1. Primary Endpoints

1.1. Lipid measurements

Serum lipids function both as the primary study endpoint and the means by which we assess study eligibility.  Pt eligibility is determined by sequential measures (see D.5.1.). Such duplicate determination for eligibility recognizes the consider­able natu­ral variation in individual cholesterol lev­els. (109) If the pt is eligible and consents to participation, duplicate fast­ing lipid profile samples drawn one week apart will be sent to the CDC-standardized reference labora­tory of Dr. Robert Nicolosi at UMass-Lowell. The mean of the results of these two venous LDL-C determinations will be used as the study baseline levels.

All physicians will have the screening fingerstick lipid profile results available at the time of the scheduled initial visit. They will not receive the results of the study baseline venous samples sent to UMass-Lowell. At 12 months all pts will have a final endpoint venous lipid profile sent to the reference laboratory. The protocols for cholesterol screening will follow the “Recommendations for improving cholesterol measurement: A report from the laboratory standardization panel of the National Cholesterol Education Program.” (110)

Screening Instrument: The Cholestech^® analyzer will be used for fingerstick cholesterol and lipid profile screening. It is both accu­rate and convenient. We used and validated this instrument in WATCH. (76) Each analyzer has been documented to achieve ± 3% for accuracy and preci­sion as compared to the CDC-stan­dardized laboratory setting.

Instrument Operators: Licensed laboratory personnel experienced in the use of the Cholestech^® analyzer.

Quality Control-UMMC: Control sera at 2 levels which bracket the 200 and 240 mg/dl decision values are run on the Cholestech^® analyzers at the beginning of each screening day. If bias is more than 5% from either control serum value, the pools are run again, and if the bias is still greater than 5%, the instrument is taken out of service until the problem is cor­rected.

Lipoprotein Measurements: Prior to obtaining blood the pt will sit for 5 minutes, since postural changes can alter serum cholesterol concentrations. Venous blood samples from an antecubital vein will be collected into 7 ml vacutainer tubes without anticoagulant, and will be centrifuged to harvest serum after separation from the clot within 2 hours. The sera will be stored at -80^o C. until transported to Dr. Nicolosi’s laboratory on a weekly basis, where they will be ana­lyzed within 1 week. Total and HDL cholesterol and TG will be analyzed using the Abbott VP Autoanalyzer and Sigma reagents. HDL-C will be measured in the super­natant after magnesium-phosphotungstate precipitation of apo B-containing lipoproteins. All assays have met the standardization criteria of the CDC-NHLBI Lipid Standardization Program. (111) A copy of a recent review of the labora­tory’s performance is included as appendix L. LDL-C will be determined indirectly as follows: LDL-C = TC - HDL-C - TG/5. (112)  LDL-C will also be measured directly utilizing the RDI LipiDirect^® Magnetic LDL Reagent kit (Reference Diagnostics Inc., Bedford, MA) (see appendix L). Briefly, this FDA-approved methodology precipitates LDL-C while leaving HDL-C and VLDL in the supernatant solution. LDL-C is obtained directly by subtracting the cholesterol concentration of the supernatant from the total cholesterol.  Therefore, in our assay of total serum LDL, 400µl of Magnetic LDL Precipitating Reagent is added to a test tube. Then 250µl of serum or appropriate reference control are added.  Samples are vortexed immediately and allowed to stand at 18 - 30^o C for 10 minutes.  Each tube is then placed on a magnetic surface for 5 minutes to allow for complete sedimentation of LDL-C. An aliquot of the supernatant is then assayed for cholesterol using the assay described above. Direct LDL-C measurement will be the primary study lipid endpoint. Total cholesterol, HDL-C, and TG will be recorded and analyzed as additional endpoints.

1.2. Dietary measurements

The premise of this study is that an intervention which leads to a reduced consumption of dietary fat, in particular SFA, will result in decreased LDL-C levels.  Therefore, assessment of diet, in particular dietary fat consumption, is of primary importance.  Because of the reduced probability of bias and concomitant ability to accurately characterize group mean nutrient intake (113-115) in relation to other assessment methods we will use computer-assisted telephone interview (CATI) 24 hour recalls (24HRs) as the assessment method for the primary dietary endpoints. A single 24HR will be conducted on a randomly selected day at the 12-month point. The 24HR will be administered within the 3-week period prior to the pt’s 12-month endpoint serum lipid measurement. The 24HRs will be administered by non-intervention RDs (currently working in our group) blinded to pt condition and trained to collect dietary data using our interview system. The 24HR-derived data is analyzed using the Nutrition Data System (NDS 2.9) (see below). Though a single 24HR is not suitable for assessing individual intake, it is appropriate for determining group means. (65) Because of the large contribution of intra-person variability in a single 24HR, analysis of change scores based on pairs of nutrient scores derived from single 24HRs results in large total variance and limited ability to detect an effect (see 11.2). (65) Analyses based on comparison of one-year data by condition are more powerful than analyses of change scores.

We also will use the 7DDR developed in WATCH, for assessment of individual change. As noted (see C.1.7. and appendix C), the 7DDR-derived nutrients agree closely with those derived from multiple 24HR. (73, 79, 88, 89) Such use allows us to examine the relation between individuals’ reported dietary change and covariates that could modify the effect of the intervention, produce an independent effect, or act as a biaser.  Such variables include education level, self-efficacy, motivation, social desirability, and stage of change.

The 7DDR also will be used by the RDs in the GNI, and by the DLMCs for telephone counseling.

The 7DDR will be given at baseline and at the 12-month point for endpoint study measures, and variably for clinical use during the year as determined by the GNI, the DLMC, and the pts’ physicians. The endpoint study 7DDRs will not be available clinically, and the clinical 7DDRs will not be used in endpoint analyses.

The Nutrition Data System/Nutrient Database

As in WATCH, for analysis of 24HRs and 7DDRs we will use the University of Minnesota Nutrition Coordinating Center's Nutrition Data System (UMNCC-NDS) software. (116) For 24 HR interviewing, this easy-to-use software accomplishes a number of data management tasks, checking for errors in range and logic and prompting interviewers for information and corrections. The system is well suited to open-ended assessment techniques such as 24HRs that demand matches for a wide array of specific food items. At the end of the interview it produces an analytic data file.

The UMNCC-NDS consists of data entry and analysis software and comprehensive food nutrient databases. It was developed in 1974 to support nutrient analyses for randomized intervention trials, and has become the leading U.S. resource for nutrient database and analysis systems for research. The database contains over 16,000 foods and 5,000 brand name products and values for 93 nutrients, allowing over 150,000 food variants, differing in preparation methods and ingredients. The database includes culturally unique foods. It is updated at least annually. We have contributed data on an assortment of foods to the NDS from our on-going studies.

2.  Secondary Endpoints

Secondary study endpoints (see section D.2.) include dietary total fat intake and body weight, and implementation of the counseling sequence by the physicians. As for SFA, total fat intake is measured utilizing the techniques described above in D.10.1.2. Body weight is measured using standard methodology, with the pt wearing only light clothing and no shoes. Physician implementation of the counseling sequence is measured using PEIs, as follows:

2.1. Survey of Patients: Interventions Delivered:

As in WATCH, during the clinical trial phase of this project a random 25% of study pts will be asked to complete a brief PEI (appendix J) assessing the type and extent of nutrition intervention provided by their physicians. All physicians receive the same training and office support program whose value we have previously demonstrated (see C.1.2), (62) and the primary purpose of the PEI here is to evaluate the effect of systems-based feedback on physician behavior. Therefore PEIs will not be conducted on the initial physician-pt encounter, but only on follow-up visits. The PEI questionnaire will be administered by telephone within 2 days following the clinic visit. In WATCH we used both in-person and telephone PEIs and obtained comparable scores. Standardizing on telephone PEIs in this study will minimize logistic difficulty and cost.

The PEI includes the critical intervention steps important for effective dietary intervention: 1. discussed cholesterol level; 2. discussed diet-cholesterol connection; 3. advised dietary change to lower cholesterol; 4. discussed past efforts to lower cholesterol; 5. discussed problems making dietary change (barriers); 6. discussed solutions to problems (resources); 7. Pt agreed to make changes or MD discussed goals; 8. gave nutrition materials; 9. referred pt for nutrition counseling; 10. set a follow-up contact. A score is determined for each pt encounter by adding up the scores for each step reported, and normalizing to a maximum score of ten. (62)

Although physicians will not be notified specifically of the status of their pts, they are likely to be aware that pts are in the Intervention Condition because of reports from the DLMC and potential comments by the pts re phone contacts and mailings. We are especially interested in seeing if the practices of the physicians are influenced, as judged by PEIs, by such prompts, and we hypothesize that there will be a favorable effect on PEI scores for condition II pt/physician encounters. There are 3 possible outcomes:

• 1.      The physicians may be unaffected by condition II feedback. In this case, PEI scores will be identical in both conditions, and the rate of GNI referral in condition I will be low.
• 2.      The physicians may be favorably affected by condition II feedback, but only as it applies to their interaction with the condition II pts. In this case there will be a differential effect, with PEI scores higher in physician encounters with condition II pts. We hypothesize that this will be the likely outcome.
• 3.      The physicians may be favorably affected by condition II feedback, and this effect may spill over into their interactions with condition I pts. In this case PEI scores in condition I pt encounters should progressively increase over the course of the study, as should the rates of referral of condition I pts to the GNI. Although this would be a desirable outcome (although probably analytically difficult), akin to safeguarding an entire population by immunizing half, we believe this is unlikely to occur. It is in part the complexity of the potential effects of the intervention on physician behavior that has led us to consider PEI outcomes only as a secondary endpoint.

3. Other Variables that are potential modifiers of the intervention effect

These include variables that may change as a consequence of the intervention (i.e., intervening variables).  They are noted below with an asterisk (*) (also see Intervention Framework figure in Appendix B).

3.1.  Physician Nutrition Knowledge and Attitudes

3.2 Patient variables

3.2.1.  Data Collected at Baseline only

3.2.1.1. Patient Demographic information: age, gender, education, occupation, ethnicity, household data.

3.2.1.2. Social Desirability and Social Approval

Social desirability (SD) is the defensive tendency of individuals to portray themselves in a manner adhering to social norms. (118, 119) Persons scoring high on the Marlowe-Crowne Personal Reaction Inventory (MCSD) tend to have acquiescent personality types that may, in turn, modify their susceptibility to interventions. (119) We have shown that increased SD scores are associated with a downward bias in the estimation of nutrient intake. (73) Because SD can function either as a biaser of nutrient intake estimates, or as a proxy for acquiescent personality type, or both, we will measure SD using the well-established MCSD. The MCSD is a 33-item true/false subscale of the Minnesota Multiphasic Personality Inventory. The internal consistency coefficient (from the Kudor-Richardson Formula) is 0.88 and test-retest reliability is 0.89. (120) Social approval (SA), the tendency to seek a positive response in a testing situation, is less focused on defensiveness. It is assessed by the 20 Likert-scale questions of the Martin-Larsen Approval Motivation Scale. (121) In the WATCH we found that responses to the 7DDR were biased by social approval. (77)  Time to complete: 8 minutes.

3.2.2. Data Measured at Baseline and One Year

The following variables are included in the baseline and one-year questionnaires:

3.2.2.1. Nutrition Stage of Change*

While we are listing this as a potential predictor or modifier, Nutrition Stage of change could also be a secondary outcome with a greater change in condition II as compared to condition I pts. Nutrition Stage of change is categorized by 2 questions concerning the individual's intention to change his nutrition be­havior, and his recent history of attempted change. (101) The responses place the pt in one of 4 stages:

Precontemplation:  The subject is not considering changing nutritional behavior within the next 6 months.

Contemplation:  The subject is seriously considering changing nutritional behavior within the next 6 months.

Ready for Action:  The subject is seriously considering changing nutritional behavior in the next month.

Action:  The subject reports having changed nutritional behavior within the past 3 months.

3.2.2.2.  Nutrition Change History

Each of the items to be used to assess nutrition change history have previously been shown to be related to  behav­ior change in our prior studies on smoking. (107)The scale includes items which assess prior change at­tempts, and the difficulty en­countered during those attempts.

3.2.2.3. Social Support*

These items measure the degree and type of support a pt is given by others in his/her efforts to change nutri­tional behav­ior. We will use the core measures used in the NCI “5-A-Day” worksite intervention trial to measure social support of family, co-workers and friends for dietary change. (122) Time to complete: 1 minute.

3.2.2.4. Other Variables Measured on Baseline and One Year Questionnaires*

Following are variables on which we will collect data using questions identical to those in WATCH.  These represent variables with which we have considerable experience: Dietary and eating behaviors; Dietary knowledge; Health attitudes; Self-efficacy; Cholesterol measurement history; Weight loss history; Smoking behavior; and Attitudes about physician advice (see Appendix I for baseline assessment).

The following two parameters are measured using separate instruments:

3.2.2.5. Physical Activity*

Physical activity will be assessed to quantify changes in energy expenditure. We will use an instrument developed in the Five-Cities Project to quantify energy expenditure (Blair/Sallis). (123) Subjects review a list of activities and their intensities and recall time spent in sleep and in moderate, hard, and very hard activities in the previous 7 days.  This instrument collects information in all domains of daily activity  and provides an estimate of weekly energy expenditure (kcal/kg/wk). The list of activities will reflect winter activities common to the North-East.  The intensity classifications of the list will conform to the metabolic equivalent (MET) activity intensity classifications in the recently compiled physical activities compendium of Ainsworth et al. (124) Time to complete: 5 minutes.

3.2.2.6. The Beck Depression Inventory*

This instrument is a standard for the assessment of depression. (125) It has well established psychometric properties, including high internal consistency (alpha = 0.86). In studies comparing responses to clinical assessments, the Pearson correlation coefficients are approximately .66. Time to complete: 3 minutes.

4.  Other outcome variables of interest

The following two variables will be measured at baseline and again at the 12-month point:

4.1. Patient Satisfaction

The Overall Evaluation of Quality Scale will be used to assess pt satisfaction with care. This is a 2-item scale consisting of an evaluation of 1) overall quality of care, and 2) outcome of care. Pts rate their responses on a 5-point rating scale: poor, fair, good, very good, excellent. The score is the mean of the 2 items. This scale compares favorably to other frequently used pt satisfaction measures. (126) Time to complete: 1 minute.

4.2. Quality of Life