Archives of Medical Research 34 (2003) 287–291

ORIGINAL ARTICLE

Clinical Findings as Predictors of Positivity of Head-Up Tilt TableTest in Neurocardiogenic Syncope

Enrique Asensio,a Jorge Oseguera,a Alvar Lorıa,b Marıa Gomez,a Rene Narvaez,a Joel Dorantes,a

Pablo Hernandez,a Arturo Orea,a Veronica Rebollara and Raymundo Ocaranzaa

aDepartamento de Cardiologıa, Clınica de Arritmias y Marcapasos, bDepartamento de Epidemiologıa Clınica,Instituto Nacional de Ciencias Medicas y Nutricion Salvador Zubiran (INNNSZ), Mexico City, Mexico

Received for publication September 6, 2002; accepted March 14, 2003 (02/173).

Background. Neurocardiogenic (vasovagal) syncope occurs frequently and can bediagnosed with the head-up tilt table (HUTT) test. Our objective in this study was toidentify clinical predictors of the positivity of HUTT test in neurocardiogenic syncope.

Methods. We conducted a prospective study of 117 cases (81 women and 36 men, 13–85years of age) referred to our Institution for HUTT testing. The ability of 10 symptomsand signs of clinical history to predict HUTT positivity were evaluated using logisticregression analysis.

Results. We observed a low rate of test-negative cases (24%) and 89 positives. Nearly allpositives (87/89) were neurocardiogenic, principally of vasodepressor and mixed types(43 and 34 cases, respectively) and a few were cardioinhibitory (10, mostly young males).Regression analysis established that dizziness, nausea, and diaphoresis in past historywere associated with HUTT positivity nearly 25 times more frequently than when absent.

Conclusions. Our three conclusions are that syncope in absence of heart diseaseaccompanied by dizziness, nausea, and diaphoresis may be treated as neurocardiogenicin settings where no HUTT is available. In addition, our low rate of negative tests mayhave been the result of our reexamining referrals prior to deciding test performance, and highfrequency of young males in cardioinhibitory syncope needs further research. � 2003IMSS. Published by Elsevier Science Inc.

Key Words: Neurocardiogenic syncope, Head-up tilt table testing, Clinical predictors.

Introduction

Syncope is a common reason for going to a hospital emer-gency room and a frequent event in the general population(1,2). On the basis of clinical history and physical examina-tion, its origin can be established presumably in a significantnumber of patients (3). Definitive diagnosis can be supportedby tests such as echocardiogram (ECG) in the case of struc-tural heart disease, electroencephalogram (EEG) in the caseof seizures, or with laboratory, imaging, and electrophysio-

Address reprint requests to: Dr. Enrique Asensio Lafuente, Clınicade Arritmias y Marcapasos, Departamento de Cardiologıa, INNNSZ, CalleVasco de Quiroga #15, Col. Seccion XVI, Tlalpan, 14000 Mexico, D.F.,Mexico. Phone/FAX: (�52) (55) 5655-3306; E-mail: easensio@avantel.net

0188-4409/03 $–see front matter. Copyright � 2003 IMSS. Published by Elsevdoi : 10.1016/S0188-4409(03)00046-8

logic tests in other types of syncope. Nevertheless, approxi-mately 40% of patients with syncope will not be properlydiagnosed (4,5). It is now known that a significant numberof episodes of syncope in these patients, ranging from 30–50% of all reported episodes, are relatively benign and aredefined as neurocardiogenic, vasovagal, or neurally medi-ated types of syncope (5). This form of syncope has lowmortality rate and good prognosis (6).

Head-up tilt table test (HUTT) has established itself as auseful test to study and classify these patients (7). The testpermits diagnosis of several entities related to autonomicdysfunctions (8). The test also allows in most instancesdifferentiation of neurocardiogenic syncope into threetypes of response (vasodepressor, cardioinhibitory, andmixed). An important feature in test interpretation is thatsymptoms experienced during testing should mimic those

ier Science Inc.

Asensio et al. / Archives of Medical Research 34 (2003) 287–291288

recorded during the patient’s past history of syncope. Thus,clinical evaluation remains an important component in diag-nosis. Use of clinical symptoms and signs to predict testpositivity could contribute to decrease superfluous testingin this area.

In this paper, we explored the ability of 10 signs andsymptoms to predict positivity of HUTT test in 117 patientsreferred to our laboratory for this purpose. In addition,we analyzed associations of these variables with type ofresponse in 87 vasovagal cases.

Materials and Methods

A prospective, descriptive, observational study was per-formed in 117 consecutive patients referred to our Institutionfor study of syncope of unknown origin during a 3-yearperiod. Patients were interviewed and examined indepen-dently of the referring physician by two cardiologists at ourService to decide whether HUTT was to be performed ifclinical presentation suggested syncope of neurocardiogenicorigin. Twenty patients with evident non-vasovagal causeof syncope were not included in the study.

We collected a complete history for each patient includingdata on family history, past medical history, and specific dataon syncope episodes. These included time elapsed sincefirst episode, number and duration of spells, characteristicfeatures of spells, any accompanying epiphenomena (sei-zures, trauma), and possible triggering event, and where,when, and how episodes presented. The 10 most frequentsigns and symptoms present in clinical history were diapho-resis, dizziness, lassitude, loss of sphincter control, nausea,paleness, paresthesia, seizures, vomiting, and trauma. Thesewere considered as present or positive when the patient hadexperienced these in at least one past episode. All HUTT-positive patients reported that signs and symptoms of theinduced syncope were similar to those of their past history.

Head-up tilt table test protocol. A two-stage tilt test wasperformed when a first nonpharmacologic stage was nega-tive. There were 74/117 (63%) patients who underwent phar-macologic testing due to a negative test in the first stage.As described elsewhere (9), the pharmacologic stage wascarried out with 5 mg of sublingual isosorbide in 62 pa-tients and with intravenous (i.v.) isoproterenol in 12 earlycases. After a 6-h fast and with 0.9% sodium chloride solu-tion (i.v.) installed, the patient had a basal rest period of 10min. At the end of the rest period, patient heart rate andblood pressure were recorded with a noninvasive sphygmo-manometer; subsequently, the patient was tilted to a 70�angle for a maximum of 21 min. Heart rate and blood pres-sure were recorded every 3 min in tilted position. The testwas terminated if the patient suffered a syncopal episodebefore 21 min elapsed; if this did not occur, the patientwas repositioned in decubitus and the pharmacologic agentwas administered. Ten minutes after drug administration,heart rate was measured. If it had increased to at least 20%

above baseline measurement, the patient was tilted again to70� for an additional 21 min or less if symptoms occurred.Diagnostic criteria used to consider a HUTT-positive testwere those generally accepted for hypotension, bradycardia,or both in the setting of neurally mediated syncope (9).Patients included in HUTT-positive cases experienced nearlyall signs and symptoms of previous patient episodes.

Statistical analysis. Analysis was performed using SPSS 8.0statistical software (SPSS, Inc., Chicago, IL, USA) utilizingconventional methods for mean and standard deviation (SD)and nonparametric tests to evaluate group differences (Mann-Whitney for two groups and Kruskal-Wallis for threegroups). Backward stepwise procedure was used to obtainbinary logistic regression model of associations of clinicalvariables with HUTT positivity; nominal logistic regressionwas used to explore associations of variables with the threetypes of neurocardiogenic syncope.

Results

Table 1 summarizes demographic and clinical data of the117 subjects partitioned by positivity of HUTT test. Nearlyone quarter of cases (28/117 � 24%) were negative and 89(76%), positive. Globally, patients were mainly middle-aged (average age, 40 years) and women (69%). Mean ofprevious syncope was 2.9 episodes with average evolutiontime of 53 months. There were no significant differencesbetween negative and positive cases concerning thesevariables.

Table 2 shows frequency of 10 clinical variables of clini-cal history selected as potential predictors for positive HUTTtest. Frequencies are given globally, as are positive andnegative HUTT. Most frequent symptoms were dizzinessand lassitude followed by nausea and diaphoresis, with fre-quencies of 50–69%. In contrast, signs such as seizures,vomiting, and loss of sphincter control had frequencies�15%. In four of 14 patients with trauma history, bonefracture or facial suture was implicated. The four mostfrequent variables showed significantly higher prevalence inHUTT-positive than in HUTT-negative group.

Logistic regression analysis (LRA) of positivity of HUTTtest vs. ten clinical variables using backward stepwise proce-dure showed that three variables (dizziness, nausea, anddiaphoresis) were important predictors of positive tilt tabletest. Table 3 shows regression coefficients that allow calcu-lating risk for positive HUTT when the patient has a historyof dizziness, nausea, and diaphoresis associated with syn-cope. According to our LRA model, a subject with these threesymptoms was nearly 25 (precisely 24.78) times more likelyto have positive HUTT than a subject without the symptoms.The Appendix at the end of this report illustrates coefficientuse of our LRA model of Table 3.

As seen in Table 4, 87 of 89 positive tests were consideredof neurocardiogenic origin; the two exceptions were postural

Clinical Predictors for Positive HUTT 289

Table 1. General data of the subjects globally and according to the outcome of tilt table testing

Negative PositiveGlobal test test Group

Variable (n � 117) (n � 28) (n � 89) differences

Gender Female 81 (69%) 19 (68%) 62 (70%) NS p � 0.86Unit Mean � SD Mean � SD Mean � SDAge Years 40 � 19 46 � 18 39 � 19 NS p � 0.07Previous episodes n 2.9 � 4.9 3.1 � 3.9 2.8 � 5.2 NS p � 0.51Evolution time Months 53 � 87 46 � 90 56 � 86 NS p � 0.45Heart rate Beats/min 71 � 16 73 � 14 70 � 16 NS p � 0.36Diastolic blood pressure mmHg 68 � 11 69 � 11 68 � 11 NS p � 0.62Systolic blood pressure mmHg 111 � 19 106 � 28 112 � 17 NS p � 0.86

Median differences evaluated by Mann-Whitney test; NS � not significant.

orthostatic tachycardia syndrome (POTS) (10) and auto-nomic failure. Divided by type, 43 (49%) of neurocardi-ogenic cases were vasodepressor type, 34 (39%) mixed, and10 (12%), cardioinhibitory. Clinical variables were not asso-ciated with type of syncope, probably because of the smallnumber of subjects with cardioinhibitory type. The sole in-tertype difference in neurocardiogenic subjects was thatcardioinhibitory variant was more frequent among youngmales.

Discussion

To our knowledge, this is the first paper in which a quantita-tive multiple relationship has been established between clini-cal findings and outcome of tilt table test. Thus, our resultsare not comparable to other prospective studies that evalu-ated clinical characteristics during syncope (11–14). Thesestudies aimed to differentiate neurally mediated syncope fromcardiac or arrhythmic syncope (15). A recent paper by Alboni

Table 2. Positivity of 10 clinical variables in subjects of studyglobally and according to outcome of HUTT

Symptomspositive Negative Positivein clinical Global HUTT HUTT Grouphistory (n � 117) (n � 28) (n � 89) differences

Dizziness 81 (69%) 12 (43%) 69 (78%) Pos � Neg 0.001Weakness 62 (53%) 10 (36%) 52 (58%) Pos � Neg 0.037Nausea 60 (51%) 7 (25%) 53 (60%) Pos � Neg 0.001Diaphoresis 59 (50%) 9 (32%) 50 (56%) Pos � Neg 0.027Paleness 32 (27%) 12 (43%) 49 (55%) NS 0.26Paresthesia 19 (16%) 3 (11%) 16 (18%) NS 0.37Seizures 17 (15%) 3 (11%) 14 (16%) NS 0.51Trauma 14 (12%) 4 (14%) 10 (11%) NS 0.67Vomiting 10 (9%) 2 (7%) 8 (9%) NS 0.76Incontinence 9 (8%) 2 (7%) 7 (8%) NS 0.90

Median differences according to Mann-Whitney test. NS � not signifi-cant p �0.05.

et al. (16) stated that diaphoresis prior to loss of conscious-ness and after recovery, and nausea during recovery, aremore frequent in neurally mediated syncope than in cardiacsyncope. This is in agreement with our observation thatnausea and diaphoresis were associated with HUTT positi-vity. On the other hand, Alboni et al. reported that theseassociations disappeared in multivariate analysis and thatvariables associated with neurally mediated syncope con-sisted of a history of �4 years between first and last syncopeplus history of pre-syncope and nausea during recovery.Although strict comparison of our results with those of Alboniet al. was unfeasible as their target value was not tilt testpositivity but neurally mediated syncope, the latter includedtilt-induced syncope in their casuistry.

We were unable to find associations of types of neurocar-diogenic variant with our 10 clinical variables. One limita-tion was our low number of cardioinhibitory cases; however,we plan to continue the study in the hope of increasingcase number.

Our rate of 24% negative HUTT cases is lower than meanof 34% reported by others (15,17–19). Our low figure mayhave been partly due to reexamination of each referred caseprior to deciding whether HUTT was to be performed.In addition, this was so despite our strategy of insisting onsimilarity of clinical data of past episodes and HUTT-induced symptoms. In fact, we believe that our approachcan increase HUTT specificity as a diagnostic tool for neuro-cardiogenic syncope so that patients without such clinical

Table 3. Logistic regression model

Clinical feature B � SE Exp B CI 95%

Dizziness 1.16 � 0.48 3.18 1.23–8.23Nausea 1.23 � 0.51 3.43 1.25–9.37Diaphoresis 0.82 � 0.49 2.28 0.87–5.98Constant �0.42

Method: backward stepwise regression; B: beta coefficient; SE: standarderror of B; Exp B: natural base (e) raised to coefficient B power; CI 95%:95% confidence interval of Exp B.

Asensio et al. / Archives of Medical Research 34 (2003) 287–291290

Table 4. Prevalence of 12 variables according to type of finding in 87subjects with HUTT positive for vasovagal syncope

Mixed Cardioinhibitory Vasodepressor Group(n � 34) (n � 10) (n � 43) differences

Age (years) 36 � 19 29 � 10 42 � 18 p � 0.051Gender 10 (29%) 7 (70%) 9 (21%) p � 0.01

(male)Dizziness 25 (74%) 9 (90%) 34 (79%) NS 0.53Weakness 18 (53%) 5 (50%) 29 (67%) NS 0.35Nausea 21 (62%) 4 (40%) 27 (63%) NS 0.40Diaphoresis 17 (50%) 6 (60%) 26 (60%) NS 0.64Paleness 22 (65%) 6 (60%) 21 (49%) NS 0.37Paresthesia 6 (18%) 3 (30%) 7 (16%) NS 0.60Seizures 5 (15%) 4 (40%) 5 (12%) NS 0.09Trauma 5 (15%) 1 (10%) 4 (9%) NS 0.76Vomiting 3 (9%) 1 (10%) 4 (9%) NS 0.99Incontinence 4 (12%) 0 3 (7%) NS 0.46

Median group differences evaluated by Kruskal-Wallis test.

features probably have syncope of a different etiology. Withregard to specificity of tilt table test, this varies from 86 to100% (9,17) with more controversial sensitivity (17–20).Confirmation by others that nausea, diaphoresis, and dizzi-ness are important predictors of positive HUTT might help tocreate a clinical diagnostic power scale useful for diagnosticpurposes in this area. One could proceed further by statingthat in settings where HUTT cannot be performed, a historyof syncope accompanied by dizziness, nausea, and diaphore-sis in absence of heart disease may justify initiation of specifictherapy for neurocardiogenic syncope. Although dizziness,nausea, and diaphoresis are nonspecific, they are associatedwith increases in vagal activity, which is pathophysiologi-cally linked to clinical manifestations of neurally me-diated syncope.

With the results of our study, we reached the followingthree conclusions: 1) past history of syncope associated withdizziness, nausea, and diaphoresis in our patients implied25-fold increase in probability of having positive HUTTtest. We believe that patients without heart disease and expe-riencing syncope accompanied by these symptoms maybe treated for neurocardiogenic syncope in settings whereHUTT test is unavailable for technical, economic, or organi-zational reasons; 2) we observed a low rate of negative tilttable testing, which may be partly due to our strategy of re-examining referred patients prior to performing the test, and3) we were unable to predict on the basis of clinical featuresthe type of neurocardiogenic syncope in our 87 positivecases. Our observation of high frequency of cardioinhibitorytype in young males needs further research.

Appendix

Logistic regression model is utilized by assigning numericexpression (a factor or multiplier) to all its B coefficients

(beta coefficients). Factors used in binary variables of lo-gistic models consist of only two, i.e., either one if the caseis positive (or present) for corresponding variable, or zero ifnegative (or absent). We shall illustrate the procedure bycalculating risk of HUTT positivity using our model inTable 4:

Log risk � �0.42 � 1.16 dizziness (DIZZ) � 1.23 nausea(NAU) � 0.82 diaphoresis (DIAPH)

Step 1. Calculate log risk for patient positive for threevariables. Because they are present, all factors would equalone and the corresponding logarithm would be the following:

Log risk Pos � �0.42 � (1.16 · 1) � (1.23 · 1)� (0.82 · 1) � �0.42 � 1.16 � 1.23 � 0.82 � �2.79

Step 2. Calculate risk of a case without variables. Factorswould be zero because they are absent and log risk would be:

Log risk Neg � �0.42 � (1.16 · 0) � (1.23 · 0)�(0.82 · 0) � �0.42 � 0 � 0 � 0 � �0.42

Step 3. Log risks are transformed to risk by using themas exponents of base of natural logarithms as follows:

Risk Pos � e2.79 � 16.281Risk Neg � e�0.42 � 0.657

Step 4. Calculate risk odds ratio (ROR) using data fromStep 3:

ROR � Risk Pos/Risk Neg � 16.281/0.657 � 24.78

Thus, a patient with dizziness, nausea, and diaphoresisis nearly 25 times more likely to have positive HUTT thana patient without these symptoms.

RORs for patients with only one or two positive variablesare shown here:

Positive symptoms ROR Positive symptom ROR

DIZZ & NAU 10.91 DIZZ 4.85DIZZ & DIAPH 7.24 NAU 5.21NAU & DIAPH 7.77 DIAPH 3.46

References1. Linzer M, Yang E, Estes M III, Wang P, Vorperian V, Kapoor W. Diag-

nosing syncope. Part 1: Value of history and physical examination.Ann Intern Med 1997;126:989–996.

2. Benditt D, Lurie K, Fabian W. Clinical approach to diagnosis of syn-cope. Cardiol Clin North Am 1997;15:165–176.

3. Sra J, Jazayeri M, Dhala A, Deshpande S, Blanck Z, Akhtar M. Neuro-cardiogenic syncope. Diagnosis, mechanism and treatment. CardiolClin North Am 1993;11:183–191.

4. Kapoor W. Workup and management of patients with syncope. MedClin North Am 1995;7:1153–1170.

5. Linzer M, Yang E, Estes M III, Wang P, Vorperian V, Kapoor W.Diagnosing syncope. Part 2: Unexplained syncope. Ann Intern Med1997;127:77–86.

6. Grubb B, Kosinski D. Disautonomic and reflex syncope syndrome.Cardiol Clin North Am 1997;15:257–268.

Clinical Predictors for Positive HUTT 291

7. Grubb B, Temesy-Armos P, Hahn H, Eliott L. Utility of upright tilt-table testing in the evaluation and management of syncope of unknownorigin. Am J Med 1991;90:6–10.

8. Grubb B, Kosinski D. Tilt table testing: concepts and limitations. PacingClin Electrophysiol 1997;20:781–787.

9. Brignole M, Alboni P, Benditt D, Bergfeldt L, Blanc J, Bloch P. Guide-lines on management, diagnosis and treatment of syncope. Eur HeartJ 2001;22:1256–1306.

10. Grubb B, Kosinski D, Boehm K, Kip K. The postural orthostatictachycardia syndrome. A neurocardiogenic variant identified duringhead-up tilt table testing. Pacing Clin Electrophysiol 1997;20:2205–2212.

11. Morillo C, Klein G, Zandri S, Yee R. Diagnostic accuracy of a low-dose isoproterenol head-up tilt protocol. Am Heart J 1995;129:901–906.

12. Oraii S, Maleki M, Minoii M, Kafai I. Comparing two different proto-cols for tilt table testing: sublingual glyceryl trinitrate versus isoprena-line infusions. Heart 1999;81:603–605.

13. Raviele A, Giada F, Brignole M. Diagnostic accuracy of sublingualnitroglycerin test and low dose isoproterenol test in patients with

unexplained syncope. A comparative study. Am J Cardiol 2000;85:1194–1198.

14. Almquist A, Goldenberg I, Milstein S. Provocation of bradycardiaand hypotension by isoproterenol and upright posture in patients withunexplained syncope. N Engl J Med 1989;320:346–351.

15. Grubb B, Wolfe D, Themesy-Amos P, Hahn H, Elliot L. Reproducibilityof head-up tilt table test in patients with syncope. Pacing Clin Electro-physiol 1992;15:1477–1481.

16. Alboni P, Brignole M, Menozzi C, Raviele A, Del Rosso A, Dinelli M.The diagnostic value of history in patients with syncope with or withoutheart disease. J Am Coll Cardiol 2001;37:1921–1928.

17. Benditt D, Ferguson D, Grubb B, Kapoor W, Kagler J, Lerman B,Maloney J, Raviele A, Ross B, Sutton R, Wolk M, Wood D. Tilt tabletesting for assessing syncope. J Am Coll Cardiol 1996;28:263–275.

18. Fitzpatrick A, Theodorakis G, Vardas P, Sutton R. Methodology ofhead-up tilt table testing in patients with unexplained syncope. Am JCardiol 1991;17:125–130.

19. Kapoor W, Brant N. Evaluation of syncope by upright tilt testing withisoproterenol. A non-specific test. Ann Intern Med 1992;116:358–363.

20. Kapoor W. Syncope. N Engl J Med 2000;343:1856–1862.