School performance and risk of suicide in early adulthood: follow-up of two national cohorts of Swedish schoolchildren.

BACKGROUND: Poor school performance and low IQ are associated with an increased risk of suicide in males; it is uncertain whether cognitive performance is a risk factor for suicide in females and whether severe mental illness influences these associations. METHOD: Record linkage study of Swedish education, population and census data with mortality and inpatient registers. Data were available for 95,497 males and 91,311 females born in 1972 and 1977 and followed up until 31 December 2005. RESULTS: 230 males and 90 females died from suicide over the follow-up period. There were strong inverse associations between school performance at age 16 and subsequent suicide risk in males (fully adjusted hazard ratio (HR) per SD increase in school performance score: 0.6 (95% CI 0.6 to 0.7; p < 0.001)) but not females: adjusted HR. 1.1 (CI 0.9 to 1.4 p = 0.50). In males there were almost four fold differences in suicide risk between children in the top and bottom fifths of the range of school grade scores. Similar associations were seen with upper secondary school performance (age 18/19 years). There was no strong evidence that deterioration in school performance between ages 16 and 18 was associated with increased suicide risk. Amongst men who developed severe psychiatric illness school performance was not associated with suicide risk. LIMITATIONS: We had limited information on the prevalence of minor psychiatric disorder in cohort members and no data on IQ for the cohort as a whole. CONCLUSIONS: Good performance in secondary (age 16) and upper secondary (age 18) school is associated with a reduced risk of suicide in men but not women. This protective effect is not seen amongst those who develop severe psychiatric illness. These differences indicate that the aetiology of suicide differs in males and females and in those with and without severe mental illness.

Association of IQ scores and school achievement with suicide in a 40-year follow-up of a Swedish cohort.

OBJECTIVE: Few studies have investigated the association of childhood IQ and school achievement with suicide. The aim of this study was to investigate the association of childhood IQ with suicide in a cohort of Swedish women and men. METHOD: 21 809 subjects born in 1948 and 1953 who completed IQ and school tests at age 13 years have been followed until 2003. Information on paternal education and in-patient care for psychosis was linked using the Swedish personal identification number. RESULTS: There were 180 suicides amongst subjects with measured IQ. High IQ was associated with reduced suicide risk among men (OR per unit increase in age-adjusted model 0.90, 95% CI 0.83-0.99), while there was no statistical evidence of an association in women (OR 1.04, 95% CI 0.90-1.20). Among men with a history of psychosis, high IQ was associated with an increased risk of suicide. CONCLUSION: Low childhood IQ at age 13 years is associated with an increased risk of suicide in men but not in women; however, amongst those with psychosis, low IQ appears to be protective.

Factor structure of the Dental Beliefs Survey in a dental phobic population.

The aim of this study was to investigate the factor structure of the 15-item Dental Beliefs Survey (DBS) in a population of dental phobic patients (n=362). Exploratory (EFA) and confirmatory (CFA) factor analyses were used. The EFA indicated three factors: 'communication', 'trust', and 'fear of negative information'. However, the EFA further suggested a general factor with all 15 items. By using the CFA, five factors were found based on the EFA solution and the original DBS dimensions. A general factor, 'social interaction distress in dental treatment', and four more narrow dimensions, 'communication', 'trust', 'fear of negative information' (originally labelled 'belittlement'), and 'lack of control', was the most adequate result with regard to theoretical and statistical properties. However, some items in the factors were partly different from the original version of the DBS. In conclusion, the DBS attempts to measure a complex phenomenon with regard to patients' perceptions and attitudes to dental care. This study has revealed psychometric properties of the DBS in a population of dental phobic patients. The important finding was a general dimension, which suggests the use of DBS as an overall measure of dental beliefs. However, more research is needed in epidemiological and clinical studies with non-phobic individuals.

Psychosomatic complaints in schoolchildren: a Nordic comparison.

AIMS: This study investigates the association between psychosomatic complaints (PSC) and background factors from children in the five Nordic countries, their families, and their social context in a model using proximal (child), intermediate (family), and distal (society) levels. Similarities and differences between the countries were compared. METHODS: A questionnaire was sent to the parents of a representative sample of 3,760 Nordic children aged seven to 12 years. The relationship between PSC and background factors was analysed using structural equation modelling. RESULTS: Headaches and stomach complaints were most common. The prevalence was higher in girls than in boys. A LISREL analysis showed that determinants of PSC varied little between countries. The most important factors associated with PSC were the mother's health and well-being, the child's mental stability and social competence, and the family's socioeconomic status. Families with low education, blue-collar work, low income, and unemployed fathers were found to be most vulnerable. High risk factors could be influenced by other intervening protective factors. The greatest country differences were found on the distal level, where macroeconomic, legislative, and social factors dominated. CONCLUSIONS: The pattern of PSC was quite similar between the Nordic countries, although there were some small differences. The results also confirm that the determinants of PSC are multidimensional. Longitudinal studies are needed to clarify the complex reality.

Determinants of psychosomatic complaints in Swedish schoolchildren aged seven to twelve years.

OBJECTIVE: The aim of this study was to investigate determinants of psychosomatic complaints (PSC) in children. The model assumes that variables are working on three levels: the proximal level (the child), the intermediate level (the family), and the distal level (the society). METHODS: Data were obtained from a large cross-sectional survey on children's health and well-being in the Nordic countries. A questionnaire was sent to the parents of a representative sample of 1,163 Swedish children aged seven to 12 years. The analysis included a multivariate analysis using the LISREL program. RESULTS: The best predictors of PSC were the mother's health, the child's mental stability, contacts with peers, and long-term illness/disability. There were many other important relationships, which act via factors, such as school satisfaction, social competence, activities, the parents' sense of coherence, and family economy. CONCLUSION: PSC in children should be seen in the context of all three levels.

Repeated measurements of mood during psychologic treatment of dental fear.

The aims of the present study were to analyze mood changes during psychologic treatment of dental fear by assessing the rate of improvement. Twenty-one patients who refused conventional dental treatment and reported extreme dental anxiety participated in the study. Levels of dental anxiety and mood were measured with the Dental Anxiety Scale (DAS) and a Mood Adjective Checklist (MACL). MACL included two dimensions, degree of relaxation (r) and pleasantness (h) as experienced in a dental situation. Mood was monitored at each treatment session from base line to termination of the therapy (eight measurements). Two different treatment modalities were used, one with a more cognitive approach (n = 9) and one emphasizing the relaxation component (n = 12). A hierarchical linear models approach was applied to analyze individual change with repeated measurements. The results showed that positive mood changes over time were statistically significant. The mean improvement in mood scores per week and session was estimated for MACL(r) and MACL(h) to be 0.14/week and 0.09/week, respectively. The growth was not affected by DAS levels or treatment mode. This study also illustrated a powerful method for analyzing a longitudinal clinical trial design with repeated measurements.

Factor analysis and reduction of a Fear Survey Schedule among dental phobic patients.

A fear survey instrument, based on the Fear Survey Schedule-II and five additional fear items, was administered to 109 patients (70 women and 39 men) on a waiting list at a specialized dental fear clinic. The fear survey was analyzed to identify its factorial structure. Five fear factors, explaining 54% of the total variance, were identified concerning areas of "illness and death", "failures and embarrassment", "social situations", "physical injuries", and "animals and natural phenomena". An ad hoc reduction of items was carried out to form a shorter, more practical to use questionnaire, which resulted in factors of four or five items with loadings greater than 0.50. The factors intercorrelated significantly (rp varying between 0.33 and 0.59) and "illness and death" correlated highly with "physical injuries" (rp = 0.59) and "animals and natural phenomena" (rp = 0.56), while "failures and embarrassment" correlated highly to "social situations" (rp = 0.54). Statistically significant, but generally lower correlations were found between each factor and the dental fear measures. The highest correlations were found between fear of "physical injuries" and dental fear. There was also a high and significant correlation between sex and fear of "animals and natural phenomena".

Multivariate analysis of fears in dental phobic patients according to a reduced FSS-II scale.

This study analyzed and assessed dimensions of a questionnaire developed to measure general fears and phobias. A previous factor analysis among 109 dental phobics had revealed a five-factor structure with 22 items and an explained total variance of 54%. The present study analyzed the same material using a multivariate statistical procedure (LISREL) to reveal structural latent variables. The LISREL analysis, based on the correlation matrix, yielded a chi-square of 216.6 with 195 degrees of freedom (P = 0.138) and showed a model with seven latent variables. One was a general fear factor correlated to all 22 items. The other six factors concerned "Illness & Death" (5 items), "Failures & Embarrassment" (5 items), "Social situations" (5 items), "Physical injuries" (4 items), "Animals & Natural phenomena" (4 items). One item (opposite sex) was included in both "Failures & Embarrassment" and "Social situations". The last factor, "Social interaction", combined all the items in "Failures & Embarrassment" and "Social situations" (9 items). In conclusion, this multivariate statistical analysis (LISREL) revealed and confirmed a factor structure similar to our previous study, but added two important dimensions not shown with a traditional factor analysis. This reduced FSS-II version measures general fears and phobias and may be used on a routine clinical basis as well as in dental phobia research.

General and Specific Abilities as Predictors of School Achievement.

The purpose of this study is to investigate the relations between aptitude variables and school achievement using a model of ability which allows simultaneous identification of general and specific abilities. A battery of 16 aptitude tests was administered in the 6th grade and course grades were collected in 17 different subject matter areas in the 9th grade (N = 866). For the aptitude tests a confirmatory factor model is fitted with a general factor (G) along with nine orthogonal, residual factors. Some of the residual factors are quite broad (Crystallized intelligence, Gc(1), and General visualization G(1)), but most are narrow factors identified by pairs of tests (e.g., V(1), Ms(1), Num Ach(1), Vz(1), S, and Cs(1)). A model is fitted to the 17 course grades as well. The model includes a general school achievement factor (GENACH) and domain specific achievement factors in areas such as science-mathematics (SCIENCE), social science (SOCSCI(1)), language (LANG(1)) and spatial-practical performance (SPATPR(1)). Relating the latent criterion variables to the latent aptitude variables it is found that some 40% of the variance in GENACH may be accounted for by G and Gc(1). However, larger proportions of variance are accounted for in the domain specific achievement factors and different aptitude factors are important in different domains. It is conclu~ded that differentiation among at least a limited number of broad abilities may be worthwhile.

The Hierarchical Organization of Cognitive Abilities: Restoring General Intelligence Through the Use of Linear Structural Relations (LISREL).

While considerable support has been obtained for the distinction, due to Cattell and Horn, between the broad factors fluid intelligence (Gf) and crystallized intelligence (Gc), there is also some evidence that Gf is equivalent to the factor of general intelligence (g). This hypothesis is investigated by using LISREL to specify higher-order models in reanalyses of three sets of psychometric data from subjects 11, 13, and 15 years old, respectively. The three studies unanimously showed Gf to be equivalent with a general factor. The discussion is centered upon the use of LISREL and the differences in results between the exploratory and confirmatory approaches to factor analysis.

Homogeneous enzyme immunoassay for thyroxine compared with a radioimmunoassay technique.

A homogeneous enzyme immunoassay (EMIT) for serum thyroxine determination has been evaluated. It was also compared to our routine method, a radioimmunoassay with use of a commercial kit, Tetra-Tab RIA. Within-day precision was very similar for the two methods, the CV ranging from about 2.5% at 150 nmol of thyroxine per liter to about 5.7% at 25 nmol/L. The two methods were compared by running 100 patients' serum samples. The resulting linear regression equation was: y = 1.088x - 17.9, with a correlation coefficient r = 0.979. The deviation from the theoretical line, y = x, wash shown to be primarily caused by the calibration of the methods, shown by running the EMIT calibrators in the radioimmunoassay method and vice versa. The choice of thyroxine method in the individual laboratory will then depend on considerations other than methodological factors.

The influence of citrate and phosphate on the Mancini single radial immunodiffusion technique and suggested improvements for the determination of urinary albumin.

When adapting the immediate bromcresol green method for urinary albumin determination a correlation study with the Mancini single radial immunodiffusion (SRID) method was performed. This study showed large disparities between the two methods, SRID giving the higher results. The unsuitability of the currently used SRID methods is demonstrated and improvements to the method are suggested. Known amounts of albumin were added to urine samples as well as to a "synthetic urine", both giving falsely elevated results with the SRID method. On investigating the different components of the "synthetic urine", it was found that the disparities were due to the influence of citrate and phosphate. On addition of citric acid or phosphate to the dilution buffer and/or the gel buffer, the results of the SRID method agreed with those of other methods and with the expected values. The findings presented in this paper can probably be extended to other immunological methods too since it seems to be the antigen-antibody reaction which is affected.

Urinary albumin determination by the immediate bromcresol green method.

It has been demonstrated recently that the reaction of serum samples with bromcresol green (BCG) reagent proceeds in two steps. Albumin is responsible for the immediate reaction while other serum proteins produce the slow reaction. In this paper the immediate BCG reaction has been used for the determination of urinary albumin concentration in patients with proteinuria by a slightly modified method with a primary pH adjustment of the urine and the use of a urine blank. Comparison of the immediate BCG method (y) with Laurell "rocket" technique (x) gave the following equation: y = 17.2 + 1.006x (n = 98; r = 0.99) mg/l. The coefficient of variation (within-day), C.V. (%), ranged between 0.9 and 2.7% depending on the albumin concentration. It is thus possible to carry out rapid, accurate and precise albumin determinations in urine samples using this simple method.

Automated serum albumin determination by use of the immediate reaction with bromcresol green reagent.

It has been demonstrated recently that the reaction of serum samples with bromcresol green reagent proceeds in two steps, albumin being responsible for the immediate reaction and other serum proteins for the slower one. I show here that the immediate reaction, used for albumin determination, can be automated. Two different automated systems are described: the LKB 2086 Reaction Rate Analyzer and the multichannel analysis system LKB 2480 Prisma. With the first system, 60 (or 120) samples per hour were run with a within-day standard deviation of 0.3 g/liter; with the second system, a feed rate of 150 samples per hour was used (SD, 0.3 g/liter, within-day). Comparison with other methods gave the following equations: y (LKB 2086) = 1.44 + 0.97 X (immunodifussion) (n = 43; r = 0.98) and y(Prisma) = 2.04 + 0.96x (manual method) (n = 73; r = 0.98). Thus albumin determinations can be done reliably, quickly, and simply at low reagent and labor cost.

Improved specificity of serum albumin determination and estimation of "acute phase reactants" by use of the bromcresol green reaction.

The reaction of serum samples with bromcresol green proceeds in two steps. Albumin is responsible for the faster (less than 1 min) reaction; the slower (30-min) reaction is a measure of "acute phase reactant(s)" in serum. Serum is simply mixed with bromcresol green reagent and the absorbance is measured twice, immediately and at 60 min. Albumin concentrations, determined from the absorbance at 0 min, correlate well with those determined by Laurell "rocket" immunoelectrophoresis; r = 0.95 with no certain deviation from unity for the slope and with a negligible difference at zero concentration. The slow reaction was expressed as deltaA% = 100 (deltaAs/deltaAv) where deltaAs and delta Av are the changes in absorbance between 60 and 0 min for the sample and a commercial control serum, respectively. The value for deltaA% correlates well with the percentage of alpha2-fraction as determined by electrophoresis on cellulose acetate, as well as with orosomucoid and ceruloplasmin, all of which are acute phase reactant(s). Whether these proteins or other acute phase reactant(s) actually cause the slow reaction has not yet been established.