The study included a total of 167 older females recruited by simple random sampling from different regions of Mauritius. Four age groups were used to categorize the participants: 60 to 69, 70 to 79, 80 to 89 and 90 to 100. Samples from various ethnic groups present in Mauritius were taken into consideration, including Indo-Mauritians and African Mauritians. Prior to the start of the project, ethical approval was granted by the Departmental Research Ethics Committee (Ref number: DHSDREC/01/2022) dated 28 ^th January 2022 at the University of Mauritius. All participants were fully informed of the project and a written consent was obtained from participants for the use and publication of collected data. Participants were assured that all of their responses would be handled with absolute confidentiality and that the survey was completely anonymous, with no mention of participant names or addresses.

The data needed to achieve the stated objectives was collected using a self-designed questionnaire. The survey was conducted in-person since it was expected that respondents could clearly articulate themselves and that many elderly people may have low levels of knowledge. Moreover, simple Creole, the most common language in Mauritius, was used to interview the participants to make sure that the questions were properly understood. Enough time was allowed for each participant to accurately answer each question. The questionnaire, which was composed primarily of closed-ended questions, consisted of five sections: I, II, III, IV and V. The questionnaire has been deposited in Figshare, a public repository. ¹⁶

Section I: Eating habits and food frequency

The food frequency section was designed by using questions from the Short Form Food Frequency Questionnaire ¹⁷ that was created by the Leeds University Nutritional Epidemiology department. It included approximately 59 items divided into the following categories: carbohydrates (grains and tubers), beans/pulses, meat and eggs, seafoods, soybeans and products, dairy products, vegetables, high-fat and high-sugar snacks, and beverages. ¹⁸ Participants could choose from seven frequency categories for each food item and scores as shown in Table 1.

Section II: Physical activity

Questions from a short version of International Physical Activity Questionnaire (IPAQ) were used to assess physical activity. ¹⁹ PAL was calculated in metabolic equivalents (MET), minute per week, and classified as low, moderate, or high intensity. The scores were as follows: low (<600 MET), moderate (600–1500 MET), and high (>1500) PAL. Table 2 shows the criteria for classification of PAL. ¹⁹

Section III: Nutrition knowledge assessment

Ten nutrition-related questions from a general nutrition knowledge questionnaire for adults were utilised and modified. ²⁰ The original questionnaire's food items were modified to reflect the Mauritian environment. The questions selected were mostly about calorie, sugar, fat and protein content. An overall score of 25 was achievable with higher scores implying participants were highly knowledgeable. A score of 1 was assigned for a right answer and a score of 0 for a wrong answer and "don't know". ¹⁴

Section IV: Anthropometric measurements

Physical measurements, including weight and height were obtained using the same tools to reduce the possibility of errors. A measuring tape and ruler was used to measure the height and the participant stood bare foot with the head horizontal against the wall. A portable bathroom scale was used to measure weight. The BMI of the participants was then determined using the formula, weight (in kilograms) divided by height (squared in meters); and were categorised as normal, underweight, overweight or obese according to WHO classification. ²¹

Section V: Personal and sociodemographic data

Age, residential area, education level, marital status, ethnic group and type of pension were all included in the sociodemographic data. Vegetarianism and eating difficulties were also assessed.

Section VI: A three-day 24-hour dietary recall

Nutritional guidelines from EPIC-Norfolk were used to create a 24-hour dietary recall form. ¹⁴ Food consumption was recorded for two weekdays and one weekend over three weeks. Typically, the 24-hour recall was carried out in the sequence of consumption starting from morning to night. This form requires a list of all meals and drinks consumed the day prior, including snacks. The recall was not conducted if a person was following a special fasting regimen or had previously attended any event where people have a tendency to overeat. On the first day of the recall, the participants were shown a variety of readily accessible household utensils in order to help them accurately report the portion sizes of their meals. The utensils used were teaspoons, tablespoons, cups and glasses. Moreover, the participants were asked to recall eating and drinking incidents by time period in order to avoid concerns with inaccurate or incomplete reporting. Three dietary recalls were used to calculate the subjects' energy intake using a food composition table mostly based on the Tanzania food composition table ²² and also with the help of the Nutritionist Pro Version 6.1, a licensed software which is accessible at the University of Mauritius laboratory (An open-access equivalent that performs a similar function is https://www.wikihow.com/). The mean energy intake was then calculated for each participant to be used for statistical analyses.

The data of this study was analysed using the software SPSS (Statistical Package for the Social Sciences, version 16.0). Percentages were used to express descriptive tests. ANOVA test, independent sample t-test, Pearson correlation and the Chi-squared test were used to show the relationship between the variables. Statistical significance was defined as P ˂0.05.

Table 3 displays the socio-demographic details of the 167 respondents by age group, area of residence, education level, marital status and ethnic group. Of the 167 respondents, 47.3% were between the ages 60 and 69, 39.5% were between the ages 70 and 79, 12.6% were between the ages 80 and 89 and only 0.6% were between the ages 90 and 100. The majority of the participants were from rural areas, which accounted for 52.7%.

The mean energy intake of the 167 respondents was 2362.87 kcal. The mean BMI was 25.86 kg/m ². According to the results obtained, BMI had a significant impact on energy intake demonstrating that the mean energy intake for overweight and obese participants is higher than that of normal and underweight participants (p ˂0.001). The energy intake of participants aged 70–79 and 80–89 was found to be higher, and this finding was statistically significant (p = 0.027). The mean energy intake was found to be lower in vegetarians, according to an independent sample t-test; however, this relationship was not statistically significant (p = 0.407). Moreover, eating difficulties had no significant impact on energy intake (p = 0.377). The study also indicated that participants who engage in low physical activity had a higher energy intake than those who engage in moderate or high physical activity. And, there was a statistically significant relationship between PAL and energy intake (p ˂0.001). Furthermore, results showed a moderate negative and significant relationship between NK scores and energy intake (p ˂0.001). It was found that the mean energy intake of participants residing in rural areas was higher; however, this relationship was not statistically significant (p = 0.074).

In general, vegetables were consumed most (M = 6.87 ± 0.50) followed by tea/coffee (M = 6.53 ± 077), butter/margarine (M = 5.50 ± 1.28), fruits (M = 5.37 ± 1.07), white rice (M = 5.25 ± 1.62), beans/pulses (M = 5.08 ± 0.71), oatmeal (M = 4.87 ± 1.65) and semi-skimmed milk (M = 1.48 ± 1.48), low fat cheese (M = 1.48 ± 1.23), brown bread (M = 1.55 ± 1.22) being the food items less frequently consumed.

Relationship between BMI and frequency of different food items consumed

The results in Table 4 show a moderate significant negative correlation, a small significant positive correlation and a moderate positive correlation between BMI and the food items listed in the table.

Relationship between NK and consumption frequency of different food items

There is a statistically significant relationship (p ˂0.001) between nutrition knowledge and education level. The results of the Pearson correlation showed that there was a small significant negative correlation at 0.01 level between NK and consumption frequency of sweet pastries/cakes, deep fried snacks and full cream yoghurt (p ˂0.001). At the 0.05 level of significance, there was a small significant negative correlation between NK and frequency of sweet biscuits/sweets (p = 0.016), chocolates (p = 0.033), butter (p = 0.048), soft drinks (p = 0.011), pasta (p = 0.040) and chicken (p = 0.035) consumption. Furthermore, a moderate significant positive correlation at the 0.01 level was obtained between NK and frequency of brown bread, brown farata, skimmed milk and fruit consumption (p ˂0.001). Another significant relationship at the 0.01 level was found between BMI and frequency of brown rice (p ˂0.001) and low-fat yoghurt (p = 0.007) consumption. This correlation was small and positive.

Effect of PAL on frequency of different food items consumed

There was a statistically significant difference between PAL and brown bread [F (2, 164) = 4.61, p = 0.011], brown rice [F (2, 164) = 3.86, p = 0.023], white rice [F (2, 164) = 3.87, p = 0.023], white bread [F (2, 164) = 4.83, p = 0.009], fish [F (2, 164) = 3.21, p = 0.043], full cream yoghurt [F (2, 164) = 5.30, p = 0.006] and soft drinks [F (2, 164) = 4.75, p = 0.010] consumption. Table 5 also demonstrates that participants who had low PAL ate a higher amount of white rice and white bread. On the other hand, those with high PAL consumed fish and white rice in greater quantities.

Relationship between residence area and consumption frequency of food items

The results of the ISTT revealed a statistically significant difference between participants living in urban and rural areas and consumption frequency of brown bread, brown rice, white rice, oatmeal, Weetabix, pasta, fish, low fat yoghurt and fruit. Table 6 also demonstrates that brown bread, oatmeal, beans/pulses, vegetables and fruit were highly consumed food items by elderly female participants living in urban areas.

Relationship between energy intake and consumption frequency of different food items

The results obtained show a moderate positive and statistically significant correlation between energy intake and white rice, chicken, full cream milk, sweet pastries, chocolates and soft drinks (p ˂0.001, significant at the 0.01 level).

A small significant positive correlation was found between energy intake and •

White bread (p ˂0.001, significant at 0.01 level)

Moreover, a moderate significant negative correlation was obtained between energy intake and brown bread, skimmed milk and fruit (p ˂0.001, significant at 0.01 level). And a small negative and significant relationship was found between energy intake and brown rice (p = 0.001, significant at 0.01 level), soybeans and products (p = 0.011, significant at 0.05 level) and low-fat yoghurt (p = 0.023, significant at 0.05 level).

A negative relationship was found between oatmeal, Weetabix, fish and energy intake (EI) while a positive relationship was observed between potato, beans/pulses, tea/coffee and EI. But these relationships were not statistically significant.

The mean EI of the 167 female elderly participants in this study was 2363 kcal/day; and this was not consistent with any recommended dietary allowances of other countries as it was much higher than that of other countries like Indonesia, Malaysia, Philippines and Thailand, which recommend 1850, 1780, 1620–1410 and 1550 kcal respectively for its female senior citizens. ²³ In the southwestern United States, elderly females reported a mean EI of 1555 ± 63.2 kcal/day, which did not support the results here. ²⁴ The increase in EI of the female senior citizens of this study may be due to significant changes in diet composition such as an increased use of cooking oils, increased consumption of carbohydrates and high-energy processed foods including cakes, instant noodles and cookies. ²⁵ However, the results of this study do coincide to a little extent with one study conducted in Mauritius in 2007 whereby the energy intake was higher (1978 kcal) when compared to dietary reference intakes for older adults. This increase in EI would offer a reasonable explanation for why being overweight was more common in the elderly female participants. ²⁶

A statistically significant relationship was obtained between age and EI in this study. The participants aged between 70 to79 and 80 to 89 had the highest mean EI (2492.83 ± 703.54, 2561.24 ± 740.26 kcal respectively) while those aged 60–69 had a mean EI of 2210.56 ± 631.68 kcal. In contrast to the results found here, which showed that the age group 60–89 had higher EI, a study in the southern United States revealed that females in the 51–70 age range had a lower mean EI value of 1571 ± 29.3 kcal. ²⁴

The mean BMI of the 167 participants was 25.86 kg/m ², which suggests that they were mostly overweight. Additionally, results indicate that as EI increases, BMI increases; the underweight individuals had the lowest EI (1299.57 ± 62.06 kcal) and obese participants had the highest EI (3326.55 ± 281.39 kcal). Therefore, this study shows that EI had a significant impact on BMI. This result supports a Chinese women's study, which found that overweight and obese women consumed a significantly higher proportion of their dietary energy from carbohydrates than lean women, indicating a link between EI and BMI. ²⁷ Similar results were observed in Perth, Western Australia, where EI in elderly women was linked to a higher BMI; but they were under-reporting their EI. ²⁸ The results of the present study has been further supported by Trichopoulou et al., (2000) ²⁹ who revealed that energy intake and BMI have a positive relationship and the BMI of females was found to be much higher than males.

This study demonstrates that the PAL of elderly females in Mauritius was low for the majority of the participants (52.10%) and there was a statistically significant relationship between PAL and energy intake. The study also showed that participants who engaged in low physical activity had a higher energy intake than those who engaged in moderate or high physical activity ( Table 5). Another study, on the other hand, reported conflicting results, arguing that there is no sufficient evidence to corroborate that physical activity has an effect on EI. ³⁰ Moreover, a systematic review showed that there was no consistent evidence to support the recommendation that physical activity may reduce the decrease in appetite and energy intake that occurs with age. ¹⁵

The results showed that there was a statistically significant relationship between NK and education level. This study demonstrates that the female elderly participants having a lower education level had least nutrition knowledge. This finding was supported by a study which found that the majority of Taiwanese female elderly participants who had no formal education and attended only primary school had poor NK. ³¹ Furthermore, a study in England discovered that those with degrees had a much higher NK than participants without formal education. ³²

The results show a moderate negative and significant correlation between NK and energy intake, indicating that EI decreases with increasing NK. A study showed that older women having a higher education level had the biggest relationship with a higher label reading score ³³ and this led to a decline in EI among them. ¹⁴ Another study demonstrated that women with a higher level of knowledge were more likely to be concerned about calories and consumed less sweets, soft drinks and salty snacks, ³⁴ implying that cutting back on these unhealthy meals will not result in an increase in energy intake.

In the current study, some food items were found to be consumed more often among female older people aged between 60 to 100. For instance, the results indicated that vegetables, tea/coffee, butter/margarine, fruit, white rice, beans/pulses, oatmeal, white bread, potato and chicken were more frequently consumed. On the other hand, food items like semi-skimmed milk, low-fat cheese, brown rice, chocolates, soft drinks, skimmed milk and oily fish were less frequently consumed. According to a previous study done in Mauritius, it was found that the older individuals consumed a lot of rice, vegetables, and little pasta because rice is one of the country's staple foods. ²⁶ The result can be further supported by a study who reported that elderly people consumed a lot of carbohydrates. ³⁵

Relationship between BMI and frequency of different food items consumed

The findings of the current study revealed that the majority of participants were classified as normal or overweight. There were also a considerable number of obese respondents (12%). This could be explained by the correlation between aging and a redistribution of both body fat and fat-free mass. With aging, loss of skeletal muscle occurs and this could be a reason why there is a bigger relative rise in intra-abdominal fat than subcutaneous or total body fat, and a relatively greater decline in peripheral fat than central fat-free mass. ³⁶ A study revealed that BMI may be influenced by the type of carbohydrates consumed. ³⁷ In the present study, there was a negative and significant correlation between consumption frequency of whole-grain carbohydrates such as brown bread, brown rice, and brown farata/chapatti and BMI. This indicates that as consumption of wholegrain carbohydrates increases, BMI decreases. Results from meta-regression on cross-sectional data support these findings, which show a negative relationship between wholegrain consumption and BMI. ³⁸ In this study, there was also a negative correlation between BMI and oatmeal and Weetabix intake but it was not statistically significant. Moreover, a positive and significant relationship was found between BMI and consumption frequency of refined carbohydrates such as white bread, white rice and white farata/chapatti in this study. This indicates that as consumption of refined carbohydrates increases, BMI increases. These results corroborate those of a Mediterranean cohort which found that white bread consumption (≥ two servings per day) had a strong relationship with the risk of becoming overweight or obese. ³⁹ A positive relationship was observed between chicken, full cream yoghurt, full cream milk, sweet pastries/cakes, sweet biscuit/sweets and chocolates intake and BMI, indicating that high consumption of these food items leads to an increase in BMI; between BMI and consumption frequency of fish, low fat yoghurt, skimmed milk and fruit, a significant negative relationship was noted. To support these findings, one study reported that low-fat dairy products were inversely connected to changes in body weight, whereas whole fat dairy products were linked to a little but substantial rise in body weight; the highest category of dairy consumption was also linked to a lower chance of being overweight and having abdominal obesity. ⁴⁰ Regarding beverage consumption (soft drinks and tea/coffee), the positive correlation between BMI reported in this study corresponds with a study on Mauritian women whereby higher intake of soft drinks resulted in high BMI. ³ Another study demonstrated that soft drink consumption was linked to an increased risk of abdominal obesity over a 10-year period. ⁴¹ However, in one study, consuming one or more portions of tea per day was linked to obesity, but the link was not statistically significant. ⁴²

Relationship between nutrition knowledge and consumption frequency of different food items

In the current study, a significant negative relationship was found between NK and consumption frequency of sweet pastries/cakes, deep fried snacks, full cream yoghurt, sweet biscuits/sweets, chocolates, butter, soft drinks, pasta and chicken. This shows that some subjects with low NK consumed high amount of these food items and vice-versa. Moreover, it was also observed that NK and frequency of brown bread, brown farata, skimmed milk, fruit, brown rice and low-fat yoghurt consumption had a positive and significant correlation. This indicates that subjects with high NK consumed more of these food items, which were healthier. A study reported that an older population with less education consistently had a higher likelihood of eating a diet high in sweets and low in protein-rich vegetables. ⁴³ However, no significant relationship was discovered between the other foods and NK. Although older persons have high nutrition knowledge, their eating habits may not be consistent with NK due to their inability to put their technical nutrition knowledge to use. ⁴⁴

Relationship between residence area and consumption frequency of food items

The results of this study revealed a statistically significant difference between participants living in urban and rural areas and consumption frequency of brown bread, brown rice, white rice, oatmeal, Weetabix, pasta, fish, low fat yoghurt and fruit. Table 6 demonstrates that brown bread, oatmeal, beans/pulses, vegetables and fruits were the highly consumed food items by the elderly female participants living in urban areas while in rural areas white rice, oatmeal, beans/pulses, vegetables and fruit were highly consumed. It can be seen that in both urban and rural areas consumption of oatmeal, beans/pulses, vegetables and fruit were high. However, the other food items showed no significant difference. Few studies have been conducted to study the relationship between residence area and frequency of food items consumed.

Relationship between energy intake and consumption frequency of different food items

Intake of energy was positively correlated with consumption of white bread, white rice, pasta, chicken, full cream yoghurt, full cream milk, sweet pastries, sweet biscuits, chocolates and soft drinks, as shown in Table 4. This indicates that consumption of these food items increases energy intake. The main function of carbohydrates is to supply the body's cells with energy and dietary fibre. ⁴⁵ However, data from the USDA's food consumption surveys from 1989 to 1991 and 1994 to 1996 showed that a higher intake of carbohydrates was the main factor contributing to increased energy intake. ⁴⁶ This corroborates the result of the current study showing that high intake of white bread, white rice and pasta increased energy intake. Chicken was found to increase energy intake among the participants. Certain parts of the chicken, such as the breasts, are frequently sold with or without the skin. Typically, thighs, wings, and drumsticks are sold with the skin still attached. Skin increases the amount of fat and calories in the chicken piece. The cooking methods for chicken including deep frying, cooking with extra fat, sugar, or salt in marinades and pan frying in oils or butter can all contribute to an increase in calories and fat. ⁴⁷ Therefore, this explains why consumption of chicken led to an increase in energy intake. This study also discovered that frequent consumption of sweet pastries, sweet biscuits and chocolates increased energy intake and this was consistent with a study conducted among Norwegian adults which classified these food items as snacks and they were among the top five in an energy-contributing food group. ⁴⁸ A study concluded that energy-dense snacks may increase energy intake and weight when consumed in higher amounts. ⁴⁹ Soft drinks frequently include sweeteners like high-fructose corn syrup, that can greatly increase EI and cause obesity ⁵⁰ and this is consistent with the current study. Moreover, a meta-analysis discovered definite correlations between soft drink use and higher energy intake and body weight. There was persistent evidence that consumers did not reduce their intake of other foods to compensate for the extra energy they get from soft drinks, leading to higher total energy intake. ⁵¹