PEDIATRICS Vol. 106 No. 3 September 2000, pp. 620-621

First Urinary Tract Infections in Swedish Children

To the Editor.

In their epidemiologic analysis of first urinary tract infections (UTIs) in Swedish children, Jakobsson et al¹ make the observation that UTI rates varied seasonally and declined over the 2-year study period. The authors do note that such rate trends are biologically implausible, but fail to acknowledge several methodologic sources of bias that may have distorted their results.

A single, mid-study population estimate was selected as the denominator for all rates, despite a 5% decline in the base population (children <2 years of age) in a single year. If this decline is attributable to a drop in the birth rate, then the person-month contribution of newborns may have declined 10% over the 24-month study period. Given such a rapid decline in the age group at greatest risk for UTI, a fixed population denominator appears to be an inappropriate denominator for the calculation and comparison of annual rates.

By using a fixed denominator for all rates, the study analysis also fails to account for the gradual attrition of some 2300 children from the study population over 2 years. This declining denominator occurs because the authors have studied only "first" UTIs. By definition, children with UTI contribute no additional person-time to the at-risk population beyond their dates of diagnosis.

The most important source of bias is the striking seasonality of births in Swedenwhere monthly birth indices are at least 30% higher in spring than early winter.^2,3 As a result, rates for any illness associated with early infancy can be expected to show substantial seasonal fluctuation. The importance of adjusting for season of birth is illustrated in a recent analysis of seasonal risks for sudden infant death syndrome in Sweden.⁴

Taken together, these factors suggest that simple monthly or annual counts of UTIs (numerator data) may not fairly represent true incidence. Accurate denominator data are needed. If the authors would consult available birth and census data, it should be possible to calculate age-specific, person-time denominators for each month of the study period. This would permit calculation of true incidence rates and an unbiased examination of rate ratios for possible seasonal or year-to-year trends.

M. Bruce Edmonson
Department of Pediatrics
University of Wisconsin-Madison
H6/440 University Hospital
Madison, WI 53792

REFERENCES

1. Jakobsson B, Esbjörner E, Hansson S Minimum incidence and diagnostic rate of first urinary tract infection. Pediatrics. 1999; 104:222-226 [Abstract/Free Full Text]
2. James WH Seasonal variation in human births. J Biosoc Sci. 1990; 22:113-119 [Medline]
3. Lam DA, Miron JA Global patterns of seasonal variation in human fertility. Ann N Y Acad Sci. 1994; 709:9-28 [Medline]
4. Haglund B, Cnattingius S, Otterblad-Olausson P Sudden infant death syndrome in Sweden, 1983-1990: season at death, age at death, and maternal smoking. Am J Epidemiol. 1995; 142:619-624 [Abstract/Free Full Text]

In Reply.

We would like to thank Dr Edmonson for his valuable comments. The main message of our paper was the high minimum incidence of UTI in the studied age group. This message stands unchanged despite the possible methodologic difficulties pointed out by Dr Edmonson. The seasonality and the decline of UTI during the study period were unexpected findings, and we appreciate the possible explanations provided by Dr Edmonson. In addition to our original data, we have now used data on the number of children born at different periods. In the new analyses the current ages and the study period were divided into intervals of 14 days, thus taking into account variations in birth rate over time. The study population was handled in the same manner. Girls and boys were studied separately. Using this approach, the results show that girls above 6 months of age run a higher risk of contracting UTI during the summer than during the winter. In fact, at 2 years of age the estimated risk during June was twice as high as that of December. After adjustment for seasonal variations in birth rate, no seasonality was observed in the male population or in the youngest girls. Therefore, the increase in UTI during the summer, reported in our paper, may partly be explained by an increase in birth rate and partly by a genuine seasonal variation in risk for girls above 6 months of age.

In the infant population, an unknown amount of UTI will not be detected. This is supported by preliminary data, where we found an increase, during the study period, in the number of UTI diagnoses in the participating centers, which was not found in the nonparticipating centers (unpublished data). Therefore, the UTI awareness may have increased in the participating centers. However, a decline in UTI awareness during the 2-year-long study period, in centers not continuously engaged in UTI research, may explain the decline in UTI observed, as discussed in our paper. Therefore, we do not believe that the true incidence of UTI in infants can be determined, even if an accurate denominator is used. Instead, we prefer a more cautious approach and to report on the minimum incidence of UTI. To our satisfaction, Dr Edmonson's useful comments indicate that our approach may have underestimated rather than overestimated the size of the population.

Birgir Jakobsson
Elisabeth Esbjörner
Sverker Hansson
Department of Pediatrics
Huddinge University Hospital
S-141 86 Huddinge
Sweden