PHYSICIAN'S EXAMINATION Examinations were conducted on all examinees by a trained physician in the mobile examination center (MEC). The protocol for this component did not detail any medical, safety, or other exclusions. Because this component was administered in the examination center, the MEC examination sample weight (WTPFEX6) should be used for data analysis. For more information on the use of sample weights in NHANES III data analysis, refer to the NHANES III Analytic and Reporting Guidelines (U.S. DHHS, 1996b). The examination was performed by a board-eligible physician. Depending on the examinee's age, the examination included observation of gait, eye, and limb abnormalities; a joint examination of upper and lower extremities; blood pressure measurements; evaluation of heart and breath sounds; Tanner staging and breast-size measurements; and bioelectrical impedance analysis (BIA) resistance and reactance measurements. Refer to Chapter 3 of the Physician Examiners Training Manual (U.S. DHHS, 1996b) and Pulse and Blood Pressure Procedures for Household Interviewers (U.S. DHHS, 1996b) for descriptions of data collection procedures and methods. Although the BIA measures were recorded during the physician's examination, these data are included in the Body Measurements section of this data file. Three sets of blood pressure measurements were taken in the examination center on examinees aged five years and over. For children and adolescents (aged 5-19 years), three Korotkoff (K) sounds were recorded: K1 (systolic); K4, muffling of pulse sounds (diastolic); and K5, disappearance of pulse sounds (diastolic). For adults (aged 20 years and over), only K1 (systolic) and K5 (diastolic) measurements were obtained. All blood pressure determinations were recorded to the nearest even number. Blood pressure measurements were also taken by trained interviewers during the household interview, on sample persons aged seventeen years and over. These data can be found in the NHANES III Household Adult Data File as part of the Household Adult Questionnaire in variables HAZA1-HAZNOK5R. Both physicians and interviewers took measurements using a mercury sphygmomanometer (W. A. Baum Co., Inc, Copiague, NY) according to the standardized blood pressure measurement protocols recommended by the American Heart Association (Frohlich, 1988). In addition to taking part in an initial training course specific to the NHANES III examination procedures, examining physicians participated in a formal, annual re-training program. Their work in the field was monitored for quality by consultants and NCHS staff who visited the mobile examination centers; monitoring occurred once a month in the earlier years and every three months during the latter part of the survey. For additional blood pressure quality control, physicians and household interviewers received a day and a half of initial blood pressure measurement training, were recertified quarterly, and were retrained annually. Training consisted of instruction in the recognition of Korotkoff sounds using videotaped blood pressure examples, comparison of blood pressure readings with those of instructors using double-headed stethoscopes, and practice with volunteer subjects. Physicians and household interviewers who exhibited end-digit preferences, high/low measurement bias, or lack of consistent measurements repeated the training until their performances were deemed satisfactory for certification. Physicians and household interviewers were given hearing tests at the beginning of their employment and yearly thereafter. Note that the coded values for "Yes" (2) and "No" (1) responses in the physician's examination vary from the usual coded values for "Yes" (1) and "No" (2) found in most other places in the NHANES III data files. They match the codes found on the Physician's Examination form (U.S. DHHS, 1996b). Also note that a number of areas (joint examination, dermatitis, chest, and heart examination) in the Physician's Examination allow for multiple responses. For example, on the wrist examination (PEP4A, PEP4A1, PEP4A2, and PEP4A3), the examining physician may have found both tenderness and swelling (PEP4A1=2 and PEP4A2=2). Consequently, the overall positive findings in the detailed questions (PEP4A1, PEP4A2, and PEP4A3) may exceed the number of positive findings reported in the lead question (PEP4A). Moreover, individuals classified as having a positive finding in a lead question may have an 8-fill, "Blank but applicable," in one of the associated detailed questions. For example, 26 cases classified as "Yes" in PEP1 (overall findings for locomotion) were classified as "Blank but applicable" for PEP1A1 (limp or shuffle) instead of "Yes, limp or shuffle" or "No findings." Here, the examining physician used "Blank but applicable" to account for situations in which he/she was unable to observe or examine the specific abnormality. When blood pressure measurement data were edited and it was discovered that examinees aged twenty years and over were missing one of a K1/K5 pair of blood pressure measurements, the remaining one was made "Blank but applicable;" there were ten such cases. However, for examinees under twenty years of age who lacked one or two of a set of K1/K4/K5 blood pressure measurements, a different approach was taken. There were 71 such cases with the overwhelming majority of the missing values being K4 and K5. Because the literature on pediatric blood pressure suggests that systolic blood pressure is the most reliable determination (Uhari, 1991), it was not desirable to delete this value if the associated K4 and/or K5 values were missing. Therefore, K1 values in the data set were retained without the corresponding K4 or K5 values. This accounts for discrepancies in counts between K1 and K5 determinations (between PEP6G1 and PEP6G3, PEP6H1 and PEP6H3, and PEP6I1 and PEP6I3). Finally, when K4 was equal to K5, K4 was coded as "Blank but applicable." The analyst should be aware of the potential for observer bias when analyzing these data. During the six years of the survey, the number of examinations performed by each physician varied widely among the twelve physicians, from as few as 199 to as many as 9,529. Data processing and editing were performed to ensure internal data consistency. Notes have been provided for variables requiring additional explanation.