Sleep and Immune Function Among People Living with Human Immunodeficiency Virus (HIV)

Abstract

People living with HIV are at increased risk for sleep disturbances. Up to 75% of the HIV-infected individuals in the United States experience sleep disturbances of some kind. Previous studies have suggested an association between patient-reported sleep disturbances and impaired immune function. This study evaluates data obtained via sleep actigraphy to evaluate the relationship between objectively measured sleep, HIV viral load, and immune function. While this study found no relationship between objective sleep and CD4+ T- lymphocyte count, higher sleep efficiency was weakly correlated with lower HIV viral loads, τ_b(93) = −.165, p = .043. More research is warranted to clarify the nature of these relationships.

Introduction

Healthy sleep is essential for optimal health and plays a vital role in chronic disease prevention and management (Badr et al., 2015; US Department of Health Human Services, 2011). There is a direct, reciprocal relationship between sleep and immune function (Cardinali & Esquifino, 2012); poor sleep worsens immune function and impaired immune function affects sleep quality (Cardinali & Esquifino, 2012; Turner & Gellman, 2013). Poor sleep health is a major contributor to the development of many sorts of chronic disabilities and disorders for all Americans, and is a comorbid condition of particular concern among people living with HIV (Badr et al., 2015; US Department of Health Human Services, 2011). More extensive research on sleep is advocated by Healthy People 2020 (US Department of Health Human Services, 2011).

The introduction of combination antiretroviral therapy (cART) has caused Acquired Immunodeficiency Syndrome (AIDS)-related mortality and morbidity to decline significantly (Detels et al., 1998; Palella Jr et al., 1998). Poor sleep is a significant contributor to disease progression and AIDS-related mortality and morbidity (Manzar et al., 2017; Phillips & Gunther, 2015); and affects up to 75% of people/persons living with HIV (PLWH) (Gutierrez et al., 2019; Rubinstein & Selwyn, 1998). Studies have found increased daytime sleep (napping), increased fragmented sleep, decreased sleep efficiency, and increased sleep latency in PLWH representing changes in sleep pattern and architecture (Byun et al., 2017; Hayduk, Kellyt, & Mitler, 1995; Lee et al., 2012; Norman et al., 1988; Reid & Dwyer, 2005; Wiegand et al., 1991). Sleep disturbances among PLWH have been associated with lower CD4+ T-lymphocyte (CD4+) counts, a strong predictor of AIDS-related mortality (Seay et al., 2013).

HIV viral load, coupled with CD4+ count, are the most reliable indicators of disease progression in HIV infection (Abbas & Lichtman, 2010; Mellors et al., 1997). The CDC has categorized five stages of HIV (0–3 and unknown) based on CD4+ count and the diagnosis of AIDS-defining opportunistic infections and neoplasms (Centers for Disease Control Prevention, 2016) which generally occur once CD4+ counts fall below 200 cells/μL (Ferreira & Ceolim, 2012). CD4+ count <50 cells/μl are associated with higher rates of opportunistic infections, HIV viral loads >100,000 copies/ml is associated with increased risk for opportunistic neoplasms (Buchacz et al., 2010).

This secondary data analysis explores the relationship between sleep, HIV viral, and CD4+ count in 104 adults living with HIV in Northeast Ohio.

Methods

Participants and setting

This study was conducted using data extracted from the first time point of the 2-year longitudinal study “Exploring Relationships among Stress, Isolation, and Physical activity (TRIP) in older adults with HIV/AIDS” carried out by Dr. Allison Webel (2014). Recruitment and screening for the TRIP study occurred via the Cleveland HIV Research Residency database, an Institutional Review Board (IRB)-approved database collecting contact information on PLWH in Northeast Ohio (Webel et al., 2014). Data collection occured between September 2012 and January 2013. Inclusion criteria were: documented HIV diagnosis, on cART, and in HIV care. Exclusions criteria were: Inability to communicate in English, diabetes mellitus, and having an artificial cardiac pacemaker or defibrillator.

Measures

Sleep was measured via wrist actigraphy (Phillips Actiwatch) and corresponding sleep diary, a validated measure of objective sleep (M. T. Smith et al., 2018). Participants wore the actigraph on their wrist for seven days and completed a corresponding sleep diary yielding data on (1) sleep duration, (2) sleep fragmentation, and (3) sleep efficiency. Sleep quantity is measured by duration (sum in minutes of total sleep from onset to end), whereas sleep quality corresponds to sleep fragmentation (time awake in minutes from sleep onset to end) and efficiency (ratio of sleep duration to time in bed) (Bagai et al., 2013; Baud, Magistretti, & Petit, 2013). Sleep efficiency above 80% is considered normal for healthy adults (Walsleben et al., 2004).

Health record abstractions were used to collect the CD4+ count (cells/μL of blood) and the HIV viral load (copies/mL of blood) (Abbas & Lichtman, 2010; Mellors et al., 1997) most temporally proximate to the time of the sleep study.

Data analysis

Data from the 104 participants of the TRIP study were analyzed using the Statistical Program for the Social Sciences (SPSS) 25.0 (IBM Corp., 2017). Descriptive statistics were conducted using absolute and relative frequency, measures of central tendency and dispersion, and normality. Frequency distributions were evaluated for coding inaccuracies, outliers, and missing data. Pearson’s correlation coefficient was used to evaluate the relationships between study variables. The variables were examined to ensure that the assumptions for correlation were met. With a skewness of 4.2 and kurtosis of 24.1 viral load was not normally distributed (Kline, 2011); therefore, a number of approaches were employed to analyze this data including, an exploratory analysis using of Kendall’s Tau-b, a nonparametric alternative to Pearson’s correlation (Schaeffer & Levitt, 1956). To transform this data, HIV viral load less than 75 were recoded as 74 as HIV viral loads less than 75 copies/mL are generally considered undetectable (United States & HIV/AIDS Bureau, 2011). Winsorization, the process of limiting extreme values to reduce the effect of outliers (Ghosh & Vogt, 2012), was used. The four largest scores (119139, 130000, 290000, and 304836) were recoded to 83701 resulting in skewness of 2.7 and kurtosis if 11.3. Secondly, viral load was then dichotomized into undetectable (HIV viral load less than 74) (76% of sample) vs. detectable (HIV viral load of 75 or greater).

Results

The study sample consisted 104 PLWH. Most were African American/Black (82.5%) males (52.9%) ranging from 20 – 64 years old. (Table 1). Sleep duration ranged from 121.3–878.5 minutes (2–14 hours) per night. Sleep fragmentation ranged from 15.5–86.4 minutes per night. Sleep efficiency ranged from 23.3–91.2% per night (Table 2).

Table 1:

Demographic Data

Characteristics	Frequency	Percent
Age (Mean=48; SD=8.727; Range=20–64)
49 years old or younger	51	49.3%
50 years old or older	53	50.8%
Gender
Male	54	52.9%
Female	48	47.1%
Marital Status
Married/Domestic Partnership	14	13.6%
Single/Separated/Divorced	89	86.4%
Education
11th Grade or Less	26	25.2%
High School or GED	28	27.2%
Some College/Technical School	33	32%
2 Years of College/AA/Technical School Training	2	1.9%
College (BS or BA)	12	11.7%
Master’s Degree	2	1.9%
Race/Ethnicity
African American/Black	85	82.5%
Other Race/Ethnicity	18	17.5%
Employment
No	89	86%
Yes	14	13.6%
Sexual Orientation
Gay	30	30.3%
Bisexual	10	10.1%
Heterosexual	55	55.6%
Other	4	4.0%
Sexual Orientation: Dummy Codes
Heterosexual	55	55.6%
Gay/Bisexual/Other	44	44.4%
Income
No Monthly Income	18	17.5%
Less than $200/month	7	6.8%
$200-$399	6	5.8%
$400-$599	3	2.9%
$600-$799	41	39.8%
$800-$999	11	10.7%
$1000 or more	17	16.5%
Health Insurance
No	5	4.9%
Yes	97	95.1%
Housing
No	11	10.7%
Yes	92	89.3%

Table 2:

Summary of Distribution

Variables	N	Mean	Median	Mode	SD	Range	Min-Max
Sleep Duration	94	352.4	358.6	121.4	112.3	757.1	121.4–878.5
Sleep Fragmentation	94	31.6	29.4	15.5	12.9	70.9	15.5–86.4
Sleep Efficiency	94	70.1	73.5	23.4	15.4	67.9	23.4–91.3
CD4+ T-Lymphocyte Count	104	609.6	520	361	375.2	1723	43–1766
HIV Viral Load	102	10239.9	33.5	20	45318.5	304816	20–304836

CD4+ count values ranged from 43–1766 cells/μL of blood (Table 2). No relationship was found between sleep and CD4+ count. (Table 3). HIV viral load values ranged from 20–304836 copies/mL of blood (Table 2). With a skewness of 4.2 and kurtosis of 24.1 viral load was not normally distributed (Kline, 2011). No relationship was found between any of the sleep variables and the HIV viral load variables (Winzorized/Dichotomized) using Pearson’s. (Table 3). Using Kendall’s Tau-b; a higher viral load was associated with decreased sleep efficiency τ_b(93) = −.165, p = .043. (Table 4).

Table 3:

Pearson’s Correlation Coefficient (r)

Variables	CD4+ T-Lymphocyte Count	Winzorized HIV Viral Load	Dichotomized HIV Viral Load
Sleep Duration	.006	−.048	−.092
Sleep Fragmentation	−.136	.065	.149
Sleep Efficiency	−.008	−.124	−.157

^*Correlation is significant at the 0.05 level (2-tailed).

^**Correlation is significant at the 0.01 level (2-tailed).

Table 4:

Kendall’s Tau_b

Variables	CD4+ T-Lymphocyte Count	Winzorized HIV Viral Load	Dichotomized HIV Viral Load
Sleep Duration	.062	−.086	−.083
Sleep Fragmentation	−.110	.136	.136
Sleep Efficiency	.055	−.165*	−.166

^*Correlation is significant at the 0.05 level (2-tailed).

^**Correlation is significant at the 0.01 level (2-tailed).

Discussion

In this cross-sectional analysis, sleep duration for study participants ranged between 2 and 14 hours a night, with an average of just under 6 hours (5.9). Adults age 18–60 should sleep at least 7 hours a night (Liu, 2016). This further supports concerns that people living with HIV are vulnerable to sleep disturbances (Rubinstein & Selwyn, 1998; Uchôa et al., 2018) and is in line with Lee et al.’s (2012) study where 290 PLWH in San Francisco slept between <2 hours - >11 hours of sleep per night with 45% of the study sample sleeping less than 6 hours a night (Lee et al., 2012). This study measured both self-reported sleep quality and objective sleep quantity using wrist actigraphy among PLWH. Interestingly, participants who reported difficulty falling asleep subjectively experienced the greatest symptom burden but objectively had actigraphy and clinical measures akin to those without sleep disturbance. However, more than half of the sample experienced severely fragmented sleep per actigraphy which was associated with lower CD4+ counts and higher HIV viral loads. However, despite the association with worsening clinical measures, only 1/3 of the sample reported fragmented sleep (Lee et al., 2012). This demonstrates the necessity of collecting both objective and subjective sleep data for a clearer picture of sleep health among PLWH.

Although evidence suggests that self-reported sleep disturbances are associated with decreased CD4+ counts among PLWH (Seay et al., 2013), sleep was not correlated with CD4+ count in this study. There was, however, an association between sleep and HIV viral load. Higher sleep efficiency was correlated with lower HIV viral loads. These findings are similar to a study comparing veterans living with HIV to veterans without HIV also founding a relationship between sleep and HIV viral load but not to CD4+ count (Womack et al., 2017).

Limitations

This study had a number of limitations to consider. As a secondary data analysis, this study was limited by the tools used and the data available in the TRIP study database (Bryman, 2016; E. Smith, 2008). For example, chart abstractions were used for laboratory results, and differences in the timing of laboratory and sleep data were unavailable for this analysis. There are also variables, such as housing, that may be confounded the observed associations, that were statically insignificant in this study, likely due to our small sample size. While actigraphy is a validated and accepted measure of sleep, polysomnography is the gold standard and may have yielded superior results (M. T. Smith et al., 2018). Convenience sampling of PLWH in Northeast Ohio leaves a potential for selection bias. Only 24% of study participants had a detectable viral load resulting in a lack of variability and issues of non-normality, requiring the use of several analytic approaches to achieve our findings. The Kendall’s Tau correlation was exploratory so was not adjusted for multiple comparisons. Finally, analyzing this data over multiple time-points would have allowed for more robust understanding of sleep, HIV viral load, and immune function in this study sample.

Implications

Adequate sleep is a critical for the health of PLWH. While it has been established that poor sleep can negatively impact immune function among PLWH, more research is warranted to clarify the nature of this relationship as well as the relationship between sleep and HIV viral load. Both objective and subjective measures of sleep are necessary to accomplish this. Future sleep studies should include a larger sample size to more adequately control for confounders, as well as objective and subjective measures of sleep with the goals of (1) improving the understanding of the relationships between objectively and subjectively measured sleep and (2) increasing the knowledge of the impact of sleep disturbances on PLWH. Refining our understanding in these areas will ultimately allow us to enhance clinical practice and interventions so as to improve overall health outcomes and quality of life for PLWH.