Share your thoughts on this presentation with #IAS2019 Operationalizing HIV prevention cascade for PWID using integrated bio-behavioral survey data from Ukraine Olga Varetska @TwitterHandle Share your thoughts on this presentation with #IAS2019

Analysis undertaken Construction of the prevention cascade. Definition of the cascade indicators Sequential cascade vs. traditional presentation Analysis of the predictors of prevention cascade outcomes age, sex, education, marital status, monthly income, duration of injection drug use in years and drug type that was injected during past 30 days Association tested using chi-square test in the bivariate analysis and multivariable logistic regression analysis Comparison between cascades for HIV-negative and HIV-positive PWID

Background Ukraine is a country in Eastern Europe with estimated HIV prevalence of 0.6% among adults HIV epidemic was initially driven by PWID, who have the highest prevalence among key populations (22.6% in 2017) and continue to play a key role in ongoing HIV transmission in Ukraine Ukraine implements a national needle and syringe program (NSP) that covers 226,469 individual PWID with the minimum prevention package and 164,140 with HIV testing Minimum prevention package includes provision of syringes, condoms and counseling on HIV prevention and referral for other health needs

Data source and key intervention 2017 integrated bio-behavioral surveillance (IBBS) data 10,076 participants, 2261 tested HIV-positive and 7,815 tested negative, (22.4% HIV prevalence). Median age - 35 years, 18% -females. Key outcome of interest - prevention of HIV transmission through unsafe injection behavior among PWID Core intervention - NSP Minimum package of services: syringes, condoms and counseling on HIV prevention and referral for other health needs

Cascade pillars Estimated number of HIV-negative* PWID size estimation Coverage of prevention services: 2a number of PWID who received either a syringe or a condom or a HIV test through the prevention programs in the past 12 months 2b number of PWID who received a syringe and a condom and a HIV test through the prevention programs in the past 12 months 3. safe injection practices : 3a use of a clean syringe and needle at last injection 3b using only clean syringes in the past 30 days 3c absence of any injection risk in the past 30 days (defined as not using shared syringes/needles, other injection equipment or containers, pre-filled syringes, and back- or front-loading) *A separate cascade was developed for HIV-positive PWID

Sequential cascade vs. traditional presentation of indicators Cascade logic: each consequent indicator is calculated as a proportion of the previous one However, in prevention the pathways along the cascade are not always linear: prevention interventions not only provide commodities, but also educate and sustainably change behaviors; commodities are available outside of the prevention programs We compare and contrast two approaches – the ‘sequential’ cascade and presentation of the same indicators separately

Prevention cascade: sequential cascade vs Prevention cascade: sequential cascade vs. traditional presentation of indicators

Analysis of the predictors of prevention cascade outcomes There was lower coverage among: younger PWID (26% in <25 compared to 48% among >45), PWID with shorter injection duration (24% in those injecting <=2 years compared to 44% among >2), and stimulant users (27% compared to 44% among opioid users). Despite lower service coverage, the younger, less experienced PWID and stimulant users reported safer behaviors than their comparison groups (46% in <25 vs 39% in >45; 46% in <=2 years vs. 37% in >2; 44% for stimulant users vs. 39% for opioid users)

Sequential cascade Total   Total Received syringe or condom or HIV test in 12 months Received syringe and condom and HIV test in 12 months Used clean syringe for last injection Used only clean syringes in 30 days Did not have injection risk in 30 days N % p-value  Total 7535 3155 42% 1375 18% 1349 1306 17% 480 6% Sex male 6360 2635 41% .028* 1134 0.236 1111 .410a 1082 .010* 400 0.723 female 1175 520 44% 241 21% 238 20% 224 19% 80 7% Age <=24 613 160 26% .000* 56 9% 0.051 55 .930a 52 8% 0.584 23 4% 0.264 25-44 6065 2587 43% 1142 1121 1086 389 >=45 857 408 48% 177 173 168 68 Injection duration <=2 years 689 165 24% 46 .339 44 .647 13 2% 0.313 >=3 years 6827 2982 1326 1300 1259 466 Family status  live alone 3123 1304 .875b 550 .068b 536 .147a 524 .105 185 .374 live w/partner 4412 1851 825 813 782 295 Education <high school 1230 504 0.704 215 0.909 204 .001* 192 16% .014* 88 .006* high school 4688 1973 867 855 836 284 >high school 1617 678 293 290 278 108 Monthly income <120 USD 2216 987 45% .004* 404 0.130 396 .868 384 .390 162 .008* 120-400 USD 4414 1820 811 796 767 274 400-800 USD 733 280 38% 130 127 126 39 5% >800 USD 172 40% 30 29 5 3% Type of drugs injected in 30 days only opioids 4712 2065 0.000 892 0.032 876 0.544 853 0.152 320 0.038 only stimulant 1057 287 27% 110 10% 109 106 48 mix or other 1766 803 373 364 347 112

Separate indicators Total   Total Received syringe or condom or HIV test in 12 months Received syringe and condom in 12 months Received syringe and condom and HIV test in 12 months Used clean syringe for last injection Used only clean syringes in 30 days Did not have injection risk in 30 days N % p-value Total HIV negative 7792 3276 42% 2584 33% 1411 18% 7583 97% 7405 95% 2882 38% Sex male 6569 2727 .028* 2154 0.106 1162 .026* 6387 0.110 6278 96% .000* 2437 0.758 female 1223 549 45% 430 35% 249 20% 1196 98% 1127 92% 445 Age <=24 640 165 26% 112 56 9% 616 0.154 607 0.635 285 46% 25-44 6268 2688 43% 2132 34% 1173 19% 6110 5959 2260 37% >=45 884 423 48% 340 182 21% 857 839 337 39% Injection duration <=2 years 712 170 24% 114 16% 47 7% 692 0.398 676 94% 0.331 316 >=3 years 7061 3098 44% 2464 1361 6873 6711 2558 Family status live alone 3269 1371 .875b 1058 32% .204b 565 17% .108b 3157 .008*,b 3111 .516b 1227 .136b live with partner 4523 1905 1526 846 4426 4294 1655 Education <high school 1260 517 41% 0.704 412 0.915 219 0.718 1211 .036* 1177 .008* 484 0.682 high school 4878 2065 1625 895 4762 4646 1778 >high school 1654 694 547 297 1610 1582 620 Monthly income <120 USD 2325 1044 .004* 807 0.114 420 0.995 2259 0.697 2192 0.500 908 120-400 USD 4547 1877 1498 827 4424 4340 1713 400-800 USD 747 286 227 30% 133 732 709 214 29% >800 USD 173 69 40% 52 31 168 164 27% Type of drugs injected in 30 days only opioids 4914 2163 1687 919 4796 0.181 4721 1862 only stimulants 1079 296 113 10% 1047 1018 470 mix or other 1799 817 670 379 1740 1666 93% 550 31% Total HIV positive  1239 55% 1088 2168 2094 725 1693 907 54% .039* 795 47% 0.052 1626 0.767 1578 .048* 546 0.674 567 332 59% 293 52% 542 516 91% 179 25 7 28% .021* 6 23 0.477 24 0.768 1795 984 865 1725 1660 440 248 56% 217 49% 410 171 Injection duration  48 17 .006* 13 .003* 45 .431b 42 88% .115b 18 0.408 2207 1220 1073 2118 2048 707 1049 579 .741b 500 .673b 1014 .177b 970 .918b 343 .491b 660 588 1154 1124 382 450 220 .002* 195 0.175 407 0.179 130 0.180 1484 818 1430 1386 480 326 201 62% 181 315 301 115 36% 957 570 60% .001* 504 53% 924 0.519 887 0.262 314 0.339 562 490 1027 998 347 184 85 76 174 169 54 46 22 43 40 87% 10 22% 1641 910 790 .007* 1577 0.617 1543 140 55 51 135 129 57 479 274 57% 247 456 422 89% 119 25%

Comparison between cascades for HIV-negative and HIV-positive PWID In comparison to HIV-negative group, HIV-positive group had: higher service coverage (55% compared to 42% for any service, and 48% compared to 33% for both syringe and condom), but somewhat lower level of safe behavior (33% compared to 38% had no injection risk in the past 30 days). similar to HIV-negative counterparts, younger and less experienced HIV-positive PWID had lower service coverage, but unlike them had less safe behaviors (28% had no injection risk in the past 30 days for <=25 years old compared to 39% in >=45 group).

Conclusion Coverage of PWID with prevention interventions remains suboptimal: At least one service was provided to 42% of the HIV-negative and to 55% of HIV-positive PWID in the past 12 months Full package of service – provided to only 18% of the in HIV-negative PWID. Likely social desirability bias in syringe sharing indicators Composite indicator on injection risk showed that prevalence of behaviors associated with HIV and HCV transmission remains significant in Ukrainian PWID It is necessary to analyze each type of risk behavior and associated factors separately

Discussion No single linear causal pathway in prevention “Motivation” pillar not directly applicable: motivation to use NSP is nearly universal among PWID because new needles are less traumatic Inclusion of only HIV negative population in the cascade calculation? Service coverage indicators could be defined in a number of ways – past 12 months? Some components of the service package are not directly relevant to the behaviors of interest (e.g. condom provision does not affect injection behavior) Standard UNAIDS indictor on last injection is very much prone to social desirability bias

Discussion: non-linear nature of the prevention cascade Some people can be on the safe side of the continuum without using any prevention service Prevention programs have multiple components that pursue different aims than safe behaviors (e.g. overall health improvement, social functioning) Safe behavior indicators in the PWID population are much higher when not linked to the program coverage - cascade representation might be misleading for program planners

Acknowledgements National HIV prevention program among PWID, implemented by Alliance for Public Health and supported by Global Fund in Ukraine Co-authors: Kostyantyn Dumchev, Olga Varetska, Yana Sazonova UNAIDS leading the prevention cascade discussion

Thank you!