Comparison of three instruments for measuring patient anxiety in a coronary care unit

Comparison of three instruments for measuring patient anxiety in a coronary care unit

Doug Elliott

This paper compares the State-Trait Anxiety Inventory (STAI), Hospital Anxiety and Depression Scale (HAD Scale) and a Linear Analogue Anxiety Scale (LAAS) for evaluating anxiety in patients with acute ischaemic heart disease. The instruments were examined for correlation, reliability and internal consistency. Strong associations were demonstrated at pre-test between the STAI and the other scales. Moderate coefficients between HAD-A and HAD-D/LAAS were also apparent. Lower correlations were found at post-test than at pre-test. At post-test, strong inter-correlations occurred for STAVLAAS. The HAD Scale demonstrated high test-retest reliability, while the STAI and LAAS were moderate in their reliability in this sample. The adequate correlation between the instruments suggest that each is a valid and appropriate measure of anxiety in this clinical sample.

INTRODUCTION

Three psychometric instruments (State-Trait Anxiety Inventory (STAI) (Spielberger et al 1983), Hospital Anxiety and Depression Scale (HAD Scale) (Zigmond & Snaith 1983) and Linear Analogue Anxiety Scale (LAAS) commonly used for evaluating anxiety in patients with acute ischaemic heart disease, were examined to determine their validity and correlation. The comparison occurred as part of an intervention study using music and muscle relaxation as strategies for reducing patient anxiety in a coronary care unit (n = 56 patients with unstable angina pectoris or acute myocard-

DOW Elliott RN. BAPPSC (Curtin), MAPPSC (Syd),Assistent Director of Nursing Research, Westmead Hospital and Lecturer, The University of Sydney, Facuky of Health Sciences, East Street, Lidcombe NSW 2141, Australia

(Requests for offprints to DE)

Manuscript accepted 31 May 1993

ial infarction). The purpose of the comparison was to determine which instrument was most appropriate in assessing anxiety levels in this specific patient aggregate. The results of the comparison study only are reported here.

BACKGROUND

Various psychometric instruments have been used in clinical measurements of anxiety. Three common instruments have been the STAI, HAD Scale, and LAAS (Elliott 1992). A number of studies have used the STAI specifically for cardiac patients (Bohachick 1984, Fielding 1980, Foster 1974, Glick 1986, Hase & Douglas 1987, Lueders Bolwerk 1990, Raleigh & Odtohan 1987, Rice et al 1986, Robinson 1988, Thomp- son et al 1987, Zimmerman et al 1988). All studies viewed the STAI as a valid and reliable instrument for the measurement of anxiety in the cardiac population.

195

196 INTENSIVE AND CRITICAL CARE NURSING

The HAD Scale was developed by Zigmond 8c Snaith (1983), and initially validated against

formal psychiatric interviews in hospital outpa-

tients. Aylard et al (1987) further validated the

instrument against other psychometric scales

(irritability, depression and anxiety scale, gen-

eral health questionnaire) in assessing anxiety

and depression in hospital outpatients. Others

have found the HAD Scale to be effective in

quantifying the anxiety of cardiac patients

(Channer et al 1985, Thompson 1989). Linear Analogue (Visual Analogue) Scales

have been used to measure subjective phe-

nomena (such as pain and anxiety) in a variety of

clinical settings. Reviews have concluded that the

scales are simple, sensitive, and reproducible

measuring instruments (Gift 1989, Huskisson

1983, Lee 8c Kiekhefer 1989, Wewers & Lowe

1990).

METHODS

The study was conducted in a seven room

coronary care unit of a major referral and

teaching hospital in Sydney, Australia. A con- venience sample of 56 volunteer subjects was

recruited from the patient population of coro-

nary care unit (CCU) admissions at the hospital.

The provisional medical diagnoses of these

patients were acute myocardial infarction

(AMI), or unstable angina pectoris (UAP). The subjects were generally simple or uncomplicated

cases. That is, the patients’ progress and recov-

ery were routine, with no complications occurring in the immediate coronary care unit period.

The comparison of instruments occurred as part of a randomized controlled trial using music and muscle relaxation audio tapes as interventions for reducing patient anxiety. Institutional ethics approval was granted for the intervention study, and copyright permission was given to the researcher to use the STAI and the HAD Scale in the study. The LAAS is a generic scale commonly used in the clinical setting. Following subject selection and informed consent, the three instruments (STAJ, HAD, LAAS) were

administered as a pre-test immediately before

the first intervention session. The average time

lapse between CCU admission and pre-test was 12h (standard deviation: 7.4h; range: 2-33h).

At the post-test, the three questionnaires were completed by the subjects after the last interven-

tion session and before transfer from the CCU.

The pre-test and post-test questionnaires were

the same instruments, other than some additional attitude and belief questions on the

post-test questionnaire.

Measuring instruments

The

1.

three instruments are described below:

STAI-A State (STAI, S-Anxiety Scale -

Form Y-l): a 20 item questionnaire with 4

response categories. Minimum and maxi-

mum scores are 20 and 80 respectively.

Previous alpha-reliability coefficients for

the STAI overall were stated as 0.83-0.92,

and for the STAI-A State scale as 0.92-0.94

(Spielberger et al 1983). Administration of

the scale under stressful and non-stressful

conditions provided evidence of construct

validity (Spielberger et al 1970). Raw scores have been categorised into low anxiety:

20-39; medium anxiety: 40-59; high

anxiety: 60-80 (Spielberger et al 1983,

Lueders Bolwerk 1990, Zimmerman et al

1989).

HAD Scale: a 14 item questionnaire with

weighted responses (O-3 for each item) divided into two separate sub-scales; 7

items each for anxiety and depression;

greater than 10 points equates to significant anxiety or depression, 8-10 points is borderline significance for either scale, and less than 7 points is not significant (Zig- mond & Snaith 1983). LAAS: a 1Ocm line, with extremes of total calm and extreme anxiety. Linear (Visual) Analogue Scales have been used to measure most commonly pain and anxiety. Following repeated testing and validation, Huskisson (1983) described the analogue scale as being a simple, sensitive, and reproducible measuring instrument.

INTENSIVE AND CRITICAL CARE NURSING 197

Others have since concurred with this

evaluation (Gift (1989, Lee & Kiekhefer

1989, Wewers & Lowe 1990).

Although ostensibly measuring the same

phenomenon, the format and style of the three instruments were different. The STAI required

subjects to agree or disagree (four options) with

particular terms. The HAD Scale required

choice of a fixed response statement from four

selections. Both questionnaires were indirect in

their assessment of anxiety. In contrast, the

LAAS required a cross on a line and asked

directly about the subject’s feeling of anxiety.

ANALYSIS

Evaluation of the psychometric instruments was

conducted using Pearson’s product moment

correlation coefficient (Pearson’s r); specifically

to test the relationship between instruments, and to examine the test-retest reliability of the indi-

vidual tools. Correlation coefficients greater

than 0.5 indicate a strong inter-instrument rela-

tionship, while coefficients of 0.3-0.5, and

0.1-0.3 indicate moderate and weak relation-

ships, respectively (Burns & Grove 1987). Inter-

nal consistency of the STAI and HAD Scale was also conducted. In addition, the STAI was evalu-

ated against normative data, and each ques-

tionnaire was reviewed against categories of borderline and significant anxiety.

RESULTS

Correlation

a) Pre-test. Pearson’s product-moment corre-

lation coefficients between instruments are outlined in Table 1. The three anxiety scales demonstrated a strong correlation with each

other. The moderate correlation of the HAD- Depression scale with the other questionnaires (0.45-0.52) was of interest, given that this scale was measuring a different phenomenon (depression) to the anxiety measuring instruments.

Table 1 Correlathn4zodchWbetwoonpsychological m~urina htruments at P*te8t

STAI HAD-A HAD-D IAAS

STAI 0.64 0.52 0.70 HAD-A 0.45 0.46 HAD-D 0.45 IJUS

6) Post-test. Correlation coefficients at post-test

are outlined in Table 2. The correlations at

post-test were lower than the pre-test for all

instruments. The STAI and LAAS did demon-

strate a continuing strong relationship, although

the STAI and HAD-A displayed a moderate,

and the HAD-A and LAAS, a weak relationship.

One would expect a strong relationship between the anxiety instruments, given that they

were theoretically measuring the same pheno-

menon. In this sample at pre-test, all anxiety scales demonstrated that strong relationship.

This result is clinically useful, considering that

the STAI and HAD instruments consist of

answering and scoring 20 and 14 items, respec-

tively. In contrast, the LAAS only requires a

cross on a line to establish a given level of anxiety.

That is, the LAAS is much more simple and

quick to use, and may provide similar informa- tion to the more complex and time-consuming

STAI and HAD instruments. Correlations of the

instruments were lower at post-test, particularly

so between the LAAS and HAD Scale. However, the scales’ stability could have been affected by

the interventions. Thus, lower correlations at

post-test may have been the result of greater

instrument sensitivity to change.

Test-retest reliability

The reliability (test-retest) analysis was per-

formed on the pre-test and post-test scores.

Table 2 CQrnktionmm Pw*Ml~l measuring instruments at port-test

STAI HAD-A HAD-D LAAS

STAI 0.46 0.33 0.67 HAD-A 0.44 0.26 HAD-D 0.19 LAAS


Table 3 Rdirbility (test-retest) coefficients (Pearson’s rl

STAI : 0.57 HAD-A : 0.79 HAD-D: 0.80 LASS: 0.81

Tests for stability of the measuring instruments

(similar results obtained on repeated administ-

ration of the instrument) are detailed in Table 3.

These results indicate that both sections of the HAD Scale were reliable (r > 0.80; Thomas

1990). The STAI and LAAS were somewhat less

stable for test-retest reliability in this sample.

Previous test-retest analysis of the STAI revealed a median correlation coefficient of 0.33

(Spielberger et al 1983). According to Spielber- ger, low stability should be expected because of

the transitory nature of anxiety states.

Internal consistency

Evaluation of the internal consistency of the

STAI was performed. The reliability coefficient

(Cronbach’s alpha) was calculated as 0.65 for the

STAI in the present study. Cronbach’s alpha values of 0.80 or greater are considered accepta-

ble levels for internal consistency (Thomas

1990). Previous studies involving college

students and military recruits demonstrated

alpha coefficients in excess of 0.90 (Spielberger

et al 1983). Internal consistency of the HAD Scale was also

performed. The inter-correlation matrix of the

Anxiety and Depression sections of the Scale both revealed extremely low values (0.34 and

-0.72, respectively), indicating that, according

to the present study’s findings, individual items

may have been measuring substantially different aspects of anxiety and depression. Other studies did not report internal consistency analyses for the HAD Scale. It would appear that evaluation of internal consistency of the HAD Scale may not be an appropriate method of analysis in this clinical sample.

Evaluation of STAI

As normative data and comparative studies were available for the STAI, further evaluation of this

instrument was conducted. Normative data for

Form Y (the STAI scale used in this study) obtained for working adults (n = 1838) indicated

a STAI-State Anxiety mean score of 34.01 (standard deviation of 9.98) (Spielberger et al

1983). Additional normative data were obtained

from general medical-surgical patients (n= 161)

for the Form X (the precursor of, and highly

correlated to the present Form Y). The mean age

of the general medical-surgical patients was 55

years (no standard deviation given). The mean

STAI-State score was 42.38 (standard deviation

of 13.79) (Spielberger et al 1983).

In comparison, the mean age for this study

was 60.6 years (standard deviation of 13.0 years).

The mean STAI-State score at pre-test was 38.64

(standard deviation of 10.9). As can be seen, the

patients in the present study were somewhat

older, and had a measurably lower anxiety score at pre-test, than the general medical-surgical

patient of previously cited studies.

Comparison of defined anxiety categories for the instruments

Raw scores from the STAI were grouped into three categories in an attempt to provide clinical

meaning to the scores (Spielberger et al 1983,

Lueders Bolwerk 1990, Zimmerman et al 1988):

0 low anxiety (scores of 20-39)

0 medium anxiety (scores of 40-59)

0 high anxiety (scores of 60-80)

Raw scores from the HAD Scale were grouped into categories defined by the instrument

developers (Zigmond and Snaith 1983):

no significant anxiety or depression (scores of less than 7 points) borderline anxiety or depression (scores of 8-10 points) significant anxiety or depression (scores of greater than 10 points).

Whilst there have been no interpretative categories set for the LAAS, a simple division of the Scale into three components was conducted in an attempt to add clinical meaning to the raw scores (low: O-33; medium; 34-66; high: 67-99). Table 4 illustrates the frequencies and percentages for

INTENSIVE AND CRITICAL CARE NURSING 199

TabI. 4 EqMmBant lMdrty cato&os at pm-teat for each instrument

Frequency (% of Sample) Low Medium Hiah

STAI 29 (52) 25 (45) 2 (4) HAD-A 32 (57) 10 (18) 14 (25) HAD-D 49 (87) 5 (9) 2 (4) LAAS 25 (45) 28 60) 3 (8)

the above categories of the STAI, HAD Scale

and LAAS at pre-test.

As can be seen, approximately half the sample

demonstrated low anxiety for the STAI, HAD-A

and LAAS, while only two or three subjects

perceived their anxiety to be high for STAI and

LAAS. A low perception of anxiety has attemp-

ted to be explained by its apparent inverse

correlation with an increased state of denial (Robinson 1988), and the postulation of denial as

an initial and common defence mechanism used

by individuals admitted to a CCU (Cassem &

Hackett 1973), although the relationship

between anxiety and denial has never been effectively demonstrated. However, 25% of

subjects were categorised as having significant

anxiety, according to the HAD-A Scale. This

percentage was much higher than for the STAI

and the LAAS, and corresponded to a lower

relative percentage for borderline anxiety in the

HAD Scale. This may indicate that the border

between the HAD categories are at a different

level to the STAI (the LAAS categories were arbitrary). In summary, very few subjects

demonstrated anxiety scores corresponding to

the high anxiety level categories, whilst the rest

of the sample were categorised into low or medium anxiety.

DISCUSSION

Reliability analyses demonstrated high test- retest reliability for both sections of the HAD Scale. The STAI and LAAS were less stable for this analysis in the present study. Spielberger et al (1983) argued that low stability is indicative of the transitory nature of anxiety states, and not a weakness of the measuring instrument (in this case, the STAI). He went on to state that alpha

coefficients were a more appropriate indicator

of reliability in the STAI than test-retest analysis.

Analysis of internal consistency for the STAI

revealed an alpha coefficient of 0.65. The figure

for the present study was lower than coefficients reported previously (Spielberger et al 1983).

The internal consistency figures for the HAD

Scale in the present study were low, indicating that this form of analysis may be inappropriate

for an instrument of this nature, which may

examine the phenomena (in this case, anxiety

and depression) from substantially different

viewpoints.

When categorising the raw scores for the

subjects’ responses into low, medium and high

anxiety groups, a majority of subjects exhibited

low anxiety. Very few subjects displayed

responses in the high or significant categories.

NURSING IMPLICATIONS

The adequate correlation between the instru-

ments suggests that they were all valid in

measuring anxiety in this clinical sample.

However, relatively low levels of anxiety were

identified in the study. That is, this sample provided a small range of scores, primarily at the

low end of the instruments’ ranges. Similar

studies have used the three psychometric instru-

ments separately with effect. If the reported

high incidence of denial in this clinical popula-

tion (Robinson 1988) is true, it may be more

appropriate to use an indirect form of anxiety

measurement (i.e. STAI or HAD), as opposed to

the more direct method of the LAAS. However, significant inter-correlation among instruments

in this study indicates that a simple measure such

as LAAS may be as effective in measuring

subjective anxiety, and would be as appropriate to use as the other two instruments, which are more complex and time-consuming. Previous literature did not compare the STAI, HAD and LAAS as instruments for use with cardiac patients or other clinical samples. Therefore, no external validation of the present study’s findings could be conducted.


References

Aylard P R, Cooding J H, McKenna P J, Snaith R P 1987 A validation study of three anxiety and depression self-assessment scales. Journal of Psvchosomatic Research 3 1: 261-268

Bohachick P 1984 Progressive relaxation training in cardiac rehabilitation: effect on psychological variables. Nursing Research 33: 283-287-

Burns N. Grove S K 1987 The nractice of nursine research: conduct, critique a;d utilization. W s” Saunders, Philadelphia, ~150

Cassem N H, Hackett T P 1973 Psychological rehabilitation of myocardial infarction patients in the acute phase. Heart 8c Lung 2: 382-388

Channer K S, James M A, Papouchado M, Rees J R 1985 Anxiety and depression in patients with chest pain referred for exercise testing. The Lancet (12th October): 820-823

Elliott D 1992 Measuring patient anxiety in coronary care. Australian Critical Care 5 (2): 12-l 5

Fielding R 1980 A note on the behavioural treatment in the rehabilitation of myocardial infarction patients. British Journal of Social and Clinical Psychology 19: 157-161

Foster S B 1974 An adrenal measure for evaluating nursing effectiveness. Nursing Research 23: 118-124

Gift A G 1989 Visual analogue scales: measurement of subjective phenomena. Nursing Research 38: 286 288

Click M S 1986 Caring touch and anxiety in myocardial infarction patients in the intermediate cardiac care unit. Intensive Care Nursing, 2 (2): 61-66

Hase S, Douglas A 1987 Effects of relaxation training on recovery from myocardial infarction. The Australian Journal of Advanced Nursing 5 (1): 1 Q-27

Huskisson E C 1983 Visual analogue scales. In Melzack R (ed) Pain measurement and assessment. Raven Press, New York, ~~33-37

Lee K A, Kiekhefer G M 1989 Measuring human responses using visual analogue scales. The Western Journal of Nursing Research 11: 128-132

Lueders Bolwerk C A 1990 Effects of relaxing music on state anxiety in myocardial infarction patients. Critical Care Nursing Quarterly 13 (2): 63-72

Raleigh E H, Odtohan B C 1987 The effect of a cardiac teaching program on patient rehabilitation. Heart & Lung 16: 311-317

Rice V H, Caldwell M, Butler S, Robinson J 1986 Relaxation training and response to cardiac catheterization. Nursing Research 35: 39-44

Robinson K R 1988 Denial and anxiety in second-day myocardial infarction patients. In: Waltz C F, Strickland 0 L (eds) Measuring of nursing outcomes: measuring of client outcomes (Vol. 1). Springer, New York, pp. 47-60

Spielberger C D, Gorsuch R, Lushene R 1970 Manual for the state-trait anxiety inventory. Consulting Psychologists, Palo Alto

Spielberger C D, Corsuch R, Lushene R 1983 Manual for the state-trait anxiety inventory: STAI (form v). Consulting Psychologists, Palo Alto, pp 4-6, 10, i.3

Thomas B S 1990 Nursina research: an exneriential approach. C V Mosby, s”t. Louis, p 104, iO7

Thompson D R, Webster R A, Cordle C J, Sutton T W 1987 Specific sources and patterns of anxiety in male patients with first myocardial infarction. British Journal of Medical Psychology 60: 343-348

Thompson D R 1989 A randomized controlled trial of in-hospital nursing support for first time myocardial infarction patients and their partners: effects on anxiety and depression. Journal of Advanced Nursing 14: 291-297

Wewers M E, Lowe N K 1990 A critical review of visual analogue scales in the measurement of clinical phenomena. Nursing Research 13: 227-236

Zigmond A S, Snaith R P 1983 The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica 67: 361-370

Zimmerman L M, Pierson M A, Marker J L 1988 Effects of music on patient anxiety in coronary care units. Heart & Lung 17: 560-566

Comparison of three instruments for measuring patient anxiety in a coronary care unit

Documents

Transcript of Comparison of three instruments for measuring patient anxiety in a coronary care unit