We report baseline cross sectional data from a randomly recruited cohort in a typical transitional community in Karachi. Our results show high rates of anthropometric and biological markers of adverse health in this community. Mean (22.4 kg/m2
) and median BMI (21.5 kg/m2
) were comparable to a population-based national estimate of average BMI for adult Pakistanis, as well as in a more recent community-based study from India 
. Mean BMI was comparable to that in other Asian populations as well 
. Greater than one in four participants was overweight or obese. This prevalence was higher than that found among urban residents from the National Health Survey of Pakistan (NHSP) 1990–94, despite the inclusion of younger adults between the ages of 15 and 24 in our sample 
. Meanwhile, the estimate for overweight and obesity was found to be similar to other South Asian populations, and much higher than in other Asian settings such as urban Indonesia and Vietnam 
. A similar pattern was observed with waist circumference and abdominal obesity, with comparable average estimates and prevalence as in other Pakistani and South Asian populations 
. The overall prevalence of hypertension in adults aged 15 years and above was 18%. This prevalence was lower than national estimates on urban Pakistanis from the last National Health Survey of Pakistan 1998 
. However, another nationally representative study looking at ethnic differences identified similar low rates among Sindhis 
. This ethnicity represents a majority of the IHCC cohort population, potentially explaining the overall lower-than-national prevalence of hypertension. These rates were also lower compared to other Asian populations 
. Another third of the population were pre-hypertensive and hence at elevated risk for the condition.
Our study was the first in Pakistan to measure glycated hemoglobin levels in randomly selected participants from the general population. Most previous studies comparing A1c levels have been carried out on diabetic or other ambulatory patients in the Pakistani hospital setting 
. Compared to a survey conducted among the adult English population, the mean A1c level found in our sample was lower than that found in South Asian immigrants (5.5% vs. 5.9%, respectively) 
. However, this difference may be partially attributed to the inclusion of younger adults aged less than 35 years in our sample who may have lower circulating glycated hemoglobin. Moreover, our study identified eight percent of the population with diabetes, and this level was higher than that identified previously in the National Diabetes Survey 1994 using 2-hour venous plasma glucose 
. More importantly, we found that an additional two in five individuals had A1c levels indicative of impaired fasting glucose or glucose tolerance. In addition only 3.7% (95% CI: 2.4–5.9) of all individuals screened for diabetes mellitus were on treatment and most of these patients (59%) were still hyperglycemic. Average total cholesterol (163.6 mg/dl) in our population was found to be similar to that in Pakistani ambulatory patients without heart disease, but lower than that found in other South Asian general populations 
. Compared to data from the NHSP 1990–94 where 13% of Pakistani adults had elevated serum cholesterol our study found a slightly higher percentage of adults with borderline-to-high cholesterol (16%), despite the inclusion of younger adults in our sample 
. This proportion was similar to that found in other urban South Asian settings 
The prevalence of current daily smoking (12%) was found to be lower in our population than previously reported for urban areas in Pakistan (15%), the difference likely owing to the inclusion of adults aged below 18 years in our sample as compared to that used by WHO 
. More alarming was the high level of smokeless tobacco use (45%) in this population. This finding was similar to previous studies in adult patients at family clinics (52%), and a squatter settlement in Karachi (nearly 40%) 
These findings suggest a high burden of non-communicable diseases maybe emerging in developing megacities like Karachi. As we follow this cohort over time, we anticipate that we will see the burden increase. The combined burden of NCD and infectious diseases on Karachi is very likely undermining economic growth, resulting in new patterns of health inequities. We recently reported very high rates of tuberculosis (350 per 100,000 population) in an area that included the communities surveyed here 
. Along with the 8% of the general population infected with HCV in this cohort, these infectious diseases and others will continue to take an unacceptable toll on lives, health expenditures and economic productivity well into the future. While a young population with increasing life expectancy rates and declining infant mortality signifies a demographic transition in Karachi, this is characterized by rapid and unplanned urbanization leading to increased social inequality in the form of low education attainment, limited access to safe drinking water, substandard housing infrastructure and crowded living conditions. Meanwhile, the public health care system has not been able to cope with rapid population growth and changes in disease patterns 
Our results concur with emerging studies in major urban centers of South Asia identifying the linkages between levels of urbanicity and lifestyles characterized by unhealthy nutrition, reduced physical activity and tobacco consumption 
. Recent studies in major urban centers of India and Sri Lanka developed an urbanicity scale to measure the linkages between urban environment and NCD risk factors. Both studies concluded that a relationship exists between urbanicity and NCD risk factors, where urbanicity represented access to markets, transport, communication, health care services, educational attainment and population density 
There are few comprehensive studies from developing countries on NCD risk factors in communities such as this one, and limited experience in the setting up of NCD and infectious disease surveillance models in low-resource urban settings. Ours is the first such study among randomly selected households from a transition population in Karachi, and it required the investment of considerable up front time and resources in participatory planning and community mobilization. This outreach effort proved to be valuable in enlisting the support of community organizations and leaders while building confidence in the Indus Hospital in whose catchment area they reside. The community engagement strategies we have developed will be invaluable in our next enrollment phase and future cohort follow up. We anticipate that our investment in the long-term community relationship will help us in designing socially acceptable and community-based participatory interventions in the future. Our findings reinforce the need to develop methodologies and tools to monitor NCD risk factors in developing countries that may help advance community-based intervention models to prevent NCDs in transition communities.
There were several limitations to our study. First, our findings are limited to baseline data. Longitudinal follow up of the cohort is needed to demonstrate longer term health outcomes. We plan to generate further data through the expansion of enrollment in phase two and follow up of the current cohort to yield a more representative sample for Karachi. Another limitation was low participation and underrepresentation of men in the health examination at the mobile health unit despite a significantly higher number of informed consents attained at the household. In the local culture, permission of the head of household was required for participation in the health examination. The household head was usually a male and was absent for extended periods for work. Among actual attendees we found more women than men because they were often at home. Proliferation of non-profit, governmental, commercial and academic research in poor and marginalized communities may have lowered the interest and motivation for participation in some households. Similar patterns of declining participation have been observed in the other parts of the world including among major epidemiological studies in developed countries 
. Known diabetes was not taken into account when estimating the prevalence of diabetes in this study as access to diabetes care in these populations is limited and individuals with diabetes likely suffer from persistent hyperglycemia. The missing data for study was less than 1% for most variables and not more than 2.5%. Missing value analysis showed that the data were missing completely at random and hence this is only a minor limitation.
The strengths of the study are significant, as we recruited the first randomly selected cohort in Karachi to generate data on the burden of NCD and infectious diseases and prevalence of associated risk factors. Accurate sampling of households in low- and middle-income megacities is challenging. The absence of recent census data necessitated the labor-intensive creation of a GIS household database allowing efficient and accurate sampling. The representativeness and success of random selection is suggested by the near-identical distribution of household member age and gender distribution to that found by the Pakistan Demographic and Health Survey 2006–07 
Our findings indicate that the future burden of disease – infectious and non-communicable - will likely increase in a population with already poor access to quality health care. Karachi is one of the fastest growing urban centers in the world, as suggested by preliminary 2011 census data 
. The social, economic and health implications of this rapid growth will be long-lasting, and there will be a great need for community-based data to plan for health promotion in these communities. Public health programming in Pakistan has been characterized by a range of vertical disease-specific programs that include the national programs for immunization, nutrition, maternal child health and programs aimed at prevention and control of malaria, tuberculosis and HIV/AIDS. The national health system has paid little attention to controlling NCDs and developing an integrated NCD surveillance model. This represents a major weakness in Pakistan’s public health planning in particular for its rapidly expanding multi-ethnic urban communities 
. Recent changes brought about by a hasty abolition of the Ministry of Health and devolution of its responsibilities to the provinces increased challenges for local public health systems, but also provide an opportunity for a new beginning 
. Our findings highlight the need for developing NCD intervention programs in coordination with existing disease control programs, and for establishing active surveillance for effective planning, implementation, and evaluation. Comprehensive control of NCD risk factors and infectious diseases in multi-ethnic Asian megacities has become an absolute necessity.