User Acceptance of HIV TIDES—Tailored Interventions for Management of Depressive Symptoms in Persons Living with HIV/AIDS

Depression and HIV

Although HIV infection and AIDS still cannot be cured, the introduction of highly active antiretroviral therapy in 1996 reduced morbidity and mortality for PLWHAs. However, it is well documented that depressed PLWHAs report more somatic symptoms, including wasting and neurologic symptoms, than do non-depressed PLWHAs. Such symptoms are associated with poor health-related quality of life, decreased functional status, and increased mortality. ^15–26 Depression is identified as the most significant mental health problem and has been reported to be prevalent in nearly 50% of PLWHAs. ^16,27–32

At the present time, no mental health-related interventions other than counseling after HIV testing are recommended by the treatment guidelines. ³³ No other mental health interventions have been included as the standard of care. Thus, it is a major challenge for clinicians to identify and provide proper care or referral for PLWHAs who are affected by depressive symptoms. Given the complexity of recurring courses of depression, plus the challenge of managing the mental health needs associated with living with HIV/AIDS as a chronic illness, it is a priority to develop an approach to identify individuals at risk of being depressed and to provide them with strategies for self management of depressive symptoms as a complement to the professional services that may be required.

Studies report that PLWHAs have a high drop-out rate from psychological counseling and support groups due to fear of stigmatization, distrust of the health care system, and lack of support for attendance. ³⁴ As a strategy for early detection of and interventions for depressive symptoms in PLWHA, HIV TIDES aims to empower users not only with increased knowledge, but also through their participation in an interactive decision making process that may enhance feelings of control in their lives. A number of studies have demonstrated that populations typically characterized as underserved can learn to use such systems and may in fact differentially benefit from their use, given appropriate design and support. ^35–37

Design of HIV TIDES

Development of the framework of HIV TIDES was guided by Social Cognitive Theory ³⁸ to provide education on key elements for managing depressive symptoms. The tailoring algorithm was based upon the results of the initial Center for Epidemiologic Studies-Depression (CES-D) assessment and the secondary assessment (Stages of Change Questionnaire, Cognitive Event Schedule, and Social Activities Questionnaire). ^39–42 Messages were developed following Kreuter’s tailoring framework concepts. ⁴³

The 20-item CES-D was the first level assessment in HIV TIDES. CES-D scores were used for the intervention pathways because the instrument is multi-factorial thus facilitating tailoring of specific self-care interventions. The instrument has sensitivity for detecting major depression between 80% and 95% when a cutoff of 16 points or greater is used. The average score for the general population is eight. The score represents the level of depressive symptoms instead of identifying specific depressive disorders. The cut-off points of 8 and 16 (total range from 0 to 60) were used to divide system users into three major groups which determined the tailoring path the users would experience (minimum, moderate, or severe level of depressive symptoms). Individuals with minimum level of depressive symptoms receive messages to reinforce their health behaviors related to physical activity, positive thoughts, pleasant social activity, and medication adherence. Those with severe level of depressive symptoms are educated with the definition of depression diagnoses and the available treatment options. They are highly encouraged to have the psychiatric evaluation done by their clinician to confirm the possible diagnosis. For individuals with moderate level of depressive symptoms, depending on the score of CES-D sub-scales, they are divided into three groups to work on increasing physical activity, practicing positive thoughts, and participating in more pleasant social activities. HIV TIDES then provides symptom education, secondary assessment, review of current status, interactive goal setting, practice planning, and the final summary of all the information provided. The users are encouraged to print the summary to keep as their personal reference and discuss it with their providers. The user can also track his or her past history with HIV TIDES every time they log into the system to review their progress.

The prototype of HIV TIDES was created with Dreamweaver MX 2004 by using HTML, JAVA script, and PHP computer language and supported by a MySQL database. The system’s compliance with the majority of usability concepts and current standards were confirmed by a heuristic evaluation and usability tests with intended end-users. ⁴⁴ HIV TIDES was iteratively refined based upon review by domain experts and the results of the usability evaluation. This included adding more content related to depression and its treatment and decreasing the readability level and word count of the tailored messages. The final prototype consists of 73 messages with an average readability level of 6.0 and word count of 103 (▶).

Technology Acceptance Model

The Technology Acceptance Model (TAM) suggests that two specific beliefs—perceived ease of use (PEOU) and perceived usefulness (PU)—determine one’s behavioral intention to use (BI) a new technology, which has been linked to subsequent usage behavior. ^45–48 It is assumed that when someone forms an intention to act, he or she will be free to act without limitation. Consequently, BI is considered as equivalent to technology acceptance in the study.

Venkatesh and Davis proposed an extension of TAM, TAM2, which incorporates additional theoretical constructs spanning social influence processes (subjective norm, voluntariness, and image) and cognitive instrumental processes (job relevance, output quality, result demonstrability, and PEOU). ⁴⁷ Later, Venkatesh (2000) published the modified TAM that was mainly focused on behavioral constructs related to PEOU. ⁴⁹ The modified model proposes control (internal and external), intrinsic motivation, and emotion as anchors that determine early PEOU of a new system. With increasing experience, it is expected that system-specific PEOU, while still anchored to the general beliefs regarding computers and computer use, will adjust to reflect objective usability, perceptions of external control specific to the new system environment, and system-specific perceived enjoyment.

Both TAM2 and modified TAM were used to build the conceptual model of HIV TIDES acceptance (▶). Two constructs in modified TAM (computer playfulness and perceived enjoyment) and four in TAM2 (voluntariness, image, output quality, and result demonstrability), were excluded due to their lack of relevance to HIV TIDES design and implementation. Sociodemographic variables, including age, education, and level of health literacy were considered variables that would have moderating effects on objective usability associated with PEOU of HIV TIDES and were added to the conceptual model.

Sample and Setting

The study received IRB approval from Columbia University Medical Center (CUMC). Users were recruited at the HIV clinic at East Harlem, New York using the IRB approved flyer. In the first two months of the pilot study, the investigator only received five subjects referred from this clinic. Consequently, an IRB modification was submitted to add an individual medical office in the Bronx to increase the number of subjects referred. Clinicians (nurse practitioners and physician assistants) in the clinic distributed recruitment flyers to their patients with HIV/AIDS. The snowball technique was also adopted. Individuals who learned about the pilot study case recruitment from their friends living with HIV/AIDS were also included. Inclusion criteria were self-identified as HIV positive and English literate. Age under 18 and pregnancy were two exclusion criteria.

Measurement

Data Collection Process

The study was conducted in the usability laboratory at CU School of Nursing. Participants spent less then ten minutes to complete all the paper-based questionnaires except for the TAM Scale before they started HIV TIDES. Those who had no prior computer experience received a brief computer skills orientation using TIDES training session. All participants worked on their own session independently. Only one participant requested assistance and that was related to clarification of a CES-D question. After the completion of the TAM Scale at the end of HIV TIDES session, a short individual debriefing session was conducted by the investigator to ensure that the user did not experience any mental distress due to the study participation. All participants completed all questionnaires. There were no missing data.

Analysis

Descriptive, correlational analyses, paired-t tests, and analyses of variance (ANOVA) were conducted using SPSS 13.0 for Windows (Chicago, IL).

Descriptive Analyses

Thirty-two PLWHAs were recruited for the study. This sample size meets minimal criteria for correlational research and was considered adequate for the pilot study. The sample was predominantly African American (AA) (69%), male (66%), in their 40s (63%), and had high school or less education (69%). More than half (56%) of the participants had only high school level health literacy. Most subjects (94%) had prior computer experience but only eight (25%) had used a computer on a regular basis. The majority (81%) had used the Internet and 66% had used e-mail before.

Participants spent an average of 10.4 minutes (SD = 5.6) completing their HIV TIDES sessions. The amount of time associated with the three tailoring paths in HIV TIDES (CES-D < 8: n = 11, Mean = 6.4, SD = 2.1; CES-D > = 8 and <16: n = 3, Mean = 24.3, SD = 0.6; CES-D > = 16: n = 18, Mean = 10.6, SD = 3.1) was significantly different (p < 0.001). Twenty-seven participants (84%) had a minimal level of depression as measured by BDI II = <13. Individuals generally reported a high level of SE-decision (Mean = 29.4, SD = 1.9) and a high level of SE-management (Mean = 44.1, SD = 6.93). Regarding medication adherence, less than half (44%) of the participants reported that they had forgotten to take their medication in the past. Few participants agreed that they were careless at times about taking their medication (n = 6, 19%), and sometimes would stop taking medication when they felt better (n = 3, 9%) or worse (n = 7, 22%). Most of the participants (n = 28, 88%) were confident or very confident of their ability to medication and have considered taking medications to be mostly or totally a part of daily routine (n = 30, 100%).

Participants reported high levels of BI (Mean = 9.5, SD = 1.2), PU (Mean = 9.5, SD = 1.2), and PEOU (Mean = 9.6, SD = 0.8) related to HIV TIDES.

Correlational Analyses

The correlational analyses focused on the relationships among BI, PU, and PEOU, as well as their relationships with other proposed factors in the conceptual model. Four factors were positively correlated with BI: PU (r = 0.61), PEOU (r = 0.61), internal control (r = 0.59), and external control (r = 0.46). Computer anxiety (r = −0.80), tailoring path (r = −0.35) and BDI II depressive symptoms (r = −0.49) showed a significant negative correlation with BI.

PU was significantly and positively correlated with PEOU (r = 0.85), internal control (r = 0.81), and SE-decision (r = 0.51) and negatively correlated with computer anxiety (r = −0.36). When the level of PEOU increased, external control (r = 0.77) increased and computer anxiety (r = −0.48) decreased. Other variables, such as SE-management, SE-adherence, and health literacy were not significantly correlated with BI, PU, or PEOU.

There were some significant differences in BI, PU, and PEOU between individuals with and without Internet experience, e-mail experience, and current computer use, as well as between AAs and non AAs (▶). Participants with Internet experience rated HIV TIDES as more useful (p < 0.05) and easier to use (p < 0.01) than those who had no prior experience. Individuals with e-mail experience reported higher levels of PEOU (p < 0.05) than individuals without e-mail experience; and those who reported current computer usage considered HIV TIDES more useful (p < 0.05) than their counterparts. AAs reported significantly higher ratings of BI (p < 0.01), PU (p < 0.01) and PEOU (p < 0.01) than non AAs (Hispanic and other).

Relationships between BI, PU, and PEOU

According to TAM, system acceptance is predicted by BI, which is related to PU and PEOU. Consistent with TAM, in this pilot study, PU and PEOU were both highly correlated with BI (r = 0.61, p < 0.01). Findings reported by Lederer et al. supported that both PU and PEOU predicted web use for work-related tasks. ⁵⁵ However, many other studies reported that only one major factor, either PU or PEOU were significantly related to BI. In studies recruiting college students, administrative staff, knowledge workers, and system developers PEOU was considered the most important factor affecting the acceptance and actual use of technology. ^47,56–57 On the other hand, PU was a strong determinant of BI but not PEOU when targeting subjects with professional backgrounds, such as physicians, and others such as online shoppers. ^58–62 Keil et al. concluded that PU is more important than PEOU in determining whether or not to use a technology especially for professionals, and explained that “no amount of ease of use will compensate for low usefulness.” ⁶³

Regarding the relationship between PU and PEOU, the results of this study indicate that there is a strong positive correlation (r = 0.85, p < 0.01). In contrast, many workplace technology adoption studies ^55,63 have found the relationship between PU and PEOU to be less significant than that with BI. Given the small sample size of this study, the investigators were not able to explore how other factors mediate the relationship between PU and PEOU.

Relationships between BI and Other Variables