Abstract
Background and Aims
Given the more than twofold increase in the prevalence of substance use disorders in the United States in the past decade, more hospital inpatients can be expected to carry substance use disorder diagnoses, necessitating evaluation of potential links to 30‐day unplanned readmissions, a marker of quality of care. This study aimed to measure the association between substance use disorder diagnoses, discharge disposition and 30‐day unplanned hospital readmissions.
Design
This retrospective cohort study extracted data from electronic health records of all inpatients. The index admission was defined as a patient's first admission in 2022.
Setting
Two urban, academic medical centers in Los Angeles, California, USA.
Participants
Among 22 108 inpatients aged 18 and over and who did not expire during the hospital stay, 7.4% had at least one substance use disorder. The median age was 58, and 56.1% identified as female. Most patients identified as white (43.3%), followed by 22.5% Hispanic/Latinx, 10.8% Asian and 9.1% Black; 14.3% identified as another race.
Measurements
The exposure was diagnosis of any substance use disorder at index admission. The outcome was 30‐day unplanned readmission.
Findings
Patients with any substance use disorder [adjusted risk ratio (aRR) = 1.24, 95% confidence interval (CI) = 1.05–1.45) and patients specifically with opioid use disorder (aRR = 1.40, 95% CI = 1.09–1.80) were more likely to have a 30‐day unplanned readmission compared with patients without substance use disorders. When assessing an interaction with discharge disposition, the association only held for patients discharged to home/self‐care (aRR = 1.33, 95% CI = 1.05–1.69). Among patients who had zero, one, two and three or more unplanned readmissions, 7.1%, 8.8%, 14.0% and 15.5% had a substance use disorder at their index admission, respectively (P < 0.001).
Conclusions
In the United States, hospital patients with substance use disorder diagnoses appear to have a higher risk of 30‐day unplanned readmission to hospital and account for a disproportionate share of patients who have multiple unplanned readmissions than hospital patients without substance use disorder diagnoses.
Keywords: 30‐day unplanned readmissions, alcohol use disorder, hospital discharge, opioid use disorder, quality of care, substance use disorders
INTRODUCTION
Unplanned readmission rates are a widely used metric of hospital performance when measuring quality of care. As highlighted by the creation of the Hospital Readmissions Reduction Program in the United States (US) Affordable Care Act, unplanned hospital readmissions are not only costly for hospitals, but also negatively impact patients' health. Unplanned readmissions interrupt the lives of patients by increasing risk of in‐hospital complications such as infections and falls, and by increasing financial burdens [1, 2, 3]. Previous research has identified a number of risk factors for unplanned readmission including clinical factors such as diagnoses, length of stay, polypharmacy and previous hospitalization as well as social factors including insurance, socio‐economic status and access to social support [3, 4, 5, 6, 7, 8].
Having a substance use disorder (SUD) may also be an additional risk factor for unplanned readmission, but is under studied [9, 10, 11]. The prevalence of SUDs in the United States has more than doubled in the past decade, increasing from 8.5% in 2012 to 17.3% in 2022 [12, 13]. Given this rise in the general population, it follows that an increasing proportion of hospitalized patients now has an SUD, regardless of primary diagnosis at the time of hospitalization [14]. Therefore, new, rigorous research is needed determine how this may impact unplanned readmissions and quality of care. Aside from two studies of patients in large, urban hospitals and another study using Medicaid claims data, other previous research showing a relationship between substance use and readmission has been primarily conducted within high‐risk subgroups including patients living with HIV and patients at psychiatric hospitals with major depressive disorder and schizophrenia [9, 10, 11, 15, 16, 17].
Although previous studies have established a link between SUDs and unplanned readmission, most are in small samples or unique subpopulations and fail to investigate differences between specific SUDs as well as possible mechanisms that may explain why patients with SUDs are at increased risk of unplanned readmission [10, 15, 16, 17]. One factor at play is that a defining criterion of SUDs is continued use despite knowledge of negative consequences [18]. Patients in the hospital and post discharge may continue to use substances, which can interfere with rehabilitation and recovery after discharge [3]. Substance use and misuse disorganizes behavior, which can also contribute to problems with adherence to medication and to rehabilitation [19, 20]. Substance use is associated with unmet social needs, including safe and affordable housing, healthy food and employment, which can further complicate simple tasks like keeping follow‐up appointments and visiting pharmacies to pick up medications [21, 22, 23, 24]. The extent to which post‐acute care following discharge can mitigate some of these disparities is not well understood and have not been studied in patients with SUDs [25, 26, 27].
In an effort to improve healthcare quality at two large academic medical centers in Southern California, this study aimed to quantify the risk of 30‐day unplanned readmission for patients with SUDs diagnoses at admission. We hypothesized that SUDs, especially alcohol use disorder (AUD) and opioid use disorder (OUD), would be associated with higher risk of 30‐day unplanned readmission, and that the strength of the association would be largest for patients discharged without post‐acute care.
METHODS
Study design
This retrospective cohort study extracted data from electronic health records (EHR) of all patients admitted to two urban academic medical centers in Los Angeles, California in 2022. Patients were excluded from the study if they were under 18 years of age, died during the hospital stay or were not eligible for readmission at the time of discharge (i.e. transferred to another acute care hospital or hospice care). The University of California, Los Angeles Institutional Review Board (IRB) reviewed this project and deemed it a quality improvement project that did not require IRB review. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline [28]. This analysis was not pre‐registered and, therefore, findings should be considered exploratory.
Measures
The primary outcome of interest was 30‐day unplanned readmission, defined as any unplanned inpatient admission to the two hospitals in this study within 30 days of discharge from a prior admission.
The primary exposure was diagnosis of any SUD at the index admission. The index admission was defined as a patient's first admission in calendar year 2022. Secondary exposures of interest included two individual SUDs recorded at the index admission: AUD and OUD. The SUD diagnosis could be in any position on the diagnosis list, which was chosen to capture as many patients with SUDs as possible. Reasons why an SUD may not be recorded as the principal diagnosis include having a more medically complex patient with multiple related reasons for hospitalization, recording a complication of an SUD rather than SUD itself as the principal diagnosis and having billing incentives to not record the SUD as the principal diagnosis even if it is the primary reason for hospitalization [29, 30].
SUDs were defined as having at least one of the following diagnoses documented during the index admission using International Classification of Diseases, Tenth Revision (ICD‐10): alcohol related disorders (F10), opioid related disorders (F11), sedative, hypnotic or anxiolytic related disorders (F13), cocaine related disorders (F14), other stimulant related disorders (F15) and other psychoactive substance related disorders (F19). Because SUDs are underdiagnosed in inpatient hospitalizations, additional diagnoses that are markers of SUDs were also included: alcoholic polyneuropathy (G62.1), alcoholic cardiomyopathy (I42.6), alcoholic fatty liver (K70.0), alcohol abuse counseling and surveillance of alcoholic (Z71.41), poisoning by, adverse effect of and underdosing of other opioids (T40.2), poisoning by, adverse effect of and underdosing of cocaine (T40.5), and poisoning by, adverse effect of and underdosing of benzodiazepines (T42.4) [29] (ICD‐10 codes used to define SUDs for individual substances are listed in Table S1).
Covariates were measured at the index admission. (1) Age: the patient's age in years at the index admission. (2) Race or ethnicity: Hispanic or Latinx, non‐Hispanic Asian, non‐Hispanic Black or African American, non‐Hispanic White and non‐Hispanic other or unknown; the other category included patients who identified as American Indian or Alaska Native, Native Hawaiian or other Pacific Islander and not otherwise specified. (3) Health insurance: commercial or private, Medicaid, Medicare, uninsured and other. The other category included military or Veterans Administration insurance, workers compensation, self‐pay and not otherwise specified. (4) Charlson Comorbidity Index: a score measuring severity of 17 comorbidities (1–6) that each predict risk of mortality within 1 year of hospitalization, with higher scores indicating greater mortality risk (range, 1–37) [31]. (5) A mental health‐related diagnosis defined as one or more of the following ICD‐10 diagnoses recorded at the index admission: schizophrenia, schizotypal, delusional and other non‐mood psychotic disorders (F20–29), mood [affective] disorders (F30–39), anxiety, dissociative, stress‐related, somatoform and other non‐psychotic mental disorders (F40–49), behavioral syndromes associated with physiological disturbances and physical factors (F50–59) and disorders of adult personality and behavior (F60–69) (Table S1). (6) Discharge disposition: home or self‐care, home health service, inpatient care, or eloped or left against medical advice; inpatient care included skilled nursing facilities, rehab facilities, long‐term care hospitals, residential care, psychiatric hospitals, substance use disorder in‐patient or residential settings and prison or law enforcement.
Statistical analysis
Patient characteristics at the index admission were described using counts and percentages for categorical variables and medians and interquartile ranges for continuous variables. Characteristics were compared for patients with and without SUDs.
A risk ratio comparing the risk of 30‐day unplanned readmission for patients with and without SUDs was computed using modified Poisson regression with robust sandwich estimators. Modified Poisson regression can directly estimate a risk ratio in cohort studies by fitting a Poisson regression model with a binary outcome. It is often superior to the better‐known log‐binomial regression, which can have convergence issues, especially when estimating an adjusted risk ratio (aRR) [32, 33]. All patients were assigned a value of one for their time at risk because all patients are followed for an equal amount of time in a cohort study. Robust sandwich estimators were used to produce valid variance estimates, as the Poisson model may produce overestimates [32].
After estimating the crude association, an additional modified Poisson regression model was used to estimate an aRR that accounted for covariates defined a priori based on literature review and prior knowledge. Covariates included age, gender identity, race or ethnicity, insurance, mental health diagnosis, Charlson Comorbidity Index and hospital. This process was repeated for the two most common individual SUDs, AUD and OUD, to assess the association between AUD and OUD with 30‐day unplanned readmission. AUD was categorized as AUD, other SUDs and no SUDs to account for poly substance use and compare those with AUD to those without any SUDs; those with AUD may have had additional SUDs. The same categorization was used for OUD. Potential effect measure modification by discharge disposition was investigated by including a product term for SUD and discharge disposition.
Last, the number of unplanned readmissions per patient in calendar year 2022 was calculated and categorized as zero, one, two or three or more unplanned readmissions. The proportion of patients with any SUD, AUD and OUD was computed among each category of number of unplanned readmissions. A secondary, exploratory analysis using a Cochran‐Armitage test for trend was used to assess the association between SUDs and number of unplanned readmissions. All analyses were conducted in R version 4.2.1, and a P‐value less than 0.05 was considered statistically significant [34].
RESULTS
This analysis included 22 108 patients; a flow diagram for inclusions is presented in Figure S1. The median age was 58 years old (IQR = 38–72) and 12 408 (56.1%) identified as female. Most patients identified as non‐Hispanic White (n = 9562, 43.3%), followed by 4971 (22.5%) Hispanic or Latinx, 2396 (10.8%) non‐Hispanic Asian, 2017 (9.1%) non‐Hispanic Black or African American, and 3162 (14.3%) identified as another or unknown race. Most patients had commercial or private insurance (9665, 43.7%), 6107 (27.6%) had at least one mental health diagnosis and the median Charlson Comorbidity Index was 1.6 (IQR = 1.0–2.3). Most patients did not use post‐acute care and were discharged to home or self‐care (n = 14 938, 67.6%). Among patients with any SUD, the median age was 53 (IQR = 38–66), 634 (38.9%) identified as female, 536 (32.9%) had private insurance and 804 (49.3%) had a mental health diagnosis. In comparison, among patients without any SUD, the median age was 59 (IQR = 38–73), 11 774 (57.5%) identified as female, 9129 (44.6%) had private insurance and 5303 (25.9%) had a mental health diagnosis (Table 1).
TABLE 1.
Characteristics of patients admitted to two urban, academic hospitals, January–December 2022.
| Characteristics | Substance use disorder | ||
|---|---|---|---|
| Total (n = 22 108) | Yes (n = 1631) | No (n = 20 477) | |
| n (%) | n (%) | n (%) | |
| Age, median (IQR) | 58 (38–72) | 53 (38–66) | 59 (38–73) |
| Female gender | 12 408 (56.1) | 634 (38.9) | 11 774 (57.5) |
| Race or ethnicity | |||
| Hispanic/Latinx | 4971 (22.5) | 341 (20.9) | 4630 (22.6) |
| Asian, non‐Hispanic | 2396 (10.8) | 62 (3.8) | 2334 (11.4) |
| Black or African American, non‐Hispanic | 2017 (9.1) | 193 (11.8) | 1824 (8.9) |
| White, non‐Hispanic | 9562 (43.3) | 802 (49.2) | 8760 (42.8) |
| Other/unknown, non‐Hispanic a | 3162 (14.3) | 233 (14.3) | 2929 (14.3) |
| Insurance | |||
| Commercial/private | 9665 (43.7) | 536 (32.9) | 9129 (44.6) |
| Medicaid | 3554 (16.1) | 583 (35.7) | 2971 (14.5) |
| Medicare | 8380 (37.9) | 480 (29.4) | 7900 (38.6) |
| Uninsured | 213 (1.0) | 13 (0.8) | 200 (1.0) |
| Other b | 296 (1.3) | 19 (1.2) | 277 (1.4) |
| Mental health diagnosis c | 6107 (27.6) | 804 (49.3) | 5303 (25.9) |
| Charlson Comorbidity Index, median (IQR) | 1.6 (1–2.3) | 1.6 (1–2.1) | 1.6 (1–2.4) |
| Hospital | |||
| Site 1 | 13 661 (61.8) | 936 (57.4) | 12 725 (62.1) |
| Site 2 | 8447 (38.2) | 695 (42.6) | 7752 (37.9) |
| Discharge disposition | |||
| Home/self‐care | 14 938 (67.6) | 854 (52.4) | 14 084 (68.8) |
| Home health service | 4103 (18.6) | 339 (20.8) | 3764 (18.4) |
| Inpatient care d | 2719 (12.3) | 333 (20.4) | 2386 (11.7) |
| Eloped/left AMA | 348 (1.6) | 105 (6.4) | 243 (1.2) |
Abbreviations: AMA, against medical advice; IQR, interquartile range.
Other includes American Indian or Alaska native (n = 98), Native Hawaiian or other Pacific Islander (n = 68) and other/unknown (n = 2996).
Other includes military/veterans administration insurance (n = 159), workers compensation (n = 70), self‐pay (n = 19) and other not otherwise specified (n = 48).
Defined as at least one of the following ICD‐10 diagnoses at index visit: schizophrenia, schizotypal, delusional and other non‐mood psychotic disorders (F20–29), mood [affective] disorders (F30–39), anxiety, dissociative, stress‐related, somatoform and other non‐psychotic mental disorders (F40–49), behavioral syndromes associated with physiological disturbances and physical factors (F50–59), disorders of adult personality and behavior (F60–69).
Inpatient care includes skilled nursing facility (n = 1652), rehab facility (n = 762), long‐term care hospital (n = 74), residential care (n = 58), psychiatric hospital (n = 155), rehab for chemical dependence (n = 13) and prison or law enforcement (n = 5).
A total of 1631 (7.4%) patients had at least one SUD at their index admission, 882 (4.0%) had AUD and 525 (2.4%) had OUD. The incidence of 30‐day unplanned readmission was 9.7% for patients with any SUD, 9.3% for patients with AUD, and 11.0% for patients with OUD (Figure 1).
FIGURE 1.
Prevalence of substance use disorders at index admissions and incidence of 30‐day unplanned readmission by substance use disorder, January–December 2022.
After adjusting for relevant covariates, patients with SUDs were 1.24 (95% CI = 1.05–1.45) times as likely to have a 30‐day unplanned readmission compared to those who did not have SUDs. Patients with OUD were 1.40 (95% CI = 1.09–1.80) times as likely to have a 30‐day unplanned readmission compared to those who did not have any SUDs. Having AUD was associated with 30‐day unplanned readmission in crude analyses [crude risk ratio (cRR) = 1.29, 95% CI = 1.04–1.60], but was not associated after adjustment (aRR = 1.20, 95% CI = 0.96–1.48) (Table 2).
TABLE 2.
Association of substance use disorders at index admission with 30‐day unplanned readmission.
| Model exposure | n (%) | cRR (95% CI) | aRR (95% CI) a | |
|---|---|---|---|---|
| Main effects models | Model 1 | |||
| Substance use disorder | 159 (9.7) | 1.35 (1.16–1.58)*** | 1.24 (1.05–1.45)* | |
| No Substance use disorder | 1475 (7.2) | 1.00 | 1.00 | |
| Model 2 | ||||
| Alcohol use disorder | 82 (9.3) | 1.29 (1.04–1.60)* | 1.20 (0.96–1.48) | |
| No alcohol use disorder | 1475 (7.2) | 1.00 | 1.00 | |
| Model 3 | ||||
| Opioid use disorder | 58 (11.0) | 1.53 (1.20–1.96)*** | 1.40 (1.09–1.80)** | |
| No opioid use disorder | 1475 (7.2) | 1.00 | 1.00 | |
| By discharge disposition | Model 1 | |||
| Home or self‐care | Substance use disorder | 72 (8.4) | 1.47 (1.16–1.85)** | 1.33 (1.05–1.69)* |
| No substance use disorder | 810 (5.8) | 1.00 | 1.00 | |
| Home health service | Substance use disorder | 40 (11.8) | 1.08 (0.79–1.46) | 1.06 (0.78–1.44) |
| No substance use disorder | 412 (10.9) | 1.00 | 1.00 | |
| Inpatient facility | Substance use disorder | 32 (9.6) | 1.03 (0.73–1.47) | 1.03 (0.72–1.47) |
| No substance use disorder | 222 (9.3) | 1.00 | 1.00 | |
| Eloped/AMA | Substance use disorder | 15 (14.3) | 1.12 (0.63–1.98) | 1.07 (0.60–1.90) |
| No substance use disorder | 31 (12.8) | 1.00 | 1.00 |
Abbreviations: aRR, adjusted risk ratio; cRR, crude risk ratio.
Adjusted for age, gender, race/ethnicity, insurance, mental health diagnosis, Charlson Comorbidity Index and hospital.
P < 0.05,
P < 0.01,
P < 0.001.
The association between having any SUD and 30‐day unplanned readmission was modified by discharge disposition after the index admission (Table 2). After adjusting for covariates, among those who were discharged to home or self‐care, patients with SUDs were 1.33 (95% CI = 1.05–1.69) times as likely to have a 30‐day unplanned readmission compared to patients without SUDs. There was no association between SUDs and 30‐day unplanned readmission among patients discharged to a home health service (aRR = 1.06, 95% CI = 0.78–1.44), an inpatient facility (aRR = 1.03, 95% CI = 0.72–1.47), or eloped (aRR = 1.07, 95% CI = 0.60–1.90).
In 2022, 19 895 (90.0%) patients had no unplanned 30‐day readmissions and 1630 (7.4%), 364 (1.6%) and 219 (1.0%) had one, two and three or more, respectively. Among patients who had zero, one, two and three or more unplanned readmissions in 2022, 1403 (7.1%), 143 (8.8%), 51 (14.0%) and 34 (15.5%) had at least one SUD at their index admission, respectively. Similarly, 773 (3.9%), 71 (4.4%), 25 (6.9%) and 13 (5.9%) had AUD, and 428 (2.2%), 56 (3.4%), 24 (6.6%) and 17 (7.8%) had OUD, respectively (Figure 2). Cochran‐Armitage tests suggest a trend between any SUD (P < 0.001), AUD (P = 0.003) and OUD (P < 0.001) and number of 30‐day unplanned readmissions in 2022.
FIGURE 2.
Prevalence of any substance use disorder, alcohol use disorder and opioid use disorder by number of 30‐day unplanned readmissions, January–December 2022. *P < 0.05, **P < 0.01, ***P < 0.001.
DISCUSSION
Findings confirmed our hypothesis and show that although the prevalence of SUDs in patients admitted for 30‐day unplanned readmissions at these two academic medical centers was less than one‐half the prevalence of past year SUDs in the general population, these patients accounted for disproportionate and significantly high 30‐day unplanned readmissions. Specifically, patients with any SUD at their index admission were 24% more likely to have a 30‐day unplanned readmission than those without. Among those with SUDs, those specifically with OUD at their index admission were 40% more likely to have a 30‐day unplanned readmission, in comparison to patients without any SUDs.
These findings extend previous studies that have observed relationships between certain SUDs and unplanned readmission, but not studies that have linked AUD to unplanned readmission [9, 10, 11, 15]. The null finding related to AUD in this study may be explained by unmeasured covariates such as housing status, income and employment, and AUD severity as well as measurement error because of underdiagnosis of AUD in inpatient settings or insufficient statistical power [9, 29]. Together with other research suggesting association between AUD and unplanned readmission and our findings of a strong bivariate association that neared statistical significance in the adjusted test, there is sufficient evidence for further research to better understand how the nuances of this potential association may be illuminated by the aforementioned factors.
The exceptionally high risks for 30‐day unplanned readmission among patients admitted with OUD were unexpected, particularly because these patients were not treated for primary OUD. These medical centers do not have organized OUD inpatient treatment units. Therefore, it is hypothesized that reasons these patients face higher readmission risks than their peers could be linked to medical conditions and comorbidities associated with opioid use such as non‐fatal overdose and infectious diseases like endocarditis, cellulitis and osteomyelitis [35, 36, 37]. Southern California reflects the rest of the country in facing increasing overdose deaths from opioids, which may account for some portion of this relationship [38]. As surprising as these findings are, they likely generalize to other academic medical centers that have no formal inpatient unit for treatment of OUD.
Novel findings showing a significant link between use of post‐acute care and 30‐day unplanned readmission risk may reflect the profound and disorganizing effects of SUDs on the overall health of patients. The finding that the relationship between having any SUD and 30‐day unplanned readmission only held for patients who were discharged to home or to self‐care may reflect disorganization at home and cumulative negative health effects when unable to consistently follow post‐discharge directions. Although the lack of association in the other three strata (home health service, inpatient facility, eloped/against medical advice) may be a result of insufficient statistical power because of smaller sample sizes for these groups, the strength of the association among those who were discharged to home or self‐care suggests that addressing SUDs during care and after discharge may be a point for innovation in improving health quality for patients, and in improving health quality metrics for the academic medical center. Especially in the setting of the ongoing national opioid overdose epidemic, starting quality improvement efforts for hospitalized patients who have a diagnosis of OUD seems particularly well reasoned.
These data do not advise the types of interventions that should be considered, but it seems reasonable to develop a method of screening for OUD among patients during admission and procedures to provide linked referrals to those patients with OUD who are discharged to home or to self‐care. Data that supports this direction come from a randomized trial of hospitalized patients with SUDs showing that patient navigation in the first 3 months following discharge significantly reduced readmissions [39]. Additionally, given that half of patients with SUDs had a documented comorbid mental health diagnosis, such interventions should take a whole‐person care approach that addresses ways mental health may interact with SUDs to affect post‐discharge quality of life and potential for readmission [40, 41, 42]. Should additional studies beyond the scope of this analysis continue to suggest that whole‐person patient navigation could be an effective intervention for reducing unplanned readmissions, health systems may look to the California Bridge program's model for scaling the provision of low‐threshold SUD treatment involving patient navigation to over 80% of California hospitals [43].
This study also identified a small, but especially high‐risk group of patients who are very frequently admitted. Despite significant reductions in the absolute number of patients who had more than one unplanned 30‐day readmission, there were sharp and significant increases in the prevalence of SUDs among patients with each increase in unplanned readmissions. Among patients with no unplanned readmissions, 7.1% had a SUD, and among patients with three or more unplanned readmissions, 15.5% had a SUD. This analysis, therefore, identified that patients with SUDs are high‐utilizers of inpatient services in comparison to patients without SUDs. Efforts to identify patients with SUDs at any point from admission to discharge as well as implementing interventions that support them during treatment and on discharge seem an important strategy for reducing repeated unplanned readmissions. Future research using statistical models is needed to build on these descriptive findings and better understand risk factors for multiple unplanned readmissions in this population.
The findings of this study are particularly compelling because they account for a variety of background factors including demographic characteristics, severity of comorbidities and mental health conditions, and all patients at two hospitals were included, yielding a large sample size of over 20 000 patients. Additionally, this study included all patients with SUDs and all types of readmissions, in comparison to previous studies that only considered primary and secondary SUD diagnoses, specific readmissions to services such as behavioral health, and particularly high‐risk groups such as patients living with HIV or patients admitted to psychiatric hospitals [9, 10, 11, 15, 16, 17]. Study findings are best understood as advisory to prospective, hypothesis‐driven efforts to assess and intervene on factors linked with SUDs and multiple unplanned readmissions in hospital settings.
Limitations
The primary limitation of this analysis is potential misclassification of both the exposure and outcome. SUDs are underdiagnosed in inpatient settings, especially if they are not the primary reason for hospitalization [29]. Therefore, even though non‐principal SUD diagnoses were included, this study has likely captured patients with the most acute and/or severe SUDs, because it seems unlikely that patients with milder SUDs would have this recorded at any position in their diagnosis list. Additionally, complications of SUDs, but not the disorder itself, often obscure the task of identifying the admitting diagnosis. Additional ICD‐10 codes such as alcoholic fatty liver, alcoholic cardiomyopathy and poisoning by, adverse effect of and underdosing of substances were included in an attempt to address this, but likely did not cover all potential diagnosis codes.
Similarly, it is likely that not all unplanned readmissions were captured, as patients may have used hospitals other than the two included in this analysis, which could contribute to an underestimate of the true association between SUDs and 30‐day unplanned readmission. This analysis did not capture mortality in the 30 days following discharge from the index admission. Although post‐discharge mortality rates are expected to be low, this could potentially cause selection bias [44, 45].
The results of this study cannot not be interpreted as causal and do not entirely account for confounding by both measured factors and unmeasured factors such as housing status and having a primary care physician [9]. Additionally, this analysis does not account for length of stay, which may influence both receipt of a substance use diagnosis and likelihood of readmission [10]. Last, our results may not be generalizable to all hospitals in Los Angeles as well as other geographic locations. The two hospitals included in this analysis are part of a large academic health system and are located in more affluent areas of Los Angeles, and therefore, do not serve a patient population that is representative of all of Los Angeles.
CONCLUSIONS
In this study, having any SUD at admission was linked to higher likelihood of 30‐day unplanned readmission. Addressing SUDs should become a priority for healthcare systems given the costly nature of 30‐day unplanned readmissions to hospitals. Although SUDs are not on the Hospital Readmissions Reduction Program list of primary diagnoses that result in reduced reimbursement for excess 30‐day unplanned readmissions, this analysis highlights the potential role that underlying SUDs may play in driving 30‐day unplanned readmissions for medically complex patients with multiple morbidities and presents a novel point of intervention to reduce costs associated with 30‐day unplanned readmissions. More importantly, the findings from this study highlight the need for interventions focused on identifying patients with SUDs, addressing needs for treatment during hospitalizations and linking patients to outpatient treatment to prevent unplanned readmissions and improve quality of care for patients.
AUTHOR CONTRIBUTIONS
Allison D. Rosen: Conceptualization (equal), data curation (equal), formal analysis (lead), methodology (lead), visualization, writing—original draft (lead). Sae Takada: Conceptualization (equal), methodology (equal), writing—original draft. Catherine Juillard: Conceptualization, writing—review and editing (equal). Yulsi L. Fernandez Montero: Conceptualization, writing—review and editing (equal). Amy M. Richards: Conceptualization, writing—review and editing (equal). Serge Ngekeng: Conceptualization, writing—review and editing (equal). Steven J. Shoptaw: Conceptualization (equal), funding acquisition (equal), supervision (equal), writing—original draft (equal). Michelle A. Bholat: Conceptualization (equal), funding acquisition (equal), supervision (equal), data curation (equal), writing—review and editing (equal).
DECLARATION OF INTERESTS
None.
Supporting information
Figure S1. Flow diagram.
Table S1. ICD‐10 codes used to define substance use disorders and prevalence in the study population at the index admission.
ACKNOWLEDGEMENTS
We acknowledge Karen Grimley, Patricia Alberto, Mary Noli and Meng Wei for their support of this work.
Rosen AD, Takada S, Juillard C, Fernandez Montero YL, Richards AM, Ngekeng S, et al. Unpacking the link between substance use disorders and 30‐day unplanned readmission. Addiction. 2025;120(12):2538–2546. 10.1111/add.70136
Funding information This work was supported by the UCLA Department of Family Medicine.
DATA AVAILABILITY STATEMENT
Data available on request due to privacy/ethical restrictions.
REFERENCES
- 1. Hosp. readmission reduction | CMS . Accessed January 15, 2024. https://www.cms.gov/medicare/quality/value-based-programs/hospital-readmissions [Google Scholar]
- 2. Siddique SM, Tipton K, Leas B, Greysen SR, Mull NK, Lane‐Fall M, et al. Interventions to reduce hospital length of stay in high‐risk populations: a systematic review. JAMA Netw Open. 2021;4(9):e2125846. 10.1001/jamanetworkopen.2021.25846 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Kangovi S, Grande D. Hospital readmissions—not just a measure of quality. JAMA. 2011;306(16):1796–1797. 10.1001/jama.2011.1562 [DOI] [PubMed] [Google Scholar]
- 4. Amritphale A, Fonarow GC, Amritphale N, Omar B, Crook ED. All‐cause unplanned readmissions in the United States: Insights from the nationwide readmission database. Intern Med J. 2023;53(2):262–270. 10.1111/imj.15581 [DOI] [PubMed] [Google Scholar]
- 5. Gruneir A, Dhalla IA, van Walraven C, Fischer HD, Camacho X, Rochon PA, et al. Unplanned readmissions after hospital discharge among patients identified as being at high risk for readmission using a validated predictive algorithm. Open Med. 2011;5(2):e104–e111. [PMC free article] [PubMed] [Google Scholar]
- 6. Tomita M, Murata K, Suzuki H, Osaki C, Matuki E, Komatuzaki K, et al. Multiple risk factors for unplanned readmissions within 1 month of hospital discharge in acute care hospitals in Japan. Int J Nurs Pract. 2024;30(3):e13235. 10.1111/ijn.13235 [DOI] [PubMed] [Google Scholar]
- 7. Rajaguru V, Han W, Kim TH, Shin J, Lee SG. LACE index to predict the high risk of 30‐day readmission: A systematic review and meta‐analysis. J Pers Med. 2022;12(4):545. 10.3390/jpm12040545 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. van Walraven C, Dhalla IA, Bell C, Etchells E, Stiell IG, Zarnke K, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. Can Med Assoc J. 2010;182(6):551–557. 10.1503/cmaj.091117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Walley AY, Paasche‐Orlow M, Lee EC, Forsythe S, Chetty VK, Mitchell S, et al. Acute care hospital utilization among medical inpatients discharged with a substance use disorder diagnosis. J Addict Med. 2012;6(1):50–56. 10.1097/ADM.0b013e318231de51 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Rowell‐Cunsolo TL, Liu J, Hu G, Larson E. Length of hospitalization and hospital readmissions among patients with substance use disorders in New York City, NY USA. Drug Alcohol Depend. 2020;212:107987. 10.1016/j.drugalcdep.2020.107987 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Mark TL, Tomic KS, Kowlessar N, Chu BC, Vandivort‐Warren R, Smith S. Hospital readmission among medicaid patients with an index hospitalization for mental and/or substance use disorder. J Behav Health Serv Res. 2013;40(2):207–221. 10.1007/s11414-013-9323-5 [DOI] [PubMed] [Google Scholar]
- 12. Substance Abuse and Mental Health Services Administration . 2013. Results from the 2012 National Survey on Drug Use and Health: Summary of national findings. NSDUH Series H‐46, HHS Publication No. (SMA) 13–4795. [Google Scholar]
- 13. Substance Abuse and Mental Health Services Administration . Key substance use and mental health indicators in the United States: Results from the 2022 National Survey on Drug Use and Health. Cent Behav Health Stat Qual Subst Abuse Ment Health Serv Adm. 2023:1–162. PMID: (HHS Publication No. PEP23–07–01‐006, NSDUH Series H‐58. https://www.samhsa.gov/data/report/2022-nsduh-annual-national-report [Google Scholar]
- 14. Trivedi C, Desai R, Rafael J, Bui S, Husain K, Rizvi A, et al. Prevalence of substance use disorder among young adults hospitalized in the US hospital: a decade of change. Psychiatry Res. 2022;317:114913. 10.1016/j.psychres.2022.114913 [DOI] [PubMed] [Google Scholar]
- 15. Grant RW, Charlebois ED, Wachter RM. Risk factors for early hospital readmission in patients with AIDS and pneumonia. J Gen Intern Med. 1999;14(9):531–536. 10.1046/j.1525-1497.1999.08157.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Lin CH, Chen YS, Lin CH, Lin KS. Factors affecting time to rehospitalization for patients with major depressive disorder. Psychiatry Clin Neurosci. 2007;61(3):249–254. 10.1111/j.1440-1819.2007.01662.x [DOI] [PubMed] [Google Scholar]
- 17. Bimerew MS, Sonn FCT, Kortenbout WP. Substance abuse and the risk of readmission of people with schizophrenia at Amanuel psychiatric hospital, Ethiopia. Curationis. 2007;30(2):74–81. 10.4102/curationis.v30i2.1078 [DOI] [PubMed] [Google Scholar]
- 18. American Psychiatric Association . Diagnostic and statistical manual of mental disorders 5th Ed., Text Rev ed.; 2022. [Google Scholar]
- 19. Appa A, Baral S, Stein B, Knight K, Gandhi M, Coffin P, et al. The IMPACT (infection management plus addiction care together) pilot: A case series of combined contingency management for substance use disorders and antibiotic adherence in the hospital setting. J Addict Med. 2023;18(2):138–143. 10.1097/ADM.0000000000001254 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Viswanathan S, Detels R, Mehta SH, Macatangay BJC, Kirk GD, Jacobson LP. Level of adherence and HIV RNA suppression in the current era of highly active antiretroviral therapy (HAART). AIDS Behav. 2015;19(4):601–611. 10.1007/s10461-014-0927-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Richards J, Kuhn R. Unsheltered homelessness and health: a literature review. AJPM Focus. 2023;2(1):100043. 10.1016/j.focus.2022.100043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Strunin L, Stone M, Jack B. Understanding rehospitalization risk: can hospital discharge be modified to reduce recurrent hospitalization? J Hosp Med. 2007;2(5):297–304. 10.1002/jhm.206 [DOI] [PubMed] [Google Scholar]
- 23. Nagata JM, Palar K, Gooding HC, Garber AK, Tabler JL, Whittle HJ, et al. Food insecurity, sexual risk, and substance use in young adults. J Adolesc Health off Publ Soc Adolesc Med. 2021;68(1):169–177. 10.1016/j.jadohealth.2020.05.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Richardson L, Wood E, Kerr T. The impact of social, structural and physical environmental factors on transitions into employment among people who inject drugs. Soc Sci Med. 2013;76(1):126–133. 10.1016/j.socscimed.2012.10.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Werner RM, Coe NB, Qi M, Konetzka RT. Patient outcomes after hospital discharge to home with home health care vs to a skilled nursing facility. JAMA Intern Med. 2019;179(5):617–623. 10.1001/jamainternmed.2018.7998 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Couturier B, Carrat F, Hejblum G. A systematic review on the effect of the organisation of hospital discharge on patient health outcomes. BMJ Open. 2016;6(12):e012287. 10.1136/bmjopen-2016-012287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Cao Y(J), Wang Y, Mullahy J, Burns M, Liu Y, Smith M. The relative importance of hospital discharge and patient composition in changing post‐acute care utilization and outcomes among Medicare beneficiaries. Health Serv Insights. 2023;16:11786329231166522. 10.1177/11786329231166522 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. Epidemiology. 2007;18(6):800–804. 10.1097/EDE.0b013e3181577654 [DOI] [PubMed] [Google Scholar]
- 29. Wakeman SE, Herman G, Wilens TE, Regan S. The prevalence of unhealthy alcohol and drug use among inpatients in a general hospital. Subst Abus. 2020;41(3):331–339. 10.1080/08897077.2019.1635961 [DOI] [PubMed] [Google Scholar]
- 30. Kerr AJ, Wang TKM, Jiang Y, Grey C, Wells S, Poppe KK. The importance of considering both primary and secondary diagnostic codes when using administrative health data to study acute coronary syndrome epidemiology (ANZACS‐QI 47). Eur Heart J ‐ Qual Care Clin Outc. 2021;7(6):548–555. 10.1093/ehjqcco/qcaa056 [DOI] [PubMed] [Google Scholar]
- 31. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40(5):373–383. 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]
- 32. Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159(7):702–706. 10.1093/aje/kwh090 [DOI] [PubMed] [Google Scholar]
- 33. McNutt LA, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol. 2003;157(10):940–943. 10.1093/aje/kwg074 [DOI] [PubMed] [Google Scholar]
- 34. R: A Language and Environment for Statistical Computing. Published online R Core Team ed.; 2021. http://www.R-project.org/ [Google Scholar]
- 35. Morita Y, Haruna T, Haruna Y, Nakane E, Yamaji Y, Hayashi H, et al. Thirty‐day readmission after infective endocarditis: analysis from a Nationwide readmission database. J Am Heart Assoc. 2019;8(9):e011598. 10.1161/JAHA.118.011598 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Upadhyaya A, Marks LR, Schwarz ES, Liang SY, Durkin MJ, Liss DB. Care cascade for patients with opioid use disorder and serious injection related infections. Toxicol Commun. 2021;5(1):6–10. 10.1080/24734306.2020.1869899 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Wildenthal JA, Atkinson A, Lewis S, Sayood S, Nolan NS, Cabrera NL, et al. Outcomes of partial oral antibiotic treatment for complicated Staphylococcus aureus bacteremia in people who inject drugs. Clin Infect Dis. 2023;76(3):487–496. 10.1093/cid/ciac714 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. California Overdose Surveillance Dashboard . Prepared by California Department of Public Health (CDPH) ‐ Substance and Addiction Prevention Branch (SAPB). Accessed February 28, 2024. https://skylab.cdph.ca.gov/ODdash/ [Google Scholar]
- 39. Gryczynski J, Nordeck CD, Welsh C, Mitchell SG, O'Grady KE, Schwartz RP. Preventing hospital readmission for patients with comorbid substance use disorder: A randomized trial. Ann Intern Med. 2021;174(7):899–909. 10.7326/M20-5475 [DOI] [PubMed] [Google Scholar]
- 40. Sreenivasan J, Kaul R, Khan MS, Malik A, Usman MS, Michos ED. Mental health disorders and readmissions following acute myocardial infarction in the United States. Sci Rep. 2022;12(1):3327. 10.1038/s41598-022-07234-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Chan WP, Smith SM, Michael C, Jenkins K, Tripodis Y, Scantling D, et al. Characterizing a common phenomenon: why do trauma patients re‐present to the emergency department? J Surg Res. 2024;303:489–498. 10.1016/j.jss.2024.09.068 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Geier TJ, Simske N, Melin S, Trevino C, Murphy P, Schroeder ME, et al. Psychiatric comorbidity in emergency general surgery patients: a prospective observational study. Trauma Surg Acute Care Open. 2023;8(1):e001117. 10.1136/tsaco-2023-001117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Samuels EA, Rosen AD, Speener M, et al. Scaling Emergency Department Opioid Use Disorder Treatment Across California to Reduce Overdose Deaths, 2019‐2023. Am J Public Health. Published Online June 27. 2024;e1–e5. 10.2105/AJPH.2024.307710 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Yu P, Chang DC, Osen HB, Talamini MA. NSQIP reveals significant incidence of death following discharge. J Surg Res. 2011;170(2):e217–e224. 10.1016/j.jss.2011.05.040 [DOI] [PubMed] [Google Scholar]
- 45. Khera R, Wang Y, Bernheim SM, Lin Z, Krumholz HM. Post‐discharge acute care and outcomes following readmission reduction initiatives: National retrospective cohort study of Medicare beneficiaries in the United States. BMJ. 2020;368:l6831. 10.1136/bmj.l6831 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Flow diagram.
Table S1. ICD‐10 codes used to define substance use disorders and prevalence in the study population at the index admission.
Data Availability Statement
Data available on request due to privacy/ethical restrictions.