Unless effectively treated, people with HIV infection suffer a long period of illness after progressing from asymptomatic infection to immune deficiency disease and death. Although there are well-established immunologic, virologic, and clinical markers or predictors of HIV disease progression, self-reported health-related quality of life (HRQoL) measures are less well understood, particularly among those patients with advanced immune disease.1 Generic measures of HRQoL, such as the SF36, SF12, Health Utilities Index Mark (HUI) 3, and EQ-5D, and disease-specific measures, such as the Medical Outcomes Study HIV Health Survey (MOS-HIV), have been used in a number of studies and found to be both reliable and valid.2-9 However, the focus in clinical trials and in medical health care has primarily been on AIDS events and death and their impact on HRQoL. To the extent that most trials with clinical outcomes in persons with HIV are powered based on these 2 outcomes, concurrent HRQoL measurements were generally accepted as measuring the impact of AIDS events and death. The direct or independent impact of non-AIDS serious adverse events (SAEs) on MRQoL has not been studied.
Increasingly effective therapeutic drugs and combination treatment regimens for HIV infection, such as highly active antiretroviral therapy, have led to increased survival rates and reduced the incidence of HIV-related complications over the past decade.10,11 There is increasing appreciation that comorbidities and adverse events that may be serious and do not specifically or directly result from progressive HIV immunodeficiency are now the most common and greatest burden of disease in persons treated for HIV infection.12,13 This recognition coincides with a change in view of HIV infection from a progressive fatal disease to a medically manageable chronic condition. As with other treatable chronic diseases,14 greater attention needs to be focused on global HRQoL issues and their attendant implications for the long-term care of HIV-positive individuals, their social adjustment to the illness, and their interaction with health care providers.
HRQoL refers to how well a person functions and to his or her perceptions of well-being in the physical, mental, and social domains of life.15 Assessing HRQoL is useful for documenting the burden of chronic disease, tracking changes in health over time, and comparing the overall effects of treatments. Although the efficacy and provision of medical treatments have prevented or delayed disease and prolonged survival in HIV-infected individuals, the duration of treatment effectiveness is sometimes limited, toxicities and side effects of HIV treatment are common, and the interaction with treatment concurrent comorbidities is increasingly complex.16 Measuring HRQoL is accepted as capturing the overall impacts of interventions on patients' functioning and well-being, which goes beyond the interpretability and clinical meaning of immunologic and virologic markers.17 Assessment of HRQoL in patients with HIV disease, treatment, and comorbidities is of increasing importance, as it is one of the only methods of reconciling the risks and benefits of prolonged therapies against a complex background of diverse morbidity.
Many studies have reported information about HRQoL in patients with HIV,18-22 and some have investigated the association between treatment activity outcome measures [primarily plasma viral load (pVL) and CD4 T-lymphocyte count] and HRQoL. These studies have shown that lower pVL and higher CD4 count are significantly associated with improvement in HRQoL.2,22-24 In serially and unsuccessfully treated patients with advanced immune deficiency, the relationship of HIV treatment activity measures to HRQoL is less clear, especially where treatments may have greater toxicities due to multiple drugs, dose and exposure factors, and host factors in drug tolerance.
AIDS events are known to have significant impact on patients' HRQoL2,22 and are primary outcomes in clinical trials evaluating patients with HIV/AIDS. Continuous virologic suppression and preservation or reconstitution of immune function remain the primary goals in treating patients with HIV/AIDS. Achievement of these treatment goals is strongly correlated with prevention of AIDS-defining opportunistic infections and malignancies, which also present as SAEs with variable impact. AIDS events are often multiple or serial, and for patients who have experienced 1 AIDS event, the impact of developing additional AIDS events on HRQoL is not well studied. Furthermore, the frequency, magnitude, and duration of the impact of non-AIDS SAEs on HRQoL have not been studied.
Our objectives were to investigate the impacts of AIDS events and non-AIDS SAEs on HRQoL and to compare the magnitude and duration of effect of these 2 types of events among patients with advanced HIV/AIDS. We hypothesized that, in addition to the HRQoL impacts of AIDS events and death, non-AIDS SAEs would also significantly impact HRQoL. Our goal was to improve the understanding of the determinants of changes in HRQoL for the benefit of both research methodologies (eg, informing sample calculations in HIV trials) and clinical management of patients with late-stage HIV/AIDS.
We used data from OPTIMA (OPTions In Management with Antiretrovirals), a recently completed multinational (United States, Canada, and United Kingdom), randomized, open, control, clinical management trial of treatment strategies for patients with advanced HIV disease for whom standard antiretroviral therapies (ARTs) have failed25 and for whom treatment prospects were limited. The trial evaluated the clinical effects of (1) retreatment intensification to 5 or more anti-HIV drugs, mega-ART, compared with standard-ART consisting of 4 or fewer anti-HIV drugs and (2) a 3-month anti-HIV treatment interruption under clinical observation compared with no interruption before treatment reinitiation. Patients were eligible to participate if they (1) were HIV positive; (2) were at least 18 years old; (3) had experienced treatment failure with at least 2 different multidrug regimens, including drugs of each nucleoside reverse transcriptase inhibitor, nonnucleoside reverse transcriptase inhibitor, and protease inhibitor classes, or laboratory evidence of resistance to drugs in each of the 3 classes; (4) had at least 3 months of current ART and were still on treatment; (5) had CD4+ T-cell counts less than 300 cells per cubic millimeter; and (6) had pVL greater than 5000 copies per milliliter (by Roche Amplicor, v1.0) or ≥2500 copies per milliliter (by bDNA: Bayer v3.0/Chiron v3.0 or polymerase chain reaction: Roche Amplicor Monitor/COBAS v1.5). Exclusion criteria included (1) pregnancy, breast-feeding, or planned pregnancy; (2) serious uncontrolled major opportunistic infection within 14 days of screening; (3) likelihood of poor protocol follow-up or if mega-ART was not feasible due to significant intolerance of many ART drugs; and (4) any medical condition or current medication that would contraindicate anti-HIV treatment as allocated in the trial. All eligible patients provided informed consent. The study was opened for enrollment on June 15, 2001, and follow-up ended on December 31, 2007. The median follow-up was 3.96 years. Because the primary focus of the present report was to compare the overall HRQoL impact of AIDS-defining events (ADEs) vs. non-AIDS SAEs, the analyses were conducted without regard to treatment allocation in the study.
Data collected at screening included demographic information (date of birth, sex, and ethnicity), viral load (VL), and CD4 cell counts. Evaluations were then performed at baseline or randomization for all patients and then at 2, 6, 12, and every 12 weeks onward for patients allocated to uninterrupted anti-HIV treatment and at 2, 6, 12, 14, 18, 24, and every 12 weeks onward for patients allocated to the 3-month anti-HIV interruption and treatment reinitiation at 12 weeks. Clinical data; laboratory data (including VL and CD4 cell count); interim history (including signs and symptoms of HIV diseases), adverse events; and their severity, medications, and HRQoL measures (including MOS-HIV, EQ-5D, and HUI3) were recorded at each scheduled visit. In addition, at 2- and 6-week “safety visits” after randomization or treatment reinitiation, VL and CD4 count data were collected. With this and other clinical information, the treating physician and the patient would consider changes to the anti-HIV treatment interruption or the treatment itself (within the standard- or mega-ART allocation) as deemed clinically appropriate.
Health-Related Quality of Life
HRQoL was evaluated with both a disease-specific measure, the MOS-HIV, and the generic preference-weighted EuroQol (EQ-5D) and Health Utility Index (HUI3) HRQoL measures.
The MOS-HIV is a brief comprehensive health status measure that has been used extensively in studies of HIV/AIDS. It is one of the first disease-specific measures available for this population and is widely used in clinical trials and evaluation studies. The questionnaire consists of 30 questions, which assess 10 dimensions of health including general health perceptions, pain, physical functioning, role functioning, social functioning, mental health, energy/fatigue, cognitive function, health distress, and QoL. In addition, 1 item assesses health transition. The MOS-HIV also yields 2 summary scores: the physical health summary score (PHS) and the mental health summary score (MHS). Higher scores on all scales indicate better functioning. This self-administered instrument asks patients to recall their health status during the past 4 weeks and takes approximately 5 minutes to complete. The MOS-HIV has been validated in several previous studies for this population.2-6
The EQ-5D was developed to describe and value current health status according to societal preferences.26 It consists of 5 dimensions and a visual analog scale. The 5 dimensions measure mobility, self-care, usual activity, pain/discomfort, and anxiety/depression, with each dimension having 3 levels. Different combinations of responses for the 5 dimensions are weighted using preferences identified by the general population. This self-administered instrument is cognitively simple, taking only approximately 1 minute to complete, and can be self-administered in several forms. Although it is prone to ceiling effects in general population studies, the psychometric properties of the EQ-5D have been shown to be robust in many disease areas.15 In HIV/AIDS, the EQ-5D has been assessed for validity in a previous study.8 We employed the US preference weights in our analysis.27
The Health Utilities Index (HUI) is a generic, preference-scored, comprehensive system for measuring HRQoL.28 The HUI questionnaire asks respondents to recall their health status in the past week when rating each of its attributes. It has evolved into 2 currently available scoring systems: the HUI2 and HUI3. We focused on the HUI3, which has 8 attributes (vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain) with 5 or 6 levels per attribute. Scoring functions have been developed from random samples of the general population in Canada. Extensive evidence supporting the reliability and validity of the HUI2 and HUI3 in a variety of diseases has been published.29 Although there is no validation study of the HUI in patients with HIV/AIDS, psychometric evidence that supports the use of HUI in this population is emerging.30
ADEs and Non-AIDS SAEs
Based on the modified 1993 US Centers for Disease Control and Prevention definitions (Morbidity and Mortality Weekly Report), ADEs were assessed clinically and adjudicated using well-established evaluation guidelines by an independent endpoint review committee that was blinded to the patients' treatment allocation. Date of diagnosis was recorded for each of the identified AIDS events.
According to the International Conference on Harmonization Harmonized Tripartite Guidelines for Clinical Safety Data Management: Definitions and Standards for Expedited Reporting, the definition of a non-AIDS serious adverse event (SAE) in the OPTIMA trial was an event that did not meet the definition of an ADE as described above and: (1) resulted in death, (2) was life threatening, (3) caused or prolonged hospitalization, and (4) resulted in persistent or significant disability or incapacity and other events considered serious by the investigator(s).
All study-related clinical events were codified, on a blinded basis, according to the standardized Medical Dictionary for Regulatory Activities (MedDRA).31 Data collection distinguished between ADEs and SAEs but did not distinguish between SAEs related to the underlying HIV immune deficiency and non-HIV events with no association. All non-AIDS SAEs were therefore reviewed by a clinical panel (S.T.B and D.W.C.) that was blinded to both randomization and specific anti-HIV treatment. Where the nature of the SAE suggested a possible association with antiretroviral medication, the panel could request confirmation of whether or not a specific drug was part of the regimen before assigning a drug- or treatment-related toxicity. Non-AIDS SAEs collected in the trial were classified as (1) HIV-related SAEs, immunologic consequences of HIV infection: non-ADEs that were likely to be due to HIV immune deficiency such as a single episode of bacterial bronchopneumonia or uncomplicated oropharyngeal candidiasis; (2) HIV-related SAEs, side effects or toxicities of anti-HIV medications: events related to ART such as drug reaction and metabolic toxicity; (3) non-HIV-related SAEs, such as motor vehicle accidents or hospitalization for thromboembolism or urinary infection considered definitely not to be related to HIV; and (4) indeterminate events that could not be ascribed to any other category.
In addition to ADEs and SAEs, we also incorporated the following baseline information on each patient: age, sex, ethnicity, employment status, HRQoL, CD4 cell counts, and pVL. Improvements in CD4 cell count and pVL were also coded at each follow-up visit if a new test score had been reported within 4 weeks of the assessment. The time to death was incorporated into our analysis by way of an indicator variable equal to 1 when an assessment was within 90 days of mortality.
To test our hypothesis, we needed to define time intervals from the event onset date (SAE) or diagnosis date (ADE) and HRQoL assessment date to identify and compare the HRQoL impact of events before diagnosis, immediately after diagnosis, and long after diagnosis. We therefore conducted an exploratory analysis using smoothing splines,32 which suggested that the HRQoL impact of an event began approximately 4 weeks before date of onset or diagnosis, lasted about 8 weeks, and was followed by an 8-week period in which the effect was still present but notably weaker.
Although patients experienced events throughout follow-up, there was a predetermined schedule for assessments. Four dichotomous variables based on the time between an event onset and assessment date were created to capture and compare the effects of events on HRQoL: (T1) if the assessment date came within 4 weeks before an onset/diagnosis date, (T2) if the onset/diagnosis date of at least 1 event fell within 8 weeks after an assessment date, (T3) if the onset/diagnosis date of at least 1 event fell between 8 and 16 weeks after an assessment date, and (T4) if the onset/diagnosis date of at least 1 event fell more than 16 weeks after an assessment date. These interval variables (T1-T4) were constructed separately for both SAEs and ADEs; however, they were not mutually exclusive-some assessments were affected by multiple events, which may have occurred in more than 1 interval.
Linear regression with generalized estimate equation method was used for both univariate and multivariate comparisons, and HRQoL measure was the outcome of interest. Age, sex, ethnicity, employment, baseline HRQoL value, baseline CD4 cell counts, baseline pVL, AIDS, non-AIDS SAE, CD4 (an increase of 50 counts/mm3 from randomization) and pVL improvements (a decrease of 0.5 log10 copies/mL), time to death (an indicator variable equal to 1 if the individual experienced mortality within 90 days after an assessment), and time since randomization were treated as covariates. For the univariate comparisons, we fitted regression models with only 1 categorical covariate as independent variable each time. Then, we estimated the mean HRQoL measure at each category and its 95% confidence interval (CI). In multivariate comparisons, all covariates mentioned above were entered in the model as independent variables.
The Z statistic was used for post hoc comparison to test the hypothesis that the impact of SAE on QoL was at least as large as the impact of AIDS events in each time interval. All analyses were executed using SAS version 9.1.
A total of 368 patients were included in the study. Patient characteristics at baseline/randomization are presented in Table 1. The mean age was 48.0 (SD = 8.5) years, 98.1% were male, 48.9% were white, and 64.4% were on social assistance or unemployed. The mean CD4 count was 127 cells per cubic millimeter (median: 107.0), and the mean log10 VL was 4.7 copies per milliliter (median: 4.8). For HRQoL measures, the mean scores of MOS-HIV PHS and MOS-HIV MHS were 41.6 and 44.7, respectively. The mean scores of the generic measure HUI3 and EQ-5D were 0.59 and 0.77. The median follow-up time was 3.96 years. A total of 240 patients (65.2%) had at least 1 non-AIDS SAE, 98 (26.6%) had at least 1 ADE, and 128 (34.8%) died during study follow-up. Among 4864 who completed follow-up assessments, the percentage of missing values among the HRQoL measures ranged from 7.8% to 9.1%.
Table 2 describes the characteristics of the observed SAEs, whereas Table 3 describes the characteristics of AIDS events. A total of 821 SAEs and 147 ADEs were observed during the study period, resulting in event rates of 60.96 and 10.91 per 100 patient-years of follow-up, respectively. The median duration of the SAEs (resolution date-onset date) was 8 days (interquartile range 3-22), 80.51% of these events resulted in hospitalization (event rate resulting in hospitalization: 49 per 100 patient-years of follow-up) and 13.88% were fatal or life threatening. A total of 240 patients (65.21%) suffered an SAE, with 110 (45.83%) of these patients suffering more than 2 events. The most common SAE diagnoses were for “infections and infestations” [247 (30.16% of events)], gastrointestinal disorders [88 (10.74%)], and blood and lymphatic system disorders [52 (6.35%)].
Ninety-eight patients (26.63% of those enrolled) suffered an AIDS event, with 11 (11.22%) of these patients suffering more than 2 events. The most common diagnoses were for esophageal candidiasis (40, 27.2%), pneumocystis pneumonia (25, 17%), and disseminated Mycobacterium avium complex (11, 7.5%). The 10 most common System Organ Classes for SAEs and diagnoses for ADEs are listed in Tables 2 and 3.
Table 4 reports the unadjusted mean scores of various HRQoL measures for different groups. On average, HRQoL scores were lower if an event occurred within any of the 4 time intervals around the assessment date. The HRQoL impact was largest for both SAEs and ADEs in T2 for each HRQoL measure and was gradually smaller in the T3 and T4 intervals. An exceptional case was the effect of ADEs on EQ-5D utility; in this case, the HRQoL impact of an ADE grew over time. Furthermore, SAEs occurring in the T1 interval also had statistically significantly lower HRQoL in comparison to cases in which no event occurred in this interval (P < 0.01 for each instrument).
Improvements in CD4 count and/or pVL accompanied improvements in HRQoL scores; however, the magnitude of these effects was small. In addition, patients within 90 days of death reported much lower mean HRQoL values than those assessed at other periods [eg, MOS-PHS: 36.5 (34.8-38.2) vs. 42.1 (41.0-43.1)].
Results of the final multivariate linear regression analyses are shown in Table 5. After adjusting for sociodemographic characteristics (age, ethnicity, and employment status) and baseline clinical information (CD4 count and pVL), we found that there was an immediate, statistically significant (at the α = 0.05 level), negative impact of both SAEs and ADEs on HRQoL scores. This finding was consistent across different instruments with the exception of the ADE impact on EQ-5D. For example, SAEs had a significant impact on mean MOS-HIV PHS and MHS scores [−4.70 (95% CI: −5.88 to −3.52); −3.5 (−4.66 to −2.34)], mean EQ-5D score [−0.06 (−0.08 to −0.04)], and mean HUI3 score [−0.10 (−0.14 to −0.07)]. On the other hand, ADEs diagnosed in the stated time frame impacted mean MOS-HIV PHS and MHS scores [−5.69 (−8.00 to −3.38); −5.24 (−7.33 to −3.15)], the mean EQ-5D score [−0.04 (−0.11 to 0.04)], and the mean HUI3 score [−0.09 (−0.14 to −0.04)].
For both event types and each instrument, we can infer from our models that HRQoL began deteriorating before the onset date of the event (T1), was at its lowest in the immediate period (T2), and then improved in the following 8-week period (T3) and thereafter (T4). In some cases, HRQoL detriments in the T3 period were not statistically significantly different from zero (ADE: MOS-PHS, EQ-5D). In most cases [exception: MOS-PHS (SAEs)], events occurring >16 weeks before the assessment date (T4) did not have a statistically significant impact on HRQoL. For example, in the regression model on the HUI3 measure, SAEs diminished HRQoL. The effect was −0.09 (−0.14 to −0.05) if the assessment preceded even the 4 weeks or less. The largest impact, −0.10 (−0.14 to −0.07), was in assessments that occurred at the time of the event. The effects were less in assessments in the 8 weeks that followed the event [−0.03 (−0.06 to 0.00)] and were not statistically significantly different from zero afterward [−0.01 (−0.03 to 0.02)].
The HRQoL impact of ADEs and SAEs across the 4 time intervals for the HUI3 is illustrated in Figure 1. Whereas the HRQoL impact of SAEs is larger in T1 and T2, the effect of ADEs is more persistent, with larger magnitudes in T3 and T4, although the result was not significant for T4 (P = 0.10). This rank ordering in estimated effects follows a similar pattern for the MOS-MHS (ADE: T2 > SAE: T2) and EQ-5D (estimated effects in T3 are equal); however, the effect of ADEs is always larger for the MOS-PHS. Furthermore, despite larger effect sizes, statistical significance was not achieved for ADEs in many intervals, most notably for the EQ-5D, in which ADEs only had a statistically significant effect on HRQoL in the T4 interval (P = 0.03).
Post hoc comparisons showed that SAE had an immediate impact at least as large as AIDS across all HRQoL measures [Z statistic-MOS-PHS: −0.81 (P = 0.21), MOS-MHS: −1.52 (0.06); EQ-5D: 1.21 (0.89); HUI3: 0.38 (0.65)]. HRQoL impacts of SAE were also not statistically significantly smaller than ADEs in T1 and T3; however, in both the MHS [−2.02 (0.02)] and HUI3 [−2.20 (0.01)], SAEs occurring >16 weeks before the assessment date (T4) had statistically significantly smaller impact on HRQoL.
We found that baseline CD4 counts and pVL had no impact on HRQoL scores. However, improvements in CD4 counts and/or pVL were significantly associated with higher HRQoL scores (MOS-PHS: P < 0.01; MOS-HIV: P < 0.01; EQ-5D: P = 0.08; HUI3: P < 0.01). Finally, mortality within 90 days of the assessment date was significantly associated with lower HRQoL scores, controlling for other covariates.
Our analysis of the impact of ADEs and SAEs on HRQoL among patients with late-stage HIV/AIDS can be summarized in the following important findings: First, ADEs occurred less frequently than SAEs in our sample of patients having both late-stage disease and limited anti-HIV treatment options. Second, although both ADEs and SAEs had significant negative impacts on HRQoL, SAEs had at least as large an immediate magnitude of detrimental effect upon HRQoL as ADEs when both were included in a multivariate linear regression model, controlling for other covariates. Third, although the immediate impacts of ADEs and SAEs were similar, we found evidence that ADEs carry a larger long-term HRQoL impact than SAEs.
Our finding of SAEs being more frequent than ADEs is supported in the literature12,33; however, the frequency of SAEs was higher in this study than previously reported. Our findings of high frequencies of SAEs requiring hospitalization can be contrasted with a study by Betz et al,33 who considered the hospitalization patterns using the HIV Research Network database. With 368 patients followed for an average of 3.66 years, in our study, the rate of SAEs, 49 per 100 patient-years, was much higher than the reported 31.3 per 100 patient-years in Betz et al. Furthermore, we found the median length of stay for hospitalization of SAEs to be somewhat longer than that reported for all hospitalizations of patients in the Healthcare Cost and Utilization Project database in 2000.34 These findings highlight the increased incidence and duration of SAEs in particular, among patients with advanced HIV/AIDS.
The magnitudes of the immediate effect of both ADEs and SAEs were clinically important at least among the EQ-5D and HUI3. Although the calculation of thresholds for minimally clinically important differences is somewhat controversial, changes in HUI3 and EQ-5D scores of 0.03 have been found to be clinically important30,35,36 using anchor-based methods. As such, the HRQoL impacts in the T1 and T3 intervals were also clinically significant; however; only ADEs had clinically significant HRQoL impacts for the EQ-5D and HUI3 instruments for the T4 interval. These findings underline the more persistent effect of ADEs on HRQoL.
Furthermore, although the magnitudes of the effects of ADEs and SAEs on HRQoL were in the same range as that reported by previous studies,8,37 some differences were found among the various HRQoL measures assessed in the OPTIMA study. In particular, ADEs had larger immediate (T2) impacts on the MOS-PHS and MHS, but smaller impacts on the preference-weighted measures (EQ-5D, HUI3), than SAEs. The fact that statistical significance was not achieved for ADEs in many intervals is primarily due to low event rates across intervals, and, in the case of the EQ-5D, a short instrument recall period, which was prone to greater volatility during acute events.38 Although this finding is tempered by the fact that the CIs in the MOS estimates for SAEs fall within the upper and lower bounds of the ADE estimates, this raises the question of responsiveness of the preference-weighted measures in comparison to the MOS-HIV measures. Wu et al8 found that the EQ-5D was less responsive than the MOS scales in detecting change due to adverse events (both ADEs and SAEs), a result for which we have demonstrated support. However, the measurement characteristics of the HUI3 have not yet been studied in this population. In any case, we are confident in stating that the immediate impact of ADEs on HRQoL was no greater than those of SAEs. Further study on the validity and responsiveness of the HUI3 in comparison to the MOS-HIV and EQ-5D measures is needed.
Improvement in CD4 (at least 50 cells/mm3) over the follow-up period had a significant and positive impact on HRQoL, which is supported in the literature,8,39 and supported the notion that efforts to improve patients' CD4 were also likely to improve HRQoL. Although improvements in pVL (0.5 log10 copies/mL) also had significant and positive impacts on HRQoL, the magnitude of the effect of pVL improvements was smaller than that of CD4 improvements, which is also consistent with previous findings.8
Others studies on the QoL of patients with HIV/AIDS have found evidence that the duration of HIV infection is negatively correlated with HRQoL.40 Our multivariate models also controlled for time to death (mortality within 90 days of assessment date), which had a statistically significant and negative impact upon HRQoL, independent of change in CD4/VL, ongoing ADEs and SAEs, and a time trend variable. This significant negative effect, independent of time trend, marks the accelerated pattern of decline in QoL in the final stages of life. This result provides a temporal dimension to findings of HRQoL being a predictor of survival among patients with advanced HIV.41,42
We tested the robustness of our results in several ways. First, we excluded improvements in CD4/pVL in an alternative model formulation due to the possible correlation with ADEs (but not SAEs), which may bias the coefficients on ADEs toward the null, thus distorting the comparison with SAEs. Exclusion of these variables had little effect on the magnitudes of the coefficients on AIDS events, and the rank ordering of AIDS event and SAE coefficient magnitudes did not change. Second, it is possible that missing study assessments due to clinical events may have biased our results if the proportion of missing HRQoL values differed between AIDS events and SAEs. We found that the frequencies of missing HRQoL values between SAEs and AIDS events were comparable (AIDS events: 16.7% vs. SAEs: 15.8%).
Our study has limitations. First, our patients were selected in the context of a randomized control trial, which raises the question of whether our patient sample was externally valid, given the stringent inclusion and exclusion criteria of the trial. Patients enrolled in OPTIMA were slightly older than patients included in an advanced HIV study by Jacobson et al,42 and a higher percentage were males. Given the high rate of mortality, we are confident, however, that a sample representative of the population of patients with late-stage HIV/AIDS was chosen. Second, the final regression models did not control for the allocated trial intervention; the effect of this covariate was not statistically significant in preliminary models, and its omission affected neither coefficient estimates nor the comparisons of the relative impacts of ADEs and SAEs. Finally, we did not attempt to further categorize and compare event types by severity as a consistent severity grading scheme across event types was not available in the context of the clinical trial. SAEs and ADEs were heterogeneous groups of events; our objective was to simply compare HRQoL impacts of these 2 event types to highlight the importance of SAEs in advanced HIV/AIDS. As our event rate was relatively high, it is likely that a representative sample of events by severity level was captured, thus providing us with consistent and unbiased parameter estimates.
As in previous studies, we found that SAEs are far more common than ADEs among patients with late-stage HIV/AIDS. Furthermore, the results of our study suggest that SAEs occurring in patients with late-stage HIV/AIDS have at least as important an immediate impact on patient HRQoL as do ADEs; however, the effect of ADEs tends to be more persistent. To date, much greater emphasis has been placed on ADEs as signaling disease progression and deterioration of patient QoL. In this context, our findings call for a greater emphasis on the detection and active prevention of SAEs in patients with late-stage HIV/AIDS. Exclusive selection of ADEs and death as the primary outcome for studies of clinical interventions may not completely capture the most meaningful clinical outcomes.
1. Kudel I, Farber SL, Mrus JM, et al. Patterns of responses on health-related quality of life questionnaires among patients with HIV/AIDS
. J Gen Intern Med
. 2006;(Suppl 5):S48-S55.
2. Jia H, Uphold CR, Wu S, et al. Predictors of changes in health-related quality of life among men with HIV infection in the HAART era. AIDS Patient Care STDS
3. Chan KS, Revicki DA. Changes in surrogate laboratory markers, clinical endpoints, and health-related quality of life in patients with the human immunodeficiency virus. Eval Health Prof
4. Lubeck DP, Fries JF. Changes in quality of life among persons with HIV infection. Qual Life Res
5. Carretero MD, Burgess AP, Soler P, et al. Reliability and validity of an HIV-specific health-related quality of life measure for use with injection drug users. AIDS
6. Paton NI, Chapman CAT, Chan S-P, et al. Validation of the Medical Outcomes Study HIV Health Survey as a measure of quality of life in HIV-infected patients in Singapore. Int J STD AIDS
7. Wu AW, Revicki DA, Jacobson D, et al. Evidence for reliability, validity and usefulness of the Medical Outcomes Study HIV Health Survey (MOS-HIV). Qual Life Res
8. Wu AW, Jacobson DL, Frick KD, et al. Validity and responsiveness of the EuroQol as a measure of health-related quality of life in people enrolled in an AIDS clinical trial. Qual Life Res
9. Bayoumi AM, Redelmeier DA. Economic methods for measuring the quality of life associated with HIV infection. Qual Life Res
10. Collier AC, Coombs RW, Schoenfeld DA, et al. Treatment of human immunodeficiency virus infection with saquinavir, zidovudine, and zalcitabine. AIDS Clinical Trials Group. N Engl J Med
11. Deeks SG, Smith M, Holodniy M, et al. HIV-1 protease inhibitors. A review for clinicians. JAMA
12. Reisler RB, Han C, Burman WJ, et al. Grade 4 events are as important as AIDS events in the era of HAART. J Acquir Immune Defic Syndr
13. Braithwaite RS, Justice AC, Chang CC, et al. Estimating the proportion of patients infected with HIV who will die of comorbid diseases. Am J Med
14. Strand V, Singh JA. Improved health-related quality of life with effective disease-modifying antirheumatic drugs: evidence from randomized controlled trials. Am J Manag Care
15. Coons SJ, Rao S, Keininger DL, et al. A comparative review of generic quality-of-life instruments. Pharmacoeconomics
16. Hammer SM, Eron JJ Jr, Reiss P, et al. International AIDS Society-USA. Antiretroviral treatment of adult HIV infection: 2008 recommendations of the International AIDS Society-USA panel. JAMA
17. Wilson IB, Cleary PD. Linking clinical variables with health-related quality of life. A conceptual model of patient outcomes. JAMA
18. Wachtel T, Piette J, Mor V, et al. Quality of life in persons with human immunodeficiency virus infection: measurement by the medical outcomes study instrument. Ann Intern Med
19. Low-Beer S, Chan K, Wood E, et al. Health related quality of life among persons with HIV after the use of protease inhibitors. Qual Life Res
20. Bing EG, Hays RD, Jacobson LP, et al. Health-related quality of life among people with HIV disease: results from the Multicenter AIDS Cohort Study. Qual Life Res
21. Campsmith ML, Nakashima AK, Davidson AJ. Self-reported health-related quality of life in persons with HIV infection: results from a multi-site interview project. Health Qual Life Outcomes
22. Call SA, Klapow JC, Stewart KE, et al. Health-related quality of life and virologic outcomes in an HIV clinic. Qual Life Res
23. Weinfurt KP, Wukkje RJ, Glick HA, et al. Relationship between CD4 counts, viral burden, and quality of life over time in HIV-1-infected patients. Med Care
24. Gill CJ, Griffith JL, Jacobson D, et al. Relationship of HIV viral loads, CD4 counts, and HAART use to health-related quality of life. J Acquir Immune Defic Syndr
25. Kyriakides TC, Babiker A, Singer J, et al. An open-label randomized clinical trial of novel therapeutic strategies for HIV-infected patients in whom antiretroviral therapy has failed: rationale and design of the OPTIMA trial. Control Clin Trials
26. Brooks R. EuroQol: the current state of play. Health Policy
27. Shaw JW, Johnson JA, Coons SJ. US valuation of the EQ-5D health states: development and testing of the D1 valuation model. Med Care
28. Torrance GW, Furlong W, Feeny D, et al. Multi-attribute preference functions. Health Utilities Index. Pharmacoeconomics
29. Feeny D, Furlong W, Torrance GW, et al. Multiattribute and single-attribute utility functions for the health utilities index mark 3 system. Med Care
30. Horsman J, Furlong W, Feeny D, et al. The Health Utilities Index (HUI®): concepts, measurement properties and applications. Health Qual Life Outcomes
31. Brown EG, Wood L, Wood S. The medical dictionary for regulatory activities (MedDRA). Drug Saf
32. Diggle PJ, Heagerty P, Liang K-Y, et al. Analysis of Longitudinal Data
. 2nd ed. New York, NY: Oxford University Press; 2002.
33. Betz ME, Gebo KA, Barber E, et al. Patterns of diagnoses in hospital admissions in a multistate cohort of HIV-positive adults in 2001. Med Care
. 2005;43(Suppl 9):III3-III14.
34. Hellinger FJ. HIV patients in the HCUP Database: a study of hospital utilization and costs. Inquiry
35. Grootendorst P, Feeny D, Furlong W. Health Utilities Index Mark 3: evidence of construct validity for stroke and arthritis in a population health survey. Med Care
36. Marra CA, Woolcott JC, Kopec JA, et al. A comparison of generic, indirect utility measures (the HUI II, HUI III, SF-6D and EQ-5D) and disease-specific instruments (the RAQoL and HAQ) in rheumatoid arthritis. Soc Sci Med
37. Wu AW, Jacobson D, Clark B, et al. Quality of life associated with valacylovir vs. high and low dose acyclovir for prophylaxis against cytomegalovirus in AIDS. Abstracts presented at: 4th Conference on Retroviruses and Opportunistic Infections; Washington D.C., January 1997.
38. Bansback N, Sun H, Li X, et al. The impact of recall time on the measurement of health utilities for acute events. Health Econ
. 2008. PMID: 18404664.
39. Liu C, Johnson L, Ostrow D, et al. Predictors for lower quality of life in the HAART era among HIV-infected men. J Acquir Immune Defic Syndr
40. Huanguang J, Uphold CR, Zheng Y, et al. A further investigation of health related quality of life over time among men with HIV infection in the HAART era. Qual Life Res
41. Cunningham WE, Crystal S, Bozzette S, et al. The association of health-related quality of life with survival among persons with HIV infection in the United States. J Gen Intern Med
42. Jacobson DL, Wu AW, Feinberg J. Health-related quality of life predicts survival, cytomegalovirus disease, and study retention in clinical trial participants with advanced HIV disease. J Clin Epidemiol
The OPTIMA Team: Trial Management Committee: Martin Schechter, MD (cochair); Lawrence R. Deyton, MD (cochair); Janet Darbyshire, OBE, MD, FRCP (cochair); Sheldon Brown, MD; Mark Holodniy, MD; Tassos C. Kyriakides, PhD; Doug Owens, MD; Wei Yu, PhD; William Cameron, MD; Joel Singer, PhD; Aslam H. Anis, PhD; Brian Gazzard, MD, FRCP; Mike Youle, MD; Malcolm Hooker, MD; Abdel Babiker, PhD; and Mark Sculpher, PhD. Trial Steering Committee: Professor Alasdair Breckenridge (chair); Paul Volberding, MD; Mark Wainberg, PhD; Kevin Schulman, MD; Don McIver; Simon Collins; Maggie Atkinson; Peter Peduzzi, PhD (ex officio); and Isabelle Schmid, PhD (ex officio). Data Safety Monitoring Board: Dame Anne McClaren (chair) (Deceased); Professor Vern Farewell; Mary A. Foulkes, PhD; Deborah Cotton, MD, MPH; and Andreas Laupacis, PhD.