Secondary Logo

Journal Logo


Comparative performance of the laboratory assays used by a Diagnostic Laboratory Hub for opportunistic infections in people living with HIV

Medina, Nardaa,b; Alastruey-Izquierdo, Anab; Mercado, Danicelac; Bonilla, Oscarc; Pérez, Juan C.c; Aguirre, Luisa; Samayoa, Blancaa; Arathoon, Eduardoa,c; Denning, David W.d,e,f; Rodriguez-Tudela, Juan Luisf; on behalf of Fungired

Author Information
doi: 10.1097/QAD.0000000000002631



In 2017, the WHO published their first guidelines for managing advanced HIV disease and rapid initiation of antiretroviral therapy (ART), which were aimed to reduce global HIV morbidity and mortality [1]. These guidelines recommend a package of interventions to diagnose and treat major opportunistic infections, which continue to be an important cause of AIDS-related mortality [1,2]. Hence, identifying people who are eligible for elements of the package of care, requires quality diagnostic services which have a crucial role to improve clinical outcomes [3–5].

According to UNAIDS, Guatemala has the largest number of people living with HIV (PLWH) in Central America, the highest proportion presenting with advanced HIV disease and high viral loads which are factors that increase the risk of opportunistic infections [6,7]. Despite the development of the HIV healthcare program and increasing coverage of ART, access to accurate testing for opportunistic infections was limited. In 2017, to provide diagnostic services for mycobacterial and fungal infections, a Diagnostic Laboratory Hub (DLH) was implemented linked with a national network of HIV healthcare facilities (HCFs). Here, we evaluate the comparative performance of different diagnostic assays employed in the DLH and describe the functions of this Health System in comparison with those provided by National Reference Laboratories (NRLs).

Materials and methods

Setting and population

From January 2017 through December 2018, diagnostic services for tuberculosis (TB), nontuberculous mycobacteria (NTM), histoplasmosis and cryptococcosis were provided by the DLH to a national network of 13 HCFs. Screening for these opportunistic infections was performed regardless of CD4+ cell count for newly diagnosed patients; patients who had abandoned ART (>90 days) and returned to care, and those on ART when an opportunistic infection was suspected. Clinical specimens were obtained in the HCFs and delivered to the DLH by means of a courier service. In addition, an electronic system was set up to capture patient data, request diagnostic services and report results.

Laboratory procedures

HCFs were requested to send sputum samples, urine, serum, and 10 ml of blood in an Isolator tube (Wampole Laboratories, Cranbury, New Jersey, USA). Additional clinical samples were also received depending on the clinical criteria. Diagnostic methods included smear for mycobacteria; cultures for fungi and mycobacteria; isolator blood culture; in-house PCR for the detection of Mycobacterium tuberculosis and Histoplasma capsulatum[8,9]; ELISA (IMMY, Norman, Oklahoma, USA) to detect Histoplasma antigen in urine and; lateral flow assay (IMMY) to detect cryptococcal antigen (CrAg). Detailed methods are described in the Supplementary material,

Role of the National Reference Laboratories and the Diagnostic Laboratory Hub

To compare the role and functions of the NRLs with the DLH, the official websites of the Center for Disease Control and Prevention (USA), the National Administrations of Laboratories and Health Institutes ‘Dr Carlos G. Malbrán’ (Argentina), Institute Gorgas Reference Central Laboratory (Panama), and the NRL (Guatemala) were searched.

Data analysis

The percentage of the diagnostic performance of each technique, with their corresponding 95% confidence intervals (CIs), was calculated against the number of total cases diagnosed. Chi-squared test was used to compare the positive proportions and significance was set at P = 0.05.

Geographic information system and Google maps tools were used to describe the location of the DLH and HCFs. Statistical analysis was performed using SPSS 19.0 software (IBM Iberica, Madrid, Spain).


Healthcare network

Guatemala has an area of 108 889 km2, with an asphalted road system of 7342 km. The national network included 13 (81.2%) of the 16 available HCFs in the country, which are mainly located in rural areas (12 out of 13; 92.3%). Nine (69.3%) of these facilities are located within an average time of 5 h to the DLH. Regarding the distribution, five (38.5%) HCFs are located in the west, four (30.7%) in the east, two (15.3%) in the north, one (7.6%) in south, and one (7.6%) in the metropolitan area.

Patients characteristic and opportunistic infections diagnosed

During 2017 and 2018, a total of 4245 PLWH were screened for opportunistic infections. Baseline characteristics are shown in Table 1. A total of 2677 (63.1%) patients were male with a mean age of 36 years (range 13–89), and 2127 (50.1%) patients were newly diagnosed with HIV infection. Patients with opportunistic infections were slightly older than without opportunistic infections (38 vs. 36 years, P = 0.002), and had a significantly higher HIV viral load (median log10 5.1 vs. 4.6 copies/ml, P < 0.0001). Only 71% of patients had CD4+ cell counts done (Table 1).

Table 1:
Baseline characteristics of the patients with opportunistic infections screening.

Seven hundred sixteen (16.9%) patients had opportunistic infections: 249 (34.7%) TB, 40 (5.6%) NTM, 227 (31.7%) histoplasmosis, 138 (19.3%) cryptococcosis, and 62 (8.6%) had multiple opportunistic infections. Among patients with multiple opportunistic infections, there were 24 (38.7%) histoplasmosis and TB, 15 (24.1%) cryptococcosis and histoplasmosis, 11 (17.7%) cryptococcosis and TB, three (4.8%) cryptococcosis and NTM, three (4.8%) TB and NTM, three (4.8%) histoplasmosis and NTM, and three (4.8%) triple infections. Almost 79% of opportunistic infections patients had less than 200 CD4+/μl defined as advanced HIV disease by the WHO (Table 1) [1].

Comparative diagnostic performance among techniques

Full screening for mycobacteria, histoplasmosis and cryptococcosis was performed in 3448 (81.2%) patients. Partial screening was done in 797 (18.8%) patients; of those 626 (78.5%) were tested for histoplasmosis, 371 (46.5%) for cryptococcosis and 272 (34.1%) for mycobacteria. The diagnostic performance results were similar between those with full and partial screening. Therefore, all cases were included to have more statistical power (Table 2).

Table 2:
Comparative performance of the techniques employed for the opportunistic infections diagnosis.

A total of 290 cases of TB, 50 of NTM, 271 of histoplasmosis and 170 of cryptococcosis were diagnosed. These numbers encompass single infections and cases with multiple opportunistic infections; thus, total cases were used to assess the comparative performance of the diagnostic techniques (Table 2).

Out of the cases tested for TB (Table 2), direct microscopy was positive in 14.1% (95% CI, 10.5–18.6), culture in 38.1% (95% CI, 32.7–43.9), isolator blood culture in 8.7% (95% CI, 5.6–13.3), and PCR in 92.6% (95% CI, 89.0–95.1). One hundred and sixty-one (55.5%) were only detected by PCR. All TB cases were diagnosed by a combination of PCR and conventional culture. NTM cases were diagnosed by culture.

Of those tested for histoplasmosis (Table 2), isolator blood culture, a longstanding high volume blood culture technique employed before the appearance of urine antigen for histoplasmosis, was positive in 36.3% (95% CI, 29.8–43.3), urine antigen in 72.3% (95% CI, 66.5–77.3), culture of respiratory samples in 8.5% (95% CI, 5.4–13.0), and PCR in 62.7% (95% CI, 56.6–69.4), sputum was the sample used for PCR in 92.6% of the samples tested. Urine antigen and PCR were the only positive test in 95 (35.1%) and 64 (23.6%) cases, respectively. Combining urine antigen and PCR assay the positive rate increased by 23.8% (72.3–96.1%). The median CD4+ cell count of those diagnosed by urine antigen testing or isolator was 33 and 20 cells/μl, respectively (P = 0.667).

For cryptococcal disease, 165 out of 170 cases were diagnosed by serum CrAg. The remaining five cases were diagnosed by cerobrospinal fluid (CSF) CrAg (two cases) and isolator blood culture (three cases). A lumbar puncture was done to 85 (51.5%) of 165 CrAg serum positive patients. A total of 55 (64.7%) cases had cryptococcal meningitis. Out of these cases, 37 (67.3%) were CrAg and culture positive, and 18 (32.7%) were only CrAg positive. Of those, 35 and 16 cases had CD4+ cell counts, with a median of 36 and 47 cells/μl, respectively (P = 0.692).

Functions of the National Reference Laboratories and the Diagnostic Laboratory Hub

A review of the main activities of NRL is summarized in Table 3. The functions listed are consistent with the role of assessment, policy development, and assurance, which contribute to the public health system. Main activities included: confirmatory/reference diagnostic testing, typing of microorganisms, assessment of antibiotic resistance, and standardization of methodologies. In Guatemala, the NRL perform most of these activities; however, do not include a Reference Laboratory for fungal infections. For the DLH the main role is to provide rapid diagnostic services and to coordinate activities to improve the management of opportunistic infections in the network. Indirectly, the DLH collects epidemiological data as a result of the diagnostic activities. This information is shared with the HCF network as well as with the Ministry of Health.

Table 3:
Main activities of the National Reference Centers and the Diagnostic Laboratory Hub in Guatemala.


Access to accurate diagnosis services is crucial to guide therapy and to improve clinical outcomes. In this study, a DLH provided rapid diagnostic services for opportunistic infections to a national network of HCFs. We used a courier service for specimen transportation and a similar strategy implemented in Uganda showed an eight-fold increase in referral samples and TB case detection [10]. In Guatemala, this system was able to provide diagnostic access to 81.2% of the HCFs in the country including the most remote ones.

Regarding TB, a low positivity rate (14.1%) of smears was found. Lower performance of sputum microscopy in HIV patients has been widely described because they frequently have paucibacillary infections [11]. However, our results were lower than those found by previous studies, where sensitivity ranged from 30 to 48% [11,12]. A low positivity rate was also found for TB culture with 38.1 vs. 62.6% in the literature [13]. Nevertheless, culture is essential to isolate the microorganism, and determine drug susceptibility, and monitor resistance rates. To increase the detection rate, it would be necessary to explore the use of different transportation systems that could improve mycobacteria recovery. TB PCR detected 92.6% of the cases, which is similar to other reports (90–100%) [14,15]. Xpert MTB/RIF and lateral flow lipoarabinomannan (LAM) assay were not available at the DLH.

Concerning histoplasmosis, the sensitivity of the urine antigen (72.3%) and the isolator blood culture (36.3%) were lower than previously reported 81–98% and 66.7–74.2%, respectively [16–19]. It is well known that urine antigen is designed to diagnose disseminated histoplasmosis and thus, it is expected to be negative in localized histoplasmosis. On the contrary, in this study clinical symptoms and signs that could have helped to classify the histoplasmosis cases were not recorded by the DLH. Therefore, we assumed that a patient with a positive urine antigen, and/or isolator blood culture had disseminated histoplasmosis. Considering this definition, the number of disseminated histoplasmosis was 205 (77.3%) out of 265 patients. Of those, 94% had a urine antigen positive. In patients with isolator and urine antigen test, seven cases had a positive isolator blood culture and negative result in urine antigen, which means that the use of both techniques increases the positivity rate by 3.5% but at a substantially higher cost. The cost of one determination of Histoplasma antigen in Guatemala is $13 vs. 22 for the isolator blood culture. Other study found that histoplasmosis may cause proteinuria [20] and other kidney damage, that in some cases, could render a false negative antigen result. An assay designed for TB in urine samples, LAM have shown that early morning urine and sample concentration can increase the sensitivity [21]. Further studies are needed to evaluate the influence of these factors. The urine antigen test has resulted in a higher detection of the histoplasmosis cases. Global access to these diagnostic tests, especially in Latin America, would increase awareness as well as the availability of an early diagnosis and treatment which will certainly increase the patient survival.

For cryptococcosis, we found that the positivity rate of the CrAg test in serum was consistent with previous reports (98–100%) [22,23]. Eighteen meningitis cases were diagnosed only by CrAg detection, highlighting the importance of this assay in the rapid diagnosis of this severe infection. CSF culture alone would have only diagnosed 67% of cryptococcal meningitis cases and at a slower response rate. Similar results were reported in other study with 10% of patients with a positive CSF CrAg having a negative culture [24]. The DLH used CrAg CSF results as India ink microscopy is less sensitive and time consuming [25].

Taking into account our findings, we recommend the following approaches. First, the sensitivity of direct microscopy for acid-fast bacilli is very low. It is a demanding technique that requires patience and expertise to get good results but provides a quick answer. Laboratories should evaluate the usefulness of this technique considering their available workforce. Second, culture is an insensitive technique for both, Mycobacteria and fungi, but in our opinion, should be maintained to recover the microorganisms for proper identification, susceptibility testing, and typing. Third, isolator blood culture has a low sensitivity and we do not recommend its use as a diagnostic tool. However, it could be useful for the recovery of Histoplasma isolates when the urine antigen is positive. Therefore, if a patient had a positive urine antigen, we would recommend asking for an urgent isolator blood culture. Fourth, TB PCR should be used because it has a good sensitivity and delivers results in a short time. However, we recommend the implementation of standardized commercial techniques, already available, such as the Xpert MTB/RIF assays. For histoplasmosis there is no commercial PCR system and, although our in-house PCR has an acceptable performance, we cannot give a clear recommendation about its use until further analysis about its comparative diagnostic performance in the diagnosis of localized histoplasmosis is done. Fifth, antigen detection techniques are the current gold standard for diagnosing cryptococcosis and disseminated histoplasmosis and should be available in every single laboratory dealing with opportunistic infections in PLWH. Without these tests, 95 (35%) disseminated histoplasmosis and 18 (32.7%) cryptococcal meningitis cases would have been missed. Currently, these methods are included in the WHO Essential Diagnostics List [26]. Sixth, the rate of patients with multiple opportunistic infections in advanced HIV in Guatemala is substantial. Therefore, optimal case-finding strategies should include screening for potential opportunistic infections irrespective of their clinical presentation, and the local prevalence of different opportunistic infections. Similar diagnostic interventions should be considered in other low resource settings.

This study also demonstrates that the DLH does not interfere with other healthcare systems. The European Center for Diseases Control (ECDC) classifies the main activities of microbiology reference laboratories into five core functions: first, diagnostic confirmation services, which include reference methods for specific pathogens with a low prevalence or not covered by the usual commercial portfolio; second, scientific advice; third, collaboration and research; fourth, provision of reference materials; and fifth, monitoring alert and response [27]. The functions of the NRLs are similar to the core activities outlines by the ECDC, which reflect the homogeneity of the Reference Laboratory activities. Guatemala NRL aims are similar but opportunistic infections diagnosis for PLWH is not included in its activity's portfolio. On the other hand, the main goal of the DLH is to provide quick diagnostics services for improving the care and management of patients as well as to aggregate the epidemiology data obtained from the diagnostic analysis. Besides, with more automation, the DLH can manage thousands of samples without a huge increase in labor costs.

The current study has several limitations. The transport time was not measured and thus we cannot determine their impact. In addition, we did not collect data of patients’ symptoms and the added value of it was not determined. Despite this, a large number of patients and events were analyzed allowing us to make clear diagnostic recommendations for other countries and states that want to improve the diagnosis of opportunistic infections in PLWH. The DLH is a promising approach to provide diagnostic services to a large community of HCFs at an affordable cost.


Authors’ contributions: N.M., A.A.-I., and J.L.R.-T. performed the research. A.A.-I. and J.L.R.-T. designed the study. L.A., O.B., and J.C.P. participated in data extraction and data cleaning. D.M. and B.S. contributed essential reagents or tools. N.M. and J.L.R.-T. analyzed the data. N.M. wrote the article. A.A.-I., E.A., D.W.D, J.L.R.-T. participated in critical revisions.

The authors would like to thank all staff members of the health care facilities of Fungired network.

The current work was supported by Global Action Fund for Fungal Infections and JYLAG, a charity Foundation based in Switzerland (E.A. received this funding under the proposal: ‘Minimising HIV deaths through rapid fungal diagnosis and better care in Guatemala’). Other contributions came from AIDS Health Foundation (AHF) Guatemala, Intrahealth International and Ministry of Health in Guatemala (MSPAS).

Fungired members: Oscar Eduardo López Pérez. Hospital La Amistad Japón-Guatemala, Izabal; Brenan Ortiz Barrientos. Hospital General San Juan de Dios, Guatemala city; Vilma Alejandrina Reyes Muñoz. Hospital Nacional ‘Dr Jorge Vides Molina,’ Huehuetenango; Gladys Sajché Aguilar. Hospital Nacional ‘Juan José Ortega’ Coatepeque, Quetzaltenango; Aura Marina Méndez Andrade. Hospital Nacional de Escuintla, Escuintla; Luis Roberto Santa Marina de León. Hospital Nacional de Malacatán, San Marcos; Ana Lucía Gómez Alcázar. Hospital Nacional de Occidente, Quetzaltenango; Eduardo Celada González. Hospital Nacional de Retalhuleu, Retalhuleu; Gustavo A. Quiñónez M. Hospital Nacional Infantil ‘Elisa Martínez,’ Izabal; Germán Orlando Cuyuch Sontay. Hospital Regional ‘Hellen Lossi de Laugerud,’ Alta Verapaz; Alba Virtud Contreras Marín. Hospital Regional de Cuilapa, Santa Rosa; María de Lourdes Fong Araujo. Hospital Regional de San Benito, Petén, Claudia Mazariegos L. Hospital Regional de Zacapa, Zacapa and Brenda Guzmán. Diagnostic Laboratory Hub, Asociación de Salud Integral, Guatemala City.

Conflicts of interest

A.A.-I. has received research grants or honoraria as a speaker or advisor from Astellas, Gilead Sciences, MSD, Pfizer, F2G, Amplyx, and Scynexis outside the submitted work. E.A. has received honoraria from GILEAD for educational conferences and participation in Advisory board meeting. D.W.D. holds Founder shares in F2G Ltd, a University of Manchester spin-out antifungal discovery company, in Novocyt, which markets the Myconostica real-time molecular assays and has current grant support from the National Institute of Allergy and Infectious Diseases, National Institute of Health Research, North West Lung Centre Charity, Medical Research Council, Global Action Fund for Fungal Infections and the Fungal Infection Trust. He acts or has recently acted as a consultant to Astellas, Sigma Tau, Basilea, Biosergen, Cidara, and Pulmocide. In the past 3 years, he has been paid for talks on behalf of Astellas, Dynamiker, Gilead, Merck, and Pfizer. He is also a member of the Infectious Disease Society of America Aspergilosis Guidelines and European Society for Clinical Microbiology and Infectious Diseases Aspergillosis Guidelines groups. All other authors declare no conflicts of interest.


1. World Health Organization (WHO). Managing advanced HIV disease and rapid initiation of antiretroviral therapy. Geneva, Switzerland: World Health Organization (WHO); 2017.
2. UNAIDS. Global HIV statistics. UNAIDS 2019. Available at [Accessed 27 June 2019]
3. Denning DW. Minimizing fungal disease deaths will allow the UNAIDS target of reducing annual AIDS deaths below 500 000 by 2020 to be realized. Philos Trans R Soc B Biol Sci 2016; 371:20150468.
4. Horton S, Sullivan R, Flanigan J, Fleming KA, Kuti MA, Looi LM, et al. Delivering modern, high-quality, affordable pathology and laboratory medicine to low-income and middle-income countries: a call to action. Lancet 2018; 391:1953–1964.
5. Alemnji G, Fonjungo P, Van Der Pol B, Peter T, Kantor R, Nkengasong J. The centrality of laboratory services in the HIV treatment and prevention cascade: the need for effective linkages and referrals in resource-limited settings. Available at [Accessed 27 June 2019]
6. Samayoa B, Aguirre L, Bonilla O, Medina N, Lau-Bonilla D, Mercado D, et al. The Diagnostic Laboratory Hub: a new healthcare system reveals the incidence and mortality of tuberculosis, histoplasmosis, and cryptococcosis of PWH in Guatemala. Open Forum Infect Dis 2020; 7:ofz534.
7. National Inform, Ministerio de Salud Pública y Asistencia Social G, USAID, Intrahealth. Informe Nacional de la Cascada del Continuo de Atención en VIH. 2018.
8. Montenegro SH, Gilman RH, Sheen P, Cama R, Caviedes L, Hopper T, et al. Improved detection of Mycobacterium tuberculosis in Peruvian children by use of a heminested IS6110 polymerase chain reaction assay. Clin Infect Dis 2003; 36:16–23.
9. Bialek R, Feucht A, Aepinus C, Just-Nübling G, Robertson VJ, Knobloch J, et al. Evaluation of two nested PCR assays for detection of Histoplasma capsulatum DNA in human tissue. J Clin Microbiol 2002; 40:1644–1647.
10. Joloba M, Mwangi C, Alexander H, Nadunga D, Bwanga F, Modi N, et al. Strengthening the Tuberculosis Specimen Referral Network in Uganda: the role of public–private partnerships. J Infect Dis 2016; 213:S41–S46.
11. Swaminathan S, Padmapriyadarsini C, Narendran G. HIV-associated tuberculosis: clinical update. Available at [Accessed 27 June 2019]
12. Claude J, Ngabonziza S, Ssengooba W, Mutua F, Torrea G, Dushime A, et al. Diagnostic performance of smear microscopy and incremental yield of Xpert in detection of pulmonary tuberculosis in Rwanda. Available at [Accessed 27 June 2019]
13. Park SH, Kim CK, Jeong HR, Son H, Kim SH, Park MS. Evaluation and comparison of molecular and conventional diagnostic tests for detecting tuberculosis in Korea, 2013. Osong Public Health Res Perspect 2014; 5:S3–S7.
14. Greco S, Rulli M, Girardi E, Piersimoni C, Saltini C. Diagnostic accuracy of in-house PCR for pulmonary tuberculosis in smear-positive patients: meta-analysis and metaregression. J Clin Microbiol 2009; 47:569–576.
15. Furini AA, da C, Pedro H, da SP, Rodrigues JF, Montenegro LML, et al. Detection of Mycobacterium tuberculosis complex by nested polymerase chain reaction in pulmonary and extrapulmonary specimens. J Bras Pneumol 2013; 39:711–718.
16. Cáceres DH, Samayoa BE, Medina NG, Tobón AM, Guzmán BJ, Mercado D, et al. Multicenter validation of commercial antigenuria reagents to diagnose progressive disseminated histoplasmosis in people living with HIV/AIDS in two Latin American countries. J Clin Microbiol 2018; 56:1–10.
17. Azar MM, Hage CA. Laboratory diagnostics for histoplasmosis. J Clin Microbiol 2017; 55:1612–1620.
18. Arango-Bustamante K, Restrepo A, Cano LE, De Bedout C, Tobón AM, González A. Diagnostic value of culture and serological tests in the diagnosis of histoplasmosis in HIV and non-HIV Colombian patients. Am J Trop Med Hyg 2013; 89:937–942.
19. Guimarães AJ, Nosanchuk JD, Zancopé-Oliveira RM. Diagnosis of histoplasmosis. Braz J Microbiol 2006; 37:1–13.
20. Kushnir MM, Crockett DK, Cloud JL, Ashwood ER, Rockwood AL. Exploratory study of proteins in urine of patients with histoplasma antigenuria. J Chromatogr B Analyt Technol Biomed Life Sci 2012; 883–884:147–154.
21. Gina P, Randall PJ, Muchinga TE, Pooran A, Meldau R, Peter JG, et al. Early morning urine collection to improve urinary lateral flow LAM assay sensitivity in hospitalised patients with HIV-TB co-infection. BMC Infect Dis 2017; 17:339.
22. Kabanda T, Siedner MJ, Klausner JD, Muzoora C, Boulware DR. Point-of-care diagnosis and prognostication of cryptococcal meningitis with the cryptococcal antigen lateral flow assay on cerebrospinal fluid. Clin Infect Dis 2014; 58:113–116.
23. Wake RM, Britz E, Sriruttan C, Rukasha I, Omar T, Spencer DC, et al. High cryptococcal antigen titers in blood are predictive of subclinical cryptococcal meningitis among human immunodeficiency virus-infected patients. Clin Infect Dis 2018; 66:686–692.
24. Ssebambulidde K, Skipper C, Rhein J. Culture-negative cryptococcal meningitis. Lancet Infect Dis 2019; 19:929–930.
25. Boulware DR, Rolfes MA, Rajasingham R, von Hohenberg M, Qin Z, Taseera K, et al. Multisite validation of cryptococcal antigen lateral flow assay and quantification by laser thermal contrast. Emerg Infect Dis 2014; 20:45–53.
26. Second WHO model list of essential in vitro diagnostics. pp. 1–17. Available at [Accessed 30 June 2019]
27. ECDC, European Centre for Disease Prevention and Control. Core functions of microbiology reference laboratories for communicable diseases. 2010; Available at: [Accessed 30 June 2019].

cryptococcosis; histoplasmosis; laboratory diagnosis; opportunistic infections; tuberculosis

Supplemental Digital Content

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.