*3.5. Results*

The mean age of the total sample was 74.8 years (95% CI = 74.49–75.12) ranging from 65 to 104, with 54.8% females, 54.7% cohabiting with someone, and 46.3% reporting low levels of education (Table 1). Some 37.2% reported being smokers or ex-smokers, 33.4% presented obesity, and 33.2% had low levels of physical activity. A total of 56.2% reported having two or more chronic conditions. Fruit and vegetable servings per day ranged from 0 to 18, with a mean of four servings (95% CI = 3.82–4.19). Participants with lower consumption (equal to or less than three servings per day) were more likely to be men, smokers, not frequently engaging in physical activity, frequent drinkers, and have a low educational level.

The minimum number of days of survival was 19 and the maximum 2688, with a mean of 2323.94 days (*SD* = 553.7). We observed 322 confirmed deceased cases (132 women and 190 men). The Kaplan–Meier estimated curves (Figure 1) showed that the level of fruit and vegetable consumption had a significant negative e ffect on survival. In the adjusted Cox proportional hazards regression model, only the interaction term *fruit and vegetable consumption\* number of chronic conditions* was found to be significant (*p* = 0.035). Table 2 presents the unadjusted and adjusted HRs and 95%CI. In the unadjusted model, both medium (HR = 1.68, 95%CI = 1.3–2.18) and high fruit (HR = 1.12, 95%CI = 1.1–1.14) consumption were significantly associated with higher risk of death. Other significant predictors of higher risk of mortality were being male, low levels of physical activity (compared with high), lower levels of education, being a smoker (current or past), and having three or more chronic conditions. The adjusted model including the interaction term between fruit and vegetable consumption and number of chronic conditions is presented in Table 2. In order to interpret the e ffect of fruit and vegetable consumption on time to death in the presence of interaction, HRs were calculated according to the number of chronic conditions (Table 3). The proportion of deceased people was significantly higher (*p* < 0.001) among those who had three or more CCs (*n* = 112, 25.8%) compared to those with none or one CC (*n* = 127, 16.3%) or two CCs (*n* = 83, 15.5%). Subjects who consumed five or more servings of fruit and vegetable per day and had two chronic conditions were at 27% less risk of mortality (HR = 0.38, 95%CI = 0.21–0.69, *p* = 0.002) compared with participants consuming three or fewer servings, while other covariates held constant. However, fruit and vegetable consumption had no impact on time to death among subjects who reported none or one CC, or three or more.

**Table 1.** Baseline characteristics of the sample and comparison between levels of fruit and vegetable consumption.


Note: CC = Chronic conditions; PA = Physical activity; 95% CI = 95% Confidence interval; Unweighted frequencies, weighted proportions and means.

**Figure 1.** Kaplan–Meier estimated curves for cumulative survival by levels of fruit and vegetable consumption.


**Table 2.** Hazard ratios, Confidence intervals and *p* values for the unadjusted and adjusted Cox proportional hazards models (*n* = 1699).

*Note:* HR = Hazard ratio; 95% CI = 95% Confidence interval; CC = Chronic conditions; PA = Physical activity. In bold, significant effect. a The adjusted model included all the variables and the interaction term simultaneously.

**Table 3.** Adjusted hazard ratios of the effect of fruit and vegetable consumption by number of chronic conditions on all-cause mortality. Cox regression model (*n* = 1,699).


Note: HR = Hazard ratio; 95%CI = 95% Confidence interval; CC = Chronic conditions. In bold, significant HR. Model included the *fruit & vegetable consumption\*number CC* interaction term and the following variables: age, gender, educational level, cohabiting, ever smoked, obesity, alcohol consumption, and level of physical activity (see Table 2).

Comparison between "medium" and "high" levels of fruit and vegetable consumption; (1) in the none or one CC group: *p* = 0.66; (2) in the two CC: *p* = 0.13; (3) and in the three or more CC: *p* = 0.596

Survival curves as a function of levels of fruit and vegetable consumption and number of chronic conditions are displayed in Figure 2 for a specific pattern of covariates (i.e., females, not cohabiting, never smoked, no education/primary school studies, non-obese, lifetime abstainers, high level of physical activity, and age 74.8 years old). For people with two chronic conditions, the probability of surviving until the end of the study was significantly greater if they consumed five or more servings per day of fruit and vegetables, compared with those who ate three or fewer.

**Figure 2.** Survival function according to the level of fruit and vegetable consumption and number of chronic conditions (*n* = 1699). The first subfigure (left) shows the survival curves according to the three levels of fruit and vegetable intake for people with none or one chronic condition, the second (middle) for those with two chronic conditions and the third (right) shows survival curves associated with fruit and vegetable consumption for respondents with three or more chronic conditions. Note: Survival functions calculated from the adjusted Cox proportional hazards model presented in Table 2. All covariates were set equal to zero.
