Ralph R. Frerichs

Depression and Physical Illness: A Multiwave, Nonrecursive Causal Model

Causal influence has frequently been assumed, but rarely demonstrated, in studies documenting a positive association between physical illness and psychological distress. The present study assesses a latent variable causal model of subjectively experienced physical illness and depression in order to disentangle reciprocal patterns of influence. The data consist of interviews from a community sample of adults (N = 744) interviewed four times over one year. For the total sample, the exogenous variables of socioeconomic status, age, and sex are shown to affect both constructs of physical illness and depression. Illness has a large, contemporaneous effect of increasing depressive symptomatology over previous levels, and depression is found to have a smaller, 4-month lagged effect of increasing levels of physical illness. Multiple group analyses reveal some differences by age and sex, but demonstrate an overall consistency in the reciprocal relationship between depression and physical illness.

The right not to know HIV-test results

Large numbers of pregnant women in Africa have been invited to participate in studies on HIV infection. Study protocols adhere to guidelines on voluntary participation after pre-test and post-test counselling and informed consent; nevertheless, women may consent because they have been asked to do so without fully understanding the implications of being tested for HIV. Our studies in Nairobi, Kenya, show that most women tested after giving informed consent did not actively request their results, less than one third informed their partner, and violence against women because of a positive HIV-antibody test was common. It is important to have carefully designed protocols weighing the benefits against the potential harms for women participating in a study. Even after having consented to HIV testing, women should have the right not to be told their result.

Measuring Depression in the Community: A Comparison of Telephone and Personal Interviews

Interview method effects in response to the Center for Epidemiologic Studies-Depression Scale (CES-D) are compared among a community sample (N = 546) of adults randomly assigned to either telephone or in-person interviews. The interviews were conducted in Los Angeles during 1980. No significant differences were observed between the two interview methods in nonresponse to symptom items, preference for specific response categories, reliability, mean level of depression, or proportion classified as depressed. Furthermore, no significant interactions were found between sociodemographic characteristics and the method of interview.

Stress, support, and depression: A longitudinal causal model

This study assesses causal relationships among stress, social support, and depression using data collected at four points in time over one year from a community sample of 740 Los Angeles County adults. A series of latent variable causal models are tested to assess effects over time spans of 4, 8, and 12 months. Depression and support are found to be moderately and highly stable, respectively, over one year, while stress is only slightly stable. Recent stress is found to increase levels of depression from previous levels. Social support is shown to have direct negative effects on current depression and indirect effects on subsequent depression. Reciprocal causal relationships are also explored.

Analysis of effects of sex and age in response to items on the CES-D scale

Analyses of differences in response to individual items in the Center for Epidemiologic Studies Depression (CES-D) scale based on sex and age breakdown are reported for a sample of 1,000 adults from Los Angeles County. Overall, results are comparable to those obtained in other population studies. Analyses performed separately for age and sex groups show response differences on individual items which affect the summated score. The interitem correlations are shown to be much higher for women that men, but overall the index assesses a comparable underlying dimension of depression for the two sexes.

Race, ethnicity, and depression: a confirmatory analysis.

This study addresses two questions: (a) Are depressive-symptom scales equally indicative of depression in different racial/ethnic groups and (b) are there mean differences between the groups in the underlying depressive disorder assessed by these scales? The data consist of interviews obtained from a large community survey of depression in Los Angeles County. Four racial/ethnic groups were considered: Anglos, blacks, English-speaking Hispanics, and Spanish-speaking Hispanics. Confirmatory factor analysis was used to test both one-factor and two-factor models of depression. An invariant factor pattern was demonstrated. No significant mean differences on a factor of Depression were found, but the groups were found to differ on a Well-Being factor. Implications for survey research on psychiatric impairment among different racial/ethnic groups are discussed.

Telephone versus in-person surveys of community health status.

Reports of physical morbidity are compared among a community sample of Los Angeles County adults (N = 546) randomly assigned to either in-person or telephone interviews. No statistically significant differences were found between the two interview methods for overall assessment of health status, illnesses reported for the previous four months, or reports of hospitalization. A significantly greater proportion of in-person respondents, however, reported the presence of restricted-activity days during the previous two weeks. This difference could not be directly attributed to sociodemographic characteristics or reported physical health status of the two samples.

Comparison of saliva and serum for HIV surveillance in developing countries

Saliva has been proposed as a non-invasive alternative to serum for HIV antibody testing. In a field study in Myanmar (formerly Burma), we evaluated such an alternative to identify the frequency of HIV infection in a surveillance programme of high-risk and low-risk sentinel groups. Duplicate vials of saliva and serum were collected from 479 high-risk and 1039 low-risk subjects. One vial of each pair was analysed blind in two laboratories, one in the USA and the other in Myanmar. The US laboratory followed WHO confirmatory strategy III with three different enzyme-linked immunosorbent assays (ELISAs), while the laboratory in Myanmar followed strategy I with one ELISA. Serum testing in the US was the gold standard. The Cambridge ELISA with saliva was a more effective surveillance tool (sensitivity 90.5%, specificity 99.5-100%) for describing the frequency of subjects with HIV antibodies than the serum ELISA supplied to Myanmar by WHO (95.9% and 98.3%, respectively). Saliva is recommended as a safe and effective alternative to serum for HIV antibody testing with ELISA in surveillance programmes in developing countries.

Nepalese origin of cholera epidemic in Haiti

Cholera appeared in Haiti in October 2010 for the first time in recorded history. The causative agent was quickly identified by the Haitian National Public Health Laboratory and the United States Centers for Disease Control and Prevention as Vibrio cholerae serogroup O1, serotype Ogawa, biotype El Tor. Since then, >500 000 government-acknowledged cholera cases and >7000 deaths have occurred, the largest cholera epidemic in the world, with the real death toll probably much higher. Questions of origin have been widely debated with some attributing the onset of the epidemic to climatic factors and others to human transmission. None of the evidence on origin supports climatic factors. Instead, recent epidemiological and molecular-genetic evidence point to the United Nations peacekeeping troops from Nepal as the source of cholera to Haiti, following their troop rotation in early October 2010. Such findings have important policy implications for shaping future international relief efforts.

Serum Lipids and Lipoproteins at Birth in a Biracial Population: The Bogalusa Heart Study

Summary: The lipid profile, total cholesterol, triglycerides, and lipoproteins (β;-, pre-β;-, and α-) of cord blood, is presented for 419 black and white infants (94% of the eligible population) born during an 18-month period in Bogalusa, Louisiana. At birth, white neonates of both sexes had higher average levels than black neonates of total cholesterol and β;-lipoproteins. White girls among the four race-sex groups had the highest cord blood levels of total cholesterol, β;-ipoproteins, and α-lipoproteins. Neither stress at delivery, birth weight, socioeconomic status, nor season of the year had an observable effect on any of the lipid or lipoprotein levels. The magnitude and direction of the relationships between the respective lipids and lipoproteins in cord blood were similar to those we have observed in preschool and school-aged children in the same community. Total cholesterol was highly correlated with both β;-and α-lipoproteins; triglycerides were correlated with pre-β;-lipo-proteins but inversely with α-lipoproteins. These observations suggest that basic biochemical relationships are already established at birth.Speculation: The lipid and lipoprotein levels in cord blood provide useful baseline data for evaluating the evolution of lipoprotein profiles in children. By elucidating those factors which serve as determinants of lipids and lipoproteins in cord blood, we may come closer to understanding the lipid regulatory mechanisms in the general population.