Swaziland HIV Estimates and Proiections Workshop Report Mountain Inn Hotel - Mbabane Swaziland On 8-10 October 2007 Prepared by: Sanelisiwe Tsela (NERCHA) Helen Atieno Odido (UNAIDS) er, / U~_"'"N_''''A_'"I'D.·_'.''S ~ ' UNICEF. UND? • UNFPA oUNDCP . Ut\IESCO 0 \'MO • WORlD B,\NK
21
Embed
Swaziland HIV Estimates and Proiections · 2012-08-27 · has been conducting antenatal clinic (ANC) sentinel surveillance, HIV increased dramatically from 3.9% to a high of 42.6%
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
The HIV Estimations and Projections workshop was a collaborative effort between NERCHA, the Ministry ofHealth and Social Welfare (MOH&SW) UNAIDS and USG,. The workshop was facilitated by Eleanor Gouws
(UNAIDS), Ray Shiraishi (CDCjUSG) and Sadhna Patel (USGjCDC) and was officially opened by NERCHADirector, Derek von Wissell. Planning and technical support for the workshop was provided by Sanelisiwe Tsela
(NERCHA), Helen Odido (UNAIDS), George Bicego (USG) and Sibongile Mndzebele (MOH&SW).
The report was prepared by Helen Atieno Odido and Sanelisiwe Tsela, with contributions from EleanorGouws, Ray Shiraishi, Sadhna Patel and George Bicego.
Swaziland HIV estimates and Projections Report, December 2007
Acknowledgements
The National Emergency Response Council on HIV and AIDS (NERCHA) would like to express profoundgratitude to UNAIDS cmdCDC/USG for providing the technical support that made this very important exercisea reality. We would especially like to thank the Ministry of Health and Social Welfare (MOH&SW) and otherpartners including CSO, HAPAC, PSI,UNICEF,WHO, MRDYA SWANNEPHA and EGPAFfor this collaborativeeffort and for their invaluable participation. We believe that the outputs of this workshop will assistthe cpuntryin making data driven decisions and subsequently strengthen the response to HIV in the country.
Derek von WissellNERCHA Director
2 Swaziland HIV estimates and Projections Report, December 2007
Abbreviations and Acronyms
AIDSANCARTCDCEPPHIV
M&EMoH&SW
NERCHASDHSSNAP
UNUNAIDS
Acquired Immune Deficiency SyndromeAntenatal Care
Antiretroviral TherapyCenters for Diseases Control
Estimation and Projection PackageHuman Immunodeficiency Virus
Monitoring and EvaluationMinistry of Health and Social WelfareNational Emergency Response Council on HIV and AIDSSwaziland Demographic Health SurveySwaziland National AIDS ProgrammeUnited Nations
Joint United Nations Programme on HIV / AIDS
Swaziland HIV estimates and Projections Report, December 2007 3
Table of Contents
Wo r ksho p and re port p re pa rations 1Ac know led g ments .....................................................................•............................................................................. 2
Ab brevi ati ons and Ac ronyms 3Tab Ie of contents 4
1. INTRODUCTION
1.1 Overview of the HIV Epidemic in Swaziland 5
1.2 Rationale for the Estimation and Projections Workshop 61.3 Wo rksho p'S 0 biecti ves 71.4 Scope of work and de live ra bles 7
1.5 Outputs of the wo r ksho p 8
2. RESULTS
2.1 Estimation and Projection Packa ge (EPP) 9
2.2 Spectrum. Estimating the demographic impact of HIV 102.3 Measuring prevalence among 15-24 age group as a proxy for incidence 14
2.3.1 Measuring Prevalence among 15-24 as a Proxy for Incidence 142.3.2 Output from Reg ression Ana lysis 16
Ap pend ices 17List of Partic ipants 19
4 Swaziland HIV estimates and Projections Report, December 2007
1. INTRODUCTION
1.1 Overview of the HIV Epidemic in Swaziland
Swaziland is one of the countries worst affected by the HIV epidemic in the world. Since 1992, the governmenthas been conducting antenatal clinic (ANC) sentinel surveillance, HIV increased dramatically from 3.9% to a high of42.6% in 2004. In 2006, ANC-based prevalence estimates showed a decline to 39.2%. Most affected by HIVare pregnant women aged 20-29 years, with HIV prevalence of 56.3% in 2004 and 48.9% in 2006, followedby those aged 30-34 (prevalence of 41 % in 2004 and 48.9% in 2006). In 2006-07 the country conducted its firstDemographic and Health Survey (DHS), including HIV testing at national level. The preliminary results confirmed thatSwaziland was in a crisis as the HIV prevalence in the population of women and men aged 15-49 was estimated at 26%.
According to the 2006-07 SDHS, women aged 15-49 are more likely to be HIV positive than men in the same
age group (31 percent and 20 percent, respectively). The findings indicate that the age patterns of HIVinfection differ for women and men. HIV prevalence is higher among women than men in age categoriesyounger than 35, while it is higher among men in the age categories 35 years and older (Figure 2). In 2006,HIV prevalence peaked at 49 percent among women in the age group 25-29, while among men, the infectionrate was at its highest level among those in the 35-39 year age category (45 percent). Interestingly, HIV
prevalence continues to be moderately high among both women and men in the age category 50 and older; forexample, around one-quarter of women and men age 50-54 were infected while 7 percent of women and 13percent of men age 60 and older were infected with HIV in 2006-07.
Figure 1: HIV Prevalence among Population Age 2 and Older byAge and Sex
Levels of HIV among pregnant women attending ANC have been uniformly high in all four regions in the
country, with the worst affected region in 2006 being Shiselweni (41.5%). Manzini region has all along beenthe worst hit but started showing some stabilization in the 2004 ANC survey. The sharp increase in Shiselweni inthe recent years has probably been
1 Preliminary findings of the Swaziland Demographic and Health Sur~ey(SDHS2007), and the 10" Roundof the National HIVSerosurveillanceinWomenattending ANC,STIClientsand TBPatientsat HealthFacilitiesinSwaziland,SurveyReport,2006.
Swazlland HIV estimates and Proloctlons Report, Doc@mber2007 5
-~o-
triggered by the recent industrialization and increased mobility in the region. National HIV prevalence amongantenatal clinic attendees over time is shownin Figure 2.
Figure 2: HIVPrevalence among antenatal clinic attendees in Swaziland 1992-2006 (Sentinel surveillance report,2006, MOH&SW)
4540
35
30
B 25t:~ 20."
~ 15a 10
5o
1992 1994 1996 1998 2000 2002 2004 2006
In order to manage the HIV epidemic in the country, the government formed the National Emergency ResponseCouncil to HIV and AIDS (NERCHA),with the mandate to coordinate the HIV multisectoral response. The Ministryof Health and Social Welfare formed the Swaziland National AIDS Program (SNAP), which leads the healthsector response to HIV. Also in place is a national multisectoral HIV and AIDS M&E Framework which spells outthe goals, objectives, indicators, data sources and reporting arrangements required for monitoring HIVprevalence as well as Swaziland's programmatic response. The country has been able to intensify thecomprehensiveHIV prevention, care and treatment efforts guided by the second national multisectoral HIV andAIDS strategic plan 2006-2008. According to the SDHS,the percentage of the adult population who hadbeen tested for HIV and knew their results in 2006 was 22% for women and 9% for men.The Ministry of Healthhas embarked on scaling up HIV Testing and counselling services at all health facilities in an effort to increasethe number of people that know their status. Other efforts to increase accessto services include the introductionof the utilization of mobile outreach services in the workplace, and including the rural communities in theprovider initiated approach to HIV testing and counselling. In addition, the Swazi government with supportfrom Global Fund has made significant progress in the past three years in making antiretroviral therapy (ART)accessible to those in need.
1.2 Rationale for the Estimation and Projections Workshop
As a result of the recent increase in treatment accessin the country which has slowed the progression from HIV toAIDS and death for many individuals, data on HIV prevalence alone is no longer sufficient to track HIV infectionrates and ensure that programmatic response efforts are appropriately targeted. To address the need forinformation to ensure effective targeting of HIV programmes the UNAIDS has been publishing global estimateson a number of HIV indicators since 2001. In March 2007, a team comprised of two national staff based atNERCHA and the UNAIDS M&E advisor attended an estimates and projections workshop organized by theUNAIDSjWHO working group on Global HIVj AIDS and STISurveillance in Pretoria, South Africa. During theworkshop, the Swaziland team together with other regional counterparts were trained on the UNAIDSestimation methodology, including Spectrum and EPp,and proceeded to generate estimates of HIV prevalenceand the demographic impact of HIV, treatment needs and coverage estimates using the revised Spectrummodel. At the time, the country had not released the preliminary HIV prevalence results from its first DHS whichare needed for calibration of national, rural and urban HIV prevalence estimates.
6 Swaziland HIV estimates and Projections Report, December 2007
In June 2007, the Ministry of EconomicPlanning and Development released the preliminary findings of the firstSwaziland Demographic and Health Survey (SDHS 2006-07). With the release of the SDHS estimates ofnational HIV prevalence it was an opportune moment for the country, supported by technical experts fromUNAIDS/Geneva and CDC/ Atlanta, to undertake the necessary analysis and modelling to make availablemore reliable estimates of HIV prevalence, the demographic impact of the epidemic, and ART needs andservice coverage. These data are valuable for planning, budgeting, and forecasting as well as for monitoringand evaluation of the response. The UNAIDS/CDC team facilitated a three-day orientation and trainingworkshop on Spectrum and EPPaimed at national staff from NERCHA and MOH&SW, UN and civil societypartners, and in particular those staff who are mandated to manage and provide data regularly.
1.3 Workshop objectives
1. Toreview the country's HIV prevalence data (including data from ANC and DHS),enter data, by site forurban and rural areas, into EPPand calibrate the prevalence curves using the latest DHS estimates ofprevalence
2. To generate uncertainty bounds on the national curve in EPP;read the national prevalence curve intoSpectrum and estimate the demographic impact of HIV
3. To generate estimates on ART needs and coverage as required for various national and UNGASSindicators
4. Toassessthe trends of prevalence among young women (age 15-24 years) attending antenatal clinicsusing regression analysis
5. To organize a consensusbuilding meeting to agree on assumptions, the process and outcomes ofSpectrum and EPPwith the NERCHA,MOH&SW, civil society, the UN and other relevant partners.
1.4 Scope of work and deliverables
1. Togenerate new curves for national, urban and rural HIV prevalence trends2. Toagree on the national HIV prevalence for 2006, with uncertainty bounds3. To assessdemographic impact of HIV in terms of numbers infected, prevalence, incidence and AIDS
related deaths4. Toassessthe potential impact of treatment on prevalence and AIDS deaths5. Torevise estimates for the following national and UNGASS indicators:
Number of Orphans (double and single)No of people in need of ARTby sexNo of people receiving ART(1st and 2nd line by age and sex)Number of children needing treatmentNumber of children receiving cotrimoxazoleNumber of new infections by sexNumber of mothers needing PMTCT
6. To perform regression analysis on the prevalence data among young women aged 15-24 yearsattending antenatal clinics from 1992-2006 to assessthe trends in prevalence.
7. To enhance understanding of EPP and Spectrum and build in-country capacity in using thesemethodologies to make national estimates of HIV and its impact.
Swaziland HIV estimates and Projections Report, December 2007 7
1.5 Outputs of the Workshop
The Estimation and Projection Package (EPP) has been developed- by the UNAIDS Reference Group onEstimates, Modelling and Projections as a tool to assist national programmes in estimating and projecting thecourse of the HIV epidemic in their countries. EPP provides direct input to Spectrum, which together withpopulation and demographic data, epidemiological assumptions and treatment coverage, produces estimatesand projections of HIV incidence, deaths, treatment needs and other AIDS impacts. At this workshop,participants applied EPP and Spectrum to data collected in Swaziland.
2Brown T, Grassly NC, Gamett G, Stanecki K. Improving projections at the country level: the UNAIDS Estimation and ProjectionPackage 2005. Sexually Transmitted Infections 2006;82 (suppUII): iii34-iii40
3Stover J, Walker N, Grassly NC, Marston M, Projecting the demographic impact of AIDS and the number of people in need oftreatment: updates to the Spectrum projection package. Sexually Transmitted Infections 2006;82 (suppl_3): iii45- iii50.
8 Swaziland HIV estimates amiH~roiections Report, December 2007
2. RESULTS
2.1 The Estimation and Proiection Package (EPP) - process to produce HIV prevalencecurves
b) EPP outputs (Figure 3 and 4)
Curves fitted to regional data showthat the national epidemic is levelling off at around 26%National prevalence for 2007 was estimated at 26.03%Uncertainty bounds on the prevalence curve are also shownin figure 4 belowThe figure below shows that prevalence in Swaziland is generally high in all four regions, withHhohho curve showing high trends until 201 O.
Figure 3: Estimated prevalence curves fitted to data from the four regions in Swaziland
40
> 30I€~0'1C
~ 20~"0ro-o-CID(.)L-ID
CL 10
o1970 1974 1978 1982 1986 1990 1994 1998 2002
••
•
2006 2010
1_- Lubombo- S_ 0d071-% -% -% -% -%
Swaziland HIV estimates and Projections Report, December 2007 9
Figure 4: Swaziland national projected curve with uncertainty range
2.2 Spectrum: Estimating the Demographic Impact of HIV
Several new assumptions were endorsed by the UNAIDS Reference Group on Estimates, Modelling andProjections at its December 2006 meeting. These assumptions have been built into the 2007 versions of theestimation software tools (EPP2007 and Spectrum 3) that were used in this Swaziland Estimatesand Projectionsworkshop.
Spectrum was used to estimate the demographic impact of HIV in Swaziland, using the assumptions andprocessesoutlined below.
a) The Process and Epidemiological assumptions in Spectrum
National prevalence curve: Calibrated using the SDHS,generated by EPPanalysis was read into
SPECTRUM IHIV progression: Net survival of people living with HIV is assumed to be 11 years (instead oflpreviously assumed 9 years) based on available cohort data (paper in press). Waybill survivalfunction assumesthat the median time from HIV infection to ART need is 8 years, and median timefrom ART need to death (in the absence of treatment) is 3 years, totaling 11 years from HIVinfection to death in the absence of treatment.
10 Swaziland HIV estimates and Projections Report, December 2007
HIV age distribution: the sex and age specific distribution of HIV prevalence from the 2006Swaziland OHS,preliminary report were used (seeTable 1).
.•.'geS-ll.jJ'
----.-- •• - "._" >.-
2.302/.8
'').1 . Il~_
FeDII:3 '~-e'").0
38.2,0489045.1
~:;.n..;, ,I' .,.l)"•..•'
''),7.,6021.1'024.11.03:o..Jl.tO
Table 1: Prevalence of HIV by age and sex from the Swaziland OHS,2006
Sex ratio: The female to male HIV prevalence ratio from the 2006-07 SOHS was used. Adultfemale and male prevalence rates from the SOHSwere 31.1 % and 19.7% respectively, resulting ina female: male ratio in 2006 of 1.58
HIV-related reduction in fertility rates: the SPECTRUMdefault estimates of the ratio of fertility ofHIV infected women to the fertility of uninfected women were changed to a ratio of 1.5 amongwomen aged 15-19 years, and a ratio of 0.95 for all women older than 20 years. Should morereliable Swazi-specific information on the impact of HIV on become available, such data will beused to update the assumptions.
Treatment coverage: Number of adults and children in need of ART who are currently receivingtreatment are presented in Table 2, together with national targets for provision of ART by 2012.The number of women receiving PMTCT services was to 8221 in 2006. It is assumed that thenumber was scaled up linearly from 2004 to 2006 and then from 2006 to about 12,000 (90%) in2012.
Number of HIV positive children receiving cotrimoxozole: the number could not be verified dueto the inability of current registers to capture this information.
Number of adults and children in need of treatment: the number were estimated from Spectrum
Duration of breastfeeding is assumed to be between 7 and 17 months. This may be revisedfollowing release of final OHSfindings and report on early 2008
Percent of HIV positive mothers who provide mixed feeding to children: 100% (as comparedto exclusive breastfeeding and replacement feeding).
b) Demographic inputs and assumptions
Provisional estimates from the national censusconducted in Swaziland in 2006 indicate that the
national population size in Swaziland is about 950,000 (final numbers as well as populationbreakdown by age and sex were not available at the time of the workshop).
Recent censusdata therefore indicate that the Spectrum population projections (based on fertilityand mortality assumptions including HIV effects) are probably too high (1.2 million). To producemore accurate estimates of population size life expectancies were adjusted d~w~ward.
Swaziland HIV estimates and Projections Report, December 2007 1 1
Projected life expectancy in the absence of AIDS, needed by Spectrum to produce a basepopulation trend, was thus altered so that it remained stable at 55 for men and 59 for womenbetween 1985 and 201 2. This resulted in an estimated population size (including the impact ofHIV) of about 1.1 miUion in 2006. Population sizes by age and sex in Spectrum will be furtheradjusted occording to the official 2006 censusdata as soon as these data are officially verifiedand endorsed.
c) Summary of Results
Tables 2 and 3 below provide HIV estimates based on the calculations and assumptionsstated above.
Specific Spectrum Outputs:
Estimated adult (15-49 year) prevalence in 2007 is about 26.1 % with an uncertainty range of24.9- 27.1 % as shownin figure 5.Total number of adults and children living with HIVin 2007: 185,005 (with uncertainty range from176,707 -192,794).Estimated number of new adult infections in 2007: 12,568 (11,258 - 13,735)Estimated total number of adult and child deaths in 2007: 11,545 (9,923 - 13,123)Estimated number of adults in need of ARTin 2007: 53,721 (44,016 - 62,099)Estimated number of children in need of ARTin 2007: 4,529 (4,217 - 4,877)Estimated number of AIDS orphans in 2007: 54,541 (46,710 - 63,400)
Figure 5: Adult HIV prevalence with 95% uncertainty ranges
Uncertainty analysis was done in Spectrum using 500 iterations
12 Swaziland HIV estimates and Projections Report, Dece.mber 2007
Table 2: HIV population, adults, mothers and children needing and receiving ART and/or PMTCT and contrimoxazole
Table 3: Summary data on estimated new infections, orphans, and AIDS deaths from 2004-2012
YEAR
New InfectionsNew Inf·ectionsAIDS Orphan s
OVCAIDS Deaths
AIDS-related
(Adults 15+)
(children) (Adults lS+)Child Deaths
'2004
123423746392239552380242795
2005
1218835154502210090283152067
2006
1229832895050810479590022374
2007
1256832235454110798888872658
2008
1306031475800211046092742711
2009
1359131136075611222894562758
2010
1409030786281911339695122783
2011
1459230286425511404294762789
2012
1477529916515411457593872771
r-....ooN...(\)
..0E(\)u(\)a+...oa.(\)
Co::
VI
Co'';::u(\)
'0...Q..
'"0coVI
(\)+-oE
'';::VI
(\)
2::::r:'"0co'No~
en
2.3 Measuring Prevalence among 15-24 as a Proxy for Incidence
One of the UNGASS goals (UNGASS indicator number 22) is to achieve a 25% reduction in HIV prevalenceamong young people, aged 15-24 years, in the most affected countries by 2005 and globally by 2010. Thisindicator can be calculated using prevalence data from pregnant women attending antenatal clinics in HIVsentinel surveillance sites in urban and rural areas.
2.3.1. Measuring Progress towards UNGASS Goals
In order to measure progress towards the above UNGASS indicator, a regression analysis can be performedon the HIV prevalence data over time: to determine if there was a change over time, to determine if the changewas statistically significant, and to determine the amount of change over time. For this analysis, data over timewere used for consistent sites only, i.e. those sites that have consistently recorded HIV prevalence over time, asshown in Figure 9 and 10 for urban and rural sites.
A minimum of three data points per site over time are generally required to perform the trend analysis. Toassessthe trends in HIV prevalence among 15-24 year old women attending ANCs in Swaziland, site-specificdata from 9 urban sites and 8 rural sites collected between 2002 and 2006 (as shown in Table 4) wereanalysed. Site-level HIV prevalence data were plotted and visual trends were inspected (Figure 9 and 10below).
Table 4: HIV prevalence data for young women (15-24) attending urban and rural antenatal clinics in Swaziland:2002-2006.
Site name 2002N2004N2006NUrban HAS
39.4719040.229219.0563HLA THIKHULU
36.7511735.7111241.8498KING SO BHUZA
41.6217351.7014744.78134MANKAYANE
48.818431.589521.9891MBABANE PHU
43.1813237.7012233.61119NHLANGANO
38.5213541.4616439.17120PIGGS PEAK PHU
39.1712044.4410841.1190SITEKI
35.5412136.8814132.08106VUVULANE'
46.21132----47.2755
Rural
DVOKOL W AKO HC
33.787432.887332.2662DWALENI
38.713151.433533.3351EMKHUZWENI HC
33.9610629.856732.8873LOMAHASHA
35.2410544.198625.4555LUYENGO
40.008542.8611233.7189MATSANJENI
40.1910733.9810337.3375NDZEVANE
32.205938.208937.0454SITHOBELA
40.006531.585728.43102
5Vuvulanenot includedinanalysis
14 Swaziland HIV estimates and Projections Report, December 2007
60
50
~ ?l-
Z; 40
c:""i; 300::2::r:: 20~ I(
10
0
2001
2002 2003 2004
Year
2005 2006
-------+-
2007
FLAS
HLATHIKHULU
KING SOBHUZA
MANKAYANE
MBABANEPHU
NHLANGANO
P1GGS PEAK PHU
SITEKl
Figure 9: HIV prevalence data by urban antenatal clinic for young women (15-24) in Swaziland,2002-2006.
50
-+- DVOKOLW AKO HC
---DWALENI
EMKHUZWENIHC~WMAHASHA-- WYENGO-- MATSANIENI
-4-NDZEVANE
SITHOBELA
10
o
2001 2002 2003 2004
Year
2005 2006 2007
Figure 10: HIV prevalence data by rural antenatal clinic for young women (15-24) in Swaziland,2002-2006.
Swaziland HIV estimates and Projections Report, December 2007 15
2.3.2 Output from Regression Analysis
Regressionanalysis was performed to assesstrends in HIV prevalence among young women (15-24) attendingurban and rural antenatal clinics. Resultsare shownfor urban and rural areas in Figures 11 and 12.
Key Finding: Predicted HIV prevalence among young people is estimated to have decreased by 17.4% and13.1 % from 2000 to 2005 and by 15.0% and 11.19% from 2002 to 2006 for urban and rural sites,respectively. The results indicate that, despite clear hopeful trends, Swaziland has not yet attained theUNGASS goal of 25% reduction in HIV prevalence among young people aged 15-24 years by 2005.
Figures 11: Urban trend with a regression line fitted against HIV prevalence data
60.0
50.0 ~
•• ;-••
.40.0 1-!-t
•--
30.0
•20.0
••10.0 0.0 I
1999
20002001200220032004200520062007
Figures 12: Rural trend with a regression line fitted against HIV prevalence data
60.0
50.0
••40.0 l ••
30.01
•t~•••
20.010.00.01999 2000
200120022003 20042005 20062007 2008
16 Swazilcmd HIV estimates and Proiedlons Report, December 2007
Appendices
Table 13: Summary output for urban regression analysis.Regression StatisticsMultiple R