Serious infections in children: an incidence study in family practice

The incidence of acute infections in children is high. Using data from the English fourth Practice-Based Morbidity Survey, children under 1 year of age reported an average of 3.3 infections in a year, and 1.2 infections in the age group 1–15 years [1]. Acute upper respiratory infection is the dominant diagnosis, which is essentially a harmless, self-limiting disease.

However, some infections can have serious consequences. For example, bacterial meningitis leads to mortality in 5% of the cases and morbidity, such as hearing loss, in 10%–15%[2]. Up to 24% mortality after meningococcal disease, being sepsis or meningitis, has been reported[3]. In developing countries, pneumonia and diarrhoea are two of the leading causes of mortality in children under the age of 5[4]. In the period of 1989–1993, infectious diseases were responsible for 7.5% of all deaths in children under the age of 1, and for 9.3% of all deaths in children between 1 and 14 years old in the Northern region of the UK[5].

Serious infections are usually defined as serious bacterial infections: pneumonia, sepsis, meningitis, pyelonephritis, bacterial gastro-enteritis, osteomyelitis and cellulitis. [6‐9]. However, pneumonia and meningitis are often caused by viruses and other viral infections can be serious as well, for example bronchiolitis. In general practice, bacterial aetiology is difficult to establish and the serious nature of the infection is sometimes determined by the need for interventions such as hospitalization or specific treatment. Therefore, in primary care pneumonia, sepsis, meningitis, bacterial gastro-enteritis, pyelonephritis, osteomyelitis and bronchiolitis can be considered serious infections.

The diagnosis of a serious infection is difficult in general practice, as the majority of the infections presented to the general practitioner is not serious and the initial presentation of serious and non-serious infections can be similar. Still, early diagnosis of a serious infection is important to avoid delay in treatment and improve prognosis [10‐12].

There is very limited information on the incidence of serious infections in general practice. Most studies were performed in hospital. They have found serious infections such as pneumonia, meningitis, sepsis and bacterial gastroenteritis to be present in up to 25% of febrile infants presenting to the emergency department[13, 14]. These results can not be transferred to general practice, because of the large differences between the populations. Other studies in general practice have reported incidences for individual infections [15‐17], but differences in health care structure limit the applicability to the Belgian context.

Therefore, we conducted this study to calculate the age-specific incidence of serious infections in children, seen in general practice in Belgium. Additionally, we determined the proportion of serious infections in the entire group of children with an acute infection and we studied seasonal influence.

Intego is a morbidity registration network, coordinated by the department of general practice of the Katholieke Universiteit Leuven. 51 general practitioners in 43 practices continuously register all diagnoses and additional data in their electronic medical records. These practices have been selected by the quality of their registration. The information is automatically extracted from the records on a yearly basis and collected in a central database. Diagnoses are classified according to a very detailed diagnostic classification list and converted into ICPC-2 codes[18, 19].

As there is free access to all physicians in Belgium, and patients are not forced to register to a practice, it is difficult to establish the practice population. This is particularly important as the incidence of a disease in a certain practice equals all new cases of this disease in that practice, in the nominator, divided by the practice population in the denominator. Until recently, only the group of patients that had consulted their GP at least once in that year has been used as denominator, called the Yearly Contact Group. But, as it was hypothesised that some factors have an influence on the consultation rate of a patient, for example age, gender or social status, it was not possible to extrapolate this Yearly Contact Group to the total practice population, because these influencing factors can be present in certain practices more than in others. Now, data of the Health Insurance Agency have been made available, by which it was possible to estimate a practice population, starting from the Yearly Contact Group and taking age, gender and region into account. This population will be referred to as the Practice Population. A more detailed description of this methodology has been published elsewhere[20].

For children, the difference between the Yearly Contact Group and the Practice Population is particularly large, because some children never consult a general practitioner but only a paediatrician. This makes the estimation of a practice population less appropriate for this age group, because the extrapolation to a larger group that never visits the general practitioner is not very useful.

For this reason, the main outcome measures are presented for both the Yearly Contact Group and the Practice Population. Secondary outcome measures are presented for the Yearly Contact Group only.

Incidences are calculated as the number of events per 1000 patient-years and based upon the data from 1998 to 2002. If a disease occurred more than once during the study period, each occurrence was calculated.

The diagnoses that were analysed are presented in table 1, with their corresponding ICPC-2 code. Serious infections are marked with *.

The ICPC-A78 code, other infection not otherwise specified, contains several diagnoses, including sepsis. We extracted the individual diagnoses of sepsis from this group and added them to the code of meningitis, resulting in a combined diagnostic category meningitis/encephalitis/sepsis. On the basis of our data, it was impossible to identify bacterial gastro-intestinal infections within the large group of all gastro-intestinal infections, neither was it possible to separate bronchiolitis from acute bronchitis. These infections could therefore not be included in the selection of serious infections. Serious infections were therefore defined as pneumonia, sepsis, meningitis, osteomyelitis and pyelonephritis.

Analyses were performed with STATA statistical software and Epi-info. 95% confidence intervals are based upon a Poisson distribution if there were 100 cases or less, and on a Normal distribution in all other cases. Differences between age groups were tested with chi² for trend. A trend over subsequent years was analysed with linear regression analysis, using the year as the independent and yearly incidence as the dependent variable. The seasonal influence was tested in SAS with the combined test for the presence of identifiable seasonality, which is a joint test of both stable and moving seasonality. In the first test, the null-hypothesis is that there is no difference in incidence between seasons. The second test tests whether the seasonal incidence has moved over the last years.

The incidence of acute infections in children is high. For every 1000 children between 0 and 14 years, over 1100 acute infections are presented to the general practitioner in Belgium every year. The vast majority of these infections are self-limiting upper respiratory tract infections.

In contrast, serious infections are relatively rare: only 12.0 serious infections per year in every 1000 children between 0 and 14 years old. This is an annual incidence of little more than 1%, and accounts for 1.1% of the selected acute infections in this age group. As we were unable to separate bacterial gastro-intestinal infections or bronchiolitis, the true incidence of serious infections is higher, especially in the youngest age group.

Incidence of acute infections seems to be declining over the last five years, which has been described before[21]. For serious infections, this evolution is less clear.

In 2000, there was an outbreak of meningococcal C infections in Western Europe, leading to widespread vaccination programmes in children. This is reflected in our results.

Incidences are highest in the youngest children between 0 and 4 years old. This is true for all acute infections (1732 per 1000 patients per year in the age group 0 to 4 years compared to 1107 in all children, which is an excess of 64%), as well as for serious infections only (21.0 between 0 and 4 years compared to 12.3 in all children, which is an excess of 75%). This peak incidence during the first years of life has been described before [22‐25]. The only exception in our data was the incidence of sinusitis, where incidence rises with increasing age. This can be explained by the fact that sinuses are not yet well developed in very young children.

Incidence rates vary during the year, with all acute infections and pneumonia being more frequent in the winter months. For other serious infections, no particular pattern could be observed, although the peak of meningitis in the fall and that of pyelonephritis in the warmer season need to be examined further.

Our results were subsequently compared to other morbidity registrations. Data from three neighbouring registrations were available, being the Transition project²¹ and the Continuous Morbidity Registration (CMR), both from the Netherlands, and the Weekly Returns Service, from the UK[21]. Data were not available from all networks for every diagnosis, except for pneumonia, but the incidences of the Intego network, the Transition project and the Weekly Returns Service are within the same range.

The CMR reports higher incidences for pneumonia as the other three registrations: in children aged 0 – 4 years, the yearly incidence is 27–30 per 1000 patients, as compared to 18.3 in the Transition project, 18.1 in our present study based on Intego and 15.7 in the Weekly Returns Service (the latter is based on children from 1–4 years old). Another outlier is the very high incidence of meningitis in children under the age of one, reported by the UK Weekly Returns Service: 9.4 infections per 1000 children per year, as compared to a yearly incidence of 1.4 per 1000 in Intego and 1.3 per 1000 in the Transition project, both based however on children from 0 to 4 years old.

But, any comparison between the different morbidity registrations is difficult. Firstly, populations are different. For example, the health care structure in the UK is totally different from that in Belgium, as there is free access to specialist care in Belgium. In general, GPs are informed about all diseases of their patients including those who went directly to a specialist. However, for young children who only have been seen by a paediatrician and not by a GP, this may not always be the case. Therefore, we used the Yearly Contact Group instead of the Practice Population for most analyses.

Although the method for estimating the Practice Population is robust as it takes several influencing factors for consultation rate into account, the true Practice Population remains undefined[26]. Even in those countries with a fixed registration to a particular GP, the true Practice Population is constantly changing. In the Belgian context, uncertainty on the true practice population is even more important. Therefore, calculations using the Practice Population should be considered with caution.

Because of free access to specialist care, our results should be interpreted as the incidence rates of infections in general practice only. Any decline in incidence could reflect a true decline or a shift in consultation practices. This 'competition' between general practice and paediatricians explains the lower incidence of meningitis in our data as compared to other morbidity registrations, because parents with very young children in Belgium tend to consult the paediatrician. The influence of differences in health care structure could be the scope of further research.

The period of registration is another influencing factor, as incidences of acute infections seem to be decreasing during the last decade[21, 27].

Finally, diagnoses are defined differently. The Transition project uses the ICPC, the Weekly Returns Service uses the ICD coding system, the Continuous Morbidity Registration an adapted version of the E list and the Intego network an extensive own list, but we recoded all diagnoses according to the ICPC-2 for analysis. For example, the Weekly Returns Service reported a higher incidence of meningitis of 9.4 infections per 1000 children less than one year old, which can be explained by the fact that this database contains unconfirmed cases.

In contrast to the total incidence of the selected acute infections, serious infections are rare in general practice, around 1% per year. Children younger than 4 years old have the highest risk for serious infections, and incidence of pneumonia peaks in the winter months.