The impact on mood and memory of consuming a probiotic containing milk drink, or a placebo, was examined as, previously, a poor mood has been found to correlate with the frequency of constipation.
A double-blind placebo-controlled trial with random allocation of subjects.
Subjects went about their normal life in the community apart from three visits to the laboratory.
One hundred and thirty-two healthy members of general population, mean age 61.8 years, volunteered in response to local media coverage. One hundred and twenty-four finished the trial.
For a 3-week period, either a probiotic containing milk drink, or a placebo, were consumed daily. Mood and cognition were measured at baseline, and after 10 and 20 days of consumption.
At baseline those who reported themselves to be less frequently constipated were more clearheaded, confident and elated. Although the taking of the probiotic did not generally change the mood, this appeared to be a reflection of the generally good mood in this sample. When those in the bottom third of the depressed/elated dimension at baseline were considered, they selectively responded by reporting themselves as happy rather than depressed after taking the probiotic. The intervention did not, however, influence the reported frequency of defaecation, probably a reflection of the initially low incidence of constipation. An unexpected and possibly chance finding was that the consumption of probiotics resulted in a slightly-poorer performance on two measures of memory.
The consumption of a probiotic-containing yoghurt improved the mood of those whose mood was initially poor. This improvement in mood was not, however, associated with an increased frequency of defaecation.
Funded by Yakult, Japan.
Probiotics are increasingly used as a food supplement, suggesting that they may offer a range of health benefits. Such products usually contain lactic acid-producing bacteria that belong to the genera Lactobacillus and Bifidobacterium. The impact on a wide range of symptoms has been reviewed, including the relief of irritable-bowel syndrome, inflammatory bowel disease, the amelioration of lactose intolerance and the prevention of bowel cancer (Kopp-Hoolihan, 2001; Broekaert and Walker, 2006). In addition, emerging research suggests that the microflora of the intestines may influence the immune system and functioning beyond the gut (Logan and Katzman, 2005). Given the wide range of proposed benefits, the present study has considered whether the taking of probiotics might result in an increase in general well-being. In particular, whether a decrease in the incidence of constipation might be associated with improved mood was examined.
There are several reports that those suffering from chronic-constipation score more poorly on measures of well-being, mood or quality of life (Wald et al., 1989, 1992; Glia and Lindberg, 1997; Norton, 2006), although the samples considered have tended to be in hospital or at least had a long-standing and severe problem. A recent speculative review suggested that probiotics may prove to be an adjuvant therapy for those with major depressive disorders (Logan and Katzman, 2005). Although a range of mechanisms was discussed, it is possible that as probiotics reduce constipation this may play a role.
The association between constipation in the general population and quality of life has received little consideration. An exception was the study of Benton (2001), who related the bodily reactions to food consumption and the general mood of healthy young women, not chosen for a history of constipation. The major finding was a consistent tendency for those reporting frequent constipation to report poorer mood. There was a significant inverse association between constipation and feeling clearheaded, calm, elated and agreeable. It was the frequency of constipation rather than other types of discomfort after eating that was associated with well-being. A poor mood was found to correlate with the frequency that constipation was reported, from once a week to less frequently than once every 6 months. The correlational nature of these data, however, prevented the conclusion that there was a causal association. There are many factors, for example, exercise, that could influence both mood and the incidence of constipation. Thus, there is a need for studies that monitor well-being while manipulating the incidence of constipation. Therefore, the present study considered the impact of consuming a probiotic-containing drink, rather than a placebo, on mood. In addition, as poor mood and poor memory are known to be related (Phelps, 2006), this aspect of cognition was also assessed.
One hundred and thirty-two members of the general population were recruited when they responded to a newspaper article inviting participation, and attended an initial meeting, where baseline scores were obtained. Of these, 126 returned after 10 days of taking the drinks, and 124 completed the trial. The data of one person who incorrectly filled in the questionnaires was discarded. Of the six who failed to return for the second meeting, two did so because of problems of attending at the time suggested, one had a death in the family, two did not like the taste of the drink and the reason in the sixth instance is unknown. The reasons for the nonattendance of the two subjects who withdrew between the second and final meeting are not known.
Inclusion criteria were that they were, by self-report, in good health and did not consume yoghurt containing live bacteria. In particular, they affirmed that they did not suffer from depression, dementia, schizophrenia or any neurological disorder, and did not have any clinically significant problems of the heart, lungs, kidney or liver; if malignancy had occurred, it had been in remission for at least 2 years. Diabetes Mellitus was acceptable only if it was controlled by diet or oral hypoglycaemics. Hypothyroidism was acceptable when stabilized by replacement therapy for more than 6 months. Hypertension was acceptable if it was treated and stable for at least 3 months.
Of those returning for the second meeting, 51 were male and 75 female. The average age was 61.8 years ±7.3 (s.d.) with a range from 48 to 79 years. As judged by details of their place of residence, their occupations and educational background, the samples were mostly well educated and reasonably affluent.
The procedure was approved by the local ethics committee and the participants gave informed consent.
Subjects attended in a series of small groups and gave background data, completing the national adult reading test (NART) and eating-associated behaviour questionnaire. Measures of memory and verbal fluency were obtained. Baseline data were obtained for mood using the profile of mood states (POMS). Memory and verbal fluency was assessed. Randomly and under a double-blind procedure, subjects were allocated to a group that consumed, on a daily basis, a probiotic-containing milk drink or a placebo.
At home, on a daily basis, mood was rated on visual analogue scales and the number of bowel movements was recorded.
The cognitive and mood tests taken at baseline were repeated in the laboratory.
As days 1–10.
The cognitive and mood tests taken at baseline were repeated.
A probiotic-containing milk drink (supplied by Yakult, Japan), contained Lactobacillus casei Shirota mixed together with water, sugar, skimmed milk powder and flavouring. Each pot was 65 ml and provided at least 6.5 × 109, that is 108 per millilitre, live bacteria. A milk-based placebo was created of similar colour and taste, but without live bacteria.
In preliminary tests, subjects were asked to rate each drink on four scales; sweet/sour, watery/creamy, pleasant/unpleasant, intense flavour/weak flavour. The assessments were made using four 100 millimetre visual analogue scales with one of the pairs of adjectives at each end of the line. With all ratings, the active drink and placebo were not significantly different. Each 65 ml bottle provided 50 kcal and 0.8 g of protein, 12 g carbohydrate and less than 0.1 g fat.
At baseline and after 10 and 20 days, mood was assessed using the questionnaire-based POMS (Lorr and McNair, 1984). The tests consist of 72 adjectives to which the subjects responded using a Likert scale by indicating the extent to which each adjective had applied to them in the last week. The test measures six basic dimensions of mood; energetic/tired, clearheaded/muddled, composed/anxious, confident/unsure, elated/depressed and agreeable/angry. These basic mood dimensions were also rated on a daily basis; every evening subjects rated their mood throughout the day on six 10 cm visual analogue scales with these pairs of adjectives at either end. The decision was made to see if those with an initially poorer mood were more likely to respond, as it seemed possible that a generally positive mood may have been masking a response to supplementation. The baseline mood scores were divided into thirds. The choice of this division reflected the generally positive mood of most of the sample; at baseline 80% were above the middle point of the depressed/elated scale; that is, their mood was initially more positive than negative. The choice of thirds reflected a compromise between the need to study those with poorer mood while considering a sufficient sample size.
Episodic memory was measured using two stories from the logical memory test of the Wechsler Memory Scale (1998). A third story was constructed of the same length with a similar number of items to be remembered. Each story was played on a tape recorder. Immediately after hearing the story, the subjects wrote down as many details as they could recall. At the end of the test session, a measure of delayed recall was obtained by asking them a second time to write down the story. A mark was given for each aspect of the story recalled.
Retrieval from long-term memory
The ability to recall the capital cities of a number of countries provided a measure of the ability to retrieve names from long-term memory (Gruneberg et al., 1977). As this test could only be used once, it was administered at the end of the study. A list of 30 countries was given. These were chosen as during pilot studies as they were not universally known or known to only a very few. With cases where no answer was produced, subjects were asked to rate on a six-point scale whether they had the feeling of either knowing or not knowing the answer they could not recall.
Tests of verbal fluency are used to measure the functioning of the frontal lobes of the brain (Alvarez and Emory, 2006). On each occasion, one of three letters was given in turn, with the instruction to write down as many words as possible in 1 min beginning with that letter. The letters used were RWA, CFL and PRW on the successive visits.
Subjects responded to the following questions by ticking one of either five or six responses chosen to be reasonable for the question, for example once a day, most days, once a week, once in 6 months, once a year, less often. The questions were adapted from Glia and Lindberg (1997) to make the terminology more easily understood.
After eating, how often do you feel sick?
After eating, do you feel stomach pain?
How often do you experience burning in your stomach after eating?
How often do you experience burning in your mouth after eating (regurgitated acid)?
How often do you feel the need to burp after eating?
How often do you have diarrhoea?
How often do you experience a bloating feeling after eating?
Do you experience the passage of wind?
How often do you suffer from indigestion?
How often do you feel you wish to empty your bowels but are unable to do so?
How often do you empty your bowels?
How frequently are you constipated?
The NART Revised is used in neuropsychological assessment when there is a need to establish pre-morbid intelligence (Nelson and Willison, 1991). A series of words, of decreasing frequency in English, pronounced in a non-standard way, are presented in a written form. The number that can be correctly pronounced predicts measures of intelligence (Hart et al., 1986).
The responses were analysed using appropriate analysis of variance (ANOVA) and analysis of covariance models. Typically, drink (probiotic/placebo) × Gender × Test session (after 10 or 20 days) with the last as a repeated-measures factor. Baseline values were used as a covariate. When memory was examined, immediate/delayed recall was included as an additional repeated-measures factor.
When the baseline measures of those who subsequently consumed the two types of drink were compared, no significant differences were found with any measure of mood or cognition. Also, the frequency with which constipation was reported to occur did not differ in those who subsequently received the probiotic (6.17±0.29) or the placebo (6.15±0.34) (a rating of 6 indicated that constipation occurs once every 6 months). The two experimental groups were well matched.
Daily ratings of mood
No general effects of taking the probiotic were found when daily mood ratings were considered. However, given the generally positive mood of the sample, further analysis considered whether there was an interaction with baseline mood: it seemed possible that those with a poorer mood would be more likely to respond. Baseline mood ratings were divided into thirds and this factor was added to the ANOVA. With five out of the six mood dimensions, the drink again failed to influence mood: agreeable/hostile (F(2,103)=0.27, NS); clearheaded/confused (F(2,104)=0.78, NS); composed/anxious (F(2,104)=0.24, NS); confident/unsure (F(2,104)=0.18, NS); energetic/tired (F(2,99)=0.42, NS). In contrast, when the depressed/elated dimension was considered, the interaction Drink × Baseline mood reached statistical significance (F(2,103)=3.82, P<0.025) (Figure 1). the bottom third of the POMS depressed/elated distribution were those at baseline who responded to the probiotic with improved mood (P<0.05).
When the six mood dimensions from the POMS were considered, on no occasion was there a general effect of taking the probiotic. However, based on findings with the daily ratings, it seemed possible that a lack of response may have reflected the generally good mood of many in the sample. Baseline mood was, therefore, divided into thirds and included as a factor in an ANOVA. With the depressed/elated scores, the interaction Drink × Test session × Baseline mood (thirds of population on depressed/elated dimension) again reached statistical significance (F(2,111)=4.19, P<0.02). The data are presented in Figure 2. The interaction reflects a significantly improved mood after 20 days in those who drank the probiotic and were, at baseline, more depressed (P<0.04).
The frequency at baseline that subjects reported suffering from constipation was examined by dividing the population into three: those reporting it weekly or more often, from 1 to 6 months, or less frequently. The mood throughout the study was considered by examining the daily ratings. Those who were constipated less frequently were more clearheaded, confident and elated, with the ratings of energetic and agreeable, just missing statistical significance (Table 1).
As the study focused on the effect of probiotics on those who failed to defaecate regularly, subjects considered were those who, during the first 5 days of the study, reported the passing of a motion less than five times, that is on average less than once a day. When the depressed/elated dimension was considered, the interaction Drink × Time reached statistical significance (F(3,33)=2.82, P<0.05). The interaction reflected a tendency for those consuming the probiotic to feel more elated rather than depressed towards the end of the study. The post hoc tests, however, failed to produce significant differences at any time point, possibly the reflection of a small sample size.
When the number of motions reported on days 1–5, 6–10, 11–15 and 16–20 were considered, there was neither a drink main effect (F(1,111)=1,45, n.s.) nor a Drink × Day interaction (F(3,333)=0.37, n.s.). The probiotic drink did not increase the frequency of defaecation in this sample.
When recall of the stories was considered, the Drink × Test session interaction reached statistical significance (F(1,119)=3.91, P<0.05). Figure 3 illustrates the interaction. Although the drinks made no difference to the memory scores after 10 days, at 20 days those who had consumed the placebo had significantly better memory scores (P<0.02).
The intelligence scores obtained from the NART and the baseline memory of the story were converted to Z scores. As intelligence and memory correlate, when these two Z scores are subtracted, a minus value can be interpreted as indicating a decline in memory with age. This variable was divided into thirds and added to the ANOVA. The interaction Drink × Level of memory decline reached statistical significance (F(2,109)=3.30, P<0.04). Post hoc analysis indicated a trend for those with failing memory to benefit from the probiotic, although the effect failed to reach significance. In those whose memory was better, the trend was for the placebo to be associated with better memory.
As a test of the ability to retrieve information from long-term memory, the ability to recall the names of capital cities was assessed. The drink did not influence the number of capitals correctly (F(1,119)=0.15, NS) and un-correctly recalled (F(1,123)=0.10, NS). However, when the capitals for which no answer was given were considered the drink comparison reached statistical significance (F(1,119)=4,52, P<0.03); those taking the placebo were more certain that they knew the answer, although unable to recall it .
With the verbal fluency scores, neither the Drink main effect (F(1,118)=0.0,. NS) nor the interaction with Time (F(1,118)=0.56, n.s.) reached statistical significance.
As the frequency with which adverse symptoms reported on the eating-associated behaviour questionnaire was low and there were no differences on individual scales, the responses to the questions were added together to create an index of food-related problems. The nature of the drink failed to influence the incidence of these problems (F(1,115)=0.34, NS); similarly, the interaction Drink × Test session was non-significant (F(2,230)=1.01, NS). There was no evidence in this population that the taking of the probiotic decreased eating-associated problems, but this finding should be placed in the context of a low baseline incidence of such problems so that an improvement was unlikely to be observed.
Maintenance of double-blind
At the end of the study, subjects were asked to guess whether they had been taking the probiotic or a placebo. Of those consuming the probiotic, 25 stated that they thought they had consumed the probiotic, 14, the placebo and 24, stated that they were unsure. After drinking the placebo, the comparable figures were probiotic 17, placebo 16 and unsure 27. The most common response was to be unsure into which group you fell. A χ2 test was non-significant. Thus, there is no reason to believe that the double-blind was broken in the study. In a series of analyses, the various measures of mood and cognition were never found to be related to what subjects believed they were consuming (rather than the drinks they were actually consuming). Therefore, there is no reason to believe that the findings of this study reflected suggestion or the expectation of the subjects.
There are several reports of an association between constipation and poor mood. In a group of the severely constipated, Glia and Lindberg (1997) found that indigestion, dyspeptic syndrome and bowel dysfunctioning correlated with a measure of well-being. Wald et al. (1989, 1992) found that those with a slow rather than normal-transit time scored more highly on several measures of psychopathology. A high incidence of psychosocial disturbance, for example, childhood problems were found in those severely constipated (Preston et al., 1984). These findings are perhaps not surprising as people who are in discomfort can be expected to have a poorer mood.
A more surprising observation was the finding of Benton (2001) when he considered a sample of the normal population that was not chosen because constipation was a problem. The reported frequency of constipation, from once a week to less frequently than once every 6 months, correlated with mood. The present study replicated this finding (Table 1). Thus, it appears that the association between the reported frequency of constipation and mood extends beyond clinical populations and extends over a long period of time. Benton (2001) found that feeling physically uncomfortable did not appear to be the mechanism in his sample. There was no association between mood and symptoms that occurred more frequently, including indigestion and discomfort.
The correlational nature of these data made interpretation difficult. It could not be assumed that there was a causal association between constipation and mood. It is easy to suggest plausible mechanisms that may influence both mood and the incidence of constipation. Diet (Benton and Donohoe, 1999) and exercise (Thirlaway and Benton, 1992; Brown et al., 2000) are known to influence both mood and constipation. Not being constipated may simply be a marker for paying attention to your diet; hence, general health and mood may be better because you are in a generally better physical state. Also, mood may influence the incidence of constipation; previously, it has been found that those who reported themselves as stressed were more likely to be constipated (Kunimoto et al., 1998). Glia and Lindberg (1997) reported that patients with chronic constipation scored more highly on measures of anxiety and depression, supporting the suggestion that they may have a higher level of psychopathology (Fisher et al., 1987; Wald et al., 1989, 1992). Thus, the association between mood and the estimated incidence of constipation (Table 1; Benton, 2001) may be psychological rather than physiological in nature, as in the present study, no association was found between the actual incidence of defaecation, as indicated by a daily diary and mood.
To explore the possibility of a causal relationship, there is a need to manipulate the incidence of constipation while looking for any changes in mood. The addition of probiotics to the diet offers the possibility of changing the incidence of constipation while monitoring mood. The present study took this approach. However, the taking of the probiotic failed to produce either a general improvement in mood or an increase in the number of motions passed. These findings need to be placed in the context of the sample studied. It is in the nature of any research that asks for volunteers for health-orientated studies that it will tend to attract the health conscious, better educated and more affluent. The present study fell into this category. On average, the intelligence scores were one s.d. above the population average and the addresses of the sample tended to be from more affluent areas. It was not a population in which constipation was a frequent problem.
In understanding the present study, two factors need to be considered: the baseline levels of mood and the incidence of constipation. In a large-scale British survey, 40% of males and 33% of females reported a daily motion (Heaton et al., 1992). Benton (2001) found that 49% of his sample reported this pattern compared with 66% in the present study. In fact, over the first 5 days of the present study, 84% of the subjects defaecated at least five times. The average frequency of reported constipation was between once every 6 months and once a year. Clearly, we are dealing with a sample in which constipation was a rare problem, a fact that probably explains the failure to find a change in the frequency of defaecation. Similarly, the baseline mood of the subjects was generally so good that in many, an improvement could not be expected. An examination of the mood ratings illustrates the phenomenon. Using a 100-point scale, with three out of the six moods, the average ratings were in the 1990s. The lowest ratings were for feeling tired rather than energetic and depressed rather than elated.
It is more reasonable to expect a demonstration of the benefits of an intervention where baseline values are low, offering considerable room for improvement. In this context, it was interesting to note that those who were at baseline in the bottom third of the elated/depressed dimension felt more happy after taking the probiotic (Figures 1 and 2). The confirmation of this finding, using two different measures of mood, adds confidence to the finding. There is, however, no evidence that the improvement in mood reflected a decrease in the incidence of constipation.
An unexpected and possibly chance finding, for which there is no obvious explanation, was that those taking the placebo had better memories after the placebo rather than the probiotic (Figure 3). However, the effects were small and need replication before a conclusion can be drawn. The recall of a story is a frequently used measure of semantic memory (Wechsler Memory Scale, 1998). The feeling of knowing ratings when recalling the names of capital cities give a subtle and sensitive measure of memory. The tip-of-the-tongue phenomenon is a familiar experience, where you know something but cannot recall it. There are many reports that when one is unable to recall some information, those who rate themselves as being more certain that they really know the answer will, at a later time, be more likely to recall it (Koriat, 2000; Young, 2004). A high feeling-of-knowing initiates a search of long-term memory that is associated with a willingness to continue searching until an answer is found. Previously, the feeling-of-knowing phenomenon has been studied using the recall of the names of capital cities, the test used in the present study (Gruneberg et al., 1977). With capitals whose names could not be recalled initially, a higher rating that the name was, in fact, known was associated with a higher likelihood of being later recalled. Those with a higher feeling-of-knowing were more likely to recall the name of the previously unremembered capital when given the first letter.
In a novel study, the choice of experimental conditions is inevitably, to some extent, arbitrary or, at best, an informed guess. The possibility exists that the intervention time was too short or that the amount of probiotic was too small. Although the dose chosen was that commercially recommended as the maintenance dose, logically such possibilities exist and could be considered in future studies. The positive results of the present study justify future consideration of larger samples chosen for either a poor mood or a problem of constipation.
The funding of this study by Yakult, Japan is gratefully acknowledged.
About this article