Survival analysis. Probability of remaining on treatment for patients receiving controlled onset, extended release (COER)–verapamil or nifedipine gastrointestinal therapeutic system (GITS).
Physical Symptom Distress Index scores. Change from baseline in SD units for individual symptoms demonstrating a significant univariate treatment effect by treatment group (positive change reflects reduced distress; P = .002 for multivariate analyses of variance). COER indicates controlled onset, extended release; GITS, gastrointestinal therapeutic system.
Top, Psychosocial quality-of-life subscale scores. Change from baseline in SD units for completers by treatment group (positive change reflects improved quality of life). Bottom, Psychosocial quality-of-life subscale scores. Change from baseline in SD units for noncompleters by treatment group (positive change reflects improved quality of life). There was a significant (P = .008) multivariate effect related to dropout status. With univariate analysis, General Health Status was significantly worse (P<.01) for all dropouts vs all completers. COER indicates controlled onset, extended release; GITS, gastrointestinal therapeutic system.
Anderson RB, Hollenberg NK, Williams GH. Physical Symptoms Distress IndexA Sensitive Tool to Evaluate the Impact of Pharmacological Agents on Quality of Life. Arch Intern Med. 1999;159(7):693-700. doi:10.1001/archinte.159.7.693
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
To examine whether the degree of stress associated with adverse physical side effects correlates with overall quality of life (QOL) and compliance rates. To determine if instruments used to assess QOL can detect differences between treatments that have no known central nervous system effects.
Patients and Methods
This randomized, double-blind, parallel group study evaluated 180 to 480 mg of controlled onset, extended release (COER)-verapamil (n = 259) or 30 to 120 mg/d of nifedipine gastrointestinal therapeutic system (GITS) (n = 269) in men and women between 21 and 80 years of age with stages 1 to 3 hypertension. A battery of questions evaluating psychological well-being and a physical symptom distress index was administered after a 4-week placebo washout (baseline) and after 10 weeks of treatment or at dropout.
Both treatments effectively lowered blood pressure, and there were no significant between-group differences in psychosocial QOL. A difference in the level of physical symptom distress was detected between treatments (P = .002; multivariate analysis of variance), with 7 significant univariate treatment effects, all favoring COER-verapamil, being noted—pedal edema, polyuria, rapid heart beat or palpitations, hives, muscle cramps, abdominal cramps, and headaches. Constipation-related distress increased significantly (P = .001) but to a similar extent with both treatments. The difference in symptom distress tended to predict compliance as there were more withdrawals in the nifedipine GITS group (n = 85) vs COER-verapamil group (n = 64) (P = .08).
Patient-assessed physical symptom distress is a sensitive, simple technique to evaluate the effect of antihypertensive medications on QOL and tolerability, as shown by its ability to detect the improvement associated with COER-verapamil. Depending on the agents involved, the Physical Symptom Distress Index may more closely predict dropout rates than the traditional psychosocial instruments, as suggested by the lower dropout rate in the COER-verapamil group. Thus, in studying treatment effects on QOL, both the distress of physical symptoms and the impact of psychosocial factors should be evaluated.
THE CLINICAL management of conditions in which symptoms are a significant component of the disease and symptom relief is the goal of treatment is simplified by the fact that symptoms provide a reminder that treatment is required. In the treatment of conditions such as hypertension, where the goal is prevention of end-organ damage through the long-term control of blood pressure and where symptoms are often absent, treatment can be more complicated. In this clinical setting, perceptions of adverse effects of medications by the patient play a prominent role in the success of treatment, as adverse effects may lead to partial or total noncompliance with a prescribed treatment regimen. Among patients who are compliant with their medications, distressing side effects of the drugs can adversely affect their quality of life (QOL) in a number of ways, including daily function, work performance, emotional status, and other important dimensions of life.1,2 As the adverse effects profile of the available agents has improved, the influence on QOL has become more subtle and thus has led to substantial interest and investigation into the assessment of QOL in therapeutic trials.2- 5
Health-related QOL refers to an individual's ability to perform a range of roles in society and to reach an acceptable level of satisfaction from functioning in those roles.6,7 There is general agreement that a comprehensive and universal QOL assessment is unlikely to ever be achieved, probably because the precise elements contributing to QOL can vary substantially in different individuals. The definition of a "good QOL" almost certainly varies with 1 of 2 broad mechanisms. One mechanism involves a direct influence on the central nervous system, with agents such as methyldopa, reserpine, and β-blockers.8 For example, β-blockers can influence affect by reducing anxiety and/or by increasing depressive symptoms.8,9 As a second mechanism, the distress induced by physical symptoms can also influence QOL, which can modify mood. Calcium channel blockers are widely held to be free of direct central nervous system adverse effects, making differences in QOL more likely attributable to the latter mechanism. Thus, a comparison of 2 widely used and studied calcium channel blockers, nifedipine and verapamil, provides an opportunity to examine the influence of physical as opposed to psychosocial distress on QOL. The goals of this study were to compare treatment effects on a physical symptom distress index, psychosocial indexes, and dropout rates.
This was a multicenter, double-blind, randomized, parallel group, placebo lead-in, dose titration study comparing controlled onset, extended release (COER)–verapamil with nifedipine gastrointestinal therapeutic system (GITS). The study consisted of 3 periods: (1) a 1- to 2-week pretreatment period, when screening procedures were completed and antihypertensive medications were tapered and discontinued, (2) a 2- to 4-week single-blind, placebo lead-in period, during which antihypertensive medications were washed out and eligibility for randomization was established, and (3) a 10-week, randomized, double-blind, double-dummy, active treatment period with patients randomized to either COER-verapamil (Covera-HS; G.D. Searle & Co, Skokie, Ill) or nifedipine GITS (Procardia XL; Pfizer Inc, New York, NY). The initial doses were 180 mg/d of COER-verapamil taken at bedtime or 30 mg/d of nifedipine GITS taken within 1 hour of awakening. If a patient's blood pressure was higher than 140 mm Hg systolic or higher than 90 mm Hg diastolic after 2 weeks of treatment, the dose was escalated. Dose escalation was continued every 2 weeks at 180, 240, and 360 to 480 mg for COER-verapamil or 30, 60, and 90 to 120 mg for nifedipine GITS until blood pressure was lower than 140/90 mm Hg or the maximum dose was reached.
Blood pressure was measured by an ambulatory blood pressure monitoring system (Spacelabs Model No. 90207; Spacelabs Inc, Redmond, Wash).10 Recordings were obtained over a 24- to 26-hour period at baseline and at the final visit. The primary end point was the average blood pressure in a 4-hour period defined as 1 hour before and 3 hours after awakening. Abbreviated results of the ambulatory blood pressure and pulse rate measurements are presented in this article. Additional details regarding cardiovascular responses to treatments have been presented elsewhere.11
Five hundred twenty-eight patients were randomized and participated in the QOL evaluations. Patients were between 21 and 80 years of age with stage 1 through middle stage 3 hypertension defined as untreated seated systolic blood pressure of 140 mm Hg or higher and lower than 190 mm Hg or seated diastolic blood pressure of 90 mm Hg or higher and less than 115 mm Hg. Patients were excluded if they had known coronary artery or cerebrovascular disease within the previous 6 months, a history of congestive heart failure, cardiac arrhythmias, secondary hypertension, renal allograft, sick sinus syndrome, or second- or third-degree atrioventricular block, serum creatinine 221 µmol/L (>2.5 mg/dL), a history of ketoacidosis or poorly controlled diabetes mellitus, or active hepatitis. They were also excluded if they regularly used any medication that would interfere with the metabolism of verapamil or nifedipine. The patients were required to read and understand the English language at least at the sixth-grade level. The study protocol and informed consent form were approved by a human ethics committee at each study center, and each patient signed a written informed consent before enrolling in the study.
The procedures and instruments used in this study were similar to those previously employed.8,9,12 In brief, a standard set of questionnaires was administered by specifically trained personnel. The battery of questionnaires included psychosocial measures (anxiety, loss of behavioral/emotional control, depression, emotional ties, general positive affect, general health status, and vitality) and a Physical Symptom Distress Index (PSDI).
The PSDI consisted of 73 items. However, 2 of the items were not used in this analysis because they applied only to men. The questionnaire presented a list of symptoms associated either with hypertension or as an adverse effect from 1 or both drugs. Each question asked the respondent, "Have you had this symptom at all, in the past month?" The respondent was to circle a number between 0 ("No, not in the past month: Didn't have it at all") and 5 ("Yes, and it has bothered me extremely"). This was a modification of the PSDI previously published.8,13
The psychological scales relied heavily on the Rand Mental Health Index including its 5 subscales of anxiety, loss of behavioral/emotional control, depression, emotional ties, and general positive affect.14 Seven other items outside the Rand Mental Health Index, published by Dupuy15 as the Psychological General Well-Being Schedule, provided 2 additional subscales (general health and vitality). The battery of instruments was administered at the time of initial enrollment in the study (to acclimate the subject to the testing procedure) at the end of the placebo washout and at the completion of the study or at the time of dropout.
Sample size was chosen based on blood pressure such that a sample size of 267 randomized patients per group provided at least 80% power to detect a between-group difference of 3 and 5 mm Hg in diastolic and systolic blood pressure, respectively.
The efficacy analysis was conducted on an intent-to-treat basis, with missing values being handled using the "last observation carried forward" method. Blood pressure and pulse rate analyses were conducted using analysis of variance with baseline value, treatment, and center included in the model. All demographic and efficacy analyses were conducted using SAS statistical software Version 6.09 (SAS Institute Inc, Cary, NC).
A survival curve was generated for the proportion of randomized patients for each treatment group remaining in the study. The Lee-Desu statistic was used to test the statistical significance of the divergence between the 2 treatment group curves.16
Quality-of-life analysis was conducted on an intent-to-treat basis. To be included in the analysis, the patient had to have a valid baseline and end-point QOL assessment. Treatment effects on baseline to end-point change in QOL were measured by 2 multivariate 1-way analyses of variance (MANOVA), 1 for the physical symptom distress and 1 for the psychosocial index. The multivariate F, as estimated by Pillai trace,17 was used as an exact test of significance in each analysis. If a main MANOVA detected an overall multivariate effect, the associated univariate effects were analyzed. The foregoing MANOVA also was repeated as a 2-way factorial analysis of treatment and dropout on symptom distress and psychosocial well-being. In addition to performing a MANOVA on the change score, a similar MANOVA was performed on the responses of the QOL questionnaires at the conclusion of the study in a cross-sectional analysis. All QOL analyses were conducted using SPSS.18
At baseline, the mean QOL values were similar between treatment groups (Table 1) and were consistent with the norms in the US population and those observed in previous studies.8,9,12- 15 Similarly, there was no significant difference in blood pressure or heart rate at baseline (Table 2). As anticipated, both treatments effectively lowered blood pressure. Details of these data are reported elsewhere.11 In brief, there was a highly significant treatment effect on the average 4-hour (1 hour before and 3 hours after awakening) systolic and diastolic blood pressure (P<.001). Diastolic blood pressure was decreased to a similar extent in each treatment group. Systolic blood pressure was decreased to a greater extent in the nifedipine GITS group compared with the COER-verapamil group (P<.001). Pulse rate decreased in the COER-verapamil group and increased in the nifedipine GITS group (P<.001 between groups). The double-product (pulse rate multiplied by systolic blood pressure) was decreased to a greater extent in the COER-verapamil group compared with the nifedipine GITS group (P<.001). Twenty-four percent of patients treated with COER-verapamil (n = 66) and 31% of patients treated with nifedipine GITS (n = 87) discontinued prematurely from the study (P = .06). For both, major reasons for withdrawal—treatment failure and adverse signs or symptoms—nifedipine GITS was worse: 45 (16%) vs 33 (12%) for treatment failure and 24 (9%) vs 18 (7%) for adverse events. Thus, both occurred one third more frequently in the nifedipine GITS group. Furthermore, most of the difference in adverse events can be accounted for by patients in the nifedipine GITS group withdrawing because of edema (6 vs 0), with no difference in withdrawals due to gastrointestinal problems (4 in the nifedipine GITS group and 5 in the COER-verapamil group). Finally, withdrawal secondary to severe adverse events was nearly twice as common in the nifedipine GITS group (9 vs 5). A survival curve depicting the proportion of randomized patients remaining in the study over time is presented in Figure 1.
A 1-way MANOVA tested simultaneously the effect of treatment on the change in distress for the 71 physical symptoms. This analysis detected a significant overall multivariate difference between the treatments (P = .002). Univariate analysis subsequently revealed 7 significant (P<.05) treatment effects (Figure 2). As anticipated, swelling of feet or ankles was substantially different between the 2 treatments (P<.0001). There was no change in the distress level related to the symptom of swelling of the feet with the use of COER-verapamil, while there was a substantial increase in distress in the nifedipine GITS-treated patients (Figure 2). Rapid heart beat and headaches were significantly reduced in the COER-verapamil-treated patients with nifedipine GITS producing no changes. Thus, some of the overall differences in the PSDI may have been secondary to a reduction in the distress of some symptoms compared with baseline in the COER-verapamil-treated patients.
The change in the distress associated with the 71 symptoms also was subjected to a 2-way MANOVA by treatment and discontinuation. This analysis revealed a significant multivariate main effect for each (treatment, P = .001; discontinuation, P = .039), but no significant multivariate interaction effect (treatment-by-discontinuation, P = .689). The same 7 symptoms as shown in Figure 2 achieved univariate significance with 1 additional symptom, "difficulty staying asleep," achieving marginal significance (P = .04). Again, all significant differences favored the use of COER-verapamil. These analyses suggest that the level of distress was increased by both the type of treatment (nifedipine GITS was significantly worse than COER-verapamil) and the patient status (dropouts were significantly worse in both treatment groups).
The distress secondary to constipation merits special emphasis. Contrary to what had been anticipated, constipation increased significantly (P<.001) in both groups. However, there was no difference between the treatments for the mean ± SEM change from baseline for constipation (−0.331 ± 0.074 vs −0.309 ± 0.064 for COER-verapamil and nifedipine GITS, respectively).
Finally, there was a substantial difference between spontaneously reported adverse events and the distress of symptoms as documented by the patients when specifically queried on the symptom checklist. Consistent with previous experience, edema was spontaneously reported as an adverse event significantly more often in the nifedipine GITS group (12.8%) compared with the COER-verapamil group (1.4%) (P<.001).8 Also as expected, constipation was reported as an adverse event significantly more often in the COER-verapamil group (14.9%) compared with the nifedipine GITS group (8.2%) (P<.05).10 Palpitations were spontaneously reported as an adverse event in a relatively small percentage of patients in both treatment groups. In contrast to spontaneously reported adverse events, the number of individuals reporting a distress level for constipation, palpitations, or edema of moderate to severe extent was substantially higher (Table 3).
In contrast to the PSDI, there was no treatment-related effect on the 7 psychosocial subscale change scores from baseline using a multivariate analysis (P = .37). However, there was a treatment by dropout interaction. A 2-way MANOVA, including dropout as a factor with drug treatment, detected a multivariate tendency for dropouts and completers to differ (P = .008). Only 1 (General Health Status) associated univariate main effect of dropout achieved significance (P = .005) (Figure 3). General Health Status was generally maintained in completers and declined in dropouts in both treatment groups, with no statistical difference being observed between treatment groups. Finally, the change in QOL scores induced by either therapy did not correlate with either the baseline psychosocial scales or PSDI. This is perhaps not surprising since, in general, the QOL of the patients enrolled in this study was good, as evidenced by their psychosocial scales and very low level of symptom distress at baseline. On average for the 71 symptoms, the mean baseline PSDI was between "no symptoms" and "symptoms present but not distressing."
An analysis was performed to determine the sensitivity of the psychosocial subscales to changes in symptom distress. The 7 symptoms that demonstrated a significant univariate treatment effect and the symptom "constipation" were divided into 2 groups for all patients, independent of their treatment. One group reported at least some distress while the other group reported no distress. The General Health Status subscale of the psychosocial scales was determined in each group. For each of these 7 symptoms, there was a highly significant (P<.001) difference in the General Health Status with lower (worse) scores being noted for patients reporting distress compared with those patients not reporting distress (data not shown). Thus, a substantial increase in symptom distress was associated with a deterioration in psychosocial QOL as measured by the General Health Status subscale.
Our study was designed to determine if QOL instruments can detect differences in commonly prescribed antihypertensive medications that probably have no central nervous system adverse effects. As anticipated, nifedipine GITS and COER-verapamil did not have a significant effect on psychosocial indexes. However, the instruments were sensitive enough to document a highly significant treatment effect on the self-reported distress produced by physical symptoms. More specifically, the distress induced by nifedipine GITS being significantly greater than for COER-verapamil. Indeed, except for constipation, there was no significant increase in symptom distress in the patients treated with COER-verapamil. Even for constipation, although the frequency was twice as great with COER-verapamil, the distress level was not significantly greater for COER-verapamil than it was for nifedipine GITS. Whether this divergence from the commonly held belief regarding verapamil and constipation is due to the formulation of verapamil or the sensitivity of the PSDI is unknown. The PSDI also tended to predict withdrawal from treatment better than the more commonly used psychosocial QOL indexes. These results agree with a previous QOL report by Palmer et al.19 Their study was smaller (n = 81) and the subjects were evaluated after a longer treatment period (4 months). Yet, they also reported an increase in symptoms in the subjects taking nifedipine. Furthermore, there was a suggestion of a difference in psychosocial QOL indexes, reaching significance (P = .05) for the cognitive function subscale, with nifedipine being worse than verapamil.
The utility of a QOL study in hypertension is based not only on its ability to distinguish adequately the influence of different medications on QOL but equally important the confidence one has in the conclusion that there is no difference between agents. For a study to be interpretable, the instruments employed must have been shown to be useful, the study design appropriate to address the issues raised, and an adequate number of patients enrolled in the study. The QOL studies also may reach false-negative conclusions if they are underpowered or use instruments inadequate to identify differences between agents.
This study avoids many of these pitfalls and constitutes a rigorous assessment of the effect of 2 widely prescribed calcium channel blockers on QOL using appropriately validated and sensitive QOL instruments. For example, in our study, if traditional psychosocial QOL indexes were used, no treatment effect would have been observed. Although the term validity is widely used in the psychosocial area, much of the validation is statistical or technical, and does not ascertain whether or not an instrument is adequate to the task.2,7 The instruments used in this study have been "validated by application" in a series of investigations.8,9,12 In a comparison of captopril, propranolol, and methyldopa, these instruments showed a QOL profile that paralleled withdrawal rates and symptom patterns.8 In a comparison of atenolol with nifedipine, the QOL assessment predicted the dropout rate.9 A similar finding was observed in this study with a negative effect on the psychosocial general health status being related to dropout status. In a comparison of captopril and enalapril, a strong correlation was shown between QOL changes with treatment and the simultaneous occurrence of spontaneous life events, coded prospectively.12 This latter study also documented that a change of 0.15 SD units for QOL indexes correlated with clinically meaningful life events.12 This experience has made it clear that the tools used in the present study are useful in the assessment of QOL and detected clinically meaningful treatment effects in patients with hypertension.
It is also important to use broad-based QOL instruments rather than those that limit the scope of the evaluation to areas one assumes may be affected by the agents based on previous spontaneously reported adverse events. By using this broad-based approach, it was possible to identify QOL changes and/or treatment differences for symptoms, such as increased urination, hives or swelling, or muscle cramps, that produce meaningful distress from the patients' perspective when they were specifically queried.
Assessment of the effect of medical therapy on physical symptoms has long been used to judge both efficacy and adverse effects. Presumably the latter will strongly influence compliance rates. However, several lines of evidence suggest that the traditional approach to measure adverse effects, ie, the events reported by the health care provider, likely underestimates the magnitude of the problem. In 1986 we reported a modification of this approach. Instead of asking the health care provider, we asked the patients if they had experienced 1 or more of 13 symptoms.8 Furthermore, if a symptom was present, we asked the patient to define the level of stress on a 4-point scale. This crude symptom distress index was sufficiently sensitive to detect a difference in the effect of methyldopa and propranolol vs captopril (P<.05). However, the sensitivity was not as great as with the psychosocial index (P<.01). An expanded assessment of the physical symptoms phase of this trial has been reported by Schoenberger et al.20 They documented that adding diuretic therapy to the 3 treatment groups substantially increases the distress of the physical symptoms, particularly for propranolol. When given as monotherapy, captopril actually numerically, but not statistically, reduced symptom distress. Although not specifically addressed, the data by Croog et al8 and Schoenberger et al20 are in agreement with our report regarding the correlation between an increase in physical symptom distress and decrease in psychosocial QOL. The data reported by Weir et al21 are also in agreement.
Such symptom distress may also impact negatively on long-term medication compliance. In this study, there was a marginal treatment effect on dropout rates (P = .08). It is possible that with a longer study a significant difference would have been documented. Of interest, both major factors contributing to the dropout rates—treatment failures and adverse events— were one third greater in the nifedipine GITS group. The reason(s) for the higher treatment failure rate in a 10-week study is (are) unclear. Whether the failure was a surrogate marker for an adverse event is not known. However, the increased withdrawal for adverse events is consistent with the QOL data, as long as one did not limit the analysis to the data obtained from the psychosocial questionnaires. Using a full range of instruments, the PSDI was significantly related to treatment, thereby predicting dropout rates.
How sensitive was the Symptom Distress Index as a measure of the impact of treatment on the overall QOL of the patient? As documented in this study, if symptom distress is increased secondary to drug therapy, there is a corresponding decrease in the psychosocial QOL. As predicted, the instruments were able to detect a substantial difference in the distress of ankle edema, from the patient's perspective, in those treated with nifedipine GITS vs those with COER-verapamil. Of relevance, the Symptom Distress Index identified an adverse effect substantially better than did the spontaneously adverse events as reported by the physician. For example, on the adverse event report, constipation was noted nearly twice as often in patients receiving COER-verapamil than those receiving nifedipine GITS. Yet, the distress due to constipation was not significantly different in the 2 groups.
How applicable are the results of this study to the general hypertensive population? The age and sex distribution are equivalent to the US population. Thus, the results are likely to be generally applicable. However, the racial composition of the study group likely reflects a somewhat larger proportion of whites. Finally, would the differences be more dramatic if a longer study had been performed? We have previously suggested that the maximum effect of antihypertensive therapy on QOL requires at least 16 to 18 weeks of therapy.8,9,12 To the extent that the PSDI is influenced similarly, we assume a longer study would have produced even more dramatic results.
In conclusion, the PSDI was found to be a sensitive technique for assessing the impact of antihypertensive therapy on QOL, as shown by its ability to detect the improvement associated with COER-verapamil. Furthermore, depending on the agents involved, it may more closely predict withdrawal rates than a traditional psychosocial QOL instrument. Thus, in studying treatment effects on QOL, both the distress of physical symptoms and the change in psychosocial factors should be evaluated. This study also adds to the growing body of evidence suggesting that the effect of drug therapy on QOL is an appropriate and accurate predictor of the future compliance of patients to a specific form of therapy.
Accepted for publication July 14, 1998.
This study was supported by a grant from G. D. Searle & Co, Skokie, Ill.
Presented in part at the American Society of Hypertension Annual Meeting, San Francisco, Calif, May 29, 1997.
Reprints: Richard B. Anderson, EdD, Commensa Inc, 86 Appleton St, Arlington, MA 02176.
Fulton, NY: Zaeem Ansari, MD.
The Minerva Consertal, Sacramento, Calif: Najam Awan, MD.
Fair Lawn, NJ: Robert Berkowitz, MD.
Arroyo Seco Medical Group, Pasadena, Calif: Edmond Clinton, MD.
Encinitas, Calif: George Dennish, MD.
Vancouver, Wash: Rohit Desai, MD. New England Clinical Research, Hampton, NH: Stephen Dyke, MD.
Rush Presbyterian–St Lukes' Medical Center, Chicago, Ill: William J. Elliott, MD, PhD.
Southern California Research, Whittier: Rob Fiddes, MD.
Albert Einstein College of Medicine, Bronx, NY: William Frishman, MD. Lawrence Clinical Research, Lawrenceville, NJ: William Garland, MD.
Division of Clinical Pharmacology, University of Southern Florida, Tampa: Stephen Glasser, MD.
Fort Wayne, Ind: Peter Hanley, MD.
Oak Ridge Medical Clinic, Lake Oswego, Ore: H. Freeman Harris, MD.
University of Miami School of Medicine, Miami, Fla: Nadar Jallad, MD, Kevin Ng, MD.
Oregon Research Group, Eugene: Kirk Jacobson, MD.
Health America, Pittsburgh, Pa: Kjel Johnson, PharmD.
Diablo Clinical Research Inc, Walnut Creek, Calif: Richard O. Kamrath, MD.
Christ Hospital/Lindner Center, Cincinnati, Ohio: Dean Kereiakes, MD.
Clinical Physiology Associates, Fort Meyers, Fla: Ronica Kiuge, MD.
Health Care Centers Inc, Newark, Del: Frederick Lahvis, MD.
Southern Drug Research, Blue Cross Blue Shield of Alabama, Birmingham: Barry K. McLean, MD.
Trenton, NJ: Rafael Levites, MD.
Omega Medical Research, Providence, RI: Dennis Mikolich, MD.
Clinical Therapeutics Corp, Coral Gables, Fla: Raphael Montoro, MD.
Orange County Heart Institute and Research Center, Orange, Calif: Joel Neutel, MD, MA, Michael A. Weber, MD.
Elkhorn, Wis: Imran Niazi, MD.
Future Scripts, Spokane, Wash: David Oakes, MD.
Veterans Affairs Medical Center, Washington, DC: Vasilios Papademetriou, MD.
Hermann Center for Chronobiology & Chronotherapeutics, Houston, Tex: Ronald Portman, MD.
Trinity Diagnostic Associates, Carrollton, Tex: Henry Punzi, MD.
Medical College of Virginia, Richmond: Domenic Sica, MD.
Louisiana Cardiovascular Research Center Inc, New Orleans: William Smith, MD.
Quality Healthcare Concepts, Eugene, Ore: David Strutin, MD.
Southern Drug Research, Health Partners, Birmingham, Ala: Gregory Sullivan, MD.
Deaconess Hospital, Institute for Prevention of Cardiovascular Disease, Boston, Mass: Geoffrey Tofler, MD.
Western Clinical Trials Inc, Whittier, Calif: Colin Walker, MD.
Rochester Clinical Research, Rochester, NY: Mervyn Weerasinghe, MD.
Androscoggin Cardiology Associates, Auburn, Me: Robert Weiss, MD.
University of Conneticut Health Center, Farmington: William B. White, MD.
Atlantic Institute of Clinical Research, Florida Health Care Plan, Daytona Beach: David Williams, MD.
Cardiology Research Associates, San Diego, Calif: Lawrence G. Yellen, MD.