fern terris-prestholt, lilani kumaranayake, catherine macphail, helen rees, charlotte watts

19
www.hivtools.lshtm.ac .uk Determinants of South African Women’s Demand for New Barrier Methods and their Distribution: Analysis of a Discrete Choice Experiment Fern Terris-Prestholt, Lilani Kumaranayake, Catherine MacPhail, Helen Rees, Charlotte Watts London School of Hygiene & Tropical Medicine Reproductive Health and HIV Research Unit, Johannesburg IAEN August 1-2, 2008, Cuernavaca

Upload: lovie

Post on 04-Jan-2016

25 views

Category:

Documents


1 download

DESCRIPTION

Determinants of South African Women’s Demand for New Barrier Methods and their Distribution: Analysis of a Discrete Choice Experiment. Fern Terris-Prestholt, Lilani Kumaranayake, Catherine MacPhail, Helen Rees, Charlotte Watts London School of Hygiene & Tropical Medicine - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Determinants of South African Women’s Demand for New Barrier Methods and their Distribution:

Analysis of a Discrete Choice Experiment

Fern Terris-Prestholt, Lilani Kumaranayake,

Catherine MacPhail, Helen Rees, Charlotte Watts

London School of Hygiene & Tropical Medicine

Reproductive Health and HIV Research Unit, Johannesburg

IAEN August 1-2, 2008, Cuernavaca

Page 2: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Background 1

Existing products for HIV prevention are failing to provide women with a means to protect themselves from HIV.

Male condom use remains subject to consent and active participation from men.

This study explores women’s demand for new barrier methods for HIV

prevention that women can initiate, and how distribution and promotion strategies can

facilitate their use.

Page 3: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Barrier methods for HIV prevention:

Existing methods Male condoms Female condoms (though not widely available)

Potential new methods Microbicides Diaphragm

Microbicide applicator containing clear gel

Diaphragm

Female Condom

Male Condom

Page 4: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Aims 1: product attributes

Policy-makers are concerned about substitution

This study explores how preferences for products vary by: The product characteristics,

their effectiveness in preventing HIV Women’s characteristics

women’s risk /need profiles and their ability to use male condoms

Page 5: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Aims 2: distribution & promotion strategies

Once an effective microbicide is found, it is critical to ensure women who need them can access them and introduce them into their relationships.

This study explores if: new barrier methods need different distribution and

promotion strategies from the existing methods distribution strategies need to be tailored to reach

different groups of women.

Aim 1 & 2 → to inform the introduction and distribution strategy of new methods.

Page 6: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Data collection

Survey

In three township communities in Johannesburg

1017 sexually active adult women (18-45)

October 2005

Questionnaire included: demographics, reproductive health histories, directly elicited preferences, and two discrete choice experiments

Page 7: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

Physical attributes choice

CHOICE Attribute

A B C

Izinto Product

Female condom

Microbicide

Ngasese Secrecy

X No, Cannot be

used without your partner knowing

Yes, Can be used

without your partner knowing

Ukwehlisa amathuba okukhulelwa Pregnancy risk reduction

95% reduction in pregnancies

1 of 20 women

becomes pregnant

95% reduction in pregnancies

1 of 20 women

becomes pregnant

Ukwehlisa amathuba okusuleleka nge HIV HIV risk reduction

95% risk reduction

19 of 20 women

remain HIV negative

35% risk reduction

7 of 20 women

remain HIV negative

Intengo Price

20 Rand

20 Rand

Ngeke

ngiguquke

kulokhu

bengikwenzile

esikhathini

esedlule

Neither,

I would not

change

what I did

last time

CHOICE

Front Back

Flesh to Flesh

Use no barrier method

Male Condom

Yes, Can be used without your partner

knowing

X No, Cannot be used without your partner

knowing

Does not prevent

pregnancy

All 20 women

become pregnant

95% reduction in pregnancies

1 of 20 women

become pregnant

0 risk reduction

0 f 20 women remain HIV

negative

95% risk reduction

19 of 20

women remain HIV negative

6 choice sets:

Here are the products and this is what they do:

Would you have used either of these products in your last sex act or would you have still done the same as you did the last time you had sex?

Neither option card

Page 8: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

Distribution and promotion strategy choice

CHOICE Attribute

A B

E fumaneha kae

Distribution channel

Spaza shop

Clinic

Mokgwa wa ho lata

Collection method

In a private room

From a person

behind a counter

Molaetsa o mo paketeng

Message on package

Extra Pleasure

HIV Prevention Hlwahlwa

Price

Free, 0 Rand

10 Rand CHOICE

Which of these ways would you prefer to collect your product? Other levels included:

Chemist Supermarket

From a shelfFrom a box or dispensing machine

Pregnancy prevention Women’s empowerment

5 Rand20 Rand

Page 9: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Estimation 1: physical attribute preferences

Nested LogitStructure:

Main effects: Prob=f(Product, HIV effec, Preg effec, secrecy, price)

Plus

Interactions level 1 (Change, not-change): Not change means use a male condom or use no condom (q104), Having had difficulties using a male condom in the past (q103), Cohabiting (q49) and

Interactions 2 (between products profiles): Product * used condomSecrecy* had difficultiesPregnancy effectiveness * cohabitHIV effectiveness * self perceived risk (RISK)HIV effectiveness * cohabitPrice * employment status

Change Not –change (C)condom or no condom

A BProduct profiles

Page 10: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Estimation 2: Distribution strategy preferences

MNL with interactions for: new products (microbicide and diaphragm) socio demographic characteristics (cohabiting, household

SES, employment status)

Prob=f( Source, Collection method, Advert, Price)

Clinic Private room HIV prevention Free

Pharmacy Person behind counter Pregnancy prevention 5

Supermarket Box/dispensing machine Enhanced Pleasure 10

Spaza Shelf Women’s empowerm. 20

Page 11: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Women’s characteristics N=1017

Condom used at last sex-act (q104) 31%

Cohabiting with sexual partner (q149) 55%

Ever experienced difficulties getting partner to use condoms (q103) 38%

Employed (emplyd) 35%

Self-perceived risk of HIV (RiskH) 23%

(RiskM) 27%

(RiskL) 24%

(RiskNo) 27%

Mean age 31.5 years

Education 45% started but did not complete secondary school

Page 12: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

What are women’s relative preferences for new barrier methods and how do these differ by women’s characteristics?  Nested Logit

 Coeff.  

Main effects Diaphragm (DG) 0.126 ***Microbicide (MD) 0.257 ***

  Female condom (FEMC) - 0.383 ***  Secrecy (SECR) 0.071 **  Pregnancy prevention effectiveness (PRG) 1.383 ***  HIV prevention effectiveness (HIV) 3.605 ***  Price (PRC) - 0.121 ***Interactions DG*Used condom 0.146 ***  BMD*Used condom - 0.048   BFEMC*Used condom - 0.098 **  BSECR*Difficulties using condom 0.101 ***  BPRG*Cohabiting - 0.144   BHIV*RISK 0.496 ***  HIV*Cohabiting - 0.334 **  PRC*employed 0.044 ***Not Change Not change (C) 0.904 ***  C* used condom 0.614 ***  C*Difficulties using condom - 0.251 ***  C*Cohabiting - 0.136 **Inclusive Value SWITCH 0.412 ***

Page 13: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Probability of choosing a microbicide with different characteristics over a female condom or neither

0

10

20

30

40

50

60

70

Lowest(HIV 35%,Preg 0%,

R20)

HIV 35%,p=10

HIV 55% HIV 75% HIV 95% Preg 95% Highest (HIV 95%,Preg 95%,

R0)

Cho

ice

prob

abili

ties

(%)

Used a condom Had not use a condom

Page 14: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

Interactions None     Products   Women's characteristics      DGM MCD Cohabit SES Emplyd

Source                  Clinic 0.29 ***    Pharmacy 0.33 ***    Spaza (corner store) -0.43 ***     +Supermarket -0.20 ***    Collection method                  Box or dispensing machine 0.04   +   +A person behind a counter -0.18 ***   --- ++  In a private room 0.11 **     -A shelf 0.02    Advertising message                  HIV prevention 0.10 **   --  Pregnancy prevention 0.06   +   +++Enhanced pleasure -0.30 ***     ++

Women’s empowerment 0.13 ***   +   -- ++

PRICE -0.04 ***         ---    

-/+ is significant at a P-value<0.1, --/++ at a P-value<0.05, and ---/+++ at a P-value<0.01. Note: From 3 separate regressions

What are women’s relative preferences for distribution strategies?

Page 15: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Preferences for advertising messages

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

None Diaphragm Microbicide Cohabiting SES Employed

Rel

ativ

e ut

ilitie

s

HIV prevention Women’s empowerment Enhanced pleasure Pregnancy prevention

Page 16: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Challenges to analysis

Potential Biases: ‘in the last sexact’: Many women responded based on best

intentions in the possible future, rather than realistic and feasible option in last sexact.

Orthogonality: The trade off between a strictly orthogonal experimental design and realism: less than efficient design non convergence of more sophisticated models : e.g. RPL with

correlations which would allow for identification of distribution strategy attributes.

HIV effectiveness very dominant attribute: 35% always chose alternative with highest HIV effectiveness

However, despite shortcomings appear to generate intuitive results.

Page 17: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Conclusions- physical attributes

Encouraging to see greater enthusiasm for switching to new products among women who are less able to protect themselves with existing methods: Women who are not using condoms Have had difficulties using condoms and Are cohabiting.

Product effectiveness clearly played an important role in women’s choices Hierarchical messages should work to help women

choose the best options for them in their situations in this population.

Page 18: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Conclusions – distribution attributes

Outlet type had strongest impact on choices but did not vary much by product or women’s characteristics: Do not need new infrastructure to distribute

new products or reach different women Advertising preferences do vary by product and

women’s characteristics: Advertising can be used to stimulate demand of

different products by different women. Suggests a social marketing approach may be

successful.

Page 19: Fern Terris-Prestholt, Lilani Kumaranayake,  Catherine MacPhail, Helen Rees, Charlotte Watts

www.hivtools.lshtm.ac.uk

Acknowledgements This study was supported by

funding from the DFID/MRC Microbicides Development Programme. UK Department for International Development (DFID). DFID supports policies, programmes and projects to promote international development. DFID provided funds for this study as part of that objective but the views and opinions expressed are those of the authors alone.

FTP, LK, HR, CW are members of the DFID Research Programme Consortium for Research and Capacity Building in Sexual and Reproductive Health and HIV in Developing Countries.

We thank all the women of Ekurhuleni

who participated in the survey and the formative qualitative research and

the enthusiastic research team who spent many an evening and weekend interviewing women.