A systematic review of the use of the Satiety Quotient

Abstract

The satiating efficiency of food has been increasingly quantified using the Satiety Quotient (SQ). The SQ integrates both the energy content of food ingested during a meal and the associated change in appetite sensations. This systematic review examines the available evidence regarding its methodological use and clinical utility. A literature search was conducted in six databases considering studies from 1900 to April 2020 that used SQ in adults, adolescents and children. All study designs were included. From the initial 495 references found, fifty-two were included. Of the studies included, thirty-three were acute studies (twenty-nine in adults and four in adolescents) and nineteen were longitudinal studies in adults. A high methodological heterogeneity in the application of the SQ was observed between studies. Five main utilisations of the SQ were identified: its association with (i) energy intake; (ii) anthropometric variables; (iii) energy expenditure/physical activity; (iv) sleep quality and quantity and (v) to classify individuals by their satiety responsiveness (i.e. low and high satiety phenotypes). Altogether, the studies suggest the SQ as an interesting clinical tool regarding the satiety responsiveness to a meal and its changes in responses to weight loss in adults. The SQ might be a reliable clinical indicator in adults when it comes to both obesity prevention and treatment. There is a need for more standardised use of the SQ in addition to further studies to investigate its validity in different contexts and populations, especially among children and adolescents.

According to the WHO, 39 % of adults were overweight and 13 % had obesity in 2016⁽¹⁾ with paediatric data being just as concerning with 340 million children from 5 to 19 years old classified with overweight and obesity worldwide⁽¹⁾. This alarming prevalence of overweight, obesity and their associated metabolic complications call for a better understanding of the mechanisms involved to propose innovative and effective weight loss strategies. Among them, the regulation of energy balance^(2,3) and the pathways involved in the control of appetite and energy intake (EI) have been of particular interest in recent years⁽⁴⁾. Both homoeostatic and hedonic mechanisms influence the motivation to eat (hunger), meal size (satiation) and post-meal suppression of hunger (satiety)⁽⁵⁾.

Indeed, a number of objective and subjective methods have been developed for the quantification and evaluation of both food intake (e.g. ad libitum test meals, food diaries) and appetite sensations (e.g. visual analogue scale; VAS). These VAS usually comprise questions pertaining to hunger ‘How hungry do you feel?’, fullness ‘How full do you feel?’, desire to eat ‘How strong is your desire to eat?’ and prospective food consumption ‘How much do you think you could eat?’, with ‘not at all’ to ‘extremely’ as labelled end points. Integrating both the energy content of food ingested during a meal and the associated change in appetite sensations, Green et al. developed a Satiety Quotient (SQ) as an indicator of the satiating efficiency of food⁽⁶⁾. The SQ is calculated by dividing the change in subjective appetite sensations in response to a meal by the energy content of the meal.

Since its development, there has been an increasing use of the SQ. While initially created as an indicator for the satiating efficiency of a meal or food, the SQ has been associated with food intake^(7–10) and body weight (BW) and composition^(9,11,12) or used as a tool to classify individuals by their satiety responsiveness^(13–15). However, the extent to which the SQ has been applied in research and its scientific and clinical relevance has yet to be examined. Therefore, the aim of this systematic review is to review the available evidence of the different contexts in which the SQ has been utilised in research, the methodologies used to calculate the SQ and to examine its clinical utility.

Methods

This review is registered in the PROSPERO database as CRD42019136442. The PRISMA guidelines were followed for the preparation of this paper⁽¹⁶⁾.

Database search

The following electronic bibliographic databases were searched: PubMed, Embase, Scopus, Web of Science, CAB Abstract Core Collection and Google Scholar. The literature search considered studies from the year 1900 to April 2020. Keyword searches were performed for ‘Satiation’, ‘Satiety response’, ‘Appetite’, ‘Hunger’, ‘Humans’, ‘Fullness’, ‘Prospective Food Consumption’, ‘Desire To Eat’, ‘Motivation To Eat’ and ‘Satiety Quotient’. The search strategy for each of the databases is detailed in Table 1. The search strategies were developed based on an analysis of the literature and were open-ended according to the nature of each database. The reference lists of the articles included were also examined to complete the search.

Table 1. Database search strategy details

Mp, title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word.

Study eligibility

Inclusion criteria

To be included in the review, studies had to use SQ. There was no exclusion criterion for the study design (cross-sectional, observational, longitudinal or interventional), population (no limit for age, weight status and associated complications and both sexes were included) and meal type (standardised or ad libitum). Published peer-reviewed studies, conference proceedings and posters (when data and design properly described), theses and dissertations were eligible.

Exclusion criteria

When data were presented in a graphical form without mean or standard deviation indicated, the corresponding author of the work was contacted to obtain complementary data. If the corresponding author did not answer or declined the query, studies were excluded. When the full text was not found and the corresponding author was unreachable or did not respond, the article was excluded.

Study selection

Titles and abstracts of potentially relevant studies were screened in duplicate for inclusion in the review and any discrepancies were collectively discussed by the authors. The same procedure was followed for the full texts. Any disagreement regarding eligibility for inclusion was discussed and a consensus made among co-authors.

Data extraction

For every included study, the following data were extracted: sample size and characteristics (sex, age, BMI), study design and aim, VAS characteristics (specific appetite sensations assessed and timing), meal characteristics, SQ equation and main SQ results.

Risk of bias

Risk of bias was independently evaluated by two authors (AF, DT) using the Cochrane risk of bias tool⁽¹⁷⁾. Risk of bias was assessed for: selection bias, performance bias, detection bias, attrition bias and reporting bias. Any discrepancies in bias coding were resolved by a third reviewer. Studies were not excluded on the basis of risk of bias.

Results

The flow diagram presented in Fig. 1 illustrates the selection/inclusion/exclusion process. The initial database search identified 1281 studies and nine additional studies were also identified. Following the removal of duplicate studies, 495 studies were identified. After review of titles and abstracts, 162 studies were excluded and eighty-five full-text were screened, leaving fifty-two included studies. Table 2 details the risk of bias analysis. Of the fifty-two studies included, thirty-three were acute studies^{(6–8,11,13–15,18–42)} and nineteen were longitudinal studies^{(9,10,12,43–59)} .

Fig. 1. Flow chart.

Table 2. Risk of bias

L, low risk; M, medium risk; H, high risk; NR, not reported.

Acute studies

Of the thirty-three acute studies, twenty-nine were conducted in adults^{(6–8,11,13–15,18–37,40,43)} and four in adolescents^{(38,39,41,42)} .

Adult acute studies (n 29)

Main aim, population and design

The main aims, populations and used designs are presented in Table 3 and fully detailed in the online Supplementary material.

Table 3. Population, design, methods and main results of adult acute studies*

VAS, visual analogue scale; SQ, Satiety Quotient; NR, not reported; AS, appetite sensation; H, hunger; BF, breakfast; EI, energy intake; T2D, type 2 diabetes; BW, body weight; F, fullness; DTE, desire to eat; PFC, prospective food consumption; S, satiety; RS, resistant starch; LSP, low satiety phenotype; T1D, type 1 diabetes; AREE, activity-related energy expenditure; SJL, social jetlag; LIE, low-intensity exercise; HIE, high-intensity exercise.

* Only protocol is detailed that is relevant to SQ.

† To convert energy values from kcal to kJ, multiply by 4·184.

Methods

Topics

Of the twenty-nine studies, 90 % (n 26) compared SQ in response to a stimulus (meal, exercise and sleep), the remaining studies^(8,13,14) used SQ to categorise their population (high or low satiety phenotype). Fifty-nine percent of the included studies (n 17) compared the SQ response to meals of different composition. Of these seventeen studies, two used liquid meals^(28,33) , fourteen solid meals^{(6,14,15,18,19,21,22,25,27,30,34–37,40)} and one study compared solid v liquid meals⁽³²⁾. Of these studies, three examined the effect of meals differing in energy content^(14,28,33) and five studies compared the effect of meals differing in macronutrient composition^{(6,15,18,19,25)} . Martini et al. ⁽²⁷⁾ compared the effect of meals differing in fibre and protein, and Au-Yeung et al. ⁽³⁰⁾ compared the effect of different amounts of protein intake via konjac glucomannan capsules and one study examined the combined effects of a modification in macronutrients, unsaturated fats, fibre and Ca⁽⁴⁰⁾. In a slightly different way, Felix et al. ⁽³²⁾ compared the effect of different kinds of rice and Finlayson et al. ⁽³⁵⁾ the effect of different tastes on appetite sensations. Defries et al. ⁽²²⁾ compared the different satiating effects of meals made from buckwheat flour or rice flour, while Felix et al. ⁽³⁶⁾ compared the different satiating effects of white rice or brown rice using four different types of rice and Kendall et al. ⁽³⁴⁾ the effect of different resistant starch compositions using beverages. Finally, in their study, Bligh et al. ⁽²¹⁾ investigated the satiating effect of two different types of Paleolithic meals compared with a reference meal.

Three of the studies investigated the influence of sleep on SQ^(20,29,31) : one examined the effect of sleep duration⁽²⁰⁾, while another examined the timing⁽³¹⁾ and a last one assessed the influence of the duration, quality and timing of sleep⁽²⁹⁾. Two of the twenty-eight studies investigated acute medication interventions^(23,26) and one assessed the effect of hormone infusions⁽²⁴⁾. Among the acute studies, two included acute exercise in their protocol and compared appetite sensations after the same exercise performed at different blood glucose levels⁽⁷⁾ and the other compared different intensities of exercise⁽³⁷⁾ or different activity related energy expenditure⁽⁴³⁾. One study investigated the effect of mental work⁽¹¹⁾ and another compared the appetite sensation response of men and women⁽⁸⁾. Finally, Drapeau et al. ⁽¹³⁾ characterised the biopsychobehavioural profiles of men with low satiety phenotype at the start of a weight loss intervention.

Visual analogue scale

Regarding the type of VAS used, 79 % (n 23) of acute studies used the pen and paper method^{(6–8,11,13,14,20,22,24,26–37,40,43)} , 10 % (n 3) used electronic VAS^(18,21,23) and three studies did not specify the type of scale used^(15,19,25) . Of the twenty-three studies using pen and paper scales, fifteen used 100 mm scales^{(6,14,20,22,24,26,27,30–36,43)} , while eight used 150 mm scales^{(7,8,11,13,28,29,37,40)} . For studies that used electronic VAS, one used 100 mm scales⁽¹⁸⁾, one used 60 mm scales⁽²¹⁾ and one did not specify the length of the scale used⁽²³⁾. The three studies that did not specify the type of scale used also did not specify the length of the scale^(15,19,25) .

Out of the twenty-nine studies, twenty-eight assessed ‘Hunger’^{(6–8,11,13–15,18–26,28–37,40,43)} , twenty-four measured ‘Fullness’^{(7,8,11,13,14,18,20–31,33,34,36,37,40,43)} and twenty investigated ‘Prospective Food Consumption’^{(7,8,11,13–15,18,20,22,24,28–31,33,34,36,37,40,43)} . ‘Desire to Eat’ was assessed in twenty studies^{(7,8,11,13–15,18,21–23,27–31,34,36,37,40,43)} and ‘Satiety’ in four studies^{(18,20,24,27)} . However, as described below, all appetite sensations measured were not used for the calculation of SQ.

Calculation of Satiety Quotient

Equations used

Of the twenty-nine acute studies included, eight used the initial equation proposed by Green et al. ^{(6,22,24,30,33–35,43)} : (appetite sensation pre-meal – appetite sensation post-meal)/EI of eating episode. This equation was slightly reworked by Drapeau et al. ⁽¹⁰⁾, who used this equation but multiplied the result by 100. Fifteen studies used the equation proposed by Drapeau et al. ^{(7,8,13,14,18–20,25,28,29,31,32,36,37,40)} . While previous studies have used similar equations, others have calculated the SQ slightly differently. Chapman et al. ⁽²⁶⁾ calculated two SQ: a prandial SQ that considered in its calculation both pre- and post-meal appetite sensations, and a post-prandial SQ only considering post-meal sensations. In their study, Martini et al. ⁽²⁷⁾ calculated three different SQ: (1) the same equation as Drapeau et al. using the pre- and post-lunch appetite sensations and energy content of lunch; (2) (appetite sensation before lunch – appetite sensation before snack)/energy content of lunch × 100 and (3) (appetite sensation before lunch – appetite sensation after snack)/(energy content of lunch + snack) × 100. More specifically, Au-Yeung et al. ⁽³⁰⁾ used the Green equation for SQ_H, SQ_DTE and SQ_PFC. For SQ_F, they subtracted fullness post-eating from fullness fasting. Salama et al. ⁽¹¹⁾ also reversed the order of subtraction between appetite sensations contrary to what was done by Drapeau, subtracting pre-meal sensations from post-meal sensations. Two studies did not specify the type of equation used^(15,21) . Finally, Thomas et al. ⁽²³⁾ used an adapted version of the equation proposed by Green and calculated ‘satiation quotient’ per quartile, reflecting the satiety capacity of a food as eaten ((quartile initial hunger – quartile ending hunger rating)/energy consumed during quartile).

Appetite sensations used

Although we have previously detailed the different appetite sensations assessed in the included studies, SQ was not calculated in each of these studies using all the assessed sensations. Twenty-five studies calculated an SQ for ‘Hunger’^{(6–8,11,13,14,19–26,28–32,34–37,40,43)} , sixteen for ‘Fullness’^{(7,8,11,13,20,21,24,27–29,31,34,36,37,40,43)} and fifteen for ‘Desire To Eat’^{(7,8,11,13,21,27–31,34,36,37,40,43)} and ‘Prospective Food Consumption’^{(7,8,11,13,20,24,28–31,34,36,37,40,43)} . Drapeau et al. ⁽¹³⁾ also calculated a mean SQ with the SQ results corresponding to the four previous appetite sensations. In three of the acute studies, an SQ for ‘Satiety’ was calculated^(20,24,27) . Hansen et al. ⁽¹⁸⁾ calculated what they named an Appetite Quotient (similar to SQ), based on composite appetite scores (with Hunger, Satiety, Fullness, Prospective Food Consumption and Desire To Eat). Gonzalez et al. ⁽³³⁾ also produced a composite SQ, whose equation is, however, not detailed. In their work, Hollingworth et al. ⁽¹⁵⁾ did not detail in the publication which appetite sensation was used to calculate the SQ.

Timing of the sensations used

For the SQ calculation, out of the twenty-nine studies, twenty-three chose to define as ‘pre-meal sensations’, the sensations recorded immediately before the tested meal^{(7,8,11,13,14,18–20,22,25,27–34,36,37,40,43)} . The remaining six studies assessed pre-lunch sensations 1 h before the meal⁽²⁶⁾, 20 min before the meal⁽²¹⁾ or 5 min before the meal⁽²⁴⁾. Three studies did not specify the timing of the VAS^(15,23,35) . Two studies also assessed appetite feelings during the meal^(23,24) . Regarding the use of post-meal appetite sensations for calculating SQ, eight studies evaluated them up to 60 min after the end of food intake^{(7,8,13,23,28,29,33,37)} , five studies up to 120 min after the end of food intake^{(20,27,32,34,36)} , 4 up to 180 min after the end of food intake^{(18,22,25,31)} and 3 up to 240 min after the end of food intake^(6,11,40) . Hopkins et al. ⁽¹⁹⁾ reported appetite sensations every hour after the end of the meal until the next meal, while Chapman et al. ⁽²⁶⁾ assessed appetite sensations up to 5 h after the end of the meal. Green et al. ⁽⁶⁾ measured appetite sensations up to 75 min after food intake, Schmidt et al. ⁽²⁴⁾ reported post-meal appetite sensations up to 25 min after the meal and finally, Harrington et al. ⁽⁴³⁾ reported post-meal appetite sensations immediately after the end of the meal. The study from Bligh et al. ⁽²¹⁾ reported appetite sensations up to 175 min after the start of food intake, while Dalton et al. ⁽¹⁴⁾ reported these sensations up to 90 min after the start of the meal. The timing of VAS is summarised in detail in Table 3.

Type of meal

Finally, SQ was also calculated in response to different meals. Among the included acute studies, thirteen used a standardised fixed meal to calculate SQ^{(7,8,13,21,22,28–30,32–34,36,37)} , while three used an individualised meal based on percentage of energy needs^(14,31,35) and six used an ad libitum meal^{(20,23–26,43)} . Six studies calculated the SQ on both types of meals: standardised and ad libitum ^{(6,11,18,19,27,40)} . One study did not specify the type of meal used to calculate the SQ⁽¹⁵⁾. Table 3 details the different meals used in the included studies.

Acute studies conducted in children and adolescents

Main aim, population and design

The main aims, populations and used designs are presented in Table 4 and fully detailed in the Supplementary materials.

Table 4. Data detailed for children and adolescents acute studies

VAS, visual analogue scale; SQ, Satiety Quotient; BF, breakfast; H, hunger; EI, energy intake; S, satiety; DTE, desire to eat; F, fullness; PFC, prospective food consumption; LED, low energy density; HED, high energy density.

* To convert energy values from kcal to kJ, multiply by 4·184.

Methods

Calculation of Satiety Quotient

Three of the included studies used pen and paper VAS^(38,39,42) , and Kral et al. did not specify the type of scale used⁽⁴¹⁾. In their studies, Thivel et al. and Fillon et al. used 150 mm scales^(39,42) and Albert et al. and Kral et al. used 100 mm scales^(38,41) . Albert et al. ⁽³⁸⁾ assessed ‘Desire To Eat ’, ‘Hunger’, ‘Fullness’, ‘Anticipated Food Consumption’, ‘Desire for specific food types’, ‘Palatability’, ‘Appreciation’ and ‘Visual appeal’. The others assessed ‘Desire To Eat ’, ‘Hunger’, ‘Fullness’ and ‘Prospective Food Consumption’^(39,41,42) .

Regarding the calculation of SQ, all of the included studies used the equation proposed by Drapeau et al. ⁽¹⁰⁾ (appetite sensation pre-meal – appetite sensation post-meal)/EI of eating episode × 100. While Albert et al. only used the immediate post-meal sensation in the equation⁽³⁸⁾, the three other studies used a mean of post-meal sensations assessed: immediately post-meal, 30 min and 60 min post-meal in Thivel et al. and Fillon et al.’s studies^(39,42) and immediately post-meal and 15 min post-meal in Kral et al. ⁽⁴¹⁾.

Although Albert et al. ⁽³⁸⁾ assessed different appetite sensations, they only calculated the SQ_H, while the three other studies calculated the SQ for each of the appetite sensations assessed: Desire To Eat, Hunger, Fullness and Satiety^(39,41,42) . All studies calculated their SQ using an ad libitum lunch meal.

Chronic studies conducted in adults

Main aim, population and design

The main aims, populations and used designs are presented in Table 4 and fully detailed in the Supplementary materials.

Methods

Topics

Eighty-four percentage of the included chronic studies investigated the SQ in response to lifestyle changes (e.g. changing from inactive to active) or physiological modifications (e.g. pre- v. post-menopause in women)^{(9,10,44–52,54–57,59)} , while three of these nineteen studies used SQ as a tool to classify the population as low and high satiety phenotype^(12,53,58) .

Two observational studies were included and examined the association between SQ and the change of EI, BW and body composition over time^(9,10) .

Among the included interventional studies, seven assessed the effect of different dietary prescriptions on SQ^{(12,44–46,55,58,59)} , while two assessed the effect of different physical activity prescriptions on SQ^(50,57) . One study investigated the effect of a prescription combining physical activity and dietary interventions on SQ⁽⁴⁷⁾. One assessed the effect of weight change on SQ⁽⁴⁸⁾ and three others more specifically on the effect of different energy restrictions on SQ change^(53,54,59) . Bédard and colleagues investigated the effect of sex on SQ⁽⁴⁹⁾ and Carbonneau et al. ⁽⁵²⁾ the effect of different nutritional labelling. Finally, the effect of probiotic⁽⁵¹⁾ or pharmaceutical⁽⁵⁶⁾ compounds on the change of SQ was also tested.

Visual analogue scale

Fifteen studies used pen and paper VAS^{(9,10,12,45–49,51–54,56,58,59)} , while the other four used electronic VAS. Of the fifteen that used the pen and paper method, six used 100 mm scales^{(45,46,54,56,58,59)} , while the others used 150 mm scales^{(9,10,12,47–49,51–53)} . With regard to electronic VAS, one study used a seven-point scale⁽⁴⁴⁾, another used a scale ranging from –3 to 3⁽⁵⁵⁾ and finally, two studies did not specify the length of the scales used^(50,57) .

Sixteen of the nineteen studies analysed ‘Hunger’^{(9,10,12,45–54,56,57,59)} and fifteen assessed ‘Fullness’^{(9,10,12,47–54,56–59)} . Thirteen studies investigated ‘Desire To Eat’^{(9,10,12,47–51,53,54,56,57,59)} and twelve assessed ‘Prospective Food Consumption’^{(9,10,12,47–51,53,54,56,59)} . Two studies used a single scale with ‘Hunger’ and ‘Fullness’ as extremes^(44,55) .

Calculation of Satiety Quotient

Equations used

Seventy-four percentage of the included studies used the following equation proposed by Drapeau et al. ^(10,13) : (appetite sensation pre-meal – appetite sensation post-meal)/EI of eating episode × 100^{(9,10,12,45,46,48–54,57,59)} . Buckland et al. ⁽⁵⁸⁾ used the same equation, but they subtracted post-meal sensation from pre-meal sensation, because they evaluated just ‘Fullness’. Hintze et al. ⁽⁵⁴⁾ reversed also the order of subtraction between appetite sensations contrary to what was done by Drapeau, subtracting pre-meal sensations from post-meal sensations, for SQ_F. Three studies used the same equation without multiplying the result by 100^(44,47,56) , and one study did not clearly specify the equation used⁽⁵⁵⁾.

Appetite sensations used

On the nineteen chronic studies, fifteen calculated SQ_H^{(9,10,12,45–48,50–54,56,57,59)} , fourteen SQ_F^{(9,10,12,44,47–49,51–56,58)} and nine SQ_DTE^{(9,10,12,47,48,51,53,54,56)} and SQ_PFC^{(9,10,12,47,48,51,53,54,56)} (see Table 5).

Table 5. Population, design, methods and main results of adult chronic studies*

VAS, visual analogue scale; SQ, Satiety Quotient; BW, body weight; BF, breakfast; H, hunger; F, fullness; DTE, desire to eat; PFC, prospective food consumption; EI, energy intake; T2D, type 2 diabetes; NR, not reported; S, satiety; RS, resistant starch; LSP, low satiety phenotype; HSP, high satiety phenotype; CER, continuous energy restriction; IER, intermittent energy restriction.

* Only protocol is detailed that is relevant to SQ.

† To convert energy values from kcal to kJ, multiply by 4·184.

Timing of the sensations used

More specifically, all studies considered as ‘pre-meal appetite sensation’ the sensations given immediately before the meal. With regard to ‘post-meal appetite sensation’, five studies used only the sensations immediately after the meal^{(45,47–49,52)} , and two studies considered the post-meal sensations as the sensations recorded 30 min after the start of ingestion^(44,55) . Others averaged appetite sensations immediately after eating with appetite sensations 1 h after eating⁽⁵⁷⁾, or every 10 min for 1 h^(10,51,53) , or every 10 min for 1 h plus 90 min and 120 min after eating⁽¹²⁾. Three studies used the average appetite sensation immediately after eating with the sensations reported every 30 min for 3 h^(9,54,59) , while Halford et al. ⁽⁵⁶⁾ and Buckland et al. ⁽⁵⁸⁾ used the same protocol but with appetite sensation evaluations every hour for 3 h and not every 30 min. Finally, Golloso-Gubat et al. ⁽⁴⁶⁾ used the average of appetite sensation at 15, 30, 45, 60, 90, 120, 150, 180 and 240 min after the meal to calculate ‘post-meal appetite sensation’. One study⁽⁵⁰⁾ indicated that it had integrated in the calculation of the post-meal sensations the sensations of appetite immediately after the meal as well as sensations assessed periodically between the two meals (Table 5).

Type of meal

Out of the nineteen included studies, seventeen calculated the SQ in response to a standardised fixed meal^{(9,10,12,46,48,51,53)} , while five used an ad libitum meal^{(44,45,47,52,55)} with one study using both type of meals⁽⁵⁶⁾. Six studies^{(49,50,54,57–59)} calculated the SQ on an individualised meal based on a percentage of energy needs.

Main results

By adopting a systematic overview of all the included studies, a large heterogeneity is observed when it comes to the purpose of using SQ. While all details are presented in Tables 3–5, five main methodological uses of the SQ can be identified: (i) the association between SQ and EI^{(7–9,12,15,18,19,21,22,25,27,32,36,40,44–46,49,54,55,58,59)} ; (ii) the association between the SQ and anthropometric variables^{(8–11,47,48,53,59)} ; (iii) the association between SQ and energy expenditure/physical activity^{(7,14,37,43,50,57)} ; (iv) the association between SQ and sleep quality and quantity^(20,29,31) and (v) SQ to classify individuals into low and high satiety phenotypes^{(13–15,40,53,58)} .

The following sections present and categorise the main results observed in the included studies. While only the main methodological aspects and results related to the use of the SQ are detailed in this section, the Tables 3–5 present the full details of the included studies.

Association between Satiety Quotient and energy and macronutrient intake

First, four of the included studies demonstrate that SQ is a predictor of food intake^(7–10). The systematic analysis of these studies shows that SQ_F^(8–10), SQ_H⁽⁷⁾, SQ_PFC⁽⁹⁾ and mean SQ⁽⁹⁾ predict EI and SQ_F predicts relative EI too (subtracting RMR from total EI)⁽⁸⁾. A distinction is made in the studies between objectively measured EI and self-reported EI using food diaries, with SQ_DTE, SQ_H, SQ_F⁽⁷⁾ and SQ_PFC⁽⁹⁾ predicting reported EI only. More specifically, according to these studies, macronutrient intake could be predicted by SQ_F, SQ_PFC and mean SQ⁽⁹⁾, and SQ_F could also predict CHO intake in food diaries⁽⁹⁾. In children, Kral et al. ⁽⁴¹⁾ suggest that energy density may influence satiety responsiveness and that SQ may predict IE.

Association between Satiety Quotient and anthropometric variables

Five of the included studies show associations between the SQ and anthropometric or body composition variables^{(8,9,11,53,58,59)} . In regard to BW, we observe that individuals with high satiety phenotype lost more BW than those with a low satiety phenotype^(12,53,58) , and we find the same conclusions regarding waist circumference in women with obesity⁽⁵⁸⁾. In fact, individuals with a high waist circumference had lower satiating effect determined by the SQ_F⁽¹¹⁾, and McNeil et al. ⁽⁹⁾ showed in their 5-year study that changes in SQ were negatively correlated with the change in waist circumference. With regard to the relationship between SQ and fat mass, Salama et al. ⁽¹¹⁾ found a positive relationship between % fat mass and SQ_F. In their longitudinal study, McNeil et al. ⁽⁹⁾ found a positive correlation between the SQ and fat mass changes (delta) over the entire study, although they found a negative correlation between year 4 and year 5.

Association between Satiety Quotient and energy expenditure/physical activity

Three of the included studies show contradictory associations between SQ and exercise or the level of physical activity^{(25,43,50,57)} . Some cross-sectional results suggest a decrease in SQ, indicating a lower satiety responsiveness, in lean individuals with high activity-related energy expenditure⁽⁴³⁾, while others show no effect of habitual physical activity level on SQ in non-obese individuals⁽²⁵⁾. In individuals with overweight and obesity, a 12-week exercise intervention led to increased satiety responsiveness to a fixed meal^(50,57) .

With regard to studies in children, it can be observed that the timing between exercise and a meal^(37,43) or the use of an energy replacement strategy⁽⁹⁾ has no effect on SQ and that no particular association was found with SQ. However, a better satiety responsiveness (higher SQ) was observed when exercise is performed just before a meal v. a rest condition⁽⁴³⁾.

Satiety Quotient to classify individuals into low and high satiety phenotypes

Six of the included studies support the SQ as a reliable tool to phenotype individuals based on their satiety responsiveness^{(12–15,53,58)} . Indeed, compared with individuals with a high satiety phenotype, individuals with a low satiety phenotype have higher EI, greater cravings for sweet foods, lower craving control, higher disinhibition and fasting Hunger, Desire To Eat and Prospective Food Consumption and exhibit a higher wanting for high-fat food^(14,15,58) . The behavioural and psychological characteristics of the low satiety phenotype are associated with a greater susceptibility to overconsumption^(14,15) . These results are also corroborated by another study, where Drapeau et al. indicate that the higher increase in cognitive restraint and a lower decrease in disinhibition in response to a weight loss intervention could increase the susceptibility of these individuals to weight gain⁽⁵³⁾, these results being in agreement with another work from Drapeau et al. ⁽¹³⁾ showing that SQ negatively correlated with the external locus for Hunger measured by the Three-Factor Eating Questionnaire. Moreover, Buckland et al. ⁽⁵⁸⁾ found a weaker control over eating and weight loss programme adherence in people with a low satiety phenotype, as well as a lower weight loss compared with people with a high satiety phenotypes.

Discussion

While there has been a growing use of the SQ in clinical studies since its development by Green and colleagues in 1997⁽⁶⁾, little attention has been paid regarding its use since then and a high methodological heterogeneity can be observed between studies. A better understanding of the SQ and its clinical implication is of particular interest since, as shown by several studies, by including both pre-meal sensation and the energy content of the meal in its calculation, it seems to provide different information than appetite sensations alone. Indeed, some studies have observed different results for appetite sensations and SQ in response to various stimuli (such as exercise or sleep for instance)^(31,37) . In that context, the present review aimed to systematically analyse the available evidence regarding the scientific and clinical use of the SQ. Fifty-two studies were included after our database search, thirty-three of them being cross-sectional/acute^{(6–8,11,13–15,18–42)} and nineteen being longitudinal^{(9,10,12,43–59)} . The large majority of the included studies enrolled adults participants with only four enrolling children and adolescents^{(38,39,41,42)} .

According to our analysis, acute studies mainly used the SQ to compare the satiating effect of different kinds of meals varying in texture (liquid and solid)^{(6,14,15,18,19,21,22,25,27,28,30,32–36,40)} , energy content^{(14,28,33,41)} or composition^{(6,15,18,19,21,25,27,30,34,36,40)} . Some of these acute investigations also assessed the effect of sleep characteristics (i.e. timing, quality or duration)^(20,29,31) , exercise^(7,37) , mental work⁽¹¹⁾, sex⁽⁸⁾ or pharmaceuticals^(23,24,26) on the SQ. Regarding the interventional studies included in our analysis, they mainly used the SQ to evaluate the effect of different dietary and/or exercise interventions^{(12,44–47,50,51,53–55,57,59)} on the SQ. Finally, some studies (acute and chronic) used the SQ to classify individuals as low or high satiety phenotypes^{(13–15,40,53,58)} .

Clinical utility and reliability of the Satiety Quotient

According to the present systematic approach, the use of the SQ might be a reliable predictor of both measured^(7–10,58) and reported^(7,9,10) EI, as well as macronutrient intake⁽⁹⁾. Studies effectively highlight higher food consumption with lower satiety responsiveness to a meal (lower SQ) in type 1 diabetes⁽⁷⁾, healthy women⁽¹⁵⁾, men and women with overweight⁽⁸⁾, premenopausal women⁽⁹⁾ and women with obesity^(54,58) . This is reinforced by other results demonstrating negative associations between SQ and BW, waist circumference as well as fat mass^(9,11,53,58) . Importantly, Drapeau et al. ⁽⁵³⁾ found a positive association between SQ and weight loss in response to an energy restriction intervention in men and women with obesity, like Buckland et al. ⁽⁵⁸⁾ in women with obesity. The SQ has been used as a clinical tool to categorise people depending on their level of satiety responsiveness to a standardised fixed meal; a low phenotype characterising people who report difficulties in appropriately recognising their appetite sensations before or after a meal⁽⁸⁾. These results are supplemented by those of Buckland et al. ⁽⁵⁸⁾, which have shown that people with low satiety phenotype have a weaker control over eating and weight loss programme adherence compared with people with high satiety phenotype. Moreover, people with low satiety phenotype prefer and consume more of high energy density food than people with high satiety phenotype⁽⁵⁸⁾. While most studies use a median split to categorise low and high satiety phenotypes, in a clinical context, a low satiety phenotype might be observed in about 10 % of patients with obesity who declare themselves as unable to detect changes in their appetite, report a weak satiety response to a meal and even show an increase in appetite after a meal for some of them⁽⁶⁰⁾. Altogether, these results suggest that the SQ is an interesting clinical indicator to identify adults at risk of overeating and thus could be used in preventive strategies and weight loss interventions. Moreover, while the literature seems to suggest the SQ and the SQ phenotype as complementary tools to already existing subjective methods (such as the evaluation of disinhibition using the Three-Factor Eating Questionnaire); providing additional information regarding the risk of overeating for instance; comparison studies are still missing and should be conducted.

Interestingly, while the SQ has been studied in the context of nutritional manipulations, some studies also examined its relationship and response to physical activity and exercise. According to these studies, moderate physical activity levels in lean individuals and exercise training in individuals with overweight and obesity are associated with a higher SQ, suggesting an improved satiety responsiveness^(43,50,57) . However, this was not the case in studies measuring SQ at an ad libitum meal in lean individuals with very high physical activity levels, one of which showing lower SQ⁽⁴³⁾ and another showing similar SQ⁽²⁵⁾ than their less active counterparts. Using a different methodology to assess the satiety response to food (preload-test meal protocol), other studies have shown that physically active individuals have better ability to adjust subsequent EI following preloads differing in energy content^(61,62) . These results, whether using the SQ or energy compensation following a preload as an indicator of satiety responsiveness, illustrate a relationship between physical activity, food intake and appetite control⁽⁶³⁾. Here again, it suggests the clinical interest of the SQ as part of multidisciplinary approaches developed to prevent and treat obesity in adults.

According to our systematic approach, only few (n 4 out of 52) studies very recently used the SQ among children and adolescents. Three of them investigated the effect of acute exercise on the subsequent satiating effect of a meal^(38,39,42) and the last, the effect of different preload energy density on satiety responsiveness. While two of these studies did not observe any effect of an acute exercise bout on the SQ calculated on the following ad libitum meal^(38,42) , Fillon et al. ⁽³⁹⁾ found increased SQ for Hunger, Prospective Food Consumption and Desire To Eat after acute moderate intensity exercise in adolescents with obesity. Kral et al. ⁽⁴¹⁾and coworkers suggested a beneficial effect of a low energy density preload on satiety responsiveness in children. In addition to the lack of available evidence regarding the use of the SQ in youth, the absence of any validation study in his population must be highlighted. Indeed, it remains unknown whether the SQ is a clinically valid and reliable tool to be used in children and adolescents. Based on the increasing interest in the appetite control of children and adolescents, particularly in those with obesity, our research group recently conducted a methodological study assessing the reproducibility of SQ and its validity as an indicator of body corpulence and composition as well as of EI in adolescents with obesity⁽⁶⁴⁾. Although SQ_H showed a relatively modest reproducibility, none of the other SQ variables were found reproducible, and no association was found with anthropometric variables, body composition or EI⁽⁶⁴⁾. This clearly calls for caution when interpreting existing results and for further studies developing reliable tools to measure the satiating effect of food in this population.

Methodological considerations

Our systematic analysis reveals a high level of heterogeneity regarding the methods used (equation used, type of meal, timing of the measurements of appetite sensations, etc.). While the SQ has been suggested as reliable and reproducible in adults, especially men with obesity (ICC for the SQ mean of 0·67)^(13,14) , more studies are needed to assess its validity and reproducibility in various contexts and populations.

While forty-three out of the forty-eight adults studies included^{(6–14,18–20,22–37,40,43–54,56–59)} used the equation initially developed by Green et al. ⁽⁶⁾, others used derived equations^{(11,23,26,27)} or did not specify the equation used^(15,21,55) . Similarly, as detailed in the tables and results section, the VAS used (e.g. 100 v. 150 mm) and the timing of the measurements of appetite sensations, with some studies only using the post-meal appetite sensation while others using the mean of the appetite sensations for up to several hours post-meal, vary between studies making any comparisons difficult. Since appetite sensations are dynamic and postprandial effects might be detected and integrated by individuals at different post-meal intervals, it would be of interest to better examine the best postprandial timing to use when calculating SQ. Importantly, while the SQ has been validated under standardised conditions and mainly using a fixed meal^(8,14) , 37·5 % (n 18) of the included studies used an ad libitum meal to calculate the SQ^{(6,11,18–20,23–27,30,40,43,44,47,52,55,56)} . Gonzalez et al. examined the accuracy of the SQ depending on the energy content of the ingested meal and observed a better reproducibility and reliability of SQ (mean SQ as well as SQ_H, SQ_F, SQ_PFC, SQ_S) in response to higher energy content compared with meals of lower energy content⁽³³⁾. Finally, while the validity of the SQ among men⁽¹³⁾ and women⁽¹⁴⁾ was suggested, it has been widely used among specific populations such as individuals with diabetes^(7,26) , premenopausal women^(9,28) , people with different levels of physical activity⁽²⁵⁾, people with overweight and obesity^{(8,10,12,13,19,24,26,29,45,47,57,59)} and shows a highly variable degree of correlations between studies (as detailed in Tables 3 and 5). Once more, this must lead us to interpret these results with caution and calls for more methodological validations.

Conclusion

While the current systematic review suggests the reliability of the SQ in adults and encourages its use as an interesting clinical tool regarding the satiety responsiveness to a meal and its changes in response to weight loss, we also encourage the adoption of a more standardised use of the SQ as well as the development of additional studies assessing its validity in several contexts and populations, especially among children and adolescents. Further studies should also be conducted to identify the potential biological markers associated with this SQ. Based on the present systematic analysis, we encourage future studies to assess SQ for Hunger, Fullness, Desire To Eat and Prospective Food Consumption after an overnight fast in response to a standardised fixed meal, without intense physical activity, and to consistently use a validated equation (such as the one initially proposed by Drapeau et al. ^(10,13) ). This would allow for more reliable outcomes and better comparisons across studies.