Browsing by Author "Vissoci, João"

Background: Primary care sensitive conditions (PCSC), a classification of illnesses that includes noncommunicable diseases (NCDs) and maternal health complications, are considered preventable through appropriate care management and interventions at the primary care (PC) level. Consistent with trends in global disease burden, PCSC are a significant contributor to avoidable hospitalizations in low and middle income countries (LMIC), which carries profound social and economic consequences. Rates of hospitalizations for primary care sensitive conditions (HPCSC) have been found to be associated with the level of infrastructure of health services delivery, health system, and socioeconomic context. This study concentrates on the Brazilian state of Minas Gerais to evaluate the current profile of HPCSC and their predictors under the universal PC program, the Family Health Strategy (FHS).

Methods: This is an ecological study based on: 1) data of PC infrastructure from 560 municipalities, collected from 2012-2013 through the Programa Nacional de Melhoria do Acesso e da Qualidade da Atenção Básica (PMAQ-AB), 2) data on rates of HPCSC available in the Hospital Information System of the Unified Health System, and 3) data on health system and socioeconomic indicators from the Brazilian Ministry of Health and the Brazilian Institute for Geography and Statistics, respectively. For the analysis, 7 groups of PCSC specifically targeted under the FHS were considered. 24 structure and process indicators were selected from the PMAQ-AB database and a principal component analysis with factor interpretability was performed, utilizing the theoretical rationale of the Starfield Model of Primary Care, to reduce and describe data dimensionality. Principal component scores were averaged by municipality, and assessed as predictors of HPCSC across municipalities in multiple regression models both individually and progressively adjusting for health system and socioeconomic variables as groups.

Results: From January-December 2012, municipalities in our sample experienced 12,078 HPCSC due to the 7 conditions chosen, with an aggregate age-adjusted rate of 112.15 per 10,000 inhabitants. The NCDs of congestive heart failure, cerebrovascular diseases, and diabetes mellitus collectively accounted for 87.56% of all hospitalizations. The best-fitting principal component model of infrastructure data consisted of 3 components that corresponded to the level of adequacy of care comprehensiveness, continuity, and coordination. In the fully-adjusted models, the strongest predictors of HPCSC per 10,000 were continuity (β= 12.44) for heart failure, comprehensiveness (β= -3.09) for cerebrovascular diseases, continuity (β= 1.45) for diabetes, continuity (β= .92) for skin and subcutaneous tissue infections, comprehensives e (β=.99) for female pelvic inflammatory diseases, and continuity (β=.74) for prenatal and postpartum conditions.

Conclusions: NCDs heavily influence incidences of avoidable hospitalizations in Minas Gerais, Brazil. Yet, our findings suggest that the community-based care models of the FHS may have the potential to mitigate the role of social vulnerability in influencing health outcomes. This project offers a model for quantifying the quality of PC infrastructure and more research is needed to validate its use in LMIC, as well as to further understand the strength and directionality of the relationship between health center, health system, and socioeconomic predictors of HPCSC.

Background: This thesis describes the perceived delays faced by patients and providers in Amazonas, Brazil while seeking or providing snakebite envenoming care in the region. Additionally, we compare the delays described by indigenous and non-indigenous patients while seeking care. Methods: This study analyzed previously recorded data from snakebite envenoming (SBE) patients and indigenous health care providers with intent to discuss their experiences. In-depth interviews and focus group transcripts were analyzed using the Three Delay Model. This model groups data into three separate delays; deciding to seek care, arriving to care, and receiving adequate care. Results: From this analysis we found that patients described many different themes for their decisions to seek care including choosing alternative methods, cultural restrictions, refusing traditional medicines and many more. The reasons behind the decisions to seek care or not were different between indigenous and non-indigenous patients. While travelling to care both study groups described the need to use multiple means of travel to arrive to distribution hospitals and the unavailability of transportation and emergency services which caused delays. Finally, once they arrived to health care facility delays were again presented by the need for multiple facilities to receive adequate care. Conclusion: The findings of this study support public health researchers push for the decentralization of antivenom. Both survivors of SBE and health care providers in the region have expressed the need for treatment to be available in their region. Due to the population of the region, any interventions whether education or political will need to consider the culture of the indigenous people to ensure positive uptake.

Background: Alcohol consumption is a major risk factor for several adverse health effects globally and is associated with a high disease burden of malnutrition in Tanzania. This study aims to: 1) assess the practicality and effectiveness of mid-upper arm circumference (MUAC) as a feasible bedside measure to detect malnutrition among adult and pediatric patients, 2) evaluate the association between alcohol use and nutritional status among adult injury patients and 3) qualitatively identify social determinants of malnutrition in Tanzania. Methods: This mixed-method study was conducted in Kilimanjaro Christian Medical Centre. Receiver Operating Characteristic (ROC) curves and logistic regression were used for quantitative data on alcohol use, body mass index (BMI), MUAC. Thematic approach was used for qualitative data on perspectives on alcohol use and its interactions with malnutrition. Results: MUAC cut-offs were determined at < 25.5 cm (BMI < 16 kg/m2) for severely underweight, < 28 cm (BMI < 18.5 kg/m2) for underweight, ≥ 30.5 cm (BMI ≥ 25 kg/m2) for overweight, ≥ 33 cm (BMI ≥ 30 kg/m2) for obese. The association between alcohol use and malnutrition (MUAC < 25.5 cm) was statistically significant. Qualitative results helped explain the association between alcohol use and malnutrition. Conclusions: MUAC is an effective tool to detect adult malnutrition to inform clinical practice in Tanzania. Polarizing attitudes towards drinking revealed by qualitative data suggested the need for alcohol awareness campaigns. Food assistance programs are needed to reduce the risk of malnutrition among vulnerable populations.

Introduction: There is a high burden of injuries in low- and middle-income countries (LMICs). Data management and storage systems are suboptimal, making it hard to share data. Also, small data sizes hinder the application of modern data science methods to draw insight. Methods: We performed a document analysis of injury data dictionaries from 4 institutions located in 4 different LMICs for discrepancies among data elements. We also compared each of the dictionaries with the World Health Organization (WHO) data set for injury (DSI) and then explored harmonization strategies of injury data, given the discrepancies. Results: Of the 949 data elements across the dictionaries, there were 6 (0.72%) shared data elements when considered by presence and 4 (0.45%) when considered by equality across the 4 dictionaries. The number of shared data elements varied when the dictionaries were compared in pairs and triads. We identified four methods of ensuring harmonization of injury data; 1) using the WHO DSI common data elements, 2) using the National Institute of Neurologic Disorders and Stroke (NINDS) common data elements, 3) using functions written in software like R and python, and 4) adopting a prospective and somewhat promising development of a framework that includes common data elements specific to injuries. We successfully harmonized three injury data sets from the Kilimanjaro Christian Medical Centre (KCMC). Conclusion: There are huge discrepancies across injury data dictionaries in LMICs. Harmonization of injury data from LMICs is however achievable and can result in a larger dataset. More research is needed in this area for the development of tools that facilitate injury data harmonization.

Background: Emergency care sensitive conditions are defined as those for which rapid diagnosis and early intervention improve patient outcomes. This thesis aimed to develop a list of important pediatric conditions in low- and middle-income countries (LMIC) to be used for further studies on pediatric epidemiology and resource utilization. Methods: A survey of 79 conditions was sent to LMIC physicians, who rated each condition on three categories (time sensitivity, preventability, and commonality) on a scale of 1-9. Responses were matched to Brazil pediatric hospitalization, ambulatory, and mortality data from 2015-2020. Results: 17 physicians completed the first Round of the survey, and 3 of these (17.65%) completed the second Round. Overall, 67 of the 79 (84.21%) were rated as highly time-sensitive and 26 (32.91%) highly preventable. Survey conditions with the highest ratings overall or country overlap (n=11), that were country-specific but highly rated in all three categories (n=8), or that comprised ~1%+ of hospitalizations (n=9), ~0.5%+ of ambulatory visits (n=6), and ~0.5%+ of mortality cases (n=8) were combined with the most common acute non-elective causes of hospitalizations (n=7) and mortality (n=9) into a list of 29 consolidated conditions overall (excluding overlap). These 29 accounted for 37.83% of hospitalizations, 8.97% of ambulatory visits, and 29.17% of mortality cases. 31 of the 79 survey conditions were age-specific and 32 context-specific. Conclusions: These 29 should be targeted in future health system utilization and burden studies. The modified Delphi approach is important in reaching provider consensus.

Background: Globally, 5 billion people lack timely access to safe and affordable surgical care, with over a fifth of them living in India. Solving India’s surgical access issues can have high returns on investment. While healthcare access and unaffordability problems are well-known in India particularly among its rural people, research on surgical care is scant. This study attempts to fill the research gap through high-resolution nationwide estimates that have direct implications for India’s national surgical plan. Methods: Secondary data analysis with a diverse geospatial and statistical toolbox was used to create the national, state, and district-level estimates in four surgical care access dimensions. The four access dimensions were: timeliness (proportion of population within 2 hours of a surgical are facility), capacity (met surgical need for operative volumes), safety (proportion of post-operative surgical site infections), and affordability (proportion of surgery-seeking households facing catastrophic expenses). A novel composite index was introduced for assessing surgical access integrating the above dimensions. Distributional and spatial inequalities in access across Indian districts and states were measured to depict regions needing policy intervention. Correlations with Sustainable Development Goals (SDG) scores were computed. Validation and sensitivity analyses were conducted to check the robustness of the findings. Results: Timely access to surgical care was achieved by > 99% of the rural population, but only 6.81% of surgical need was met. SSI proportion was 0.19% and 60.99% of surgery-seeking households faced catastrophic health expenditure. Heterogeneities in these dimensions were observed at state and district-levels. Significant rural-urban differences were observed in surgical care access dimensions and other considered surgical care variables. The Zadey-Vissoci Access to Surgical Care Index (ZV-ASCI) depicted limited access across several states and districts. Within-state distributional inequality in ZV-ASCI was about three times that of between-states. We found limited support for spatial autocorrelations and identified the low access district clusters. For aspirational districts, whose development is high on the national agenda, ZV-ASCI was not correlated with SDG composite score. Conclusions: Our methodological workflow has high translational value for global surgery research in low-and-middle-income countries. For India, these are the first such nationwide findings that can direct the development of a National Surgical, Obstetric, and Anesthesia Plan (NSOAP). The proposed index can encourage buy-in from policymakers and raise surgical care on the global and national agenda.

Background: The impact of malnutrition on the outcomes of hospitalized adult patients in resource-limited settings is not fully described. We aimed to report our observation of the comparison made between mid-upper arm circumference (MUAC) and injury outcomes in adults admitted to the Emergency Department of a hospital in Moshi, Tanzania. Methods: This study was a secondary analysis based on an on-going Trauma registry collected by researchers from the Duke Global Health Institute at Duke University in collaboration with the Kilimanjaro Christian Medical Centre (KCMC) from April 2018 until April 2020. Results: Females were significantly more likely to have higher MUAC scores than males (B=1.90; SE=0.49; p-value=0.000). Compared to single patients, those married were more likely to have higher MUAC (B=1.28; SE=0.48; p-value=0.007). Compared to advanced education, patients identified with basic education were less likely to have higher MUAC (B=-1.46; SE=0.42; p-value=0.000). Although glasgow outcome was not significantly associated with MUAC score in our univariate analysis (B=-1.20; SE=1.46; p-value=0.41), in our adjusted model, lower good recovery were less likely to have higher muac scores then other glasgow outcomes (B=-2.02; SE=0.73; p-value=0.000). Conclusions: Undernutrition in hospital patients is often unrecognized and there is a need for simple means of screening to facilitate targeted nutritional intervention. Further research is needed to understand the pathophysiology of malnutrition during acute illness and validate MUAC cut-off points for hospitalized adults.

Background: Globally, many low-income settings lack diagnostic tools to handle prognosis of TBI patients. In such settings, development of generalizable predictive models which indicate likelihood of patient outcomes may help improve decision-making for physicians and health care providers.

Methods: An analysis of a Rwanda TBI registry (n=682) was conducted to determine key predictors of TBI mortality. A previously developed prognostic model of a Tanzania TBI registry (n=3209) was subsequently implemented in the Rwanda TBI registry for external validation. 8 different machine learning models were implemented in the Rwanda dataset. Subsequently, 6 Tanzania models and a combined model aggregating the Tanzania model predictions were used to compare and predict Rwanda patient outcomes.

Results: The predictive models developed in Rwanda had satisfactory predictive ability, with the best performing model, Ridge Regression having an AUC of 90% (CI: 89.3%-90.7%). The models developed in Tanzania and used to predict outcomes within the Rwanda dataset showed similar predictive ability, the AUC of the best performing Random Forest model, 91.3% (CI: 88.0%-94.6%) and the combined Tanzania machine learning model, AUC 91.9% (CI: 88.7%-95.1%).

Conclusions: The results from the Tanzania and combined models indicate satisfactory predictive ability and generalizability. The ability of the models to hold similar predictive power in an external dataset, with the use of indicators collectable at triage suggests potential applicability in other low-resource settings.

Introduction: Acute coronary syndromes (ACS) result in significant morbidity and mortality in low-and-middle-income countries (LMICs). Fifty percent of deaths in this region are from a cardiac etiology. Not much is known about the epidemiology of ACS in Brazil. Our aim was to describe the correlation between social determinants of health and access-to-care markers as related to ACS mortality and its geographic distribution in the country. Methods: Using the Brazilian National Health Database (DATASUS) and other nationally aggregated data sources, socioeconomic (SE) parameters, cardiovascular risk (CV) factors and an accessibility index for high complexity cardiac care centers (with hemodynamic monitoring and cardiac interventions) were obtained. To account for spatial dependency, geographic weighted regression (GWR) analysis was performed for all the predictor variables with respect to the outcome of deaths. Results: There were 776,449 ACS-related deaths from 2012 to 2018. The highest ACS mortality rate was in the South region of Brazil (104.7 per 100,000 population). The GWR analysis showed regional variability of socioeconomic factors as correlated with ACS mortality. A low accessibility-index in the North and Northeast regions of Brazil was strongly associated with ACS deaths. Conclusions: Spatial analysis allows for estimation of the local heterogeneity in the relationship between SE components, CV risk factors and access-to-care markers as related to ACS mortality. Such analyses allow for improved understanding of the burden of ACS in Brazil.