Non-linear Effects Of Age Research Articles

Chronological age, cognitions, and practices in European American mothers: A multivariate study of parenting.

The authors studied multiple parenting cognitions and practices in European American mothers (N=262) who ranged from 15 to 47 years of age. All were 1st-time parents of 20-month-old children. Some age effects were 0; others were linear or nonlinear. Nonlinear age effects determined by spline regression showed significant associations to a "knot" age (approximately 30 years), with little or no association afterward. For parenting cognitions and practices that are age-sensitive, a 2-phase model of parental development is proposed. The findings stress the importance of considering maternal chronological age as a factor in developmental study.

Genotype-by-sex interaction in the aetiology of type 2 diabetes mellitus: support for sex-specific quantitative trait loci in Hypertension Genetic Epidemiology Network participants

While there are sex-related differences in both the prevalence of type 2 diabetes mellitus and disease risk factors, there is only limited research on sex-specific influences on type 2 diabetes aetiology within the same study population. Thus, we assessed genotype-by-sex interaction using a liability threshold model in an attempt to localise sex-specific type 2 diabetes quantitative trait loci (QTLs). Hypertensive siblings and their offspring and/or parents in the Hypertension Genetic Epidemiology Network of the Family Blood Pressure Program were recruited from five field centres. The diabetic phenotype was adjusted for race, study centre, age and non-linear age effects. In total, 567 diabetic individuals were identified in 385 families. Variance component linkage analyses in the combined sample and stratified by sex and race were performed (SOLAR program) using race-specific marker allele frequencies derived from a random sample of participants at each centre. We observed a QTL-specific genotype-by-sex interaction (p=0.009) on chromosome 17 at 31 cM, with females displaying a robust adjusted logarithm of odds (LOD) of 3.0 compared with 0.2 in males and 1.3 in the combined sample. Three additional regions demonstrating suggestive evidence for linkage were detected: chromosomes 2 and 5 in the female sample and chromosome 22 (adjusted LOD=1.9) in the combined sample. These findings suggest that multiple genes may regulate susceptibility to type 2 diabetes, demonstrating the importance of considering the interaction of genes and environment in the aetiology of common complex traits.

Maternal chronological age, prenatal and perinatal history, social support, and parenting of infants.

The role of maternal chronological age in prenatal and perinatal history, social support, and parenting practices of new mothers (N=335) was examined. Primiparas of 5-month-old infants ranged in age from 13 to 42 years. Age effects were zero, linear, and nonlinear. Nonlinear age effects were significantly associated up to a certain age with little or no association afterward; by spline regression, estimated points at which the slope of the regression line changed were 25 years for prenatal and perinatal history, 31 years for social supports, and 27 years for parenting practices. Given the expanding age range of first-time parents, these findings underscore the importance of incorporating maternal age as a factor in studies of parenting and child development.

Adjusting for Non-Linear Age Effects in the Repeat Sales Index

A true constant quality real estate price index should measure the general change in price level free from any change in quality over time. In recent years, the repeat-sales method has been widely used to construct constant quality property price indices. Since buildings depreciate over time, a simple repeat-sales index would underestimate the growth in property prices. The major problem of controlling the effects of age constant in a repeat-sales model arises from the exact multicollinearity between the age variable and the time dummy variables. In this study, we derive a solution that is theoretically sound and practical by allowing the age effects to be non-linear. In case of leasehold properties, we further incorporated interest rates into the model because the effects of age on real estate prices depend theoretically on interest rates. A sample of residential units in Hong Kong sold more than once from Quarter 2 of 1991 to Quarter 1 of 2001 (more than 11,000 repeat sales pairs) are used for the empirical analysis.

The Effects of Bilateral Tax Treaties on U.S. FDI Activity

The effects of bilateral tax treaties on FDI activity have been unexplored, despite significant ongoing activities by countries to negotiate and ratify these treaties. This paper estimates the impact of bilateral tax treaties using both U.S. inbound and outbound FDI over the period 1966-1992. Our estimates suggest a statistically significant average annual increase of FDI activity for each additional year of a treaty which ranges from 2% to 8%, depending on the FDI activity measure and empirical framework we employ. Examination for nonlinear effects of treaty age on FDI activity suggests that while treaties have an immediate impact on FDI flows, there is a substantial lag before treaty adoption positively affects FDI stocks and affiliate sales. Finally, our results suggest that bilateral tax treaties have an effect on investment outside of the withholding tax rates they alter, which may include the commitment and risk reduction effects of these treaties.

Age–period–cohort effects on inequalities in psychological distress, 1981–2000

In the closing decades of the twentieth century, changes in population sociodemographics took place that might be thought to have an adverse influence on the nation's psychological distress. Here, we examine the stability of social and gender inequalities in psychological distress throughout the 1980s and 1990s. The study uses data from the 1958 National Child Development Study and the 1970 British Cohort Study collected when the cohort members were aged between 23 and 42. Multilevel logistic regression models were used to examine the effects of social class, gender, age, period and cohort on psychological distress as measured by the Malaise Inventory. We identify clear social inequalities in psychological distress during 1981-2000 that reduced in magnitude over this period. Non-linear age effects were observed: psychological distress improves in early adulthood but declines again on approaching mid-life. The 1970 cohort had poorer psychological distress than the 1958 cohort. Although women had higher rates of psychological distress than men, gender differences reduced in magnitude. Declining rates of women's psychological distress over time have not been matched in men. A reduction in social inequalities over time was also observed. Improvements in the psychological health of those in manual occupations were not equalled among those in non-manual occupations. Both social and gender inequalities have narrowed in the last two decades of the twentieth century.

Does the menopause influence the risk of bacteriuria?

The objective of this study was to test the hypothesis that the risk of bacteriuria is increased as a result of estrogen deprivation following the menopause. All midstream urine samples (MSU) sent to the King's College Hospital department of microbiology by general practitioners in 1997 were assessed. Bacteriuria was diagnosed when the bacterial count was >10(5) organisms/ml. Logistic regression analysis was performed to investigate the effects of age and sex on the likelihood of having a positive result. Non-linear effects of age were investigated, with interest focusing in particlar on the time around the menopause. There were 15,392 MSU samples analyzed; 11,811 (77%) were from women and 3581 (23%) from men. In both sexes the proportion of positive results increased with increasing age (P<0.0001). The specimens taken from women were significantly more likely to be positive than those taken from a man of the same age (P<0.0001). In women there was no evidence of any non-linear relationship between age and the log odds of a positive result. A plot of the proportion of positive results versus age did not suggest any departure from a linear relationship at or following the menopause. In conclusion, the increased risk of bacteriuria which occurs as women get older appears to happen gradually as a result of the aging process, rather than as the result of pathophysiological changes in the urogenital tract that take place at or following the menopause.

Age-period-cohort analysis of lung cancer mortality in Japan, 1960-1995.

The mortality data on lung cancer in Japan from 1960 to 1995 was analysed based on an age-period-cohort (APC) model. Though the APC model has an 'identifiable problem' caused by the relationship of age, period and cohort parameters, non-linear components of them revealed their original (separated) effects. They were: (1) non-linear age effects had a peak in 55-59 and 60-64 years old in males and 50-54 in females, (2) non-linear period effects were very small in both genders, (3) non-linear age and period effects were small enough to neglect compared with their linear effects, and (4) there were five parts of trends in Japanese lung cancer mortality in both genders in the non-linear birth cohort effects. The 1961-65 birth cohort effect seemed to increase differently from previous birth years. This trend should be monitored carefully.

Age at first alcohol use: a risk factor for the development of alcohol disorders.

This study aimed to describe the natural course of DSM-III-R alcohol disorders as a function of age at first alcohol use and to investigate the influence of early use as a risk factor for progression to the development of alcohol disorders, exclusive of the effect of confounding influences. Data were obtained from a community sample (N=5,856) of lifetime drinkers participating in the 1990-1991 Mental Health Supplement of the Ontario Health Survey. Survival analyses revealed a rapid progression to alcohol-related harm among those who reported having their first drink at ages 11-14. After 10 years, 13.5% of the subjects who began to drink at ages 11 and 12 met the criteria for a diagnosis of alcohol abuse, and 15.9% had a diagnosis of dependence. Rates for subjects who began to drink at ages 13 and 14 were 13.7% and 9.0%, respectively. In contrast, rates for those who started drinking at ages 19 and older were 2.0% and 1.0%. Unexpectedly, a delay in progression to harm was observed for the youngest drinkers (ages 10 and under). Hazard regression analyses revealed a nonlinear effect of age at first alcohol use, marked by an elevated risk of developing disorders among subjects first using alcohol at ages 11-14. First use of alcohol at ages 11-14 greatly heightens the risk of progression to the development of alcohol disorders and therefore is a reasonable target for intervention strategies that seek to delay first use as a means of averting problems later in life.

Determinants of Progression of HIV Infection in a Greek Hemophilia Cohort Followed for Up to 16 Years After Seroconversion

Our objectives are to describe the progression of HIV disease and to assess the influence of hemophilia-related variables, age at infection, and antibodies to cytomegalovirus infection (anti-CMV) in a Greek cohort of 158 HIV-1-positive hemophilic men, who received prospective follow-up for up to 16 years after infection. A total of 79 patients had died, representing a cumulative progression rate of 72.4% (95% confidence interval [CI], 56.6-83.3). A significant proportion of the mortality (30%) resulted from conditions not formally related to AIDS, with liver failure and cerebral hemorrhage predominant. At 16 years after seroconversion, 66 patients had developed clinical AIDS, a cumulative progression rate of 58.2% (95% CI, 47.1%-86.3%) whereas 15 years after infection 81.5% (95% CI, 74.2%-87.9%) of the patients had AIDS or a CD4 cell count <200 cells/mm3. Hemophilia-related variables or presence of anti-CMV were not significantly associated with disease progression. Age at infection was a strong prognostic factor for all three endpoints. Appropriate modeling showed a nonlinear age effect, with a steeper increase of relative hazard for patients >40 years of age at seroconversion. The age effect remained significant even after controlling for current CD4 cell count. Further investigation is required to elucidate the mechanisms of the age effect and the contribution of HCV coinfection on the disease progression.

Meta-analyses of age-cognition relations in adulthood: estimates of linear and nonlinear age effects and structural models.

A meta-analysis was conducted on 91 studies to derive a correlation matrix for adult age, speed of processing, primary-working memory, episodic memory, reasoning, and spatial ability. Structural equation modeling with a single latent common cognitive factor showed that all cognitive measures shared substantial portions of age-related variance. A mediational model revealed that speed of processing and primary-working memory appear to be important mediators of age-related differences in the other measures. However, not all of the age-related influences were mediated. An examination of quadratic age effects and correlational patterns for subsamples under and over 50 years of age revealed that (a) negative age-cognition relations were significant for the 18- to 50-year-old sample and (b) the age-related decline accelerated significantly over the adult life span for variables assessing speed, reasoning, and episodic memory.

Non-linear effects in dose-time analyses: application of modern statistical techniques

Background and purpose: The non-linear effect of overall treatment time and total dose on the outcome of fractionated head and neck radiotherapy is of interest. This and other non-linear effects can be investigated using recently developed statistical techniques. This article provides an illustration of the ability of these statistical methods. Materials and methods: Recently developed statistical methods (Hastie, T.J. and Tibshirani, R.J. Generalised Additive Models. Chapman-Hall, London, 1990), called generalized additive models, are applied to data from the patterns of fractionation study (Withers, H.R., Peters, L.J., Taylor, J.M.G., Owen, J.B., Morrison, W.H., Schultheiss, T.E., Keane, T., O'Sullivan, B., van Dyk, J., Gupta, N., Wang, C.C., Jones, C.U., Doppke, K.P., Myint, S., Thompson, M., Parsons, J.T., Mendenhall, W.M., Dische, S., Aird, E.G.A., Henk, J.M., Bidmead, M.A.M., Svoboda, V., Chon, Y., Hanlon, A.L., Peters, T.L. and Hanks, G.E. Local control of carcinoma of the tonsil by radiation therapy: an analysis of patterns of fractionation in nine institutions. Int. J. Radiat. Oncol. Biol. Phys. 33: 549–562, 1995) of tonsil cancer. These techniques enable one to develop models which more accurately represent the relationship between multivariate predictor variables and the outcome variable in a regression analysis. These data driven methods allow the effect of each predictor variable on the outcome to be nonlinear and estimated from the data. This is achieved by replacing the standard linear model combination of predictor variables, such as ‘α 0 + α 1dose + α 2time’ by ‘ S 1(dose) + S 2(time)’, where S 1 and S 2 are smooth non-linear functions of dose and time, respectively, which are estimated from the data. Results: In the pattern of fractionation study these methods indicate that the effect of total dose on the probability of local recurrence is linear, but there is a suggestion that the effect of overall treatment time is non-linear. There is no effect of dose and time on the latency time to recurrence of those who do recur, but there is a weak suggestion of a non-linear effect of patients age, with younger patients recurring earlier. Conclusions: Generalized additive models provide a flexible and powerful means of exploring non-linear effects in experimental data.

Meta-analyses of age-cognition relations in adulthood: Estimates of linear and nonlinear age effects and structural models.

Meta-analyses of age-cognition relations in adulthood: Estimates of linear and nonlinear age effects and structural models.

Age, period, and cohort size effects on self-reported alcohol, marijuana, and polydrug use: Results from the National Youth Survey

Age, period, and cohort size effects are calculated for alcohol, marijuana, and polydrug use for seven cohorts of youth, aged 11–17 in 1976 and 18–24 in 1983. Data are taken from the National Youth Survey, a longitudinal prospective probability sample. Use of cohort size, rather than year of birth, is based on existing theory (the Easterlin hypothesis) and allows parameter estimation by ordinary least-squares regression. The results confirm the existence of a nonlinear age effect with a peak at about age 20, a finding consistent with the hypothesis of maturational reform. The Easterlin cohort size hypothesis also receives support, with larger birth cohorts having higher rates of substance use for most of the dependent variables. Period effects are also present and appear to be nonlinear with a peak at 1980 for marijuana use; they are monotonically positive for alcohol and polydrug use. In several respects, these results for alcohol and illicit substance use differ from those for other types of illegal behavior with respect to age, period, and cohort size effects.

Genetic and environmental sources of variation in physical fitness.

The technique of path analysis was used to assess inherited and environmental variance components in physical fitness indicators measured in 1630 subjects from 375 families of French descent living in the greater Québec city area. For that purpose, submaximal power output (PWC150/kg), muscular endurance, muscular strength, reaction time and movement time were evaluated during a visit of the family to the laboratory. Inter-class correlations in various types of relatives were computed from scores adjusted for linear and non-linear effects of age and sex by a regression procedure (Y = age + sex + (age X sex) + age2). Correlations were then used in the path analytic BETA model which allows the partition of transmissible variance (t2) into genetic (h2) and cultural (b2) components. Results indicated that t2 accounted for 18% (movement time) to 63% (muscular strength) of the phenotypic variance. The contribution of genetic factors was found to be negligible for PWC150/kg and movement time, and accounted for about 20% of the phenotypic variance for reaction time and muscular endurance and 30% for muscular strength, while non-transmissible variance (1 - t2) accounted for 37% (muscular strength) to 82% (movement time) of the phenotypic variance. These results suggest that biological variation observed in the physical fitness level of a healthy population is mainly associated with non-transmissible environment factors and that the contribution of heredity is moderate and clearly lower than previously reported.

Non-linear effects of age on nerve conduction in adults

The effects of age on conduction and amplitude in median and ulnar (motor and sensory), superficial radial, sural and common peroneal (motor) nerves of adults were prospectively investigated. Four routinely recorded parameters, including conduction velocity, amplitude and duration, were considered for each sensory nerve and measurements of conduction velocity, terminal motor latency and amplitude were made for each motor nerve. The resulting 25 sets of data were analysed using both linear and quadratic regression. The three sets of terminal motor latency data showed no age dependence while in six other instances there was a linear relationship with age (three of these being motor amplitude). The remaining 16 sets of data were statistically best represented by quadratic analysis. The shapes of the paraboli were remarkably similar, reaching a maximum (conduction velocity and amplitude) or minimum (duration) value in the fourth decade and thereafter declining or rising respectively at an accelerating rate. With only one exception, quadratic curves of the same parameter were shown to be parallel. Tables of normal data which make accurate adjustments for the aging effect can be constructed from this analysis for use in routine clinical investigation.

Selenium, cadmium, zinc, and copper in human kidney cortices and post mortem indices of hypertension

Kidneys from 32 autopsied Caucasian human subjects aged 16-60 were frozen then lyophilized while the flasks were kept in insulating containers. (Subjects with evidence of extensive weight loss, chronic renal failure, or carcinoma were not included). Replicate samples of cortex were removed, weighed, and wet-ashed in HNO3-HClO4. Zn, Cd, and Cu were estimated by atomic absorption spectrophotometry with the 2261 Å line used for Cd background correction. Se was estimated by fluorescence with DAN in a Turner fluorimeter, but a #74 Kodak Wratten filter was added to minimize the 500 nm interference noted by E. Pickett (personal communication, 1980).Means and variability for the three elements in kidney cortex were similar to those in a North Carolina study as was Cd/(Cd+Zn)×100. Cortical Cd (P<0.05), Zn (P<0.10), and Cd/(Se·Zn) (P<0.05) increased with age, whereas Se showed no significant relationship with age. Age(2), nonlinear effects of age, had a slight influence upon Cu (P<0.05) only if data were not adjusted for gender. There was no influence of ageper se upon Cu with or without adjustment for gender. The scatter diagrams of element concentrations plotted vs age contained several provocative "outlier" values.A positive association of kidney cortex Cd concentration, or Cd/(Se ·Zn), with postmortem indices of hypertension existed only if age, gender, and age(2) were omitted from the multiple regression equation. This adjustment was not included in a similar study of North Carolina cases and appears to be the source of the major difference in their respective inferences drawn about the positive relationship of kidney cortex Cd with evidence of hypertension. This difference in statistical models does not however account for the failure of the West Virginia sample to indicate a protective role of kidney cortex Se suggested for the North Carolina subjects. Larger samples, drawn from regions differing in Se abundance, will be necessary to test the latter question adequately.

Infants' scanning of geometric forms varying in size

Infants 1, 2, and 3 months of age were shown a series of plane geometric shapes of three different sizes while their eye movements were unobtrusively monitored with an infrared corneal reflection eye-tracking system. The size of the form significantly influenced the pattern of visual inspection. Contrary to earlier reports, there were no age differences in the spatial distribution of fixations to the geometric stimuli. Nonlinear age effects were found in some parameters of eye movement control.