Exploring the Use of Self-Management Strategies for Antiretroviral Therapy Adherence Among Women with HIV in the Miami-Dade County Ryan White Program

Abstract

Women with HIV (WWH) face increased difficulties maintaining adherence to antiretroviral therapy (ART) due to a variety of demographic and psychosocial factors. To navigate the complexities of ART regimens, use of strategies to maintain adherence is recommended. Research in this area, however, has largely focused on adherence interventions, and few studies have examined self-reported preferences for adherence strategies. The purpose and objectives of this study were to explore the use of ART self-management strategies among a diverse sample of WWH, examine demographic and psychosocial differences in strategy use, and assess the association between strategies and ART adherence. The current study presents secondary data of 560 WWH enrolled in the Miami-Dade County Ryan White Program. Participants responded to questionnaire items assessing demographic and psychosocial characteristics, use of adherence strategies, and ART adherence during the past month. Principal component analysis identified four categories among the individual strategies and multivariable binomial logistic regression assessed adherence while controlling for individual-level factors. The majority of WWH reported optimal ART adherence, and nearly all used multiple individual strategies. The number of individual strategies used and preferences for strategy types were associated with various demographic and psychosocial characteristics. Adjusting for demographic and psychosocial characteristics, optimal ART adherence during the past month was associated with the use of four or more individual strategies. When conducting regular assessments of adherence, it may be beneficial to also assess use of adherence strategies and to discuss with WWH how using multiple strategies contributes to better adherence.

In the United States (US), nearly a quarter of people with HIV are women (Kaiser Family Foundation, 2020; Office on Women’s Health, 2021). Compared to men, women with HIV (WWH) are more likely to experience difficulties with adherence to antiretroviral therapy (ART) and viral suppression (Benning et al., 2020; Centers for Disease Control and Prevention, 2022b). There are many contextual and personal factors associated with low ART adherence among WWH. This population is more likely to belong to a racial/ethnic minority group, and thus, are vulnerable to unique barriers to treatment adherence such as obtaining less education, experiencing income and housing instability (Beer et al., 2016; Centers for Disease Control and Prevention, 2022a), facing HIV-related stigma and low social support (Chandran et al., 2019; Loutfy et al., 2012; Vyavaharkar et al., 2007; Waldron et al., 2021), bearing greater responsibility for childcare (D. Merenstein et al., 2009; D. J. Merenstein et al., 2008; Waldron et al., 2021), and reporting mental health difficulties (Beer et al., 2016; Waldron et al., 2021) due to the intersectional nature of these experiences. Consequently, understanding how WWH navigate the complexities of self-managed ART regimens in light of these experiences is a necessary area of exploration.

It is recommended that people with HIV improve ART adherence by adopting various strategies such as setting a schedule; using pill boxes, calendars, diaries, alarms, or mobile applications; anticipating changes in daily routines; and recruiting help of family members (U.S. Department of Health & Human Services, 2022). Research in the area of treatment self-management, however, has largely focused on adherence interventions across various chronic conditions including asthma (Eakin & Rand, 2012; George, 2018), diabetes (Sapkota et al., 2015; Vallis et al., 2023; Williams et al.,2014), hypertension (Izeogu et al., 2020; Xu & Long, 2020), and HIV (Areri et al., 2020; Kanters et al., 2017; Simoni et al., 2008), and few studies of people in self-managed chronic treatment have examined self-reported preferences for adherence strategies. Exploring the use of self-management strategies among WWH is an understudied area of research. One study that investigated strategy use among Black WWH found that 80% of participants used multiple strategies which were associated with different demographic and psychosocial factors (Houston & Osborn, 2015). Specifically, the use of pillboxes was associated with higher education, calendars and other visual reminders were associated with greater extrinsic adherence motivation, and watch alarms were more common among younger participants (Houston & Osborn, 2015).

No study to our knowledge, however, has examined other important contextual and personal factors for WWH including HIV-related stigma, social support, childcare responsibility, and anxiety, or examined the association between strategy preferences and HIV outcomes. The purpose of the current research was to explore the use of ART self-management strategies among a diverse sample of WWH. The main objectives of this work were to (1) identify preferences for various self-management adherence strategies among the sample, (2) examine demographic and psychosocial differences in preferences for different strategies, and (3) assess the association between the use of strategies and ART adherence.

Methods

Research Setting, Participants, and Procedures

In 2020, the South Florida metropolitan that includes Miami-Dade County ranked first in the US in HIV infections and diagnoses with approximately 25,595 people living with HIV (Centers for Disease Control and Prevention, 2022a). At that time, 24.1% of people with HIV in Miami-Dade were women, and over 80% of infections were among Hispanic and non-Hispanic Black individuals (Behavioral Science Research Corporation, 2022b; Sullivan et al., 2020). Nationally, the Ryan White HIV/AIDS Program served over half of people with HIV (Health Resources and Services Administration, 2020); in Miami-Dade, specifically, 19% of Ryan White clients were women (Behavioral Science Research Corporation, 2022a). The current study uses data from a larger cross-sectional study examining the relationship between women-centered HIV care and treatment adherence, retention in care, and viral load suppression among adult WWH who were enrolled in the Miami-Dade County Ryan White program. A convenience sample of 567 participants was surveyed in the larger study, and 560 were included in the current sample. Seven participants were removed due to extensive missing data (n = 1), identifying as a transgender man (n = 1) or transgender woman (n = 3), or not reporting gender (n = 2). Participants were categorized into four racial/ethnic groups: Non-Hispanic Black (excluding Haitians), Hispanic (of any race), Haitian (of any race), or other race/ethnicity (i.e., Non-Hispanic White, Asian, American Indian and Alaska Native, or other race/ethnicity). Data collection occurred from June 2021 to March 2022, and surveys were conducted by telephone due to the COVID-19 pandemic. The survey was administered in English, Spanish, or Haitian Creole based on participant preference. All participants provided verbal informed consent and received a $50 Walmart gift card for participation. All study procedures were approved by the Florida International University IRB.

Measures

Demographic and psychosocial factors.

The survey was developed using the Andersen Behavioral Model for Vulnerable Populations (Andersen, 1995) and its adaptations for HIV (Christopoulos et al., 2011). Demographic data for the current study included gender, race/ethnicity, US or foreign born, age, percent of federal poverty line for household income, and education level. Psychosocial factors included disclosure of HIV status to at least one person (i.e., people in the household, intimate partners, extended family, friends, and co-workers); feelings of HIV stigma related to concerns about privacy and disclosure (Reinius et al., 2017) categorized into low, moderate, and high stigma based on quartile range; perceived social support (Chesney et al., 2000); childcare responsibility (D. Merenstein et al., 2009; D. J. Merenstein et al., 2008); housing instability in the past year; and anxiety and depression symptoms assessed with the GAD-7 and CESD-R-10, respectively, and categorized using recommended cutoffs (Andresen et al., 1994; Spitzer et al., 2006).

ART adherence.

ART adherence was calculated as a percentage of days ART treatment was correctly taken during the past 30 days (Wilson et al., 2020). A cut-off score of 95% adherent days (approximately 29 of the past 30 days) was used to indicate optimal vs. suboptimal adherence. Based on findings of Leach et al. (2021), a lower threshold of 90% adherent days (approximately 27 of the past 30 days) was also considered in sensitivity analyses.

ART self-management strategies.

Eleven individual strategies for ART self-management were measured with a self-report questionnaire item derived from previous in-depth qualitative interviews conducted among WWH (Ramirez-Ortiz et al., 2023). Participants responded “yes” or “no” to each item.

Data Analysis

All analyses were conducted in SPSS 26. Descriptive statistics were conducted for ART adherence and all demographic, psychosocial, and strategy variables. Principal component analysis was conducted to categorize individual adherence strategies. Independent t-tests and one-way ANOVAs with post hoc analyses assessed demographic and psychosocial differences in mean number of individual strategies. Chi-square tests of independence interpreted with standardized adjusted residuals were used to assess differences in use of single vs. multiple strategies, endorsement of strategy types, and the number of strategy types used based on demographic and psychosocial factors. Adjusted binomial logistic regressions assessed the association between strategy types and optimal ART adherence (≥95% adherent days). Adjusted binomial logistic regressions were also conducted to assess the association between number of strategies used and the odds of optimal ART adherence controlling for demographic and psychosocial factors. For all regression analyses controlling for demographic and psychosocial factors, multiple imputation using the MICE package in R was conducted to address data missingness. Exploratory binomial logistic regression analyses assessed the association between odds of optimal adherence and combinations of strategy types endorsed by at least 5% of the sample, and between optimal adherence and using 4 or more individual strategies. Sensitivity analyses were conducted using a lower threshold of 90% adherent days.

Results

Study Population

Of the 560 WWH, 35.4% were Non-Hispanic Black, 33.2% Hispanic, 28.0% Haitian and 3.4% were of another race/ethnicity. Participants ranged in age from 22–90 years of age with most being 50 years and older (63.6%). ART adherence was high among the sample with 82.3% reporting optimal ART adherence in the past month on ≥95% adherent days. Other demographic and psychosocial characteristics are presented in Table 1.

Table 1.

Participant characteristics and differences in mean number of individual strategies used

Characteristics	Total n (%)	Mean strategies (SD)	p-value
Race/ethnicity			< .001***
Non-Hispanic Black	198 (35.4)	4.15 (1.93)a
Hispanic, of any race	186 (33.2)	4.78 (1.83)b
Haitian, of any race	157 (28)	3.80 (1.35)a
Other race/ethnicity	19 (3.4)	3.31 (1.92)a
US born			.291
Yes	203 (36.3)	4.12 (1.96)
No	357 (63.7)	4.30 (1.69)
Age in years (range = 22–90)	M = 52.21, SD = 10.54		.014*
22–34 years old	36 (6.4)	5.06 (1.77)a
35–49 years old	168 (30)	4.10 (1.86)b
50–90 years old	356 (63.6)	4.21 (1.75)b
Income level (% of the federal poverty line)			.368
Less than 100%	245 (43.8)	4.11 (1.76)
100 to 199%	198 (35.4)	4.37 (1.89)
200 to 299%	79 (14.1)	4.35 (1.61)
300% or greater	37 (6.6)	4.05 (1.86)
Education level			.219
Some high school or less	163 (29.1)	4.12 (1.75)
High school graduate or trade school	212 (37.9)	4.16 (1.84)
Some college or more	184 (32.9)	4.42 (1.78)
Adherence to ART in the past month (≥95%)			.003**
Yes	461 (82.3)	4.34 (1.75)
No	99 (17.7)	3.75 (1.93)

Note. Total sample, N = 560. M = mean. SD = standard deviation. Those born in Puerto Rico and other US territories are classified as not US born. Adherence to ART in the past month was calculated as a percentage of days ART was correctly taken during the past 30 days; cut-off points of 90% and 95% adherent days were used to indicate optimal vs. suboptimal adherence. At a threshold of 90% adherent days, 89.9% (n = 503) were considered adherent; frequency of adherence at a threshold of 95% adherent days is reported in the table. Missing responses (I don’t know/refused): Income (n = 1), Education (n = 1). Statistical analyses to assess mean differences in individual strategy use: Independent t-tests reported for dichotomous variables; one-way ANOVAs reported for all other variables. Post hoc analyses for ANOVA results were conducted with Tukey HSD pairwise comparisons. For each significant ANOVA test, significant pairwise comparisons where p ≤ .05 are indicated when superscript letters are different from each other (i.e., a vs. b).

^*p ≤ .05

^**p ≤ .01

^***p ≤ .001

Use of Self-Management Strategies Among WWH

Number of individual self-management strategies.

The number of individual strategies used by participants ranged from zero to 10 strategies (M = 4.23, SD = 1.79). The majority (93.6%, n = 524) used multiple (i.e., two or more) individual strategies with nearly half of these participants reporting use of four (22.3%) or five individual strategies (22.6%) in their self-management plan. Individual strategies that were most popular included a set routine, planning ahead when traveling, a set schedule, a spare dose, a weekly or monthly pillbox, and automatic refills (see Table 2). Just eight women reported not using any strategies.

Table 2.

Endorsement of ART self-management strategies

Strategy Type	Total sample (N = 560)	Multiple strategy users (n = 524)	Single strategy users (n = 28)	Strategies for ART Self-Management	n (%)†
	n (%)	n (%)	n (%)
Routine Strategies	518 (92.5)	500 (95.4)	18 (64.3)	You have a set routine (for example, every day after dinner).	469 (83.8)
				You have a set schedule (for example, every day at 1 PM).	393 (70.2)
Plan Ahead Strategies	480 (85.7)	473 (90.3)	7 (25)	You plan ahead when you travel.	434 (77.5)
				You keep a spare dose with you at all times.	346 (61.8)
				You set up automatic refills of your HIV medication.	226 (40.4)
Analog Strategies	261 (46.6)	259 (49.2)	2 (7.1)	You use a weekly or monthly pillbox.	231 (41.3)
				You use a paper calendar or diary to mark dosages taken.	40 (7.1)
				A family member, friend, or partner reminds you to take your HIV medication every day.	52 (9.3)
Cell Phone Strategies	137 (24.5)	136 (26)	1 (3.6)	You set a daily alarm reminder on your cell phone.	81 (14.5)
				You received daily text message reminders on your phone.	67 (12)
				You use a cell phone application to set reminders or track dosages taken.	32 (5.7)

Note. Column percentages are presented for the total sample, multiple strategy users, and single strategy users. Participants who reported use of two or more individual strategies were categorized as multiple strategy users whereas those who reported use of one individual strategy were categorized as single strategy users. Eight participants did not use any strategies.

^†Percentages reflect endorsement across the total sample.

Types of self-management strategies and patterns of use.

Using principal component analysis with direct oblimin rotation, four combinations of the individual strategies were observed. These combinations represent categories, or “strategy types,” which we have called “routine strategies,” “analog strategies,” “cell phone strategies,” and “plan ahead strategies” (see Table 3 for principal component analysis results). The strategy types utilized by most women in our sample were routine strategies and plan ahead strategies; approximately half incorporated use of analog strategies and a quarter incorporated cell phone strategies (see Table 2). Among those only using one individual strategy, routine strategies (64%) and plan ahead strategies (25%) were the most popular. However, most participants used multiple combinations in their strategy plans; the most common were the use of two (40.5%, n = 221) or three (33.2%, n = 181) strategy types. Most popular among participants using two combinations was routine and plan ahead strategies (81.9%), and among users of three combinations was routine, plan ahead, and analog strategies (97.8%). About 15% (n = 82) of women relied on use of all four strategy combinations.

Table 3.

Principal component analysis loadings of adherence strategy items

Items	Rotated Component Coefficients
	1	2	3	4
1. You have a set routine (for example, every day after dinner).	.79	.01	.16	−.14
2. You have a set schedule (for example, every day at 1 PM).	.68	.14	−.18	.24
3. You plan ahead when you travel.	.12	.71	.06	.24
4. You keep a spare dose with you at all times.	.16	.77	.14	.25
5. You set up automatic refills of your HIV medication.	−.10	.71	.05	−.08
6. You use a weekly or monthly pillbox.	.01	.15	.63	.13
7. You use a paper calendar or diary to mark dosages taken.	.01	.11	.43	.50
8. A family member, friend, or partner reminds you to take your HIV medication daily.	.004	.03	.78	.13
9. You set a daily alarm reminder on your cell phone.	−.20	.19	.10	.63
10. You received daily text message reminders on your phone.	.09	.10	.02	.69
11. You use a cell phone application to set reminders or track dosages taken.	.15	.10	.20	.60

Note. Major loadings for each item are bolded. Extraction method: Principal component analyses. Rotation method: Direct Oblimin. Eigenvalues > 1. Components: (1) Routine strategies, (2) Plan ahead strategies, (3) Analog strategies, (4) Cell phone strategies.

Demographic and Psychosocial Characteristics Associated with Self-Management Strategy Use

Number of individual self-management strategies.

As reported in Tables 2 and 4, the number of individual strategies used by our participants was significantly higher among participants who were Hispanic, 22–34 years old, disclosed HIV status to anyone or specifically to people in their household, reported lower levels of HIV stigma, were “extremely happy” with their social support, or have experienced housing instability. Use of multiple strategies was less likely for participants who were Non-Hispanic Black (91%) or other race/ethnicity (82%) (vs. 97% Hispanic and 98% Haitian, χ2 = 16.45, p < .001), US born (90% vs. 98% foreign born, χ2 = 16.21, p < .001), or “not at all happy” with their social support (82% vs. 89%–98% somewhat to extremely happy, χ2 = 15.48, p = .004). No other demographic or psychosocial differences were found in use of multiple individual strategies.

Table 4.

Psychosocial characteristics and differences in mean number of individual strategies used

Characteristics	Total n (%)	Mean strategies (SD)	p-value
Disclosure of HIV status
Disclosed to at least one person			.049*
Yes	527 (94.1)	4.27 (1.81)
No	33 (5.9)	3.64 (1.54)
Disclosed to people in household			.028*
Yes	343 (61.3)	4.37 (1.84)
No	162 (28.9)	4.02 (1.61)
Disclosed to intimate partners			.072
Yes	266 (47.5)	4.40 (1.88)
No	84 (15)	3.98 (1.82)
Disclosed to extended family			.128
Yes	346 (61.8)	4.33 (1.83)
No	207 (37)	4.09 (1.75)
Disclosed to friends			.325
Yes	204 (36.4)	4.35 (1.84)
No	345 (61.6)	4.19 (1.78)
Disclosed to coworkers			.570
Yes	54 (9.6)	4.33 (1.68)
No	402 (71.8)	4.18 (1.86)
HIV stigma	M = 3.14, SD = .66		.003**
Low stigma	160 (28.6)	4.36 (2.03)a
Moderate stigma	241 (43)	4.41 (1.79)a
High stigma	131 (23.4)	3.84 (1.47)b
Satisfaction with social support	M = 3.87, SD = 1.04		.003**
Not at all happy	17 (3)	4.41 (2.40)
Somewhat happy	36 (6.4)	3.67 (1.77)a
Moderately happy	106 (18.9)	3.96 (1.84)a
Very happy	191 (34.1)	4.30 (1.80)
Extremely happy	164 (29.3)	4.69 (1.68)b
Number of children			.615
No children	402 (71.8)	4.20 (1.81)
1 to 2 children	76 (13.6)	4.21 (1.60)
3 or more children	82 (14.6)	4.41 (1.89)
Difficulty with childcare			.652
No children	402 (71.8)	—
Yes	26 (4.6)	4.42 (1.68)
No	132 (23.6)	4.32 (1.75)
Housing instability			.019*
Yes	75 (13.4)	4.77 (2.15)
No	485 (86.6)	4.15 (1.72)
Anxiety and depression symptoms
No significant anxiety or depression symptoms	379 (67.7)	4.21 (1.79)	.307
Significant anxiety symptoms only	8 (1.4)	4.75 (1.04)
Significant depression symptoms only	81 (14.5)	4.56 (1.80)
Both significant anxiety and depression symptoms	54 (9.6)	4.50 (2.07)

Note. Total sample, N = 560. M = mean. SD = standard deviation. Significant anxiety symptoms were indicated by scores ≥10 on GAD-7 and significant depression symptoms were indicated by scores of ≥10 on CESD-R-10. Disclosed to at least one person was created based on endorsement of any disclosure item. “Not applicable” responses in Disclosure variables: Disclosed to people in household (n = 55), Disclosed to intimate partners (n = 206), Disclosed to extended family (n = 6), Disclosed to friends (n = 9), Disclosed to coworkers (n = 103). Missing responses (I don’t know/refused): Disclosed to intimate partners (n = 4), Disclosed to extended family (n = 1), Disclosed to friends (n = 2), Disclosed to coworkers (n = 1), HIV stigma (n = 28), Satisfaction with social support (n = 46), Anxiety and depression symptoms (n = 38). Statistical analyses to assess mean differences in individual strategy use: Independent t-tests reported for dichotomous variables; one-way ANOVAs reported for all other variables. Post hoc analyses for ANOVA results were conducted with Tukey HSD pairwise comparisons. For each significant ANOVA test, significant pairwise comparisons where p ≤ .05 are indicated when superscript letters are different from each other (i.e., a vs. b).

^*p ≤ .05

^**p ≤ .01

^***p ≤ .001

Types of self-management strategies.

Figure 1 depicts differences in use of each combination of individual strategies by demographic and psychosocial factors. Use of routine strategies was significantly lower among participants of other race/ethnicity and those who have disclosed their HIV status to friends. Use of plan ahead strategies was significantly higher among Hispanic participants and those who have at least some college education. Analog strategies were significantly more common among participants who were Hispanic, 22–34 years old, disclosed HIV status to people in household, intimate partners, extended family, or friends, or had experienced housing instability. However, analog strategies were less commonly used by Haitian participants. Use of cell phone strategies was significantly more common among participants who were Hispanic, foreign born, 22–34 years old, had experienced housing instability, or reported both significant depression and significant anxiety symptoms. By contrast, cell phone strategies were less preferred among participants who were Haitian or other race/ethnicity, 50–90 years old, experienced high levels of stigma, or reported no significant anxiety or depression symptoms.

Fig. 1.

This figure depicts the distribution of participants reporting use of the four types of strategies by demographic and psychosocial factors. An asterisk (*) indicates chi-square test of association is significant: a. Race/ethnicity – use of routine strategies (p = .004), use of plan ahead strategies (p < .001), use of analog strategies (p = .001), use of cell phone strategies (p < .001); b. US born – use of routine strategies (p = .024), use of cell phone strategies (p = .048); c. Age – use of analog strategies (p = .022), use of cell phone strategies (p < .001); e. Education level – use of plan ahead strategies (p = .004); f. Disclosure of HIV status – use of routine strategies (disclosure to friends, p = .023), use of analog strategies (disclosure to people in household, p = .006; disclosure to intimate partners, p = .033; disclosure to extended family, p = .019; disclosure to friends, p = .025); g. HIV stigma – use of cell phone strategies (p = .013); k. Housing instability – use of analog strategies (p = .045), use of cell phone strategies (p = .002); l. Anxiety and depression symptoms – use of cell phone strategies (p = .009)

Number of self-management strategy types.

Figure 2 depicts differences in the number of strategy types used in self-managed plans by demographic and psychosocial factors. Number of strategy types used was associated with the following characteristics: Participants of other race/ethnicity were more likely to not use any strategies, Haitian participants were more likely to use two strategy types, and Hispanic participants were more likely to use all four strategy types. Likewise, use of four types was significantly more common among foreign born participants compared to US born participants who were more likely to use one or three types. While women 22–34 years old were significantly more likely to use all four types, this pattern of use was less common among those 50–90 years old. Participants who received no more than some high school education were more likely to use two strategy types and less likely to use all four. Use of all four types was more common among participants who received at least some college education. Those less satisfied with social support (i.e., “not at all happy” and “somewhat happy”) were significantly more likely to use just one strategy type while those “extremely happy” with their social support were less likely to report this pattern of use. Lastly, those who have experienced housing instability were more likely to use all four strategy types while those who did not face housing instability were more likely to use two strategy types.

Fig. 2.

This figure depicts the distribution of participants reporting use of 0 to 4 strategy types by demographic and psychosocial factors. An asterisk (*) indicates chi-square test of association is significant: a. Race/ethnicity (p < .001); b. US born (p = .033); c. Age (p = .001); e. Education (p = .050); h. Satisfaction with social support (p = .003); k. Housing instability (p = .004)

Use of Self-Management Strategies and ART Adherence

Controlling for the effect of each strategy type, the use of routine strategies was significantly associated with increased odds of optimal adherence on ≥95% adherent days (aOR = 2.45, p = .011, 95% CI [1.23, 4.87]) while use of plan ahead strategies (aOR =1.16, p = .625, 95% CI [.63, 2.14]), analog strategies (aOR = 1.23, p = .362, 95% CI [.79, 1.94]), and cell phone strategies (aOR = 1.05, p = .860, 95% CI [.61, 1.80]) had positive, non-significant associations. We conducted exploratory analyses to examine the association between optimal adherence and all possible combinations of strategy types with at least 5% endorsement from our participants. In unadjusted binomial logistic regressions, there were no significant associations between increased odds of optimal adherence and using routine strategies only (7.3%, OR = 1.05, p = .916, 95% CI [.45, 2.43]), routine and analog strategies (4.5%, OR = 1.13, p = .822, 95% CI [.38, 3.38]), routine and plan ahead strategies (32.9%, OR = .83, p = .413, 95% CI [.53, 1.30]), or all four strategies (14.6%, OR = .87, p = .638, 95% CI [.48, 1.57]). The only combination of strategy types that significantly increased the odds of optimal adherence was a combination of routine, analog, and plan ahead strategies (32.7%, OR = 2.01, p = .008, 95% CI [1.20, 3.38]). However, this combination did not retain an independent and significant effect on the odds of optimal adherence (aOR = 1.67, p = .062, 95% CI [.98, 2.91]) when adjusting for the number of individual strategies used (aOR = 1.16, p = .03, 95% CI [1.02, 1.32]).

In multivariable analyses, the number of individual strategies used significantly contributed to increased odds of optimal adherence above and beyond demographic and psychosocial factors (aOR = 1.24, p = .003, 95% CI [1.08, 1.42]). Those using a combination of routine, analog, and plan ahead strategies reported using an equivalent of three to eight individual strategies, and the overall sample average was four individual strategies. As such, a series of exploratory binomial logistic regression models assessed the extent to which using an average or above-average amount of independent strategies was associated with increased odds of optimal adherence. Overall, 187 (33.4%) women reported using zero to three individual strategies, 124 (22.1%) used four individual strategies, and 249 (44.5%) used five to 10 individual strategies. Using between zero and three individual strategies significantly reduced the odds of optimal adherence (OR = .51, p = .003, 95% CI [.33, .79]). As reported in Table 5, using five to 10 individual strategies significantly increased the odds of optimal adherence above and beyond all demographic and psychosocial factors while using four individual strategies had a positive, non-significant effect. Reporting greater satisfaction with social support was also significantly associated with increased odds of optimal adherence.

Table 5.

Unadjusted and adjusted logistic regression models predicting ART adherence

	ART adherence
	90% threshold			95% threshold
	aOR	p-value	95% CI	aOR	p-value	95% CI
Number of strategies used
0-3 strategies	Ref
4 strategies	2.54	.026 *	[1.12, 5.77]	1.76	.081	[.93, 3.30]
5-10 strategies	3.35	<.001 ***	[1.65, 6.81]	2.47	.001 ***	[1.42, 4.30]
Race/ethnicity
Hispanic ethnicity	Ref
Black race	.63	.347	[.24, 1.65]	.97	.942	[.45, 2.12]
Haitian ethnicity	1.29	.585	[.52, 3.17]	1.07	.843	[.54, 2.13]
Other race/ethnicity	.44	.297	[.09, 2.06]	.98	.970	[.26, 3.73]
U.S. born
No	Ref
Yes	1.02	.964	[.42, 2.49]	.68	.304	[.33, 1.42]
Age
22–34 years old	.23	.006 **	[.08, .65]	.43	.053	[.18, 1.01]
35–49 years old	.52	.052	[.27, 1.01]	.77	.318	[.45, 1.29]
50–90 years old	Ref
Income level (% of federal poverty level)
Less than 100%	Ref
100% to 199%	1.33	.431	[.65, 2.70]	.86	.591	[.51, 1.47]
200% to 399%	1.25	.656	[.47, 3.31]	1.31	.501	[.60, 2.84]
300% or greater	.38	.057	[.14, 1.03]	.53	.157	[.22, 1.28]
Education level
Some high school or less	Ref
High school graduate or trade school	1.05	.900	[.46, 2.39]	1.10	.759	[.69, 2.05]
Some college or more	.57	.192	[.26, 1.32]	.66	.196	[.36, 1.24]
Disclosure to at least one person
No	Ref
Yes	.22	.159	[.03, 1.18]	.36	.130	[.10, 1.35]
HIV stigma
Low stigma	Ref
Moderate stigma	1.32	.452	[.64, 2.70]	1.05	.855	[.60, 1.84]
High stigma	.73	.436	[.33, 1.62]	.84	.604	[.43, 1.63]
Satisfaction with social support	1.19	.228	[.90, 1.58]	1.28	.030 *	[1.02, 1.60]
Number of children
No children	Ref
1 to 2 children	1.32	.554	[.52, 3.35]	1.09	.815	[.54, 2.18]
3 or more children	.96	.934	[.41, 2.23]	.69	.257	[.37, 1.31]
Housing instability
No	Ref
Yes	1.19	.699	[.50, 2.82]	.77	.428	[.40, 1.47]
Anxiety and depression score
No significant anxiety or depression symptoms	Ref
Significant anxiety symptoms only	.78	.821	[.09, 7.01]	1.36	.779	[.16, 11.58]
Significant depression symptoms only	.96	.907	[.45, 2.04]	1.02	.961	[.54, 1.90]
Both significant anxiety and depression symptoms	1.73	.363	[.53, 5.59]	1.72	.242	[.69, 4.25]

Note. OR = odds ratio, aOR = adjusted odds ratio, 95% CI = 95% confidence intervals, Ref = reference group. Data was imputed for all variables with missing data using predictive mean matching on 10 imputed datasets. At 90% adherent days threshold, 89.9% (n = 503) were considered adherent. At 95% adherent days threshold, 82.3% (n = 461) were considered adherent. Bold indicates significant results.

^*p ≤ .05

^**p ≤ .01

^***p ≤ .001

Threshold of 90% Adherent Days.

Sensitivity analyses were conducted using a lower threshold of 90% adherent days. A majority of women (89.9%, n = 503) were classified as adherent. Using between zero and three individual strategies significantly reduced the odds of optimal adherence (OR = .38, p < .001, 95% CI [.32, .66]). On the other hand, using between four and 10 individual strategies significantly increased the odds of optimal adherence above and beyond all demographic and psychosocial factors (see Table 5). Being 50-90 years old compared to 22-34 years old was also significantly associated with increased odds of optimal adherence.

Discussion

Nearly all participants used multiple individual strategies with preferences varying by contextual and individual factors relevant to WWH, as found in previous research (Houston & Osborn, 2015). The current study findings also indicate that utilizing many diverse self-management strategies contributes to high levels of ART adherence. Given the paucity of self-reported data on use of adherence strategies, the results of this study are novel and contribute to an area of research that is understudied.

This work builds on previous work in the current intervention-focused literature by assessing various aspects of strategy plans incorporated into self-managed ART regimens among WWH. In addition to individual strategies, we also considered preferences for combinations of individual strategies to better conceptualize the diversity of strategy plans utilized by the women in our sample. The combinations of individual strategies that our participants largely relied on were the routine strategy type and the plan ahead strategy type. We found that the combination of routine strategies, in particular, significantly contributed to ART adherence above and beyond plan ahead strategies, analog strategies, and cell phone strategies. Our findings contrast with a large portion of previous research in this area which has focused on the effectiveness of strategies more aligned with the analog and cell phone strategy categories examined. However, despite the use of pillboxes as standard in HIV care and the growing interest in the use of technology to promote adherence (Kanters et al., 2017; Office of AIDS Research, 2021; Petersen et al., 2007; Simoni et al., 2008), these types of strategies were not as common among our sample.

Additionally, our results demonstrate that using a greater number of strategies independently and significantly contributed to optimal ART adherence in multivariable analyses above and beyond any specific combination of strategy types. In particular, using five to 10 independent strategies was significantly associated with adherence with a threshold of 95% adherent days, and four or more strategies were associated with a lower threshold of 90% adherent days. However, using less than four strategies was associated with suboptimal adherence. WWH using a greater number of individual strategies were also using a greater number of strategy types. These findings suggest that no particular combination of strategies need be imposed on WWH, but rather using a greater variety of individual adherence strategies and types of strategies contributes to better adherence. It is possible that some strategies are utilized as backups in moments where routine or other preferred types of strategies are not sufficient to maintain adherence when disruptions in daily life occur. As such, HIV care medical case managers and other medical professionals providing adherence counseling should be familiar with a wide variety of strategies that may interest diverse populations of WWH (U.S. Department of Health & Human Services, 2022) and not just those strategies in the spotlight of current research. Adherence counselors should work with clients in care to identify four or more strategies that would be helpful to them. This counseling should be individual, patient-centered, and based on personal circumstances and preferences.

The current study builds on previous research by exploring a more comprehensive set of individual and contextual factors relevant to the diverse experiences of WWH. This population faces increased difficulties in maintaining adherence due to a variety of intersecting demographic and psychosocial factors. For example, compared to the non-Hispanic Black, Haitian, and other race/ethnicity participants, Hispanic participants reported greater diversity in their strategies to maintain ART adherence including a greater number of both individual strategies and strategy types as well as the use of less endorsed strategy types like cell phone strategies. Foreign born participants also reported more diverse use of strategies than US born participants, but this is likely due to the large proportion of Hispanic participants in this group; fully exploring this relationship is beyond the scope of this study. Characteristics associated with greater social vulnerability like nondisclosure of HIV status, HIV-related stigma, and low social support were identified as barriers to using a greater variety of strategies. Experiences with housing instability and mental health difficulties, however, were unexpectedly associated with diverse strategy plans, potentially reflecting efforts to maintain stability during particularly disruptive life experiences. Younger participants who were between 22 and 34 years old were more likely to use a greater number of strategy types including analog and cell phone strategies. Low endorsement of cell phone strategies is likely explained by the large percentage of participants ages 50 years and older in the sample. Age differences may also explain racial/ethnic disparities in the use of cell phones considering Hispanic participants were on average younger than non-Hispanic Black and Haitian participants. However, it is unclear why analog strategies were less popular among older participants, and why younger participants reported poorer adherence despite diverse use of strategies. Overall, these findings indicate that clinical work with WWH should consider and value individual needs and preferences in decisions to use strategies to improve adherence to self-managed ART regimens.

Limitations

This study is not without limitations. Due to the cross-sectional and secondary nature of these data, causality between factors cannot be determined. The complexity of ART regimens (e.g., number of medications/doses) was also unmeasured. Further, there was potential for selection bias as participants were recruited with convenience sampling. For example, those who consented to participate might be more engaged in care than those who chose to not participate. Likewise, these findings may not be generalizable to WWH not in care or those who are not clients of the Ryan White Program in Miami-Dade County. Another limitation of our data is the large number of “not applicable” responses in the disclosure of HIV status items. For example, 206 participants responded “not applicable” to the item assessing disclosure to intimate partners because they did not have any intimate partners. To address this issue, we used a combined variable that represented disclosure to at least one person. However, non-disclosure among our sample was overall uncommon. Future research among women with HIV should aim to also recruit women who have difficulties with adherence, are not in care, or have not disclosed their status to others. Lastly, data were collected during a heightened period of restrictions due to the COVID-19 pandemic. As such, COVID-19 was a unique context with potential adverse impacts on psychological functioning, social functioning, and HIV outcomes among WWH.

Conclusions

Diversity in one’s “ART toolbox” of strategies may provide the flexibility needed to better incorporate self-managed ART regimens into the lives of WWH. In pursuit of optimizing the HIV care continuum for adults and attaining targets for ending the HIV epidemic in the US (International Advisory Panel on HIV Care Continuum Optimization, 2015; Office of AIDS Research, 2021; The White House, 2021), medical case managers and other medical professionals providing adherence counseling must educate clients on the many self-management strategies available to them. For WWH who are struggling with adherence or demonstrating unsuppressed viral loads, those providing adherence counseling can assist clients with building a system of adherence strategies by recommending a natural time-of-day or activity-associated routine strategy and other strategies that complement clients’ circumstances and current needs. Further, regular assessments of adherence to HIV treatment should incorporate evaluation of barriers to ART adherence in order to provide collaborative and constructive adherence support, and to adapt the client’s adherence system as psychosocial disruptions and difficulties emerge. Support for WWH should utilize a ground-up approach, engage clients in conversations about preferences for strategies, counsel clients that multiple strategies contribute to better adherence, and link necessary social and economic resources and mental health services. Decision trees and other decision support tools can assist those providing adherence counseling in mapping out barriers to adherence and choosing strategies that best meet the needs of clients.

Data Availability:

Data are not publicly available.