In the context of a community health needs evaluation, we investigated the prevalence and age and gender distribution of noncommunicable disease in a peri-urban shantytown in Peru. Overall, there was a low prevalence of self-reported, diagnosed noncommunicable disease, although the prevalences of depression (12%) and chronic respiratory disease in the population (8%) signal that these conditions may contribute significantly to the NCD burden. Perhaps most alarming was the high frequency of excess weight, particularly among women; our findings of a 53% prevalence of overweight/obesity overall and 54% prevalence of abdominal obesity in women indicate a major risk for development of a noncommunicable disease. With regard to other risk factors for NCD, inadequate fruit and vegetable consumption by the majority (92%) of participants and a 15% prevalence of hypertension—the majority undiagnosed—in this population demonstrate the importance of these risk factors as potential contributors to NCD risk in this community.
Our estimates of NCD morbidity are generally similar to those of prior studies conducted in informal settlements in Peru [12, 15, 16]. However, our population prevalence of depression was substantially lower than the prevalence of probable mental illness estimated in the PERU-MIGRANT shantytown cohort (12 vs. 33-38%) . This difference could be explained by our definition of depression as having received a diagnosis of depression from a healthcare provider, which likely underestimated the prevalence of depression in our population. The difference also may be due to the fact that in the PERU-MIGRANT study, current rather than lifetime prevalence was measured; recall bias may have resulted in underreporting of lifetime depression diagnoses. The measure employed in the PERU-MIGRANT study may additionally have resulted in an overestimate of population depression prevalence due to its inability to distinguish between depression and anxiety . Our estimate of lifetime depression prevalence is, however, similar to that estimated for the adult population of metropolitan Lima (18%) . As compared to mental health research conducted in slum populations outside of Peru, the prevalence of depression in the present study is less than half of that measured in older adults  and women  living in slums in India, and is lower than the prevalence of self-rated poor or fair mental well-being measured in two Bangladeshi slum populations [33, 34]. These differences are likely attributable in part to the fact that in other studies the definition of depression or poor mental health did not require having received a physician diagnosis.
The only other study to our knowledge to report on the prevalence of chronic respiratory disease in a adult slum population, conducted in India, found similar prevalences of asthma symptoms (10%) and chronic bronchitis (8.5%) . In this population, asthma symptomology, but not chronic bronchitis, was associated with female gender and increasing age. It is possible that the heterogeneity of conditions captured by our chronic respiratory disease measure obscured condition-specific gender and age differences that may have been present in our study population.
The prevalence of arthritis measured in the present study is slightly lower than the prevalence of osteoarthritis of the knees found in a Bangladeshi adult urban slum population , which may be attributable to differences in the working conditions and physical demands of occupations, particularly for men, in these populations. Diabetes prevalences estimated in slum populations in Kenya [37, 38] and India  are similar to that of the present study while they are higher in Bangladesh (8%)  and in an elderly Indian population (18%) . The paucity of cancer and myocardial infarction data from adult slum populations limits our ability to compare our findings regarding these outcomes.
Similarly to the prevalences of NCD conditions, the prevalences of risk factors for NCD estimated in this study are also comparable to those of prior studies in Peruvian informal settlements [12, 13, 17, 18, 42]. Our population prevalences of hypertension, overweight status, and obesity are slightly lower than those estimated for the PERU-MIGRANT shantytown cohort, however, the age-specific prevalences of these conditions in our population suggest that this difference is due to the slightly younger age of our study population. The prevalence of overweight and obesity estimated in this study was similar to that of other slum populations in Nigeria  and Kenya  but greater than that of a different Kenyan slum  and greater than Indian  and Bangladeshi  slum populations. These differences in the prevalence of overweight and obesity are likely due to a number of factors and may be related to the progress of the epidemiologic transition or characteristics specific to the particular slum setting. In contrast to the findings of the PERU-MIGRANT study , women did not have a significantly greater prevalence of obesity defined by BMI, however, the greater prevalences of abdominal obesity in women in our study suggest that this finding may have been due to the limitations of our sample size rather than a true difference in results. Like the finding of the current study, research in Kenyan [37, 45] and Indian  slum populations found that women were more likely to have abdominal obesity while data from an Indian study provided evidence that abdominal obesity increased with age . Age and gender associations with abdominal obesity were not assessed in the Nigerian or Bangladeshi studies and the association with age was not assessed in the Kenyan studies. The association we observed between female gender and abdominal obesity may be the result of gender norms affecting individual health-related behaviors such as physical activity. The association of abdominal obesity with age likely reflects the cumulative effects of unhealthy diet and insufficient physical activity over the lifespan which we did not investigate in the current study but may further investigate in future NCD studies in this population.
Comparing to slum populations outside of Peru, our population prevalence of hypertension was similar to that of populations in Kenya  and India , but was considerably less than the prevalence of 38% measured in a Nigerian slum population  and considerably greater than the approximately 2% prevalence measured in a Bangladeshi slum . As in the present study, hypertension was associated with increasing age in previous studies conducted in Peru , India , Kenya  and Nigeria ; this association was not evaluated in the Bangladeshi population. This age-related trend with hypertension likely reflects the stiffening of blood vessels through ageing, although it could also reflect uncontrolled confounding by factors that are associated both with age and hypertension.
Like other research in informal settlements in Peru, we found a greater prevalence of binge drinking and smoking among men [17, 18], which is similar to patterns of hazardous drinking and tobacco use in the adult population of the Lima metropolitan area  and which likely reflect differences in gender-based social expectations . Similar gender differences in hazardous drinking and tobacco use have been noted in non-Peruvian slum populations [37, 39, 43, 45].
Because this study was conducted in the context of a broader health needs assessment, we designed our study instruments with the goal of balancing three objectives: 1) to capture a large number of health indicators on a broad spectrum of topics to describe the community burden of disease, 2) to collect valid measures of NCD-related variables, and 3) to limit participant burden. Although we did obtain an overall picture of the burden of communicable and noncommunicable disease in this population, we were unable to collect a number of NCD and NCD risk-related variables or to use more time-consuming, higher validity measures for each variable. The self-reported nature of the NCD variables, for example, likely led to underestimates of the prevalence of these diseases. With regard to NCD-related risk behaviors, only one aspect of unhealthy diet—adequate fruit and vegetable consumption—was measured, and we did not measure this variable using a high validity measure such as a food frequency questionnaire. Future investigations of diet-related NCD risk factors could additionally collect information regarding salt intake and saturated fat and trans-fat consumption, two other important aspects of diet-related NCD risk . Due to concerns about survey length, we also did not measure physical activity in our survey, and the inclusion of this risk factor will be important in future community surveys both as an important risk behavior related to NCD and to measure the impact of interventions and improvements in community infrastructure. The inclusion of biochemical measures would have allowed us to better measure certain NCDs and associated risk factors such as diabetes and high cholesterol and therefore should be included in future NCD studies in this community. Although our population estimates were weighted to account for the greater proportion of non-participation among males selected to participate in this study, the difficulty we encountered in contacting and in obtaining consent from potential male participants highlights the importance of coordinating the schedule of data collection with the availability of less accessible segments of the population and the potential need for participation incentives in order to reduce underrepresentation of men in the sampling of similar populations. Our relatively small sample size may have precluded the identification of other differences in NCD and risk factor prevalence by age and gender.
Despite these limitations, this study had a number of strengths. To our knowledge, this is the first population-based study to describe the burden of noncommunicable disease and related risk factors in a Peruvian shantytown. Aside from the randomized sampling and novel population of this study, other strengths include the objective measurement of BMI, abdominal obesity, and hypertension, and the use of a standardized questionnaire to assess NCD risk behaviors.