SEARCH

SEARCH BY CITATION

Keywords:

  • harm avoidance;
  • major depression;
  • recurrence;
  • remission;
  • Temperament and Character Inventory

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Aim

Previous studies consistently reported increased harm avoidance (HA) assessed with the Temperament and Character Inventory (TCI) in patients with major depressive disorder (MDD). However, such findings may have been related with depression severity and number of depressive episodes. The aims of the present study were twofold: to examine TCI personality profile in remitted MDD (DSM-IV) patients and to compare TCI personality between MDD patients with single episode (SGL-MDD) and those with recurrent episodes (REC-MDD) in order to elucidate personality profile associated with recurrence.

Methods

TCI was administered to 86 outpatients with remitted SGL-MDD (12 male and 17 female patients; mean age 43.2 ± 12.1 years) and REC-MDD (26 male and 31 female patients; 40.3 ± 11.6 years), and 529 healthy controls (225 men and 304 women; 43.4 ± 15.5 years), matched for age, sex and education years. Logistic regression analyses were performed in which single/recurrent episodes of depression were the dependent variable and age, sex, age of onset, family history of psychiatric disease and TCI scores were entered as possible predictors.

Results

The remitted MDD patients had significantly higher scores on HA (P < 0.001) and lower scores on self-directedness (P < 0.001), compared with the controls. HA (P = 0.03), its subscore, fatigability (P = 0.03), and family history of psychiatric disease were found to be positive predictors for recurrence.

Conclusion

There are differences in personality profile between remitted MDD patients and controls, and between remitted REC-MDD and SGL-MDD patients, suggesting that they are trait markers. HA and fatigability might be useful to assess risk for recurrence of depression.

Personality profile in depression has been discussed in different cultural and racial backgrounds by various explanatory models and is considered to be important in clinical practice. According to Kraepelin, the premorbid personality of depressive illness contains scrupulosity and patience.[1] Boyce et al. developed the Interpersonal Sensitivity Measure (IPSM), finding hypersensitivity to interpersonal rejection as a personality feature in depression.[2] Accumulating evidence suggests that introversion and neuroticism, assessed with the Eysenck Personality Inventory (EPI) and the NEO-Five Factor Inventory (NEO-FFI), are global personality features in depression.[3-5]

The psychobiological model of personality was proposed[6] and the Temperament and Character Inventory (TCI) was developed by Cloninger et al.[7] TCI is composed of four temperaments (novelty seeking [NS], harm avoidance [HA], reward dependence [RD], and persistence [PS]) and three characters (self-directedness [SD], cooperativeness [CO], and self-transcendence [ST]). Temperament refers to a heritable bias to basic stimulus-response and is considered to be related with different monoamine neurotransmitter functioning in the brain system,[6] while character reflects cognitive processes and the self-concept, which develop during the lifespan.[7] The TCI has been widely used in researches to examine personality aspects of psychiatric disorders, such as schizophrenia,[8] bipolar disorder (BPD)[9] as well as unipolar depression. Our group previously reported differences in personality profile between patients with unipolar major depressive disorder (MDD) and those with bipolar II, suggesting the possible use of TCI to discriminate between the two disorders.[10] An important methodological issue is that personality profile assessed with a questionnaire is influenced by the state of the subject. Previous studies have consistently reported that patients with depression had higher HA scores than healthy controls;[11-14] however, high HA and low SD correlated with depression severity at the time of personality assessment.[10, 11, 13-17] In this context, TCI personality profile in depression could be explained by several proposed models, including trait-, state-effects, and scar model.[3] The scar model is that personality features change as consequences of depression experience and remain in remitted patients. Another issue is the possible effects of ethnicity and its culture on the relation between personality and the development of depression. To our knowledge, there is only one study that compared TCI personality between patients with remitted MDD and healthy controls in Asian countries, indicating that only HA was significantly different and higher in the former than in the latter group.[16]

Recurrence of depression is of great clinical importance. It is reported that 40–85% of depression is recurrent.[18, 19] Attempts have been made to predict recurrence in depressed patients from demographic, clinical and psychological factors.[20] Some studies found age of onset and family history of psychopathology to be associated with recurrence.[20-22] Personality features might be linked with recurrence of depression. Indeed, high neuroticism measured by the Brief Revised Eysenck Personality Questionnaire was associated with recurrent episodes.[23] To our knowledge, there is only one study that compared TCI personality between remitted depressed patients with single episode and those with recurrent episodes,[12] reporting that the latter group demonstrated higher HA as compared with the former group.

The aims of the present study were twofold: to examine TCI personality profile in remitted MDD patients, minimizing the possible effects of current mood state, and to compare TCI personality between MDD patients with single episode (SGL-MDD) and those with recurrent episodes (REC-MDD) in order to elucidate personality profile associated with recurrence.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Subjects

Subjects were 86 patients with remitted MDD and 529 healthy controls. The patients were recruited from the National Center of Neurology and Psychiatry (NCNP) Hospital. The hospital is located in the western part of metropolitan Tokyo. Patients were diagnosed according to the DSM-IV criteria,[24] based on clinical interviews and medical records by the consensus of at least two experienced psychiatrists. Depression severity was assessed by an experienced psychiatrist using the 17-item Hamilton Depression Rating Scale score (HDRS-17)[25] and only patients in remission (defined as HDRS-17 ≤ 7) were included. The patients with remitted MDD (38 male and 48 female patients) were divided into two groups, the SGL-MDD (12 male and 17 female patients), defined as a group of patients with only a single episode of MDD, and REC-MDD (26 male and 31 female patients), defined as a group of patients with two or more episodes of MDD. Healthy controls were recruited through advertisements in free local information magazines, by word of mouth and by announcements on our website. All healthy controls were screened by psychiatrists using the Japanese version of the Mini-International Neuropsychiatric Interview[26, 27] and additional unstructured interviews to confirm no past or current history of psychiatric illness or contact with psychiatric services. Participants were excluded from both the patient and control groups if they had a history of severe head injury or a serious physical disorder. All participants were biologically unrelated Japanese who resided in the western part of metropolitan Tokyo. Written informed consent was obtained from each subject for participation in the study. The present study was approved by the ethics committee of the NCNP.

Assessment of personality profile and family history

The Japanese version of the TCI[7] translated by Kijima et al.[28] was administered to all patients and controls. All patients were in remission when TCI was administered to them. The TCI is a self-report questionnaire composed of 240 items with yes or no answers. Regarding family history, information was obtained from patients about the presence or absence of any psychiatric disorders followed by the access to psychiatric service in the patients' first- and second-degree relatives.

Statistical analysis

spss 11.0 (spss Japan, Tokyo, Japan) was used to conduct all statistical tests. Continuous variables are expressed as mean ± SD. Statistical differences in categorical variables were evaluated by χ2-test or Fisher's exact test when appropriate. As normality or homogeneity of variance assumptions in distribution were often unmet, differences in continuous variables in demographic data and the TCI scores between: (i) patients with remitted MDD versus healthy controls; and (ii) the SGL-MDD group versus the REC-MDD group versus healthy controls, were examined by using the Mann–Whitney and Kruskal–Wallis tests, respectively. After comparing TCI scores among three groups by Kruskal–Wallis test, post-hoc pairwise comparisons with the Mann–Whitney test were performed. To examine the possible association between recurrence of depression and TCI personality profile, the forward stepwise logistic regression analyses were carried out based on a likelihood-ratio test with inclusion and exclusion criteria of P = 0.05 and P = 0.1, respectively. We performed logistic regression analyses in which single/recurrent episodes of depression were the dependent variable and age, sex, age of onset, family history of psychiatric disease and TCI scores were entered as possible predictors. Omnibus test was used to test the model fit for the logistic regression models. Nagelkerke R2 was used to estimate the percent of explained variance in the model. Statistical significance was set at two-tailed P < 0.05.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Demographic and clinical characteristics of patients and controls

The demographic and clinical characteristics of patients and controls are shown in Table 1. There were no significant differences in age, sex or education years among the three groups (i.e. REC-MDD, SGL-MDD and controls). There was no significant difference in mean HDRS-17 score between the REC-MDD and SGL-MDD groups. These two groups were similar in many clinical variables examined; however, age of onset was significantly earlier and the rate of positive family history of psychiatric disorder was significantly higher in the REC-MDD group than in the SGL-MDD group.

Table 1. Characteristics in patients with remitted major depressive disorder stratified by recurrent episodes and healthy controls
 Remitted major depressive disorderHealthy controls (n = 529)Mann–Whitney or χ2-testKruskal–Wallis test or χ2-test
Recurrent (n = 57)Single episode (n = 29)Total (n = 86)Single vs recurrent depressionTotal depression vs controlsSingle vs recurrent depression vs controls
  1. Bold font indicates statistical significance.

  2. HDRS-17, 17-item Hamilton Depression Rating Scale; IMIeq, imipramine-equivalent dose of antidepressants.

Demographics       
Sex (male/female)26/3112/1738/48225/304χ2(1) = 0.14χ2(1) = 0.083χ2(2) = 0.22
P = 0.82P = 0.82P = 0.90
Age (years)40.3 ± 11.643.2 ± 12.141.2 ± 11.843.4 ± 15.5U = 671U = 2155χ2(2) = 1.40
P = 0.16P = 0.43P = 0.50
Education (years)14.8 ± 2.215.2 ± 4.814.9 ± 3.315.3 ± 2.8U = 749U = 20509χ2(2) = 2.64
P = 0.46P = 0.14P = 0.27
Clinical features       
Age of onset (years)29.7 ± 12.236.9 ± 10.632.1 ± 12.1 U = 512  
P = 0.004
Family history of any psychiatric disorder (yes/no)27/304/2531/55 χ2(1) = 9.40  
P = 0.004
Comorbid dysthymic disorder (n)202 Fisher's exact test  
P = 0.55
Lifetime suicide attempt (yes/no)6/512/278/78 Fisher's exact test  
P = 0.71
Lifetime hospitalization to psychiatric ward (yes/no)12/453/2615/71 χ2(1) = 1.53  
P = 0.25
Lifetime electroconvulsive therapy (yes/no)2/550/292/84 Fisher's exact test  
P = 0.55
HDRS-17 score3.0 ± 2.72.5 ± 2.92.8 ± 2.8 U = 709  
P = 0.27
Medication status       
IMIeq (mg/day)45.8 ± 104.914.2 ± 45.435.1 ± 90.3 U = 707  
P = 0.12
Benzodiazepine use (n)241236 χ2(1) = 0.004  
P = 1.00
Lithium use (n)718 Fisher's exact test  
P = 0.26

Comparison in TCI scores between total patients with remitted MDD and healthy controls

The total remitted MDD patients showed a significantly higher mean score on HA (P < 0.001) and lower mean score on SD (P < 0.001) than the controls. Mean NS score in the remitted MDD patients was lower than that in the controls at a trend level (P = 0.077) (Table 2).

Table 2. Comparisons of TCI scores in patients with remitted major depressive disorder with single and recurrent episodes, and healthy controls
 Remitted patients with MDDHCAnalyses
REC (n = 57)SGL (n = 29)Total (n = 86)n = 529Mann–Whitney test (Total depression vs controls)Kruskal–Wallis testPost-hoc pairwise comparisons with Mann–Whitney test
  1. Bold font indicates statistical significance.

  2. HC, healthy controls; MDD, major depressive disorder; REC, recurrent MDD; SGL, single-episode MDD; TCI, Temperament and Character Inventory.

Temperament       
Novelty seeking19.0 ± 5.221.0 ± 4.119.7 ± 5.020.6 ± 4.4U = 20048χ2(2) = 6.8REC < HC
P = 0.077P = 0.034
Harm avoidance25.1 ± 6.621.3 ± 6.223.9 ± 6.717.8 ± 6.3U = 11804χ2(2) = 55.5REC, SGL > HC
P < 0.001P < 0.001REC > SGL
Reward dependence14.9 ± 3.514.7 ± 4.014.8 ± 3.715.3 ± 3.6U = 20947χ2(2) = 1.5 
P = 0.237P = 0.467
Persistence4.6 ± 1.93.8 ± 2.24.3 ± 2.04.4 ± 1.8U = 21653χ2(2) = 4.5 
P = 0.468P = 0.106
Character       
Self-directedness23.4 ± 8.625.5 ± 8.824.1 ± 8.729.2 ± 6.6U = 14925χ2(2) = 27.4REC, SGL < HC
P < 0.001P < 0.001
Cooperativeness27.8 ± 6.228.1 ± 5.327.9 ± 5.928.6 ± 5.1U = 21506χ2(2) = 0.7 
P = 0.416P = 0.715
Self-transcendence9.7 ± 4.810.5 ± 4.310.0 ± 4.610.5 ± 5.3U = 21808χ2(2) = 1.2 
P = 0.538P = 0.550

Comparison in TCI scores between remitted MDD patients with single episode and recurrence

TCI scores of the three groups (REC-MDD, SGL-MDD and controls) were compared using Kruskal–Wallis test and post-hoc Mann–Whitney test for pairwise comparisons (Table 2). Regarding NS, the REC-MDD group, but not SGL-MDD group demonstrated lower scores (U = 12012, P = 0.011) than the controls. Regarding HA, both REC-MDD and SGL-MDD groups had significantly higher scores than the controls (U = 6520, P < 0.001 and U = 5283, P = 0.005, respectively). Regarding SD, both REC-MDD and SGL-MDD groups had significantly lower scores than the controls (U = 9171, P < 0.001 and U = 5754, P = 0.023, respectively). When REC-MDD and SGL-MDD groups were compared, there was a significant difference in HA (U = 544, P = 0.010), but not in the remaining temperaments or characters.

Logistic regression analyses to find predictors of recurrence of major depression

Table 3 shows results of the forward stepwise logistic regression analyses to predict REC-MDD or SGL-MDD status. In model 1 in which age, sex, age of onset, family history and scores of seven TCI dimensions were entered as predictor variables, family history and HA were significantly positive predictors for being REC-MDD. Model 1 demonstrated an appropriate goodness of fit (Omnibus test χ2 = 15.2, P < 0.001; Nagelkerke R2 = 0.23). To examine if any subscales of HA dimension could be predictors for being REC-MDD, model 2 was analyzed where age, sex, age of onset, family history and four subscores of HA were entered as predictor variables. Model 2 showed that family history and fatigability (HA4) were significantly positive predictors for REC-MDD, and demonstrated an appropriate goodness of fit (Omnibus test χ2 = 15.2, P = 0.001; Nagelkerke R2 = 0.23).

Table 3. Final models of risk prediction of recurrence of depression from the scores of the Temperament and Character Inventory, in addition to age, sex, age of onset and family history of psychiatric disorder
VariableβSEOdds ratio (95%CI)P-value
  1. Model 1 – Logistic regression, including age, sex, age of onset, family history, NS, HA, RD, PS, SD, CO, ST as independent variables. Nagelkerke R2 = 0.23.

  2. Model 2 – Logistic regression, including age, sex, age of onset, family history, HA1, HA2, HA3, HA4 as independent variables.

  3. Nagelkerke R = 0.23.

  4. CI, confidence interval; CO, cooperativeness; HA, harm avoidance; HA1, anticipatory worry; HA2, fear of uncertainty; HA3, shyness; HA4, fatigability; NS, novelty seeking; PS, persistence; RD, reward dependence; SD, self-directedness; ST, self-transcendence.

Model 1    
Family history1.660.615.27 (1.58–17.5)0.007
Harm avoidance0.080.041.09 (1.01–1.17)0.03
Model 2    
Family history1.540.614.67 (1.40–15.5)0.01
Fatigability0.230.111.26 (1.02–1.56)0.03

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

The present study was conducted to compare personality in patients with remitted MDD and healthy controls as measured by TCI. All patients were in remission at the time of TCI assessment; therefore, the possible influence of mood and any other symptoms on TCI, which has been reported by previous researches,[12-15] could be minimized in the present study. Another aim of the current study was to examine the relation between TCI personality and recurrence of depression, comparing the remitted MDD patients with single episode and those with recurrent episodes. We found that there were significant differences in TCI personality dimension scores between the remitted patients with MDD and healthy controls, and between the remitted REC-MDD and SGL-MDD patients. Logistic regression analyses showed that HA and its subscore, fatigability (HA4), positively predicted REC-MDD.

An advantage of our study is that the sample size (n = 615 in total subjects, consisting of 86 patients and 529 healthy controls) was relatively large, compared to the precedent studies on TCI personality in remitted MDD patients (summarized in Table 4). In addition, total remitted MDD patients and control subjects were well matched for age, sex and education years, and these demographic variables and clinical characteristics, including depression severity, comorbidity and medication status, were also well matched for between the REC-MDD and SGL-MDD groups.

Table 4. Summary of scores of seven dimensions of the Temperament and Character Inventory in remitted depression
 CountrySample size (n)TemperamentCharacter
PatientsControlNSHARDPSSDCOST
  1. a

    Richter et al. reported the temperament and character inventory results of patients with depression just before discharge and healthy controls, dividing patients into two subgroups according to with or without comorbid anxiety disorders and according to with single or recurrent episodes.[12] Their data were transformed, referring to the mean, standard deviation and sample size described in the paper, to analyze the difference between total patients and healthy controls and we showed the results of the t-test with the transformed data.

  2. CO, cooperativeness; HA, harm avoidance; NS, novelty seeking; PS, persistence; RD, reward dependence; SD, self-directedness; ST, self-transcendence.

Richter et al., 2000[12]aGermany126126[DOWNWARDS ARROW][UPWARDS ARROW]  [DOWNWARDS ARROW] [DOWNWARDS ARROW]
Smith et al., 2005[29]UK5289 [UPWARDS ARROW]  [DOWNWARDS ARROW]  
Nery et al., 2009[13]USA1545    [DOWNWARDS ARROW]  
Takahashi et al., 2013[16]Japan31174 [UPWARDS ARROW]     
Present studyJapan86529 [UPWARDS ARROW]  [DOWNWARDS ARROW]  

Patients with remitted MDD were found to have significantly higher HA and lower SD compared to healthy controls. As shown in Table 4, all directions of TCI results in patients as compared to controls were consistent with the previous studies.[12, 13, 16, 29] All previous studies except for the smallest study[13] found higher HA in patients with remitted depression. Similarly, all but one[16] found lower SD in remitted patients. Overall, our finding is consistent with these previous studies regardless of ethnic differences. Several studies have found that higher HA and/or lower SD in patients with current depression compared to controls remained even after patients achieved remission.[11, 12] A sib-pair study showed the high heritability of HA and SD, reporting that both depressed probands and their never-depressed siblings had higher HA and lower SD than controls.[17] Furthermore, a prospective study showed that elevated levels of neuroticism assessed by the short form of the Brief Revised Eysenck Personality Questionnaire were predictive of future depressive episodes.[30] The neuroticism is correlated with HA in TCI,[31] which is reasonable as both personality traits reflect anxiety proneness and vulnerability to stressful situations. Taken together, our findings suggest that elevated HA and decreased SD may, at least in part, be trait markers associated with vulnerability to depression. However, our results may be explained in part by the scar model. A longitudinal study showed that neuroticism increased after experiencing an episode of depression, which provided evidence for the scar model,[32] although the other longitudinal study failed to support it.[4] There are few studies investigating TCI personality before, during and after the episode of depression. Further longitudinal prospective studies are warranted to elucidate state, trait or scarring effect on TCI personality in depression.

Many risk factors for the recurrence of depression, including not only clinical and biological factors[33] but also psychosocial and psychological factors, have been reported.[20] Accordingly, we hypothesized that there was an association between personality and recurrence. As compared to controls, TCI scores were significantly altered on two temperaments (low NS and high HA) and one character (low SD) in patients with REC-MDD and on one temperament (high HA) and one character (low SD) in those with SGL-MDD. Previous studies suggested that the combination of the two altered personalities, high HA and low SD, was specific to MDD,[14, 29] which was congruent with the present study. Early age of onset in chronic MDD was associated with recurrence.[20, 21] In addition, the high morbidity risk of depression in the relatives was associated with recurrent depression of probands.[20, 22] The logistic regression analyses with stepwise selection method to avoid multicollinearity showed that HA and its subscore, fatigability were found to be positively associated with REC-MDD as well as family history. The association between high HA and recurrence of depression is consistent with previous reports,[13] while to our knowledge, this is the first report that found an association between high fatigability and recurrence of depression. A high score on fatigability means feeling fatigue easily and not feeling vigorous, which can be linked plausibly with the scar model, that is, the result of experience of recurrent depression. Taken together, although a prospective study is needed to assess the value of personality profile predicting recurrence of MDD, the statistically significant risk indicators for recurrence of depression that we found, HA and fatigability, can be useful to assess risk for recurrence of depression in patients with first-episode depression and may be related with the underlying vulnerability of recurrent depression. In clinical settings, physicians should follow first-episode patients longer if the patient has high HA and fatigability as well as family history of any psychiatric disease.

Recurrence of depression is related with bipolarity.[34, 35] Therefore, it is possible that symptoms of BPD would appear in the patients in life later on, especially in the REC-MDD group. In precedent studies, patients with BPD demonstrated higher HA score, compared to controls.[10, 36] However, there is no study in which TCI personality traits were compared between BPD and the subtypes of MDD, such as the REC-MDD. Further personality research might clarify the relation between BPD and recurrence of depression.

The mean age of onset was relatively high in this study population, compared to the previous epidemiologic study.[37] It is reported that age of onset affected the comorbidity rate of personality disorder in depression,[38] while a study by Celikel et al. found that age of onset did not correlate with TCI scales.[14] To examine an influence of age of onset on the current results, we subdivided each of the REC-, and SGL-MDD groups into two subgroups using each mean age of onset as the cut-off point. There was no significant difference in HA score between the early-, and late-onset-REC-MDD groups or between the early-, and late-onset-SGL-MDD groups (data not shown). Although it cannot be entirely denied that late onset would influence the results of TCI in general, we would like to argue that at least the currently obtained difference between the REC- and SGL-MDD groups in HA score using age- and sex-matched samples, cannot solely be attributable to the relatively late age of onset.

The current study has several limitations. First, the sample size in the present study is not very large, which may cause type II errors. Second, the observed differences on TCI personality between the REC-MDD and SGL-MDD groups could be due to the effect of antidepressants. However, Gelfin et al. reported that antidepressant had no effect on personality characteristics,[39] although findings have been inconsistent.[40] Third, we chose the MDD patients with remission, regardless of the basal score of HDRS, that is, the initial severity of depression when they had been in a depressed state, which is reportedly a significant risk factor for non-remission.[41] Therefore, the remitted MDD patients recruited in the current study might be biased in terms of initial severity of depression. Further studies with a larger sample of drug-free and clinically remitted MDD patients, using additional variables, such as initial severities of depression and anxiety, may be needed to extend and confirm the present results. Fourth, due to the cross-sectional nature of the study, we could not conclude any causal relation between TCI personality profile and MDD. In addition, family history data was not obtained by any structured interview. However, our results clearly suggested the association between recurrence of depression and family history of any psychiatric disorders. Lastly, the majority of the healthy controls were volunteers who participated in this research of their own accord, implying that sampling bias, favoring, for example, exploration or outgoing, could exist.

In conclusion, we obtained evidence for elevated HA and decreased SD in remitted MDD when compared to healthy controls in Japanese subjects. Dividing remitted MDD patients into two subgroups according to presence or absence of recurrence, there were differences in personality profile when compared with each other and with controls. Furthermore, HA and its subscore, fatigability, were found to be positively associated with REC-MDD. Our findings may contribute to elucidating the underlying vulnerability of recurrence of depression and to early intervention and prevention of recurrence of depression by clinically using high HA and fatigability as risk indicators.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

This study was supported by an Intramural Research Grant for Neurological and Psychiatric Disorders of the NCNP, Health and Labour Sciences Research Grants, and ‘Understanding of molecular and environmental bases for brain health’ carried out under the Strategic Research Program for Brain Sciences by the Ministry of Education, Culture, Sports, Science and Technology of Japan. We thank Yuki Mitsuyama, Junko Iida, Manami Sakairi, Kentaro Yoda and Miku Fujita for assistance in the neuropsychological tests and recruitment of participants and all the volunteers for their participation. All authors declare that they have no conflicts of interest.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References