Algorithm for the grouping of contiguous geographical zones

Health impact evaluation of boron in drinking water: a geographical risk assessment in Northern France

OBJECTIVE

To evaluate health impact of boron in drinking water.

METHODS

A regional scale geographical study in Northern France was conducted. Assessment of boron blood levels in a group of 180 healthy individuals and correlation with boron content in drinking water were followed by an assessment of health indicators such as birth rates, mortality rates, and sex ratios in zones of different boron content in drinking water.

RESULTS

After necessary adjustments, men living in municipalities with more than 0.30 mg/L of boron in drinking water had elevated but not significant boron blood levels compared with those living in municipalities with boron water levels of less than 0.30 mg/L (159.1 versus 123.0 ng/g; p > 0.05). The standardized birth ratio adjusted for the reference geographic zone and calendar time period was 1.07 and 1.28 in the low and high (>0.3 mg/L) boron content municipalities, respectively. The birth rate in municipalities with high boron content in drinking water was higher than that of the reference geographic zone and of the French general population (p < 10(-4)). The standardized mortality ratio adjusted for the reference geographic zone and calendar time period was 0.94 and 0.92 in the low and high boron content municipalities, respectively. The mortality rate in municipalities with high boron content in drinking water was less than that of the reference geographic zone and of the French general population (p < 10(-3)). No statistical difference was noted in the male-female sex ratios between the different municipality zones (p = 0.45).

CONCLUSIONS

The results of this study do not support the idea of a deleterious effect of boron on human health, at the boron water level contents found in this specific region. In fact, there is a tendency toward a beneficial effect with low-dose environmental exposure (less than 1 mg/L of boron) in drinking water.

Spatial partitioning using multivariate cluster analysis and a contiguity algorithm

Spatial analysis of epidemiological data can be a useful tool for identifying patterns of disease occurrence and can provide substantial support for prevention and control strategies. To obtain the greatest spatial resolution, it is important to use the smallest available areal units with homogeneous population. However, small areas usually have a small population, introducing spurious variability in the chosen indicators of disease occurrence. This paper describes an approach for combining small geographical units to stabilize mortality rates by pooling information across areas according to specified risk profiles. The procedure is based on a principal component analysis, followed by a cluster analysis of social-economic indicators to classify the risk profile of each small area. The classification is used in an algorithm to join neighbouring areas with similar profiles until an estimated population size is achieved. We applied this method to two Administrative Regions of the city of Rio de Janeiro, Brazil, using the census tracts as the basic areal unit. Census tracts were classified according to four socioeconomic categories distributed spatially as a mosaic, where tracts of differing categories neighbour each other. The aggregation algorithm produced a new partition of the region studied, with the created areal units preserving the internal socioeconomic homogeneity.

Aggregation of existing geographic regions to diminish spurious variability of disease rates

The availability of large data sets together with the growth in power and storage capabilities of computers have made the analysis of the spatial distribution of disease rates an increasingly important tool in public health research. Use of existing geographic divisions or groupings tends to result either in unstable estimates of disease rates if the corresponding populations are small or in loss of spatial resolution if the areas are unnecessarily large. This paper describes a computer algorithm for combining existing geographic areas into regions with populations large enough to diminish spurious variability in disease rates while limiting the loss in resolution. The method is demonstrated using Medicare hospital admissions data for pneumonia and central nervous system cancer. Disease rates were calculated for both predefined regions and those generated by the algorithm and their frequency distributions were compared. The algorithm produces more stable rates over a variety of diseases and provides substantially more flexibility than the use of predefined aggregations.

Assessing spatial patterns in disease rates

We describe the empirical performance of three indices of spatial autocorrelation (Moran's I, Geary's c and a rank adjacency statistic D) in the analysis of regional cancer incidence data. Heterogeneity in regional population sizes and age structure leads to variable precision in estimated rates; the usual methods for assessing I, c and D, which ignore such heterogeneity, are shown to be liberally biased, especially for c and D. The power of these indices to detect likely disease patterns is estimated by stimulation; the power is quite variable, depending on the exact pattern assumed, although I tends to have the highest power. The null distributions appear quite robust in small samples, even when several regions have no observed case. Preliminary work on the Ontario cancer registry showed generally unimportant effects on the spatial analysis of variation in case registration rates or missing residence data.

Use of computerised general practice data for population surveillance: comparative study of influenza data

OBJECTIVE

To assess the potential for using routine computerised general practice data for surveillance of illness.

DESIGN

Comparison of the incidence of influenza during the 1989 epidemic derived from a computerised database with that derived from the Royal College of General Practitioners's weekly returns service--a well established predominantly manual surveillance system.

SETTING

433 general practices throughout the United Kingdom that used a commercial computer system linked to a central databank.

MAIN OUTCOME MEASURE

Incidence of influenza.

RESULTS

The slope of the influenza epidemic curve was essentially the same whether derived from the routine computerised data or royal college's weekly returns service data, and the computerised data were geographically consistent. Throughout the study period, however, the computer derived incidence was between one third and one quarter of that derived from the royal college's system (which is served by practitioners trained in surveillance methods). The peak weekly rates were 164 cases per 100,000 for the computerised system and 583 cases per 100,000 for the royal college's surveillance system.

CONCLUSIONS

The apparent underreporting in the routine computerised data probably reflects lack of motivation and experience in disease surveillance and haphazard computer entry (particularly of consultations that took place outside of the surgery and consultations that did not result in a prescription), along with overestimation of the population under surveillance. Nevertheless, routine computerised surveillance allows rapid data collection from a large number of practices over a wide geographical area and would greatly augment existing methods.