*Norio Yasui-Furukori, MD, PhD, Department of Neuropsychiatry, Graduate School of Medicine, Hirosaki University, Hirosaki 036-8562, Japan. Email: firstname.lastname@example.org
Aims: A tree-drawing test acts as both a projective psychological examination as well as a supplementary psychodiagnostic tool. There is little information relating the characteristics of schizophrenia and the tree-drawing test. The present study compared the structural and morphological differences in the results of the tree-drawing test between schizophrenic patients and healthy individuals, as well as between schizophrenic patients who responded well to treatment and those who responded poorly.
Methods: The subjects included 202 chronic schizophrenic patients and 113 healthy individuals. The schizophrenic patients were categorized as ‘good responders’ or ‘poor responders’ based on their response to medical treatments. The tree-drawing test was performed on all subjects. The tree drawn by each subject was analyzed structurally and morphologically.
Results: There were significant differences between the trunk and branches drawn by schizophrenic patients and those drawn by healthy controls. There were no significant differences between the good responders and the poor responders in any aspect of the tree drawings. Multiple regression models showed that the ratio of the tree area to the total area of the drawing paper, the width of the trunk, the trunk base opening, and the size of the branch ends were significantly associated with schizophrenia.
Conclusion: The present study suggests that the trees drawn by schizophrenic patients are significantly different from those drawn by healthy individuals, but among schizophrenic patients, it is difficult to distinguish between good responders and poor responders using the tree-drawing test.
SCHIZOPHRENIA INVOLVES SEVERAL psychiatric syndromes including positive and negative symptoms, cognitive dysfunction, excitement, and anxiety–depression. These psychopathological symptoms are usually evaluated using the brief psychiatric rating scale (BPRS)1 or the positive and negative syndrome scale (PANSS).2 However, these assessments may be stressful for subjects, as patients are required to be interviewed for about an hour. In addition, the BPRS and the PANSS do not always cover all of the psychiatric syndromes of schizophrenia.
The Baum test (a tree-drawing test), a projective psychological examination and a supplementary psychodiagnostic tool, was systematized by Charles Koch in 1949.3 The Baum test has distinct advantages: first, the examination usually takes less than 30 min; second, because examinees tend to think that the examination is a test for drawing ability,4 they are more willing to accept the exam; third, the test is simple. Thus, the examination provides significant information to the examiner without as much stress as with other projective psychological examinations, such as the Rorschach test.
Several Japanese studies have been performed, analyzing the use of this examination technique in patients with schizophrenia.5–13 Previous studies analyzed the trees drawn based on impressions of the drawings, such as whether they were positive, negative or normal.14 However, there were a few studies that attempted to quantify the trees drawn based on structure or morphology, suggesting that the structural and/or morphological differences in trees drawn are factors by which schizophrenic individuals and healthy subjects can be distinguished.9,15,16 A recent study suggested that a larger trunk-to-crown ratio and trunk-end opening were characteristic of schizophrenic subjects, but there were also differences in these factors between paranoid type and non-paranoid type schizophrenic patients.12 However, this finding was inconclusive due to the small number of subjects (46 cases and 53 controls), and it was particularly problematic due to the variety of psychopathological symptoms associated with schizophrenia. In addition, few studies have determined whether or not the tree-drawing test predicts the response to treatment in schizophrenic patients.
The present study compared the morphological differences of trees drawn by healthy individuals and those drawn by schizophrenic patients to confirm the utility of the tree-drawing test (TDT) as a supplementary means of diagnosing schizophrenia. Additionally, the morphological differences of the drawn trees were compared between patients who responded poorly to treatments and those who responded well: an indicator of long-term prognoses.
A total of 202 schizophrenic patients who had been diagnosed according to the DSM-IV17 and who had received medication as outpatients (n = 63) or inpatients (n = 139) were recruited from Hirosaki-Aiseikai Hospital (n = 96), Huyoukai Hospital (n = 69), and Kuroishi-Akebono Hospital (n = 34). Additionally 113 age- and sex-matched healthy individuals working in medical institutions were recruited as controls. The mean (± SD) ages of the schizophrenic patients (114 men, 88 women) and the healthy individuals (66 men, 47 women) were 53 ± 11.7 and 44.6 ± 11, respectively. There were no significant differences between the healthy individuals and the schizophrenic patients in any factors. The mean (± SD) duration of illness, BPRS score, and chlorpromazine equivalents of the schizophrenic patients were 26.2 ± 14.0 years, 45.0 ± 19.9 points, and 1108 ± 725 mg/day, respectively.
The schizophrenic patients were divided into two groups according to their treatment response (good responders and poor responders) based on Brenner's Definition of Treatment Response Levels,18 which was defined by Brenner et al. for rating treatment refractoriness. The rating scale consists of a combination of the Clinical Global Impressions scale (CGI), the Brief Psychiatric Rating Scale (BPRS) and the Independent Living Skills Survey (ILSS). Levels 1 and 2, ‘remission’, indicate no need for a formal rehabilitation program. Levels 3 and 4 indicate responsiveness to a learning-based rehabilitation program. Level 5 suggests the need for a continuous, individually oriented strategy; for intensive trials of atypical or adjunctive antipsychotics; and for a team approach to case management. Levels 6 and 7 clearly indicate longer-term hospitalization with empirical trials of untried pharmacological or psychosocial treatments.18 All patients in this study had been hospitalized at least once due to acutely exacerbated psychosis with CGI scores of 5 or greater. Therefore, patients at level 3 (CGI = 3) after the treatment were regarded as good responders. We defined level 3 or lower as a good responder, and level 5 or higher as a poor responder. To emphasize the difference between good responders and poor responders, level 4, which is the level with the highest frequency in Brenner's Definition of Treatment Response Levels, was excluded. The number of patients at levels 1, 2, 3, 5, 6 and 7 were 22, 34, 56, 43, 41 and 6, respectively. The numbers of good responders and poor responders were 112 and 90, respectively. Table 1 shows the characteristics of the schizophrenic patients. There were no significant differences between the good responder group and the poor responder group with regard to age or duration of illness.
Table 1. Characteristics of patients with schizophrenia
Good responder (n = 112)
Poor responder (n = 90)
NS, not significant.
52.6 ± 11.5
53.5 ± 11.8
Duration of illness (years)
21.2 ± 14.5
30.6 ± 12.0
Brief psychiatric rating scale
28.4 ± 7.3
61.5 ± 13.9
P < 0.001
Chlorpromazine equivalents (mg/day)
542 ± 262
1604 ± 630
P < 0.001
The present study was carried out after obtaining approval from the Ethics Committee of Hirosaki University School of Medicine. After a full description of the study to the subjects, written informed consent was obtained.
The present study used the TDT according to Koch's original method.3 The subjects were asked to draw ‘a single fruit-bearing tree’12 on a sheet of A4-sized drawing paper. Brenner's Definition of Treatment Response Levels and the TDT were rated on the same day.
Analysis of the drawn trees
The tree drawn by each subject was quantified and classified by both structural analysis and morphological analysis. As shown in Figure 1, the structural analyses of the drawn trees quantified the results in the following way: (A) ratio of trunk to crown [trunk length (mm)/whole length of tree (mm)] (B) ratio of left side to right side [left width of trunk (mm)/whole width of trunk (mm)], and (C) tree size relative to page space [whole length of tree (mm) × whole width of tree (mm)/paper size (mm)]. These three points approximately represent the whole structure of the drawn trees and were selected to easily understand the spatial and locational character of the tree.
The morphological classification of the drawn trees was performed based on the following analysis: (i) shape of the trunk (0 points if the trunk was a double line, 1 point if the trunk was a single line); (ii) width of the trunk (0 points if the trunk width was two fingers or more, 1 point if the trunk width was less than one finger, 2 points if the trunk was a single line); (iii) top-end of the trunk (0 points if the top of the trunk was closed, 1 point if the top of the trunk was open, 2 points if the trunk was a single line); (iv) base-end of the trunk (0 points if the base of the trunk was natural, 1 point if the base of the trunk was open or closed, 2 points if the trunk was a single line); (v) shape of the branches (0 points if the branches were double lines, 1 point if the branches were single lines); (vi) end of the branches (0 points if the branch ends were closed or hidden in the crown, 1 point if the branch ends were open, 2 points if the branches were a single line); (vii) branch crossing (0 points if the branches crossed diagonally, 1 point if the branches crossed at a right angle); (viii) root visibility (0 points if the underground roots were not visible, 1 point if the roots were visible). These eight points were selected to evaluate the detail and the development of the drawings. Figure 2 shows a few classification examples of drawn trees. One rater carried out the morphological classification in order to decrease inter-rater variability.
A t-test and anova were used to analyze differences in structure between the trees drawn by the schizophrenic and control groups, between good responders and poor responders, and between good responders, poor responders, and the control group. The study used the χ2-test and Fisher's exact test for frequency analyses of the morphological differences.
A logistic regression model using a stepwise procedure was used to identify differences between healthy individuals and schizophrenic patients in order to investigate the relationship between the characteristics of the drawn trees and existing schizophrenia. A logistic regression model using a forced entry was used to identify differences between good responder and poor responder patients to investigate the relationship between response and non-response to treatments when a significant association was not obtained using the logistic regression model using a stepwise selection. Independent variables included in the model were the three factors of the structural analysis and the eight factors of the morphological analysis. Furthermore, an examination of predictive value was performed, and a logistic regression equation using correlation factors was used.
Differences were assessed with two-tailed tests, with an alpha level of 0.05. All analyses were performed using spss 15.0J for Windows (spss Japan, Tokyo, Japan).
Table 2 shows the frequency of the structural and morphological factors among good responders, poor responders and the healthy controls. The areas of the trees drawn by the schizophrenic group were significantly smaller than those of the healthy control group. There were significant differences in the frequency of several morphological factors between the schizophrenic and healthy control groups. The trees drawn by the schizophrenic group were more likely to have a single line trunk, a narrow trunk width, a trunk with an open top, an unnatural expression of the base of the trunk, single line branches, and right angle crossings of branches and the trunk. No differences in roots were found between the schizophrenic and control groups.
Table 2. Three factors of the structural analysis and eight factors of the morphological analysis
n = 113
GR vs PR
GR vs PR vs C
Total vs control
n = 112
n = 90
n = 202
Trunk length (mm) / whole length of tree (mm).
‡ Left width of trunk (mm) / whole width of trunk (mm).
Whole length of tree (mm) × whole width of tree (mm)/ paper size (mm).
χ2 was used.
†† Fisher's exact test was used.
C, control; GR, good responders; NS, not significant; PR, poor responders.
Table 3 shows the results of the logistic regression analysis with stepwise selection of the healthy individuals and the schizophrenic patients. There were significant differences between the healthy individuals and the schizophrenic patients in the combination of four factors (the percentage of tree area on the drawing paper, the trunk width, the trunk base-end, and the branch ends).
Table 3. Wald statistic in logistic regression analysis to predict schizophrenia
Whole length of tree (mm) × whole width of tree (mm) / paper size (mm).
0 points if the trunk width was two fingers or more, 1 point if the trunk width was less than one finger, 2 points if the trunk was a single line.
0 points if the base of the trunk was natural, 1 point if the base of the trunk was open or closed, 2 points if the trunk was a single line.
0 points if the branch ends were closed or hidden in the crown, 1 point if the branch ends were open, 2 points if the branches were a single line.
There were no differences between the good and poor responders in any of the tree drawing factors (Table 2). Additionally, no differences were found in any of the factors among patients at level 2 or lower, defined as in ‘remission’ in Brenner's Definition of Treatment Response Levels, or among patients at level 5 or higher (data not shown). In addition, there were no significant differences in any of the factors between the good responder and poor responder groups even as a result of logistic regression analysis with forced entry, as shown in Table 4. There were no correlations between any of the BPRS scores and any of the TDT morphological factors.
Table 4. Wald statistic in logistic regression analysis to predict response to treatment
The results of this study show that the ratio of the tree area to the drawing paper area, the trunk width, the base-end of the trunk, and the branch ends in the drawn trees are considered to be associated with schizophrenia. The results suggest that the trees drawn by schizophrenic patients have the following characteristics: the size is small, the trunk is a single line or narrow and is open at the base, and the branches are single lines or open at the end. These characteristics may reflect the fact that schizophrenic patients tend to be socially withdrawn, they tend to show a reduced ability in reality testing, and they are often aggressive and impulsive.4
While there were several significant differences between the trees drawn by the schizophrenic and healthy control groups, no differences were found between the good responder and poor responder groups within the larger schizophrenic group. Furthermore, the trees drawn by remitters according to Brenner's Definition of Treatment Response Levels were not different from those drawn by severely resistant and refractory schizophrenia patients. This shows that even those whose positive and negative symptoms had improved or remitted according to clinical assessment tended to draw severely disordered trees, similar to those drawn by the poor responder group, suggesting that the tree drawings may not be a ‘state marker’ but possibly a ‘trait marker’ for schizophrenia.
Several studies have compared the characteristics of the tree-drawing test between healthy individuals and schizophrenic patients.9,16,19,20 It was not easy to compare the results between this study and those previous studies because of different analytical methods used, however Taniguchi et al.19 reported that the size of the tree drawn is influenced by aging.
The present study was the first study to compare tree-drawing test results when distinguishing responses to treatments in schizophrenic patients using a clear definition. However, it was found that the tree drawings of good responders were not significantly different from those of poor responders, suggesting that the tree-drawing test can evaluate impairments that ordinary clinical assessments are unable to reflect.
Inadomi et al.12 reported that there were significant differences between paranoid schizophrenics and non-paranoid schizophrenics in the results of a tree-drawing test. In that study, the ratio of trunk to crown in non-paranoid schizophrenics was higher than that in paranoid schizophrenics and healthy individuals, which was not in line with our findings. The end of the trunk in paranoid schizophrenic tree drawings tended to be open, whereas it was not open in trees drawn by non-paranoid schizophrenic or control subjects, which is partially in accordance with our results. However, our results suggest that characteristics other than those related to the findings described by Inadomi et al.12 are also important keys in distinguishing between schizophrenic and healthy subjects.
In patients with schizophrenia, neurocognitive deficits have been associated with long-term outcome. On average, these patients perform one to two standard deviations below healthy individuals on neurocognitive measures, such as those assessing attention, executive function, memory, and processing speed.21,22 Moreover, many studies have reported that cognitive functions of schizophrenic patients remain impaired after the improvement of positive or negative symptoms.23,24 Therefore, we hypothesized that the tree-drawing test enables a psychopathology, such as a cognitive dysfunction, to be expressed, even when positive and negative symptoms have improved as evaluated by ordinary clinical assessment. Although patients had been evaluated as good responders by clinical assessment, the trees drawn by the good responders were severely disordered because the cognitive function of schizophrenia remains impaired.
There are several limitations to the present study. This study was a retrospective study. The assessment point was only one point, and the symptoms were not pursued longitudinally. Morita et al.20 reported that a growth index can change through even the short-term clinical course of schizophrenia. Inpatients and outpatients were mixed, although cognitive function in outpatients may be better than that of inpatients. Therefore, further study will be needed in which the symptoms of schizophrenia are monitored, inpatients are separated from outpatients, and the prospective method is addressed. We were not able to present the inter-rater reliabilities because the morphological classification was performed by one rater to eliminate inter-rater variability. Further investigation on the inter-rater reliabilities and variability are required. The present study compared only schizophrenic and healthy individuals; however, examiners of the tree-drawing test are required to discriminate not only between schizophrenia and health, but also between schizophrenia and other disorders and between other disorders and health. Further investigations that apply the present method will be required to discriminate schizophrenia from other disorders and to discriminate other disorders from healthy individuals.
In conclusion, the present study suggests that the trees drawn by schizophrenic patients, for example, the ratio of the tree area to the total area of the drawing paper, the width of the trunk, the opening of the trunk base, and the size of the branch ends, are significantly different from those drawn by healthy individuals, however it is difficult to distinguish between good responders and poor responders using the tree-drawing test.