Factor structure and validity of Type D personality scale among Indian (Tamil-speaking) patients with acute myocardial infarction.

BACKGROUND: Type D personality has been linked to negative outcomes following acute myocardial infarction (AMI). Our objective was to determine the factor structure, reliability, and validity of the Type D personality construct among people with AMI in the Indian clinical setting.
MATERIALS AND METHODS: In a cross-sectional study conducted between August 2016 and July 2017, 200 consecutive patients with AMI admitted to a tertiary care center completed the Tamil version of the Type D scale-14 (DS-14). The Eysenck Personality Questionnaire Revised-Short Form (EPQR-S) was also administered to check the convergent/divergent validity.
RESULTS: The prevalence of Type D personality was 24%. Results of the principal components analysis revealed a four-factor solution for the Tamil version of the DS-14 which explained more than 75% of the variability. Strong convergent validity with the neuroticism subscale of EPQR-S (r = 0.84, P < 0.001) and divergent validity with the extraversion subscale (r = -0.75, P < 0.001) was noted. The two subscales of the Type D scale, negative affectivity and social inhibition (SI), demonstrated good reliability (Cronbach's alpha of 0.85 and 0.76, respectively). Dropping item no 14 from the DS-14 led to significant increase in internal consistency (Cronbach's alpha 0.81) for the SI subconstruct.
CONCLUSION: Among an Indian Tamil-speaking clinical sample of patients with AMI, Type D is a valid construct and can be assessed using the regional version of the DS-14 which showed a four-factor structure and good psychometric properties. Item no. 14 of the DS-14 scale may need modification for the Indian setting. Copyright:

INTRODUCTION

Ischemic heart disease is one of the leading causes of mortality and disease burden worldwide, measured by Disability-Adjusted Life Years lost.[1] Ethnically, South Asians are at an increased risk of coronary artery disease (CAD) due to a combination of lifestyle-related and pathophysiological factors.[2] Figures from India over the last four decades show a consistent increase in the prevalence of CAD.[34] Owing to its assuming epidemic proportions, a number of psychosocial risk factors, apart from the “traditional” risk factors, are being investigated to assist enhanced management.

In this regard, the role of personality factors in CAD has received significant research attention. The first links between personality and heart disease arose, nearly six decades ago,[5] from the concept of Type A personality and its purported role in the pathogenesis of CAD, based on classic cohort studies.[67] However, subsequent work on Type A personality showed contrasting results with relation to the incidence of CAD[89] as well as prognosis following acute coronary syndromes,[1011] across cultures and settings. These negative findings spawned a search for the “pathogenic” ingredients of Type A personality as well as exploration of other personality dispositions such as negative affectivity (NA).[12] The latter denotes a tendency to experience negative emotions such as depression, anxiety, and hostility-aggressiveness on a consistent basis.[13]

The Type D or distressed personality was first postulated by Denollet[14] to measure a stable tendency to jointly experience NA and social inhibition (SI). The construct, initially, borne out of evidence that linked negative emotions (such as anger-hostility)[1516] and reduced levels of perceived social support[1718] with increased risk of cardiac events, are now thought to be an independent predictor for development and prognosis of CAD.[19] However, there are negative studies too[20] and growing concerns about the conceptualization and cross-cultural validity of the Type D construct, just as for Type A personality.[21]

The 14-item Type D scale-14 (DS-14), a short and low burden measure was developed by Denollet[14] to assess the construct of Type D personality. The original version showed good psychometric properties and a two-factor structure which has since been replicated in the general population, as well as cardiac and noncardiac medical populations.[22] However, the literature is dominated by findings from Western settings and studies from the South Asian settings are restricted to China,[23] Korea,[24] Hong Kong,[25] and Taiwan.[26] Most of these studies have replicated the two-factor structure and confirmed the reliability and validity of the Type D construct. The Indian culture is inherently pluralistic with different religious, dietary, linguistic, and social interaction patterns, probably a relic of various ruling dynasties in its different parts.

Hence, we sought to examine the factor structure and validity of the Type D personality scale among Tamil-speaking patients with acute myocardial infarction (AMI) attending the services of a tertiary care center in South India. This will establish the utility of the DS-14 scale in the Indian setting and provide a background for further work on the role of Type D personality in the pathogenesis of CAD in our setting.

MATERIALS AND METHODS

Setting and participants

The study was conducted in collaboration between the departments of psychiatry and cardiology at a teaching cum tertiary care hospital in South India. The hospital has all the specialty and super-specialty departments functioning out of a single campus. Most patients who avail services belong to the low-income category and hail from the rural background with agriculture being the predominant vocation in the region.

A convenience sample of 200 consecutive participants aged between 18 and 65 years admitted with AMI (non-ST-elevation myocardial infarction and ST-elevation myocardial infarction) diagnosed by a cardiologist and admitted to the cardiac and coronary care unit of the hospital were recruited between August 2016 and July 2017. We excluded participants with psychosis and those who were unfamiliar with the local language (Tamil). Written informed consent in the vernacular language was obtained from each of the participants. The study protocol was approved by the Institutional Ethics Committee.

Instruments

We used a semi-structured proforma to collect relevant sociodemographic and clinical details. This proforma also covered all the known risk factors for CAD (such as smoking, diabetes, hypertension, dyslipidemia). For the diagnosis of metabolic syndrome, we used the International Diabetes Federation criteria.[27] In addition, all patients (n = 200) also completed the following instruments:

1. DS-14 – This is the most widely used self-administered instrument for measuring Type D personality.[14] It consists of 14 items, equally distributed between the two-component constructs, NA, and SI. Items are answered on a five-point Likert scale ranging from 0 to 4. Thus, the scale yields two subscale scores ranging from 0 to 28. The presence of Type D personality is defined by a cutoff score of ≥10 on both the subscales. Translation of the DS-14 questionnaire into the local language (Tamil) was carried out following the guidelines recommended by the World Health Organization for translating research tools[28] and involved forward and back-translation by bilingual experts, who also resolved any discrepancies in the process. A panel of five experts including three psychiatrists, a clinical psychologist, and a cardiologist examined the translated version for semantic, conceptual, and technical equivalence. The final Tamil version showed good consistency with the English version

Eysenck Personality Questionnaire Revised-Short Form (EPQR-S)[29]– This is a 48-item self-reported questionnaire, with equal items for the four subconstructs, namely neuroticism, psychoticism, extraversion, and lie. Each item has a binary response of “yes” or “no” and was scored 1 or 0. Each subscale score thus ranges from 0 to 12. For the study, the EPQR-S underwent forward translation to Tamil and back-translation to English by two different bilingual experts. Inconsistencies were resolved by the experts through mutual discussion till consensus. The translated version showed good internal consistency for the neuroticism (Cronbach's alpha 0.79) and extraversion subscale (Cronbach's alpha 0.77), which were used to establish convergent validity with the Type D scale.

To minimize patient burden, the interview was carried out in multiple sittings if desired by the patient.

Statistical analysis

Statistical Package for the Social Sciences Version 16.0 (SPSS for Windows, Version 16.0. Chicago, SPSS Inc.) was used for analysis. Continuous data were expressed as mean with standard deviation or median with interquartile range, whereas frequencies and percentages were used to express categorical variables. Correlation between symptom scores was assessed using Pearson's or Spearman's rank correlation depending on data distribution.

Factor analysis, using the extraction method of principal component analysis with varimax rotation, was conducted to assess the factor loadings of the different items of the DS-14. The sequence followed for factor analysis included the following steps: initial preparation of a correlation matrix to check whether the data could be factor analyzed or not, deciding the number of factors to be retained (based on visual inspection of the scree plot as well as selecting components with initial Eigenvalue more than 1) and rotating the factors by using varimax rotation to improve the interpretability of the factors[30] The Kaiser–Meyer–Olkin (KMO) measure and Bartlett's test of sphericity were used to assess the sampling adequacy for factor analysis. An item loading of more than 0.5 was considered as the threshold for loading on a factor. If any item had a loading of ≥0.5 or more than one factor, then it was assigned to the factor on which it had a higher loading.

For convergent validity, the correlation of Type D personality scale with Eysenck Scale Neuroticism was assessed, while for divergent validity, the correlation of Type D personality scale with Eysenck Scale Extraversion was assessed. For reliability, the Cronbach's alpha and the average inter-item correlation were computed for the two subconstructs of Type D scale separately.

RESULTS

Sample characteristics Table 1

The mean age of the sample (n = 200) was 55.0 (±12.1) years. Majority of the respondents were male (n = 142, 71.0%), had received formal schooling at or below the tenth grade (n = 168, 84.0%), and had a diagnosis of ST-segment elevation type myocardial infarction (n = 164, 82.0%). Comorbid diagnosis of diabetes mellitus was present in 63 patients (31.5%), while 59 (29.5%) were suffering from concurrent hypertension. Alcohol use was reported by 82 (41%) patients, while 93 (46.5%) were smokers. Type D personality and metabolic syndrome were positive in 48 (24%) and 125 (62.5%) patients, respectively.

Reliability, convergent, and divergent validity

Among the myocardial infarction patients, the internal consistency (Cronbach's alpha) was 0.85 for the NA subscale and 0.76 for the SI subscale. The average inter-item correlation was 0.63 for the NA and 0.47 for the SI subscale. For item no 14, belonging to the SI subscale, a negative correlation with other items on the subscale was noted. Dropping this item led to an increase in the Cronbach's alpha for SI subscale to 0.81. None of the other items, when dropped, increased the Cronbach's alpha value by >0.1.

A correlation coefficient of 0.844 (P < 0.001) suggested that Type D personality scale converged with construct of neuroticism subscale on the EPQR-S. A correlation coefficient of −0.746 (P < 0.001) suggested that Type D personality scale was opposite to the construct of extraversion.

Factor structure of type D scale

The initial correlation matrix showed that all the items had a minimum correlation of 0.3 with at least one other item. This and the value of the determinant (3.175) showed the factor analyzability of the matrix. The KMO measure of sample adequacy was 0.85 and the Bartlett's test of sphericity value (χ2) was 2004.20 (P < 0.001), suggesting sample adequacy. A four-factor solution was deemed as the best possible solution (one-factor solution did not have four items loaded on it, two-factor solution had one item with cross-loadings and two items which did not load on any factor, and three-factor solution had one item each with cross-loading and not loading on any factor). This model explained 75.6% of the variance. Figure 1 shows the Scree plot and the rotated component matrix of the items is shown in Table 2. Most of the items (1, 2, 3, 4, 5, 6, 7, and 9) loaded on the first factor. Three items (8, 10, and 13) loaded on the second factor. Two items (11 and 14) loaded on the third factor, while one item (12) loaded on the fourth factor.

Figure 1

Scree plot

Table 2

Rotated component matrix of the factor analysis of Type D personality scale

DS-14 items	Factor 1	Factor 2	Factor 3	Factor 4
Item 1	0.795	0.329	−0.204	0.092
Item 2	0.750	0.395	0.182	0.235
Item 3	0.851	−0.055	−0.183	0.016
Item 4	0.820	0.215	0.164	0.268
Item 5	0.771	0.441	0.215	−0.218
Item 6	0.790	0.373	0.253	0.113
Item 7	0.628	0.390	0.166	−0.416
Item 8	0.465	0.633	0.087	0.000
Item 9	0.718	0.088	0.081	0.192
Item 10	0.062	0.807	−0.296	0.179
Item 11	0.457	0.163	0.752	0.019
Item 12	0.282	0.026	0.109	0.866
Item 13	0.167	0.603	0.163	−0.118
Item 14	−0.168	−0.085	0.923	0.086

DS-14 – Type D scale-14

DISCUSSION

The study represents the first attempt to examine the factor structure and validity of Type D personality construct as well as its assessment with the Tamil version of the DS-14 scale in the Indian clinical setting. A four-factor structure was obtained for the Type D scale, as opposed to the traditional two-factor structure which has been replicated in Western,[31323334] Middle Eastern,[35] and the few available Southeast Asian studies.[242526] The scale demonstrated good psychometric properties such as acceptable internal consistency for both subscales as well as strong convergent/divergent validity. Thus, in Tamil-speaking Indian patients with AMI, the validity of the Type D construct is confirmed.

The Tamil version of the DS-14 appears to be a reliable and valid measure of Type D personality. The psychometric properties compare well with those reported for the DS-14 scale in Chinese, Taiwanese, Belgian, and Dutch samples.[14252633] The deletion of individual items on the scale did not lead to an increase in the internal consistency (Cronbach's alpha) by more than 0.1 for all except one of the items in the scale. This indicates good consistency between items in the respective subscales.

A notable exception to this was item no 14 (“When socializing, I don't find the right things to talk about”) which had a negative inter-item correlation with other items on the SI subscale finding. The Cronbach's alpha with this item deleted rose to 0.81 and the average inter-item correlation of the SI scale of Type D, with item-14 included, was a moderate 0.47. As such, it seems that this item may not be appropriate for the Indian populace. To put it in another way, not having the right things to talk about in a social setting may not be considered unacceptable in India. This may be attributed to cultural reasons. In the Indian culture, similar to other Asian “high context” cultures, it is customary not to talk with strangers and meaning in social interactions is often derived implicitly.[36] This may be the reason for failure of item no. 14 in the Indian setting. Similar to our findings, previous studies from the Asian setting[2426] have also found that certain items on the DS-14 were not doing well in their setting and may need modification.

It is known that exploration of psychological scales in different communities, ethnicities, or languages may result in a factor structure different from the original scale.[3738] In the present work, we could not replicate the original two-factor structure for Type D scale[14] which has subsequently been reproduced in different settings.[243135] Instead, we obtained a different four-factor structure for Type D construct. The four factors in our sample could be labeled as follows: Factor 1 – “NA/withdrawn,” Factor 2 – “reticent,” Factor 3 – “reclusive,” and Factor 4 – “worry.” Factor 1, which combined items related to the central concepts of Type D (namely, NA and SI) was responsible for maximum proportion of explained variance of the four-factor structure (47%). It extracted five items from the original Factor 1 (NA) as well as three items from the original factor 2 (SI).

Factor 2 (“reticence”) extracted three items, two from original Factor 2 (SI) and one from original Factor 1 (NA). Two items, both from original factor of SI loaded onto Factor 3 (“reclusive”). The fourth factor (“worry”) extracted one factor from original dimension of NA, which however loaded separately. These differences seem to imply that an average Type D Indian individual may combine or integrate gloominess, unhappiness, and irritability together with features of SI when faced with a social situation. Items that suggested “reticence” loaded separately and combined items from both original factors. This suggests that reticence is a separate quality that may be valued differently, not necessarily in a negative light, in the Indian culture.

Factors 3 (“reclusive”) and 4 (“worry”) each had items from one of the original factors only but were extracted separately. The probable reasons behind this may relate to acceptance of social reclusiveness as an individual right among Indians and worry as something that is different in meaning and context from the rest of the items. Our findings not only emphasize the importance of cross-cultural exploration of factor structures but also highlight the importance of culture and language nuances in evaluating psychological constructs.

The prevalence of Type D personality was 24% in our sample. This figure is comparable to that from Western setting where the prevalence quoted ranges from 25% to 36%[31394041] and non-Western settings where the corresponding figures are a tad lower and range from 20% to 31%.[242526] Thus, the Type D personality is broadly applicable cross-culturally.

Our study has certain implication for clinical practice and research. First, there is scope to modify the DS-14 scale in the Indian setting to enhance its utility. This may be taken up by psychosomatic researchers. Second, given the robust association between Type D personality and negative outcomes in CAD, the use of this instrument may identify a high-risk subgroup within Indian CAD patients. Third, given the unique cultural patterns of interaction in the Indian Society, more research is needed to outline the potential moderators in the relationship between Type D personality and CAD about which little non-Western data are available. Potential candidate variables in this context include perceived social support, medical comorbidities, and quality of life. Fourth, given that personality offers itself to modification, the role of preventive strategies aimed at modifying Type D personality attributes needs further examination in the Indian setting.

The present work has certain limitations. The study findings are drawn from a hospital-based sample of patients with AMI admitted to a tertiary care center, and therefore, generalization to other settings should be done with caution. Future studies should recruit both hospital- and community-based samples for better representativeness. The lack of an apparently healthy control group precludes any conclusions about the association between Type D personality and CAD in the Indian clinical setting. We did not perform a test–retest reliability, and therefore, this cannot be commented upon. Nevertheless, this is the first attempt at validation of the Type D personality construct in Indian culture. The sample size was sufficiently large for the factor analysis as indicated by the KMO coefficient. Further validation work should involve larger samples with diverse patient groups (hospital-based vs. community-based, with diabetes vs. without diabetes, etc.) to delineate the prevalence and role of Type D personality in this population.

CONCLUSION

The findings from the present study confirm and extend the validity and reliability of the Type D personality construct to an Indian (Tamil) sample of AMI patients. The Tamil version of the DS-14 is a reliable and valid measure to assess this personality construct. The prevalence of Type D in AMI was comparable with other Asian studies. Notably, the DS-14 scale appears to have four-factor structure and item no. 14 does not do well in the Indian culture. Being a pluralistic society, further replication of our findings from the Indian culture, particularly from diverse settings, is warranted. Given the robust links between Type D personality and a range of poor cardiac and health outcomes in CAD as well as the increased ethnic vulnerability of South Asians to this problem, further work from India is required to identify novel psychosocial preventive and treatment targets to optimize clinical outcomes in CAD.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

Table 1

Baseline characteristics of the sample (n=200)

Variable	Mean±SD or frequency (%)
Age	55.0±12.1
Gender
Male	142 (71.0)
Female	58 (29.0)
Education
≤10th grade	168 (84)
>10th grade	32 (16)
Occupation
Unemployed	109 (54.5)
Employed	91 (45.5)
Marital status
Single	27 (13.5)
Married	173 (86.5)
Past history of CAD
Yes	11 (5.5)
No	189 (94.5)
Family history of hypertension
Yes	62 (31)
No	138 (69)
Diagnosis
NSTEMI	36 (18)
STEMI	164 (82)
Metabolic syndrome
Yes	75 (37.5)
No	125 (62.5)
Type D personality
Yes	48 (24)
No	152 (76)

SD – Standard deviation; CAD – Coronary artery disease; STEMI – ST-segment elevation type myocardial infarction; NSTEMI – Non-STEMI