Types of Sleep Problems in Adults Living with HIV/AIDS

Participants and Procedures

After approval of the study protocol by the Committee on Human Research, 350 adults living with HIV in the San Francisco Bay Area were recruited and enrolled using posted flyers at HIV-related clinical and community sites. The flyer indicated recruitment for a study about “symptoms and genetics” and was not specific to any particular symptom. Written informed consent was obtained prior to participation. Eligibility criteria included English-speaking adults at least 18 years of age, diagnosed with HIV for ≥ 3 months, not working nightshift schedules, and not pregnant within the prior 3 months. Study visits were conducted at the university's outpatient Clinical Research Center. After excluding subjects who tested positive on urine sample screening for illicit drugs (n = 31), subjects who could not provide a specimen (n = 1), and subjects with missing questionnaire or sleep diary data (n = 5) or actigraphy data (n = 27), the analysis for this sample included 290 adults with data for actigraphy and self-report measures.

In addition to completing demographic and symptom questionnaires, the assessment included wearing a wrist actigraph monitor for 72 continuous hours during weekdays, and anthropometric measures of height, weight, and circumference of neck, waist, hips, and thighs. Participants provided a urine sample for toxicology screening using RediCup (Redwood Toxicology Laboratory, Inc, Santa Rosa CA, USA) and a copy of their most recent laboratory results for CD4+ T-cell count and viral load. Anti-retroviral therapy (ART) was determined by review of the participant's report of current medications. This analysis includes self-report and actigraphy measures of their sleep in relation to demographic and clinical characteristics.

Measures

Statistical Analysis

Descriptive statistics were used to summarize demographic and clinical characteristics. The 72 h of continuous actigraphy data were analyzed for each individual using Action4 software (Ambulatory Monitoring, Inc. Ardsley, NY). This software allows for automatic sleep scoring as described in the methods section. These data were entered into SPSS, Version 15.0 software (SPSS, Inc, Chicago IL, USA) for group analyses. ANOVA with Scheffe post hoc testing was used to compare short sleepers (< 6 h total nocturnal sleep by actigraphy measures) with typical sleepers (6.0-7.9 h) and long sleepers (≥ 8 h).

Actigraphy recording of wake after sleep onset (WASO) was averaged over the 3 nights, and WASO > 15% was used to dichotomize the sample. In the uncontrolled home, without a sleep technician to document lights out time, pressing the actigraphy event marker to indicate lights can be unreliable and invalid on any given night. Therefore, the participant's diary report of time to fall asleep was averaged over the 3 nights, and sleep onset latency (SOL) > 30 min was used to dichotomize the sample. Group differences in types of sleep problems were assessed with χ² tests of independence for categorical variables and either one-way ANOVA or independent t-tests for continuous variables. Post hoc testing compared each sleep problem group to the reference group of good sleepers using either pairwise χ² tests adjusted for the number of comparisons (categorical variables) or Dunnett test (continuous variables). Transformations were conducted to normalize skewed distributions. Spearman correlation coefficients (r) were use to estimate associations between continuous variables with non-normal distributions. Significance was set at p < 0.05 for all analyses.

Sample Characteristics

Demographic and clinical characteristics for the 290 participants included in this analysis are presented in Table 1. The sample was ethnically diverse and predominantly male, reflecting the local population of adults living with HIV infection. Over half (59%) of the women in the sample were African American. Most participants had been living with HIV for many years; 70% were currently receiving anti-retroviral therapy (ART); 51% had received an AIDS diagnosis; and most were unemployed (86%) or receiving medical disability assistance (75%).

Sleep Quantity

The sample's total sleep time across the 3 nights ranged from < 2 to > 11 h/night (mean = 6.2 ± 1.6 h). Almost half of the sample (45%, n = 131) averaged < 6 h of sleep per night; another 42% (n = 121) slept ≥ 6 h, but less than 8 h; and 13% (n = 38) slept an average ' 8 h per night (Table 2).

“Long sleepers” are typically defined as adults sleeping ≥ 9 h, but in this sample, only 9 participants met this criterion, and they did not differ from the 29 participants sleeping 8-9 h per night on other variables of interest. Therefore, these 2 groups were collapsed into a single group of participants sleeping ≥ 8 h per night for the purpose of this analysis. The primary difference between participants sleeping ≥ 8 h and the groups obtaining less sleep at night was that the longer sleepers reported more depressive symptoms on the CES-D. Long sleepers also spent more time in bed at night, had better sleep efficiency, and slept more during the day, although their perceived sleep quality on the PSQI did not differ from the other groups. Long sleepers were more likely to be Caucasian and college educated, but were otherwise similar to participants sleeping less on demographic and clinical variables.

In contrast to the long sleepers, participants sleeping < 6 h/night spent less time in bed, had lower sleep efficiency, and slept less during the day than longer sleepers, but did not report significantly more sleep disturbance or worse symptom experience. However, short sleep was associated with lower CD4+ T-cell counts (398 ± 256 vs. 494 ± 269, t₂₇₅ = 3.03, p = 0.003) and higher viral load values (2.09 ± 1.99 vs. 1.58 ± 1.91, t₂₆₉ = 2.14, p = 0.033). Short sleepers were also more likely to be African American, have less education, and have lower income.

Problems Falling Asleep

Difficulty falling asleep, defined as an average SOL > 30 min, was reported by 100 participants (34% of the sample). Only 15% of the participants (n = 42) averaged < 10 min to fall asleep by their diary recording over the 3 nights. Self-reported SOL was highly variable (range 1 to 160 min) and not normally distributed (30.2 ± 26.4, median = 21.7). Diary report of SOL was correlated with the usual amount of time to fall asleep reported on the PSQI (Spearman r = 0.60, p < 0.001), indicating that the 3 nights of monitoring were typical of the participants' usual sleep onset latency during the prior month. Self-reported SOL was also weakly correlated with actigraphy estimates of SOL (Spearman r = 0.13, p = 0.025), WASO (Spearman r = 0.13, p = 0.024) and time in bed (Spearman r = 0.22, p < 0.001). Diary reports of time to fall asleep were unrelated to total sleep time at night and daytime napping. Self-reported SOL was also unrelated to reported use of alcohol, caffeine, and tobacco, as well as measures of BMI, neck circumference, or waist circumference.

Problems Maintaining Sleep

Wake after sleep onset (WASO) was calculated from wrist actigraphy as the percentage of time spent awake after falling asleep and could range from 0 (no awake time after sleep onset) to 100% (awake the entire night). Fragmented sleep or difficulty maintaining sleep was common in this sample, with 56% (n = 161) having WASO above 15%. The 3-night mean WASO ranged from 0.63% to 78.9% (mean = 20.9% ± 14.8%, median = 17.5%), and was weakly correlated with habitual sleep efficiency on the PSQI (Spearman r = −0.15, p = 0.016), suggesting that the WASO observed during the 3-day monitoring period was somewhat consistent with the participants' self-report of their typical amount of sleep fragmentation. WASO was inversely correlated with actigraphy measures of total sleep time at night (Spearman r = −0.71, p < 0.001) and daytime sleep quantity (Spearman r = −0.23, p < 0.001). There was no relationship between WASO and self-reported use of tobacco, alcohol or caffeine, or measures of BMI, neck circumference, or waist circumference.

Group Comparisons

Although there was a weak association between difficulty falling asleep and difficulty maintaining sleep (Spearman r = 0.13, p = 0.024), the 2 sleep problems were largely independent. As such, the sample was divided into 4 groups based on the presence or absence of the 2 sleep problems. There were 41 (14%) who had difficulty falling asleep only, defined as taking > 30 min to fall asleep on average as reported in the sleep diary. There were 102 (35%) who only had a problem maintaining sleep, averaging > 15% WASO recorded by actigraphy. There were an additional 59 participants (20%) who had both types of sleep problems. The reference group consisted of 88 participants (30%) who had neither problem and were considered “good sleepers” for group analyses.

The groups were compared and results are reported in Tables 1 and 3. The groups differed on demographic factors, such as race, education, income, and employment status, but did not differ by gender, age, or self-reported use of tobacco, alcohol, or caffeine. There were small clinical differences between the groups, and significant differences on measures of sleep and symptom experience. As shown in Table 3, all 4 groups had a mean PSQI score > 5 indicating significant sleep disturbance. Each sleep problem group was compared to the good sleepers and can be summarized as follows:

Trouble falling asleep only: Participants who reported trouble falling asleep but had no difficulty staying asleep (n = 41) reported significant sleep disturbance on the self-report measures, despite obtaining the most sleep of the 4 groups, both at night (7.3 ± 1.1 h) and during the day (1.3 ± 1.3 h). In fact, this was the only group that obtained more sleep according to actigraphy than they reported on the PSQI. Only 10% of the adults in this group obtained < 6 h of sleep per night, and nearly one-third (29%) obtained ≥ 8 h of sleep at night. This group also reported the highest symptom burden, the most symptoms of anxiety and depression, and the highest ratings of morning fatigue. The demographic and clinical characteristics of this group were similar to the good sleepers. Women with difficulty falling asleep had slightly lower waist circumference than good sleepers, but there was no group differences on anthropometric measures among men or the small sample of transgender adults.

Trouble staying asleep only: Participants who had trouble maintaining sleep, but no trouble initially falling asleep (n = 102) represented the largest proportion of the sample (35%). On average, this group obtained the least amount of sleep (5.3 h), although on the PSQI they reported usually obtaining 7.1 h of sleep. While their sleep efficiency was low (65%), they reported less sleep disturbance on the self-report measures than participants with trouble falling asleep, and they had the lowest symptom burden (total number of symptoms, as well as symptoms of depression, anxiety, and fatigue) of all 4 groups. Participants who only had trouble staying asleep were more likely to be African American and less likely to be Caucasian, and tended to have less education and lower incomes than the good sleepers. While the differences were not statistically significant, their CD4+ T-cell counts were somewhat lower and their viral load counts were slightly higher than the good sleepers, and they were slightly less likely to be receiving anti-retroviral therapy.

Trouble falling and staying asleep: Participants who had trouble both falling and staying asleep (20%, n = 59) obtained an average of 5.5 ± 1.4 h of sleep at night and 0.7 ± 0.6 h of sleep during the day, and had an average sleep efficiency of only 64%. In contrast to the previous group, participants with both sleep problems reported more sleep disturbance on the self-report measures and also reported more symptom burden (overall as well as symptoms of depression and anxiety specifically). This group reported a surprisingly low rate of sleep medication use (19%) given the severity of their sleep problems. This group was more likely to be African American and less likely to be Caucasian, less likely to be employed, and tended to have less education and income than the good sleepers. This group had the lowest CD4+ T-cell count and the highest viral load values, and was significantly less likely to be on anti-retroviral therapy compared to the group of good sleepers.