Research Department of Human Nutrition and Center for Advanced Food Studies, The Royal Veterinary and Agricultural University, Frederiksberg, Denmark

Correspondence to: B Buemann, Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, Rolighedsvej 30, 1958 Frederiksberg, Denmark. E-mail: BBU@KVL.DK

BACKGROUND: Alcoholic beverage drinking may increase total energy intake at a meal by various mechanisms and this effect may depend on the sort of beverage.

OBJECTIVE: To test the effect of wine, beer and a soft drink served with a normal meal on food and total energy intake in non-obese men.

DESIGN: A supper meal consisting of three consecutive dishes was presented to 22 young men. Ad libitum energy intakes (EI) of the meal were measured at three different occasions in a cross-over design with red wine, lager beer or a carbonated soft drink. This was done in two studies with different design. In the first study the beverages were supplied ad libitum and in a second study the intake of the beverages was fixed: beer and soft drink at 9 ml/kg body weight and wine isoalcoholic to beer, 3.185 ml/kg body weight.

RESULTS: In the ad libitum beverage study total EI was higher with wine than with the soft drink and beer (P<0.05). In the fixed beverage study differences in total EI did not reach statistical significance (P=0.14), although the intake of goulash was higher with wine and beer than with the soft drink (P<0.005).

CONCLUSION: These data indicate that alcoholic beverages, and wine in particular, may enhance total EI at a meal relative to a soft drink, when served with no restriction.

International Journal of Obesity (2002) 26, 1367-1372. doi:10.1038/sj.ijo.0802069

alcohol; appetite; beer; compensation; energy intake; food intake; wine

Introduction

Average annual per capita consumption of alcohol in countries such as France and Germany is about 12 l according to retailers statistics.¹ This corresponds to more than 800 kJ per day. It is therefore not surprising that much interest has been focused on the impact of alcohol on energy balance. However, the impact of the consumption of different beverages on energy balance and body weight is far from clear. Observational studies relating adiposity measures to reported alcohol intake are not consistent but paradoxically tend to show negative associations in women.² However, observational studies may be difficult to interpret as they may be confounded by a variety of factors such as lifestyle, social status³ and biased reporting. Dietary experiments therefore seem to be required to address the effect of alcohol beverage drinking on energy balance. However, such studies are subject to other sorts of limitations. Voluntary food consumption at a single meal after the intake of an alcoholic beverage is a frequently used design to test the acute effect of alcohol on energy intake.^4,5,6,7,8 However, by using such an approach there are several issues to consider in relation to the choice of the non-alcoholic reference beverage. The non-alcoholic beverage may contain less or the same amount of energy as the alcoholic preload. If the lack of energy caused by alcohol removal in the reference beverage is not substituted by other macronutrients, findings of incomplete compensation for the alcohol energy in the test drink does not necessarily mean that alcohol has unique non-satiating properties. Such design may simply demonstrate a general poor satiating capacity of dissolved macronutrients.⁹ On the other hand, if the reference beverage is supplemented with other macronutrients to achieve an energy content identical to that of the alcoholic drink the satiating effect of alcohol would be compared to these macronutrients. The time that the test beverages are ingested in relation to the ad libitum meal may also be important. Thus, the possible impact of ethanol on food intake may depend on its metabolic effects, its psychogenic and its organoleptic effects, which may come into effect at different times. If the test beverage is ingested separately as a pre-load, organoleptic interactions may be abolished. Such a design is preferred by most investigators, probably because the organoleptic effects of alcohol may differ with regard to the taste qualities of the accompanying meal, thus making interpretation of the results very complicated. However, such a protocol may not fully reflect real life as most people also consume wine and beer during the meals.

In the present study we compared energy intakes at identical ad libitum three-course supper meals while red wine, beer or a carbonated soft drink were served with the food. Two different studies were performed. One where the subjects were allowed to drink the beverages ad libitum with the meals and one where fixed quantities, of which two-thirds were ingested prior to the meal, were consumed. The subjects' impression of the appearance and taste qualities of the courses was assessed to see whether the different beverages had varying effects on the subjects' perception of the meal which, in turn, might influence appetite and the amount eaten.

Subjects and methods

Energy intake was measured at an ad libitum supper meal accompanied by red wine (Jumilla, a Spanish medium dry Valpolicella with 13 vol/vol percent alcohol and an energy per volume of 3.0 kJ/ml), lager beer (Carlsberg Hof with 4.6 vol/vol percent alcohol and an energy per volume of 1.66 kJ/ml) or carbonated non-alcoholic soft drink (Sprite Regular with an energy per volume of 1.73 kJ/ml). Two different studies were performed both with the different beverages administered in a sequence which was randomized amongst the subjects. In the first study the beverages were provided ad libitum (up to limit of 750, 1667 and 1500 ml for wine, beer and carbonated soft drink, respectively) with the whole meal. In the second study the beverages were provided in fixed quantities corresponding to 3.19 ml/kg body weight for wine, and 9.00 ml/kg body weight for beer and carbonated water. Hence, wine and beer were provided in isoalcoholic amounts. To mimic a typical social gathering in this study one-third of the beverage was administered 40 min before the start of the meal, the next third 15 min before the meal and the last third together with the second course of the meal. The subjects were instructed to finish each of the two first beverage servings within 10 min, while the last serving was consumed while eating the second course. Water was served ad libitum throughout the entire meal.

Twenty-two, apparently healthy, male subjects (mostly medical students, age 20-33 y, weight 60-99 kg, body mass index (BMI) 19.9-25.5 kg/m²) volunteered for both studies. Twenty-one of them were identical in the two studies. All of the subjects reported a moderate weekly alcohol consumption, except for one subject who did not indicate any consumption. Most of the subjects reported drinking both beer and wine with a higher alcohol intake from Friday to Sunday than during the rest of the week. In an attempt to standardize energy balance prior to the experiments the subjects consumed at home the same breakfast and lunch meals, provided by the project, before each test. Energy content corresponded to 20 and 30% for breakfast and lunch respectively of the subjects' estimated daily energy requirements calculated as 1.7´BMR, where BMR (kJ) was predicted as 64´body weight (kg)+2840 according to Sandström et al.¹⁰ The breakfast consisted of white bread, cheese and raspberry jam and contained 60 and 26 energy-percent carbohydrate and fat, respectively. Lunch was composed by rye bread, white bread, tuna, ham, liver paste, cheese, mayonnaise, cucumber and orange juice and had an energy composition of 52 and 30 energy-percent carbohydrate and fat. The subjects were instructed to reproduce the time schedule for these meals before all the experiments and to maintain the same level of physical activity for a 2 day period prior to the tests. All experiments were conducted on Tuesdays, Wednesdays or Thursdays and we succeeded in testing each subject on the same day of the week in all the six trials except for three subject-tests.

The ad libitum test meal was identical in the two studies and consisted of three courses: a pasta salad with ham, cream and yoghurt, peas and peppers, a goulash dish served with white bread and a chocolate sponge cake. The subjects filled their plates from a large bowl containing an abundant quantity of pasta salad or goulash, which were homogeneously prepared to allow an accurate re-weighing of leftovers. The subjects were told to eat ad libitum until they felt comfortable and not to be concerned about leftovers. They were never informed that the purpose of the experiments was to test food intake. They were told that the studies were conducted to evaluate the influence of the different beverages on their perception of the meal and that their food intake and appetite were only measured as they could affect how the meal was experienced. The subjects were placed at a distance of more than 2 m from each other, and they were not allowed to communicate during the experiments. No restrictions were imposed on the time the subjects spent on each course. The subjects indicated their sensation of hunger, satiety, fullness, prospective eating, desire for something salty, desire for something sweet, desire for fatty food, thirst and general comfort on 100 mm long visual analog scales (appetite-VAS). VAS were also used for the subjects' evaluation of the general 'appetizingness', taste, smell, appearance and after-taste of each of the courses (hedonic-VAS). Energy intake of beverages was calculated by using energy density measures provided by the manufacturer. Energy intake from the test food was calculated using DANKOST version 2.0 dietary assessment software (Danish Catering Centre Ltd). The study was approved by the Ethics Committee of Frederiksberg and Copenhagen and the subjects gave written informed consent according to the declaration of Helsinki II which states that the subject may withdraw from the study at any time without consequences.

Study I (ad libitum beverage intake study)

The subjects reported to the department at 17:00 h. and were seated at individual tables in a dining room. At 17:05 h the subjects completed appetite-VAS forms. Subsequently, all the three courses were shown to the subjects and they were informed that an unlimited amount of each course was available. At 17:15 h the pasta salad and the beverage were served. The beverage and the quantity offered were as follows: one bottle of wine (750 ml), five bottles of beer (333 ml each) or six bottles of carbonated soft drink (250 ml each). The bottles were all placed on the table in front of the subjects. The subjects filled their own glasses from the bottles. Instructions were given to consume as much of the beverage as they pleased during the meal, but were not offered beverage in excess of the amount presented at the introduction of the meal. The subjects indicated when they were finished with the first course and it was removed. Then the subjects made a hedonic-VAS evaluation of the first course before the second course was provided. This procedure was repeated between the second and third courses where hedonic-VAS were completed for both the goulash and bread. The third course was also evaluated by hedonic-VAS. Ten minutes after the termination of the third course appetite-VAS forms were completed.

Study II (fixed beverage intake study)

The beverage was administered according to a combined pre-load and within meal protocol. The subjects reported to the department at 16.30 h and were seated at tables in the dining room. After completion of appetite-VAS one-third of the prescribed amount of the beverage was consumed within 10 min. At 17:00 h the second third of the beverage was ingested. Appetite-VAS were completed after finishing each of the two beverage servings. At 17:15 h the first course was served and the protocol of study I was now repeated, except that the beverage consumption was fixed as the last third of the prescribed quantity and that its time of consumption was confined to the second course. Moreover, in contrast to study I, tap water was available ad libitum throughout the whole meal.

Statistics

Total energy intake, energy intake from food and beverage and hedonic-VAS were compared for wine, beer and soft drink by a split-plot analysis if data were normally distributed. If the overall analysis showed significance the Tukey pairwise multiple comparison procedure was applied to determine which of the levels differed. Friedman repeated measures analysis of variance on ranks was used if data were not normally distributed. In this case the Student Newman Keuls test was applied as the post-hoc test. These analyses were performed using the SIGMASTAT program (Jandel Scientific, Erkrath, Germany). For the beverage ad libitum study, associations between the effect of beer and wine on food energy intake compared to soft drink and the corresponding changes in beverage energy intake were tested by Pearson correlation analysis. Appetite-VAS were tested by a two-way split-plot analyses with beverage type and time of recording as within subject factors. Differences from the first recording, before any beverage or food was ingested, were used as out-come variables. SPSS 7.5 for Windows (SPSS Inc., Chicago, IL, USA) was used for the two latter analyses.

Results

Energy intakes in the ad libitum beverage intake study are shown in Figure 1. In this study total energy intake was higher when wine was administered, both in comparison with beer and with the soft drink. This was explained both by a higher energy intake of wine, which significantly differed from that of the soft drink, and by higher average of total food intake which failed to reach statistical significance. The latter was partly attributable to a significantly higher cake intake. No relationships were observed between changes in beverage energy intake from alcoholic beverages vs soft drink and the corresponding changes in food energy intake (P>0.3). Beverage intake in ml was higher with beer (909±81 ml) than with the soft drink (753±48 ml), which again was higher than with wine (570±37 ml; P<0.05 for all three comparisons). Alcohol intake was 41.8±3.7 ml with beer and 74.1±4.8 ml with wine (P<0.0005). Figure 2 shows energy intakes in the fixed beverage intake study. There was no significant difference in total energy intake of food between the three beverages. As beer and wine were administered in isoalcoholic amounts the energy intake of the wine was by design lower than that of beer and soft drink. The intake of the goulash was higher with both alcoholic beverages than with the soft drink. Intake of the supplied water was higher with wine (529±54 ml) than with beer (396±52 ml), which again was higher than with soft drink (318±54 ml; P<0.05 for both comparisons). However, the lower beverage liquid intake with wine by design was not fully compensated for by a higher intake of water. This resulted in a higher total liquid intake both with beer and soft-drink compared to wine (soft-drink vs beer vs wine: 1002±63 vs 1080±59 vs 772±55 ml, P=0.001).

In the ad libitum beverage intake study no differences in response to the meal between wine, beer and soft-drink were found with regard to appetite-VAS parameters, except for general comfort, where a significant increase compared to pre-meal assessment was seen after the meal with wine but not after beer or soft-drink (changes from pre-meal: soft drink vs beer vs wine: -4.6±4.4 vs 0.6±4.4 vs 11.1±3.7 mm). Such an effect of wine was not observed in the fixed beverage intake study, where thirst was the only parameter where an overall difference in the response to the meal between the beverages could be found. The negative effect on thirst scores was significantly less after wine compared both to soft drink and beer (P<0.01). This could be due to the lower liquid intake during the wine test. No hedonic-VAS differed significantly between the beverage types in the two studies, but general 'appetizingness' of the goulash was rated borderline significantly better (P=0.06) in the wine test than when soft drink was given in the ad libitum beverage study (soft drink vs beer vs wine: 58±5 vs 67±3 vs 71±4 mm).

Discussion

The ad libitum beverage intake study demonstrates that if an unlimited amount of wine is offered to younger men much energy can be consumed from the beverage itself without any compensation in food intake of an accompanying meal. In the present beverage ad libitum study six of the 22 subjects surprisingly drank 750 ml wine (2.3 MJ), ie the whole bottle, which was the maximum allowed for ethical reasons, despite that fact that the subjects had reported a moderate habitual alcohol intake. We cannot exclude that wine consumption would have been even greater in some of these subjects if more wine had been available. On the other hand, the whole bottle of wine which was presented with the meal may unconsciously have been regarded as a 'serving size'. This could be in disagreement with normal wine-drinking behavior where a wine bottle typically is shared. In contrast, only two subjects reached the upper limit of 1667 ml (2.7 MJ) for beer consumption. Hence, the reduction in beverage volume with wine compared to soft drink was insufficient to fully compensate for the higher energy density of the wine. As no additional liquid was offered in this experiment, thirst may have had a greater driving force on wine compared to beer intake relative to the energy content of the beverages. This effect may have been enforced by a diuretic effect of the higher alcohol intake with wine. Furthermore, in contrast to wine, the urge for beer and soft drink may have been limited by gastric distention caused by CO₂. However, fullness did not score differently between the beverages.

In spite of the greater energy intake as fluid no compensation was observed in pasta salad, goulash and bread intake, and the intake of cake was even higher with wine than with soft drink. Sensory specific satiety may have suppressed the intake of the cake in the soft drink trial. Total energy intake was therefore higher with wine. Furthermore, by correlation analyses we were unable to detect any relationship between beer and wine induced changes in beverage energy intake compared to soft drink and the corresponding changes in food energy intake. This indicates that even the subjects who exhibited the greatest increase in energy intake from beverage in the wine test did not compensate by a reduction in food intake. This may be another example of the poor or absent compensation for liquid energy that has been demonstrated in previous investigations.^9,11

The increased intake of cake with wine could be attributable to a specific hyperphagic effect of the alcohol intake being 77% higher with wine than with beer. The increase ratings of general comfort with wine but not with beer or soft drink may have been due to a frank lightheadedness caused by a higher plasma concentration of alcohol. This may in turn have relieved some eating restraints. Alternatively, the comfort of the subjects during beer and carbonated soft drink consumption might have been attenuated by gastric distention caused by the CO₂. Although liquid intake from beer was higher than from wine in the ad libitum study, average energy intake from beer was between that of soft drink and wine, and it did not differ significantly from either of these beverages. This is due to the fact that the beer was less energy dense than the wine as its energy per volume only was 55% of that of the wine.

In the fixed beverage study the pre-defined alcohol intake both with beer and wine was 31.5 ml, which only corresponded to 75 and 43%, respectively, of the amounts consumed in the ad libitum study. In this study, the intake of goulash was higher with beer than with soft drink. This could not be related to a lower volume or energy content of the beer compared to soft drink because these factors were similar by design. The intake of goulash was also higher with wine, but in neither case did the difference in total energy intake compared to soft drink reach statistical significance. Hedonic effects of beer and wine might have made the goulash more palatable and so caused the higher intake compared to the soft drink test. That wine may have such an effect was implied by a trend for the goulash to achieve a better VAS rating of general 'appetizingness' during the ad libitum wine test compared to soft drink. A psychogenic effect of alcohol facilitating food intake may be another explanation.

Facilitation of total energy intake by alcoholic beverages beyond what can be explained by a lack of compensation for liquid energy has been indicated by other studies. Thus a negative compensation for the energy in an appetizer consisting of beer or wine at a subsequent ad libitum lunch meal has been reported. This was in contrast to a partial compensation (about 30%) when isoenergetic and isovolumetric liquid appetizers containing carbohydrate, fat or protein were given.⁶ The higher intake of lunch after alcohol was a result of both an increased eating rate and eating time. No compensation occurred for the increased intake with alcohol compared to the non-alcoholic appetizers during the rest of the day, according to subjects' own food intake diaries. In two other studies, alcoholic beer given as a preload resulted in a significantly higher intake of a test lunch presented after 20-30 min compared to nonalcoholic beer demonstrating a negative compensation for the energy in alcohol.^5,8 This hyperphagic effect appeared not to be related to expectations or conditioned learning about alcohol drinking and eating as a debriefing of the subjects indicated that most subjects were not conscious about the lack of alcohol in the nonalcoholic beer.⁵ A higher voluntary energy intake has also been found following a high fat appetizer including alcohol compared to an isocaloric low fat non-alcoholic appetizer with the same volume.¹² As the beverages in these studies were administered separately from the test meal the results may be explained by metabolic and/or psychogenic mechanisms rather than by organoleptic interactions with the beer or wine. Although reservations must be made, these findings support the theory that the body differentiates between energy as alcohol and other sorts of liquid energy.

Other studies, however, have failed to demonstrate a specific facilitating effect of alcohol on energy intake. In a two week in-patient study, where prescribed beverages were altered between alcoholic and glucose beverages and no beverages, the day-to-day compensation in ad libitum food intake for the energy in the drinks was incomplete. However, the level of compensation for the alcohol and glucose beverages was similar.¹³ In this study the subjects' attention was diverted from the fact that food consumption assessment was the true objective of the experiments and possible feeding restraint factors may therefore have been of less importance, leaving less opportunity for an unrestraining effect of alcohol to influence the results. Disinhibition of restraint may, however, in some cases, not be involved in the increased food intake with alcohol drinking as alcoholic fruit juice compared to non-alcoholic fruit juice has been reported to promote food intake in unrestrained but not in restrained men.⁴ The 2-week in-patient study was also in line with findings from a 7-day self record study where food intake seemed to be independent of alcohol intake, both on a between-subjects and between-meals level, yielding higher total energy intakes with alcohol.¹⁴ On the other hand there was no implication in that study of a true facilitation of food intake by alcohol, although the meals were more prolonged if consumed with alcoholic beverages. Apparently only one long-term free-living study has tested the effect of prescribed moderate alcohol drinking on total energy intake and body weight and composition. In that study, a 6 week period with a daily intake of 170 ml red wine did not alter body weight or body fat, which also was unaffected by 6 weeks with total alcohol abstinence.¹⁵ Furthermore, 3 day dietary records showed similar energy intakes in wine drinking and abstinence periods. Although the intervention period may have been too short and the food intake measures too inaccurate to detect minor changes in total energy intake, the total energy of the administered wine would be about 21 MJ, corresponding to an increase of about 750 g in body weight. To this figure should be added a greater energy intake from food provided that the wine had facilitated food intake, and in this case the change in body weight would probably have been detectable. The study therefore indicates that some compensation rather than facilitation occurs when wine is introduced in a fixed moderate amount into the habitual diet.

On a long-term basis a compensation in food intake to moderate daily alcohol intake is therefore likely to occur, at least if the beverages are consumed with meals. It therefore seems to be important to dissociate single-meal studies such as the present, which may show an acute hyperphagic effect of alcohol, from the more prolonged effects on energy balance of regular moderate drinking. Furthermore, the high intake of wine (>300 ml for all subjects except one) provoked by the present beverage ad libitum study may be applicable to a bingeing pattern of alcohol consumption. Such drinking behavior may affect food intake differently from regular moderate drinking. Finally, we did not allow communication between the subjects during the meals to avoid mutual encouragement to drink. Although this may improve environmental control it may at the same time be a limitation as social interactions may stimulate both eating and drinking. Allowance for social effects might therefore have amplified the hyperphagic effect of alcohol found in this study.

In conclusion, the present data suggest that acute compensation in food intake may be absent even when a high liquid energy intake is induced by ad libitum wine drinking. Due to the lower energy density of beer, and possibly also due to its lower alcohol concentration and consequently smaller alcohol intake by the subjects, the facilitating effect of beer on total energy intake in the ad libitum study may have been less pronounced compared with red wine. However, approximately three units of alcohol either as beer or wine appeared to be sufficient to exert a modest hyperphagic effect in the fixed beverage study, although gastric distention by carbonated beverage drinking or effects on taste perception of food cannot be ruled out as an alternative explanation. More long-term intervention studies with the introduction of beer and wine vs soft drinks, using body weight and composition as outcomes, are needed to elucidate whether compensation will occur for the energy in different alcoholic beverages on a chronic basis.

Acknowledgements

We would like to thank our kitchen staff, dieticians and Charlotte Jakobsen for their invaluable contributions. We are grateful for Cristina Cuthbertsons for linguistic corrections. This work was supported by: The Danish Brewers Association, The Toubro Foundation and The Grethe Pedersen Foundation.

1 Brewers and Licensed Retailers Association Statistical Handbook 1995, Brewing Publications Ltd.

2 Hellerstedt WL, Jeffery RW, Murray DM. The association between alcohol intake and adiposity in the general population. Am J Epidemiol 1990; 132: 594-611. MEDLINE

3 Männistö S, Uusitalo K, Roos E, Fogelholm M, Pietinen P. Alcohol beverage drinking, diet and body mass index in a cross-sectional survey. Eur J Clin Nutr 1997; 51: 326-332. Article MEDLINE

4 Yeomans MR, Hails NJ, Nesic JS. Alcohol and the appetizer effect. Behav Pharmac 1999; 10: 151.

5 Yeomans MR, Phillips MF. Failure to reduce short-term appetite following alcohol is independent of beliefs about the presence of alcohol. Nutr Neurosci 2002; 5: 131-139. MEDLINE

6 Westerterp-Plantenga MS, Verwegen CRT. The appetizing effect of an apéritif in overweight and normal-weight humans. Am J Clin Nutr 1999; 69: 205-212. MEDLINE

7 Poppitt SD, Eckhardt JW, McGonagle J, Murgatroyd PR, Prentice AM. Short-term effects of alcohol consumption on appetite and energy intake. Physiol Behav 1996; 60: 1063-1070. MEDLINE

8 Hetherington MM, Cameron F, Wallis DJ, Pirie LM. Stimulation of appetite by alcohol. Physiol Behav 2001; 74: 283-289. MEDLINE

9 DiMeglio DP, Mattes RD. Liquid versus solid carbohydrate: effects on food intake and body weight. Int J Obes Relat Metab Disord 2000; 24: 794-800. Article MEDLINE

10 Sandström B, Aro A, Becker W, Lyhne N, Pedersen JI, Pórsdóttir I (eds). Energi. In Nordiska näringsstofrekommendationer vol 28, Nord, 1996, pp 47-66.

11 Mattes RD. Dietary compensation by humans for supplemental energy provided as ethanol or carbohydrate in fluids. Physiol Behav 1996; 59: 179-187. MEDLINE

12 Tremblay A, St-Pierre S. The hyperphagic effect of a high-fat diet and alcohol intake persists after control for energy density. Am J Clin Nutr 1996; 63: 479-482. MEDLINE

13 Foltin RW, Kelly TH, Fischman MW. Ethanol as an energy source in humans: comparison with dextrose-containing beverages. Appetite 1993; 20: 95-110. MEDLINE

14 de Castro JM, Orozco S. Moderate alcohol intake and spontaneous eating patters of humans: evidence of unregulated supplementation. Am J Clin Nutr 1990; 52: 246-253. MEDLINE

15 Cordain L, Bryan ED, Melby CL, Smith MJ. Influence of moderate daily wine consumption on body weight regulation and metabolism in healthy free-living males. J Am Col Nutr 1997; 16: 134-139.

Figure 1 Energy intakes at ad libitum beverage study. § P<0.05 vs carbonated soft drink. *P<0.05 vs beer and carbonated soft drink. Statistical significances tested by Tukey (in case of normality) or Student-Newman-Keuls test (in case of lack of normality) post hoc tests if the overall repeated measures analyses showed significance.

Figure 2 Energy intakes at fixed beverage study. §P<0.05 vs carbonated soft drink. Statistical significances tested by Tukey (in case of normality) or Student-Newman-Keuls test (in case of lack of normality) post hoc tests if the overall repeated measures analyses showed significance.