Search

Abstract:  Mortality rates for a decade in Brisbane are analysed for dependence upon atmospheric factors. Time filters are applied to both the dependent and independent variables, and several models are developed to enable prediction, especially for weekly intervals. Statistically, deaths are observed to increase with colder and less humid weather with winds from a westerly, direction. Overall, taking account of both synoptic and seasonal influences, >90% of cumulative deviations from mean death rates are explained. Some differences are also noted in the association of death with the weather between sexes, age groups and causes of death.

Associations between various anthropometric indicators and breast cancer were investigated in a community-based case-control study conducted among 456 premenopausal women (135 cases) and 656 women who were at least 10 years past menopause (185 cases) in Brisbane, Australia, from 1981 to 1985. Among postmenopausal women, increased risk was associated with greater weight at age 25 years (p trend = 0.002) and with a heavier maximum weight (p trend = 0.003), but not with recent weight. Results were similar for relative weight (weight (kg)/height (m)1.5). Overall, weight gain was unrelated to the occurrence of breast cancer, although risks were elevated among postmenopausal women who weighed less at age 25 (< 54 kg) and subsequently experienced a large weight gain (> 13 kg). Among premenopausal women, there was little indication that body size or change in weight was related to breast cancer. These results suggest that both heaviness in early adulthood and a large maximum weight at any time are associated with elevated risks for postmenopausal breast cancer.

Much of the evidence that supports a relation between a positive family history of and increased risk for colorectal cancer is based on information obtained exclusively from patients. There have been few assessments of the accuracy of such data. The validity of self-reported family history of colorectal cancer was assessed in the course of a case-control study of colorectal adenomas conducted among patients aged 20-75 years who underwent colonoscopy in Brisbane, Australia between 1980 and 1985. Family histories reported by a subsample of 237 colonoscopy patients (74 cases and 163 controls) were compared with relatives' medical records and death certificates. Patients' reports of colorectal cancer in 90 relatives were confirmed for 70 (77.8%; 95% confidence interval (CI) 67.8-85.9). Among 124 reports by patients of relatives who had other abdominal cancer or bowel conditions, 114 (91.9%; 95% CI 85.7-96.1) were confirmed to be correct, while 10 (8.1%) were found to be colorectal cancer. Finally, 105 (99.1%; 95% CI 94.9-100.0) of a random sample of 106 completely negative reports by patients were confirmed to be correct. Overall, 77% of positive family histories (any positive relatives) were confirmed, and it was estimated that 98% of negative family histories (no positive relatives) were correct. Cases were slightly more accurate than controls in reporting both positive and negative histories among their relatives. By extrapolation of these results to the total sample of 1,244 patients in the larger case-control study, sensitivity of self-reported positive family history was estimated to be 0.87 among cases and 0.82 among controls, and specificity was estimated to be 0.97 in both groups.

This study validates the concept that a standard (static) weightbearing lateral x-ray, taken with an orthoposer device, in the patient's angle and base of gait is clinically similar to findings seen during ambulatory (dynamic) gait. Sixty female subjects requiring diagnostic x-rays underwent videofluoroscopy of their walking gait. Using a single-frame, shuttle-advance video recorder, the appropriate fluoroscopic video frame was identified. The calcaneal pitch angle was measured using a digitized program. A high repeatability of the measuring process and high correlation between x-ray and fluoroscopic results suggest the value of taking weightbearing angle and base of gait lateral x-rays.

Calcaneal pitch has been considered to be an indirect measure of subtalar joint function. The aim of this pilot study was to assess changes in the calcaneal pitch angle during dynamic gait. Sixty female subjects underwent videofluoroscopy to obtain 27 usable gait cycle data. A single-frame, shuttle-advance video recorder was used to identify midstance of the gait cycle. The calcaneal pitch angle was measured during three midstance periods. The study confirms findings from video and forceplate analysis and reintroduces videofluoroscopy as a gait research tool.

We investigated the relationship between cutaneous malignant melanoma and multiple sunburns in the Queensland population. Interview data were gathered from 236 case-control pairs concerning their lifetime experience of severe sunburns, their occupational and recreational sun exposure, and their skin type. Excluding the lentigo maligna melanoma subtype, an association between multiple sunburns and melanoma was evident. After controlling for other major risk factors there was a significant dose-response relationship (P less than 0.05): the estimated relative risk associated with 2-5 sunburns in life was 1.5, and with 6 or more was 2.4. This observation extends the hitherto circumstantial evidence of a causal relationship between exposure to solar ultraviolet radiation and melanoma, and suggests that precautionary measures could prevent the development of this disease in a proportion of cases in fair-skinned populations.

The true survival rates for the various forms of childhood cancer are best determined from a population-based study rather than from the results of clinical trials. Population-based survival rates have been calculated for four periods between 1956 and 1980 in Queensland. There was a significant improvement in survival for children who developed cancer after 1973 compared with those diagnosed before this date. There has however been no significant improvement in the survival rate for childhood cancer overall, or for acute lymphoblastic leukaemia since 1973. Over the 25 year period significant trends in survival rates were seen in acute lymphoblastic leukaemia, non-Hodgkin's lymphoma, Hodgkin's disease, Wilms' tumour, medulloblastoma, and retinoblastoma. No such trend was seen for acute non-lymphoblastic leukaemia, neuroblastoma, rhabdomyosarcoma, juvenile or anaplastic astrocytoma, brain stem glioma, histiocytosis X, or bone tumours. There is a need for continuing research into better methods of treatment of childhood cancer.

A case-control study of breast cancer, diet, and lactational history was carried out in Brisbane, Australia, between 1981 and 1985. Cases were recruited from both the private and the public health care sectors. The breastfeeding histories of 459 cases and 1,091 controls selected from the electoral roll were analyzed. An odds ratio of 0.85 (95% confidence interval 0.55-1.30) for any lactation versus no lactation was observed, with no suggestion of a monotonic trend with increasing duration. This estimate was only slightly altered by multiple logistic regression analysis. A statistically significant nonlinear association was noted for breastfeeding of the first liveborn child (chi 2 = 14.8, df = 6; p less than 0.05), with a slightly elevated odds ratio for lactations of less than one month, a lowered odds ratio for lactations of one month to three months, and a weakly elevated odds ratio in women breastfeeding for more than nine months. Results for premenopausal women were very similar to those for postmenopausal women. Taken together with other recent studies, these results suggest that lactation may play a modest direct or indirect part in reducing the risk of breast cancer.

Using the methodology of an earlier study of socioeconomic mortality gradients, we partitioned Brisbane City into five strata of equal size on the basis of suburb scores derived from aggregate socioeconomic census data. Numbers of deaths by stratum, age, sex and cause were obtained from mortality files. For almost all causes, mortality gradients had not changed between 1976–1979 and 1980–1987. A new category, medically‐preventable death under age 65, had lower rates in higher‐ranking suburbs. Potential years of life lost (PYLL) per unit of population, age‐standardised, were also computed by stratum and cause. External causes of death were the main contributors to PYLL among men, with a strong socioeconomic gradient, while neoplasms were most important among women, with little evidence of a social class effect. It is estimated that, in urban Australia, the annual number of additional deaths under age 65 due to socioeconomic circumstances is over 2000 for males and over 1000 for females. This study provides a baseline against which the programs of health advancement initiated in the mid‐1980s may be evaluated, or conversely, the effects of societal changes assessed.

Problems in defining social class and its relationship to health are discussed and the results of a study of infant mortality from 1976 to 1979 in socioeconomically ranked suburbs of Brisbane are presented. The ranking is according to a score developed from aggregate suburb data gathered at the 1976 and 1981 censuses; the population at risk was ascertained from published annual suburb birth figures, and numbers of death by suburb, age and cause, from computerised mortality files. Significantly higher mortality rates in lower ranking suburbs were observed. Possible explanations for, and implications of, the findings are discussed.

For the purposes of an area analysis of social class and mortality (from 1976 to 1979) Brisbane City was partitioned into five strata with roughly equal populations ranked on the basis of a socioeconomic suburb score derived from aggregate census data. Deaths by suburb, age, sex and cause were available from annual computerised mortality files and population age and sex distributions were computed from census statistics and inter‐censal estimates. Mortality rates were found to be higher in lower‐ranking strata overall, and for circulatory, external (such as accidental) and respiratory causes but not for neoplasms other than lung cancer. The implications of these results are discussed.

Routine data on mortality and hospital activity were used to estimate changes in coronary heart disease mortality, case fatality, and hospitalizations in the Australian state of Queensland over the decade 1971-1980. Acute myocardial infarction (International Classification of Diseases (ICD) 410) and other ischemic heart disease (ICD 411-414) were considered separately. For acute myocardial infarction, age-adjusted total mortality declined by about one fourth in both men and women; in-hospital mortality decreased somewhat more and deaths out of hospital correspondingly less. The age-adjusted case fatality ratio fell by the same amount (29%), in both sexes. Similar trends occurred at all ages. Admission rates decreased 11% in men and 18% in women. Similar patterns were evident for other ischemic heart disease except for admissions, which rose 23% among men and remained at the same level in women. These findings suggest that both declining incidence, particularly in the form of fewer deaths out of hospital, and improvements in care may have contributed to the general decline in coronary heart disease mortality in this community. Without direct measures of incidence or changing disease severity, the relative contributions of each factor cannot be examined.

We describe the attitude and views of general practitioners towards the menopause and hormone replacement therapy (HRT) in metropolitan Brisbane, Australia.

5. “Introduction to Statistical Inference,”by E. S. Keeping. New York: van Nostrand (1962), pp. xi + 451.

This paper considers the duration of seek times for a skip-serially processed file on a moveable head disc pack. It is shown that under certain assumptions about the generation of requests for records, the distributions of both the number of records to be processed and the number of distinct cylinders to be accessed per run are binomial. Probabilistic arguments from the theory of order statistics are used to derive the distribution of the number of cylinders separating consecutive cylinders accessed. A general functional form is introduced for the relation between inter-cylinder distance and duration of seek time, and used to derive a general expression for the expectation of total seek time for a scan of the file. Results are computed for a numerical example. These show, first, the advantage of software which ensures that successive scans operate in opposite directions; and second, the danger of serious inaccuracy (especially at low hit-ratios) if simple approximations are used based on the value of the seek time function at the mean inter-cylinder distance.

Critical values for Bartholomews test of the equality of means against an alternative of simple order are available only in special cases when the number of groups exceeds four. In this paper a Taylor expansion of the null distribution of the test statistic is developed, from which adjusted critical values in the general case are derived. The adequacy of the approximation is examined when the number of groups is four or six.

Motivated by a problem involving correlation due to genetic relationships within a sample, we investigate the effect of within-sample dependence and between-sample independence on the variance of the Mann–Whitney–Wilcoxon statistic, given that the marginal distributions are identical in each sample. The dependence increases the variance of the statistic by a factor equal in order to the combined sample size when three specified joint marginal distributions are chosen. The results for the two-sample statistic are extended to the mean of the Kruskal–Wallis statistic, when between-sample independence and within-sample dependence are assumed. The effect is similar in this case in that the mean of the statistic is increased by a factor of the same order as the sample size. A numerical example is given.

A formula for an upper bound to the propartional loss of information due to using some value of the ratio of the block component in inter-block variance to intra-block variance is derived. It is shown that this loss is small if the value used is not too discrepant from the true ratio.