Riviste UNIMI
Not a member yet
21278 research outputs found
Sort by
«Revised and corrected by the Author»: The 1947 Edition of Amo, dunque sono by Sibilla Aleramo
No full textIl contributo indaga le varianti fra la redazione pubblicata nel 1947 del romanzo Amo, dunque sono di Sibilla Aleramo rispetto alla princeps del 1927. Dalla collazione delle due edizioni emerge infatti un lavoro di sfrondamento e sottrazione del testo, a cui contribuì anche Franco Matacotta, all’epoca compagno di Aleramo. Il risultato è una redazione non solo meno corposa, ma anche rimaneggiata per rimuovere elementi stilistici e di contenuto considerati inadatti al nuovo clima culturale e politico del dopoguerra.This paper examines the textual variants between the 1947 edition of Sibilla Aleramo’s novel Amo, dunque sono and its first edition, published in 1927. A comparison of the two editions reveals a process of abridgement and omission, to which Franco Matacotta – Aleramo’s partner at the time – also contributed. The result is an edition not only less substantial, but also revised to eliminate stylistic and thematic elements considered unsuitable for the new post-war cultural and political climate
Nuovi dati sulla formazione del patrimonio dei ‘venerabili monasteri di S. Maria di Licodia e di S. Nicolò l’Arena’ in Sicilia: tre privilegi inediti dei sovrani aragonesi (1336-1361)
No full textThe article examines the formation of the patrimony of the monasteries of Santa Maria di Licodia and San Nicolò l\u27Arena in Sicily, moving from the analysis of three unpublished privileges granted by the Aragonese sovereigns between 1336 and 1361. The study is based on documents preserved in the Tabulario of the ‘venerable monasteries’, which reveal the strategies adopted by the monks to protect and expand their land holdings through royal privileges, deeds of gift, purchases and sales, permutations and testamentary bequests. The privileges studied include a letter from King Peter II settling a dispute over an olive grove; a privilege from Infante Giovanni confirming a land donation; and a ratification from Queen Constance of Aragon on a land exchange. Not only do these documents illustrate the economic transactions carried out by the monks, but they also offer insight into the political and religious interests of the Aragonese rulers, the internal dynamics of the Sicilian aristocracy, and relations with the local notables. The article thus attempts to contribute to the understanding of the historical and socio-economic dynamics of Aragonese Sicily and the formation of the large monastic estates on the southern slope of Etna in the 14th century.L’articolo esamina la formazione del patrimonio dei monasteri di Santa Maria di Licodia e di San Nicolò l’Arena in Sicilia, muovendo dall’analisi di tre privilegi inediti concessi dai sovrani aragonesi tra il 1336 e il 1361. Lo studio si basa su documenti conservati nel Tabulario dei \u27venerabili monasteri\u27, i quali rivelano le strategie adottate dai monaci per tutelare ed espandere i loro possedimenti terrieri attraverso privilegi regi, atti di donazione, compravendite, permute e lasciti testamentari. I privilegi studiati comprendono una lettera di re Pietro II che risolve una disputa su un uliveto; un privilegio dell’infante Giovanni che conferma una donazione di terre e una ratifica della regina Costanza d’Aragona su una permuta di terre. Questi documenti non solo illustrano le operazioni economiche compiute dai monaci, ma offrono anche uno spaccato degli interessi politici e religiosi dei sovrani aragonesi, delle dinamiche interne all\u27aristocrazia siciliana e dei rapporti con il notabilato locale. L’articolo tenta di contribuire, così, alla comprensione delle dinamiche storiche e socio-economiche della Sicilia aragonese e alla formazione delle grandi proprietà monastiche nel versante meridionale dell\u27Etna nel XIV secolo
Lo spazio della paternità nei Carmina di Venanzio Fortunato
No full textL’articolo indaga lo spazio che al vasto tema della paternità è concesso da Venanzio Fortunato nella sua silloge poetica. Prendendo le mosse da alcune considerazioni a carattere generale sulla figura del pater e sulla sua autorità nei secoli di passaggio dal mondo romano a quello medievale, è possibile delineare le diverse tipologie di figure paterne presenti nei Carmina: vescovi, re, genitori comuni. Questi protagonisti dei versi di Venanzio esplicano il loro ruolo in maniera alquanto diversa, ad esempio ora come pastori del popolo di Dio, ora come guide di un regno ma anche come padri di futuri eredi al trono, ora come genitori in ansia per la sorte della loro prole. La sensibilità poetica di Venanzio permette altresì di cogliere il ventaglio di significative emozioni che arrischiscono i legami esistenti tra i patres e i loro figli, che siano carnali o spirituali.The article examines the space given by Venantius Fortunatus to the vast theme of fatherhood in his poetic sylloge. Starting from some general considerations on the figure of the pater and his authority in the centuries of transition from the Roman to the medieval world, it is possible to delineate the different types of paternal figures present in the Carmina: bishops, kings, ordinary parents. These protagonists of Venantius\u27s verses fulfil their role in very different ways, now as shepherds of the people of God, now as rulers of a kingdom but also as fathers of future heirs to the throne, now as parents worried about the fate of their offspring. Venantius\u27s poetic sensibility also allows us to grasp the range of significant emotions that underlie the bonds between patres and their children, whether carnal or spiritual
The Algerian-Italian relations: A strategic partnership in a turbulent geopolitical context
No full textLe relazioni algerino-italiane: un partenariato strategico in un contesto geopolitico turbolento
Negli ultimi venticinque anni, le relazioni tra l’Algeria e l’Italia si sono evolute in uno dei partenariati bilaterali più strutturati e resilienti che collegano il Nord Africa e l’Unione Europea. Il Trattato di Amicizia del 2003 ha istituzionalizzato il dialogo politico e la cooperazione economica, ponendo al centro l’interdipendenza energetica, fondata sul gasdotto TransMed. Oltre al settore degli idrocarburi, i legami storici e il costante dialogo ad alto livello hanno contribuito a gestire le divergenze politiche e a preservare la continuità diplomatica. Tuttavia, il partenariato rimane asimmetrico e settoriale: la politica estera italiana è incardinata negli impegni dell’UE e della NATO, mentre l’Algeria mantiene una forte adesione ai principi di sovranità e di non ingerenza. Ne consegue che la cooperazione risulta più stabile sul piano economico che in quello della sicurezza. Le relazioni algerino-italiane rivelano così le asimmetrie delle interazioni Nord Africa-UE, dove la cooperazione pragmatica coesiste con gerarchie persistenti e con l’intreccio tra resilienza autoritaria e pragmatismo democratico.Over the past twenty-five years, Algerian-Italian relations have evolved into one of the most structured and resilient bilateral partnerships linking North Africa and the European Union. The 2003 Treaty of Friendship institutionalized political dialogue and economic cooperation, with energy interdependence – anchored in the TransMed pipeline – forming the core of mutual interests. Beyond hydrocarbons, historical ties and sustained high-level exchanges have helped manage political divergences and preserve diplomatic continuity. Yet the partnership remains asymmetrical and sectorial: Italy’s policies are framed by EU and NATO commitments, while Algeria upholds principles of sovereignty and non-interference. Consequently, collaboration is more consistent in the economic than in the security domains. Algerian-Italian relations thus reveal the asymmetries of North Africa-EU interaction, where pragmatic cooperation coexists with enduring hierarchies and the interplay of authoritarian resilience and democratic pragmatism
Prevalence of Anemia Among Pregnant Women Attending Antenatal Care in Dhamar Governorate, Yemen: A Cross-Sectional Study
No full textIntroduction
Anemia, a chronic condition affecting millions globally, poses greater risks to pregnant mothers, particularly in regions like Yemen. However, data on pregnancy-related anemia in Yemen is scarce. This study aimed to gauge anemia prevalence in Yemen\u27s Dhamar governorate, identifying associated pregnancy-related risk factors.
Methods
Through a questionnaire-based cross-sectional study involving a total of 157 pregnant participants, data on sociodemographics (age, education, income) and health history (gestational age, parity, hemorrhage) were collected. Utilizing SPSS v25, frequencies, percentages, and binary logistic regression (OR, AOR) with 95% confidence intervals (p<0.05) were calculated.
Discussions
Results highlighted 17% of pregnant women in medical centers had anemia (43% of which are mild, 23% moderate, 33% severe). Significantly, rural residence and lower socioeconomic status were linked (p=0.021, 0.015) with AORs of 0.305 (CI: 0.111-0.835) and 3.86 (CI: 1.294-11.542).
Conclusion
In conclusion, anemia affects pregnant women, particularly those in rural areas with lower incomes. Addressing this requires targeted medical interventions to mitigate anemia\u27s prevalence and impact during pregnancy
Deficits in Alcohol Enforcement Legislation in Ireland: Bottomless Brunches
No full textAlcohol remains a clear and present danger to population health in Ireland. In response to this threat, the Irish Government passed the Public Health (Alcohol) Act, 2018. In addition to introducing Minimum Unit Pricing (MUP) and alcohol warning labels, this legislation also prohibits the selling of alcohol at a reduced price and the sale of alcohol in a manner likely to encourage alcohol consumption. However, this paper identifies numerous examples of ‘bottomless brunches’ in Ireland wherein unlimited alcohol is provided for a fixed price within a certain time period. A robust response from the Health Services Executive (HSE) and the Government is required to respond to alcohol as a Commercial Determinant of Health (CDoH)
Application of Logistic Regression and Random Forests to Assess the Relevance of Chrononutrition Information for Prediction of Overweight in the INRAN-SCAI 2005-2006 Nutrition Survey
No full textIntroduction
Obesity is a world wide crisis nowadays it contributes to a number of serious diseases like type 2 diabetes[1] and coronary heart disease. Chrononutrition and in particular timing of eating, has been shown to have an influence on obesity in studies with animal models[2]. Previous human epidemiological studies have already investigated the association between timing and regularity of eating with BMI [3,4]. However to date no study specifically assessed the added value of using data on the timing and regularity of calorie intake for predicting obesity, on top of total calorie intake across the day, when this is information is available in nutritional epidemiological studies like surveys with diet diaries.
Objectives
The target population of this study are Italian adults aged 18 to 64 years old from a past cross sectional nutrition survey INRAN-SCAI[5]. The aim of the study is to compare the performance, in predicting overweight (BMI>25), of models based on 6 day time intervals of calorie intake, to models based on the total calorie intake across the day and comparing prediction performance of logistic regression and random forests using ROC curves.
Methods
Data collected during the Italian nationally representative food consumption survey INRAN-SCAI in the years 2005-2006 was used to make the predictive models. The data used for our analysis was collected from 2312 adults aged 18-64, in the form of a diet diary over 3 consecutive days as well as sociodemographic questionnaire. The days were divided into 6 time intervals, which correspond to the times of 3 main meals and the intervals between them (6am-9am; 9am-12pm; 12pm-3pm; 3pm-7pm; 7pm-10pm; 10pm-6am).
Three different types of models were compared in this study: (i) models trained on the mean energy intake and irregularity of the 6 time intervals, (ii) models trained on the 6 time intervals but using repeated measures from the 3 days and (iii) models trained using the mean energy intake and irregularity for the whole day.
Logistic regression models were built using the whole data sample as well as separating the training(70%) and testing(30%) set for the model, with overweight (chosen due to the small number of strict obesity cases) as outcome and including as predictors also several available sociodemographic variables (including physical activity) chosen by a preliminary application of a forward variable selection algorithm. Sensitivity, specificity, negative predictive value, positive predictive value and error rate were calculated for two different cut off points, one being 0.5 and the second being the optimal cutoff point maximizing the specificity and sensitivity simultaneously. Cross validated models (10 fold) were also generated. The ROC curves corresponding to all the models were compared
For all 3 conditions random forest models were also generated. Their ROC curves were compared and metrics such as specificity, sensitivity NPV, PPV and error rate were in turn calculated.
Results
When logistic regression models were trained on 100% of the data sample (including 34.6% overweight subjects), the models using time intervals repeated measures or means both performed better than the model using a day mean. The AUC for repeated measures was 0.7664, very similar to AUC for interval means (0.7639), the AUC for the mean of the day model was 0.7520. The difference between them was statistically significant both when tested across the whole ROC curve and when tested for optimal cutoff point and 0.5 cutoff point. At the optimal cutoff point specificity, NPV, PPV and error rate performed better in the models using time intervals, whereas sensitivity was higher in the model using day mean. For the 0.5 cutoff the models with intervals performed better in all values except for specificity, which was roughly the same for all models.
After separating the training set and test set, the differences in AUC between the models were no longer statistically significant. Similar result was obtained for cross validated models, the differences in AUCs were slightly larger than when separating the test and training sets by hand but the confidence intervals were overlapping. All the metrics (sensitivity, specificity, PPV, NPV and error rate) for training and test sets for all 3 models were very close, at both mentioned cutoff points, the only one which stood out was sensitivity in the repeated measures at 0.5 cutoff (40.7% vs. 37.1%, 35.5%).
The random forest models also did not show a significant difference between the AUC corresponding to the 3 conditions. The sensitivity for the random forest models was generally low (38% for mean of the day, 32.9% for means of intervals, 36.1% for repeated measures). NPV was much higher for the mean of the day model at 86.3%, compared to around 70% for others while the rest of the metrics, also except for sensitivity, performed better in the models using time intervals. Comparison of sensitivity/specificity/ROC for random forests and 0.5 cut-off of logistic models are shown in Figure 1.
Conclusions
The study provides an insight into how chrononutritional data based on repeated diet diaries summarized by calories consumed in 6 time intervals (as defined in previous work [3,4] ) compares in terms of overweight prediction with using just mean total calorie intake in the day, while also comparing the performance of a machine learning technique like random forest with a classical method like logistic regression. The area under ROC was significantly higher when using time intervals (repeated or averaged) compared to whole day only when they were trained on 100% of the data sample, and this result became nonsignificant after a 30% test set was randomly taken from the sample, both within the training and test sample.
Chrononutrition information in this case allowed for a significantly better prediction of overweight only when testing on the same data as the model was trained on and may thus be attributable to overfitting. Hence the results, despite being based on a nationally representative sample, don’t seem to be generalizable for public health purposes. On the other hand, population heterogeneity may also be hindering prediction. Another limitation is the cross-sectional nature of the INRAN-SCAI nutrition survey. In conclusion, timing/regularity of eating may still capture useful information to study overweight/obesity but a properly powered prospective study is warranted to truly assess its potential different impact/relevance also for subgroups of the population
Intensity and Type of Physical Activity and Semen Quality in Young Healthy Men Living in Brescia, North Italy
No full textINTRODUCTION
A decline in semen quality has been documented over recent decades, possibly linked to rising rates of obesity, unhealthy diets, and exposure to environmental toxins, although the exact causes remains a topic of debate [1]. Recreational physical activity (PA) has been associated with numerous heath benefits, including the prevention of chronic diseases and the promotion of overall well-being, and it is strongly recommended by the World Health Organization (WHO) and various Scientific Societies.
Physical activity is hypothesized to have a positive effect on semen quality due to its favorable impact on metabolic and endocrine functions [2]. However, excessive physical activity may have the opposite effect, potentially leading to reduced semen quality and fertility. This may be due to impaired function of the hypothalamic-pituitary-gonadal axis, increased oxidative stress, and chronic inflammation [3–5].
AIMS
To investigate this topic and add evidence, we conducted a cohort study aimed to assess the relationship between PA intensity and semen quality – measured by sperm concentration, total and progressive motility, and proportion of normal morphology cells - in a population of healthy young men living in Brescia, North Italy.
MATERIALS AND METHODS
A prospective study within the FAST randomized trial was conducted between April 2018 and June 2019 [6]. Semen quality parameters were assessed at the enrollment (baseline) and again after 4 and 8 months. Each semen sample was collected in a sterile container via masturbation after a period of sexual abstinence lasting at least 3 days and no more than 5 days. Samples were delivered to the laboratory within 30–40 minutes of collection, and a portion of each sample (<50 μl) was immediately processed for semen analysis (spermiogram). Additionally, a 20 ml blood sample was collected from each participant under fasting conditions.
PA was assessed at baseline and after 4 and 8 months using the International Physical Activity Questionnaire (IPAQ), which evaluates various types of activity - including walking, moderate-intensity, and vigorous-intensity activities - and estimates the total energy expenditure expressed in Metabolic Equivalent of Tasks (METs).
Due to the longitudinal nature of the data, a linear mixed model with robust variance estimation was used to assess the association between PA with sperm concentration. A generalized linear mixed model with a Poisson distribution was applied to evaluate total, progressive motility and normal morphology cell counts, using the total number of cells as an offset.
Restricted spline regression models were fitted to model the potential nonlinear shape of the associations between total PA and semen parameters.
RESULTS
A total of 143 young healthy men (median age 20 years, IQR 19-21 years) participated in the study. The majority were engaged in moderate (45%) or high (43%) recreational PA, with a median expenditure of 1,960 (95% confidence interval, 1,055–3,182) Metabolic Equivalent of Tasks in min/wk.
The main results are presented in Table 1. An increase in total sperm motility (IRR 1.11, 95% CI, 1.05-1.17) and normal morphology (IRR 1.18, 95% CI, 1.03-1.35) was observed among participants engaged in moderate PA. Conversely, an inverse association was observed for walking and vigorous-intensity PA. No association was observed between PA and sperm concentration.
An inverse U-shape relationship was identified, with the highest values of total sperm motility and normal morphology occurring at intermediate levels of PA. No statistically significant trend was found for sperm concentration, although a U-shaped relationship association was suggested by the restricted cubic spline model.
CONCLUSIONS
Our findings are consistent with several studies previously conducted on healthy young men from the general population, as well as on male partners of infertile couples, which have shown that individuals engaging in moderate-to-high levels of physical activity tend to exhibit better semen quality compared to those with sedentary lifestyles or very high levels of activity. The results support current recommendations to engage in moderate physical activity to promote overall health, including improvement in semen quality [7]. Future research should investigate the mediating role of DNA methylation in the relationship between physical activity and semen quality
Outdoor Particulate Matter, Cardiovascular Health and Incidence of Coronary or Ischemic Stroke Events: A Population-Based Study
No full textIntroduction
Long-term exposure to outdoor particulate matter of size<2.5 microns (PM2.5) has been associated with increased risk of cardiovascular diseases (CVD) incidence [1], although with substantial between-study heterogeneity. In addition, pathways and factors of individual susceptibility to PM2.5 have been poorly characterized so far. The American Heart Association (AHA) proposed the Life’s Simple 7 (LS7) metric of cardiovascular health by combining lifestyle (not smoking, healthy diet, engaging in sufficient physical activity) and health (normal body weight, maintaining low values of blood pressure, total cholesterol and glucose) components [2]. The mediating role of LS7, and whether the risk excess due to PM2.5 is exacerbated in individuals with a “poor” cardiovascular health, remain to be ascertained. Aims
To investigate the association between long-term exposure to PM2.5 and the incidence of first coronary heart disease or ischemic stroke event, and the contribution of LS7 on explaining the risk excess, in a population-based cohort in Northern Italy. Furthermore, to assess the role of LS7 as an effect modifier for the association.
Methods
Study populationThe RoCAV study [3] is a population-based cohort of n=3777 50+ years old residents in the city of Varese (Lombardy region, northern Italy) at the time of recruitment (2013-2016; 64% participation rate). For the aims of these analyses, we retained individuals free of coronary heart disease and stroke at baseline (n=3313, 62% men). Cardiovascular healthParticipants underwent a comprehensive baseline examination, assessing cigarette smoking (MONICA questionnaire), dietary intake (EPIC food frequency questionnaire), habitual physical activity (Baecke questionnaire), as well as clinical and laboratory parameters (fasting blood lipids and glucose, blood pressure and anthropometric measures). From these, we calculated the LS7 metric on a 0-to-14-points scale [2], further categorized as poor (0-4 points), intermediate (5-9 points) and ideal (10-14 points) [4]. We categorized the two LS7 components of lifestyle (LS7-ls) and health (LS7-h) in a similar way. PM exposureMonthly concentrations for PM2.5 over the period 2000-2019 were retrieved from the EXPANSE project models [5], at a spatial resolution of 25mt. Individuals’ concentrations were attributed from spatial linkage at the residential address at baseline, geo-referenced; the exposure metric was the average concentration in the 12 months before the month of baseline visit.
Study endpointsIndividuals were followed-up through record linkage with Electronic Health Records (hospital discharge and mortality) provided by the Local Health Agency. We selected discharge codes suggestive of myocardial infarction and unstable angina (ICD-IX codes: 410-411) or elective coronary revascularization; cerebrovascular infarction (ICD-IX 433 or 434) or endarterectomy with stenosis at cerebral or pre-cerebral arteries (ICD-IX 433.1, 433.3, 434.0). Selected records were then reviewed to identify and retain the first index case. Fatal cases were identified from underlying causes of deaths suggestive of coronary deaths (ICD-X codes: I21-I25) or ischemic stroke (I63). Censorship occurred at the date of death from other causes or emigration outside the study Region, as ascertained by contacting the municipality of residency. The study endpoint is the occurrence of first coronary heart disease or ischemic stroke, fatal or non-fatal, before Dec 31st, 2022. Statistical analyses
Due to the low number of events in women, the analyses were carried out on men and women combined, and on men only. We first estimated the rate ratios for the LS7 categories (“ideal” as reference) from Poisson regression models, adjusting for age and sex. Then, we estimated the Hazard Ratios (HR) and 95%CI for 1 interquartile range (1.93 µg/m3) increase in PM2.5 using nested Cox models with attained age during follow-up on the time scale, and adjusting for: sex and education (Model1), and further for LS7 (Model2). We computed the percent explained by LS7 on the log scale as 100*[log(HRM2)-log(HRM1)]/[log(HRM1)]. Finally, we investigated individual susceptibility by adding to Model2 a LS7*PM2.5 interaction and by reporting the corresponding p-value from a Wald chi-square test (2df). These analyses were repeated for LS7-ls and LS7-h components. ResultsMean±SD PM2.5 concentrations were 18.6±1.8 µg/m3. During 7.1 years of median follow-up time, we observed n=196 CVD events for a rate of 8.6 per 1,000 person-years (men: n=150 events, 10.8 per 1,000 py). “Poor” vs. “ideal” LS7 resulted in a 2.34-fold (95%CI: 1.33-4.10) increased event rate (men: 2.13, 1.12-4.06). In Model1, PM2.5 was associated with increased CVD risk, in the overall sample (HR=1.11, 95%CI: 0.95-1.29) and in men (HR=1.19, 95%CI: 1.00-1.43). Further adjustment for LS7 explained 13.4% (men and women) and 7.2% (men) of the risk excess, respectively, mainly due to the LS7-ls component. We found evidence of interaction between PM2.5 and the LS7-ls component (Wald test p-values: 0.03 [men and women] and 0.007 [men]). In men, the HRs for PM2.5 in the “poor” (40% of the sample) and in the “ideal” (15% of the sample) LS7-ls categories were 1.49 (95%CI: 1.11-1.98) and 0.67 (0.45-1.00), respectively. Conversely, no interaction was observed for LS7, nor for the LS7-h component. ConclusionsIn our cohort, representative of contemporary outdoor PM2.5 levels in a North Italian population, long-term exposure to PM2.5 was associated with an increased risk of first major CVD event, especially in men. AHA LS7 explained only a small proportion of the association. Adults engaging poor lifestyles are more susceptible to the detrimental effect of PM, and can benefit the most from policies reducing pollution levels.
 
A Random Forest Algorithm For Identifying Risk Factors For Multimorbidity In The UK Biobank Cohort
No full textIntroduction: High-income countries are undergoing significant demographic shifts, characterized by population decline and progressive aging. These transformations are associated with an increase in the prevalence of chronic diseases, which often coexist, worsening individuals’ quality of life and increasing healthcare costs. Identifying the factors that contribute to the onset of multimorbidity is particularly complex, as these factors often interact with each other and cause multiple effects across different diseases.
Objectives: This study aimed to identify the main risk factors for multimorbidity within a large UK cohort using a fully nonparametric ensemble method. This approach makes no assumptions about the underlying relationships between variables and allow managing high-dimensional data while preventing overfitting.
Methods: We analyzed data from the UK Biobank cohort, which includes detailed information on socioeconomic status, lifestyle, anthropometric measures, and environmental exposures collected at recruitment, along with disease occurrence obtained through linkage with hospital admissions (primary and secondary diagnoses), death records, and cancer registries. Multimorbidity was defined as the presence of at least two chronic conditions from a list developed through an international consensus using a modified Delphi method [1]. To assess the role of 18 candidate variables in predicting the onset of multimorbidity over a five-year follow-up, we applied a random forest algorithm adapted for survival analysis within a competing risk framework [2], considering two competing events: the development of multimorbidity and death prior to its onset. The candidate variables included: white British/Irish ethnicity (Yes/No), qualification level, average total household income before tax (adjusted for household size and categorized into quintiles), area-level index of multiple deprivation (deciles), body mass index (kg/m2), waist circumference (cm), pack-years of smoking, alcohol drinking (g/day), healthy diet score (ranging from 0 to 5, based on the intake of fruit, vegetables, fish, whole grains, processed and red meat), walking (at least 10 min, number of times a week), moderate physical activity (at least 10 min, number of times a week), vigorous physical activity (at least 10 min, number of times a week), particulate matter air pollution 2.5 (PM2.5) (µg/m3), PM2.5-10 (µg/m3), PM10 (µg/m3), NO2 (µg/m3), average exposure to evening (7:00 pm – 11:00 pm) or night noise (11:00 pm – 7:00 am) (dB). Results were summarised using out-of-bag partial dependence plots and variable importance (VIMP) metrics.
Results: Of the 422,344 individuals included in the cohort, aged between 39 and 73 years, we selected 137,565 participants who were free from the conditions included in the definition of multimorbidity at the time of recruitment and for whom risk factor information was available. During the five-year follow-up, 4384 individuals developed multimorbidity (2740 males, 1644 females). The five-year cumulative incidence was 3.9% in males and 2.6% in females. Among individuals who developed multimorbidity during follow-up, the main conditions observed were cancer (52.4% of males and 52.1% of females), arrhythmias (44.7% of males and 28.5% of females) and coronary artery disease (42.1% of males and 24.8% of females). Based on VIMP metrics, the strongest predictors in men were smoking, waist circumference, and sleep duration; in women alcohol, smoking, and waist circumference. Five-year cumulative incidence was higher for heavy smokers (sex-specific 95th percentile of pack-years) (males: 6.3%, females: 4.0%) compared to non-smokers (males: 3.5%, females: 2.4%); for individuals with elevated waist circumference (sex-specific 95th percentile) (males: 6.1%, females: 5.2%) versus those with median values (males: 3.9%, females 2.6%); for heavy alcohol drinkers (sex-specific 95th percentile) (males: 4.6%, females: 4.0%) versus median intake (males: 3.8%, females: 2.4% ); for those sleeping 4 hours/day (males: 6.3%, females: 4.2%) or 10 hours/day (males: 6.5%, females: 4.5%) versus 7 hours/day (males: 3.7%, females: 2.5%). Diet, physical activity, and air pollution had smaller impacts.
Conclusions: Preventive interventions targeting smoking, abdominal obesity, and heavy alcohol consumption among middle-aged adults in the UK and likely in other high-income countries, may substantially reduce the incidence of multimorbidity. Such interventions could improve the health trajectory and burden of disease of future older populations. In addition, promoting adequate sleep duration appears to be beneficial and should be integrated into public health recommendations