Portuguese validation of the Short Health Anxiety Inventory: Factor structure,reliability,and factor invariance

Abstract

The Short Health Anxiety Inventory is a brief instrument to assess health anxiety widely used across countries; however, no validated version is available for Portuguese-speaking population. Factorial structure, reliability, and equivalency factor with the Spanish version were analyzed with Portuguese adolescents aged 14–18 years. A Portuguese adolescent cohort (N = 629) and a comparative Spanish adolescent cohort (N = 1502) were evaluated. The original two-factor version was the best fitting model for the Portuguese version. The reliability was excellent. Complete measurement invariance across both countries was supported. The Portuguese version of the Short Health Anxiety Inventory is a valid screening inventory to assess health anxiety in adolescents.

Keywords

adolescence health anxiety invariance Portugal Short Health Anxiety Inventory validation

Health anxiety (HA) involves preoccupation and fear with having or acquiring a serious illness based on misinterpretation of ambiguous and unexplained somatic symptoms (Hadjistavropoulos et al., 2012, 2014). Individuals with HA are easily alarmed about their health status and tend to check their body repeatedly for signs of illness. Despite appropriate medical evaluation, preoccupation and fears related to health persists, which causes distress and deteriorates multiple areas of functioning (American Psychiatric Association (APA), 2000). Abnormal illness behavior, inadequate coping strategies, and benefits from sick role contribute to maintain HA (Hart and Björgvinsson, 2010).

The prevalence of HA in community samples ranges from 1.3 to 10 percent, and from 3 to 8 percent in ambulatory medical samples with no gender differences (APA, 2013). Adolescence and early adulthood are critical periods to develop this disorder (Barlow et al., 2001). During the adolescence, somatic complaints are highly prevalent, which tend to co-occur with HA (Fergus and Valentiner, 2011; Goodwin et al., 2013). High levels of HA reduce severely personal well-being (physical symptoms and preoccupation on health), social relationships (spend less time with friends), and academic performance areas (school absenteeism related to medical appointments) and increase the use of medical resources (Asmundson et al., 2010; Bobevski et al., 2016; Rieffe et al., 2009; Tomenson et al., 2012). HA presents a high comorbidity with depressive and anxiety disorders, such as generalized anxiety disorder, panic disorder, and obsessive compulsive disorder (Abramowitz et al., 2007b; APA, 2013). Early diagnosis is important to reduce the negative impact of HA on the personal, social, and occupational functioning of the affected individual and avoid abusing medical resources and services.

Salkovskis et al. (2002) developed the 18-item Short Health Anxiety Inventory (SHAI), a shortened version of the 64-item Health Anxiety Inventory, in order to design a brief screening tool to detect HA, which is easy to implement in any setting. Based on the cognitive-behavioral model of HA (Salkovskis et al., 2002; Warwick and Salkovskis, 1990), SHAI measures fear of illness and fear of negative consequences of an illness using a two-factor structure. SHAI is one of the most widely used self-reports for assessing HA symptoms in community and clinical contexts (Alberts et al., 2013; Hedman et al., 2015b). Compared to other measures of health anxiety—such as Illness Attitude Scales (Kellner et al., 1987) and Whiteley Index (Pilowsky, 1967)—the SHAI includes a broader range of symptoms of HA and discriminates between individuals with high HA and those with somatic illness without exaggerated health concerns (Hedman et al., 2015a). Recently, the Somatic Symptom Disorder-B Criteria Scale (SSD-12) was developed based on the Diagnostic and Statistical Manual of Mental Disorders (DSM) (Toussaint et al., 2016). This new conceptualization of somatization-related disorders needs more evidence and presents some limitations. According to Barsky (2016), seriously ill medical patients may score high on the B criterion of the somatic symptom disorder in the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5) (excessive thoughts, feelings, and behaviors associated with health concerns or somatic symptoms) (APA, 2013), even when their psychological response to the medical illness is not disproportionate.

Previous research has shown that the SHAI is a valuable tool in the assessment of HA among adolescents and youth from the United Kingdom (Salkovskis et al., 2002), the United States (Abramowitz et al., 2007a; Wheaton et al., 2010), Japan (Yamauchi et al., 2009), Hungary (Köteles et al., 2011), China (Zhang et al., 2015), and Spain (Morales et al., 2015); but also samples of adults from the United States (Abramowitz et al., 2007b), Canada (Alberts et al., 2011), and Germany (Bailer et al., 2013).

There is considerable consensus on the adequacy of the two-factor structure of the SHAI (fear of illness and fear of negative consequences of an illness factors), although Abramowitz et al. (2007a) proposed a three-factor model for the SHAI in a study with American undergraduates. These factors are (1) illness likelihood factor, (2) illness severity factor, and (3) body vigilance factor. The SHAI has been highly used across samples; however, no studies have studied the invariance of this tool across countries and cultures. Only Alberts et al. (2011) proved loading invariance of the SHAI across clinical and non-clinical population. Since the SHAI is a highly used tool to assess HA (Hedman et al., 2015b), more research needs to be conducted on its factorial structure and invariance across samples and countries. To our knowledge, there is no validated version of the SHAI in Portuguese population to assess HA; therefore, the aims of this study are threefold: (1) test the latent structure of HA in a sample of Portuguese adolescents, (2) analyze the reliability of the Portuguese version of the SHAI, and (3) test the factorial invariance (FI) of the Portuguese version with respect to a previous validated version of the SHAI in a population with similar cultural background, such as the Spanish one.

Method

Study population and design

The sample was composed of 2131 adolescents aged 14–18 years, students at 28 high schools. Of them, 629 were resident in six cities from Portugal and 1502 from five provinces in Spain. Of the whole sample, most of them had married parents (71.2%) and 20 percent had divorced or separated parents. Table 1 shows the sociodemographic characteristics of the sample by country.

Table 1.

Sociodemographic characteristics of the sample (N = 2131) by country.

Characteristics	Portugal (n = 629)	Spain (n = 1502)	Contrast
No. (%) male	269 (42.8)	767 (51.1)	12***
Mean age (SD), years	17.05 (0.79)	14.91 (0.95)	48.04***
14 years, N (%)	0 (0)	606 (40.3)	1187.50***
15 years, N (%)	9 (1.5)	542 (36.1)
16 years, N (%)	145 (24.8)	248 (16.5)
17 years, N (%)	239 (40.9)	88 (5.9)
18 years, N (%)	191 (32.8)	18 (1.2)
No. (%) family situation
Married parents	428 (68.1)	1179 (78.5)	111.50***
Separated or divorced	148 (23.5)	294 (19.6)
Parents who live together	9 (1.4)	8 (0.5)
Single parents	7 (1.1)	20 (1.3)
Lost one or both parents	37 (5.9)	1 (0.1)

SD: standard deviation.

Contrast: Student’s t-test for quantitative variables and χ² test for qualitative variables.

***

p < 0.001.

The cultural adaptation of the SHAI in Portugal was conducted based on the Spanish validated version (Morales et al., 2015). Permissions from the Regional Ministry of Education of the Autonomous Region of Madeira (Portugal) and the ethics committee of the Miguel Hernández University (Spain) were obtained for this study. In the first phase, the inventory in Spanish was translated into Portuguese by the method of translation and retro-translation by experts in clinical and health psychology. A bilingual translator translated the original items. Then, a bilingual member of the research team conducted the back translation. Members of the team confronted both versions and agreed when there were discrepancies. Later, two specialists in clinical and health psychology revised both translations independently and then the final version was agreed. A small sample of Portuguese scholars (n = 5) completed the questionnaire to check the understandability of the items and they had the possibility of make suggestions to improve the clarity of the items. No changes were necessary at that point. The last version of the SHAI in Portuguese was applied to students from secondary schools in Portugal and Spain. The implementation phase was conducted between January and July 2012 in both countries. A total of 28 high schools from six different areas from the autonomous region Portugal (n = 10) and five provinces of Spain (n = 18) participated in the study.

Permission of the participating high schools and consents from the parents of minor participants were obtained. Informed consent was obtained from all individual participants included in the study. Participation was voluntary, and data were treated confidentially. Students were asked to answer individually the items. Doubts that arose at the time were solved by a researcher. The evaluation lasted 1 hour and took place in the class during the academic schedule. After a month of the first application, participants from Portugal were reassessed at the schools to calculate the test–retest reliability of the Portuguese version of the SHAI.

Measures

SHAI

SHAI evaluates the level of HA, which is understood as the concern related to the health and fear of negative consequences of illness. This instrument consists of 18 items with four response options scored with 0 (no symptoms), 1 (mild symptoms), 2 (severe symptoms), or 3 (very severe symptoms). People diagnosed with hypochondriasis obtain higher scores than those with anxiety disorders, patients with some form of physical illness, and those without anxiety disorders. The original study reported two-factor structure of the SHAI with an excellent intern consistency: Cronbach’s alpha of 0.89 for the total scale, 0.84 for F1 (Fear of Illness), and 0.72 for F2 (Negative Consequences on an illness) (Salkovskis et al., 2002). Test–retest reliability was adequate (0.76). Reliability and validity of this inventory has been supported across the countries with clinical and non-clinical populations.

Sociodemographic characteristics (ad hoc)

Sex, age, and family situation were evaluated

Statistical analysis

Lavaan package for structural equation modeling version 0.5–12 (BETA) (Rossel, 2012) was used to test the classical latent structure of the SHAI with Portuguese adolescents. The factor structure of the SHAI was confirmed following the Hu and Bentler (1999) criteria. A proper adjustment of the model was determined by values greater than 0.90 for comparative fit index (CFI) and Tucker–Lewis index (TLI), less than 0.08 for root mean square error of approximation (RMSEA), and 3 or less for χ²/gl. As suggested by Finney and DiStefano (2006), the normality distributional assumption was checked by Mardia’s (1974) multivariate kurtosis index, in order to select the most appropriate method of estimation for confirmatory factorial analysis (CFA). Because of the lack of normality of the data, diagonally weighted least squares (DWLS) estimator was used. DWLS is a method recommended because of its robustness with ordinal data, small samples, and even in cases of violations of normality (Forero et al., 2009; Mîndrilă, 2010).

CFA models were specified including covariance between the errors of specific items on the identification of physical changes based on theoretical and substantial issues: items 1 and 3 (E1, E3), 2 and 3 (E2, E3), 3 and 10 (E3, E10), and 10 and 13 (E10, E13) (Figure 1). Item 1 refers to the worry about health, and item 3 refers to the awareness of bodily sensations or changes; so that a person who is aware of bodily sensations (evaluated in item 2) may be more worried about his or her health (item 1). The interpretation of aches/pains as appropriate or not to people of the similar age (item 2) is related to the awareness and worry about bodily sensation (item 3). If someone interprets that a specific pain body is common on his or her age, he or she probably will not be so worried than if he or she interprets it is not common. Someone who is aware of bodily sensations (item 3) is more likely that he or she wonder what it means bodily sensations (if he or she has them) (item 10). Finally, invested time in wondering what means bodily sensations (item 10) may be related to the difficulties to think other things rather than unexplained bodily sensations (item 13).

Figure 1.

Two-factor structure of the SHAI with Portuguese sample.

The Statistical Package for Social Sciences (SPSS, v 21) was used for sample descriptive analyses. The effect size of the differences in characteristics of the sample from Portugal and Spain and gender was calculated using Cohen’s (1988) d index. Ordinal alpha coefficient was estimated because it is considered a more accurate estimation of reliability for ordinal response scales than Cronbach’s alpha (Gadermann et al., 2012; Zumbo et al., 2007). A two-way random-effects analysis of variance intraclass correlation coefficient (ICC) for absolute agreement was calculated to estimate the test–retest reliability of the SHAI for baseline to day 30. An ICC test–retest of 0.60 was set as the minimum acceptable ICC value in this study, as suggested by other researchers (Anastasi, 1998; Baumgartner and Chung, 2001).

EQS 6.1 was used to calculate FI that implies configural invariance and measurement invariance (MI). Configural invariance indicates whether the factorial structure is similar across the groups. MI refers to the degree in which the parameters of the instrument are similar across groups (Byrne, 2008). MI includes three levels: metric or weak, strong, and strict. A moment analysis and maximum likelihood, robust (ML, robust) estimator was used because of the lack of multivariate normality (Mardia = 34.28 in Portugal, and 93.01 in Spain). When the difference in the CFI between the two models is equal to or less than 0.01, all restrictions are accepted, and the process of analyzing invariance continues. Conversely, when the difference in the CFI between the both models is greater than 0.01, the least restrictive model is accepted and rejects the other one, that is, the more restrictive model does not hold (Cheung and Rensvold, 2002). Completed cases were included for the analyses.

Results

CFA

The indexes of adjustment of the single-factor model showed a poor fit to the data, as follows: χ² (131) = 504.36, p ⩽ 0.001, CFI = 0.88, TLI = 0.86, and RMSEA = 0.067 (0.061, 0.074). CFA indicated that the two-factor model—proposed by the original authors (Salkovskis et al., 2002)—showed a good fit of the data: χ² (130) = 368, p ⩽ 0.001, CFI = 0.92, TLI = 0.91, and RMSEA = 0.054 (0.047, 0.060); however, indices for the three-factor model were not acceptable (See Abramowitz et al., 2007b): χ² (129) = 782.37, p ⩽ 0.001, CFI = 0.78, TLI = 0.74, and RMSEA = 0.090 (0.084, 0.096).

Based on the adjustment of the models and previous studies with similar results (Abramowitz et al., 2007b; Köteles et al., 2011; Yamauchi et al., 2009), we selected the two-factor solution proposed by the original authors (Salkovskis et al., 2002) as the best fitting model. The model is composed of two factors: Fear of Illness (items 1–14) and Negative Consequences of an illness (15–18). The correlation between both factors was moderate for the Portuguese (0.61). All item loadings were higher than 0.51, with the exception of item 7 (“Ability to control the thoughts on health”) for Fear of Illness factor (0.46) (Figure 1). All other analyses were performed based on the two-factor structural model.

CFA for the Spanish sample was carried out. According to the Portuguese model, the adjustment for the single-factor model was poor: χ² (131) = 1144.24, p ⩽ 0.001, CFI = 0.86, TLI = 0.83, and RMSEA = 0.072 (0.068, 0.076). Indices for the two-factor were acceptable as follows: χ² (153) = 688, p ⩽ 0.001, CFI = 0.92, TLI = 0.91, and RMSEA = 0.054 (0.050, 0.057). However, the three-factor model did not fit the data: χ² (129) = 2238.39, p ⩽ 0.001, CFI = 0.69, TLI = 0.64, and RMSEA = 0.10 (0.101, 0.108). The model composed of two factors was selected based on the indices of adjustment, which coincides with the original version (Salkovskis et al., 2002), and a previous validation of the model in Spain (Morales et al., 2015).

Internal consistency and analysis of items

Table 2 shows the reliability and properties of the items of the Portuguese and Spanish versions of the SHAI. In the Portuguese version, the corrected item-total correlations $(r_{i t}^{c})$ were higher than 0.30 in all items—criterion recommended by Nunnally et al. (1967)—except items 2, 3, and 16. In the Spanish version, the corrected item-total correlations $(r_{i t}^{c})$ were higher than 0.30 in all items, except item 1. Since the removal of these items did not increase alpha significantly, based on the clinic criteria, we decided to keep all of them in the final Portuguese version. The mean values of the items—except item 5—were lower than the theoretical mean (1.5), which is not surprising using a screening instrument with non-clinical population. The values of standard deviation (SD) ranged between 0 and 1, indicating an acceptable variability in responses.

Table 2.

Psychometric characteristics of the Portuguese and Spanish versions of the SHAI.

Country	Item	M	SD	$r_{i t}^{c}$	α-i
Portugal	Item 1	1.63	0.76	0.37	0.82
	Item 2	1.01	1.18	0.26	0.83
	Item 3	1.83	0.87	0.20	0.83
	Item 4	0.96	0.85	0.59	0.81
	Item 5	1.67	1.08	0.47	0.82
	Item 6	0.66	0.73	0.51	0.81
	Item 7	0.77	0.83	0.72	0.80
	Item 8	0.40	0.71	0.53	0.81
	Item 9	0.60	0.65	0.57	0.81
	Item 10	1.32	1.05	0.34	0.82
	Item 11	0.89	0.81	0.43	0.82
	Item 12	0.53	0.67	0.62	0.81
	Item 13	0.95	0.85	0.53	0.81
	Item 14	0.97	0.59	0.45	0.81
	Item 15	0.84	0.85	0.50	0.81
	Item 16	0.80	0.73	0.27	0.83
	Item 17	0.73	0.92	0.52	0.81
	Item 18	0.58	0.82	0.44	0.82
	α	0.82
	Item M (SD)	0.95 (0.91)
	FI M (SD)	14.25 (5.69)
	NC M (SD)	2.97 (2.29)
	Total M (SD)	17.22 (6.88)
Spain	Item 1	1.37	0.67	0.29	0.88
	Item 2	1.05	0.98	0.36	0.88
	Item 3	1.42	0.87	0.35	0.88
	Item 4	0.82	0.83	0.55	0.87
	Item 5	1.15	0.98	0.48	0.87
	Item 6	0.34	0.61	0.61	0.87
	Item 7	0.56	0.75	0.69	0.86
	Item 8	0.33	0.71	0.64	0.87
	Item 9	0.48	0.68	0.65	0.87
	Item 10	0.84	0.93	0.52	0.87
	Item 11	0.43	0.67	0.61	0.87
	Item 12	0.33	0.62	0.71	0.86
	Item 13	0.76	0.80	0.59	0.87
	Item 14	0.95	0.60	0.44	0.87
	Item 15	0.76	0.87	0.51	0.87
	Item 16	0.64	0.73	0.51	0.87
	Item 17	0.50	0.81	0.59	0.87
	Item 18	0.43	0.72	0.53	0.87
	α	0.88
	Item M (SD)	0.73 (0.19)
	FI M (SD)	10.83 (5.72)
	NC M (SD)	2.33 (2.27)
	Total M (SD)	13.15 (7.12)

SHAI: Short Health Anxiety Inventory; M: mean; SD: standard deviation; $r_{i t}^{c}$ : corrected item-total correlation; α-i: ordinal alpha if the item is removed; α: ordinal alpha; FI: fear of illness; NC: negative consequences of an illness; Total: SHAI total score.

The total ordinal alphas for the Portuguese and Spanish versions were excellent. It was moderately higher in the version of Spain (ordinal α = 0.88) compared with that of Portugal (ordinal α = 0.82). The reliability of each of the subscales was also adequate in both versions, being higher in the subscale Fear of Illness compared to the Negative Consequences of subscale an illness.

Factorial invariance

Table 3 shows the fit of the four models calculated to test the level of factorial invariance (FI) between the Portuguese and the Spanish versions of the SHAI. The first step was to calculate the configural invariance model. The model specified in the CFA (Figure 1) was used to estimate the two-dimensional model (nested) for Portugal and Spain. After weak and strong invariance models were confirmed (Table 3), strict invariance—in which the variances of errors are restricted—was estimated. The model showed a good fit, noting that the CFI value remained stable compared to the previous model. These results confirm that the versions of the SHAI from Portugal and Spain reached a strict level of invariance.

Table 3.

Goodness-of-fit indexes for different levels of invariance.

Level of invariance	S-B χ²	df	p	CFI	ΔCFI	RMSEA	90% for RMSEA
Level of invariance	S-B χ²	df	p	CFI	ΔCFI	RMSEA	LL	UL
Configural invariance	696.16	260	0.0001	0.921	–	0.040	0.036	0.043
Weak invariance	733.75	276	0.0001	0.917	−0.004	0.039	0.036	0.043
Strong invariance	993.54	292	0.0001	0.922	0.005	0.041	0.038	0.045
Strict invariance	1002.23	297	0.0001	0.922	0	0.047	0.044	0.050

S-B χ²: Santorra–Bentler Scaled chi-square; df: degree of freedom; CFI: comparative fit index; ΔCFI: change in CFI between models; RMSEA: root mean square error of approximation; LL: lower limit; UL: upper limit.

Test–retest reliability

For the 540 (87% of retention) who completed the SHAI twice, ICC test–retest coefficients were 0.69 for total score (confidence interval (CI) 95%: 0.63, 0.74; F = 3.24; p ⩽ 0.001), 0.60 for Fear of Illness (CI 95%: 0.64, 0.74; F = 3.32; p ⩽ 0.001), and 0.62 for Negative Consequences of an illness (CI 95%: 0.55, 0.68; F = 2.65; p ⩽ 0.001). These ICC coefficients indicated a relatively satisfactory level of test–retest reliability (⩾0.60).

Age and gender differences

The average total score of the SHAI was higher in the sample recruited in Portugal (M = 17.25, SD = 6.87) compared to Spanish one (M = 13.15, SD = 7.11; p < 0.001, d = 0.58). Portuguese adolescents had higher scores on the subscales Fear of Illness (Portugal: M = 14.27, SD = 5.68 vs Spain: M = 10.82, SD = 5.71; p < 0.001, d = 0.59), and Negative Consequences of an illness (Portugal: M = 2.97, SD = 2.29 vs Spain: M = 2.33, SD = 2.27; p < 0.001, d = 0.28) compared with Spanish ones. Men from Portugal had higher scores on the overall scale (p < 0.001) and in the respective subscales (p < 0.001) compared to men from Spain. However, women from Spain had higher scores on the overall scale (p < 0.001) and subscales Fear of Illness (p = 0.005) and Negative Consequences of an illness (p < 0.001) than women from Portugal. HA was unrelated to age in Portugal (p = 0.10) and Spain (p = 0.06). Student’s t-test for independent means indicated that there were no differences in the level of HA between those aged 14–15 years and those aged 16–18 years (p = 0.10) in the sample from Portugal and Spain (p = 0.08).

Discussion

The CFA confirmed the good fit of the two-factor structure for the Portuguese version of the SHAI, according to the first study of Salkovskis et al. (2002) with patients with hypochondria, other clinical problems and students: Fear of Illness and Negative Consequences of an illness. Subsequent studies that test the latent structure of the SHAI confirmed the two-factor structure of the inventory in the United States with adult patients diagnosed with hypochondria (Abramowitz et al., 2007b) and college students without item 13 (Wheaton et al., 2010), in Japan (Yamauchi et al., 2009) and Hungary with undergraduate students (Köteles et al., 2011), in Germany with clinical and non-clinical adult population (Bailer et al., 2013), and recently in Spain with high school adolescents (Morales et al., 2015) and in China with undergraduate healthy medical students (Zhang et al., 2015). The adjustment of the two- and three-factor models of the SHAI has been compared in studies with clinical population in the United States (Abramowitz et al., 2007b) and non-clinical population in Japan (Yamauchi et al., 2009), Hungary (Köteles et al., 2011), and Spain (Morales et al., 2015). Despite the good fit of the three-factor model in most of these studies, the two-factor model proposed by Salkovskis et al. (2002) is commonly selected by the researchers because of its clarity and parsimony. The correlation between both factor was moderate (r = 0.61), indicating that both are related although they measure different issues of HA. It was consistent with the study of Morales et al. (2015) (r = 0.54).

The Portuguese version of the SHAI showed an adequate total reliably (ordinal α = 0.82), similar to the studies conducted in the United Kingdom (α = 0.89) (Salkovskis et al., 2002), Hungary (α = 0.83) (Köteles et al., 2011), and Japan (α = 0.85) (Yamauchi et al., 2009), slightly lower than in the United States (α = 0.96) and Spain (ordinal α = 0.93) (Morales et al., 2015), but higher than the version validated recently in China (Cronbach’s α = 0.74) (Zhang et al., 2015). In this study, the temporal stability of the response was adequate (ICC > 0.60 between both assessments). Other studies found similar results (Salkovskis et al., 2002; Zhang et al., 2015).

This study indicates a strict level of invariance between the Portuguese and Spanish versions of the SHAI. This implies that the results of both versions are comparable, with a minimum bias measure (Dimitrov, 2010). Total scores of the SHAI were higher in the sample from Portugal (M = 17.25; SD = 6.87) compared to the one from Spain (M = 13.15; SD = 7.11). In both cases, mean values were distant from 27, which is the cut-off score recommended by Alberts et al. (2013) when 0–3 scale range is used. Cultural, social, and economic factors can explain that Portuguese adolescents reported higher HA symptoms than their Spanish peers. Portugal and Spain national models of health system—whose funding is mainly public (The Beveridge model) —present significant differences in health care access. Unlike Spain, health system in Portugal includes patient copayment for doctor visits (Subdirección General de Información Sanitaria e Innovación, 2014). Since 2012, the cost doubled due to the economic crisis (from €3.80 to €10 for primary care emergency, and from €9.60 to €20 for specialized emergency) (Subdirección General de Información Sanitaria e Innovación, 2014). The cost of access to health resources can reduce the excessive costs associated with patients with HA, but it suggests that it does not reduce the concern about health from those affected by HA. Although it is not the aim of this study to compare the group means across the studies without having checked the FI, it is noteworthy that the level of HA of the Portuguese adolescents was similar to the young Chinese people with anxiety (M = 16.97; SD = 8.27) (Zhang et al., 2015). Another significant aspect is that the degree of HA of the Portuguese adolescents was higher than that reported by other studies with undergraduate students, as Abramowitz et al. (2007a) (M = 12.48; SD = 6.79) and Wheaton et al. (2010) (M = 10.79; SD = 6.38) in the United States, Yamauchi et al. (2009) in Japan (M = 14.04; SD = 6.90), and Morales et al. (2015) in Spain with high school students (M = 10.48; SD = 5.70) (see Albert et al., 2013).

This study has several limitations that provide directions for future research. The Portuguese version of SHAI was validated with non-clinical sample. It would be necessary to conduct a study to test the psychometric properties of this inventory with clinical sample from Portugal. Convergent and divergent validity of SHAI with other measures that assess HA in adolescent population of Portugal must be studied in future studies. Despite the limitations of this study, the outcomes provide evidence of the structure and reliability of the Portuguese version of SHAI to assess HA in adolescents. This is the first study to adapt and validate the SHAI with a Portuguese population and test equivalency factor with a previous validated version. Findings confirm the two-factor structure of the Portuguese version of SHAI, an adequate internal consistency and temporal stability, and its equivalence to the version validated recently in Spain (Morales et al., 2015). These results are promising for the use of the SHAI as a brief screening instrument to detect HA in Portuguese adolescents. The validation of SHAI in Portugal fills an important gap in the evaluation of HA with Portuguese-speaking population from the cognitive-behavioral perspective.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Support for this research was provided by the Regional Agency for the Development of Research Technology and Innovation in Portugal, Ref. 1080/1974, the Foundation for Research and Prevention of AIDS in Spain, Ref. FIPSE 360971/10, and by the Program VALi+d for Research Staff training of the Council of Culture, Education and Science of the Valencian Autonomous Government of Spain, Ref. ACIF/2012/132.

References

Abramowitz

Deacon

Valentiner

(2007a) The short health anxiety inventory: Psychometric properties and construct validity in a non-clinical sample. Cognitive Therapy and Research 31(6): 871–883.

Abramowitz

Olatunji

Deacon

(2007b) Health anxiety, hypochondriasis, and the anxiety disorders. Behavior Therapy 38(1): 86–94.

Alberts

Hadjistavropoulos

Jones

. (2013) The short health anxiety inventory: A systematic review and meta-analysis. Journal of Anxiety Disorders 27(1): 68–78.

Alberts

Sharpe

Kehler

. (2011) Health anxiety: Comparison of the latent structure in medical and non-medical samples. Journal of Anxiety Disorders 25(4): 612–614.

American Psychiatric Association (APA) (2000) Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) (4th edn). Text revision. Washington, DC: American Psychological Association.

American Psychiatric Association (APA) (2013) Diagnostic and Statistical Manual of Mental Disorders (DSM-5®). Washington, DC: American Psychological Association.

Anastasi

(1998) Psychological Testing (6th ed). New York: Macmillan.

Asmundson

Abramowitz

Richter

. (2010) Health anxiety: Current perspectives and future directions. Current Psychiatry Reports 12(4): 306–312.

Bailer

Rist

Müller

. (2013) German validation of the short health anxiety inventory (SHAI). Verhaltensther Verhaltensmodif (in Druck), 34(4): 378–398.

10.

Barlow

Durand

Herrejón

JLN

(2001) Psicología anormal: Un enfoque integral. México, DF: Thomson Learning.

11.

Barsky

(2016) Assessing the new DSM-5 diagnosis of somatic symptom disorder. Psychosomatic Medicine 78(1): 2–4.

12.

Baumgartner

Chung

(2001) Confidence limits for intraclass reliability coefficients. Measurement in Physical Education and Exercise Science 5(3): 179–188.

13.

Bobevski

Clarke

Meadows

(2016) Health anxiety and its relationship to disability and service use: Findings from a large epidemiological survey. Psychosomatic Medicine 78(1): 13–25.

14.

Byrne

(2008) Testing for multigroup equivalence of a measuring instrument: A walk through the process. Psicothema 20(4): 872–882.

15.

Cheung

Rensvold

(2002) Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling 9(2): 233–255.

16.

Cohen

(1988) Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.

17.

Dimitrov

(2010) Testing for factorial invariance in the context of construct validation. Measurement and Evaluation in Counseling and Development 43(2): 121–149.

18.

Fergus

Valentiner

(2011) The short health anxiety inventory and multidimensional inventory of hypochondriacal traits: A comparison of two self-report measures of health anxiety. Cognitive Therapy and Research 35(6): 566–574.

19.

Finney

DiStefano

(2006) Non-normal and categorical data in structural equation modeling. In: Hancock

Mueller

(eds) Structural Equation Modeling: A Second Course, pp. 269–314. Greenwich, CT: Information Age.

20.

Forero

Maydeu-Olivares

Gallardo-Pujol

(2009) Factor analysis with ordinal indicators: A Monte Carlo study comparing DWLS and ULS estimation. Structural Equation Modeling 16(4): 625–641.

21.

Gadermann

Guhn

Zumbo

(2012) Estimating ordinal reliability for Likert-type and ordinal item response data: A conceptual, empirical, and practical guide. Practical Assessment, Research and Evaluation 17(3): 1–13.

22.

Goodwin

Fairclough

Poole

(2013) A cognitive–perceptual model of symptom perception in males and females: The roles of negative affect, selective attention, health anxiety and psychological job demands. Journal of Health Psychology 18(6): 848–857.

23.

Hadjistavropoulos

Alberts

Asmundson

(2014) Health anxiety. In: Storch

McKay

(eds) Obsessive-Compulsive Disorder and its Spectrum: A Life-Span Approach, pp. 205–222. Washington, DC: American Psychological Association.

24.

Hadjistavropoulos

Janzen

Kehler

. (2012) Core cognitions related to health anxiety in self-reported medical and non-medical samples. Journal of Behavioral Medicine 35(2): 167–178.

25.

Hart

Björgvinsson

(2010) Health anxiety and hypochondriasis: Description and treatment issues highlighted through a case illustration. Bulletin of the Menninger Clinic 74(2): 122–140.

26.

Hedman

Lekander

Ljotsson

. (2015a) Optimal cut-off points on the health anxiety inventory, illness attitude scales and Whiteley index to identify severe health anxiety. PLoS ONE 10(4): e0123412.

27.

Hedman

Ljótsson

Andersson

. (2015b) Psychometric properties of internet-administered measures of health anxiety: An investigation of the health anxiety inventory, the illness attitude scales, and the Whiteley index. Journal of Anxiety Disorders 31: 32–37.

28.

Bentler

(1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling 6(1): 1–55.

29.

Kellner

Abbott

Winslow

. (1987) Fears, beliefs, and attitudes in DSM-III hypochondriasis. The Journal of Nervous and Mental Disease 175(1): 20–25.

30.

Köteles

Simor

Bárdos

(2011) Validation and psychometric evaluation of the Hungarian version of the short health anxiety inventory (SHAI). Mentálhigiéné És Pszichoszomatika 12(3): 191–213.

31.

Mardia

(1974) Applications of some measures of multivariate skewness and kurtosis in testing normality and robustness studies. Sankhyā: The Indian Journal of Statistics Series B 36: 115–128.

32.

Mîndrilă

(2010) Maximum likelihood (ML) and diagonally weighted least squares (DWLS) estimation procedures: A comparison of estimation bias with ordinal and multivariate non-normal data. International Journal of Digital Society 1(1): 60–66.

33.

Morales

Espada

Carballo

. (2015) Short Health Anxiety Inventory: Factor structure and psychometric properties in Spanish adolescents. Journal of Health Psychology 20(2): 123–131.

34.

Nunnally

Bernstein

Berge

JMT

(1967) Psychometric Theory. New York: McGraw-Hill Education.

35.

Pilowsky

(1967) Dimensions of hypochondriasis. The British Journal of Psychiatry 113(494): 89–93.

36.

Rieffe

Villanueva

Adrián

. (2009) Quejas somáticas, estados de ánimo y conciencia emocional en adolescentes. Psicothema 21(3): 459–464.

37.

Rossel

(2012) Lavaan: An R package for structural equation modeling and more. Version 0.5—12 (BETA). Available at: http://users.ugent.be/~yrosseel/lavaan/lavaanIntroduction.pdf

38.

Salkovskis

Rimes

Warwick

. (2002) The health anxiety inventory: Development and validation of scales for the measurement of health anxiety and hypochondriasis. Psychological Medicine 32(5): 843–853.

39.

Subdirección General de Información Sanitaria e Innovación (2014) Los sistemas sanitarios en los países de la UE: Características e indicadores de salud 2013. Madrid: Ministerio de Sanidad, Servicios Sociales e Igualdad.

40.

Tomenson

McBeth

Chew-Graham

. (2012) Somatization and health anxiety as predictors of health care use. Psychosomatic Medicine 74(6): 656–664.

41.

Toussaint

Murray

Voigt

. (2016) Development and validation of the somatic symptom disorder-B criteria scale (SSD-12). Psychosomatic Medicine 78(1): 5–12.

42.

Warwick

Salkovskis

(1990) Hypochondriasis. Behaviour Research and Therapy 28(2): 105–117.

43.

Wheaton

Berman

Franklin

. (2010) Health anxiety: Latent structure and associations with anxiety-related psychological processes in a student sample. Journal of Psychopathology and Behavioral Assessment 32(4): 565–574.

44.

Yamauchi

Matsuoka

Himachi

. (2009) Development and validation of Japanese version of the short health anxiety inventory. Japanese Journal of Psychosomatic Medicine 2(1): 833–840.

45.

Zhang

Liu

. (2015) The reliability and validity of a Chinese-version short health anxiety inventory: An investigation of university students. Neuropsychiatric Disease and Treatment 11: 1739–1747.

46.

Zumbo

Gadermann

Zeisser

(2007) Ordinal versions of coefficients alpha and theta for Likert rating scales. Journal of Modern Applied Statistical Methods 6(1): 4. Available at: http://digitalcommons.wayne.edu/jmasm/vol6/iss1/4/