Specific Aims. The comparative risk of adverse depressive symptoms or worsened health-related quality of life (HRQoL) due to beta-blockers (BBs) versus calcium antagonists (CAs) is based primarily on weaker study designs (i.e., case reports and case series, case-control trials) or relatively small studies (i.e., Phase II and Phase III drug approval trials). However, no large-scale, prospective, randomized clinical trials have directly compared CAs with BBs with respect to risk of depression or depressive symptoms or their impact on patient’s HRQoL.

Specific Hypotheses for Objective #1. Six specific two-tailed hypotheses will be tested to meet the study’s objectives. Subjects randomized to the CA treatment arm will be different from those randomized to the non-CA treatment arm on measures of the:

1. level of self-reported depressive symptoms (overall score on the Center for Epidemiologic Studies—Depression Scale [CES-D]);

2. somatic domain of the CES-D;

3. negative affect domain of the CES-D, and;

4. positive affect domain of the CES-D.

Specific Hypotheses for Objective #2. Subjects randomized to the CA treatment arm will be different from those randomized to the non-CA treatment arm on measures of:

5. Functional status;

6. Health-related Quality of Life, especially measures of energy/vitality subscale of the SF-36.

Study significance. Nearly 40 percent of the United States population over the age of 65 is treated for hypertension. The most important treatment modality for hypertension is pharmaceuticals. In 1997, the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure published their Sixth Report. The committee recommended BBs and diuretics as initial therapy for Stage 1 and Stage 2 hypertension. The primary reason for recommending BBs over calcium channel antagonists was that CAs “have not been used in long-term controlled trials to demonstrate their efficacy in reducing morbidity and mortality,” whereas BBs have been shown to be efficacious in long-term trials. JNC-VI also pointed out that when therapeutic effectiveness is equal, consequences on patient’s HRQoL are especially important. One such adverse consequence purported to be associated with multiple antihypertensive medications is mental depression. However, no large-scale, prospective, randomized clinical trials have directly compared CAs with BBs with respect to risk of depression or depressive symptoms, suicidality or their impact on patient’s HRQoL.

The proposed study will clarify if there is excess risk of depressive symptoms and decline in quality-of-life among patients with pre-existing CAD. It will strengthen previous literature by using more rigorous study methodologies and larger sample sizes to obtain adequate power. This study will emend several methodological concerns including:

• Prospective versus retrospective study design;

• Random allocation to treatment group versus observational study design;

• Rigorous case definition using multiple indicators with a focus on the mood symptoms and an ability to differentiate mood from CNS side effects;

• Adequate statistical power; and,

• Evaluation of a population (CAD) already at higher risk depression to determine whether there is excess risk attributable to the pharmacologic treatment.

Study Design. SADD-Sx is an independent, but cooperative adjunct to INVEST. INVEST evaluates only the cardiovascular outcomes of the CA and non-CA hypertension treatment strategies. SADD-Sx has the same randomized trial study design as INVEST. The initial study population is the 16,000 subjects randomized to either the CA or non-CA arms of INVEST. This proposed adjunct study uses standardized self-reported questionnaires to obtain information about depressive symptoms and HRQoL from patients enrolled in the INVEST study.

SADD-Sx inclusion criteria. In addition to the INVEST inclusion criteria, subjects must meet additional criteria before they are included in the SADD-Sx study. First, INVEST's protocol allows for a non-beta-blocker option in the non-CA arm. Subjects must be prescribed atenolol at randomization to be eligible for SADD-Sx. Among the first 900 patients enrolled in INVEST, over 85 percent were prescribed atenolol if they were randomized to the non-CA treatment arm. Second, the subjects must live in the United States or Canada. Third, they must be able to read and understand Spanish or English. All three of the survey instruments have English and Spanish translations. SADD-Sx exclusion criteria. The primary method of data collection is mailed survey, so, subjects must be able to read at the 8th grade level. If not, then they will be surveyed using CATI. To be interviewed they must be able to hear and interact over the telephone. Subjects must have a permanent address or telephone number to be contacted. Patients maintained on pre-randomization non-study antihypertensives will be excluded. Next, subjects must have a permanent address or telephone number to be contacted. Finally, persons not prescribed atenolol or verapamil as part of the non-CA or CA treatment strategy will be excluded.

Recruitment procedures. INVEST investigators will identify subjects who meet our eligibility criteria and send their names and addresses to the SADD-Sx investigators. These persons first will be sent a survey at randomization. They will be sent a follow-up survey within two weeks of the index date. If they do not respond within a week, they will be followed up and recruited by telephone. If they consent, they will be administered a computer-assisted telephone interview.

Sources of Data and Data Collection. SADD-Sx will collect depressive symptom reports and HRQoL from subjects via telephone or mail survey. The primary data collection instruments, in their order of importance, are the (1) Center for Epidemiologic Studies - Depression (CES-D) Scale, (2) Medical Outcomes Study - Short Form 36 (SF-36), and (3) ADL/IADL functional status.

Two "types" of outcome measures are used in this study: (1) depressive symptoms and (2) HRQoL.

Outcome #1. Self-rated Depressive Symptoms: Center for Epidemiologic Studies – Depression Scale (CES-D). The CES-D is a 20-item rating scale designed to measure the respondent’s current level of depressive symptoms. The CES-D is widely used and extensively validated. Scores range from 0 to 60; higher scores indicating more depressive symptoms. Scores 16 or higher are generally used to distinguish persons with a sufficient amount of depressive symptoms so that they closely resemble clinically depressed patients in treatment.

Outcome #2: Functional Status and related Health-related Quality-of-Life measures. It is important to the finding of “depression” that HRQoL and functioning measures be included in the study. The DSM-IV states that “the (depression) episode must be accompanied by clinically significant distress or impairment in social, occupational, or other important areas of functioning. (pg 320). The two scales that measure HRQoL, including measures of patient’s functional status, are included in the questionnaire.

Functional Status: Ability to Perform Activities of Daily Living. The primary measure of functional status will compare differences in ADLs and gives a range of functional activities from mild-to-moderate-to-severe functional impairment. Health-related Quality of Life. The secondary outcome is general health-related quality of life (HRQoL) and it is measured using the Medical Outcomes Study Short Form 36 (SF-36). The SF-36 is a 36-item measure of health status developed for use in clinical research, health policy evaluations and general population surveys that takes about 5 to 10 minutes to complete. The reliability and validity of both the telephone and mail versions of this instrument have been extensively reviewed. The SF-36 measures HRQoL across eight domains and the “vitality” domain measures the extent that persons are feeling energetic and full of pep rather than feeling tired and worn out.

The primary predictor variable is the subject’s hypertension treatment strategy. INVEST subjects are randomly assigned to a non-CA (0) or calcium antagonist treatment strategy (1). The SADD-Sx sample is selected from those in the non-CA arm prescribed a BB. The CA treatment strategy may include addition of HCTZ and ACE inhibitor (trandolapril) if the single drug does not control blood pressure. The non-CA strategy is initiated by use of atenolol. HCTZ and trandolapril may be added if blood pressure is not controlled by the beta-blocker alone.

Control Variables. Control variables are associated with the outcome variables and might otherwise confound the findings were they not considered.

1. Risk factors for depression. Relevant risk factors other than gender include: age, health status, and other medications that may cause depression or other cognitive impairments (e.g., anticholinergics, neuroleptics, anxiolytics, tricyclic antidepressants) at baseline. Questions regarding current and past history of depression are included in the survey.

2. Subject’s baseline on outcome variables. Outcome measures are going to be measured on multiple occasions. The best predictor of a later outcome usually is the subject’s earlier (e.g., baseline) status on the outcome. An important strategy in this study is to evaluate the influence of the hypertension treatment strategy controlling for the subject’s baseline status and increases the statistical power.

Analysis Plan. The analytic strategy uses two approaches. First, data of all patients who complete the study “per the protocol” (PP) will be evaluated. They will have complete data for all of the independent and dependent variables. This strategy uses statistical techniques appropriate for change over time (e.g., comparing mean scores using analysis of covariance. The second strategy analyzes the data using an “intention to treat” (ITT) strategy, including techniques appropriate for censored data (e.g., survival analysis or Cox proportional hazards techniques). This takes into account patients dropping out of the study because of death, inability to contact for follow-up, refusals, or missing data on critical measures. If the results of both strategies are equivalent, then the conclusions are much strengthened. If the results are different, however, then factors potentially associated with “dropping out” of the study (e.g., disease severity, prior psychiatric history) will be determined and examined further for their confounding effects.

Multivariate statistical methods. Familiar methodologies such as regression analysis (including analysis of covariance) and logistic regression will be used to assess the difference in an outcome measure of interest between experimental and control, while adjusting for control variables and baseline measures. To evaluate change over time, we have chosen to use a residuals regression approach, rather than using change scores (from baseline measures) for the reasons outlined in Bohrnstedt. For the regression analyses, control variables will be entered into the equation in the first step and the treatment strategy will be examined second. The change in R2 will be examined. A statistically significant change in R2 indicates that the treatment arms differed and improves prediction of the depressive symptoms and HRQoL outcomes. IF DATA QUALITY PERMITS, models will be fitted using techniques appropriate for the serial, longitudinal nature of the data.

Sample size. This section describes the sample sizes needed to test the hypotheses. It is organized around the two categories of outcome variables (depressive symptoms and HRQoL) and describes the economies gained by using longitudinal data.

Self-rated depressive symptoms. Using baseline data reduces variation and results in a smaller effective sample size. For example, based on measures of mean differences and variation from our previous research, a sample size of 784 subjects in each group would be needed for the CES-D measure without baseline data. However, since this is a longitudinal study and there is significant correlation between subject’s status on baseline and subsequent visits, the sample size needed to detect a significant difference is actually lower.

Radloff found the CES-D test-retest correlation to be 0.54 after six months and 0.49 after 12 months. If the intercorrelation between the baseline and subsequent visit measurement of CES-D were 0.5, the number of subjects needed at the end of the study would be 392 rather than 784 for a small effect size. Assuming a 60 percent completion rate (per protocol group), 653 persons would need to be enrolled in each group at baseline to obtain 392 valid responses at the end of the study. A 60 percent completion rate is conservative because these subjects prior commitment INVEST and should be higher than a typical ambulatory-based population.

Functional status. Thomas and Lichtenstein suggest that one’s score on the functional status measure in a given year is highly correlated with one’s score on the following year (r=0.70). Conservatively, if the correlation between the baseline and subsequent measures of physical functioning were 0.5, the number of subjects needed at the end of the study would be 337 rather than 673 for a small effect size (ES=.20).

Health-related quality of life. The SF-36 is stable over time with correlations ranging from 0.43 to 0.90. Using the smallest 6-month correlation (0.4) and assuming similar effect sizes to depressive symptoms and functional status outcomes, 400 subjects will be needed in each group to detect an effect size of 0.20.

Combinations of drugs. Subjects assigned to the BB or CA treatment will take the same two adjunct medications (hydrochlorothiazide and trandolapril) if the primary medication does not control their hypertension. The INVEST protocol has standardized dosing recommendations. Even so, there will be several combinations of doses and medications possible during the titration and BP stabilization phase. So, while the obstacle of multiple dissimilar antihypertensive combinations is not entirely overcome, it is reduced. Patient's medication combination and dose also will be included in the analysis and there is sufficient sample size to evaluate these issues.

The Work Plan. SADD-Sx will take a little over two years to complete. Patient self-reported baseline data collection can begin as soon as April 1999. Subjects will be enrolled for one-year and they will be enrolled over a one-year time period. So, patients enrolled in Month X will finish in Month X+12 and patients enrolled in Month X+12 will finish in Month X+24. For example, patients enrolled in April 1999 will be closed out in April 2000. Patients enrolled during April 2000 will be closed out in April 2001. Each patient will have 12-months exposure to medication. Baseline data collection will end approximately April 2000. Analysis of baseline data will begin soon after that date, including cleaning, verifying and building data analytic files. This process will be completed by August 2000. Six-week data collection will begin in late May or early June 1999 and will end in late-May or early June 2000. Cross-sectional and longitudinal analytic data files (baseline and six-week) will be built by December 2000. Collection of six-month data will begin in November or December 1999. It will last approximately one year or until November or December 2000. Close-out self-report data collection will begin in January or February 2001 and will last until January or February 2002. After analytic data sets are built, they will be combined with the INVEST automated data set. This process should begin around December 2001 and should last for approximately two months. Once the longitudinal data set containing all of the data is complete, the analysis should take approximately six-months (until July or August 2002) and the final report will be submitted September 30, 2002.