We estimate incidence and mortality rates and standardize them by age group, sex, and availability of health services. Using the standardized rates, we analyze the social distribution of NCDs, finding that the NCDs are concentrated in municipalities with higher socioeconomic levels, indicating that higher levels of socioeconomic development do not necessarily lead to better health in Bolivia.
Several factors could explain our results. One likely explanation is that the living and working conditions can lead to harmful lifestyles, leading to NCDs . For instance, urban areas in developing countries have generally experienced an increase in NCDs due to the adverse effects of globalization. Thus, a considerable proportion of global marketing promotes tobacco, junk food, and alcohol consumption. Similarly, rapid and unplanned urbanization changes people’s way of life through increased exposure to shared risk factors, such as unhealthy diets, sedentarism, air pollutants (including tobacco smoke), and harmful alcohol use .
In Bolivia, the NCDs with the highest incidence are hypertension and diabetes. Following the aggregated pattern, these diseases are concentrated in municipalities with higher socioeconomic status. As we describe below, hypertension and diabetes are contingent on behavioral risk factors such as unhealthy diets and sedentary lifestyles, more likely adopted in urban settings.
Hypertension originates from genetic and environmental factors. Although the genetic predisposition to hypertension is not modifiable, this disease is highly preventable due to the strong influence of environmental risk factors and lifestyle, among which are: excessive weight gain (which leads to overweight and obesity), unhealthy diet, excessive consumption of sodium, and insufficient consumption of potassium, high-stress levels, insufficient physical activity, and excessive alcohol consumption [29, 30].
Changing people’s environment and lifestyles can prevent Diabetes Mellitus. In particular, the probability of getting sick with Type 2 diabetes, which corresponds to 90% of cases worldwide, is strongly associated with overweight and obesity. The most relevant risk factors are excess adipose tissue, high body mass index, abdominal and visceral fat . So, regular physical activity [32, 33] and diets low in saturated fat or trans-fat can reduce the incidence of this disease [31, 34]. In some cases, adopting these lifestyles may be more effective than pharmacological solutions; the difficulty lies in making these behavioral changes permanent and effective in vulnerable populations .
Furthermore, hypertension, diabetes, and obesity are among the main risk factors for other cardiovascular diseases - ischemic arrest, hemorrhagic arrest, ischemic heart disease, heart failure, peripheral arterial disease, and Chronic Kidney Diseases [35–37]. Also, some risk factors, including a diet high in saturated fats, affect the likelihood of getting breast and prostate cancer [38–42], two of the most common cancers in Bolivia. Consistent with the idea that diseases sensitive to harmful lifestyles explain the inequality of NCDs, we also found positive concentration indices in cancer, other cardiovascular diseases, and Chronic Kidney Diseases (column 2, Table 1).
Wagstaff decomposition analysis showed that urbanization contributes to the concentration of hypertension and diabetes in municipalities with high socioeconomic status. Higher levels of urbanization come with improvements in socioeconomic conditions, but they also increase access to processed or ultra-processed foods (convenience stores) and reduce access to recreational spaces (parks) . In this way, urbanization favors adopting unhealthy lifestyles, which would increase hypertension  and diabetes . In addition, several studies find that an improvement in socioeconomic status can reduce the incidence of NCDs. However, if this increase reaches the point of allowing families to transition to a diet with higher calorie content, it can produce the opposite effect .
The presented evidence suggests that poor municipalities with indigenous populations and nutritional deficits (anemia) are not likely to suffer from hypertension or diabetes (See Table 3). Poor municipalities may be in an earlier phase of the epidemiological transition, where infectious diseases are still more prevalent than NCDs. For example, some infectious diseases related to soil-transmitted helminths increase iron deficiency anemia, which is more common in lesser developed countries . Alternatively, we can interpret the contribution of the percentage of the indigenous population to hypertension and diabetes inequality as evidence of horizontal inequalities, i.e., inequalities between culturally defined groups. In this case, indigenous populations' culturally specific economic conditions or behavior may reduce the chances of developing these diseases .
On the other hand, we observe that education reverses the inequality of hypertension and diabetes, being particularly relevant for the first disease (See Fig. 3). Municipalities, where higher educational levels accompany a higher socioeconomic status, do not concentrate incidences of hypertension or diabetes as high as municipalities of similar socioeconomic status but with lower educational levels. For example, evidence from other countries shows that education reduces hypertension  and sedentary lifestyle , one of the main risk factors for diabetes. In addition, policies that create physical spaces and times to perform exercises in neighborhoods, educational units, and workplaces are more effective if accompanied by health education, support groups, and counseling . Similarly, although its effect is minor, access to basic sanitation also reverses the concentration rate, probably due to its association with the quality of the urban environment.
Rheumatoid arthritis escapes the pattern described above as it concentrates in poor municipalities. Furthermore, the relationship between behavioral risk factors and the incidence of this disease is not as clear. Rheumatoid arthritis is an autoimmune disease of chronic inflammation that destroys joints and bones, causing disability and early mortality. The most mentioned risk factors are genetic, hormonal, dietary, sex, infectious agents, and smoking . Recently, the disease has been linked to abdominal obesity and environmental factors associated with people’s occupation [49, 50], among them, stress, moisture, vibration, asbestos, fertilizers, crops and foliage, and mineral dust (silica) [51–54]. People living in rural areas of the country, usually poor, tend to be exposed to some of these environmental factors, possibly explaining rheumatoid arthritis’ negative CI. The higher incidence of rheumatoid arthritis in poor municipalities could lead to poverty traps in which the disease, when not treated on time, generates disability, which generates greater poverty .
Finally, the heterogeneity of NCDs incidence in Bolivia suggests implementing policies that prioritize municipalities with high incidence rates. Most of the diseases included in the study concentrate in municipalities with high socioeconomic status, and they share common risk factors. Likely, actions aimed at reducing the inequality of highest-incidence diseases will affect most of the other diseases. The main risk factors for these diseases coincide with the modifiable factors prioritized by the WHO . Hence, its strategy to reduce the prevalence of NCDs could be a reasonable starting point for designing municipal policies in Bolivia.
Despite our data processing and estimation efforts, the quality and availability of data limit the study results. We use epidemiologic surveillance records from the HMI in Bolivia, whose validity relies on health services' accessibility and correct identification. Therefore, NCD’s incidence rate could be either under or overestimated depending on local health services conditions. We standardized the incidence rates by health service availability to reduce this bias in the inequality analysis. Nevertheless, it remains a concern for the aggregated estimates where we cannot control for the overall performance of the HMI. In addition, it also concerns the disease-specific rates as data quality could vary by disease.
On the other hand, we had no information at the municipal level of social determinants directly related to the incidence of NCDs, such as access to processed food, recreational spaces, variables of environmental contamination. Hence, we focused on more general social determinants, limiting our capacity to understand the underlying mechanisms of NCD’s inequality.