Association of Baltic Sea and Mediterranean diets with frailty phenotype in older women, Kuopio OSTPRE-FPS study

Abstract

Purpose: To evaluate the association between Baltic Sea diet (BSD) and Mediterranean diet (MED) with frailty.

Methods: This was a secondary analysis on the osteoporosis risk factor and prevention-fracture prevention study on 440 women aged 65-72 years. Frailty was ascertained with the presence of 3-5 and prefrailty 1-2 of the following criteria: weight loss ≥ 5%, low life satisfaction score, walking speed ≤ 0.51 m/s, handgrip strength divided by body mass index ≤ 0.67 kg/m² and physical activity ≤ 2 h/week. Women answered to questionnaires on lifestyle factors and 3-day food record. BSD score was ascertained using intake of nine and MED score of eight foods or nutrients components from food record. Multinomial logistic regression models adjusted for age, energy intake, smoking, living status, marital status and intervention group evaluated associations between MED and BSD with frailty phenotype status.

Results: At 3-year follow-up, 206 women (46.8%) were prefrail and 36 (8.2%) were frail. After adjusting for confounders, a tendency was found between BSD per standard deviation (SD)-unit increase and lower likelihood of frailty (β = 0.62, 95% CI = 0.38-1.01, P = 0.057). Further, MED per SD-unit increase was associated with lower likelihood of prefrailty (β = 0.74, 95% CI = 0.6-0.9, P = 0.009). Consumption of vegetables was lower in frail (31.5 ± 36.0 g/day) and prefrail women (37.1 ± 42.0 g/day) than in non-frail women (48.6 ± 40.7 g/day) (P for trend = 0.041).

Conclusions: Positive behavioral characteristics such as following MED and BSD may be associated with lower likelihood of prefrailty and frailty in older women. However, further longitudinal analyses are warranted.

Keywords: Baltic Sea diet; Frailty; Mediterranean diet; Older women.

Fig. 1

Likelihood of prefrailty and frailty for Baltic Sea diet tertiles (tertile 1 is referent = 1.0). Odds ratios and 95% confidence interval (Error bars represent the CI) were calculated using multinomial logistic regression analysis. Model was adjusted for age (years), energy intake (kJ/day), smoking status (never, current and former smokers), living status (living alone, live with another person and live in retirement home), marital status (unmarried, cohabiting, married, divorced, widow) and calcium and vitamin D supplementation intervention group. The reference category was referent (not frail). Unadjusted P for trend = 0.042. Adjusted P for trend = 0.066

Fig. 2

Likelihood of prefrailty and frailty for Mediterranean diet tertiles (tertile 1 is referent = 1.0). Odd ratios and 95% confidence interval (Error bars represent the CI) were calculated using multinomial logistic regression analysis. Model was adjusted for age (years), energy intake (kJ/d), smoking status (never, current and former smokers), living status (living alone, live with another person and live in retirement home), marital status (unmarried, cohabiting, married, divorced, widow) and calcium and vitamin D supplementation intervention group. The reference category was referent (not frail). Unadjusted P for trend = 0.005. Adjusted P for trend = 0.004