Alcohol consumption and illicit drug use: associations with fall, fracture, and acute healthcare utilization among people with HIV infection

Abstract

Background:

Given alcohol and/or other drug (AOD) use occurs among people with HIV (PWH), we examined its association with falls and fall-related outcomes and if frailty moderates the association.

Setting:

Northeastern US city.

Methods:

We analyzed an observational cohort of PWH with current or past AOD use. Alcohol measures were any past 14-day heavy use, average alcohol/day, and days with heavy use. Drug use measures were past 30-day illicit use of cocaine, opioids, and sedatives. Repeated cross-sectional associations were estimated with separate multivariable GEE regression models for each fall-related outcome.

Results:

Among PWH (n=251; mean age 52 [standard deviation=10]), 35% reported heavy alcohol use, 24% cocaine, 16% illicit opioids, 13% illicit sedatives, 35% any fall; 27% were frail. Heavy alcohol use was associated with a fall (AOR=1.49, 95%CI: 1.08, 2.07), multiple falls (AOR=1.55 95%CI: 1.10, 2.19), and fall/fracture-related emergency department (ED) visit or hospitalization (AOR=1.81, 95%CI: 1.10, 2.97). Higher average alcohol/day and more heavy drinking days were associated with multiple falls. Illicit sedative use was associated with a fall, multiple falls, and ED/hospitalization and opioid use with fracture. Frailty moderated the association of heavy alcohol use and a fall (AOR=2.26, 95%CI 1.28, 4.01 in those frail) but not in those not frail.

Conclusion:

The effect of AOD use on falls and fall-related outcomes was most pronounced with alcohol, particularly among frail PWH. Heavy alcohol, illicit sedative, and illicit opioid use are high-priority targets for preventing falls and fall-related consequences for PWH.

INTRODUCTION

People aging with HIV are more likely to experience declining cognitive and physical functioning (1,2) than the general population, with a substantial proportion of people with HIV (PWH) under age 65 reporting limitations with daily activities such as grocery shopping and transportation (3). Reasons include multimorbidity, persistent depression, social determinants of health, and HIV infection itself (1–4). Physical functioning is also affected by incident falls and fall-related consequences. A meta-analysis of PWH found a pooled annual frequency of 26% for at least one fall (and 14% for multiple falls) - comparable to fall rates in the general population a decade older (5). The consequences of a fall can be devastating, including fracture, immobility, hospitalization, and loss of independence (6). Fall prevention is particularly important for PWH who are at higher risk of frailty, a physiological condition that itself is a risk factor for falls (7) and confers vulnerability to decline after an injury (8). Despite this, fall prevention strategies for PWH are understudied and extrapolated from older general populations. While there is growing recognition that preservation of functioning should be an important aspect of HIV care (1–3), PWH have expressed concern that fall prevention is not adequately addressed by their healthcare providers (9,10).

Among the most important but minimally examined risk factors for falls are alcohol and other drug (AOD) use. Beyond the risk of a fall caused by an occasion of heavy alcohol use, chronic effects of alcohol use that overlap with HIV can contribute to a higher risk of falls. For example, long term effects of both alcohol and HIV include sensory peripheral neuropathy, neurocognitive impairment (11–14), and loss of muscle mass (sarcopenia) and strength (15–18). As a result, problems with coordination, balance, and gait are not uncommon with heavy alcohol use and long-standing HIV infection (19,20). Autonomic dysfunction is an under-recognized complication of HIV (21) as well as heavy alcohol use (22). Any of these mechanisms can raise the risk of a fall or contribute to the development of frailty (23,24), thereby increasing fall risk (7,25).

Despite multiple pathways for alcohol to increase fall risk in PWH, research on alcohol and falls is limited. Study findings vary by level of alcohol use, method of detection, fall severity, and sample characteristics such as age, sex, and HIV disease severity (11,25–27). Earlier studies have used data from clinical encounters, subject to under-detection of substance use (28). Fall prevention guidelines cite alcohol use as a contributing factor but provide no specific recommendations about the level of consumption that increases fall risk (29,30). Assessments with validated instruments administered in a standardized manner would offer a level of detail necessary to assess the full range of alcohol use that increases the risk of falls, fractures, and healthcare utilization.

An even larger research gap exists for the relationship of illicit drug use and falls. Associations between falls and illicit drug use have been observed in a few populations with HIV (26,27), however this association has largely been examined without information about drug type or recency of use, using unclear time frames (e.g., “current” vs “former”) (25,31). Research in general populations suggests a relationship between opioid use and fracture (32,33). Knowing whether certain illicit drug types are associated with falls would be valuable for counseling patients and designing fall prevention interventions.

We previously reported that greater severity of an alcohol use disorder (i.e., number of diagnostic criteria) and drinking more than 6 drinks in a day at least once in the past year were both associated with greater risk of a fall among PWH (34). We expand the scope to examine if detailed measures of alcohol use quantity and frequency (average daily consumption, number of heavy drinking days) and illicit drug use (opioid, cocaine, or sedative) are associated with a fall, multiple falls, fracture, and fall or fracture-related emergency department (ED) visit or hospitalization. Given the close relationship between falls and frailty, we also examined whether frailty is a mediator and/or moderator of any of these associations.

METHODS

Study design

The Boston ARCH Frailty, Functional Impairment, Falls, and Fractures (4F) Study is a longitudinal study of PWH (34) with at least one of the following inclusion criteria: i) unhealthy alcohol use (Alcohol Use Disorders Identification Test for Consumption score 3+ for women, 4+ for men] in the past 12 months (35); ii) any illicit drug use in the past 12 months; or iii) enrollment in a prior cohort study of PWH with an AOD use disorder or ever injection drug use (36). Participants were recruited from the prior cohort study and adult primary care and HIV clinics at an l urban academic medical center in the northeast U.S between 2018 and 2020.

Data Collection

Trained research associates administered standardized interviews to participants. Gait speed and grip strength were measured for frailty evaluations as previously described (37). Comorbidities, CD4 count, and HIV viral load suppression were collected from the electronic health record (EHR). Standardized interviews and physical assessments were repeated annually except for information on AOD use and fall, fracture, and fall/fracture-related utilization, which were assessed every 6 months.

Outcomes

The primary outcome was fall (any). Secondary outcomes were number of falls (1, 2, 3 or more), fracture (any), and any fall/fracture-related ED visit or hospitalization. All outcomes were self-reported and referenced the previous 6 months. We used the AIDS Clinical Trials Group (ACTG) falls form, including the fall definition as “an unexpected event, including a slip or trip”, in which the participant lost his/her balance and “landed on the floor, ground or lower level, or hit an object like a table or chair,” not including falls that were from a major medical event or an overwhelming external hazard (26,38,39).

Independent variables

Three alcohol use measures were examined and referenced the previous 14 days: i) any heavy use (primary) defined as exceeding National Institute on Alcohol Abuse and Alcoholism (NIAAA) recommended daily drinking limits (≥5 U.S. standard drinks in a day for men, ≥4 for women) or weekly limits (>14 drinks in a week for men, >7 for women) (40); ii) number of days with heavy use (exceeding NIAAA daily limits); and iii) average daily consumption (grams/day with one U.S. standard drink considered to be 14 grams of ethanol). Quantity and frequency of alcohol consumption was assessed with the Timeline Followback, a validated, self-reported, calendar-based method of assessing daily alcohol use (41,42).

Three separate measures of drug use in the past 30 days were assessed with the Addiction Severity Index: any illicit opioid use, cocaine use, and illicit sedative use (43). Illicit opioid and illicit sedative use included use of opioids or sedatives prescribed to the participants but used in amounts greater than prescribed.

The following covariates were included in the analyses: age, gender, self-identified race/ethnicity, body mass index (BMI), immune suppression (CD4<200), HIV viral load suppression (viral load<200 copies), homelessness (any night in a shelter or street) (44), Charlson-comorbidity score (45), and significant depressive symptoms (Center for Epidemiologic Studies Depression Scale score 16+) (46). The latter three covariates were analyzed as time-varying, while data from the participant’s baseline assessment was used for the other covariates.

Frailty (Fried’s frailty assessment) was a time-varying variable, defined as meeting ≥3 criteria of slow gait speed, low grip strength, self-reported unintentional weight loss, exhaustion, and low physical activity level (37).

Statistical analysis

Associations between AOD use and fall outcomes were analyzed using repeat cross-sectional analyses, pooling data from baseline and 6-month visits for up to 3.5 years for a total of 1174 observations. Generalized estimating equation (GEE) logistic regression models were used to account for multiple observations per participant. We performed separate adjusted cross-sectional analyses for each alcohol measure and each outcome. All models also included separate measures of illicit opioid, illicit sedative, and cocaine use, and all covariates described above. None of the independent variables and covariates were highly correlated (correlation coefficients <0.40).

To test the hypothesis that heavy drinking is associated with the three dichotomous outcomes (i.e., fall, fracture, and fall/ fracture-related ED/hospital utilization), we used separate GEE logistic regression models with a 2-level variable for heavy alcohol use (yes vs no) as the main independent variable. We compared fit statistics for the three different correlation structures to determine the best structure to use. The same strategy was used to examine the three dichotomous illicit drug use exposures.

To examine the association of heavy alcohol use and the number of falls, we used GEE ordinal logistic regression to model the probability of being in a category with a higher number of falls (1, 2, or >3). Average daily alcohol consumption and number of heavy drinking days were modeled as linear predictors. Because homelessness and low BMI could be on the casual pathway of the impact of AOD use on falls, we performed sensitivity analyses removing homelessness and BMI covariates from the multivariable regression models.

We explored whether frailty (yes vs no) was a moderator (if the strength of the relationship between AOD use and fall outcomes differed among frail vs non-frail participants). We used frailty data collected at the same research interview when available and at the previous interview when not available. We also examined whether frailty was a mediator of any of the significant associations. Based on a Baron and Kenny conceptual model (47), we considered frailty to be a mediator if: a) the AOD measure was significantly associated with frailty; b) the association between the AOD measure and fall outcome was attenuated when accounting for frailty; and c) frailty was significantly associated with the fall outcome when accounting for the AOD measure.

RESULTS

Characteristics of the study sample at baseline

The mean age of the cohort (n=251) was 52 years (standard deviation [SD] 10) with fewer women (33%) than men (Table 1). Most self-identified as either Hispanic (21%) or Black, non-Hispanic (52%). Fifteen percent experienced homelessness. HIV viral load suppression (<200 copies/mL) was almost universal (87%). About a quarter met criteria for frailty (27%).

Table 1:

Baseline characteristics of people with HIV with a history of alcohol and drug use for the overall study sample and stratified by any heavy alcohol use ^a

	Total sample	Heavy alcohol use
		yes	no
	n=251	n=89	n=162
Age, mean (standard deviation)	52 (11)	51 (11)	53 (10)
Female sex	82 (33%)	29 (33%)	53 (33%)
Race/ethnicity b
Hispanic	53 (21%)	14 (16%)	39 (24%)
Black, not Hispanic	131 (52%)	55 (62%)	76 (47%)
White, not Hispanic	47 (19%)	13 (15%)	34 (21%)
Homeless	38 (15%)	13 (15%)	25 (15%)
Underweight (BMI <18.5)	6 (3%)	2 (2%)	4 (3%)
Obese (BMI >=30)	79 (33%)	30 (37%)	49 (31%)
HIV viral load suppression (<200 copies)	208 (87%)	71 (84%)	137 (88%)
Immune suppression (CD4<200)	21 (7%)	8 (9%)	13 (8%)
Frail c	63 (27%)	26 (31%)	37 (24%)
Depressive symptoms d	136 (54%)	52 (58%)	84 (52%)
Alcohol use, past 14 days
Any heavy use	89 (36%)	89 (100%)	0 (0)
Alcohol grams per day, median (IQR)	3 (0, 22)	35 (18, 59)	0 (0, 2)
Heavy drinking days, median (IQR)	0 (0, 2)	3 (1, 6)	0 (0, 0)
Alcohol use disorder, past year	106 (42%)	71 (80%)	35 (22%)
Illicit drug use e, past 30 days
Opioid use	40 (16%)	15 (17%)	25 (15%)
Cocaine use	61 (24%)	33 (37%)	28 (17%)
Sedative use	32 (13%)	14 (16%)	18 (11%)
Opioid use disorder, past year	39 (16%)	13 (15%)	26 (16%)
Cocaine use disorder, past year	67 (27%)	30 (34%)	37 (23%)
Sedative use disorder, past year	13 (5%)	6 (7%)	7 (4%)
Number of falls
1	30 (12%)	9 (10%)	21 (13%)
2 or more	57 (23%)	26 (29%)	31 (19%)
Fracture, any	15 (6%)	9 (10%)	6 (4%)
ED or hospitalization for fall or fracture	30 (12%)	16 (18%)	14 (9%)

^aNumber of participants (proportion of sample) unless otherwise specified. Heavy alcohol use defined as exceeding NIAAA weekly limits (>7 drinks/week for women, >14 drinks/week for men) or daily limits (≥4 drinks in one day for women, ≥5 in one day for men).

^bMultiracial/other: 20 (8%) of total sample

^cFried’s frailty criteria. Gait speed was missing for 26 participants at baseline.

^dCenter for Epidemiologic Studies Depression Scale score 16+.

^eIncludes opioid (or sedative) medication prescribed to the participant but used in greater amounts than prescribed.

A substantial proportion of the sample reported heavy alcohol use (89/251, 35%) (Table 1); 80% (71/89) of whom met criteria for an alcohol use disorder. Cocaine (24%) was the most prevalent illicit drug, followed by illicit opioids (16%) and illicit sedatives (13%). Polysubstance use was common. A substantial proportion of participants with any heavy alcohol use (n=89) also used cocaine (37%), illicit opioid (17%), or an illicit sedative (16%) (Supplemental Table 1). Among those with cocaine use (n=61), 82% had a cocaine use disorder. Similarly, 82% of those with any illicit opioid use (n=40) had an opioid use disorder. Only about half of those with illicit sedative use (n=32) had a sedative use disorder (43%).

Approximately one-third (87/251, 35%) of the sample had at least one fall, most of whom had ≥2 falls (57/87, 66%). Fracture and fall/fracture-related ED/hospitalization were infrequent (15/251, 6%; 30/251, 12%, respectively). Figure 1 shows the proportion of falls among those with and without heavy alcohol, cocaine, opioid, and sedative use.

Figure 1:

Proportion reporting a fall stratified by use of heavy alcohol, illicit opioid, cocaine, and illicit sedative

Adjusted associations between AOD use and falls

Heavy alcohol use was associated with a fall (AOR 1.49, 95% Confidence Interval [CI]: 1.08, 2.07); multiple falls (AOR 1.55, 95%CI: 1.10, 2.19); and fall/fracture-related ED/hospitalization (AOR 1.81, 95%CI: 1.10, 2.97) but not fracture (Table 2). There was a positive association between the number of falls and average daily consumption (AOR 1.08, 95%CI: 1.03, 1.13 for each increase of 14 grams of alcohol per day) and heavy drinking days (AOR 1.05, 95%CI: 1.01, 1.10 for each additional heavy drinking day).

Table 2:

Multivariable associations between alcohol and drug use and fall, number of falls, fracture, and ED/hospitalization for a fall or fracture among people with HIV ^a

	Fall, any	Number of falls b	Fracture, any	ED/hospitalization for a fall or fracture
Alcohol use
Heavy use, any c	1.49 (1.08, 2.07)	1.55 (1.10, 2.19)	1.23 (0.66, 2.27)	1.81 (1.10, 2.97)
Average daily consumption d	1.04 (0.99, 1.09)	1.08 (1.03, 1.13)	1.08 (1.00, 1.17)	1.07 (0.99, 1.15)
Number of heavy drinking days e	1.03 (0.99, 1.07)	1.05 (1.01, 1.10)	1.07 (1.00, 1.16)	1.06 (1.00, 1.13)
Illicit drug use f
Opioid use	1.38 (0.88, 2.16)	1.17 (0.69, 1.99)	2.22 (1.08, 4.57)	0.88 (0.39, 1.98)
Sedative use	1.73 (1.01, 2.96)	2.21 (1.32, 3.71)	1.66 (0.55, 5.05)	2.54 (1.12, 5.80)
Cocaine use	0.76 (0.46, 1.24)	1.13 (0.65, 1.95)	1.15 (0.44, 2.99)	0.67 (0.34, 1.32)

^aAdjusted Odds Ratios (95% Confidence Intervals) were calculated using separate GEE logistic regression models for each dichotomous outcome and GEE ordinal logistic regression for the outcome, number of falls. All models include one alcohol measure, opioid, sedative, cocaine use (as 3 separate measures), demographics, homelessness, HIV disease markers, body mass index, and depressive symptoms. AOR for opioid, sedative, and cocaine use presented in this table are results of the multivariable regression analyses of heavy alcohol use. Bolded text indicates p<0.05.

^bNumber of falls was a categorical variable: 0, 1, 2, or ≥3 falls. AOR indicates odds of reporting a higher category of number of falls

^cTimeline Follow back, past 14-day use. Exceeds NIAAA weekly limits (>7 drinks/week for women, >14 drinks/week for men) or daily limits (4+ drinks in one day for women, 5+ in one day for men).

^dAdjusted Odds Ratio for an increase of 14 grams per day

^eAdjusted Odds Ratio for each additional day of heavy alcohol use (exceeding daily limit).

^fAddiction Severity Index, past 30 days. Parameter estimates for each drug type are reported from the model of heavy alcohol use

Illicit sedative use was associated with a fall (AOR 1.73, 95%CI: 1.01, 2.96); multiple falls (AOR 2.21 (95%CI: 1.32, 3.71); and fall/fracture-related ED/hospitalization (AOR 2.54, 95%CI: 1.12, 5.80). Illicit opioid use was associated with fracture (AOR 2.22, 95%CI: 1.08, 4.57) but not with a fall (AOR 1.38, 95%CI: 0.88, 2.16) or multiple falls (AOR 1.17, 95%CI: 0.69, 1.99). There were no significant associations with cocaine and any of the outcomes. The AOR for all illicit drug use measures were not substantially different when modeling the other alcohol exposures (i.e., average daily consumption and heavy drinking days). Results of regression analyses without covariates for homelessness and weight loss were also not substantially different (data not shown).

Homelessness was associated with higher odds of a fall, multiple falls, fracture, and fall/ fracture-related ED/hospitalization (Table 3). Depressive symptoms were also associated with increased odds of a fall, multiple falls, and fall/fracture related ED/hospitalization.

Table 3:

Results of fully adjusted GEE regression models of the association of heavy alcohol use and drug use with fall, number of falls, fracture, and ED/hospitalization for a fall or fracture among people with HIV ^a

	Fall, any	Number of falls e	Fracture, any	ED/hospitalization for a fall or fracture
Heavy alcohol use b	1.49 (1.08, 2.07)	1.55 (1.10, 2.19)	1.23 (0.66, 2.27)	1.81 (1.10, 2.97)
Illicit drug use c
Opioid use	1.38 (0.88, 2.16)	1.17 (0.69, 1.99)	2.22 (1.08, 4.57)	0.88 (0.39, 1.98)
Sedative use	1.73 (1.01, 2.96)	2.21 (1.32, 3.71)	1.66 (0.55, 5.05)	2.54 (1.12, 5.80)
Cocaine use	0.76 (0.46, 1.24)	1.13 (0.65, 1.95)	1.15 (0.44, 2.99)	0.67 (0.34, 1.32)
Older age	1.02 (1.00, 1.05)	1.02 (1.00, 1.05)	1.02 (0.97, 1.06)	1.05 (1.01, 1.09)
Female vs male	1.58 (1.02, 2.42)	1.69 (1.06, 2.71)	1.55 (0.56, 4.27)	2.02 (1.07, 3.80)
Race/ethnicity d
Hispanic	0.42 (0.22, 0.82)	0.51 (0.25, 1.00)	0.91 (0.34, 2.45)	0.33 (0.13, 0.81)
Black, non-Hispanic	0.41 (0.25, 0.69)	0.41 (0.25, 0.68)	0.56 (0.29, 1.08)	0.40 (0.21, 0.77)
Multiracial or other	1.10 (0.50, 2.43)	1.18 (0.49, 2.81)	1.39 (0.31, 6.30)	1.77 (0.68, 4.58)
Homeless	1.82 (1.13, 2.93)	2.15 (1.27, 3.63)	2.60 (1.09, 6.23)	3.09 (1.59, 6.03)
BMI
Obese	1.35 (0.85, 2.13)	1.26 (0.78, 2.05)	1.52 (0.57, 4.07)	1.21 (0.63, 2.32)
Underweight	1.08 (0.54, 2.14)	0.92 (0.42, 2.00)	3.16 (0.66, 15.05)	1.97 (0.49, 7.99)
CD4<200	1.37 (0.72, 2.61)	1.19 (0.60, 2.37)	2.17 (0.97, 4.85)	1.53 (0.72, 3.25)
HIV viral suppression	0.92 (0.48, 1.77)	0.85 (0.41, 1.76)	1.36 (0.44, 4.20)	0.90 (0.38, 2.13)
Charlson score
10+ vs 6	0.97 (0.51, 1.84)	1.34 (0.64, 2.82)	0.48 (0.11, 2.02)	0.79 (0.29, 2.16)
8-9 vs 6	0.98 (0.55, 1.75)	0.98 (0.50, 1.93)	0.58 (0.14, 2.40)	0.76 (0.31, 1.85)
7 vs 6	0.80 (0.46, 1.41)	0.81 (0.43, 1.52)	1.16 (0.39, 3.51)	0.77 (0.33, 1.78)
Depressive symptoms	1.60 (1.13, 2.27)	2.28 (1.57, 3.31)	1.17 (0.65, 2.10)	1.66 (1.08, 2.54)

^aAdjusted odds ratios (95%Confidence Intervals) were calculated using separate GEE logistic regression models for each outcome and each alcohol or drug use measure. Number of falls was analyzed using GEE ordinal logistic regression. Models include all variables listed in the table. Bolded text indicates p<0.05.

^bTimeline Followback for past 14 days. Exceeds NIAAA weekly and/or daily limits.

^cAddiction Severity Index, past 30 days. Includes misuse of opioid and sedative medications prescribed to the participant.

^dWhite, non-Hispanic is referent

^eMultiple falls modeled as a 3-level categorical variable: 1, 2, 3 or more falls.

Moderation analyses indicated that the effect of heavy alcohol use on any fall was in those with frailty (AOR of 2.26 [95%CI 1.28, 4.01]) and not among non-frail participants (AOR of 1.26 [95%CI 0.85, 1.85]). There was no evidence of effect modification by frailty in any other analysis (p-values for interaction terms >0.1), nor was it a mediator of any of the significant associations.

DISCUSSION

Among this sample of PWH with a range of AOD use and a mean age of 52, about one-third reported at least one fall in the past 6 months, 27% were frail, and 13–35% reported substance use or heavy alcohol use. Our findings supported the clinical consequences of heavy alcohol use and illicit sedative use, as each was associated with a higher odds of fall, multiple falls, and fall or fracture-related ED/hospitalization. Higher average daily alcohol use (i.e., increase of 14 grams a day) and each additional heavy drinking day were associated with more falls. The association between heavy alcohol and/or illicit sedative use with fall-related complications requiring acute healthcare utilization suggests that falls were clinically meaningful. Illicit opioid use was associated with higher odds of fracture. Frailty moderated the association of heavy alcohol use and fall, such that those who were frail had more than twice the odds of a fall with heavy alcohol use. Taken together, these findings suggest that the relationship between AOD use and falls is specific to substance type and in the case of alcohol, related to the intensity and pattern of use, and frailty status.

In contrast to prior studies in PWH linking falls to heavy alcohol use using AUDIT-C questionnaire (27,34),our study used the Timeline Followback method, a more detailed measure of daily consumption. This allowed us to examine both the frequency of heavy use and average drinks per day as a continuous measure rather than the fixed categorical measures of alcohol assessed by the AUDIT-C. This is important because the biochemical effects of alcohol are affected by drinking frequency in addition to the overall volume of use (48). Despite these differences, it appears that the association of heavy drinking and falls is consistent when defining heavy use by AUDIT-C or by NIAAA recommended drinking limits. Our study also extends the literature by demonstrating associations of heavy drinking with recurrent falls and fall or fracture-related acute medical utilization with analyses that accounted for illicit drug type with validated assessments. The results of the moderation analyses of heavy alcohol and fall risk are concerning and indicate that avoiding heavy alcohol use is particularly important for PWH who are frail and therefore, more at risk of fall-related complications on functioning, ambulation, and management of comorbid conditions.

Our study also builds on the findings by Womack et al. who found higher risk of serious falls in persons with a drug use disorder based upon EHR data (27) by demonstrating an association with specific drug types (i.e., illicit sedative use) using a well-validated instrument (Addiction Severity Index) with a wider range of fall severity. We found this association among people with any illicit sedative use - less than half of whom met criteria for a sedative use disorder. While illicit sedative use is more likely in younger age groups, the prevalence has increased in all age groups (49). In the current study, illicit sedative use included both non-prescribed and prescribed sedatives – the latter meeting the definition if used in greater amounts than prescribed (sedative “misuse”). Older adults in the general population are more likely to misuse prescribed sedatives than non-prescribed sedatives (50) to treat insomnia and physical health conditions (51). This highlights a need for attention to somatic symptoms as well as inquiry about overuse of prescribed sedative medications.

While prior studies have found an association between any opioid use and falls (51), we found that illicit opioid use was associated with greater odds of fracture but not falls. We may have been underpowered for the analysis of falls regarding illicit opioids. The association of opioids and fracture has also been reported in studies of women with and at high risk for HIV (52) and general populations (33) including a meta-analysis of opioid medications prescribed for pain (53). Opioids have also been linked to decreased bone formation and low bone density (54,55).

Homelessness was one of the strongest factors associated with falls in analyses that adjusted for AOD use and depressive symptoms and the only exposure associated with all outcomes. This is notable because falls have been increasingly recognized as health risks related to homelessness in general populations (56,57), but one of the few studies to find this among PWH. A fracture or fall-related injury may have profound impact for PWH experiencing homelessness given that walking is likely a primary mode of transportation to go to work, medical appointments, and social service agencies. Like other fall prevention strategies that target modifiable environmental hazards, supplying housing could have considerable impact on reducing the risk of falls, fracture, and costly healthcare utilization.

We also found that depressive symptoms were independently associated with higher odds of a fall, more falls, and ED/hospitalization. Depression and medications used for treatment of depression are well-established risk factor for falls (25,26,39), another potentially modifiable factor.

Several limitations merit consideration when interpreting the study findings. A participant may have fallen and subsequently stopped or decreased his/her AOD use. If there were a substantial number of “sick quitters”, the effect would be to reduce the differences associated with heavy alcohol use and illicit drug use. Therefore, the “true” odds of a fall with heavy AOD use may be higher than this study’s estimate. An AOD use biomarker could have confirmed the absence of recent drinking/use in persons who report abstinence or low-level use, though the biomarker would reflect use over a narrower timeframe. We also did not determine whether use of alcohol or illicit sedative occurred immediately before a fall or fracture. This is reasonable as we were interested not only with the direct effect of AOD use prior to a fall but also with long-term effects of AOD use such as impaired balance, frailty, or slower processing speed. More broadly, the time period of the substance use measures and falls was not identical. It would be difficult to obtain detailed information about daily alcohol consumption or the number of heavy drinking days over a 6-month interval to match fall outcomes. To our knowledge, there are no validated instruments for obtaining this information. Additionally, while our alcohol measures spanned a 2-week interval, the vast majority (80%) met criteria for an AUD over the previous 12 months, with a similar pattern for the three types of illicit drug use and drug use disorders suggesting that heavy alcohol and illicit drug use extended beyond these narrow intervals. We did not include information on polypharmacy or prescribed opioid or sedative medications in the analyses. The cross-sectional nature of the analyses precludes our ability to draw conclusions about the temporal relationships between AOD use and falls and fall/fracture-related outcomes. Since all participants were HIV positive, this study does not provide comparison information about AOD use and falls in general populations. Finally, our findings may not be generalizable to other HIV populations with fewer alcohol and drug use disorders.

This study has important strengths. Consideration of multiple substances is important, given that polysubstance use has become more of the norm especially among people who drink in heavy amounts (58). Having data on the full spectrum of AOD use (i.e., from at-risk use to a use disorder) and frailty is a strength, because data on heavy drinking and illicit drug use that does not meet criteria for a clinical diagnosis of a AOD use disorder and frailty status is often not available in large administrative datasets. Distinguishing the time interval of illicit drug use is another contribution to the literature. We also note that the AOD use measures were collected with well-validated instruments in a research setting aimed at limiting underreporting of AOD use.

In this cohort of PWH with AOD use, approximately one-third had sustained a fall in the past 6 months. We found strong associations between heavy alcohol use, illicit drug use with falls and falls-related complications. Feedback to patients about the risks associated with heavy alcohol and illicit sedative use may be valuable, particularly for PWH who are frail. While avoiding heavy alcohol use may lower fall risk, the study findings also suggest benefits with decreasing the volume of drinking in a day and the number of heavy drinking days. Heavy alcohol, illicit sedative, and illicit opioid use are high-priority modifiable targets for preventing falls and fall-related morbidity among people aging with HIV.