HIV infection in hospitalized patients and Medicaid enrollees: the accuracy of medical record coding

Hospital Use by Persons With HIV in the 21st Century: A 5-State Study

OBJECTIVE

The purpose of this study is to determine whether reductions in hospital utilization observed immediately after the availability of highly active antiretroviral therapy (between 1995 and 2000) have persisted into the 21st century.

DATA SOURCES

Data on all human immunodeficiency virus (HIV)-related hospital admissions in 5 states (California, Florida, New Jersey, New York, and South Carolina) in 2000, 2005, 2010, and 2013 were obtained from the State Inpatient Database, which is administered by the Agency for Healthcare Research and Quality. In addition, data on the number of persons living with HIV were obtained from the Centers for Disease Control and Prevention and from the California Department of Public Health.

STUDY DESIGN

This study compares the average number of hospitalizations per person living with HIV in each of the 5 states as well as the average cost for hospital care per person with HIV in 2000, 2005, 2010, and 2013.

RESULTS

The total number of hospitalizations by persons with HIV in the 5 study states fell by one third between 2000 and 2013 even though the number of persons living with HIV increased by >50%.

CONCLUSIONS

Persons with HIV disease were 64% less likely to be hospitalized in 2013 than they were in 2000. In addition, the probability of a person with HIV being hospitalized fell 44% between 2000 and 2010 and 29% between 2010 and 2013.

In Four ACA Expansion States, The Percentage Of Uninsured Hospitalizations For People With HIV Declined, 2012-14

This study examines the influence of the Affordable Care Act's optional state Medicaid expansion on insurance coverage and health outcomes for hospitalized patients with HIV. I used data from the State Inpatient Databases of the Healthcare Cost and Utilization Project for all hospitalizations of patients with HIV from 2012 through the first six months of 2014 in four states that expanded their Medicaid programs and two states that did not. I found that the percentage of hospitalizations of uninsured people with HIV in the four expansion states fell from 13.7 percent to 5.5 percent in the study period, while the percentage in the two nonexpanding states increased from 14.5 percent to 15.7 percent. I also found that hospitalized patients with HIV who did not have insurance were 40 percent more likely to die during their hospital stays than comparable patients with insurance.

HIV Patients in the HCUP Database: A Study of Hospital Utilization and Costs

This study examines the utilization of hospital care by HIV patients in all hospitals in eight states (California, Colorado, Florida, Kansas, New Jersey, New York, Pennsylvania, and South Carolina), and examines the cost of hospital care for HIV patients in six of these states (California, Colorado, Kansas, New Jersey, New York, and South Carolina). The eight states in the sample account for more than 52% of all persons living with AIDS in the United States; the six states account for 39%. The unit of observation in both studies is a hospital admission by a patient with HIV. Hospital data were obtained from the Healthcare Cost and Utilization Project (HCUP), State Inpatient Database (SID), which is maintained by the Agency for Healthcare Research and Quality (AHRQ). The HCUP contains hospital discharge data and is a federal/state/industry partnership to build a multistate health care data system. Using multivariate analytic techniques and data from 2000, results indicate that cost and length of a hospital stay vary significantly across states after accounting for a patient's gender, insurance type, race, age, and number of diagnoses, as well as the teaching status and ownership category of the hospital.

Hospitalization rates of people living with HIV in the United States, 2009

OBJECTIVES

We determined hospitalization rates and disparities among people with HIV, which may have been underestimated in previous studies, as only those in medical care were included.

METHODS

We estimated the hospitalization rate of people with diagnosed HIV infection in the U.S. in 2009 using two nationally representative datasets. We took the number of hospitalizations from the Nationwide Inpatient Sample and searched each discharge for International Classification of Diseases, Ninth Revision codes for HIV infection and opportunistic infections (OIs). We divided the number of hospitalizations by the number of prevalent diagnosed HIV cases estimated by CDC to produce hospitalization rates, and then compared those rates using Z-tests.

RESULTS

The estimated nationwide hospitalization rate was 26.6 per 100 population. Women had a 51% higher rate than men (35.5 vs. 23.5 per 100 population, p=0.002). Black people (31.2 per 100 population, p=0.01) had a 42% higher rate, and Hispanic people (18.2 per 100 population, p=0.23) had an 18% lower rate than white people (22.1 per 100 population) of hospitalization for any illness. Of hospitalizations with an OI, females with HIV had a 50% higher rate than males with HIV (5.0 vs. 3.4 per 100 population, p=0.003). Black people with HIV (4.7 per 100 population, p<0.001) had a 72% higher rate and Hispanic people with HIV (2.9 per 100 population, p=0.78) had a similar rate of hospitalization with an OI compared with white people with HIV (2.7 per 100 population).

CONCLUSIONS

Hospitalization rates among people living with HIV in the U.S. are higher than have been previously estimated. Substantial gender and racial/ethnic disparities in hospitalization rates exist, suggesting that the benefits of antiretroviral therapy have not been realized across all groups equally.

Validation of case-finding algorithms derived from administrative data for identifying adults living with human immunodeficiency virus infection

Objective

We sought to validate a case-finding algorithm for human immunodeficiency virus (HIV) infection using administrative health databases in Ontario, Canada.

Methods

We constructed 48 case-finding algorithms using combinations of physician billing claims, hospital and emergency room separations and prescription drug claims. We determined the test characteristics of each algorithm over various time frames for identifying HIV infection, using data abstracted from the charts of 2,040 randomly selected patients receiving care at two medical practices in Toronto, Ontario as the reference standard.

Results

With the exception of algorithms using only a single physician claim, the specificity of all algorithms exceeded 99%. An algorithm consisting of three physician claims over a three year period had a sensitivity and specificity of 96.2% (95% CI 95.2%–97.9%) and 99.6% (95% CI 99.1%–99.8%), respectively. Application of the algorithm to the province of Ontario identified 12,179 HIV-infected patients in care for the period spanning April 1, 2007 to March 31, 2009.

Conclusions

Case-finding algorithms generated from administrative data can accurately identify adults living with HIV. A relatively simple “3 claims in 3 years” definition can be used for assembling a population-based cohort and facilitating future research examining trends in health service use and outcomes among HIV-infected adults in Ontario.

The cost and incidence of prescribing errors among privately insured HIV patients

BACKGROUND

With the rapid growth in the volume of HIV-related studies that address drug interactions, appropriate medication regimens, and when and how to alter drug regimens, it is challenging for physicians to stay informed. Physicians require knowledge about all drugs taken by HIV patients in order to assess accurately the benefits and risks of various drug combinations.

OBJECTIVE

To examine the cost and frequency of antiretroviral prescribing errors among a sample of privately insured patients with HIV disease.

METHODS

Data were obtained from the MarketScan Commercial Claims and Encounter Database created by the Medstat Group Inc. The MarketScan database contains claims data for inpatient care, outpatient care, physician services and prescription drugs in benefit plans sponsored by >50 large employers in the US. This study compared data from the 1999-2000 MarketScan database with those from the 2005 MarketScan database. The 2005 MarketScan database includes 12,226 HIV enrollees who received antiretroviral drugs. This study compared the claims experience of HIV patients who filled a prescription for a drug combination that is not recommended by the US Department of Health and Human Services Panel on Antiretroviral Guidelines for Adults and Adolescents with the claims experience of patients who did not receive such a prescription.

RESULTS

In the 1999-2000 database the most common inappropriate drug combination involved the co-administration of a protease inhibitor (PI) and the lipid-lowering drug simvastatin, and 1% of patients experienced this type of error. In the 2005 database, only 0.4% of patients (46 of 12,226) experienced an inappropriate combination of simvastatin and a PI while 5.3% of patients (644 of 12,226) received atazanavir and tenofovir without ritonavir (referred to herein as 'boosting errors'). Patients who experienced a boosting error incurred higher annual costs than patients who took ritonavir along with tenofovir and atazanavir ($US 20,927 vs $US 16,704). Because atazanavir was approved by the US FDA in June 2003, medication errors involving atazanavir were not relevant in 1999 and 2000. Overall, it was found that HIV patients were three times as likely to experience an inappropriate drug combination in 2005 than they were in either 1999 or 2000 (5.9% vs 1.9%), and that this increase is attributable to boosting errors. In addition, the prevalence rate of HIV in the 2005 MarketScan database was almost triple that in the 1999 MarketScan database (0.14% vs 0.05%).

CONCLUSION

This study indicates that those who provide care to HIV patients must be vigilant in their efforts to provide patients with a drug therapy regimen that minimizes the chance of an adverse reaction and maximizes the potential to control viral replication.

The Changing Pattern of Hospital Care for Persons Living With HIV

OBJECTIVE

To compare inpatient utilization and costs by persons living with HIV in 2000 with inpatient utilization and costs in 2004.

DATA SOURCES

Data on 91,343 hospital discharge abstracts representing all HIV-related admissions in 6 states (California, Florida, New Jersey, New York, South Carolina, and Washington state) in 2000 and data from 72,829 hospital discharge abstracts representing all HIV-related admissions in the same states in 2004 are used. These data were obtained from the Agency for Healthcare Research and Quality's Healthcare Cost and Utilization Project, and they were combined with data on the number of persons living with HIV that were obtained from the Centers for Disease Control and Prevention and 2 state departments of health.

STUDY DESIGN

This study compares the hospital care received by persons living with HIV in 6 states in calendar year 2000 with the hospital care received by persons living with HIV in calendar year 2004 in the same 6 states. This study also compares population-based measures of hospital utilization (ie, to measure the average utilization of hospital care per person living with HIV in each state) across the 6 states.

RESULTS

This study found that the average age of a hospitalized patient with HIV rose from 41 to 44 years and that the average number of diagnoses rose from 6.0 to 7.4. Moreover, it was found that the average number of admissions per person living with HIV fell 39% and that the percentages of female and black patients with HIV remained the same.

CONCLUSIONS

Hospitalized patients living with HIV are getting older and sicker, although the average number of admissions per person living with HIV continues to fall.

Hospitalizations for Metabolic Conditions, Opportunistic Infections, and Injection Drug Use Among HIV Patients

BACKGROUND

Rapid changes in HIV epidemiology and highly active antiretroviral therapy (HAART) may have resulted in recent changes in patterns of inpatient utilization.

OBJECTIVE

To examine trends in inpatient diagnoses and mortality in HIV patients.

DESIGN/SETTING/PATIENTS

Serial cross-sectional analyses of HIV patients hospitalized in 1996, 1998, and 2000, using hospital discharge data from the Healthcare Costs and Utilization Project for 12 states. Each hospitalization was classified as an opportunistic illness, complication of injection drug use (IDU), liver-related complication, ischemic heart disease, cerebrovascular disease, non-Pneumocystis carinii pneumonia (PCP), diabetes, or chronic hepatitis C virus (HCV).

MAIN OUTCOME MEASURES

Number of hospital admissions, inpatient mortality.

RESULTS

We evaluated 316,963 admissions that occurred between 1996 and 2000, with an overall mortality of 7%. Hospitalizations for opportunistic infections significantly decreased from 40% to 27% of all HIV-related admissions. The overall proportion of IDU complications remained relatively stable (6%) each year. Hospitalizations increased for liver-related complications from 8% to 13% and for chronic HCV from 1% to 5% in this period. The number of hospitalizations for cerebrovascular disease and for ischemic heart disease was relatively negligible in all years. Overall, inpatient mortality decreased between 1996 and 2000. Relatively higher mortality was observed among African Americans, Hispanics, those with Medicaid, those with Medicare, and the uninsured, however. Opportunistic infections and liver-related complications were associated with greater inpatient mortality.

CONCLUSION

Results do not show a significant recent rise in HIV-related inpatient utilization. Admissions to treat opportunistic infections have declined precipitously, consistent with the effects of HAART. Although not dramatic, liver-related disease is an increasing cause of hospitalization in HIV+ patients.

Inappropriate drug combinations among privately insured patients with HIV disease

OBJECTIVE

The objective of this study was to examine inappropriate drug combinations among privately insured patients with HIV disease.

DATA

Data were obtained from the MarketScan Commercial Claims and Encounter Database for the years 1999 and 2000.

METHODS

Each of the 2110 person-years of data examined in this study represents the claims experience of an enrollee with HIV disease who filled an antiretroviral medication prescription in either 1999 or 2000 for a protease inhibitor or nonnucleoside reverse transcription inhibitor. This study compares the claims experience of patients with HIV who filled a prescription for an inappropriate drug combination as specified in guidelines jointly issued by the U.S. Department of Health and Human Services and the Henry J. Kaiser Family Foundation with the claims experience of patients who did not.

RESULTS

An inappropriate drug combination was found in approximately 2% of the person-years of data, and persons who experienced an inappropriate drug combination had higher claims costs. One half of all of the inappropriate drug combinations involved a single lipid-lowering agent (simvastatin). Protease inhibitors decrease the activity of the enzyme that metabolizes simvastatin, and high concentrations of simvastatin have been associated with muscle damage. We found that patients who received protease inhibitors and simvastatin were more likely to experience muscle damage.

CONCLUSION

Persons with HIV have compromised immune systems and often take many medications. Thus, the risk and consequences of medication errors are severe, and both providers and patients should carefully monitor drug regimens to ensure that they are both safe and efficacious.

The epidemiology, treatment patterns, and costs of cytomegalovirus retinitis in the post-haart era among a national managed-care population

To examine the epidemiology, treatment patterns, and costs of cytomegalovirus (CMV) retinitis treatment in the post-HAART (highly active antiretroviral therapy) era, a retrospective cohort study was performed using data from US managed-care plans from 1997-2002. Cases with CMV retinitis were defined by requiring diagnosis codes for HIV (or AIDS), CMV, and retinitis and claims for anti-CMV treatment. Costs of oral, intravenous, and intraocular treatment periods were examined. The incidence of enrolled HIV or AIDS cases increased from 7 per million members in 1997 to 150 per million members in 2001. The incidence of CMV retinitis decreased from 23 per 10,000 HIV or AIDS cases in 1997 to 8 per 10,000 HIV or AIDS cases in 2001. The average duration of a CMV episode was 192 days and the average cost was 19,576 US dollars. In a multiple linear regression model adjusting for age, gender, insurance type, geographic region, HAART use, and co-existing AIDS-defining illnesses, intraocular and oral treatment periods saved 7135 and US dollars and 6866 US dollars, respectively, per treatment period compared with intravenous treatment (P < 0.05). The incidence of CMV retinitis decreased in this managed-care population during the post-HAART era. Use of oral or intraocular treatment saves costs compared with intravenous treatment in a managed-care environment.

Sensitivity of an AIDS case-finding algorithm: who are we missing?

OBJECTIVE

The objective of this study was to determine the sensitivity of an AIDS case-finding algorithm.

METHOD

This study applied the AIDS case-finding algorithm to paid Medicaid claims linked to New Jersey AIDS surveillance data and assesses its sensitivity across subgroups of patients.

FINDINGS

Of the 7183 cases with confirmed AIDS, based on the state's registry information, 95% (n = 6818) were correctly detected using all Medicaid claims, including pharmacy. For patients in a community-based waiver program, covered by Medicare, diagnosed with a severe mental illness, or continuously enrolled in Medicaid, with regular contact with the medical system, the algorithm identified almost all patients. To further evaluate algorithm performance, it was used with 2 groups of interest to researchers. For AIDS patients in the last 6 months of life, 88% were correctly detected without use of pharmacy claims and 95% when pharmacy claims were included. For pregnant women, data from the 6 months before the latest delivery date identify 27% of pregnant women without pharmacy claims and 41% when pharmacy claims were included. Using claims made after the latest delivery date, 81% of pregnant women were detected without pharmacy claims and 93% when pharmacy claims were included

CONCLUSION

Results demonstrate that a multilevel screen can be used with Medicaid claims to effectively to detect most patients with AIDS. Detection is lower for some subgroups, and the absence of pharmacy claims can compromise detection.

Antiretroviral therapy and health care utilization: a study of privately insured men and women with HIV disease

OBJECTIVE

To compare the use of antiretroviral therapy and other health care resources by women and men with HIV disease who are privately insured.

DATA SOURCES

Data were obtained from the MarketScan Commercial Claims and Encounter Database produced by the Medstat Group, Inc., of Ann Arbor, Michigan. This database includes eligibility files as well as claims data for inpatient care, outpatient care, physician services, and prescription drugs for enrollees in employer-sponsored benefit plans for 24 large employers around the nation.

STUDY DESIGN

Examine utilization by 2,026 privately insured persons (1,494 men and 532 women) with HIV disease in calendar year 2000 under the age of 65.

PRINCIPAL FINDINGS

Using a simple comparison, we found that 71 percent of men (68.7 to 73.3 percent is 95 percent confidence interval) and 39 percent of women (35.1 to 43.5 percent is 95 percent confidence interval) with HIV disease received antiretroviral therapy. We also found that the average annual drug cost was $9,037 for a man ($8,372 to $9,702 is 95 percent confidence interval) and $3,893 for a woman ($3,476 to $4,490 is 95 percent confidence interval). Furthermore, we found that the out-of-pocket expenses comprised 10 percent of total expenses for men ($1,617 out of $16,405) and 4 percent for women ($405 out of $10,397).

CONCLUSION

There are major differences in the utilization and cost of health care between privately insured men and women with HIV disease.

Recent Trends in HIV-Related Inpatient Admissions 1996–2000

BACKGROUND AND OBJECTIVES

HIV-related inpatient utilization declined immediately after the diffusion of highly active antiretroviral therapy (HAART), but some studies suggest that admission rates may have recently begun to increase. Using comprehensive hospital discharge data from 7 states, this study examines trends in HIV-related inpatient admissions and length of stay (LOS) from 1996 through 2000.

METHODS

We identified HIV-related admissions by ICD-9-CM diagnosis codes in the range from 042 to 044. Analyses assessed differential patterns of change over time, depending on state, gender, race/ethnicity, and insurance.

RESULTS

HIV-related inpatient admissions generally declined each year, but the rate of decline diminished recently. A similar pattern held for trends in inpatient LOS. Admissions for white male patients and for patients with private insurance showed the greatest decreases and the least leveling of the trend. The proportion of HIV admissions to total admissions was highest for black men and lowest for white women. In contrast to the period from 1993 through 1996, the proportion of HIV admissions covered by Medicare was greater than the rate of privately insured admissions.

CONCLUSIONS

Although there is no substantial evidence for widespread increases in admissions during this period, results suggest that the trend in HIV-related hospital admissions is level in recent years. Racial/ethnic disparities in inpatient utilization persist. Further analysis of the impact of treatment failure or HAART-related complications on HIV admissions is warranted.

Analysis of a population-based Pneumocystis carinii pneumonia index as an outcome measure of access and quality of care for the treatment of HIV disease

OBJECTIVES

A population-based Pneumocystis carinii pneumonia (PCP) Index was developed in New York City to identify geographic areas and subpopulations at increased risk for PCP.

METHODS

A zip code-level PCP Index was created from AIDS surveillance and hospital discharge records and defined as (number of PCP-related hospitalizations)/(number of persons living with AIDS).

RESULTS

In 1997, there were 2262 hospitalizations for PCP among 39 740 persons living with AIDS in New York City (PCP Index =.05691). PCP Index values varied widely across neighborhoods with high AIDS prevalence (West Village =.02532 vs Central Harlem =.08696). Some neighborhoods with moderate AIDS prevalence had strikingly high rates (Staten Island =.14035; northern Manhattan =.08756).

CONCLUSIONS

The PCP Index highlights communities in particular need of public health interventions to improve HIV-related service delivery.

Location, race, and hospital care for AIDS patients: an analysis of 10 states

This study is the first statewide comparison of hospital utilization and inpatient mortality rates for people with acquired immune deficiency syndrome (AIDS). Data from 120,772 hospital discharge abstracts for all AIDS-related admissions in 10 states (California, Colorado, Florida, Iowa, Kansas, Maryland, New Jersey, New York, Pennsylvania, and South Carolina) in 1996 were combined with data on the number and the racial and ethnic characteristics of all people living with AIDS (PLWAs) in each state. These data were used to derive population-based estimates of the use of hospital services per PLWA and of inpatient mortality rates in each state. Multivariate analyses examined sources of variation in inpatient length of stay and inpatient mortality. The primary finding of this study is that hospital utilization rates and inpatient mortality rates for people with AIDS vary substantially across states and among racial and ethnic groups within states even after adjusting for severity of illness. Blacks and Hispanics had longer hospital stays and were more likely to die in the hospital than whites. State-level policies, such as home and community-based waiver programs and enhanced HIV reimbursement rates, significantly affected hospital use.

Trends in HIV-related inpatient admissions from 1993 to 1997: a seven-state study

BACKGROUND AND OBJECTIVES

Reports of declining HIV-related inpatient use have typically been based on limited or local data. Using comprehensive hospital discharge data from seven states, this study examines trends in HIV-related inpatient admission rates and lengths of stay from 1993 through 1997.

METHODS

We identified HIV-related admissions by the International Classification of Diseases, ninth edition (ICD-9-CM) diagnosis codes in the range 042 to 044. Analyses assessed effects of state, gender, race/ethnicity, insurance, and time period; they also examined differential patterns of change.

RESULTS

HIV-related inpatient admission rates rose between 1993 and 1995 but began declining steadily starting in late 1995. This general pattern occurred for all states, demographic groups, and insurers. The magnitude of the decline varied; admissions for white males and for patients with private insurance showed the greatest decreases. Admission rates were highest for African-American males and lowest for white females. Lengths of stay declined steadily; trends did not differ by demographic group or payer.

CONCLUSIONS

Results document the recent decline in HIV-related hospital admission rates. Relative declines in admissions parallel reported disparities in access to new antiretroviral therapies. Racial/ethnic disparities in inpatient use persist.

Estimating the national cost of treating people with HIV disease: patient, payer, and provider data

OBJECTIVE

Existing estimates of the national cost of treating all people with HIV disease use data from a sample of people with HIV disease to extrapolate the cost of treating all people with HIV disease (patient-based approach). This study derives estimates using two novel approaches (i.e., payer-based and provider-based) and compares these with existing estimates.

DATA SOURCES

These include the Health Insurance Association of American and the American Council of Life Insurance 1996 HIV survey, the 1996 State Inpatient Databases (SID) maintained by the Agency for Healthcare Research and Quality, and the IMS America Ltd. survey of independent and chain drugstores.

PRINCIPAL FINDINGS

The cost of treating all people with HIV disease in 1996 was between $6.7 and $7.8 billion U.S., and the average annual cost of treating a person with HIV disease was between $20,000 and $24,700 U.S.

CONCLUSIONS

Analysts should derive estimates of the cost of treating people with HIV disease using several different approaches. K

Frequency and predictors of medically attended injuries in HIV-infected children

The extent to which medically attended injuries complicate the clinical course of HIV-infected (HIV+) children is unknown. In a cohort of HIV+ children delivered from 1985 to 1990 and aged less than 60 months, we determined medically attended injuries per 100 child-years, Injury Severity Scores (ISS), and predictors of medically attended injuries by using New York State Medicaid claims from 1986 to 1992 linked to birth certificates. Injury rates and ISS were compared to those of a population of black, inner city children aged less than 60 months from emergency room records. HIV+ children had slightly more injuries (19.3 vs. 16.8/100 child-years) but similar ISS (2.4 vs. 2.3). Predictors of injuries in HIV+ children included younger maternal age (24/100 child-years, p = 0.008) and delivery outside of New York City (29/100 child-years, p = 0.02). Illicit drug use and alcohol use were associated with greater ISS while cocaine use was associated with a higher rate of possibly intentional injuries. Medically attended injuries affected one in five HIV+ children in our cohort annually, slightly more than the comparison population. Specific maternal and birth characteristics such as substance abuse and younger age at delivery may help target at-risk children.

Incidence and duration of hospitalizations among persons with AIDS: an event history approach

OBJECTIVE

To analyze hospitalization patterns of persons with AIDS (PWAs) in a multi-state/multi-episode continuous time duration framework.

DATA SOURCES

PWAs on Medicaid identified through a match between the state's AIDS Registry and Medicaid eligibility files; hospital admission and discharge dates identified through Medicaid claims.

STUDY DESIGN

Using a Weibull event history framework, we model the hazard of transition between hospitalized and community spells, incorporating the competing risk of death in each of these states. Simulations are used to translate these parameters into readily interpretable estimates of length of stay, the probability that a hospitalization will end in death, and the probability that a nonhospitalized person will be hospitalized within 90 days.

PRINCIPAL FINDINGS

In multivariate analyses, participation in a Medicaid waiver program offering case management and home care was associated with hospital stays 1.3 days shorter than for nonparticipants. African American race and Hispanic ethnicity were associated with hospital stays 1.2 days and 1.0 day longer than for non-Hispanic whites; African Americans also experienced more frequent hospital admissions. Residents of the high-HIV-prevalence area of the state had more frequent admissions and stays two days longer than those residing elsewhere in the state. Older PWAs experienced less frequent hospital admissions but longer stays, with hospitalizations of 55-year-olds lasting 8.25 days longer than those of 25-year-olds.

CONCLUSIONS

Much socioeconomic and geographic variability exists both in the incidence and in the duration of hospitalization among persons with AIDS in New Jersey. Event history analysis provides a useful statistical framework for analysis of these variations, deals appropriately with data in which duration of observation varies from individual to individual, and permits the competing risk of death to be incorporated into the model. Transition models of this type have broad applicability in modeling the risk and duration of hospitalization in chronic illnesses.

Cigarette smoking and maternal-child HIV transmission

We investigated the association of cigarette smoking with maternal-child HIV transmission, adjusting for illicit drug use, maternal clinical status, and delivery factors. Vital statistics birth data were linked to the New York State Medicaid HIV/AIDS Research Database for HIV-infected women delivering a liveborn singleton from 1988 through 1990. Follow-up of these children was accomplished by Medicaid data > or = 2 years after birth, and their HIV status was ascertained by a clinically based classification. The adjusted relative risk or hazard (RH) of transmission for maternal factors was determined from Cox models. The overall transmission was 24.5% for the 901 maternal-child pairs. Smokers comprised 40% of women with data on smoking (n = 768); their transmission rate was 31% versus 22% for nonsmokers (p = 0.02). In the entire cohort, the adjusted RH of transmission for smokers was 1.45 (95% confidence interval [CI] 1.07-1.96); among women with advanced HIV, the adjusted RH was even higher (RH = 1.71; 95% CI 1.14-2.58). Users of cocaine (15% of the cohort) or of mixed or unspecified illicit drugs (28%) had higher transmission rates in unadjusted analysis (33%, p = 0.06 and 31%, p = 0.06 respectively); after adjustment for smoking and other maternal factors, neither cocaine (RH = 1.04 (95% CI 0.66-1.63)) nor mixed nor unspecified drug use (RH = 1.13 (95% CI = 0.75-1.70)) was significantly associated with transmission. Our data document an association of cigarette smoking during pregnancy with an increased risk of maternal-child HIV transmission that can be added to the growing list of complications caused by cigarette smoking.

Association of maternal HIV infection with low birth weight

We evaluated factors associated with low birth weight (LBW) in an HIV-infected cohort (n = 772) and a general sample (n = 2,377) of women delivering a live singleton in federal fiscal years 1989 and 1990 while enrolled in New York State Medicaid. The association of LBW and HIV infection was studied in logistic models, controlling for illicit drug use, demographic characteristics, adequacy of prenatal care, and medical risk factors. Overall, 29% of the HIV-infected women had a LBW infant compared to 9.3% of the general sample (p < 0.001). The adjusted odds of LBW for HIV-infected women were twofold higher than for uninfected women [odds ratio (OR) = 2.04 and 95% confidence interval (Cl) = 1.54, 2.69]. Odds of LBW were also increased for illicit drug users (OR = 2.16; 95% CI = 1.59, 2.94), cigarette smokers (OR = 1.81; 95% CI = 1.37, 2.39), and African-American versus non-Hispanic white women (OR = 1.89; 95% CI = 1.31, 2.72). Lower odds appeared for women with adequate prenatal care (OR = 0.54; 95% CI = 0.42, 0.68). Among only women with full-term deliveries, the association of HIV with LBW remained strong as we found nearly threefold greater odds of LBW for HIV-infected women. This study indicates that HIV-infected women have an increased risk of bearing a L.BW infant, even after adjusting for the effects of drug use, health care delivery, and other social and medical risk factors.

Prenatal care and birth outcomes of a cohort of HIV-infected women

Adequate prenatal care has been linked to improved birth outcomes in general populations but has not been assessed in HIV-infected women. We examined longitudinal claims files and vital statistics records for women in the New York State Medicaid HIV/AIDS data base delivering a singleton from 1985 through 1990. Adequacy of the self-reported number of prenatal visits was assessed by the Kessner index. In logistics models, we estimated the association of prenatal care, illicit drug use, and other maternal characteristics with three outcomes; low birth weight, preterm birth, and small-for-gestational-age. Of 2,254 singletons delivered by this HIV-infected cohort, 28% were low birth weight, 23% were preterm birth, and 20% were small for gestational age. Two-thirds had inadequate prenatal care. Non-drug users had 57 and 26% lower adjusted odds of low birth weight and preterm delivery than drug users. The adjusted odds of low birth weight and preterm birth for women with an adequate number of prenatal visits were, respectively, 48 and 21% lower than for women with inadequate care. Adequate prenatal care was also associated with a 43% reduction in the odds of small-for-gestational-age. An adequate number of prenatal visits by women in this HIV cohort was associated with a significant reduction in all three adverse birth outcomes, but most had inadequate prenatal care. These data support strengthening efforts to bring pregnant, HIV-infected women into care.

Completeness of tuberculosis case reporting, San Juan and Caguas Regions, Puerto Rico, 1992

Completeness of tuberculosis case reporting in Puerto Rico was assessed. Cases diagnosed among hospitalized, tuberculosis, and human immunodeficiency virus clinic patients during 1992 were retrospectively reviewed. Hospital discharge diagnoses, pharmacy listings of patients receiving anti-tuberculous medications, laboratory and acquired immunodeficiency syndrome registry data were used for case finding in selected hospitals and clinics. Identified cases were matched to the health department TB case registry to determine previous reporting through routine surveillance. Records of unreported cases were reviewed to verify tuberculosis diagnoses. Of 159 patients with tuberculosis, 31 (19.5%) were unreported. A case was defined according to the Centers for Disease Control and Prevention definition. Unreported cases were less likely than previously reported cases to have specimens that were culture positive for M. tuberculosis, 14 of 31 (45.2%) compared with 111 of 128 (86.7%). Excluding the laboratory, tuberculosis diagnoses in acquired immunodeficiency syndrome registry patients had the highest predictive value of finding tuberculosis (94.1%), followed by tuberculosis clinic records (71.7%), and pharmacy listings (45.6%). Tuberculosis discharge diagnoses, however, yielded the largest number of unreported cases (14). Health care providers should be educated regarding the importance of promptly reporting all suspected TB cases regardless of results of laboratory testing.

Using administratively collected hospital discharge data for AIDS surveillance

The objectives of this study were twofold: to improve methods of identifying possible and acquired immunodeficiency syndrome (AIDS)-related hospital discharges in administrative databases and to measure AIDS-reporting completeness in Massachusetts both overall and by subgroup. We used fiscal year 1988 discharge data from the Massachusetts Rate Setting Commission (RSC) and data from the Massachusetts AIDS Reporting System (ARS). We identified 3362 discharges of adult patients (> 12 years old) from the RSC file that had diagnosis codes which are human immunodeficiency virus (HIV)-specific (042.x, 043.x, 044.x, or 795.8) or pertain to AIDS-defining "manifestations." Medical records of 650 patients apparently not reported to the ARS were reviewed. THe best set of codes overall consisted of either (a) the 042.x code or (b) the 043.x, 044.x, or 795.8 code plus selected manifestation codes (sensitivity, 93%; specificity, 86%; predictive value positive, 71%). Of the 927 AIDS cases identified from the 3362 discharges, only 36 had not been reported. AIDS cases among women (odds ratio (OR) = 2.9; 95% confidence interval (CI): 1.33 to 6.33), intravenous drug users (OR = 4.2; 95% CI: 2.20 to 8.02), and persons residing outside the Boston metropolitan area (OR = 2.3; 95% CI: 1.18 to 4.57) were more likely to be unreported than those among comparison groups.

Patterns of ambulatory care for AIDS patients, and association with emergency room use

OBJECTIVE

We examined the association of patterns of ambulatory care for AIDS patients with any use of the emergency room (ER) and the monthly rate of ER visits in the six months after AIDS diagnosis.

DATA SOURCES/STUDY SETTING

The study population was obtained from the New York State Medicaid HIV/AIDS Research Data Base and includes patients diagnosed with AIDS from 1983 to 1990.

DATA COLLECTION/EXTRACTION METHODS

To examine patterns of care and ER use not leading to hospitalization, we studied patients who survived at least six months after their first AIDS-defining diagnosis. The data base included person level information on visits to different provider sites and patient demographic and clinical characteristics.

STUDY DESIGN

We defined the dominant provider as the site delivering the majority of ambulatory care for patients with a minimum of four ambulatory visits in the six months after AIDS diagnosis. Dominant providers were classified by specialty and setting: generalist physician; general medicine clinic; AIDS specialty clinic; and other specialty clinic or physician (e.g., cardiology). Patients without a dominant provider were grouped into those with four or more visits and those with fewer than four visits. Regression analysis was used to estimate relationships between ER use and patterns of ambulatory care and patient demographic and severity of illness characteristics.

PRINCIPAL RESULTS

The study population included 9,155 AIDS patients aged 13 to 60 years at diagnosis, continuously Medicaid-enrolled, and surviving at least six months after AIDS diagnosis. Among those with four or more visits (56 percent), over 70 percent had a dominant provider. Overall, 39 percent of the study population visited the ER while, in the group with four or more visits, 53 percent of those without a dominant provider had an ER visit. Patients without a dominant provider were estimated to have 32 percent higher odds of ER use than patients with a dominant provider. Among patients with a dominant provider, patients with a generalist or primary care clinic dominant site of care were estimated respectively to have 18 percent and 23 percent lower odds than patients with an AIDS specialty clinic as the dominant site of care. Drug users had higher odds of ER use, as did women.

CONCLUSIONS

In this Medicaid AIDS population, a dominant provider delivering the majority of a patient's care was associated with less use of the ER by the patient. Among patients with a dominant provider, ER use was lowest for those with a primary care provider. Further examination of the type and availability of ambulatory services in AIDS specialty clinics and primary care settings, as well as more detailed information on patient characteristics, may reveal reasons for these patterns of ER use.

Patterns of Medicaid expenditures after AIDS diagnosis

This article examines average monthly Medicaid expenditures after diagnosis of acquired immunodeficiency syndrome (AIDS) for the diagnosis, mid-illness, and death intervals, as well as Kaplan-Meier estimates of expenditures from AIDS diagnosis to death. A clinical severity measure (the Severity Index for Adults with AIDS [SIAA]) designed to be predictive of patient survival was applied to a population of continuously enrolled New York State Medicaid patients who survived at least 6 months after being diagnosed with AIDS. Our findings suggest that groups of more seriously ill patients who appear to have more intense demand for health care services, especially over the diagnosis and mid-illness intervals, can be identified using the SIAA.