The HIV Research Network: An Update on Findings

1
The HIV Research Network An Update on Findings

• Richard D. Moore, MD, MHS
• Kelly A. Gebo, MD MPH
• August 30, 2006

2
Background Determining HIV/AIDS Costs and Health
Service Utilization in the U.S.

• AIDS Costs and Services Utilization Study
  (1991-92)
• Multisite sample of AIDS patients in care
• HIV Costs and Services Utilization Study
  (1996-98)
• Nationally representative sample of 2,864 HIV
  patients in care
• Probability sample permits strong inferences to
  national population source of important data on
  delivery of HIV care in U.S.
• Lessons Learned from HCSUS
• Over 120 papers evaluating health services
  utilization, health care costs, disparities in
  access to care, and clinical issues
• Recruiting sample expensive and time-consuming
• Sample becomes unrepresentative of population
  over time, unless refreshed
• Can be difficult to obtain medical records from
  providers not linked to study

3
HIV Research Network (HIVRN)

• Network of HIV primary care providers who collect
  and transmit clinical and health services
  delivery data from HIV-infected persons in care
  for aggregate analyses of clinical and health
  services outcomes of the HIV care in the U.S.

4
Network Research Objectives

• Key issues to address
• Evaluate health care utilization and costs of
  providing these services
• Access to care and disparities in utilization
• Quality of care and patient safety
• Substance abuse/Mental health
• Clinical disease progression and outcomes
• Adult and Pediatric

5
HIV Research Network
6
HIVRN Collaborators

• Adult Sites
• Victoria Sharp- St. Lukes Roosevelt, NYC
• W. Christopher Mathews- UCSD, San Diego
• Philip Keiser- Parkland Hospital, Dallas
• James Hellinger- Community Medical Alliance,
  Boston
• Patrick Nemecheck- Nemecheck Health Renewal,
  Kansas City
• P. Todd Korthuis- OHSU, Portland
• Charurut Somboonwi Tampa General Health Care,
  Tampa
• Robert Beil- Montefiore Medical Center, NY
• Lawrence Hanau- Montefiore Medical Center, NY
• Lawrence Crane- Wayne State University, Detroit
• Norm Markowitz Henry Ford Hospital, Detroit
• Silver Sisneros Alameda County Consortium,
  Oakland
• Roberto Corales Community Health Network,
  Rochester
• Peter Sklar- Drexel University, Philadelphia
• Kelly Gebo- Johns Hopkins, Baltimore

• Pediatric Sites
• Stephen Spector-UCSD, San Diego
• Aditya Gaur St. Judes, Memphis
• Richard Rutstein- CHOP, Philadelphia
• Victoria Sharp-St. Lukes Roosevelt, NYC
• Data Coordinating Center
• Johns Hopkins
• Richard Moore
• Kelly Gebo
• George Siberry
• Funding Sources
• AHRQ
• SAMHSA
• HRSA
• OAR

7
HIVRN Governance

• Steering Committee comprised of all
  Investigators annual meeting, quarterly
  conference calls
• Executive Committee of Project Officers and DCC
  staff monthly meetings.
• Data Subcommittee approves proposals for
  analyses regular conference calls and email
  correspondence.
• Dissemination of Data/Results
• Internet HIVRN data available (www.AHRQ.gov/data/h
  ivnet.htm)
• Intranet website for the HIVRN members

8
Methods

• Sites individually collect information
  electronically and by chart abstraction
• De-identified information sent to Central Data
  Coordinating Center (DCC)
• Data cleaned, quality assured
• Reports sent back to sites for confirmation of
  data interaction with sites to resolve data
  issues
• Compatible, longitudinal, multisite database
  created and maintained at the DCC

9
Clinical Data

• Age, sex, race, HIV transmission risk factors
• AIDS-defining illness
• CD4, Viral loads
• Antiretroviral drugs
• Opportunistic illness prophylaxis drugs
• Hepatitis (HCV, HBV) serologies
• Date of death, cause
• Other lab data lipids, glucose, renal function
• Other drugs
• Resistance tests

10
Resource Utilization Data

• Acute/chronic hospital care
• Admission/Discharge dates
• Diagnoses
• Outpatient Visits
• Dates of service
• Diagnoses
• CPT Coding
• Emergency Department
• Substance Abuse/Mental Health Visits
• Insurance

11
Characteristics of the Adult sample in CY 2004
(N16,745)
12
Characteristics in CY 2004
Insurance Coverage
Ryan White CARE Act Funding at 14/15 adult sites
13
Topics Addressed

• Healthcare Utilization
• IP/OP utilization adult and peds
• ED Utilization
• IP diagnoses
• Overall costs of care
• MH-SA analyses
• ETOH usage
• Effect of MH/SA on Health Care Utilization
• Effect of MH/SA on VL suppression
• Quality of care
• Disparities in Access to Care
• Rural
• Clinical Analyses
• Virologic Suppression (Adult and Peds)
• Pain
• Quality of Life

14
Health Services Utilization and Costs of
Care Among HIV-Infected Adults in Care 2000-2003

15
Health care utilization (CY 2000 - 2002)
Significantly different
Fleishman Med Care 2005
16
Health care utilization (CY 2000-02)
Significantly different
Fleishman Med Care 2005
17
Hospitalization Rates
AIDS-Related Pneumonia, PCP GI Pancreatic
diseases, liver diseases Mental Health
Substance-related, affective disorders Circulatory
Carditis, hypertension
Rate (per 100 Person Years)
Betz Medical Care 2005
18
Total Costs of HIV Care by CD4
Gebo CROI 2006
19
Comprehensive Costs of HIV Care by CD4 Count
Gebo CROI 2006
20
Changes from HIVRN utilization data

• We are currently utilizing data from E.R. visits
  to ascertain various modes which patients use to
  access care
• (1) those who use E.R. and
• (2) those who use the urgent care clinic for
  primary care.
• With this data we will be able to identify
  clients who need help in obtaining primary care
  in our clinic
• Kathleen Clanon, M.D., Alameda County Medical
  Center
• Our monthly collection of CD4 count, viral load
  values, and missing values has encouraged
  clinicians to more closely track both the
  patients in the clinic, and patients who have
  missed appointments and are late for quarterly
  clinical and lab monitoring. This has resulted
  in additional efforts to track patients who have
  missed visits.
• James Hellinger, M.D. Community Medical
  Alliance, Boston, MA

21
Substance Abuse and MH Analysis
22
Alcohol Consumption

• 40 reported any ETOH use within the preceding
  month
• 10 of the sample reported hazardous/binge
  drinking
• Of those who consumed EtOH, 25 were HBD

23
Prevalence of Illicit Drug Use

• 41 Current Drug Use
• 33 Former Drug Use (not within 6 mos)
• 27 Never drug use

24
Drug Use by Drug Type
25
Multivariable Factors Associated with Any
Drinking in Past Month
Chander RAS 2005
Adjusted for PCP visits and site of care
26
Multivariate factors Associated with HBD
Chander RAS 2005
Adjusted for site
27
HAART Usage by EtOH Use
plt.001
28
Percent of Patients With Any Inpatient Medical
Hospitalization, CY 2001
Himelhoch CROI 2005
29
Adjusted Odds Ratios of Any Inpatient
Hospitalization
Adjusted for CD4 and site
30
Percent of Patients with RNAlt400
Chander SGIM 2005
31
Adjusted Odds Ratios of Virologic Suppression
Adjusted for CD4, site
32
Quality of CareHAART and OI Prophylaxis
Utilization
33
Pharmacy Utilization

• HAART Usage (CD4lt350) 91
• PI Backbone 68
• NNRTI Backbone 63
• PCP (2 or more CD4lt200) 88
• MAC (2 or more CD4lt50) 87

34
Factors Associated with HAART Usage (CD4lt350)
Adjusted for site of care, insurance
Gebo et al JAIDS, 2005
35
Factors Associated with PCP Prophylaxis (CD4lt200)
Gebo Medical Care 2005
Adjusted for site of care, insurance
36
Factors Associated with MAC Prophylaxis (CD4lt50)
Adjusted for site of care, insurance
Gebo Medical Care 2005
37
Urban vs. Rural Demographics
Wilson IDSA 2006
38
Clinical Factors
Wilson IDSA 2006
39
Pharmaceutical Utilization
40
Medical Utilization
41
Clinical Changes from PCP/MAC Project

• Projects in the works now include a red flag
  letter that notifies docs of particular
  deficiencies (such as lack of PCP or MAI
  prophylaxis, patients on triple nuke therapy and
  regimens that have incorrect dosing or contains
  meds that shouldn't be used together).
• Robert Beil, MD- Montefiore Medical Center
• 'The data obtained.has been helpful in
  identifying other opportunities to improve and
  comply with HIV/AIDS national guidelines.
  Tracking the CD4 and meds listed on the same page
  is a reminder to start the patient on prophylaxis
  as needed.
• John Jovanovitch, MD - Henry Ford Hospital
  System, Detroit, MI

42
Current Analyses

• ED Utilization
• Medical Utilization by SA/MH/HCV status
• Pain and QOL Analyses
• Pediatric Diagnoses Across Time
• SA Analyses
• Effect of SA tx on adherence
• Effect of PCP counseling to those with SA d/o

43
Conclusions

• Functioning multisite network collecting HIV data
  in cost efficient manner
• Near real time data collection with quick
  feedback to sites
• Have been able to conduct rapid analyses to
• Address disparities in care and safety issues
• Evaluate Health care utilization patterns
• Clinical outcomes
• Data from the HIVRN has been useful for
• Allocation of healthcare resources
• Appropriation of health care costs dollars
• Improvement of HIV treatment strategies

44
Lessons Learned

• An admission/diagnosis code/insurance status in
  Kansas City is not the same as in Boston
• Communication is essential
• Authorship rules
• Research proposals
• Intranet Website
• Site visits are invaluable
• Data collection and QA processes can lead to
  improved clinical care

45
Acknowledgments

• Patients at each site
• Site Investigators and data managers
• Project collaborators within and external to
  HIVRN
• Project Officers
• Funding Agencies
• AHRQ
• HRSA
• SAMHSA
• OAR

46
(No Transcript)
47
Conclusions

• Inpatient, but not outpatient utilization
  significantly decreased between 2000 and 2002
• IP rates higher in blacks, Hispanics, women, and
  more immunosuppressed
• OP rates highest in Hispanics, older patients and
  more immunosuppressed
• Most common reasons for hospitalization are OIs,
  particularly pneumonia and PCP
• HIV care is expensive, over 57K/yr for CD4lt50
• ART drugs most costly category, except for
  sickest patients

48
Conclusions

• Prevalence of EtOH use (40) is high
• Factors associated with ANY EtOH use male sex, gt
  HS education, and current drug use,
• Of those who drink, 25 are HBD
• Factors associated with HBD CD4gt200, current or
  former illicit drug use
• HBD adversely affects adherence to HAART

49
Methods

• Study Design Cross Sectional Analysis
• Population HIV adults followed at 4 sites in
  the HIVRN, with on site psychiatric care
• 4 sites represent nearly 40 of subjects followed
  in HIVRN
• Median sample size per site of 1922 (range
  256-2136).
• Definitions
• SMI was defined using the VA Serious Mental
  Illness Treatment and Research Evaluation Centers
  definition and includes ICD-9 codes
  295.0-9296.0-1296.4-8297.0-3297.8-9298.0-9.
• IDU was defined as a history of injection drug
  use
• Time Period January 2001 December 31, 2001
• Data Demographic, clinical resource utilization
• Outcome Inpatient utilization
• Independent Variables SMI/IDU
• Analysis Number of medical admissions and median
  length of hospital stay (LOS) using chi squared
  and constructed logistic regression models to
  adjust for potential confounders.

50
Baseline Demographic Characteristics by MH/SA
Category
51
Baseline Demographic Characteristics of 2001
Sample
52
Among Those Hospitalized, Percent with gt1
Hospitalization, CY 2001
53
Adjusted Odds Ratios of gt1 Hospitalization
Adjusted for site
54
Conclusions

• As compared to those without SMI or IDU,
  individuals with SMI IDU who receive outpatient
  HIV care
• (1) Less likely to receive HAART
• (2) More than 2.5 times the odds of being
  medically hospitalized during the CY 2001.
• (3) More than 2.5 times the odds of being
  medically hospitalized more than once in CY 2001

55
Methods

• Study Design Retrospective cohort study
• Population HIV adults followed at 10 sites in
  the HIVRN
• East (4), Midwest (2), South (2), and West U.S.
  (2)
• 8 Academic, 2 Community Based
• Time Period January 1 December 31, 2001
• Outcome HAART, PCP, and MAC utilization
• Analysis Logistic regression

56
Conclusions

• High rates of HAART usage, OI prophylaxis
• Racial and HIV Risk factor disparities associated
  with HAART usage persist
• Males more likely to receive PCP prophylaxis than
  females, however no demographic disparities for
  MAC
• Number of outpatient visits factor most strongly
  associated with receiving adequate HAART, OI
  prophylaxis

57
Conclusions

• Rural patients are more likely to be white and
  IDU than urban patients
• Medical and pharmaceutical utilization was
  similar between 2 groups
• Exception Rural less likely to receive PCP
  prophylaxis
• Clinical outcomes excellent, with VL suppression
  similar in both groups

58
Implications

• HIV care delivered by urban providers to rural
  patients is equivalent to that provided to urban
  patients
• Future work needed to see if there are
  differences in quality of care if providers
  travel to patients or patients to providers

59
Implications

• Improving adherence to outpatient visits may
  improve access to life saving therapies
• Future studies needed to assess if disparities
  are a result of provider, patient or
  provider/patient interaction

60
Methods

• Study Design Cross Sectional
• Population HIV adults followed at 5 sites in
  the HIVRN
• East (1), South (2), and West U.S. (2).
• 5 Academic Based
• Time Period January December 31, 2004
• Patients categorized using the University of
  Washington rural health categorization schema
  into rural (lt 10K) and urban (gt 100K)
• Data Demographic, resource utilization, pharmacy
  use
• Outcomes
• Demographics
• Quality of Care as measured by resource use

61
Methods

• Using home zip codes in 2004, 13,941 patients
  from 5 HIV Research Network sites (1 Eastern, 2
  Southern, 2 Western) were categorized using the
  University of Washington rural health
  categorization schema into rural (lt 10K) and
  urban (gt 100K) clinical and demographic
  characteristics, inpatient and outpatient
  utilization, quality of care, and virologic
  suppression were compared using chi-squared tests
  for categorical variables and t-tests for means.

62
Methods

• Analyses approval process
• Research question posed via intranet
• Reviewed by DSC and feedback to investigator
• Research team assembled
• Deidentified data disseminated to investigator
• Final analyses at DCC to adjust for site
• Reports/abstracts/manuscripts prepared by
  initiating author and participating co-authors
• Dissemination of Data/Results
• Internet HIVRN data available (www.AHRQ.gov/data/h
  ivnet.htm)
• Intranet website for the HIVRN members
• Submission of research ideas, suggestions for new
  variables
• Abstracts, posters, papers
• Variable List
• All contact information

63
Interview (conducted in 2003-04)

• 951 adult and 300 pediatric interviews
• Oversampled women and Hispanics
• Topics assessed include
• Detailed health care utilization data
• Case management, home care, pharmacy, CAM
• Adverse Drug Events
• Illicit Drug and Alcohol Usage
• Adherence to ART
• HIV related symptoms and quality of life
• Mental Health and Substance Abuse treatment
  utilization

64
Presentations

• Presentations (examples)
• American Public Health Association
• American Academy of Pediatrics
• Academy Health
• Conference on Retroviruses and Opportunistic
  Illness
• Infectious Diseases Society of America
• International AIDS Society
• International Conference on Scientific Basis of
  Health Services Research
• International Society for Quality in Health Care
• International Society of Pharmacoepidemiology
• Pediatric Academic Societies Meeting
• Robert Wood Johnson Clinical Scholars Meeting
• Society of General Internal Medicine

65
Manuscripts

• 2002 JAIDS health care utilization and costs
• 2004 JAIDS disparities in access to HAART
• 2005 Medical Care
• Longitudinal update of health care utilization in
  adults
• Diagnoses associated with hospitalization
• Rates of OI prophylaxis
• Pediatric health care utilization
• Pediatric suppression of HIV viral load
• Manuscripts in press/under review at
• AJPH
• AIDS Care
• Alcoholism Clinical and Experimental Research
• Health Services Research
• HIV Research
• JAIDS
• Medical Care

66
Methods

• Study Design Cross Sectional Survey
• Population HIV adults followed at 14 sites in
  the HIVRN, oversampled Hispanics/Women
• East (6), Midwest (3), South (2), and West U.S.
  (3).
• 7 Academic, 7 Community Based
• Time Period January December 31, 2003
• Data Demographic, resource utilization, QOL,
  ETOH, illicit drug use, adherence
• Outcomes
• Any ETOH consumption in past 1 month
• Hazardous/Binge Drinking
• gt 14 drinks/wk in women, gt21 drinks/wk in men
  gt5 drinks at once
• Analysis Logistic Regression