Abstracts Vol. 1 [International Conference on AIDS (11th: 1996: Vancouver, Canada)]

Monday, July 8, 1996

Mo.C.204 - Mo.C.21 I

Lessons Learned: Improving the quality and specificity of data in AI Sae (.t istries is
attainable through modest initiatives t hat utilize sound epidemiolo ti: t,.'qes and analysis.
Robert J. Rybicki, CDPFI, 37 35 S tate Street, Chicago, IL 60604l i A
Telephone: [phone] Fax.[phone]
Mo.C.204
EFFECT OF THE 1993 AIDS CASE DEFINITION CHANGE ON REPORT DELAY IN
LOS ANGELES COUNTY
Am Rock Woh, Paul Sin on2, ioron Bunchit IDepart ment:,. i i, rrvices Los
Angeles. CA. 2Centers for 1)isease Conlrl and Prevention, Atlari,, A
Objective:To describe the impact of the 1993 AIDS case definit:,, c> irc,, on the timeiness of AIDS case reporting in Los Angeles County (LAC).
Methods: Report delay (RD) was defined as the interval from d.i i f AIDS diaginosis to
date of case report. RD wr ilyed rye eport, race, seo,,. f HIV exposure,
facility of diagnosis, nd case definition category (pre-1993 definii.- s e-panded 1993 criteria) for all AIDS cases reported rn I AC from January 1985 thro, 1 De sember 1995
(nr30,548).The Wilcoxon two sample test was used to compare ' iian R)D across variables of interest.
Results: Median RD wa 4 months for all cases and steadily increased irom I month in 1985
to a peak of 8 months in 1993, decreasing to 4 months and 3 months n 1994 and 1995,
respectivelyThis trend was paralleled by the trend in the number of AiDS ca es reported
annually Median RD did no t vary by ra ce, sex and HIV exposure mode but was slightly
shorter for cases diagnosed at Ciounty (3 months) vs. non Counly facilities (4 mronths;
p<0.0 I). For cases reported 1993 95, median RD did not vary for cases meeting the pre1993 definition (5 months) vs. cases meeting only the expanded 199, ceria' (5 months and
6 months for irmrmunologic [Is Dl count _200cells/pl] and clinical cases, respectively).
Conclusions: Temporal vrii tion in RD) should be considered when evaluating local trends
in AIDS surveillance data and developing AIDS case projections, as s done nationallyThe
increased RD observed in 1993 most likely reflects the increased ioriloai and the
reporting of prevalent cases associated with the expansion of the case definition that year
The decline in RD in 1993-95 does not appear to be attributabe to more timely reporting of cases that were reported based solely on the AIDS-defirn i, crter i ntroduced in
1993.This may in part reflect the fact that laboratory reporting of rimmunologic cases is
not required in LAC.
Amy Rock Wohl, Ph.D., 600 S. Commonwealth Ave., Suite 805, Los Ange(s,. A 90005.
Telephone: (2 t13) 351 8 140. Fax (2 ) 4187 4683
Mo.C.205
AIDS UNDER-REPORTING MAY DISTORT THE EPIDEMIC:THE QUEBEC EXPERIENCE
Remis, Rober t S, Meunier L,Vandal AC Palrer RWtI, Desrochers 13. SilAIDS Prevention
and Control Pr cgr SiReional Pubi iHealth Department, Montreal. uebec, Canada
Objectives: To determine te ipact of re trospective active cas-id,1 ' on tre nds in AIDS
incidence in Quebec.
Methods: Based on istudy of the completeness of AIDS reportng in 19941, we identified
acute carewrdhospitals with the largest number of previously unreported AIDS cases. Using
predictive ICD 9 codes, we identified,and validated the medical files of persons treated at
these hospitals from 21987 to 1994. AIDS cases so identified were compared with cases
reported through the passive surveillance system. AIDS incidence for eici risk category
adjusted for report ing idelays was compared with and without the cases iderntified in the
active surveillance proiect.
Results: Completeness of AID)S reporting in
Quebec was 75% (total AIDS cases to end
1995: 3943). Five hospitals with ia high level of YarlyAIDS incidence among PHEC
413 with alivse__
under-reporting and a s ubstantial AIDS case suvellan _load were identified, all in Montreal. As a result 30 -
of case-finding to end 1995 in 3 hosp itals an 20 without adic
additional 285 AIDS cases were identified, 133 sutvoSiateh
cases had been previously reported by these 10
hospitals.The distribution of risk category for 0
the "active" cases differed from t hat of passive 8 0 85 90 95
cases: 27.7% were among persons born in Year of diagnosis
HIV endemic countries (PHE(:) compared to
9.4% for the passive cases (p 10 '); for injection
drug users, it was /7.0% versus 1 9% (p 0.02). Relative AIDS incadce inii' Montreal compared to the rest of Quet)ec increased from 8.7 to 9.2 with the iaclu,'ion oft the active
cases. Finally, the shatpe of the AIDS incidence curve among PHEc.changed with the addition of the active cases (see graph).
Conclusions: In our study we observed differential underreporting, o AIDS tases, in particular among persons born in I IV endemic countries and among inic ti)n drug users. Under
reporting of AIDS my not only underestimate the magnitude of th, FHIV epidemic, it may
also distort the tl-ends In AI)S randl HIV ncidence.
Dr: R.S. Remis, I6 I6 Ren l oots'te ouost Suie 300A. Montreal P i anida i3H I P8
Tel: [phone] Fax: (5) 93D 502 E mait: md22nrusic i
Mo.C.206
RECORD LINKAGE ESTIMATES OF REPORTING COMPLETENESS FOR AIDS IN
CANADA INDICATE INCREASING RATES OF INCOMPLETE REPORTING
Rieketts Maui-a h itE h 1, I to P''Fri Ht"tool t. ' rMR**,
Sutherland FD. 'Laborato (ente for Disease Control, Canalt 'ives ty of Ottawa,
Canada ***Statistic aad,(),w. nd.
Objectives: To deter mine the xiri of reportin sompletene, 'D i Canada.
Methods: A probabst -eord lilage based on common field Iris date of birth,
dates of death, ty and province of residence, diagjosis at time f.lth) was conducted
between the Natiorna AIDS (,se Reportin Surveillance Systen (ACRSS) and Statistics
Canada Meortahty Dat Base (CMDB) for all causes of death to the end I 1'92 and to the
end of Sept. 1994. Captue re' c1)itre was ued to calculate reportirg compileeness.
Historical re modelling if I-eportiig co mpleteness was used to Ic, iate the itnpact of
reporting delay Tests for independet e were applied.

Results: Overall reporting completeness (RC) was found to be 86%. RC varied by province
from 60 - 95% and by sex from 87%
95 (men) to 77% (women). Long reporting,_ __.,""delays contributed to underreporting by
0,... " - 9 over 10% early in the epidemic. RC by;7 Oct. 994 decreased from 94% in those
85-'.t9,deceased prior to 1986 to 86o in those
r0 deceased prior to 1993.Thirs may be due
I to reporting delays, but historical recon
structions of reporting delay, created by
70I r.... truncating the ACRSS data then succes
710. 13,ep19 22 25 28 31 3 sively adding new quarters indicate flatten
Date of hReCrpt of Report at ACRSS...r....o...e......,ing at lower levels of RC rates over time.
Evidence of dependence may indicate the
overa ll RC estimate is high. Dependence has likely increased n recent years, wh ichwould systema ically overestimate RC over time and may conceal a stronger trend than indicated here.
Conclusions: Overall RC is high, but this study indicates decreasing reporting completeness.
Delayed reports may eventually increase the final RC.
Maura N. Ricketts, Laboratory Centre for Disease Control, LCDC Building, 0603E I, Ottawa,
Ontario KIA OL2 Phone: [phone] Fax: [phone] email: [email protected]
Mo.C.210
HIV TESTING BEHAVIOUR OF CANADIANS
Sutherland Don*, Archibald CP*, Houston S*, Palmer R**, Major C***, Browne J***
*Bureau of HIV/AIDS & STD, Health Canada, Ottawa; **HIV Protect Centre, Sunnybrook
Hospital,Toronto: ***Ontario Ministry of Health,Toronto.
Objective: To examine the HIV testing behaviour of Canadians aged 15 yr and oeider with
respect to the proportion of the population tested, reason for testing, and the demographic
and ' ocial/behavioural factors assoaciated with testing.
Methods: A Canada-wide random-digit-dialling telephone survey of people aged >_15 yr
was conducted in December of 1995 (n-=3123).Trained interviewers obtained basic demographic information and asked questions on a wide range of health issues, including questions on HIV testing behaviour: ever tested, reason for and venue of most recent test, HIV
status, and presence of risk factors for HIV infection. Chi-square analysis was used to assess
univeriate associations and logistic regression was used to examine the effect of multiple
variaales on HIV testing behaviour: Men and women were analyzed separately
Results: Of the 1395 men and 1728 women interviewed, 17% and 15%. respectively, had
ever been tested for HIV. Overall, 5 I% of those tested were tested within the past two
year'. Individuals aged 20- 34 yr were most likely to have been tested (25% of men, 28% of
worren) and individuals aged 35-50 were next most likely (20% of risen, 14% of women).
Ever testers were more likely to be currently unmarried (62% vs 46%), have a household
income below $30,000 (50% vs 4II%), and to live in communities with populations over I
millioen (43% vs 32%) (all p-values<). Only I I% (I 3% of men, 9% of women) were tested
non nominally; 78% of both men and women were tested at a doctor's office or hospital.
The most common reasons for getting tested were to monitor general health (40% of
men 32% of women) and because the individual felt they were at risk for HIV infection
(30% of each gender). Among those tested, 0/259 women and 3/237 men were HIV+.
Additional data from this survey will be available in February 1996 and further analyses,
including multivariate analyses, will be presented.
Conclusions: Nearly a quarter of young to-middle aged adults in Canada have been voluntarily tested for HIV and half of them have done so in the past two years. Understanding
HIV testing behaviour will help with the planning and evaluation of prevention and counsellirng services, and also with the interpretation of information firom databases of HIV+
tests Such interpretation is essential for effective national surveillance of HIV infection.
Don Sutherland, Postal Locator 0202A,Tunney's Pasture, Ottawa, Canada K IA OL2
Tel: [phone] Fax: [phone] E-mail: [email protected]
Mo.C.21 I
ACCEPTABILITY OF ANTENATAL HIV-I SCREENING
Kiari J*, NduatiRuth*, John G*.**, Gakuru C*, Musia (J*, Bwayo J*, Rogoe K*, Kamau K*,
Kreiss J**. University of Nairobi Kenya* University of Washington Seattle, USA**
Objectives:To evaluate the acceptability of antenatal HIV I screening in Nairobi City Council
Clinics and to determine factors associated with failure to return for HIV I test results.
Methods: Pregnant women presenting to two Nairobi City Council Clinics for routine antenata screening were offered testing for HIV I after pretest counselling. A standardized questionnaire was administered which evaluated sociodemographic and obstetric history and
attitudes toward HIV-I testing.Women who accepted HIV I testing had blood collected for
HIV- I serology in addition to routine testing for hemoglobin level and syphilis serology
Women were advised to return in two weeks for results and a brief interview was conducted at this time to reevaluate attitude towards HIV I testing.
Results: Of the 399 women offered HIV-I testing. 397 (99%) agreed to testing and 379
(95%) felt that all pregnant women should be tested. Of 397 women tested. 276 (70%)
retu ned for results.Women who failed to return for HIV I test results were more likely to
have had a previous pregnancy (OR 1.6 95% ClI.0 2.5). HIV I seropositive women were
signiicantly less likely to return to obtain test results (OR 0.5, 95% CI 0.3 0.7) than HIV I
seronegative women.
Conclusions: Although women n this popslatron almost urersal y agreed 3o HIV-I testing,
30ts did sot reeturn for results. HIV- I seropeosirse wonton were Icss lIkery to rotornc for
results.This may have an impact on perinatal intervention strategies to decrease transmission of HIV I which depend on antenatal HIV I screening.
Dr Ruth Nduati, Department of Pediatrics, Unversity of Nairobi, P O. Box 19676.
Nairobi Kenya

a'
0)
D
C
>
0
C
u
0
a)
0)
C
0
c--
ro
C
0
r
C
0)
32