AFISE Adherence to antiretroviral medication in Ro HIV

HESPER HAN Adherence to antiretroviral medication in Romanian HIV patients: is it about taking or timing? Objective We describe the development and validation in Romanian of the first self-report tool that assesses adherence to medication by taking into consideration two complementary behaviors: taking the medication and timing of intake. Background Antiretroviral treatment (ART) requires high adherence to achieve therapeutic benefits; ideally, patients need to administer daily all doses recommended (taking adherence) and do so at exact time intervals (timing adherence). These two elements have been previously measured through electronic monitoring (EM), which is less easily applicable to routine care. Developing a self-report questionnaire that targets taking and timing adherence would facilitate clinical assessment and the effective planning of adherence interventions in clinical settings. Methods We developed the ProMAS-Rom-HIV, a 12-item questionnaire on adherence to ART, based on ProMAS, an 18-item Dutch adherence questionnaire validated for chronic conditions with item response theory methodology. The binary response scale of ProMAS (“yes/ no”) was replaced with a 7 point response scale, to investigate the feasibility of increasing the amount of information obtained. The ProMAS items were back-translated to Romanian, adapted to ART, and piloted with a sample of 10 patients via cognitive interviewing. Patients responded to a form that contained both socio-demographics and the adherence questionnaire and results of most recent viral load tests (+/- 12 months from questionnaire completion date) were collected from medical files. As suggested by the World Health Organization, a cut-off point of 1000 copies per ml was set to distinguish between HIV detectable and undetectable levels. We used Mokken Scaling Analysis, Factor Analysis, classical test theory analysis and cluster analysis to examine the structural validity of ProMAS-Rom-HIV. Reliability was tested with Cronbach’s alpha, Guttmann’ s lambda 6, beta, omega. Regarding criterion validity, we hypothesized that adherence will be negatively correlated with viral load values and tested this hypothesis via Wilcoxon rank sum tests. Results Sample: Reliability: We included 104 Romanian HIV patients with the following characteristics: 63% women, mean age 31 +-7, 20% with detectable viral load, mean age since HIV diagnosis 11.12, 27% employed, 55% not parents, 40% living in rural areas (see table 1). Characteristic Value Count )%)/ mean (SD) Education Elementary 9 (8.7) High school 42 (40.4) Middle 28 (26.9) None 1 (1) University 24 (23.1) Number of children 0 58 (55.8) 1 31 (29.8) 2 10 (9.6) 3 5 (4.8) Family status Extended family 52 (50) Independent family 44 (42.3) Other situations 8 (7.7) Work status Employed 29 (27.9) Unemployed 69 (66.3) Occasional work 6 (5.8) Relationship status Single 33 (31.7) Stable relationship 71 (68.3) Other treatment taken for other None 89 (85.9) conditions or comorbidities Psychiatric medication 3 (2.9) Tuberculosis medication 2 (1.9) Other medication 10 (6.9) The values for reliability were satisfactory, with omega=.87 for the “taking scale” and .75 for the “timing scale”, as shown in table 4 Scale Taking adherence Timing adherence Calpha 0.863 0.728 G6 0.857 0.667 Beta 0.766 0.609 Omega 0.871 0.753 Table 4: Reliability indices for both scales The items were weakly associated with each other (Spearman’s rho=.23), as seen in figure 1. Figure 1 Heatplot Spearman correlations between item scores Table 1: Demographic and medical data of the sample (N= 104) Structural validity: Out of the 12 items tested, eight items reflected two separate dimensions: timing and taking adherence with 3 and 5 items, respectively (see table 2). Both scales had good psychometric properties: the first scale had a homogeneity value H(se)=.74(.07) as shown in table 2 and the value for the second subscale was of .62(.08) - table 3. Item H se Item H se ProMas.1 0.755 (0.087) ProMas.3 0.615 (0.091) ProMas.6 0.711 (0.090) ProMas.9 0.567 (0.113) ProMas.7 0.776 (0.090) ProMas.10 0.659 (0.074) Subscale 2 has a homogeneity value H(se) = 0.616, (0.084) ProMas.8 0.773 (0.087) ProMas.11 0.715 (0.086) Table 3: Item homogeneity values for subscale 2 Subscale 1 has a homogeneity value H(se) = 0.741, (0.071) Table 2: Item homogeneity values for subscale 1 Criterion ##   validity: ##  Wilcoxon rank sum test with continuity correction  ##   ##   ##  Wilcoxon rank sum test with continuity correction  ## data:  MostData$ProMAS.Taking[MostData$datenv.prms >= ‐365] by MostData$viral.load[MostData$datenv.prms >= ‐365]  ##   ## W = 777, p‐value = 0.04379  ## data:  MostData$ProMAS.Taking[MostData$datenv.prms >= ‐365] by MostData$viral.load[MostData$datenv.prms >= ‐365]  ## alternative hypothesis: true location shift is not equal to 0 ## W = 777, p‐value = 0.04379  ## alternative hypothesis: true location shift is not equal to 0 People with undetectable viral load were more likely to show high adherence for both timing (W=857, p=.01) and taking (W=777, p=.04), as illustrated in figure 2 (bubble chart cross tab). Authors Ana-Maria SCHWEITZER Fundatia Baylor (Baylor Foundation), Romania aschweitzer@baylor-romania.ro English During the last month it so happened that I simply forgot to take (one of ) the medica- tion During the last month it so happened that I took (one of ) my medication at a different moment than indicated by the doctor (or the moment I usually take it) During the last month it so happened that I took less medication than prescribed by my doctor During the last month I took a break from taking (one of ) my medication During the last month I decided to stop tak- ing (one of ) my medicines During the last month I changed by myself the ti Z[و ۙHو H^HYYX][ۂ\[H\[۝]\[Y]H ۙHو H^HYYX[\]H]\Y[[\X[\[H\[۝]\\[Y]HYZH ۙHو H^HYYX[\܈B^KY\p ΰX]0X\X[ܜ[][ۜ]Y[][pܙ\”X[X[[[[XH[HZHXH[[\]\H[\BHZ]HX]H [[[HYYX[Y[x)[[[XH[HXH[[\]H[ZHX]H [[[HYYX[Y[[HH[HܘHX][ZKXH\ܝ[ ]HX]HX]HHؚKBZJx)[[[XH[HXH[[\]HX]HXZHKB[HYYX[Y[HX]ZKXH\܋B[8)[[[XH[H[HX]]^H[K[ZHXB[[[HYYX[Y[x)[[[XH[H[H\]H\[][[[[YYX[Y[x)[[[XH[H[H[X][\ܘHB\H[ZHX]HYYX[Y[[x)[[[XH[HXH[[\]H[ZHX]H [[[HYYX[Y[HXZH\]HX]HؚKBZx)[[[XH[HXH[[\]H\]HYKBX[Y[[HH[XYx)XH NPTTWUH][\‘Y\p NXpX\0ܛXZ[Y\[p [p[\pܙ\pp\[0YY\p NXpX\0ܛXZ[Y\[p [p[\pܙ\pp\[0YZ^HY[XHR[]XH^[܈ ^[܈[][ۊKX[XBZ^[܋\X[XK“ZYZH\B\X\[H܈XXYZ\S[]\]Hو\YYY[\]\[“ZYZK\P[[^[H[XBX[\X\[\ܛX[H\X\ TTK]YH\\[ۈ H[]\]K[ۋ[B[^[K[XP[][[یKY\p XpX\0ܛX[Z[Y\[p [p[\pܙ\pp\[0YY\H XH\ܛX܈Z[[[Z[ [H[\Hܙ\H[\[YY\p XpX\0ܛX[Z[Y\[p [p[\pܙ\pp\[0Yۘ\[ۜ΂\\H\]Y\[ۛZ\H]X\HYX\\\[Z[[Z[\\\]BZ][ܜ]\HY\[K][\HY[YHوZ[H\Hܝ]Y\B[ۛZ\K]H[XX[]K\Y\]X]H܈\H[H\H[[X[][˂\[HPTTKRU[[X[XXH[\][\[[ۈX[]Y\˂[HZ][ܜ[HYX\\YXH[\\ܝ HXYX\Y\܈[Z[˜[Z[Y\[H[HY[YYY܈]Y[ܛ\[[܈XX[]YX[]Y[[\]Y]H]\[YK][ܙHX\H[\[[ۜˈH\[Y] BYYZ[\[Y ]][\\[Y]H[Y][ۈYZ[SH]H[[\[X[۝^