health Research Authority REC (Ref: 17/YH/0144). Informed consent has been obtained from all subjects involved in the study at recruitment of the cohort.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Technical Series on Safer Primary Care: Multimorbidity. Availabe online: https://www.who.int/publications/i/item/9789241511650 (Accessed on 05 Apr 2025).
- 2.Multimorbidity: a priority for global health research. Availabe online: https://allcatsrgrey.org.uk/wp/wpfb-file/82222577-pdf/ (Accessed on 15 Mar 2025).
- 3.Making more of multimorbidity. an emerging priority. Lancet (London, England). 2018;391:1637. 10.1016/s0140-6736(18)30941-3. [DOI] [PubMed] [Google Scholar]
- 4.Velek P, Luik AI, Brusselle GGO, Stricker BC, Bindels PJE, Kavousi M, et al. Sex-specific patterns and lifetime risk of multimorbidity in the general population: a 23-year prospective cohort study. BMC Med. 2022;20:304. 10.1186/s12916-022-02487-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Violan C, Foguet-Boreu Q, Flores-Mateo G, Salisbury C, Blom J, Freitag M, et al. Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational studies. PLoS One. 2014;9:e102149. 10.1371/journal.pone.0102149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kingston A, Robinson L, Booth H, Knapp M, Jagger C. Projections of multi-morbidity in the older population in England to 2035: estimates from the population ageing and care simulation (PACSim) model. Age Ageing. 2018;47:374–80. 10.1093/ageing/afx201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Dekker LH, de Borst MH, Meems LMG, de Boer RA, Bakker SJL, Navis GJ. The association of multimorbidity within cardio-metabolic disease domains with dietary patterns: a cross-sectional study in 129 369 men and women from the Lifelines cohort. PLoS One. 2019;14:e0220368. 10.1371/journal.pone.0220368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kyprianidou M, Panagiotakos D, Faka A, Kambanaros M, Makris KC, Christophi CA. Adherence to the Mediterranean diet in Cyprus and its relationship to multi-morbidity: an epidemiological study. Public Health Nutr. 2021;24:4546–55. 10.1017/s1368980020004267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Vicinanza R, Bersani FS, D’Ottavio E, Murphy M, Bernardini S, Crisciotti F, et al. Adherence to Mediterranean diet moderates the association between multimorbidity and depressive symptoms in older adults. Arch Gerontol Geriatr. 2020;88:104022. 10.1016/j.archger.2020.104022. [DOI] [PubMed] [Google Scholar]
- 10.Lachat C, Hawwash D, Ocké MC, Berg C, Forsum E, Hörnell A, et al. Strengthening the reporting of observational studies in epidemiology-nutritional epidemiology (STROBE-nut): an extension of the STROBE statement. PLoS Med. 2016;13:e1002036. 10.1371/journal.pmed.1002036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cade JE, Burley VJ, Alwan NA, Hutchinson J, Hancock N, Morris MA, et al. Cohort profile: the UK women’s cohort study (UKWCS). Int J Epidemiol. 2017;46:e11. 10.1093/ije/dyv173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Webster J, Greenwood DC, Cade JE. Risk of hip fracture in meat-eaters, pescatarians, and vegetarians: results from the UK women’s cohort study. BMC Med. 2022;20:275. 10.1186/s12916-022-02468-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Riboli, E.; Kaaks, R. The EPIC Project: rationale and study design. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol. 1997; 26 (Suppl 1): S6–14, 10.1093/ije/26.suppl_1.s6. [DOI] [PubMed]
- 14.Spence, M.; Cade, J.; Burley, V.; Greenwood, D. Ability of the UK Women's Cohort Study Food Frequency Questionnaire to rank dietary intakes: a preliminary validation study. In Proceedings of Proceedings of the Nutrition Society, London; p. 117A.
- 15.Cade JE, Burley VJ, Greenwood DC. The UK women’s cohort study: comparison of vegetarians, fish-eaters and meat-eaters. Public Health Nutr. 2004;7:871–8. 10.1079/phn2004620. [DOI] [PubMed] [Google Scholar]
- 16.Food Standards Agency (FSA). Food Portion Sizes, 3rd ed.; Mills, A., Patel, S., Crawley, H., Eds.; The Stationary Office: London, UK, 2002.
- 17.Trichopoulou A, Kouris-Blazos A, Wahlqvist ML, Gnardellis C, Lagiou P, Polychronopoulos E, et al. Diet and overall survival in elderly people. Br Med J. 1995;311:1457–60. 10.1136/bmj.311.7018.1457. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Trichopoulou A, Orfanos P, Norat T, Bueno-de-Mesquita B, Ocke MC, Peeters PH, et al. Modified Mediterranean diet and survival: EPIC-elderly prospective cohort study. BMJ. 2005;330:991. 10.1136/bmj.38415.644155.8F. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zhang Y, Chen H, Carrillo-Larco RM, Lim CCW, Mishra SR, Yuan C, et al. Association of dietary patterns and food groups intake with multimorbidity: a prospective cohort study. Clin Nutr ESPEN. 2022;51:359–66. 10.1016/j.clnesp.2022.07.019. [DOI] [PubMed] [Google Scholar]
- 20.Shang X, Peng W, Wu J, He M, Zhang L. Leading determinants for multimorbidity in middle-aged Australian men and women: a nine-year follow-up cohort study. Prev Med. 2020;141:106260. 10.1016/j.ypmed.2020.106260. [DOI] [PubMed] [Google Scholar]
- 21.Godfray HCJ, Aveyard P, Garnett T, Hall JW, Key TJ, Lorimer J, et al. Meat consumption, health, and the environment. Science. 2018. 10.1126/science.aam5324. [DOI] [PubMed] [Google Scholar]
- 22.Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol. 2004;57:1288–94. 10.1016/j.jclinepi.2004.03.012. [DOI] [PubMed] [Google Scholar]
- 23.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–83. 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 24.Song G, Li W, Ma Y, Xian Y, Liao X, Yang X, et al. Nutrient intake and risk of multimorbidity: a prospective cohort study of 25,389 women. BMC Public Health. 2024;24:696. 10.1186/s12889-024-18191-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.The National Statistics Socio-Economic Classification User Manual. Availabe online: https://www.ons.gov.uk/ons/guide-method/classifications/archived-standard-classifications/soc-and-sec-archive/the-national-statistics-socio-economic-classification--user-manual.pdf (Accessed on 28 Mar 2025).
- 26.Darin-Mattsson A, Fors S, Kåreholt I. Different indicators of socioeconomic status and their relative importance as determinants of health in old age. Int J Equity Health. 2017;16:173. 10.1186/s12939-017-0670-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ)-short and long forms. Availabe online: http://www.ipaq.ki.se/scoring.pdf (Accessed on 28 Mar 2025).
- 28.Vega-Cabello V, Struijk EA, Caballero FF, Yévenes-Briones H, Ortolá R, Calderón-Larrañaga A, et al. Diet quality and multimorbidity in older adults: a prospective cohort study. J Gerontol A Biol Sci Med Sci. 2024. 10.1093/gerona/glad285. [DOI] [PubMed] [Google Scholar]
- 29.Wang Q, Schmidt AF, Lennon LT, Papacosta O, Whincup PH, Wannamethee SG. Prospective associations between diet quality, dietary components, and risk of cardiometabolic multimorbidity in older British men. Eur J Nutr. 2023;62(7):2793–804. 10.1007/s00394-023-03193-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Dominguez LJ, Donat-Vargas C, Sayon-Orea C, Barberia-Latasa M, Veronese N, Rey-Garcia J, et al. Rationale of the association between Mediterranean diet and the risk of frailty in older adults and systematic review and meta-analysis. Exp Gerontol. 2023;177:112180. 10.1016/j.exger.2023.112180. [DOI] [PubMed] [Google Scholar]
- 31.Itsiopoulos C, Mayr HL, Thomas CJ. The anti-inflammatory effects of a Mediterranean diet: a review. Curr Opin Clin Nutr Metab Care. 2022;25:415–22. 10.1097/mco.0000000000000872. [DOI] [PubMed] [Google Scholar]
- 32.Dayi T, Ozgoren M. Effects of the Mediterranean diet on the components of metabolic syndrome. J Prev Med Hyg. 2022;63:E56-e64. 10.15167/2421-4248/jpmh2022.63.2S3.2747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Rodríguez-Molinero J, Migueláñez-Medrán BDC, Puente-Gutiérrez C, Delgado-Somolinos E, Martín Carreras-Presas C, Fernández-Farhall J, et al. Association between oral cancer and diet: an update. Nutrients. 2021. 10.3390/nu13041299. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Mahajan H, Lieber J, Carson Mallinson PA, Bhogadi S, Banjara SK, Kinra S, et al. The higher dietary inflammation is associated with a higher burden of multimorbidity of cardio-metabolic and mental health disorders in an urbanizing community of southern India: a cross-sectional analysis for the APCAPS cohort. Human Nutrition & Metabolism. 2024;36:200254. 10.1016/j.hnm.2024.200254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Rohrmann S, Linseisen J. Processed meat: the real villain? Proc Nutr Soc. 2016;75:233–41. 10.1017/s0029665115004255. [DOI] [PubMed] [Google Scholar]
- 36.Mirmiran P, Teymoori F, Farhadnejad H, Mokhtari E, Salehi-Sahlabadi A. Nitrate containing vegetables and dietary nitrate and nonalcoholic fatty liver disease: a case control study. Nutr J. 2023;22:3. 10.1186/s12937-023-00834-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Yoo W, Zieba JK, Foegeding NJ, Torres TP, Shelton CD, Shealy NG, et al. High-fat diet-induced colonocyte dysfunction escalates microbiota-derived trimethylamine N-oxide. Science. 2021;373:813–8. 10.1126/science.aba3683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Malcomson FC, Mathers JC. Nutrition and Ageing. Subcell Biochem. 2018;90:373–424. 10.1007/978-981-13-2835-0_13. [DOI] [PubMed] [Google Scholar]
- 39.Tazzeo C, Zucchelli A, Vetrano DL, Demurtas J, Smith L, Schoene D, et al. Risk factors for multimorbidity in adulthood: a systematic review. Ageing Res Rev. 2023;91:102039. 10.1016/j.arr.2023.102039. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and analyzed during the current study are not publicly available due to privacy/ethical restrictions, but are available from the UKWCS data access committee on reasonable request.