SECOND ANNUAL POPULATION HEALTH
POSTER SESSION
SELECTED ABSTRACTS
The following are selected abstracts from the 2002-2003 Annual Poster
Session sponsored by the Population Health Program Student Organization.
Over 50 posters were presented. The abstracts included below demonstrate
the breadth and depth of faculty-student research collaboration typical of
the M.S./Ph.D. Program in Population Health.
Poster
# 01: The Medical Costs of Genital Warts in a Set of Private U.S. Health
Plans.
Insinga R,
Dasbach E, Carides G
Objective
and Rationale: This study
examines the medical costs, physician visits, and duration of an episode
of care for genital warts (GW), among privately insured individuals
generating health care claims within a set of U.S. health plans. Only one
previous study, conducted in the Netherlands, has examined the costs
associated with care for GW in a population setting.
Data and
Methods: Inpatient, outpatient
and pharmacy payments were derived from the Medstat Marketscan® database,
containing fee-for-service health insurance claims from U.S. privately
insured individuals. From an eligible sample of 785,944 individuals,
incident cases for 1998 were identified using ICD-9 code 078.11 (condyloma
acuminatum). Point estimates and 95% confidence intervals were estimated
using a 2-stage conditional method, to account for administrative
censoring.
Results:
We identified 536 individuals
initially diagnosed with GW in 1998. On average, GW episodes resulted in
3.1 (95% C.I.: 2.8, 3.4) physician visits, cost $436 (95% C.I.: $376,
$496), and lasted 92.7 (95% C.I.: 85.3, 100.2) days. Costs were slightly
greater for males than females ($477 vs. $404), as was the duration of
care (102.6 vs. 84.8 days), however physician visits were similar across
groups (3.1 vs. 3.1 visits). Costs were highest for males age 30-44
($598) and females age 15-29 ($494).
Conclusions: Mean wart costs
across age and gender subgroups ranged from $300-600, with most costs
accruing during the first 3 months following the initial visit. Mean
costs and physician visits were highest among young adults.
Poster # 02: Positive
Association Between IgG AND IgA Antibodies to Chlamydia Pneumoniae
and Acute Myocardial Infarction in Young Males in the U. S. Military.
C.M. Arcari,
K.E. Nelson, M. Krauss, C.A. Gaydos
Despite the progress in prevention,
diagnosis, and treatment, coronary heart disease (CHD) still remains the
leading cause of death in the United States, and established risk factors
are unable to explain fully half of all CHD cases. The role of infections
in the pathogenesis of CHD has been postulated. This study addresses the
hypothesis that previous C. pneumoniae infection is a significant
independent risk factor for CHD.
Data and specimens were collected from an
ongoing and well-established prospective cohort of United States Military
active-duty personnel. A nested case-control study was conducted using
300 cases and 300 matched controls. Results of this study to date
demonstrate a significant association between high titer IgG and IgA
antibodies to C. pneumoniae and acute MI.
Despite current prevention efforts, CHD
continues to be the number one cause of death in the United States. It is
an important challenge to identify new modifiable risk factors, which
could be used in prevention strategies to reduce the morbidity and
mortality associated with CHD. If C. pneumoniae infection is an
independent risk factor for CHD, then prevention of infection using
antibiotics or vaccine may lower the incidence of CHD.
Poster # 03: Cervical Cancer in Wisconsin: Increasing
Incidence and and Mortality Among Younger Women.
John R.
Pfister, Laura Povinelli, and
Daniel F.I. Kurtycz
While there has been a substantial
decrease in the incidence of cervical cancer and associated mortality over
the past several decades primarily due to the success of mass screening
programs employing the Papanicolaou (Pap) test, 4,205 women in the United
States, including 69 in Wisconsin, died of cervical cancer in 1999.
Wisconsin’s 1990 public health plan set an objective to reduce the
age-adjusted mortality rate due to cervical cancer by 50% between 1985 and
2000. To assess Wisconsin’s progress toward this goal, data from several
sources were evaluated: 1) cervical cancer mortality data from the CDC
WONDER on-line databases, 2) in-situ and invasive cervical
cancer incidence rates calculated from data published by the Wisconsin
Bureau of Health Information, 3) prevalence rates of cervical squamous
intraepithelial lesions on Pap tests conducted by the Wisconsin State
Laboratory of Hygiene on more than 500,000 women attending family planning
clinics, and 4) Pap testing data from the Behavioral Risk Factor
Surveillance System (BRFSS) surveys. Preliminary analyses reveal
increasing cervical cancer mortality rates among middle-aged women,
increasing incidence of in-situ and invasive cancer and relatively high
prevalence of intraepithelial lesions in young women, racial disparities
in both incidence and mortality, and lower Pap test utilization by the
less educated. The need for improvements in current cervical cancer
prevention efforts, and possible strategies, will be discussed.
Poster # 04: Epidemiology
of Menarche in a Population‑based Incident Cohort of Young Women with Type
1 Diabetes.
K.K. Danielson,
C.I. Allen, M. Palta, and D.J. D'Alessio for the Wisconsin Diabetes
Registry Project.
Whether type 1 diabetes and its management
affect age at menarche is not yet understood. Therefore, we 1) determined
mean age at menarche in a population‑based incident cohort of 188 young
women with type 1 diabetes diagnosed between the ages of 1 ‑ 29, and 2)
examined the effect of diabetes management on age at menarche in the
subgroup of 90 girls diagnosed before menarche. Data on age at menarche,
diabetes management, and body‑mass index (BMI) were collected with
biannual questionnaires and at exams. Glycosylated hemoglobin (GHb)
levels were determined from blood samples collected up to three times each
year. For the cohort, mean age at menarche was 12.78 + 1.33 (range:
9.59 ‑ 18.10). This was not significantly different from the average age
at menarche in the United States population reported by the National
Health Examination Survey (p= .92). Age at diagnosis of type 1 diabetes
was not associated with age at menarche (r=‑.05, p=.46). For the subgroup
of 90 girls, data on diabetes management and BMI in the three years prior
to menarche were analyzed. Mean GHb was significantly associated with age
at menarche (r= .21, p=.03). In multivariable linear regression,
controlling for mean BMI and race, mean GHb remained a significant
predictor (=.11, p=.04). Mean blood glucose checks, insulin injections,
and insulin dose were not associated with age at menarche. While mean age
at menarche for all young women with type 1 diabetes in our population did
not differ from the U.S. population, our findings among the subgroup
diagnosed prior to menarche suggest that glycemic control influences age
at menarche.
Poster # 05: Method of
Detection of Carcinoma in situ of the Breast According to
Histology.
Amy Trentham-Dietz
and Polly Newcomb
We
investigated patterns of detection of carcinoma in situ of the
breast (BCIS) among cases interviewed as part of two case-control studies.
Cases were Wisconsin residents, ages 45-74 years and interviewed by
telephone about 1 year after diagnosis. The interview included questions
regarding breast cancer risk factors as well as tumor detection
(mammography, self/ partner, clinical breast exam, or other). Study 1
included BCIS cases interviewed in 1987-1991 (N=322), while Study 2
included cases accrued in 1997-2001 (N=888). In Study 1, 69% of cases were
classified as ductal (ICD-O 85002), 22% as lobular (ICD-O 85202), 3% as
NOS (“not otherwise specified”, ICD-O 80102 & 81402), and 6% had other
histology codes. In Study 2, 70% of BCIS cases were ductal, 11% were
lobular, 3% were mixed ductal/ lobular (ICD-O 85212, 85222), 5% were NOS,
and 11% had other codes. In Study 1, 67% of ductal cases were detected
through mammography, as compared with 76% of lobular cases. In Study 2,
the percent of ductal cases detected through mammography increased to 87%,
while the percent of lobular cases detected through mammography was
essentially unchanged (77%). These data suggest that BCIS detection varies
according to histology. Detection patterns for BCIS also likely depends on
trends in characterization by pathologists, and the sensitivity and
prevalence of mammography. Further analysis will evaluate risk factors in
relation to detection of different histologic types of BCIS.
Poster # 06: Title:
Estimating the Contributions of Screening and Improved Treatments to the
Breast Cancer Mortality Reduction.
Dennis
G. Fryback,
Marjorie A. Rosenberg, Patrick L. Remington, Amy Trentham-Dietz, Natasha
K. Stout, Vivek Puttabuddhi, Vipat Kuruchittham
Breast cancer mortality has
declined in Wisconsin over the past decade, from 29 to 23/100K, and
similarly in the US. Increasing
use of screening mammography, improvements in treatment, and changing
demographics and risk factors have each been posited to explain part or
all of this decline but the separate effects of each are as of yet
unknown. To understand the relative contributions of screening, treatment
and risk factors in explaining trends in breast cancer incidence and
mortality in Wisconsin we developed a discrete-event simulation model of
the epidemiology of breast cancer. The simulation model uses theory and
data from many disciplines and sources to generate population level
statistics about breast cancer incidence and mortality in Wisconsin. By
varying model inputs, hypotheses about the contributions of screening
mammography and treatment to reducing breast cancer mortality can be
explored. Preliminary results exploring these contributions are presented
and indicate that screening and treatment improvements in the past 20
years have roughly independent and equal contributions to the observed
breast cancer mortality reduction.
Poster # 07: Workload,
Quality of Working Life and Quality of Care among Intensive care Nurses.
Ayse P. Gurses
and Pascale Carayon
Several studies show that high workload in
intensive care units (ICUs) may lead to compromised quality of care and
patient safety. In this study, we investigate the impact of workload on
quality of working life and quality of care among intensive care nurses.
We also identify the situational constraints in an ICU that increase nurse
workload unnecessarily and hinder performance. Situational constraints are
factors in a work environment that inhibit people from optimally
performing their jobs. A conceptual model that illustrates the
relationships among the variables of situational constraints, workload,
quality of working life, and quality of care is proposed.
Poster # 08: Shared
Decision-making and Inappropriate Antibiotic Use.
Elizabeth Cox
Introduction Inappropriate
antibiotic prescribing is commonplace and
can result from physician
misperceptions about patient expectations for antibiotics. Shared
decision-making (SDM) makes patient expectations explicit and thus may
decrease inappropriate antibiotic prescribing. SDM occurs when the
physician and patient share all decision-making steps simultaneously, with
a two-way exchange of information and preferences as well as agreement on
the decision to be implemented. Currently, no method exists to measure
SDM. We propose to develop a measure of SDM including a conceptual model
of the components of SDM (decision-making and physician/patient
partnership) as well as measures for these components. Subsequent work
will examine the association between SDM (as measured by the developed
instrument) and rates of inappropriate antibiotic prescribing.
Methods
Videotapes were collected for 100 children’s acute care visits to seven
family physicians and eight pediatricians selected to represent a breadth
of physician and patient demographics. Patient and parent demographics as
well as physician practice and personal characteristics were obtained by
self-report. Physician involvement in decision-making is coded with
OPTION instrument. Parent and child involvement in decision-making is
coded with OPTION instrument adapted to parent and child tasks in
decision-making. Partnership for physician-child and physician-parent is
coded with Roter Interaction Analysis System. Confirmatory factor
analysis will be performed for the four measures of decision-making and
the two measures of partnership.
Results
Data
collection and data coding with OPTION are 85% complete. Of participating
physicians, 31% are minority ethnic backgrounds as are 25% of the
children. Nearly 50% of children are less than 2 years of age.
Conclusions
Data
collected has a broad representation of physician and patient
demographics.
Future research
Completion of data collection and SDM instrument development is
anticipated in the next 9 months. A second study using the new
instrument will collect additional data
on 200 children’s
acute care visits and will examine the association between SDM and rates
of inappropriate antibiotic prescribing.
Poster # 09: The Subtle
Tyranny of Shared Decision-Making.
Kathryn Flynn,
Maureen Smith, and Elizabeth Cox
Implementation of shared
decision-making is increasing as a preferred model of the provider-patient
relationship . Shared decision-making is consistent with patient-centered
care and has the potential to reduce inappropriate care and cost.
However, while patients report a desire for greater information exchange
with their providers, they may not always want to participate in
decision-making. Wholesale adoption of a shared decision-making model is
potentially unethical and harmful, especially for older or sicker
patients. It is unclear when shared decision-making is appropriate.
Relationships may function effectively without shared decision-making
depending on whether the patient has a usual source of care and the length
and strength of that relationship. The objective of this project is to
understand how patient characteristics and the provider-patient
relationship affect the appropriateness of shared decision-making.
Poster # 10: Managed
Care Payments to Hospitals for Older Patients with Acute Stroke.
Maureen Smith
Managed care organizations (MCOs) have rapidly multiplied and become major
players in the health care marketplace. As a result, the use of different
payment mechanisms to control the costs of hospital care has increased and
concerns have been raised about the potential negative impact of these
financial arrangements on care decisions. However, hospital organizational
characteristics may be more important in determining care for acute stroke
(e.g., the hospital's level of specialization) and may confound the
relationship between MCO financial arrangements and care decisions. The
goal of this analysis is to identify how hospital organizational
characteristics relate to financial arrangements for stroke care. The
sample included persons age 65 or older who were hospitalized for acute
stroke during 1998, 1999 or 2000 and were enrolled in a large national MCO.
This resulted in a sample of 5,387 stroke patients and 480 hospitals in 11
regions across the country. Data included MCO enrollment and claims data
plus the American Hospital Association's annual survey of hospitals.
Hospital payment type was defined as per diem, per stay, percent, or
other. Hospital characteristics included the degree of specialization,
centralization, integration, as well as bed size, teaching status, and
formal written contracts with MCOs. Multinomial logistic regression was
used to examine the relationship between hospital characteristics and
payment type for each stroke patient while accounting for clustering of
patients within hospitals and adjusting for patient characteristics (age,
sex, previous stroke, comorbidities, year of hospitalization). These
results will be used in a subsequent analysis to examine the relative
importance of hospital characteristics and MCO financial arrangements to
care decisions for acute stroke.
Poster # 11: Current Practice Patterns in the Primary
Care Management of Patients with Hypertension in a Midwestern HMO.
MaryAnn Steiner, Nilay Shah,
Bryan Becker and Curt Johnson
This
analysis is part of a quality improvement project to determine the utility
of the medication possession ratio as a predictor of clinical outcomes in
patients (pts) with hypertension. This abstract describes key demographic
and clinica data and physician practice patterns observed in a
hypertensive pt population enrolled in a midwestern HMO. The study was a
retrospective records review of 708 adult pts treated for hypertension by
primary care providers (PCPs) from Jan-Dec 2000. 50.4% and 44.5% of pts
were treated by family practitioners and internists, respectively. Mean pt
weight was 93.2 kg. 72.9% of pts were >50 yrs old. Mean number of clinic
visits was 5.69. Recorded medical problem lists indicated 16% of pts had
diabetes (DM), 42.2% hyperlipidemia, 38.5% obesity, 13.3% depression, and
only 4.8% renal disease. Lab data were documented in 45.8% of charts. Scr
was recorded in 112 charts (15.8%). 110/112 of pts had Scr <2.0, however
106/112 had Clcr (MDRD) <90. 23% of DM pts had no recorded labs. 62% of DM
pts had recorded Scr. 28/29 DM pts with S albumin recorded had Clcr (MDRD)
<90. 12.4% of DM pts had S. albumin <3.5. 57.5% of DM patients had
hyperlipidemia. Over 80% of pts >age 50 had no Scr recorded. 48.3% of all
pts had last measured BP <140/90. 23.9% of DM pts had last measured BP
<130/80. These data indicate there are many opportunities for improved
care of these hypertensive pts. Laboratory data were recorded in <50% of
all pts and 77% in DM pts. Renal insufficiency was often undetected or not
documented. Scr was measured in 62% of DM pts and <20% of pts > age 50,
both groups at high risk for kidney disease. By calculating Clcr we
determined that reduced Clcr was extremely common in pts with Scr data
available. BP control was suboptimal, with <50% of all pts achieving
<140/90 mmHg. Less than 25% of DM pts achieved NKF Consensus target BP
goals. Opportunities exist for improved orchestration of clinical care and
follow-up for at risk CKD population. This analysis can be used to design
quality improvement programs for hypertension management and CKD detection
within this managed care organization.
Poster # 12: The Relationship Between Discontinuities in
Usual Source of Care and Health Care Access, Quality and Utilization.
Bartell JM, Smith MA
Background:
The increasing complexity and change in the U.S. health care system can
result in substantial disruptions in continuity of care. These
disruptions appear, in part, as either a lack of usual source of care or,
increasingly, a change in usual source of care. Little is known about the
impact of these discontinuities on individuals’ perceptions or use of
health care, particularly individuals who experience a change in usual
source of care. It is critical to understand the consequences of this
important source of instability in our health care system. The
purpose of this study is to determine how these different types of
discontinuities are related to perceived access, assessments of health
care, and utilization patterns.
Methods:
Study data was from the 1998-1999 Community Tracking Study
Household Survey, a nationally-representative,
telephone-administered survey of civilian, non-institutionalized
individuals. The main independent variable, usual source of care in
the last 12 months, was measured as no usual source of care, no change in
usual source of care, and change in usual source of care due to health
insurance, quality or other reason. Access to care was measured as
self-reported unmet needs or postponed care. Satisfaction, ratings of
last physician visit and trust in current physician were measured as the
percent of respondents giving the highest rating on a 5-point scale from
1=”very satisfied” to 5=”very dissatisfied.” Utilization information was
collected for the prior 12 months. All regression analyses accounted for
complex survey design and were adjusted for the possible confounders of
age, gender, marital status, rural/urban, race, education, family size,
income, health status, insurance type, managed care, preference for risk
and willingness to trade cost for provider choice.
Population Studied:
Our analysis sample included 48,551 adult
respondents.
Results:
Discontinuities in care were reported by 25% of all respondents: 14%
reported no usual source of care and 11% reported a change in usual source
of care in the last 12 months. Respondents with a change in usual source
reported the most access problems: 12.5% reported unmet needs compared
with 9.5% for no usual source and 6.6% for no change in usual source
(p<.0001) while 32.8% reported postponed care compared to 24.8% and 22.0%
(p<.0001). Those with a change in usual source or no usual source
reported lower satisfaction (51.1% and 51.0% versus 68.0%, p<.0001) and
lower ratings of their last health care visit (29.0% and 34.2% versus
39.8%, p<.0001) compared with those who had no change in usual source.
Utilization of both outpatient and inpatient care was highest among those
who changed their usual source. Within the subgroup of those who changed
their usual source, ambulatory care utilization was higher in those who
changed due to quality issues than in those who changed due to health
insurance.
Conclusions:
Our data suggest that discontinuities in
care may be important determinants of perceived access, utilization and
quality of care for patients.
Policy Implications:
Further analysis is needed to determine the mechanisms through which
discontinuities influence these outcomes.
Poster # 13: Practice Patterns in the Management of
Febrile Infants: Role of Timing of Residency Training and Timing of
Literature Publication.
Cox
ED,
Smith MA, Bartell JM
Work analyzes the contribution of relevant literature publication during
residency training as a predictor of physician practice patterns in the
management of febrile infants.
Poster # 14: Family Socioeconomic Status, Parental
Support, and Initiation of Smoking and Drinking in Adolescence and Young
Adulthood.
Vanessa H. Newburn
It is well-known that
socioeconomic status (SES) and social support structures are related to
health outcomes in adults, and that differential participation in
health-risk behaviors partially mediates these relationships. Less is
known about how SES and social support predict health-behavior choices in
children and adolescents, particularly whether these factors influence age
of initiation. The objectives of this study were to examine whether
family SES and quality of parental relationships during childhood are
related to age of initiation of smoking and drinking behaviors, and
whether parental support potentially mediates the relationship between SES
and choice of risky behaviors. Data on childhood experiences for 3,032
adults ages 25-64 who participated in the 1995-1996 National Survey of
Midlife Development in the U.S. (MIDUS) were analyzed using structural
equation modeling techniques. Consistent with our hypotheses, we found
that higher levels of parental support were significantly related to older
ages of initiation of smoking and drinking behaviors, and that higher
levels of family SES were related to increasing parental support, older
ages of initiation of smoking, and younger ages of initiation of drinking
behavior. Parental support was a stronger predictor of age of initiation
of risk-behaviors than family SES, and the direct effect of higher family
SES as a predictor of younger ages of drinking initiation was reduced when
parental support was modeled as a potential mediating mechanism. Because
health risk-taking behavior in childhood is related to risk-taking
behavior and excess preventable morbidity and mortality in adulthood,
understanding the determinants of health and lifestyle choices in
childhood should be an important public health priority.
Poster # 15: The role of
uncertainty in the breastfeeding decision for black and white women.
Kathryn
Flynn
Despite strong scientific
evidence of breastfeeding’s superiority over artificial infant feeding
methods in terms of health and cognitive benefits as well as decreased
financial burden, breastfeeding initiation rates remain well below
national guidelines, particularly for black women. Previous research has
not considered the influence of uncertainty in social and financial
support over the life course on the breastfeeding decision nor the
differences in the life course between black and white women. This
uncertainty may explain some of the apparent mismatch between the
recognized health benefit/cost ratio and the decision by many women not to
breastfeed. I analyze 703 black and 1,958 white women aged 18-45 from the
1995 National Survey of Family Growth (NSFG) who had a first birth between
1983-1995. I develop a binomial logistic regression model accounting for
complex sample design to estimate the log odds that a woman will
breastfeed her first child. Uncertainties in social or financial support
are important determinants of breastfeeding behavior over and above
current employment or marital status, though black and white women respond
to key events over the life course in markedly different ways.
Poster # 16: Cancer care capacity of health service
areas as a contextual determinant of mortality among women diagnosed with
breast cancer.
Indiana Strombom,
Maureen Smith, Pat Remington, Amy Trentham Dietz
Background:
Hospital-level and payer-level characteristics have been shown to have an
association with increased risk of individual mortality. Among cancer
patients, typically, these results arise from studies that did not include
either important individual level variables, like tumor characteristics at
diagnosis, or relevant community level variables, like socioeconomic
characteristics or detailed variables describing the cancer care-specific
context.
Purpose:
We hypothesize that community cancer care capacity has an independent
association with overall mortality among women diagnosed with breast
cancer, after adjustment for both community socioeconomic characteristics
and for a comprehensive set of individual-level prognostic factors. This
hypothesis will be tested within the framework of Andersen’s Behavioral
Model of Health Services Utilization.
Methods:
Telephone interview data of Wisconsin case participants in the Women’s
Health Study are used in this analysis. Women diagnosed with breast
cancer between 1988 and 1994 (n=7,179) were identified through the
Wisconsin cancer registry and invited to participate; they were passively
followed up until 1998 for mortality ascertainment. Individual-level data
were combined with data from the American Hospital Association, 1990
census, physician licensing and hospital data, aggregated at the level of
Health Service Area of residence (HSA), the contextual unit of analysis.
HSA’s are defined by zip code clusters characterized by the Dartmouth
Atlas project. Community cancer care capacity variables included hospital
diagnostic and therapeutic oncology services, cancer program approval
status and other affiliations, hospital admissions volume, and the density
of primary care physicians and oncology specialists. We analyzed the
individual and contextual data using multivariate logistic regression with
robust Huber-White standard error estimators, adjusting for socioeconomic,
demographic and behavioral factors, as well as tumor characteristics and
reported health care use.
Results:
Twenty two percent of the participants were deceased by the end of
follow-up. Deceased participants were more likely to have lower levels of
education, to have been diagnosed at non-localized stages of breast cancer
and to have detected the tumor themselves, rather than through mammography
screening. After multivariate adjustment, increasing density of primary
care physicians was associated with lower mortality and exhibited a
dose-response relationship that was statistically significant in the upper
two quintiles. The volume of hospital admissions and the existence of
women’s health programs in the area were associated with individual
overall mortality but no clear trends were evident. Notably, the density
of oncology-related specialists and cancer program approval status were
not significant in the adjusted model.
Conclusions and policy implications:
After adjusting for individual level factors and HSA-level socioeconomic
status, the cancer care services offered within the HSA of residence for
women diagnosed with breast cancer had an independent relationship with
individual overall mortality. Analyses of cancer care capacity may
elucidate important effects of our health care system on individual
outcomes for women with breast cancer, over and above individual-level
risks, and may provide support to possible pathways currently under
study.
Poster # 17: Relating
caregiver burden to communicative and cognitive deficits in stroke
patients.
Melissa Nelson
and Maureen Smith
Caregivers of stroke patients have unacceptably high levels of burden,
leading to physical, mental, and social health problems. Little is known
about whether impairment in a stroke patient’s communication and cognitive
abilities interact with other stroke deficits to heighten caregiver
burden. The objective of this research is to examine whether stroke
patients’ communication and cognitive abilities modify the impact of their
physical, social, and emotional deficits on caregiver burden. We
hypothesize that impaired communicative or cognitive deficits in a stroke
patient will increase perceived burden more than objective burden for
patients of equivalent physical, social, and emotional deficits. Data
were obtained from approximately 400 stroke patients and their caregivers
enrolled in a large HMO in the Minneapolis/St. Paul metropolitan area.
In-person and telephone interviews of patients or their proxies and their
caregivers were conducted 2 months after discharge from the hospital for
their stroke. Perceived caregiver burden was measured using the Sense of
Competence Questionnaire, which measures satisfaction with the impaired
person, one’s own performance, and the impact on the caregiver’s personal
life. Objective caregiver burden was measured as number of hours per week
devoted to caring for the stroke patient. Stroke patients’ deficits were
measured with the stroke adapted Sickness Impact Profile. Multivariate
regression will be used to relate caregiver burden to stroke patient
deficits. Understanding the relationship between stroke patients’
communicative and cognitive deficits and caregiver burden will allow us to
target caregivers in need of additional support.
Poster # 18: A Model for Assessing Local Tobacco Control
Coalitions.
Barbara Hill
and D. Paul Moberg
Problem/Objective - To better understand
the role of local coalitions in a comprehensive tobacco control program,
it is necessary to develop a strategy to systematically collect, analyze
and disseminate meaningful data regarding the coalitions. The information
obtained can be used for program evaluation, program enhancement and to
inform technical assistance efforts. Ideally, this information should be
collected with a minimum reporting burden to coalitions and eventually
linked to community outcomes.
Methods – The Coalition
Reporting System (CRS) was designed to gather information about coalition
structure and operation. The CRS consists of information from the
coalition characteristics survey, activities survey and from site visits.
Information from the CRS is used to better understand how coalitions are
structured and organized to accomplish their goal of reducing tobacco use
in their community, to compare characteristics, activities and challenges
of coalitions with those of other coalitions in order to facilitate better
networking and to assess training and technical assistance needs.
Results – Only 18% of coalition
coordinators work full time, most coalition members are professionals
representing agencies and 80% of coalitions meet monthly and meet during
the day. Early site visit results prove helpful in providing a fuller
picture of coalition functioning. Survey results have been used to inform
training and technical assistance to coalitions.
Discussion – A strategy was developed to
assess local efforts with a minimum burden to coalitions. This strategy
included collecting and analyzing information using surveys and site
visits. The approach provides a multi-dimensional picture of coalitions
that can assist with accountability, program development, and technical
assistance.
Poster # 19: Evaluation of a Youth-Focused Anti-Smoking
Media Campaign.
Patrick Remington,
Ann Olen, Ann Christiansen, Catheryn Brue, Eden Schafer
Learner Outcome:
Participants will be able to describe the effects of a modest media
campaign on changing the tobacco-related awareness, attitudes, and
behaviors of youth ages 11-14.
Objective:
The Thomas T. Melvin Youth Tobacco Prevention and Education Program in
Wisconsin provided $1 million per year to fund a media campaign to lower
rates of tobacco use among youth 11-14. Due to limited funding, the
campaign was limited to one region. The evaluation objective was to
determine if the campaign was reaching its target audience, and
influencing their attitudes and behaviors regarding tobacco. The campaign
messages were 1) smoking causes wrinkles, 2) smoking makes you tired, and
3) tobacco industry lies.
Methods:
We conducted in-school surveys in public middle schools of the funded
districts, and in control schools. The survey was administered in the
fall of 1998, as the integrated media campaign began. The follow-up
survey was conducted in spring of 1999. The effectiveness of the media
was determined by calculating prevalence rate ratios (RRs) in the
intervention versus control communities.
Results:
Awareness of the media campaign was very high compared to the control
communities and remained high among youth in intervention communities.
Attitudes regarding tobacco use were more positive in intervention than in
control communities. Compared to the control communities, campaign
awareness was greatest for causing wrinkles (RR=9.9, 95% CI 7.8-12.6),
industry lying (RR=6.2, 95% CI 5.1-7.6), and making tired (RR=3.8, 95% CI
3.3-4.3). Attitudes were greatest for causing wrinkles (RR=2.1, 95% CI
1.8-2.4), but not significantly different for making tired (RR=1.2) and
industry lying (RR=1.1).
Discussion:
This modest campaign had effects on awareness and attitudes, especially
those related to wrinkles. The lack of effect on behavior is expected,
due to limited funding and short duration.
Poster # 20: Interactive Health Communications to Help Prevent
Relapse to Smoking.
Mark E.
Zehner, Stevens S. Smith, Jose
A. Valdez, David H. Gustafson, Michael C. Fiore, Devayani S. Pophali, and
Timothy B. Baker
Smoking cessation programs
often achieve 70% initial abstinence, but produce only 10-30% long-term
success. A key to efficacious relapse prevention may be to sustain
treatment over the period of significant relapse risk. One means to
delivering such time-sensitive treatment is via home computer. For
example, during a course of smoking cessation treatment, computer-based
interactive health communication (IHC) applications can provide continuous
information and support, identify and intervene with users at risk of
relapse, and allow them to self-tailor frequency and dose of intervention
during period of need. This study reports usage data from an on-going
clinical trial evaluating the internet-based Comprehensive Health
Enhancement Support System (CHESS) smoking relapse prevention program.
Smokers motivated to quit were randomized to either: (1) brief treatment
(BT) of nine weeks open-label bupropion SR and three behavioral counseling
sessions or (2) BT and 12 weeks free, in-home access to the CHESS program
(BT-CHESS). CHESS tracked subject (N=62) usage throughout the
intervention period. 90% of participants used the CHESS program during the
first week post-quit, 48% during week 9 (concurrent with end of
pharmacotherapy), and 42% during week 12. Participants accessed CHESS an
average of 6.1 and 5.0 times per week for weeks 1 and 2, respectively and
between 3.5 and 4.4 times per week subsequently. These preliminary data
indicate smokers are using the CHESS program to assist their smoking
cessation effort. This type of IHC program has benefits for users
(continuous access, etc.) and health care systems (decreased personnel
time, etc.).
Poster #21: Population-Based Assessment: Use and Cost of
a New Benefit for Smoking Cessation.
Marguerite Burns,
Marjorie Rosenberg, and Michael Fiore
Uncertainty about the use and cost of
insurance coverage for smoking cessation treatment (SCT) is a barrier for
health care purchasers’ adoption of such coverage.
Presented here are the results from the first year, of a three-year
observational study designed to reduce this uncertainty for public health
care purchasers.
In January 2001, the State of Wisconsin
introduced SCT insurance coverage for its ~ 200,000 employees, retirees
and dependents. Employee self-report data were collected to measure
smoking status, benefit awareness and use using the Consumer Assessment of
Health Plan Survey in Spring 2002. Employee/retiree households were
randomly selected and stratified by health plan. A 61% response rate was
achieved, and data from ~ 6,000 respondents were available for analysis.
2001 pharmaceutical claims data were collected from 12 of 17
health insurance carriers.
14.2% of respondents reported smoking
every day or some days. 28% of smokers reported awareness of the SCT
benefit. 14.2% of smokers, or 2% of all
state employee health plan members, reported using the pharmaceutical
benefit “during the past 12 months.” Claims data suggest that 9.2% of
smokers and 1.3% of all members used the benefit. The benefit’s PMPM cost
was $0.15 during its first year.
The benefit use rate and $0.15 PMPM cost
are consistent with those reported in the literature. Cost is lower than
the State of Wisconsin’s estimate, $0.24 PMPM. It suggests that the
addition of an evidence-based SCT benefit results in modest additional
costs for the purchaser.
Poster # 22 Economic
Evaluation in Clinic Trials: Calculating Hospitalization Costs from
Clinic Trial Data.
Henry Henk
OBJECTIVES: Economic
evaluation is increasing common in clinical trials. Often, individuals’
health care costs are not observed in these trials, rather health care
cost estimates are often calculated from observed resource use and unit
prices estimated from other sources. The level of detail collected in
clinical trials varies and determines which resource costing methods can
be used. Using hospital inpatient data, this study compares three
resource costing methodologies that utilize varying levels of information
about hospitalizations. METHODS: As part of project TrEAT, an
alcohol-related intervention study, hospital primary discharge data were
collected for HMO patients. Data from HCUP National Inpatient Sample (NIS)
2000 are used for estimating unit prices. Three resource costing methods
are applied: (1) a per day unit price over all hospitalizations; (2) a
unit price per day for each DRG; (3) a unit price per day for each primary
ICD-9-CM discharge diagnosis. Inpatient costs are calculated as the
product of these unit prices and the observed inpatient days. Inpatient
costs for the intervention and control groups are compared. RESULTS: For
the 1-year period following study enrollment, method 1 yields control and
intervention group averages of $969($5472) and $491($2916), respectively.
Method 2 produced mean and standard deviations of costs that were
approximately twice as large, $1912($11391) and $1086($7509) for the
control and intervention group, respectively. Method 3 produced mean and
standard deviations larger than method 1 but smaller than method 2:
$1382($7481) and $891($5966) for the control and intervention group,
respectively. Differences between the intervention arms are greater using
method 2, ($826 versus $478 and $491). CONCLUSIONS: The level of
resource use detail can affect the results of economic evaluation of
clinical trials. Costing hospitalizations using DRG level data resulted
in larger differences between intervention arms than methods using
ICD-9-CM level data or a fixed per diem amount.
Poster # 23: A
measurement error model with a Poisson distributed covariate.
Liang Li,
Mari Palta, Jun Shao
We study a
linear model in which one of the covariates is measured with error. The
surrogate for this covariate is the event count in unit time and has a
Poisson distribution with the unobserved covariate value as the rate. We
show that ignoring this measurement error leads to inconsistent estimators
of the regression coefficients and propose a set of unbiased estimating
equations to correct the bias. The method is computationally simple and
does not require using supplemental data as is often the case in
measurement error analysis. No distributional assumption is made for the
unobserved covariate. The model and method can be extended to accommodate
longitudinal data in a straightforward way. The proposed method is
illustrated with an example from the Wisconsin Sleep Cohort Study.
Poster # 24: A Bayesian Cost-Effectiveness Analysis of
Implantable Cardioverter Defibrillators (ICD).
MA
Rosenberg,
DG Fryback, GE Flood, B O'Brien, A Johnson-Masotti
Background & Objective.
Previous clinical trials have examined the efficacy of treatment with ICD
versus conventional anti-arrhythmic drug therapy to prevent death in
patients at risk of ventricular fibrillation or sustained ventricular
tachycardia. The Canadian Implantable Defibrillator Study (CIDS)
randomized over 600 patients to ICD versus amiodarone and found a small
benefit of ICDs compared to amiodarone in reducing mortality in patients
observed for 6.5 years. After suitable IRB approval, we obtained the data
for the first 430 patients randomized in CIDS for whom economic cost data
were also collected, and used Bayesian parametric models for survival and
cost to compute the cost per life-year saved of ICD versus amiodarone
treatment projected to a 10 year time horizon.
Methods. Costs were modeled parametrically for each calendar
quarter using multiple two-part models and projected to a 10 year time
horizon. Survival was modeled using a Weibull distribution. The models
were estimated simultaneously allowing co-varying parameters. Statistical
uncertainty in estimated cost-per-life-year saved is assessed by drawing
20,000 samples of incremental costs and incremental life years saved based
on the joint posterior distribution of estimated parameters.
Results. The median incremental cost-per-life-year saved is
$133,171 and the cost-acceptability curve indicates less than 15% chance
the incremental cost-per-life-year saved by using ICDs versus amiodarone
falls below $50,000.
Conclusions. We conclude under conventional standards the use of
ICD is not cost-effective compared to amiodarone treatment for these
patients generally. Future work may examine specific subsets of patients
where previous exploratory analyses have indicate potentially greater
benefits of ICDs. Bayesian methods provide great flexibility in
estimation of complex cost and survival models and application of results
to assess uncertainty in conclusions about cost effectiveness.
Poster # 25: Correlates of reported breathing pauses in
Hmong residents of Wisconsin.
P Vue,
T Young, D Austin, L Finn
Introduction: Hmong young and middle-aged
adults, like other South-east Asian groups, are at high risk for Sudden
Unexplained Nocturnal Death Syndrome ( SUNDS), a condition of unknown
etiology. Hmong culture links SUNDS with the experience of tsog chuam,
a nocturnal attack by a spirit that paralyzes the victim, crushes the
chest and takes the breath away. The experience is suggestive of
REM-related sleep disorders and sleep apnea, but the characteristics of
sleep problems in SE Asians is unknown. The aim of our study was to
investigate the correlates of sleep disorder symptoms in Hmong residents
of Wisconsin.
Methods: A sample was identified from a
statewide Hmong directory and sent a questionnaire on sleep habits,
problems, and other factors ( in English and Hmong) by mail. Logistic
regression was used to model correlates of reporting breathing pauses
during sleep a few nights or more/month.
Results: Thus far, 455 questionnaires
were successfully completed. The prevalence of habital snoring (22%) was
similar to that reported in other adult populations, but the prevalences
of frequent ( several nights/month) restless sleep (26%), cateplexy (10%),
nightmares ( 15.4%) and potentially REM-related symptoms including sleep
paralysis on wakening (7%)or sleep onset (8%), were high. The strongest
predictors of reported breathing pauses (a few times/month or more vs
rarely or never) were the experience of a night-spirit crushing the chest
(odds ratio=6.4, p<.001) and sleep paralysis ( odds ratio=5.5, p=.0002).
Males were more likely to report breathing pauses, but the odds ratio was
not statistically significant.
Conclusions: The data, although
preliminary, suggest the sleep characteristics of Hmong differ from non-SE
Asian populations, and that the tsog chuam experience often linked with
SUNDS may be related to sleep-related breathing pauses.
Poster # 26: Changes in off-CPAP Apnea-Hypopnea Index (AHI)
over time in women CPAP users.
Peterson AG,
Young T, Finn LA, Austin, D
Introduction:
Some studies of sleep apnea patients have shown that off-CPAP AHIs of CPAP
users after long-term CPAP therapy are reduced, suggesting a sustained
improvement in sleep-disordered breathing. However, others have shown no
significant difference in AHI from pre-treatment to a single
post-treatment off-CPAP Nocturnal Polysomnogram (NPSG). One explanation
for this discrepancy is that a “first night effect” exists where the
pre-treatment AHI is artifactually high because of disrupted sleep,
leading to reduced values at follow-up. In addition, a single baseline
or follow-up measurement may fail to reflect sleep-disordered breathing
status, due to intra-subject variation.
Our aim for this analysis was to use
multiple measures of AHI over time to examine the effect of CPAP use on
off-CPAP AHIs. We also examined weight changes over time, theorizing that
CPAP users might have health benefits leading to weight loss as part of a
natural reinforcement cycle.
Methods:
Our study population consists of 186 female participants in the Women’s
Sleep Study (WSS) a subset of the Wisconsin Sleep Cohort Study (WSCS),
contributing up to seven in-home sleep studies each. For each study, CPAP
users are asked to sleep without CPAP. All NPSGs were performed using
Compumedics, Inc. P-Series portable sleep recording equipment, at
approximately 6-month intervals. Women who have used CPAP (N=17) have
contributed 92 sleep studies and the remaining 169 Non-users have
contributed 795 sleep studies for this analysis.
Results:
We found that the average AHIs of both the CPAP Users (off-CPAP) and
Non-users did not vary significantly over time. The average change in AHI
over 6-month intervals among CPAP users was –0.12 events/hour of sleep
compared to an average change in AHI among Non-users of 0.11 (NS). The
average change in weight over 6-month intervals among CPAP users (0.61 Kg)
versus Non-users (-0.056 Kg) was marginally statistically significantly
different (p=0.07).
Conclusions:
CPAP use did not result in significantly
lower AHIs over time. The use of CPAP did not appear to be associated
with weight loss. In fact, the CPAP users were slightly more likely to
gain weight during each 6-month interval than Non-users.
Poster # 27: Role of Sleep Complaints in Initiation of
Hormone Replacement Therapy.
T.
Young, L Finn, A
Peterson
Introduction:
Observational studies have not consistently shown that objective or
subjective sleep is better in menopausal women who use hormone replacement
therapy (HRT), compared with those who do not. One explanation for the
discrepancy is that women with poor quality sleep, compared to those
without sleep problems, are more likely to seek HRT. Our aim was to
investigate sleep problem frequency prior to initiation of HRT with
prospective data.
Methods:
Sleep problems, use of HRT, and other data have been recorded daily by a
sample of midlife women enrolled in the Wisconsin Sleep Cohort Study.
Out of the women enrolled in this study,
34 began HRT after entering the study. These women (users) were matched
by age with 34 women who did not begin HRT (nonusers). The 3-month period
before HRT initiation for the users and the same time period for her
matched control were examined for the occurrence of the following sleep
complaints: restless sleep, trouble getting to sleep, trouble staying
asleep and nightmares. Two outcomes were computed for these sleep
complaints: 1) the average number of days with sleep complaints in the 3
month exposure period and 2) whether or not at least two months with 5 or
more days of trouble getting to or maintaining sleep or restless sleep was
reported.
We used t-tests to compare the
mean number of days of each sleep complaint and Chi-square tests to
compare whether or not women reported at least two months with 5 or more
days of sleep complaints between those who did and did not initiate HRT.
Results:
Women did not differ in their average number of days with sleep complaints
or the percentage reporting at least two months with 5 or more days of
sleep complaints in the previous three months given their HRT initiation
status. On average (se), women who initiated HRT reported 0.7 (1.1), 2.6
(0.9), 0.6 (0.4), and 1.7 (0.4) days of trouble getting to sleep, trouble
maintaining sleep, nightmares, and restless sleep, respectively. This is
compared to women who did not initiate HRT who reported 1.0 (1.8), 3.0
(1.1), 0.4 (0.2), and 1.9 (0.5) days of trouble getting to sleep, trouble
maintaining sleep, nightmares, and restless sleep, respectively.
Additionally, 30% of women who initiated HRT reported at least two months
with 5 days or more of trouble getting to or maintaining sleep or restless
sleep out of the previous 3 months. This is no different from women who
did not initiate HRT, of whom 29.4% reported at least two months with 5 or
more days of trouble getting to or maintaining sleep or restless sleep out
of the previous 3 months.
Conclusions:
Our data did not suggest a role of sleep complaints in the initiation of
hormone replacement therapy in midlife women.
Poster # 28: Sleep
Disordered Breathing and Type II Diabetes Mellitus.
K.J.Reichmuth, D.Austin,
T.Young, and J.Skatrud.
Introduction: We investigated
the increased risk of diabetes mellitus (DM) associated with
sleep-disordered breathing (SDB), using cross-sectional and longitudinal
data from a population-based sample of 779 men and 664 women enrolled in
the Wisconsin Sleep Cohort Study.
Methods:
Health interviews and polysomnography were conducted at baseline and up to
2 follow-up studies. SDB was measured by the number of apnea and hypopnea
events per hour of sleep (AHI) and diabetes status was determined by
self-reported diagnosed diabetes. Subjects with type 1 DM or unknown type
were excluded. Adjusted odds ratios (OR) of prevalent DM for 2 levels of
SDB were estimated by multiple logistic regression using generalized
estimating equations. To determine incidence rates by SDB status, subjects
with DM on their first visit or with a single visit were excluded from the
data. The data were stratified by BMI at baseline. Fisher exact tests were
used to test significance.
Results:
|
Incidence of Newly Diagnosed DM
Stratified by BMI |
|
Apnea-Hypopnea
Index |
|
< 5 |
5-15 |
≥15 |
|
BMI < 30 kg/M2, n
= 493, p = 0.026 |
(n) % |
(2) 0.46 % |
(1) 2.27 % |
(1) 9.09 % |
|
BMI  30
kg/M2, n = 309, p = 0.049 |
(n) % |
(11) 5.29 % |
(4) 6.90 % |
(7) 16.28 % |
The odds of DM for subjects with AHI ≥ 15
vs. <5 was significantly increased, even after adjustment for age, sex,
and body habitus (OR = 2.08, 95% CI = 1.09 – 3.95).
Conclusion:
Our preliminary prospective data show a statistically significant increase
in type II DM in people with evidence of SDB compared to those without SDB,
regardless of obesity status. This raises the possibility that SDB may be
an independent risk factor for type II DM.
Poster # 29: Self-Rated Health in Young Persons with
Type 1 Diabetes in Relation to Risk Factors from a Longitudinal Study.
Guan-Hua Huang
It has become accepted that Type 1
diabetes treatment should both alleviate the physical complications of
disease and improve overall quality of life (QOL). The Wisconsin Diabetes
Registry Study is a population-based cohort study that has followed
individuals longitudinally from Type 1 diabetes diagnosis since May 1,
1987. This study provides us a unique opportunity to examine QOL in
children, adolescents and young adults with Type 1 diabetes and its
relationship with both socio-demographic and clinical risk factors. In
the report, we use ordinal-scaled self-rated global health as a measure
of QOL and propose a random-effects model to draw inferences on
individuals regarding the relationship of longitudinally measured QOL with
multiple risk factors. Results show that male gender, higher parental
socioeconomic level, younger age at diagnosis of diabetes, shorter
diabetes duration, no
hospitalization in the preceding 6 months, lower glycosylated hemoglobin
level, and questionnaires answered by other than the subject are
independently associated with reporting better health. We also found that
individuals varied in their reported health even after adjusting for all
identified risk factors. This variability reflects that individuals have
different perceptions about health, and the cross-sectional study that
combines different health perceptions across individuals could bias the
result.
Poster # 30 Trends in
blood pressure among children and adults during the first 10 years of type
1 diabetes.
Allen C,
Zhang H, Palta M, LeCaire T, Huang G-H, DAlessio DJ for the Wisconsin
Diabetes Registry Project. Madison, WI.
Individuals with type 1 diabetes are at
high risk of developing hypertension. However, little is known about
blood pressure changes in the first 10 years of diabetes when glycemic
control becomes well established and pathological changes may be
initiated. A population-based cohort of children and adults newly
diagnosed with type 1 diabetes was followed for up to 10 years to examine
if blood pressure levels differed from the general population and to
determine the effects of diabetes duration and glycemic control on blood
pressure. Demographic and family health history information were
collected at enrollment. Blood pressure was measured at the 4-6 months,
4, 7 and 9 years after diagnosis. Participants were asked to provide
blood samples for glycosylated hemoglobin testing 3 times each year or at
regular clinic visits. Age-specific mean systolic and diastolic blood
pressures (DBP) were similar between diabetic individuals and a general
population sample (NHANES II). However, regression analysis of the
diabetes cohort data showed that diabetes duration and glycosylated
hemoglobin were significantly related to DBP after controlling for age,
sex, race, height and body mass index. The relationship between duration
and DBP changed with age. Systolic blood pressure was significantly
related to duration, but not significant related to glycemic control.
These data suggest that there are not
detectable blood pressure differences between diabetic individuals and the
general population in the first 10 year of diabetes, but that diabetes
does effect diastolic blood pressure via duration and hyperglycemia. The
duration effect is not uniform for all ages.
Poster # 31: Prospective
Study of Blood Pressure and Urinary Albumin Excretion in a Diabetes
Incident Cohort.
Allen C,
Palta M, LeCaire T, Brazy P, Huang G-H, D’Alessio DJ for the Wisconsin
Diabetes Registry Project.
The
temporal relationship of hypertension and microalbuminuria, the earliest
recognized marker for diabetic kidney disease, is not clear. Describing
the relationship of blood pressure and urinary albumin excretion in the
beginning of type 1 diabetes is critical for understanding the natural
history of diabetic kidney disease. Blood pressure, urinary albumin
excretion (UAE) rates, and blood glucose control measurements as well as
sociodemographic, and medication data were collected from a
population-based incident cohort of children, adolescents and young adults
with type 1 diabetes. Overall levels (means) and trends (slopes) for
blood pressure and UAE in the first 5 years of diabetes were compared with
the corresponding quantities at 6-10 years after diabetes diagnosis.
Persistent microalbuminuria by 10 years duration was low in this
population (3%). Early UAE (µg/min) mean significantly predicted late
mean diastolic blood pressure (adjusted = 3.44, 95% confidence interval:
1.41, 5.47). UAE slope did not predict blood pressure means or slopes.
Neither systolic or diastolic means or slopes significantly predicted UAE
means or slopes. Our finding provides further support for the hypothesis
that changes in the kidney occur in the first few years of type 1 diabetes
and that these changes predict increase in diastolic blood pressure.
Poster # 32 Pulmonary Function Testing in the Pediatric
Age Group.
Hana Said,
For the Newborn Lung Project
We will demonstrate pulmonary function
testing using different methods.
- Impulse
oscillation is a relatively new method for testing lung mechanics
(pressure and flow) that requires minimal subject cooperation. Fourier
analysis is used to extract components of respiratory resistance
and reactance.
- Spirometry is
considered the most fundamental test of lung function (flow and volume)
and is an effort-dependent test. Key parameters are FEV1
(expiratory volume in 1 second) and FVC (forced expiratory
volume). For example, asthma tends to lower the ratio of these two.
Peak flow meters are portable
electronic devices that follow the same concept as spirometry and are used
to monitor respiratory function in asthmatic patients.
Poster # 33: Why you
can't sleep well in the mountains.
Andy Lovering, Jimmy Fraigne, Witali
Dunin-Barkowski, Ed Vidruk & John Orem.
The decreased availability of oxygen
(hypoxemia) at high altitude leads to an increase in breathing
(hyperventilation) in an effort to compensate for this decrease. This
hyperventilation leads to a fall in blood levels of carbon dioxide (hypocapnia
). As a consequence, at altitude, one simultaneously experiences
both hypoxemia and hypocapnia. NREM and REM sleep are both reportedly
disrupted at high altitude. Because administration of supplemental oxygen
prevents this disruption, it is commonly believed that hypoxemia is the
crucial factor responsible for poor sleep quality at altitude. Overlooked
is the fact that not only does oxygen administration prevent hypoxemia,
but it also prevents hyperventilation and consequently, hypocapnia. In
view of this, we asked, ”is it the hypoxemia or hypocapnia which causes
sleep disruption at altitude?” Our study of 4 cats indicates that
hypocapnia causes a significant disruption of REM sleep under both
hypoxemic conditions as well under normal levels of
oxygenation.
