Article Text

Download PDFPDF

Incidence of insulin dependent diabetes mellitus in Karachi, Pakistan


OBJECTIVES To determine the incidence of insulin dependent diabetes mellitus (IDDM) among children aged up to 16 years residing in the city of Karachi, Pakistan, during the five years from 1989 to 1993.

DESIGN Retrospective study of incidence using hospital and clinic records.

SETTING The city of Karachi, Pakistan.

SUBJECTS Children satisfying standard criteria for the diagnosis of IDDM, attending treatment facilities for the first time during the study period.

MAIN OUTCOME MEASURES The incidence of IDDM in this population and its variation by age and sex.

RESULTS The incidence of IDDM in this population is 1.02/100 000 per year, which is one of the lowest incidence rates yet reported.

CONCLUSIONS The very low incidence of IDDM, contrasted with the substantially higher incidence among migrants, supports the view that environmental factors are the major determinants of variations in the incidence of this condition between populations.

  • insulin dependent diabetes mellitus
  • disease incidence
  • Pakistan.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The incidence of childhood insulin dependent diabetes mellitus (IDDM) is known to vary widely between countries. Rates among children aged 16 years and under vary from over 40/100 000 in Finland to less than 1/100 000 in Japan.1 2 In most European registries the rate is between 5 and 20/100 000 per year, with the higher rates generally in northern Europe and lower rates in southern Europe, apart from Sardinia.1 The reasons for this variation are unclear and the relative contributions of genetics and environment are a subject of debate.

The incidence of IDDM in India and Pakistan is not known, but it is believed to be low. This is suggested by several prevalence studies from India and Pakistan.3-5 No study of the incidence of IDDM has yet been published from any part of the Indian subcontinent, however.

The city of Bradford in West Yorkshire has a large population of children whose parents were migrants from rural Pakistan in the 1950s and 1960s. Children with IDDM from this group have the same high risk HLA markers and the same autoantibodies as other local children with IDDM.6 These children of migrants had a very low incidence of IDDM in the late 1970s. This has risen steadily since then and is now only slightly lower than that found in other local children.7 8

To explore these results further and to obtain the first set of incidence data from the Indian subcontinent, we carried out a study of the incidence of IDDM in the city of Karachi, Pakistan. Karachi is a large city, and a former capital, on the coast of Pakistan. The population is almost entirely Moslem. It is an important industrial centre and a major port.


All of the facilities in Karachi treating children with IDDM were identified by questioning local doctors. All of the larger facilities agreed to cooperate with this study; however, two small private hospitals declined to take part. Each identified centre was visited. Cases were identified using clinic and ward records and hospital notes were abstracted onto a standard form.

Children aged 16 years or under at diagnosis, diagnosed between 1989 and 1993 and normally resident in the city of Karachi at the time of diagnosis, were eligible for inclusion. Routine demographic information was recorded for each child, including dates of birth and diagnosis, sex, home address, and family history of diabetes.

It was not possible to apply capture-recapture methods to estimate our completeness of ascertainment as no independent secondary source of cases could be found.9 Pharmacy records for insulin prescriptions were not available, either centrally or in the pharmacies themselves. The records of the Diabetes Association of Pakistan were investigated, but their facilities provide treatment for members and, although it was an important primary source for cases, there was negligible overlap with other clinics. Primary care records are not centralised and patients often consult several different doctors and a range of traditional healers. No central records of hospital admissions exist.

The most recent census of Pakistan was in 1981 and this serves as the basis for population estimates. The census bureau made available to us local estimates for the population of Karachi by sex and five year age groups for each year from 1981 to 1995 (Pakistan Bureau of the Census, personal communication). These were derived by the Census Bureau from a simple growth model. Calculations were also performed using the 1981 census as a base to give a partial check on the impact of errors in the population estimates on our results.


The 1981 census recorded 2.5 million children aged 16 years and under in Karachi, 1.2 million girls and 1.3 million boys. The estimated 1991 population was 4.1 million children, 2 million girls and 2.1 million boys.

Two hundred and forty cases were identified aged 16 years or under and diagnosed between 1989 and 1993. Two hundred and eight were eligible as they were resident in Karachi at the time of diagnosis. The other 32 children had moved to Karachi for treatment after diagnosis elsewhere. There were 107 boys and 101 girls. The incidence per 100 000 per year was 1.01 for boys, 1.04 for girls and 1.02 in total, one of the lowest figures yet recorded (table 1). If the 1981 population is taken as a base the total rate is 1.67/100 000 per year.

Table 1

Rates (number of cases) of IDDM/100 000 children/year in Karachi from 1989 to 1993 inclusive

The peak incidence was between the ages of 10 and 12 years in both sexes, which is similar to the pattern usually seen in registries from Western Europe10 (fig 1). The highest rate was for girls aged 12 years. There was a slight female excess, rather than the male excess found in most registries, but this was not statistically significant (rate ratio 1.03; 95% confidence interval 0.78 to 1.36).

Figure 1

Incidence of IDDM/100 000 children/year in Karachi from 1989 to 1993 plotted against age and smoothed with a three year moving average.


Diabetes is one of the most important chronic diseases of childhood.11 A better understanding of its aetiology might prove beneficial to public health. Knowledge of the incidence of IDDM among children in different parts of the world is an important tool for epidemiology. Patterns of international variation in the incidence of IDDM provide material with which to develop and test hypotheses about the origins of this serious disease. This is the first such report from the Indian subcontinent.

This study confirms the exceptionally low incidence of IDDM in Pakistan and validates our earlier study of migrants from Pakistan.7 Western Europeans and inhabitants of the northern part of the Indian subcontinent share common ancestry and close genetic and linguistic links,12 so these observations suggest the importance of environmental factors in the origins of IDDM. There are many important environmental differences between the UK and Pakistan, including differences of diet, wealth, climate, and child health. Which of these contribute to the observed difference in rates is not known.

Although our study certainly missed some cases it is likely that most of the incident cases have been recorded. Although it was not possible to confirm the completeness of our case ascertainment directly, our searches of the available records were extensive and thorough. It is unlikely that a large proportion of cases reaching medical treatment centres could have been missed. It is the opinion of doctors in Karachi that almost all cases do reach medical care. Only 6% of the children in this study had attended traditional healers for their disorder (S Hanif and S Ahmed, unpublished observations). Furthermore, even an ascertainment rate as low as 50%, far worse than is likely, would not modify our conclusions significantly. The population estimates are also of unknown reliability. They are derived for use by planners and others working in the city, and are the best figures available. Even large errors in population estimates would not affect our conclusions.

This study shows the value of careful comparative studies of disease incidence between countries. Establishing that a difference exists between two countries and that it disappears in migrants from one country to the other is only a beginning. Understanding the reasons for occurrence is a possible route to a better understanding of the origins of IDDM and further work to this end is planned.