RESEARCH
Personality characteristics in the patients of obsessive compulsive disorder
Krishan Kumar, Rajneesh Kumar1, Gurpreet Kaur2, Brahmdeep Sindhu3, Sachin4
Clinical Psychologist, State Institute of Mental Helath, Pandit Bhagwat Dayal Sharma Post Graduate Institute of Medical Sciences, Rohtak, Haryana, India
1Neurologist, Paras Hospital, Gurgaon, Haryana, India
2Scientist ‘C’, Defence Institute of Psychological Research, Defence Research & Development Organisation, Delhi, India
3Senior Psychiatrist, General Hospital, Gurgaon, Haryana, India
4Research Scholar, University of Rajasthan, Jaipur, Rajasthan, India
Abstract
Background The relationship between obsessional personality traits and obsessive compulsive disorder (OCD) has long been the subject of debate. Although clinicians have asserted for nearly a hundred years that such a relationship exists, empirical investigations have failed to provide consistent support; however, none of these empirical investigations have undertaken analyses that control for the effect of mood variables. Employing a non-clinical sample, some psychologists found that when mood variables are taken into account, a unique relationship between obsessional traits and obsessional symptoms emerges.
Material and methods A replication was undertaken on a large group of individuals with OCD.
Results After the effects of depression and anxiety were removed from a correlational analysis, obsessional symptoms were found to be significantly associated with obsessional and passive aggressive traits.
Conclusion OCD was not associated with any other grouping of traits as specified in the revised third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R) Axis II classification system.
Kumar K, Kumar R, Kaur G, Sindhu B, Sachin. Personality characteristics in the patients of obsessive compulsive disorder. Dysphrenia. 2012;3:144-8. Epub 2012 Feb 22.
Keywords: Obsessional symptoms. Obsessional traits. Passive aggressive traits. Depression. Anxiety. Doubting.
Correspondence: keshusony@rediffmail.com
Received on 24 January 2012. Accepted on 22 February 2012.
Introduction
The relationship between obsessive compulsive disorder (OCD) and various forms of the ‘obsessional personality’ has been under discussion for nearly a hundred years.[1] Opinions vary with respect to the exact nature of this relationship. Traditionally, the presence of a premorbid obsessional personality was thought to favour the development of obsessional symptoms;[2] however, more recently, it has been suggested that obsessional personality features may develop after the onset of OCD as part of a general coping response.[3] Theoretically, a mixture of these positions is also possible, whereby specific traits potentiate the emergence of particular symptoms and vice versa. There have been virtually a few researches examining the mechanisms that govern the relationship between obsessional traits and symptoms. This is largely because investigations have failed to move beyond repeated and unsatisfactory attempts at confirming that such a relationship exists at all.
The first detailed description of a personality type thought to favour the development of obsessive compulsive symptoms was provided by Janet[1] in his “Les Obsessions et la Psychaesthenie”. The “psychaesthenic state” was characterised by a broad range of personality features, many of which still appear in its contemporary successor, obsessive compulsive personality disorder (OCPD).[4] The full complement of psychaesthenic features included feelings of incompleteness, doubt, an inner sense of imperfection, a need for uniformity and order, pedantry, a restricted range of emotional experience, excessive cleanliness, poor thought control, and a fondness for collecting things.
Some five years after Janet’s publication, Freud[2] also endorsed the distinction between an obsessional personality and obsessional symptoms. For Freud, an ‘obsessional neurosis’ was more likely to arise in an individual showing features of the ‘anal erotic character’, which marked the coincidence of three traits: obstinacy, parsimony, and orderliness. Although the theoretical framework employed by Freud to explain the development of both the ‘anal erotic character’ and ‘obsessional neurosis’ is widely recognised as weak, all three features of the ‘anal erotic character’ have survived to be included in the diagnostic criteria for OCPD.
In the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV),[5] OCPD is described as “a pervasive pattern of preoccupation with orderliness, perfectionism, and mental and interpersonal control, at the expense of flexibility, openness, and efficiency”. The diagnostic criteria employed are very similar to those used in the tenth revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10)[6] for the anankastic personality. A meaningful distinction between OCPD and the anankastic personality cannot be made, as both sets of criteria are clearly designed to capture the same clinical entity.
Although the obsessional personality has been accepted as a clinical reality,[7] its relationship with OCD remains contentious. Some studies suggest that the relationship exists,[8] while others do not.[9] After reviewing the literature, Baer and Jenike[10] conclude that the majority of patients with OCD have at least one personality disorder; however, OCPD is in the minority, often occurring less frequently than mixed, dependent, avoidant, and histrionic types. This failure to find a consistent relationship between obsessional traits and symptoms is rather perplexing to clinicians, who, on the whole, have tended to favour the idea that OCD is associated with ‘a certain type’ of character, be that the ‘obstinate’ or ‘vacillating’ types described by Lewis[11] or the person of ‘strict conscience’ described by Rachman and Hodgson.[12] Moreover, principle component analyses of items included in measures of OCD tend to yield trait like factors. For example, Cooper and Kelleher[13] conducted several such analyses on the Leyton Obsessional Inventory (LOI).[14] Common to all analyses were three components or factors. Two of these reflected the cardinal features of OCD, namely, washing and checking; however, a third reflected a ‘feeling of incompleteness’. It is interesting to note that the incompleteness component of the LOI bears a strong resemblance to Janet’s[1] ‘sentiment d’incompletude’.
Rosen and Tallis[15] have suggested that previous investigations failed to find a specific relationship between OCPD and OCD because the effects of mood were not taken into account. Controlling for mood effects in the analysis of data is of great importance insofar as anxiety and depression may obscure the specific relationship between OCPD and OCD. Employing a non-clinical population, significant correlations were found between obsessional symptoms, as measured by the Maudsley Obsessional-Compulsive Inventory (MOCI),[16] and ten out of 13 personality disorders as measured by the Personality Diagnostic Questionnaire-Revised (PDQ-R);[17] however, once the effects of anxiety and depression were removed from the analysis, only one significant relationship remained. This was between OCPD and OCD.
The results reported by Rosen and Tallis[15] appear to support the view that OCPD and OCD are positively related; however, the data were collected from a non-clinical population. It is possible that the method of analysis employed, although capable of yielding promising results on a non-clinical population, would not yield similar results when applied to a large clinical group.
Method
The study was carried out during 2009-10 at various private hospitals in Delhi NCR, India.
Sample: Seventy seven individuals suffering from OCD were taken for the present study. Fifty-one were female and 26 were male. All had volunteered their names for entry onto a research data base; initial inclusion on this data base was determined if they had scored 11 or above on the MOCI.[16] A sub-sample of 46 respondents was randomly selected for interview. All but one met the revised third edition of DSM (DSM-III-R)[4] criteria for OCD. This suggests that 98% of the subjects participating in the present study would meet DSM-III-R criteria for a diagnosis of OCD. The mean age of the 77 respondents was 39 years (standard deviation, SD = 10.017). The minimum age was 18, while the maximum age was 66.
Instruments: PDQ-R[17] is a 151-item, self-administered, forced-choice; true/false diagnostic instrument measuring all DSM-III-R[4] Axis II personality disorders and also includes the provisional diagnoses of sadistic and self-defeating personality disorders. It yields aglobal PDQ-R score indicating overall personality pathology in addition to giving individual threshold scores on each of the Axis II criteria. Its criteria are identical to DSM-III-R, Axis II.
MOCI[16] is a 30-item, self-administered, forced-choice, true/false measure of obsessional symptoms. The MOCI yields a total, global obsessionality score in addition to scores on four subscales: checking, washing, slowness, and doubting. It is stable psychometrical tool[16] and has been used in a variety of studies to assess obsessional symptomatology in both clinical and non-clinical samples.[18,19]
Beck Depression Inventory (BDI)[20] was included as a measure of depression. It consists of 84 self-evaluative statements grouped into 21 categories that assess the cognitive, motivational and physiological symptoms of depression based on a one-week time frame. For each item there is a graded series of four alternative statements ranging from neutral to a maximum level of severity. The items are scored from zero to three so that the total BDI score range is from zero to 63. High test-retest values have been reported[20] and it has been found to have high levels of internal consistency.[21]
Spielberger State Trait Anxiety Inventory: Trait Scale, Form Y (STAI-Y)[22] was used as a measure of trait anxiety and consists of a 20-item self-report measure developed to assess the separate concepts of state and trait anxiety. These are scored on a four-point scale of intensity ranging from ‘almost never’ to ‘almost always’. The trait items are expected to be relatively stable and free from the influence of situational stresses. The STAI has been found to possess high internal consistency, and the trait form has demonstrated good test-retest reliability for intervals up to three months.[22]
Results
The means and SDs for the MOCI, MOCI subscales, the BDI and the STAI-Y are shown in table 1. All are in ranges suggesting clinical status. Pearson’s correlation coefficients were calculated for the data. All tests were two-tailed. Given the statistical problems associated with analyses employing multiple correlations, only those relationships significant at p < 0.01 and p < 0.001 are reported. MOCI total scores were significantly correlated with PDQ-R total scores and scores for nine out of 13 personality disorders (antisocial, avoidant, borderline, histrionic, narcissistic, obsessive compulsive, passive aggressive, sadistic, and self-defeating). The washing subscale was significantly correlated with PDQ-R total scores, and scores for borderline, histrionic, obsessive compulsive, passive aggressive and self-defeating personality traits. The doubting subscale was significantly correlated with PDQ-R total scores, and scores for avoidant, borderline, dependent, histrionic, narcissistic, obsessive compulsive, passive aggressive and self-defeating personality traits. The checking subscale was significantly correlated with PDQ-R total scores, and scores for borderline, histrionic, narcissistic, obsessive compulsive, passive aggressive, sadistic and self-defeating personality traits. All of these relationships are shown in table 2. When partial correlations were calculated, controlling for the effects of depression and anxiety (as measured by the BDI and the STAI-Y), most of these significant relationships disappeared. The MOCI total scores remained significantly correlated with PDQ-R total scores (r = 0.288, P < 0.01) and scores for obsessive compulsive (r = 0.320, P < 0.01) and passive aggressive personality traits (r = 0.302, P < 0.01) only. Doubting subscale scores remained significantly correlated with PDQ-R total scores (r = 0.321, P < 0.01) and scores for obsessive compulsive (r = 0.413, P < 0.01), passive aggressive (r = 0.471, P < 0.01) and self-defeating (r = 0.301, P < 0.01) personality traits. These relationships are shown in table 3.
Table 1. Mean and standard deviations (in parentheses) for the MOCI, MOCI subscales, BDI and the STAI-Y
|
MOCI |
Total |
14.13 (5.59) |
|
Slowness |
2.83 (1.34) |
|
|
Washing |
3.82 (2.74) |
|
|
Doubting |
4.57 (1.84) |
|
|
Checking |
4.69 (2.27) |
|
|
BDI |
|
17.39 (11.43) |
|
STAI-Y |
|
54.04 (12.85) |
Table 2. MOCI and personality disorder bivariate correlation coefficients
|
Personality Disorder |
MOCI total |
Slowness |
Washing |
Doubting |
Checking |
|
PDQ-R total |
0.568** |
0.144 |
0.316* |
0.589** |
0.409** |
|
Antisocial |
0.288* |
0.148 |
0.158 |
0.191 |
0.259 |
|
Avoidant |
0.341* |
-0.081 |
0.241 |
0.375* |
0.232 |
|
Borderline |
0.464** |
0.235 |
0.273* |
0.415** |
0.332* |
|
Dependent |
0.260 |
0.025 |
0.185 |
0.325* |
0.108 |
|
Histrionic |
0.446** |
0.152 |
0.281* |
0.401** |
0.290* |
|
Narcissistic |
0.350* |
-0.044 |
0.197 |
0.338* |
0.275* |
|
Obsessive Compulsive |
0.514** |
0.192 |
0.330* |
0.591** |
0.348* |
|
Paranoid |
0.169 |
-0.031 |
-0.031 |
0.168 |
0.235 |
|
Passive Aggressive |
0.491** |
0.213 |
0.266* |
0.611** |
0.336* |
|
Sadistic |
0.317* |
0.034 |
0.199 |
0.222 |
0.276* |
|
Schizoid |
0.108 |
-0.096 |
0.033 |
0.237 |
0.007 |
|
Schizotypal |
0.258 |
0.062 |
0.258 |
0.243 |
0.220 |
|
Self-defeating |
0.477** |
0.167 |
0.477** |
0.552** |
0.339* |
*P < 0.01, **P < 0.001
Table 3. MOCI and personality disorder correlation coefficients with BDI and STAI partial out
|
Personality disorder |
MOCI total |
Slowness |
Washing |
Doubting |
Checking |
|
PDQ-R total |
0.288* |
0.066 |
0.152 |
0.321* |
|
|
Antisocial |
0.195 |
0.149 |
0.097 |
0.045 |
0.180 |
|
Avoidant |
0.047 |
-0.195 |
0.102 |
0.093 |
-0.013 |
|
Borderline |
0.227 |
0.156 |
0.136 |
0.179 |
0.135 |
|
Dependent |
0.009 |
-0.004 |
0.055 |
0.075 |
-0.091 |
|
Histrionic |
0.252 |
0.111 |
0.167 |
0.169 |
0.120 |
|
Narcissistic |
0.102 |
-0.109 |
0.054 |
0.067 |
0.102 |
|
Obsessive compulsive |
0.320* |
0.176 |
0.216 |
0.413** |
0.170 |
|
Paranoid |
0.042 |
-0.069 |
-0.049 |
0.032 |
0.165 |
|
Passive aggressive |
0.302* |
0.184 |
0.149 |
0.471** |
0.155 |
|
Sadistic |
0.153 |
-0.020 |
0.106 |
0.039 |
0.146 |
|
Schizoid |
-0.162 |
-0.216 |
-0.101 |
0.032 |
-0.210 |
|
Schizotypal |
0.109 |
0.003 |
-0.005 |
0.109 |
0.094 |
|
Self-defeating |
0.202 |
0.098 |
0.060 |
0.307* |
0.120 |
*P < 0.01, **P < 0.001
Discussion
The present study demonstrated that obsessional symptoms, as measured by the MOCI, are associated with a wide range of personality traits, as measured by the PDQ-R. Significant associations were found between the MOCI and traits subsumed under the following DSM-III-R Axis II headings: antisocial, avoidant, borderline, histrionic, narcissistic, obsessive compulsive, passive aggressive, sadistic, and self-defeating. When subsequent analyses were undertaken controlling for the effects of depression and anxiety, only two significant relationships were preserved. That is between the MOCI and obsessive compulsive personality traits and between the MOCI and passive aggressive personality traits. The doubting subscale of the MOCI accounted for much of the relationship observed between MOCI total and PDQ-R total scores. The doubting subscale was also significantly correlated with self-defeating personality traits after partialling out the effects of anxiety and depression.
In an initial investigation employing a non-clinical population,[15] the same method of analysis demonstrated a significant association between obsessional symptoms and obsessive compulsive personality traits only. This suggests that increasing severity of obsessional symptoms captures a specific association with passive aggressive characteristics. The relationship between obsessional symptoms and passive aggressive personality traits will be considered before the main findings of the present study are discussed in detail.
It should be noted that Passive Aggressive Personality Disorder (PAPD) is no longer listed in the Axis II section of the DSM-IV. Evidence in favour of the construct was considered insufficient to warrant inclusion as an official diagnostic category; however, PAPD may appear in future editions of the DSM under a new heading, i.e. ‘Negativistic Personality Disorder’. Given that PAPD has been excluded from the current DSM and that the PDQ-R was based on the DSM-III-R system, DSM-III-R will be used as a basis for the present discussion.
PAPD is described as: “A pervasive pattern of passive resistance to demands for adequate social and occupational performance.” Although individual criteria show some overlap with obsessional traits and symptoms (e.g. procrastination and slowness), the overlap is no more significant than can be found in several other personality disorders. For example, sensitivity to criticism is found in Avoidant Personality Disorder[23] and is also strongly associated with compulsive checking.[24]
Overlapping criteria may be less relevant than the beliefs that are shared by individuals with OCD and PAPD. Central beliefs associated with PAPD include: “being controlled or dominated by others is intolerable”, “I have to do things my own way”, and “it is best not to express my anger directly but to show my displeasure by not conforming”.[23] These reflect several common features found in OCD, namely, preoccupation with control,[25] rigidity[26] and difficulty expressing anger.[27,28] It is interesting that the inadequate occupational performance described in PAPD is also a feature of OCD. For example, Rachman[27] suggests that avoidance of excessive responsibility in individuals with OCD often leads to underachievement at work.
Although some beliefs associated with PAPD may be relevant to OCD, the passive aggressive style of social interaction may represent a coping response. As such, traits associated with PAPD may develop as a consequence of existing obsessional problems. Many patients with OCD suffer a unique conflict that arises between rigidity and difficulty expressing anger. If an individual with OCD insists that significant others participate in certain rituals, and anger cannot be employed to overcome resistance, then alternative strategies to gain compliance must be adopted. A passive aggressive attitude may provide the means of encouraging collusion without violating the self-imposed prohibition of the expression of anger. Shafran et al.[29] found that 60% of relatives are involved to some extent with the rituals of an affected family member. This suggests that individuals with OCD are able to develop sophisticated and effective methods of manipulating those who share their domestic environment.
The main finding of the present study suggests that obsessional traits and symptoms are associated, consistent with some observations recorded in the clinical literature; however it should be noted that the relationship between OCD and OCPD appears to be largely attributable to scores on the doubting subscale of the MOCI. This has important implications with respect to how the concept of the obsessional personality is interpreted.
There is certainly some overlap between the items included in the doubting subscale of the MOCI and the obsessional trait items included in the PDQ-R, particularly those that appear to reflect constructs such as perfectionism and moral sensitivity. Nevertheless, the vast majority of items do not overlap. There are no specific ‘doubt’ items included in the PDQ-R, and the doubting subscale of the MOCI does not include any specific items that reflect rigidity, dedication to work, indecisiveness, and restricted range of affect, parsimony, and difficulty discarding possessions. In addition, although it might be argued that the doubt related items on the MOCI reflect a trait, Hodgson and Rachman[16] underscore the efforts made during the development of the instrument to exclude items that might reflect traits. As such, all of the subscales of the MOCI should be regarded as measures of symptoms or complaints.
In sum, the main finding from the present study suggests that a relationship exists between a subset of obsessive compulsive symptoms and the personality traits that constitute OCPD. The doubting subscale of the MOCI includes items such as “I have a strict conscience”, “I tend to get behind in my work because I repeat things over and over again”, and “I usually have serious doubts about the simple everyday things I do.” These suggest that the obsessional patient with OCPD traits is likely to exhibit a symptom cluster characterised by unwanted intrusive thoughts (particularly those that violate a strict moral code), repetition (particularly when accompanied by a sense of incompletion), and doubts about everyday actions. This is consistent with Rachman and Hodgson,[12] who, although questioning the validity of the obsessional personality, suggested that specific obsessional traits might be related to specific obsessional symptoms.
It is relatively easy to see how moral sensitivity will result in much of the ‘stream of consciousness’ being regarded as unwanted or intrusive. Similarly, a trait such as perfectionism will more than likely produce symptoms that reflect concern over detail; however, it is unclear why obsessional traits such as restricted affect or parsimony should be related to doubts about everyday actions or ‘repeating’. It seems plausible therefore, to suggest that the diagnostic criteria of OCPD conceal a set of core personality features that relate very closely to the symptom subset captured by the doubting subscale of the MOCI. This set of core personality features might form the basis of a reformulated obsessional personality type. Although there is insufficient evidence to infer all the elements of such a personality type, moral sensitivity could certainly be granted a central position.
The determinants of moral sensitivity may be closely associated with elements of the obsessional belief system already under investigation, most notably, exaggerated responsibility, thought action fusion, and extended personal influence.[27,30,31] If the core features of a reformulated obsessional personality could be understood in terms of underlying beliefs, then such beliefs might serve as legitimate targets for modification in therapy. As such, the treatment of patients exhibiting doubt related symptoms might routinely include specific ‘schema focused’ components.
Further reading
Oodun RN, Tripathi MN, Yadav JS. Clinical profile of “schizo-obsessive” disorder: a case report. Dysphrenia. 2012;3:190-1.
References
1. Janet P. Les Obsessions et la Psychaesthenie. Paris: Alcan; 1903.
2. Freud S. Character and anal eroticism. In Collected Papers, 2. London: Hogarth; 1908.
3. Berg CZ, Rapoport JL, Whitaker A, Davies M, Leonard H, Swedo SE, et al. Childhood obsessive compulsive disorder: a two-year prospective follow-up of a community sample. J Am Acad Child Adolesc Psychiatry. 1989;28:528-33.
4. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed, rev. Washington, DC: American Psychiatric Association; 1987.
5. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.
6. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva: World Health Organization; 1992.
7. Pollak JM. Obsessive--compulsive personality: a review. Psychol Bull. 1979;86:225-41.
8. Black A. The natural history of obsessional neurosis. In Beech HR, editor. Obsessional States. London: Methuen; 1974.
9. Steketee GS. Personality trait and disorders in obsessive--compulsives. J Anxiety Disord. 1990;4:351-64.
10. Baer L, Jenike MA. Personality disorders in obsessive compulsive disorder. Psychiatr Clin North Am. 1992;15:803-12.
11. Lewis A. Problems of Obsessional Illness. Proc R Soc Med. 1936;29:325-36.
12. Rachman S, Hodgson R. Obsessions and compulsions. Englewood Cliffs, NJ: Prentice-Hall; 1980.
13. Cooper J, Kelleher M. The Leyton Obsessional Inventory: a principal components analysis on normal subjects. Psychol Med. 1973;3:204-8.
14. Cooper J. The Leyton obsessional inventory. Psychol Med. 1970;1:48-64.
15. Rosen KV, Tallis F. Investigation into the relationship between personality traits and OCD. Behav Res Ther. 1995;33:445-50.
16. Hodgson RJ, Rachman S. Obsessional-compulsive complaints. Behav Res Ther. 1977;15:389-95.
17. Hyler SE, Rieder RO. PDQ-R: Personality Diagnostic Questionnaire-Revised. New York: New York State Psychiatric Institute; 1987.
18. Sher KJ, Frost RO, Kushner M, Crews TM, Alexander JE. Memory deficits in compulsive checkers: replication and extension in a clinical sample. Behav Res Ther. 1989;27:65-9.
19. Salkovskis PM, Harrison J. Abnormal and normal obsessions--a replication. Behav Res Ther. 1984;22:549-52.
20. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry. 1961;4:561-71.
21. Strober M, Green J, Carlson G. Utility of the Beck Depression Inventory with psychiatrically hospitalized adolescents. J Consult Clin Psychol. 1981;49:482-3.
22. Spielberger CD, Goruch RL, Lushene R, Vagg PR, Jacobs GA. Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press; 1980.
23. Beck AT, Freeman A, Pretzer J, Davis DD, Fleming B, Ottaviani R, et al. Cognitive therapy of personality disorders. New York: Guilford Press; 1990.
24. Rachman S. Obsessional-compulsive checking. Behav Res Ther. 1976;14:269-77.
25. McFall ME, Wollersheim JP. Obsessive-compulsive neurosis: a cognitive-behavioural formulation and approach to treatment. Cognit Ther Res. 1979;3:333-48.
26. Guidano VF, Liotti G. Cognitive processes and emotional disorders: a structural approach to psychotherapy. New York: Guilford Press; 1983.
27. Rachman S. Obsessions, responsibility, and guilt. Behav Res Ther. 1993;31:149-54.
28. Tallis F. Obsessions, responsibility and guilt: two case reports suggesting a common and specific aetiology. Behav Res Ther. 1994;32:143-5.
29. Shafran R, Ralph J, Tallis F. Obsessive-compulsive symptoms and the family. Bull Menninger Clin. 1995;59:472-9.
30. Salkovskis PM. Obsessional-compulsive problems: a cognitive-behavioural analysis. Behav Res Ther. 1985;23:571-83.
31. Tallis F. Obsessive compulsive disorder. A cognitive and neuropsychological perspective. Chichester: Wiley; 1995.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.