The epidemiology of infantile hypertrophic pyloric stenosis in Sweden 1987–96
- aDepartment of Pediatric Surgery, Drottning Silvias Barn-och Ungdomsjukhus, Sahlgrenska University Hospital, S-416 85, Gothenburg, Sweden, bDepartment of Pediatric Surgery, Astrid Lindgrens Sjukhus, Karolinska Vägen, S-171 76, Stockholm, Sweden, cDepartment of Mathematics and Statistics, University of Gothenburg, Gothenburg, Sweden
- Dr Hedbäck
- Accepted 12 July 2001
AIMS To find out whether the incidence of infantile hypertrophic pyloric stenosis (IHPS) has changed over the past decade, and if so, to investigate possible contributory factors.
METHODS All infants undergoing pyloromyotomy for IHPS in Sweden between 1987 and 1996 were studied. Using the national patient registers the yearly incidence was determined and evaluated in relation to sex, latitude, urbanisation, and type of surroundings by use of a Poisson model.
RESULTS There was a substantial decline from 2.7/1000 to 0.85/1000 over the time period. The incidence in the south was almost three times greater than in the north.
CONCLUSION The declining incidence and geographical difference suggest that environmental factors are of importance in this disorder.
For some years, we have had an impression that there has been a decline in the incidence of infantile hypertrophic pyloric stenosis (IHPS). This study was performed to investigate this observation, and if a decline has taken place, to investigate factors that might have contributed to it. Changes in the incidence of IHPS, both upwards1-6 and downwards,7-9 have previously been reported.
Patients and methods
All patients cared for as in-patients at medical facilities in Sweden are registered with the National Patient Registry which is maintained by the Swedish National Board of Health and Welfare. Using this registry, we identified all patients with a diagnosis of IHPS (No. 750F, ICD-9)10 during the years 1987–96. The patient series studied consisted of those individuals who were reported to have undergone pyloromyotomy (No. 4410, Classification of Surgical Interventions).11 For reasons of patient confidentiality, the registry did not allow determination of the precise age at surgery; however, all patients were less than 1 year old. The inclusion date of a patient was the date of arrival at the hospital before surgery. No one was included in the patient series more than once.
In each case patients were registered with respect to date of arrival at the hospital, sex, county, and local district of residence. The total number of patients treated each year was determined as was the total number of births each year, together with their sex and geographical location.12 Sweden consists of 24 counties subdivided into 288 local districts.
The probability (F) of a child undergoing pyloromyotomy for IHPS during the first year of life was estimated using a Poisson model13 incorporating the following factors: sex (X), time as year and date (Y), and latitude (Z), so that: Equation 1where β0 is a constant, X = 0 for boys, and X = 1 for girls, Y is the number of days after 31 December 1986, and Z is the latitude (in degrees) of the main town or city of the county in question. The β coefficients, their standard errors (SE), and their two sided p values were estimated according to the maximum likelihood method.
Equation 1 was modified by adding β4Y·Z, an interaction factor between time and latitude, so that: Equation 2Using equation 2 allowed the investigation of time trends in relation to latitude. In this modified function, all β coefficients were estimated separately for boys and girls instead of using β1X.
A second modification was performed by adding β5U and β6V, where U and V are factors considering other geographical differences: Equation 3As in equation 2, all β coefficients were estimated separately for boys and girls instead of using β1X. U is defined as follows: 1 = cities with >200 000 inhabitants; 2 = suburbs around these cities; 3 = cities with >100 000 inhabitants; 4, 5, 6, and 7 = town districts with >60 000, >30 000, >15 000, and >10 000 inhabitants, respectively; 8 = villages of <10 000 inhabitants; 9 = very sparsely inhabited districts. V is defined as: 1 = coastal districts; 2 = inland districts situated more than 100 km from the coast; and 3 = remote districts (mainly mountains and forests).
Between 1987 and 1996, a total of 2157 infants (1738 boys, 419 girls), underwent pyloromyotomy for IHPS (see table 1). Tables 2, 3, and 4 present the β coefficients with their standard errors and p values for equations 1, 2, and 3, respectively. Figure 1 shows the estimated probability of undergoing pyloromyotomy for a male and female infant during 1987 to 1996 if born in the south or north of Sweden. The p values of the β coefficients for sex (X), time trend (Y), and latitude (Z) were all significant (p < 0.0001). There was an overall decline in the incidence of IHPS during the study period, although when investigated separately the trend of decline for girls was not significant. The decline was not found to be related to latitude (Y·Z). Regarding the geographical factors studied (U and V), there were no significant differences, although there was a trend towards a lower incidence in boys in urban areas.
This study provides clear evidence of a decline in the incidence of IHPS between 1987 and 1996 in Sweden. The incidence of IHPS was almost three times higher in the south of Sweden than in the north. The registers employed for this study provide complete and reliable data: all medical facilities in Sweden are required to submit data on inpatients, their diagnoses, and surgical interventions; data submission is virtually complete for the past several years. This is possible because Swedish medical care is highly centralised in one nationwide organisation. The validity of the registers of the National Board of Health and Welfare has been confirmed in previous studies.14 Furthermore, the fact that the patients identified from the National Patient Registry coincided closely with those who were identified as having undergone pyloromyotomy supports the validity of the registers used. Only one discrepancy was found—an adult patient with incorrect diagnosis of IHPS. The reporting procedures in paediatric and surgical clinics did not change during the study period. Therefore the change in incidence could not be explained by registration procedures.
Previous epidemiological studies have shown various different trends. No change was identified in three studies—in Denmark from 1950 to 1984,15 in British Columbia during the 1970s,16 and in Western Australia from 1971 to 1984.17 An increasing incidence was found in Minnesota from 1950 to 1984,1 in central Scotland and other parts of UK,2-5 and in Germany during the 1970s and 1980s.6 Three previous studies showed a decline like ours—in Northern Ireland before 1970,7 in Saskatchewan from 1970 to 1985,8 and in New York State from 1983 to 1990.9 The latter study was by far the largest, including 4063 cases, and showed a fall in incidence from 2.4 to 1.7/1000 births. White race and male gender were associated with an increasing incidence of IHPS; low birth weight, increased maternal age, and high birth order were associated with a reduced incidence. A study from California on 1963 cases of IHPS during 1983 to 1988 showed no clear decline, although some other findings were confirmed: the incidence among white infants was 2.4 and among Hispanic infants 1.8/1000 births while it was 0.7 and 0.6/1000 births among black and Asian infants, respectively.18 The highest incidences have been reported in the UK, ranging from 3.5 to 8.8/1000 births.1-3 In the Saskatchewan study bottle feeding was 2.9 times more prevalent among the IHPS infants.8
From the results of these various studies, it appears that IHPS occurs with a varying frequency over time and in different regions with incidences ranging between 1.7 and 8.8/1000 births. Occurrence depends on genetic factors, gender and race being affected differently, although minor environmental factors like type of feeding also appear to be important.
In our study, the finding that the incidence of IHPS was almost three times higher in the south of Sweden compared to the north could indicate unknown environmental factors of importance. We found that the substantial decline in incidence occurred throughout the whole country during the ten year study period. This change in incidence might be explained if IHPS is being transmitted like an infectious disease, although the low incidence would suggest that individual susceptibility would have to be variable and infrequent. This susceptibility may depend on both hereditary and environmental factors. It is impossible to say whether environmental or genetic factors are most important in determining the difference in incidence between northern and southern Sweden. The only obvious difference between the north and south is the difference in climate. The incidence of IHPS at the end of the study period, especially in northern Sweden, is the lowest reported so far among white infants.
This study was supported by grants from the Göteborg Children's Hospital Research Foundation, and The Wilhelm and Martina Lundgren's Research Foundation II.