Published on December 15, 2008
The Effects Race and Income on HIV/AIDS Infection Is HIV/AIDS More Prevalent in African Americans? Final Project STAT 5990 / HINF 6030 course Masters in Health Informatics By ABEL MEDHANIE GEBREYESUS - B00498815 SUNIL NAIR - B00492855 Dalhousie Health Informatics Halifax, Nova Scotia December 12, 2007 1
Abstract Objective: Are the African-American people more susceptible to HIV/AIDS infection and be hospitalized? Does race and level of income have an impact on increasing a chance of HIV/AIDS infection and as a result of having poor mortality? These are the questions we would try to answer by this statistical analysis. Methods: This is a population based study, where data were collected through National Hospital Discharge Survey (NHDS). The NHDS dataset has discharge records from all participating U.S hospitals. This study includes 125,165 discharge data of HIV/AIDS infected patients hospitalized in the year 2005. A Multivariate analysis using Poisson distribution was conducted in order to examine the relationship of race and income level on high incidence of HIV/AIDS infections in African-American population. Results: Although African Americans (blacks) represent only the 13% of the United States population, of the total patients hospitalized due to HIV/AIDS, 55.87% of those were blacks (p-Value=0.02). 92.54% of those hospitalized were either not insured or were covered by government insurance coverage, which could belong to a low income group (p-Value<.0001). Conclusion: Race and Income has a significant influence on susceptibility to HIV/AIDS infections; Afro-Americans (Blacks) are 1.33 times more likely to be infected than whites. A significant finding is that the income level didn't change race's effect on HIV infections. Race has a significant effect on HIV infections or is an important predictor of incidence of HIV infections independent of the income. In other words, irrespective of the income level being black and poor increases the changes of being infected with HIV/AIDS. 1. Introduction In the Global summary of AIDS epidemic published in December 2007, the WHO estimates that there are 33.2 million people living with HIV in this world today. There are more than 6800 new infections per day, and almost 96% of those are from the low and middle income countries.  A plethora of research work has been carried out in underdeveloped and developing countries that suggest an association between poverty and a higher incidence rate of HIV/AIDS.  In the industrialized nations most of the research has been focused towards associating the infection with race, sexual behavior, and drug abuse with increase prevalence of HIV infection. 2
According to official estimates, around 1.2 million American are living with HIV. As a result, more and more people are hospitalized. Among those around 1.2 million Americans, who are living with HIV/AIDS are, the most affected are those of African Americans, in terms of race. According to a Center for Disease Control (CDC) fact sheet, “in the United States, the HIV/AIDS epidemic is a health crisis for African Americans. At all stages of HIV/AIDS—from infection with HIV to death with AIDS—blacks are disproportionately affected compared with members of other races and ethnicities”. In the United States in 2002, HIV/AIDS was among the top 3 causes of death for African American men aged 25–54 years and among the top 4 causes of death for African American women aged 25–54 years. It was the number 1 cause of death for African American women aged 25–34 years  Therefore we will determine the influence of race and income on increased incidence of HIV/AIDS in Afro-Americans. Based on this, the null hypothesis is: There is no effect of race and income on HIV infection. This means that we are going to test whether or not there's a difference of HIV infection count or rate among different races, at the same time, we are going to check whether or not there is an association between the HIV infection count/rate and income, as we control the patients' age, sex and Marital Status. 2. Literature Review BASTARDO et al in their study: “Relationship between Quality of Life, Social Support and Disease-related Factors in HIV-infected Persons in Venezuela”, examined the relationships among health-related quality of life (HRQL), social support, sociodemographic factors and disease-related factors in persons infected with the HIV living in Venezuela. This exploratory study was designed to assess the HRQL and levels of social support of HIV-infected persons living in Venezuela; and examine the relationships of quality of life, social support and demographic and disease-related factors in such persons. The researchers conclude that there is an important association between social support and HRQL in HIV-infected persons in Venezuela. Another Canadian study by Canadian HIV/AIDS Legal Network says: “The links between poverty and HIV/AIDS go in two directions. In one direction, poverty contributes to people’s vulnerability to HIV, exacerbates the impact HIV/AIDS has upon them, and leads to greater illness and early death. Going in the other direction, the experience of HIV/AIDS by individuals, households and communities leads to an intensification of poverty. As a result, HIV/AIDS frequently impoverishes people in such a way as to intensify the epidemic itself” 3
In another detailed study, a U.S. Department of Health and Human Services, Office of the Surgeon General’s report indicates, many African Americans live in segregated neighborhoods, and poor African Americans tend to live among other African Americans who are poor. “Poor neighborhoods have few resources, a disadvantage reflected in high unemployment rates, homelessness, crime, and substance abuse,” it underlines. The report says: “African Americans are more likely than whites to live in severe poverty, with incomes at or below 50 percent of the poverty threshold; the African American rate of severe poverty is more than three times the white rate. Children and youth are especially affected; while the national poverty rate for U.S. children is nearly 20 percent, almost 37 percent of African Americans 18 and younger live in poor families.” This report also stated about the relationship between such poverty and poor health conditions in African Americans. The study by Tony L. Whitehead from University of Maryland is also noted, “in the United States, ethnic minority groups, particularly African Americans (and Hispanics) suffer disproportionately in morbidity and mortality from the human immunodeficiency virus and the acquired immunodeficiency syndrome (HIV/AIDS).” In his “Urban Low-Income African American Men, HIV/AIDS, and Gender Identity” indicates African Americans are more prone to infection and the living style of blacks is hustle oriented with low income survival. The HIV/AIDS epidemic is concentrated in poor communities, where African Americans are disproportionately represented. From 2000 to 2004, new AIDS cases in the United States increased by less than one percent; however, new cases in the South, where poor and black communities are found largely, increased by nine percent. Jennifer Kates and Alicia Carbaugh of the Kaiser Family Foundation (February 2006) showed that the picture is gloomy for African Americans in the HIV/AIDS situation. 3. Method Data Source NHDS data are collected from a sample of inpatient records acquired from a national sample of participating hospitals in the U.S. Our data set consisted of discharge records of HIV/AIDS patients as per the ICD-9 disease classification code. There were 125,165 (n) HIV/AIDS related discharges in the year 2005. Design Since this is a survey data, we needed to weight each observation to get the outcome - HIV counts. For this study we have focused on Poisson regression model (PRM) in order to examine the count of HIV infection in our sample. The Poisson regression models are basic models for count data analysis. The GENMOD procedure of SAS was employed to do the PRM. The 4
/DIST=POISSON option tells SAS to use the Poisson distribution. It is essential to do a good diagnostic check of whether or not the Poisson distribution is a good fit for this count outcome. We performed the Goodness–of-fit test by investigating the Value/DF value for the model deviance (the measure of discrepancy between observed and fitted values). For large samples like the NHDS dataset, a model with a good fit to the data will have a Value/DF value close to 1, and as we found out this value in the first model is bigger than 1, suggesting a somewhat poor fit. We found that the existing model is over-dispersed; therefore we refit the model with a Pearson scaling factor to adjust the over-dispersion. We used the over-dispersed Poisson models to analyze the effects of race and income on HIV infection, controlling the other effects such as patients’ age, sex, region and marriage. The Value/DF value is 1 at the row of “Scaled Pearson X2” in the second model as seen in the table below. Criteria for Assessing Goodness of Fit Criterion DF Value Value/DF Deviance 861 142383.6668 165.3701 Scaled Deviance 861 473.7625 0.5502 Pearson Chi-Square 861 258763.2600 300.5380 Scaled Pearson X2 861 861.0000 1.0000 Log Likelihood 1735.6473 The variables used: # Variable Type Label 4 Age Number Age in years, months or days 5 Sex Number patient sex 6 Race Number Patient race 14 Owner Number Ownership of hospital 12 Region Number Geographic region of hospital 15 Weight Number Analysis weight 7 Marstat Number Marital status As age is normally distributed it was left as a continuous variable. As the NHDS datasets does not carry the income data, we have used the Ownership of the hospital as an indicative of the level of income of the patients. Ownership of hospital indicates whether the hospital is private, government or is non-profit 5
charitable trust management. Typically, poor patients frequent the non-profit charitable hospitals who cannot afford to purchase private medical insurance. We have conducted a full Poisson model including all chosen variables. We also have tried to fit the model without race to check how the coefficient of income changes and another model without income to check how the coefficient of race changes so that we could analyze the influence between race and income on HIV infection occurrences. Sex, age and marital status do not appear to be influential. There is no difference between married patients and others that we called 'single' even though the p-value of marriage is <0.0001 which comes from ‘undisclosed’ group and ‘singles’ group's comparison. We have excluded marital status because of the large number of ‘undisclosed’ as this could be due to the fact that patients were reluctant to disclose their sexual preferences. 4. Results and discussion While it is widely recognized that poverty, or low income, is associated with poor health, even in rich societies, the nature of the relationship between income and health status is not clearly understood. Especially, in big health issues and crises, like HIV/AIDS, it is important to look at the relationship of income and the possibility of HIV infections. But, who is poor or rich is very relative term and defined differently by different social or economic orientations. However, for this study, we are going to classify using simple definition. Although it is still a matter of controversy, we look at the ownership variable. As all of the subjects or the population is hospitalized, we counted those who go to the proprietary hospitals, coded as 1, labeled as high income and those who hospitalized in government owned or in non- profit organizations, coded as 2 and 3, respectively, labeled as low income AIDS patients. A number of studies focus specifically on measures of very low income, or poverty. They find that persistent poverty appears to be most damaging for health. Those people who are persistently poor have worse health outcomes than those who experience poverty only occasionally or not at all (BENZEVAL et al). In our study, parallel to income, race is also taken into consideration. In this case, we will examine which race is more infected and what is the income level to this race and based on such results, we will conclude how the relationship is of income affected races to be more infected. 6
Descriptive Statistics of Variables Used in the Model Variable Frequency Percent Sex Male 77839 62.19 Female 47326 37.81 Region Northwest 42169 33.69 Midwest 12948 10.34 South 54867 43.84 West 15181 12.13 Race Black 69924 55.87 White 30780 24.59 Others 6210 4.96 Unknown 18251 14.58 Income/Ownership of Hospital High 9340 7.46 Low 115825 92.54 Marital Status Married 11494 9.18 Single 65815 52.58 Undisclosed 47856 38.23 The data has 62.19 % male and 37.81% female population. These include from 15 years old to 75 years old, and the peak age of hospitalization due to HIV/AIDs is 45. As the age increases, the hospitalization also increases until it reaches 45 years old. Then, it declines after 45. 7
Peak age of hospitalization Although African Americans (blacks) represent only the 13% of the United States population, this specific group is affected more than any races with 55.87% of blacks are hospitalized. This shows there is clear difference among different races; and it shows how the blacks are affected by HIV/AIDS more than any other races. For instance, whites are around 80 of the United States population. However, only 24.59% of them are hospitalized. In another study also, the National Institutes of Health stated that although African Americans represent only the 13% of the United States population, this specific group is affected more than any races with approximately 46% of new HIV infections and 50% of reported AIDS cases. Another interesting factor of the result, most of these African Americans also depend on government funded health insurance coverage, like on Medicare and Medicaid. This factor also can show their status to some extent. As most of them are depending on the government funded health insurance, they have less income in their daily life. Although it needs more study these interrelated factors has some indication on weather there is a direct relationship between income and infection. 8
Among those infected and hospitalized African American, the majority are single, in terms of their marital status. They are almost one third of the total hospitalized African Americans with 30.99%. This trend is also the same in whites, because among those 24.59 hospitalized whites, 10.89% are singles. Taking into account this information, both in blacks and whites, singles are more affected. This could be most of them are homosexuals. Or the data may put homosexuals, lesbians or other form of partners as singles. Whites Blacks Others Married 3.03 5.02 1.14 Single 10.89 30.99 3.23 Widowed 0.04 1.30 0.46 Divorced 1.77 0.29 0.40 Separated 0.70 2.46 0.04 Not stated 8.15 15.80 14.28 Total % 24.59 55.87 19.54 Adjust for Over-dispersion The exponential of the estimate (coefficient) represents the difference or ratio of the expected HIV infections between the two levels (compared category vs. reference). For example, race is declared as a categorical predictor in the model, we need to compare blacks, with others and whites. Compared category vs. Referenced category Effects/Predictors Estimate Exponential p-Value Race Black 0.29 1.33 0.02 Others -0.38 0.68 0.12 Whites (reference) . . . 9
From above table, we see that the predicted mean count of HIV infections for blacks is about 1.33 times that for whites (controlling for income and the other variables in the model), which is significant (p-value =0.02) at alpha=0.05 level, while the predicted mean HIV infections for others is about 0.68 times than that for whites, but the difference is not significant (p=0.12). Since the p-Value for both race and income are <0.05, we can reject the hypothesis that race and income does not have an influence on increased incidence HIV/AIDS. Race 95% Confidence Interval (0.04, 0.528) at alpha=0.05 High Income 95% Confidence Interval (-1.35, -058) at alpha=0.05 The Type III likelihood ratio tests for the predictors in the model (similar to F-tests in an analysis of variance setting) allowing us to get an idea of whether the effects (categorical or continuous) are significant or not in overall. The p-value for race is 0.0027 in the Type 3 analysis, which tells us that race is a significant effect or predictor on the mean count of HIV infections in overall. Check whether race will change without income in model Effects/Predictors Estimate Exponential p-Value Race Black 0.33 1.39 0.01 Others -0.30 0.68 0.23 Whites (reference) . . . Finally we wanted to check whether the two variables race and income are related and whether controlling one will have any impact on the other. When we did not consider Income, the estimate for blacks was 0.33 and its p-value was 0.01. The ratio was 1.39, slightly higher than that in model 2 (1.33, controlling income). With this we can conclude that income didn’t change race’s effect on HIV infections significantly. In other words, race and income are significant predictors on the HIV infections independently. And there is no influence of level of income on the race having an effect as a pre-disposing factor to HIV infections. 10
5. Conclusion The primary focus of our investigation was to determine the association of income level and race with higher incidence of HIV/AIDS infection. We would examine the effect of different determinants, for example, is it because of particular communities poverty, their socio-economic status, their ethnicity or their gender that they are more predisposed to getting HIV/AIDS. The research findings suggests that there is an important association between race and income level (poverty) that place people at risk of HIV infection and subsequent disease progression. Among the group of race, Blacks have a much higher risk of infection than the other groups combined. We also established that most HIV/AIDS patients are from a low income group. The mortality rate of low-income HIV-positive people is higher than higher incomes and more education.  The main reason for this finding could be that low-income HIV-positive patients are more likely to be covered by Medicaid or Medicare or are uninsured. HIV/AIDS analytical study is a highly complex methodology because of the continuously changing nature of the disease and the varied characteristics of the different at-risk groups, and also because the epidemic is high in the most marginalized of communities. Accessing these communities for research purposes can itself be a formidable challenge. There are several other challenges faced by the HIV/AIDS researchers in developed nations, like in the US. Generally healthcare and hospital datasets does not contain direct information on income. Most of income related data is gathered by direct interviews. Perhaps the striking observation from our study was that the level of income has no relation to race as far as the HIV infections are concerned. What this means is that an African-American who belongs to a higher income group could still be more susceptible to HIV/AIDS infection. 11
Bibliography  2007 AIDS epidemic update. UNAIDS. November 2007. http://www.unaids.org/en/HIV_data/2007EpiUpdate/default.asp  Anderson RN, Smith BL. Deaths: leading causes for 2002. National Vital Statistics Reports 2005;53(17): 67–70.  CDC. “Racial/ethnic disparities in diagnoses of HIV/AIDS---33 states, 2001--2004.” MMWR 2006;55:121--5.  CDC. “Racial/ethnic disparities in diagnoses of HIV/AIDS---33 states, 2001--2004.” MMWR 2006;55:121--5.  Fleming, Patricia (07/01/2006). quot;The Epidemiology of HIV/AIDS in Women in the Southern United Statesquot;. Sexually transmitted diseases (0148-5717), 33 (7), p. S32.  Bastardo, Y.M., and Kimberlin, C.L., Relationship between quality of life, social support and disease-related factors in HIV-infected persons in Venezuela, AIDS Care, Volume 12, Number 5, 1 October 2000 , pp. 673-684(12)  http://www.aidslaw.ca/publications/interfaces/downloadFile.php?ref=107  http://mentalhealth.samhsa.gov/cre/ch3_current_status.asp  ibid  ibid  Whitehead, TONY L., Urban Low-Income African American Men, HIV/AIDS, and Gender Identity, Department Of Anthropology, University Of Maryland  ibid [13 Jennifer Kates and Alicia Carbaugh of the Kaiser Family Foundation; African Americans and HIV/AIDS, February 2006  BENZEVAL, MICHAELA; TAYLOR JAYNE and KEN JUDGE, Evidence on the Relationship between Low Income and Poor Health: Is the Government Doing Enough? Fiscal Studies (2000) vol. 21, no. 3, pp. 375–399  Journal of Health Care for the Poor and Underserved, United Press International reports (United Press International, 11/1 12
Calcification Inhibitors in CKD and Dialysis Patients
The Effect Race and Income on HIV AIDS infection in African-Americans - Sunil Nair Health Informatics Dalhousie ... African-Americans - Sunil Nair Health ...
... to HIV SOUTHERN ILLINOIS UNIVERSITY AT infection ... health care for African Americans ... HEALTH SCIENCES. HIV risk behaviors and AIDS ...
... Sunil Nair Health Informatics Dalhousie University. ... The Effect Race and Income on HIV AIDS infection in ... Project Health Informatics Sunil Nair.
Critical care triage. PubMed Central. Christian, ... Sunil; Salins, Naveen ... The effect of nursesâ€™ preparedness and nurse practitioner status on ...
mimetypeMETA-INF/container.xml1.0 urn:oasis:names:tc:opendocument:xmlns:container content.opf application/oebps-package+xml 2011 AHA_ebook_v1_split_000.htm ...
Contract Talk: Reading PS Form 1840 Reverse ... Assistant Secretary for Health, ... 4:15 pm Using the AEC for Ancillary Income Reed B. Hogan, ...
weekly menu - Innovation Place ... Lecture Gastroenterology and Public Health in the Era of ... race t-shirt вЂў Post-race breakfast ...