Distance from care predicts in-hospital mortality in HIV-infected patients with severe sepsis from rural and semi-rural Virginia,USA

Abstract

There are few data regarding outcomes from severe sepsis for HIV-infected patients living in rural or semi-rural settings. We aim to describe the characteristics and predictors of mortality in HIV-infected patients admitted with severe sepsis to the University of Virginia located in semi-rural Charlottesville, Virginia, USA. We queried the University of Virginia Clinical Data Repository for cases with ICD-9 codes that included: (1) infection, (2) acute organ dysfunction, and (3) HIV infection. We reviewed each case to confirm the presence of HIV infection and severe sepsis. We recorded socio-demographic, clinical, and laboratory data. We used a generalised linear mixed-effects model to assess pre-specified predictors of mortality. We identified 74 cases of severe sepsis in HIV-infected patients admitted to University of Virginia since 2001. The median (IQR) age was 44 (36–49), 32 (43%) were women, and 56 (76%) were from ethnic minorities. The median (IQR) CD4+ T-cell count was 81 (7–281) cells/µL. In-hospital mortality was 20%. When adjusted for severity of illness and respiratory failure, patients who lived >40 miles away from care or had a CD4+ T cell count <50 cells/µL had > four-fold increased risk of death compared to the rest of the study population (AOR = 4.18, 95% CI: 1.09-16.07, p = 0.037; AOR = 4.33, 95% CI: 1.15–16.29, p = 0.03). In HIV-infected patients from rural and semi-rural Virginia with severe sepsis, mortality was increased in those that lived far from University of Virginia or had a low CD4+ T cell counts. Our data suggest that rural HIV-infected patients may have limited access to care, which predisposes them to critical illness and a high associated mortality.

Keywords

HIV AIDS sepsis immunosuppression opportunistic infection mortality rural population intensive care unit

Introduction

The rural prevalence of HIV has steadily increased over the last two decades and by 2009 there were approximately 51,505 people living with HIV infection in rural areas of the United States.¹ Access to HIV care is often limited in these settings and patients living in rural areas are less likely to receive antiretroviral therapy (ART) than patients in urban centers.² Poorly controlled HIV infection is associated with increased mortality in rural settings; however, the source of this increased mortality is not fully described.

Rural patients are more likely to have advanced HIV infection at the time of their entry into care and it is possible that opportunistic infections play a role in accelerating their mortality.³ These infections can progress to severe sepsis, which is a systemic inflammatory response syndrome characterised by infection, organ failure, and critical illness with an associated mortality of 30–40%.⁴ Recent reports have detailed sepsis as a common and increasing cause of admission to intensive care units for HIV-infected patients, but there are few reports of outcomes from severe sepsis in these patients.^5,6

Most reports of HIV-infected patients with severe sepsis come from large urban centers. There is a paucity of data from non-urban centers such as the University of Virginia (UVa) hospital, which serves a largely rural and semi-rural patient population. The UVa Ryan White HIV Clinic cares for patients from 52 counties in rural western Virginia (nearly 24,000 square miles) with 19 of these counties designated as underserved by the Department of Health and Human Services.⁷ Therefore, we aimed to determine predictors of mortality for HIV-infected patients with severe sepsis admitted to UVa. We were specifically interested in determining whether increased distance from UVa was associated with increased mortality.

Materials and methods

We obtained approval for this study through the University of Virginia Institutional Review Board (IRB). Since this was a retrospective study using data obtained from hospital charts, the IRB waived the need for written informed consent from the participants. We analysed data from patients with HIV infection and severe sepsis admitted to UVa hospital since 2001 when medical information first became electronically retrievable. The UVa Health System is a 577-bed hospital with approximately 28,000 inpatient admissions and 112,000 outpatient visits a year.⁸ We collected data from the UVa Clinical Data Repository (CDR), a UVa Health System warehouse of computerised records containing patient information. We screened patients in the CDR using the ninth revision of the International Statistical Classification for Diseases (ICD-9) codes for (1) infection, (2) acute organ dysfunction, and (3) HIV infection.⁴

After identifying potential cases, we reviewed individual records to verify each patient’s HIV infection and severe sepsis diagnosis. We defined severe sepsis as microbiologically confirmed or clinically suspected infection, the presence of two or more systemic inflammatory response syndrome (SIRS) criteria (temperature <36℃ or >38℃; heart rate >90 beats per minute; white blood cell (WBC) concentration <4000 cells/µL or >12,000 cells/µL; or respiratory rate >20 breaths per minute) and acute organ dysfunction.^9,10 We defined cardiovascular failure as hypotension with a systolic blood pressure <90 mmHg, a mean arterial pressure (MAP) <70 mmHg, or other documentation of hypotension requiring intravenous fluid resuscitation. We recorded socio-demographic, clinical, and laboratory data and calculated Severe Organ Failure Assessment (SOFA), Acute Physiology And Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology II (SAPS II), and Glasgow Coma Scores (GCS).^11–14

We pre-specified potential risk predictors based on the results of exploratory analyses and previously known published risk factors. To account for correlations between multiple observations of the same subject, we used a generalised linear mixed-effects model to assess significant associations between pre-specified risk predictors and mortality. We further determined which risk predictors to include in the final models using variable clustering analysis based on an appropriate similarity matrix of the candidate predictors.¹⁵ Due to sparseness of data for the distance from care, for ease of interpretation we dichotomised the distance from care into >40 miles or not based on a sensitivity analysis which determined 40 miles as the optimal cut-off value for distance. The distance from care was associated with CD4+ T-cell concentration (p = 0.003). To avoid collinearity, we developed two separate regression models to assess effects of predictors of mortality. Both models were adjusted for the same set of other covariates. A generalised C-index, or area under the receiver operator characteristic (ROC) curve, was utilised as a measure of overall predictive discrimination, which is defined in this study as the ability to separate those patients who died from those who did not. An ROC curve area of 0.5 indicates no discrimination while an area of 1.0 indicates perfect discrimination. Nominal significance level was set at p < 0.05. All statistical analyses were performed using SPSS software (IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.) and R version 3.1.2 (The R Foundation for Statistical Computing, Vienna, Austria).

Results

We confirmed 74 cases of severe sepsis in 64 HIV-infected patients including one case that was not initially identified by the CDR but was identified during the chart review. One patient had four separate episodes of severe sepsis and seven patients had two episodes of severe sepsis. There were 32 (43%) women patients and 56 (76%) patients from ethnic minorities including 49 (66%) non-Hispanic black patients and five (6.8%) Hispanic patients. The median (IQR) age was 44 (36–49) years. The median (IQR) distance from UVa was 36 (0–79) miles and the median (IQR) length of stay in the hospital was ten (5–24) days (Table 1).

Table 1.

Summary statistics of demographic and clinical data at admission.

	N = 74
Demographics
Age, median (IQR)	44 (36–49)
Female, n (%)	32 (43)
Miles from UVa, median (IQR)	36 (0–79)
Living >40 miles from care, n (%)	32 (43)
Minority, n (%)	56 (76)
Days in hospital, median (IQR)	10 (5–24)
HIV
CD4+ T-cell count (cells/µL), median (IQR)	81 (7–281)
Viral load (copies/mL), median (IQR)	59,900 (136–334,846)
Currently on ART, n (%)	28 (38)
New HIV diagnosis, n (%)	9 (12)
Severe sepsis
Temperature ℃, median (IQR)	38 (36–39)
Heart rate, median (IQR)	110 (97–121)
Respiratory rate, median (IQR), N = 71^a	20 (18–24)
MAP (mmHg), median (IQR)	81 (67–97)
Focus of infection, n (%)
Pulmonary	34 (46)
Systemic	12 (16)
Genital/Urinary	11 (15)
Central nervous system	8 (11)
Gastrointestinal	5 (7)
Musculoskeletal	2 (3)
Skin and soft tissue	2 (3)
Cardiovascular	1 (1)
GCS, median (IQR)	15 (13–15)
APACHE II median (IQR), N = 65^a	17 (13–23)
SOFA, median (IQR), N = 65^a	5 (3–7)
SAPS II median (IQR)	36 (30–46)
Laboratory data
WBC (1000 cells/µL), median (IQR)	9 (5–16)
Hemoglobin (g/dL), median (IQR)	11 (10–13)
Hematocrit (%), median (IQR)	35 (28–38)
Platelets (1000 cells/µL), median (IQR)	178 (104–244)
Lactic acid (mmol/L), median (IQR), N = 36^a	2 (1–3)
Sodium (mmol/L), median (IQR)	134 (130–138)
Potassium (mmol/L), median (IQR)	4 (3–5)
Bicarbonate (mmol/L), median (IQR), N = 69 ^a	21 (18–24)
Blood urea nitrogen (mg/dL), median (IQR)	21 (15–41)
Creatinine (mg/dL), median (IQR)	2 (1–3)
Total bilirubin (mg/dL), median (IQR), N = 71^a	1 (0–1)
Albumin (g/dL), median (IQR), N = 71^a	3 (3–4)

UVa: University of Virginia; GCS: Glasgow Coma Scores; WBC: white blood cells

Due to the occurrence of missing data, numbers are less than total N.

Patients had a median (IQR) temperature of 38 (36–39)℃, heart rate of 110 (97–121) beats per minute, respiratory rate of 20 (18–24) breaths per minute, and WBC concentration of 9000 (5000–16,000) cells/µL. The most frequent foci of sepsis-causing infections were the lungs (46%) followed by the genitourinary tract (15%) and the central nervous system (11%). There were 12 (16%) infections that were considered systemic without a primary focus. Frequently encountered organ dysfunction included acute renal failure (50%), central nervous system failure (45%), and cardiovascular failure (41%). Respiratory failure occurred in 26% and was more likely in patients with a pulmonary focus of infection (OR = 4.90, 95% CI: 1.55–15.52, p = 0.007). Vasopressors were required in six (8%) patients within 24 hours of admission. Of the 51 (69%) patients with laboratory-confirmed infections, 30 (59%) pathogens were bacterial of which 14 were Gram-positive and 16 were Gram-negative. There were 13 (25%) fungal, four (8%) mycobacterial, and four (8%) viral infections.

HIV-related data

The median (IQR) CD4+ T-cell count was 81 (7–281) cells/µL. The median (IQR) HIV copies/mL was 59,900 (136–334,845) and there were 16 (22%) patients with an undetectable viral load at admission. Patients were receiving ART at the time of admission in 28 (38%) cases and 12 (16%) patients started ART during their admission. There were nine (12%) patients who were unaware of their HIV status and 30 (41%) patients with a past medical history of opportunistic infection, including 13 (17%) patients with a history of Pneumocystis jirovecii pneumonia (PCP).

Predictors of mortality

In-hospital mortality was 20%. In the exploratory analysis, age, living >40 miles from UVa, temperature, SAPS II score >40, septic shock, platelets <100,000/µL, serum lactate, bicarbonate, albumin, CD4+ T cell concentration <50 cells/µL, failure of three or more organs, and respiratory failure were all predictive of in-hospital mortality (p < 0.05; Table 2). Mortality for patients living ≤40 miles away was 10% compared to 34% for patients living >40 miles away (OR = 4.98, 95% CI: 1.38–17.89, p = 0.014).

Table 2.

Exploratory analysis of clinical predictors for in-hospital mortality.

	OR	95% CI	p
Demographics
Age	0.53	0.29–0.97	0.039
Gender (Women vs. Men)	0.59	0.17–2.04	0.406
Living >40 miles from UVa	4.98	1.38–17.89	0.014
Minority	1.36	0.33–5.58	0.666
Substance abuse	0.42	0.13–1.43	0.167
Length of stay (in log10 scale)	0.83	0.38–1.80	0.636
HIV
CD4+ T-cell count <50 cells/µL	5.36	1.49–19.31	0.010
Viral load (copies/mL)	1.64	0.99–2.70	0.054
Currently on ART	0.78	0.23–2.67	0.695
New HIV diagnosis	2.21	0.48–10.21	0.310
Severe sepsis
Temperature ℃	0.19	0.05–0.68	0.011
Heart rate	1.57	0.65–3.76	0.316
Respiratory rate	0.73	0.45–1.17	0.189
MAP (mmHg)	1.05	0.35–3.20	0.929
Focus of infection
Pulmonary	2.92	0.87–9.77	0.083
Systemic	0.75	0.15–3.80	0.732
Central nervous system	1.11	0.36–3.39	0.854
Gastrointestinal	0.98	0.10–9.32	0.988
Musculoskeletal	4.14	0.24–70.70	0.326
GCS	0.73	0.52–1.02	0.064
Bacterial infection	11.14	1.30–95.23	0.028
APACHE II >24	3.07	0.84–11.26	0.091
SOFA >4.5	2.17	0.64–7.30	0.212
SAPS II >40	5.36	1.48–19.40	0.011
Septic shock	7.13	1.08–46.83	0.041
Laboratory data
WBC (1000 cells/µL)	0.35	0.09–1.43	0.143
Hemoglobin (g/dL)	0.72	0.33–1.60	0.426
Hematocrit (%)	0.78	0.31–1.97	0.601
Platelets <100,000/µL	4.29	1.26–14.64	0.020
Lactic acid (mmol/L)^a	2.97	1.38–6.40	0.005
Sodium (mmol/L)	0.75	0.35–1.63	0.467
Potassium (mmol/L)	1.44	0.77–2.72	0.257
Bicarbonate (mmol/L)^a	0.46	0.23–0.91	0.025
Blood urea nitrogen (mg/dL)	1.23	0.84–1.82	0.292
Creatinine (mg/dL)	0.87	0.51–1.50	0.622
Total bilirubin (mg/dL)^a	1.14	0.95–1.37	0.171
Albumin (g/dL)^a	0.25	0.10–0.63	0.003
Organ failure
Single-organ failure	0.38	0.11–1.31	0.125
Two-organ failure	0.57	0.14–2.31	0.430
Three-organ failure	4.48	1.35–14.83	0.014
CNS failure	1.11	0.36–3.39	0.854
Respiratory failure	4.99	1.48–16.81	0.010
Renal failure	0.60	0.19–1.93	0.393

ART: antiretroviral therapy; CNS: central nervous system; GCS: Glasgow Coma Scores; WBC: white blood cells

Due to the occurrence of missing data, numbers do not add up to total N.

Distance from UVa was associated with CD4+ T-cell count. Compared to patients with a CD4+ T cell count ≥50 cells/µL, patients with a CD4+ T-cell count <50 cells/µL were almost five times more likely to live >40 miles from UVa (OR = 4.70, 95% CI: 1.72–12.86, p = 0.003). Therefore, to avoid collinearity in the model, we analysed these two variables in separate models for our primary outcome of in-hospital mortality. After we adjusted for severity of illness and respiratory failure, both variables remained statistically significant predictors of in-hospital death (living >40 miles from UVa: AOR = 4.18, 95% CI: 1.09–16.07, p = 0.037; CD4+ T-cell count <50 cells/µL: AOR = 4.33, 95% CI: 1.15–16.29, p = 0.03; Tables 3 and 4). The generalised C-index for the model with distance from the care and the model with CD4+ T-cell count were 0.79 and 0.81, respectively, which indicated that both models had very good predictive discrimination.

Table 3.

A generalized linear mixed-effects analysis of clinical predictors for in-hospital mortality including living >40 miles from UVa.

	OR	95% CI	p
Living >40 miles from UVa	4.18	1.09–16.07	0.037
Respiratory failure	3.46	0.93–12.79	0.063
SAPS II >40	3.27	0.83–12.85	0.090

UVa: University of Virginia.

Table 4.

A generalised linear mixed-effects analysis of clinical predictors for in-hospital mortality including CD4+ T-cell count <50/uL.

	OR	95% CI	p
CD4+ T-cell count <50/uL	4.33	1.15–16.29	0.030
Respiratory failure	2.54	0.63–10.25	0.190
SAPS II > 40	3.75	0.93–15.13	0.064

Discussion

In the first study of severe sepsis in a rural or semi-rural HIV-infected patient population, we have shown that increased distance from a tertiary care center such as UVa is predictive of in-hospital mortality in patients with HIV infection and severe sepsis. We also found that patients that lived far from UVa had lower CD4+ T-cell counts. Severe sepsis is a serious illness in this population, with an associated in-hospital mortality of 20%, which was similar to urban cohorts.^6,16,17

There are several potential barriers to providing care for rural HIV-infected patients. Local medical communities may have little experience with HIV medicine and resources for ART may be limited. Travel to facilities with HIV medicine experience is often expensive and difficult to coordinate for rural patients. This is true for patients attending the UVa HIV clinic where 41% have a low socio-economic status.⁷ These barriers to care often lead to compromised immunity and opportunistic infections for rural HIV-infected patients.² In our study, we found an inverse relationship between CD4+ T-cell count and distance from care at UVa. Many study patients also had a history of opportunistic infections, which suggests a delayed diagnosis of HIV infection and limited access to care for this rural population. Like other studies, we found the risk of death from severe sepsis increased with decreasing CD4+ T-cell counts.^18,19 Use of ART could improve immune function, however patients were receiving ART at admission in only 28 (38%) of identified cases of severe sepsis. As well as issues related to transport and local HIV care, rural HIV-infected patients also suffer disproportionate stigma against HIV infection, and often lack access to substance abuse and psychological services that could promote ART adherence.²⁰

Not only do HIV-infected rural patients often lack access to care for HIV infection, but they are more likely to have limited access to hospitals with critical care expertise, which is important because admission to intensive care units (ICUs) with a high case load has been associated with improved outcomes compared to low volume ICUs.^21–23 Accordingly, mortality for patients admitted with severe sepsis to small rural hospitals may be increased due to lack of critical care expertise compared to tertiary referral hospitals.²⁴ Additionally, minorities and Medicaid patients, who suffer disproportionate rates of HIV infection, are at increased risk of limited access to emergency care.^25,26 When local critical care is unavailable, patients are often transferred to regional tertiary care centers such as UVa. However, this may lead to delays in intensive care management and worse outcomes as increased distance from facilities with expertise in critical care has also been associated with increased mortality.^25–27

Even when available, HIV-infected patients are often less likely to be admitted to an ICU and receive ICU interventions, and have a greater in-hospital mortality than those without HIV infection.^17,28,29 This discrepancy may be due in part to disparities in health insurance between HIV-infected and uninfected persons as uninsured patients are less likely to receive ICU-level care than patients with insurance.³⁰ For example, in one study, HIV-infected patients with PCP with Medicaid insurance were less likely to receive ICU care.³¹ It is also possible that patients in rural areas have a higher threshold for seeking medical care and may present to hospital with more advanced illness compared to urban patients.²⁰

Successful rural healthcare in general and HIV care in particular likely require an integrated approach to surmount barriers to care. One mechanism by which to bridge the medical resource gap in rural areas is telemedicine. Telemedicine can bring together a multidisciplinary team of subspecialists to conduct clinics to successfully deliver HIV care to incarcerated patients and patients in remote settings.^32,33 Mobile telephone text messaging is also effective in enhancing adherence to ART compared to standard care.³⁴ Substance abuse is common in rural Virginia and HIV-infected patients and the UVa HIV clinic has shown that bidirectional text messaging is a successful strategy to encourage adherence to ART in HIV-infected substance abusers.^7,35–37 Delivering high-quality critical care to rural patients is also challenging. Solutions to this problem include addressing local workforce shortages, efficient and specialist-led retrieval systems, as well as adjunctive responses via telemedicine.³⁸ Remote telemedicine ICUs can also assist with screening and management of rural patients with severe sepsis and lead to improved outcomes.^39–41

This study had several limitations. Our sample size was relatively small but similar to other studies of severe sepsis in HIV-infected patients.^16,42–44 We identified one case during chart review and it is possible that our case definition for screening by the CDR missed additional cases, but we do not expect that bias was introduced by our screening mechanism. Due to the retrospective nature of the study, not all data, e.g. APACHE II scores, were available for each patient. We used distance from UVa as a proxy for rural location, but it is possible that some patients living far from UVa may reside in an urban setting. However, we did review the county location for each patient. Our data suggest that low CD4+ T-cell concentrations accounted for the difference in mortality for patients living far from UVa but it is possible that other unmeasured variables explain this difference. In fact, it is quite likely that the reason for poorer outcomes for patients living far from UVa was multifactorial and related to both HIV-specific and critical care needs. Despite these limitations, our data shine light on an infrequently studied patient population of patients with severe sepsis and HIV-infection.

Conclusions

Rural HIV-infected patients may have limited access to care, which predisposes them to critical illness and a high associated mortality. Predictors of in-hospital mortality included distance from UVa and a low CD4+ T-cell count. For the first time, our data suggest that HIV-infected patients living far from care have worse outcomes when presenting with severe sepsis than patients living near care. Improved management of HIV infection in rural populations can help prevent infections that could lead to devastating illness such as severe sepsis. Access to care earlier in the disease progression could also potentially improve outcomes in this population by decreasing the severity of illness at admission.

Footnotes

Acknowledgements

Bruce Ellsworth provided socio-demographic information for patients with these data missing from the electronic medical record; the UVa Public Health Sciences, Division of Biomedical Informatics, provided access to the Clinical Data Repository for data collection; and the data were presented in part at the 20th International AIDS Conference (AIDS 2014) in Melbourne, Australia. The University of Virginia Harrison Institution provided funding to EEE and CCM to support this study.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

HIV surveillance in urban and nonurban areas. Centers for Disease Control and Prevention 2012.

Cohn

Berk

Berry

et al.

The care of HIV-infected adults in rural areas of the United States. J Acquir Immune Defic Syndr 2001; 28: 385–392.

Ohl

Tate

Duggal

et al.

Rural residence is associated with delayed care entry and increased mortality among veterans with human immunodeficiency virus infection. Med Care 2010; 48: 1064–1070.

Angus

Linde-Zwirble

Lidicker

et al.

Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001; 29: 1303–1310.

Coquet

Pavie

Palmer

et al.

Survival trends in critically ill HIV-infected patients in the highly active antiretroviral therapy era. Crit Care 2010; 14: R107–R107.

Japiassu

Amancio

Mesquita

et al.

Sepsis is a major determinant of outcome in critically ill HIV/AIDS patients. Crit Care 2010; 14: R152–R152.

Schafer

Brant

Gupta

et al.

Intimate partner violence: a predictor of worse HIV outcomes and engagement in care. AIDS Patient Care STDS 2012; 26: 356–365.

Satyanarayana

Heysell

Scully

et al.

Mycobacterial infections in a large Virginia hospital, 2001-2009. BMC Infect Dis 2011; 11: 113–113.

Bone

Sibbald

Sprung

. The ACCP-SCCM consensus conference on sepsis and organ failure. Chest 1992; 101: 1481–1483.

10.

Levy

Fink

Marshall

et al.

2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003; 31: 1250–1256.

11.

Vincent

Moreno

Takala

et al.

The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Int Care Med 1996; 22: 707–710.

12.

Knaus

Draper

Wagner

et al.

APACHE II: a severity of disease classification system. Crit Care Med 1985; 13: 818–829.

13.

Le Gall

Lemeshow

Saulnier

. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 1993; 270: 2957–2963.

14.

Teasdale

Jennett

. Assessment of coma and impaired consciousness. A practical scale. Lancet 1974; 2: 81–4.

15.

Harrell F. Regression modeling strategies. Corrected ed. New York: Springer Verlag, 2001.

16.

Greenberg

Lennox

Martin

. Outcomes for critically ill patients with HIV and severe sepsis in the era of highly active antiretroviral therapy. J Crit Care 2012; 27: 51–57.

17.

Mrus

Braun

et al.

Impact of HIV/AIDS on care and outcomes of severe sepsis. Crit Care 2005; 9: R623–R630.

18.

Akgun

Tate

Pisani

et al.

Medical ICU admission diagnoses and outcomes in human immunodeficiency virus-infected and virus-uninfected veterans in the combination antiretroviral era. Crit Care Med 2013; 41: 1458–1467.

19.

Jacob

Banura

Baeten

et al.

The impact of early monitored management on survival in hospitalized adult Ugandan patients with severe sepsis: a prospective intervention study. Crit Care Med 2012; 40: 2050–2058.

20.

Reif

Golin

Smith

. Barriers to accessing HIV/AIDS care in North Carolina: rural and urban differences. AIDS Care 2005; 17: 558–565.

21.

Gaieski

Edwards

Kallan

et al.

The relationship between hospital volume and mortality in severe sepsis. Am J Respir Crit Care Med 2014; 190: 665–674.

22.

Glance

Osler

et al.

Impact of patient volume on the mortality rate of adult intensive care unit patients. Crit Care Med 2006; 34: 1925–1934.

23.

Peelen

de Keizer

Peek

et al.

The influence of volume and intensive care unit organization on hospital mortality in patients admitted with severe sepsis: a retrospective multicentre cohort study. Crit Care 2007; 11: R40–R40.

24.

The outcome of patients with sepsis and septic shock presenting to emergency departments in Australia and New Zealand. Crit Care Resusc 2007; 9: 8-18.

25.

Hsia

Srebotnjak

Kanzaria

et al.

System-level health disparities in California emergency departments: minorities and Medicaid patients are at higher risk of losing their emergency departments. Ann Emerg Med 2012; 59: 358–365.

26.

Shen

Hsia

. Changes in emergency department access between 2001 and 2005 among general and vulnerable populations. Am J Public Health 2010; 100: 1462–1469.

27.

Durairaj

Will

Torner

et al.

Prognostic factors for mortality following interhospital transfers to the medical intensive care unit of a tertiary referral center. Crit Care Med 2003; 31: 1981–1986.

28.

Casalino

Wolff

Ravaud

et al.

Impact of HAART advent on admission patterns and survival in HIV-infected patients admitted to an intensive care unit. AIDS 2004; 18: 1429–1433.

29.

Curtis

Yarnold

Schwartz

et al.

Improvements in outcomes of acute respiratory failure for patients with human immunodeficiency virus-related Pneumocystis carinii pneumonia. Am J Respir Crit Care Med 2000; 162: 393–398.

30.

Kemp

Stepp

. Palliative care for patients with acquired immunodeficiency syndrome. Am J Hosp Palliat Care 1995; 12: 14–27.

31.

Curtis

Bennett

Horner

et al.

Variations in intensive care unit utilization for patients with human immunodeficiency virus-related Pneumocystis carinii pneumonia: importance of hospital characteristics and geographic location. Crit Care Med 1998; 26: 668–675.

32.

Waldura

Neff

Dehlendorf

et al.

Teleconsultation improves primary care clinicians' confidence about caring for HIV. J Gen Intern Med 2013; 28: 793–800.

33.

Young JD, Patel M, Badowski M, et al. Improved virologic suppression with HIV subspecialty care in a large prison system using Telemedicine: an observational study with historical controls. Clin Infect Dis 2014; 59: 123–126.

34.

Horvath

Azman

Kennedy

et al.

Mobile phone text messaging for promoting adherence to antiretroviral therapy in patients with HIV infection. Cochr Database Syst Rev 2012; 3: CD009756–CD009756.

35.

Ingersoll

Dillingham

Reynolds

et al.

Development of a personalized bidirectional text messaging tool for HIV adherence assessment and intervention among substance abusers. J Subst Abuse Treat 2014; 46: 66–73.

36.

Wunsch

Nakamoto

Behonick

et al.

Opioid deaths in rural Virginia: a description of the high prevalence of accidental fatalities involving prescribed medications. Am J Addict 2009; 18: 5–14.

37.

Wunsch

Nuzzo

Behonick

et al.

Methadone-related overdose deaths in rural Virginia: 1997 to 2003. J Addict Med 2013; 7: 223–229.

38.

Peake

Judd

. Supporting rural community-based critical care. Curr Opin Crit Care 2007; 13: 720–724.

39.

Boots

Singh

Lipman

. The tyranny of distance: telemedicine for the critically ill in rural Australia. Anaesth Intensive Care 2012; 40: 871–874.

40.

Loyola

Wilhelm

Fornos

. An innovative approach to meeting early goal-directed therapy using telemedicine. Crit Care Nurs Q 2011; 34: 187–199.

41.

Rincon

Bourke

Seiver

. Standardizing sepsis screening and management via a tele-ICU program improves patient care. Telemed J E Health 2011; 17: 560–564.

42.

Amancio

Japiassu

Gomes

et al.

The innate immune response in HIV/AIDS septic shock patients: a comparative study. PLoS One 2013; 8: e68730–e68730.

43.

Lee CC, Chu FY, Ko WC, et al. Different clinical presentation of community-onset bacteremia among human immunodeficiency virus-infected and human immunodeficiency virus-uninfected adults in the ED. Am J Emerg Med 2014; 32: 1241–1247.

44.

Silva

Jr dos Santos

. Sepsis in AIDS patients: clinical, etiological and inflammatory characteristics. J Int AIDS Soc 2013; 16: 17344–17344.