Age and IQ Explained Working Memory Performance in a RCT with Fatty Fish in a Group of Forensic Inpatients

Abstract

Objectives

To investigate the effect of a long-term fatty fish intervention on a pure cognitive mechanism important for self-regulation and mental health, i.e. working memory (WM), controlling for age and IQ.

Design

A randomized controlled trial.

Setting

A forensic facility.

Participants

Eighty-four young to middle aged male forensic inpatients with psychiatric disorders.

Intervention

Consumption of farmed salmon or control meal (meat) three times a week during 23 weeks.

Measurement

Performance on WM tasks, both accuracy and mean reaction time, were recorded pre and post intervention.

Results

Performance on a cognitive functioning tasks taxing WM seemed to be explained by age and IQ.

Conclusion

Fatty fish consumption did not improve WM performance in a group of young to middle aged adults with mental health problems, as less impressionable factors such as aging and intelligence seemed to be the key components. The present study improves the knowledge concerning the interaction among nutrition, health and the aging process.

Introduction

A relationship between fish consumption and cognitive functioning in the elderly has been suggested (1). Although the precise mechanisms involved are not clearly understood, it can be assumed that nutrients such as fatty acids and vitamin-D are of importance since fish is a source of these nutrients (2). However, the relationship between fish consumption and cognitive functioning in young to middle aged adults is less convincing. Recently an observational study by Gijselaers et al. (3) found no relationship between fish consumption and cognitive performance in a group of normal healthy young and middle-aged adults. However, the question remains whether regular fish consumption can improve cognition in less healthy young to middle aged adults, e.g., individuals characterized by severe mental health problems. More knowledge about the effect of fish consumption on cognition in vulnerable young to middle aged individuals will have important implications with regard to the insight concerning the impact of nutrition throughout the life span as well as clinical implications. Thus, to establish causal relationships between fish consumption and cognition, experimental intervention studies are needed.

Cognitive functioning, and particularly executive functioning such as working memory (WM), is very important for optimal functioning and adaptation in everyday life. According to Baddeley (4), WM can be regarded as a system underlying thought processes (e.g., decision making, reasoning and planning future actions), which stores and maintains information in the short term. WM has shown to be associated with mental health, e.g., depressed patients have shown poorer WM performance compered to healthy controls (5). WM is very sensitive to aging (6). It has been assumed that executive functioning starts to decline by the age of 65 years. However, De Luca et al. (6) studied performance on a range of cognitive tasks taxing different components of executive functioning, including WM, over the life span. Participants ranged in age from 8 to 64 years and the results showed that executive functions overall seemed to decline much earlier than expected. Executive performance seemed to be at its height between 20 and 29 years of age. From 29 there was a slight decline, however a significant decline was observed for the middle aged adults, i.e., by the age of 50-64. To prevent cognitive decline, as well as aggravation in mental health problems caused by cognitive decline, there is a need for identification of effective and healthy intervention strategies. Thus, the question is whether fatty fish interventions can strengthen WM performance in a group of young to middle aged adults with mental health problems. However, due to the close relationship between aging and WM performance, age should be accounted for.

Importantly, the WM “component” of executive functions is of particular importance for carrying out other executive function tasks, such as planning, e.g., the Tower of Hanoi (TOH) test. As the task difficulty increases so does the WM load (7). Recently, it was found that regular fatty fish consumption improved performance only on the more difficult move problems on the ToH task, i.e., subtasks with high demands to WM capacity. This was especially true for participants with a history of alcohol/drug abuse (8). Based on these results one can assume that the improved performance on the planning task could be related to improved WM capacity. To gain more insight into the fish-cognition relationship it is important to investigate whether there is a causal relationship between fatty fish consumption and WM performance. However, it has been a matter of debate whether executive functioning is linked to intelligence. Research concerning this relationship has been inconsistent and it has been argued that some executive functions may be more associated with intelligence than others (9). Importantly, a relationship between WM performance and intelligence has been found (10). To gain new insight into the fish and WM relationship, controlling for intelligence is of particular importance.

The aim of the present study was to investigate whether fatty fish consumption would improve WM in young to middle aged adults characterized by mental health problems when adjusting for age and IQ. Due to the previous findings (8), we also wanted to investigate the effect of fish consumption on WM accounting for earlier alcohol/drug abuse.

Methods

Participants

In the current study 84 male forensic inpatients participated. The mean age was 41 (range: 21-60). The present study is part of a larger project concerning nutrition and mental health. As reported previously (8) the participants were characterized by some kind of psychiatric illness (e.g., personality disorders, affective disorders) classified according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR). As also reported in Hansen et al. (8) a total of 47% had a history of alcohol/drug abuse in combination with another psychiatric diagnosis.

102 inpatients responded to the invitation, but initially 95 volunteers were randomly assigned into the two groups. Before randomization to the Fish group (FG) or the Control group (CG) participants were matched on age, IQ (Wechsler Adult Intelligence Scale-Third Edition (WAIS-III) (11) and Psychopathy Checklist List-Revised (PCL-R) score (12). IQ above 75 was used as an inclusion criterion. For randomization a computerized random number generator (in Excel) was used to assign each of the matched pairs. The random allocations to the groups were completed after all participants were enrolled and had completed baseline testing. For each of the WM sub-tasks extreme scores due to inattention to the tasks were removed prior to data analyses (see Figure 1). Because there is little consensus on how to address missing data, and the risk of introducing a bias, missing data were not replaced (13).

Figure 1.

Presents the study progress

Cognitive task

In the present study computerized version of the n-back task using the E-prime system (Psychology Software Tools INC., Pittsburgh, PA) was used to measure WM. The task consisted of four different sub-tasks; 0-back, 1-back, 2-back and 3- back. Thus, the tasks varied in difficulty from the easy 0-back to the more difficult 3-back task. However, all sub-tasks consisted of a continuous flow of letters. For the 0-back task the participants were instructed to press the “1” key on the keypad as quickly as they could every time they saw the letter “h. If a letter other than “h” appeared they had to press “2” as quickly as possible. This task can be regarded as a simple reaction task as it does not require manipulation of information in the WM (14). For the 1-back task the participants had to press the “1” key if the letter on the screen matched the letter presented immediately before it. If the current letter on the screen did not match the letter that came right before it, they were told to press. For the 2-back task they were instructed to press “1” if the letter matched the one they saw two targets previously, and otherwise. The 2-back task adds a more complex element of memory since subjects must keep in mind the last two letters that were seen on the screen. This task requires focused attention and complex manipulation of new information while keeping something in the WM (15). The 3-back task was the most challenging in that the participants had to press “1” if the letter on the screen matched the letter presented three targets previously. For each task condition 40 stimuli were presented in a randomized order. The stimuli on the screen disappeared when a response was given. The participants were instructed to be as fast and accurate as possible. Number of correct responses (accuracy) and mean reaction time (mRT) were recorded by the computer. Reaction time data was log transformed (16). Participants were tested individually.

All participants underwent the same experimental test procedure prior to group assignment (pre-test, between May and July of 2008), and after the food intervention period (post-test, between January and March 2009).

Intervention

Both study protocol and experimental procedures were approved by the Ethics Committee at Sand Ridge (April 10, 2008). They were in compliance with the Helsinki Declaration and the US Federal Regulations. Volunteers were given written and oral information about the study. All participants signed an informed consent form, and they were instructed about the option to withdraw from the study at any time for any given reason without penalty. All participants were subjected to a food intervention. The FG was given farmed Atlantic salmon (Salmo salar L.) for dinner (portion size of 150 – 300 grams) thrice weekly for a period of 23 weeks (September-February). The standard portion was 300 g three times a week; however during the final four- weeks of the study, they were served portion sizes of 150 grams of salmon because we ran out of fish by the end of the study. Our calculations were based on a portion size of 200g. However, the participants were used to eat more than 200g for dinner. Since we wanted to do the posttesting while they were still eating the fish we decided to give them 150 g fish per meal while we did the post-testing. The CG was given an alternative meal (e.g., chicken, pork, beef, etc.) having the same nutritional value as they normally received with the same frequency and duration. Importantly, fatty fish (limited to tuna) was served only about once a month as part of the routine diet for patients.

Thus, the habitual dietary intake was low in omega-3. The participants' baseline levels of sum EPA and DHA in red blood cells (RBC) given as omega-3 index (in percentages of RBC total fatty acids) was at baseline 3.2% in the intervention group and 3.5% in the control group. After the study, the omega- 3 index increased significantly to 7.3% in the intervention group and was unchanged in the control group (3.8%) (17). The omega-3 index is used as an risk factor for coronary heart disease and the proposed omega-3 index risk zones are: high risk, <4%; intermediate risk, 4-8%; and low risk, >8% (18). By using this omega-3 index, the levels of EPA and DHA indicate low intake in all participants at baseline.

Double portions of all meals served during a week were collected in six consecutive weeks during the intervention period to gain dietary information and to determine selected nutrient of the menus at the institution. The menus were repeated after 12 weeks. Food and beverages from a whole week were mixed together into one sample (i.e., one sample from the CG and one sample from the FG). Table 1 shows a description of analyzed selected nutrients of control and intervention (fish) total diet. As the present study is part of a larger project investigating the effects of nutrition/fatty fish consumption on health, changes in vitamin D and omega-3 from pre-to post-test have been reported elsewhere (17).

Table 1.

Shows means and standard deviations for energy, protein, vitamin D, sum EPA+DHA and fat for the total diet in the control and intervention group

	Control (meat) diet (N=6)	Intervention (fish) diet (N=6)
Energy (kcal/100g)	219±22	198±16
Protein (g/100g)	9.3±1.0	9.1±0.7
Vitamin D(µg/100g)	2.9±0.7	2.9±0.7
Sum EPA+DHA (mg/100g)	<0.01	72±6
EPA (mg/100g)	<0.01	22±3
DHA (mg/100g)	<0.01	50±3
Fat (g/100g)	13.3±0.5	14.3±1.0

The content of several undesirable substances was also determined in the Atlantic salmon. The level of mercury was 22 μg/kg, and the level of dioxins and dioxin-like PCBs was 0.48 ng TEQ/kg; both are far below the EUs upper limits of 500 μg/kg and 6.5 ng TEQ/kg in fish, respectively. Taking into account the amount of salmon consumed per week during the weeks with the highest salmon intake, the intake of dioxin and dioxin-like PCBs per week represents 31% of the tolerable weekly intake (TWI) in a person weighing 100 kg (19). Persons with higher body weight will have a correspondingly lower percentage of TWI.

Statistical analysis

Changes in cognitive performance were analyzed by a twoway analysis of covariance (ANCOVA), 2(FG and CG) × 2(pre- and post-test), with age and IQ as continuous predictors (covariates). Because of the results gained from the ANCOVA, a dichotomization of the participants, independent of fish consumption, was made by the median split of age (median=42) and IQ (median=95). Thus, differences between <42 and >42 years of age and IQ <95 and >95 were investigated by t-tests for independent samples. To follow up Hansen et al. (8) participants were further divided into four groups based on their history of alcohol/drug abuse, i.e., Fish-alcohol/drug abuse, Fish-non-alcohol/drug abuse, Control-alcohol/drug abuse, and Control-non-alcohol/drug abuse. These analyzes were also conducted by ANCOVA, with age and IQ as covariates.

Results

Descriptive results

t–tests for independent samples showed no differences in age t(83)=0.45, p=0.65 (FG: 41.76±9.72; CG: 40.84±9.14) or IQ t(83)=0.10, p=0.92 (FG: 95.79±11.10; CG: 96.05±12.85). Preliminary correlations revealed no relationship between the two covariates, age and IQ (r=-.017, p=0.88).

Means and standard deviations for cognitive performance, i.e., accuracy and mRT on the n-back task for the CG and the FG are shown in Table 2.

Table 2.

Shows means and standard deviations for cognitive performance, i.e., accuracy and mean reaction time (mRT) on the n-back task for the Control and the Fish group. mRT is log transformed

	Control group			Fish group
	Pre-test	Post-test	N	Pre-test	Post-test	N
Accuracy
0-back	38.07±2.78	38.88±1.35	43	38.66±1.11	38.20±4.79	41
1-back	36.25±4.34	36.68±4.69	41	36.39±4.95	37.58±2.75	38
2-back	33.71±5.31	34.39±4.38	41	34.20±35.50	35.50±3.33	40
3-back	30.21±5.05	31.07±4.23	42	31.38±5.09	32.00±3.46	39
mRT
0-back	6.43±0.21	6.41±0.23	43	6.47±0.21	6.48±0.25	41
1-back	6.76±0.28	6.73±0.30	41	6.73±0.31	6.72±0.30	38
2-back	7.00±0.34	7.00±0.35	41	7.00±0.33	6.96±0.32	40
3-back	7.08±0.36	6.98±0.35	42	7.03±0.34	7.01±0.31	39

Number of correct responses on the WM task

The ANCOVA showed no significant effect of fish consumption on the 0-back task, controlling for age and IQ F(2,79)=1.27, p=0.29, ηp2=0.031. Nor were any effects of age or IQ found F(2,79)=0.47, p=0.63, ηp2=0.012 and F(2,79)=0.17, p=0.85, ηp2=0.004, respectively. For the 1-back task no effect of fish consumption was found, F(2,74)=0.56, p=0.57, ηp2=0.015. However, there was a significant effect of IQ F(2,74)=4.41, p=0.016, ηp2=0.106, but not age F(2,74)=2.27, p=0.11, ηp2=0.058. The ANCOVA for the 2-back task revealed no effect of fish consumption F(2,76)=1.19, p=0.31, ηp2=0.030. But, there was a significant effect of IQ F(2,76)=12.35, p=0.001, ηp2=0.245 and age F(2,76)=3.63, p=0.031, ηp2=0.087. Looking at the 3-back task, the results showed no effect of fish consumption, F(2,76)=1.01, p=0.37, ηp2=0.026. Again the results showed a relationship between performance on the n-back task and IQ, F(2,76)=11.203, p=0.001, ηp2=0.228. No effect of age was found, F(2,76)=1.06, p=0.35, ηp2=0.027.

Mean reaction time to correct responses on the WM task

For the mRT on the 0-back task the ANCOVA revealed no effect of fish consumption F(2,79) = 1.058, p = 0.35, ηp2 = 0.026. However, there was a significant effect of IQ F(2,79) = 10.88, p = 0.001, ηp2 = 0.216, and age F(2,79) = 15.45, p = 0.001, ηp2 = 0.281. The same pattern of results was found for the mRT on the 1-back task. No effect of fish consumption F(2,74) = 0.154, p = 0.86, ηp2 = 0.004, was found, but again there was a significant effect of IQ F(2,74) = 15.18, p = 0.001, ηp2 = 0.291 and age F(2,74) = 7.54, p = 0.001, ηp2 = 0.169. For the 2-back task there was no effect of fish consumption F(2,76) = 0.364, p = 0.70, ηp2 = 0.009 or IQ F(2,76) = 2.33, p = 0.10, ηp2 = 0.058. However, the results revealed a significant effect of age F(2,76) = 3.323, p = 0.041, ηp2 = 0.080. For the most difficult task, the 3-back task, the results revealed a similar pattern. No effect of fish F(2,76) = 0.667, p = 0.52, ηp2 = 0.017 or IQ F(2,76) = 1.182, p = 0.31, ηp2 = 0.030 was found. However, there was a significant effect of age, F(2,76) = 7.211, p = 0.001, ηp2 = 0.160.

t–test for independent samples further revealed significant differences between <42 and >42 years of age, and IQ <95 and >95, independent of fish consumption, see Table 3.

Table 3.

Shows differences between <42 and >42 years of age and IQ <95 and >95, in cognitive performance at pre-and post-test

Pre-test	Age<42	N	Age>42	N	t-value	p
Accuracy
0-back	38.00±2.79	41	38.70±1.19	43	-1.50	0.137
1-back	37.58±3.61	36	35.26±5.11	43	2.29	0.025*
2-back	35.84±3.04	38	32.28±7.21	43	2.83	0.005*
3-back	31.18±4.54	39	30.40±5.55	42	0.68	0.50
mRT
0-back	6.37±0.19	41	6.52±0.20	43	-3.37	0.001*
1-back	6.66±0.27	36	6.81±0.30	43	-2.30	0.024*
2-back	6.92±0.31	38	7.08±0.34	43	-2.20	0.031*
3-back	7.01±0.34	39	7.10±0.36	42	-1.19	0.235
Post-test
Accuracy
0-back	38.17±4.79	41	38.91±1.32	43	-0.97	0.33
1-back	37.89±2.19	36	36.47±4.80	43	1.64	0.10
2-back	35.92±3.11	38	34.07±4.35	43	2.18	0.03*
3-back	32.28±3.74	39	30.81±3.92	42	1.73	0.09
mRT
0-back	6.34±0.20	41	6.54±0.24	43	-4.11	0.001*
1-back	6.63±0.26	36	6.81±0.31	43	-2.68	0.009*
2-back	6.89±0.28	38	7.06±0.36	43	-2.34	0,021*
3-back	6.90±0.30	39	7.09±0.33	42	-2.81	0.006*
Pre-test	IQ<95	N	IQ>95	N	t-value	p
Accuracy
0-back	38.33±2.67	39	38.38±1.58	45	-0.09	0.925
1-back	35.81±4.98	36	36.74±4.30	43	-0.90	0.371
2-back	31.84±7.52	38	35.81±2.96	43	-3.19	0.002*
3-back	29.44±5.22	36	31.84±4.74	45	-2.16	0.033*
mRT
0-back	6.53±0.21	39	6.37±0.18		3.70	0.001*
1-back	6.85±0.29	36	6.66±0.27	43	3.15	0.002*
2-back	7.07±0.36	38	6.94±0.30	43	1.73	0.088
3-back	7.12±0.41	36	7.01±0.29	45	1.46	0.149
Post-test
Accuracy
0-back	38.10±4.96	39	38.93±1.10	45	-1.09	0.277
1-back	35.97±5.16	36	38.07±1.92	43	-2.47	0.016*
2-back	33.76±3.91	38	35.98±3.65	43	-2.64	0.010*
3-back	30.25±3.73	36	32.53±3.74	45	-2.73	0.008*
mRT
0-back	6.54±0.23	39	6.36±0.22	45	3.53	0.001*
1-back	6.86±0.30	36	6.62±0.25	43	3.95	0.000*
2-back	7.05±0.29	38	6.92±0.35	43	1.79	0.077
3-back	7.03±0.34	36	6.97±0.32	45	0.79	0.435

Looking at the four groups based on earlier alcohol/drug abuse, controlling for age and IQ, the ANCOVA revealed a pattern of results that was exactly the same as described for the FG and the CG.

Discussion

The present RCT revealed that fatty fish consumption did not affect WM performance in a group of young to middle aged adults with mental health problems. Age and IQ seemed to explain most of the variance.

For the easy 0-back task the ANCOVA showed that accurate performance was not associated with any of the covariates or fish consumption. Importantly, the 0-back task has often been used as a control condition or a non-executive functioning task, because it does not require manipulation of information in the WM (14). Looking at the three other sub-tasks, IQ seemed to explain most of the variance with regard to accurate performance. A distinction between the two easiest sub-tasks (i.e., 0- and 1-back) and the two most difficult sub-tasks (i.e., 2- and 3-back) can also be made because of the complexity and memory load associated with the two most difficult tasks (20). However, the 1-, 2-, and 3-back tasks require all on-line monitoring and updating of WM (14), but the 2- and 3-back tasks require on-line monitoring and updating in addition to manipulation of remembered information (14,20). Thus, regular fish consumption did not enhance WM in a group of young to middle aged adults.

No relationship between fish consumption and processing speed was found. Again the results seemed to be explained by age and/or IQ. For the 0- and 1-back task both age and IQ seemed to be of importance. For the two most difficult tasks only age seemed to explain processing speed. Age-related cognitive decline has been explained by decrements in WM capacity and processing speed. These declines have further been explained by gray and white matter loss as well as other biological changes in the brain such as cell and dopamine loss (see 21 for an overview). It has been assumed that vitamin D, one of the key nutrients found in fatty fish, is important for cognitive processing speed, especially in middle aged and older men (22), because vitamin D is important for brain myelination (23). Notably, the vitamin D status in the FG was close to the optimal level (for US population: >75 nmol/L) at post-test, while the CG had a suboptimal vitamin D status (55 nmol/L) (cf. 17). Also the level of omega-3 were significantly higher in the FG compared to the CG (cf. 17), and some studies have shown a relationship between omega-3 and cognitive functioning as well (see 24 for an overview). Especially the DHA has been shown to be very important for cognitive functioning. Stonehouse at el. (25) found that DHA supplementation improved RT to WM in a group of healthy males aged 18-45. As shown in Table 1 the diet with fish served in the present study contained a high level of DHA. Thus, the present study investigating WM, a pure mechanism of cognition did not find support for these arguments. Accounting for the history of earlier alcohol/drug abuse did not alter the results.

Importantly, the median age was 42, and independent of fish consumption the results revealed a trend towards a distinction between those who were <42 and >42 years old. The results suggest that even at the age of 42 there seems to be a decline in processing speed, but also in accuracy on the 2- and 1-back (see Table 3). Participants with IQ >95 had better performance than participants with IQ <95. This was especially true for the accuracy, but also for reaction time on the easier 0-and 1-back tasks. There was no relationship between IQ and reaction time on the more difficult 2- and 3-back tasks, which can be explained by a speed-accuracy trade off.

Thus, the improved performance on the TOH task found in Hansen et al. (8) does not seem to be related to a close relationship between fish consumption and WM as assumed. However, Hansen et al. (8) also found a relationship between fish consumption and improved performance on the Iowa Gambling Task (IGT), a complex decision-making task, which performance has shown to be intricately related to sleep. Sleepdeprivation has shown to decrease performance, while sleep intervention has been shown to improve performance on the IGT. The effects have been explained by brain activity during REM sleep and consolidation of memory (26). It should be mentioned that fatty fish consumption influenced sleep in Hansen et al. (17). Performance of both TOH and IGT used in Hansen et al. (8) relies on a set of executive components (i.e., WM, inhibition and set-shifting), understanding and codebreaking. Taken together these results suggest that fish interventions may improve performance on tasks depending on understanding/codebreaking and “a good night's sleep,” rather than pure cognitive mechanisms such as WM (associated with less impressionable factors, such as aging and intelligence). This hypothesis requires more investigation. Executive functioning, such as planning and decision-making, is crucial in order to adapt successfully to everyday demands (27, 28). To better understand the relationship between fish consumption and cognition, future research should also examine other core executive components such as inhibition and set-shifting.

There are some general limitations and issues related to dietary intervention studies, such as the impossibility of blinding. Moreover, there is always a risk of dropouts, especially if the intervention is long-term, and the participants may get tired of eating fish several times per week.

The present study, which to our knowledge is one of few RCT with fatty fish, has some important strengths. In order to minimize the blinding problems, both the FG and the CG received a meal, three times per week, which was different from what they usually got at the institution. When the FG got their fish meal, the CG was served an alternative meal. Importantly, both groups received some kind of intervention or “special treatment.” We also collected double portions to compare the composition of control and intervention diets (Table 1). The present study was conducted in an institution, with a similar environment for all the participants, and we had completely control of the consumption of the intervention and control meals. Thus, the present study has a high degree of compliance. Importantly, in the present study only one participant dropped out due to missing meals and two withdrew from the project. Thus, the overall dropout rate in the present study was insignificant.

Conclusion

The current study, a robust RCT investigating the relationship between regular fish consumption and WM, controlling for age and IQ, did not find any clear relationship between fish consumption and WM performance. However, further investigation focusing on the effect of fish consumption on other executive components (i.e., set shifting and inhibition) is needed. Since there is reason to believe that early cognitive decline can aggravate mental health conditions, more knowledge about the fish-cognition relationship will have important health and clinical implications.

Ethical Standards

The study was in compliances with the Helsinki Declaration and the US national regulations

Conflicts of Interest

The authors declare no conflict of interest.