Durability and Long-term Clinical Outcomes of Fecal Microbiota Transplant Treatment in Patients With Recurrent Clostridium difficile Infection

A total of 137 patients were contacted after fecal microbiota transplantation (FMT) for recurrent Clostridium difficile infection. At median 22 months follow-up, 82% were free of C. difficile recurrence and 95% would undergo FMT if they had recurrent C. difficile.

Abstract

Background

Fecal microbiota transplant (FMT) appears safe and effective for treatment of recurrent Clostridium difficile infection (RCDI). However, durability, long-term clinical outcomes, and patient satisfaction after FMT are not well described.

Methods

Eligible patients who received FMT for RCDI at Emory Hospital between 1 July 2012 and 31 December 2016 were contacted via telephone for a follow-up survey. Of 190 eligible patients, 137 (72%) completed the survey.

Results

Median time from last FMT to follow-up was 22 months. Overall, 82% (113/137) of patients at follow-up had no recurrence of C. difficile infection (CDI) post-FMT (non-RCDI group) and 18% (24/137) of patients had CDI post-FMT (RCDI group). Antibiotic exposure for non-CDI infections after FMT was more common in the RCDI group compared to the non-RCDI group (75% vs 38%, P = .0009). Overall, 11% of patients reported improvement or resolution of diagnoses not related to CDI post-FMT, and 33% reported development of a new medical condition or symptom post-FMT. Ninety-five percent of patients (122/128) indicated that they would undergo FMT again, and 70% of these 122 reported that they would prefer FMT to antibiotics as initial treatment if they were to have a CDI recurrence.

Conclusions

In this follow-up survey of outcomes after FMT at a median of 22 months follow-up, 82% of patients had durable cure of CDI. Patients with recurrence had more post-FMT antibiotic exposure, underscoring the need for thoughtful antibiotic use and a potential role for prophylactic microbiome enrichment to reduce recurrence.

Clostridium difficile infection (CDI) can lead to diarrhea that can cause significant morbidity and mortality [1]. It is the most common healthcare-associated infection in the United States, and rates of recurrent CDI (RCDI) appear to be rising [2]. Disruption of the normal gastrointestinal microbiota via antibiotics or other healthcare exposures is a key step in CDI pathogenesis that allows C. difficile to proliferate in the intestine and produce toxins, leading to acute inflammation and diarrhea [3, 4]. In severe cases, CDI can progress to sepsis, pseudomembranous colitis, toxic megacolon, and death [5]. Current first-line treatments for CDI are oral vancomycin or metronidazole. However, recurrence after completion of a first treatment course ranges from 15% to 20% and increases up to 60% after the first recurrence [6]. Given the intestinal microbiota’s role in CDI susceptibility, fecal microbiota transplant (FMT) has emerged as a therapy to modify this risk without the use of antibiotics. However, the long-term efficacy of FMT in reducing RCDI is not well described.

FMT involves transfer of a healthy donor’s stool to a patient’s gastrointestinal tract in an attempt to restore a healthy microbiota [7]. Several studies have indicated that the procedure is safe and effective for CDI, with resolution rates ranging from 85% and 90% [8, 9]. A 2012 multicenter long-term follow-up study that described clinical outcomes in 77 patients included a secondary cure rate of 98%. The study also assessed the development of new medical conditions and improvement of previous medical conditions during the post-FMT period [10, 11]. However, understanding of long-term safety, patient satisfaction, and durability, defined as absence of CDI after FMT, remains limited by small sample sizes and mean follow-up time of less than 1 year. As FMT is becoming increasingly used for RCDI [12], we sought to estimate long-term clinical outcomes of FMT recipients, durability of the procedure, and patient satisfaction.

METHODS

This was a single-center follow-up study of all RCDI patients who underwent FMT at Emory University between July 2012 and December 2016. The FMT protocol has been approved at Emory Healthcare by the Medical Practices, Infection Prevention, Antibiotic Utilization, and the Pharmacy and Therapeutics committees since July 2012. FMT was performed in both outpatient and inpatient settings using either nasogastric tube or colonoscopy. Informed consent for FMT was obtained from all patients or their medical decision makers. Demographic and clinical data were collected by chart extraction and follow-up questionnaire (see Supplementary Materials) administered as a scripted telephone interview. The questionnaire was developed by the coauthors (Y. M., C. K., T. D.) based on previous publications [10, 11] and was not validated. Retrospective chart review and survey questionnaire were approved by the Emory University Institutional Review Board (approval 71277).

The primary study outcome was RCDI cure at the time of the follow-up survey, defined as absence of RCDI after FMT. Patients were assigned to the RCDI group if they had recurrence of diarrhea that was confirmed to be C. difficile positive by polymerase chain reaction (PCR) after FMT. Patients without diarrhea after FMT or with diarrhea that was C. difficile negative by PCR testing were assigned to the non-RCDI group. Pre-FMT data that were collected included patient’s health status at the time of FMT (outpatient, hospitalized, or nursing home facility), number and type of failed CDI antibiotic treatment courses, and donor relationship to patient. Post-FMT data that were collected included recurrence of diarrhea (all cause and CDI related), antibiotic exposures after FMT (classified by antibiotic class and number of antibiotic courses), and history of hospitalization or surgery post-FMT. Improvement or worsening of preexisting medical conditions or development of any new medical conditions or symptoms was recorded. To assess patient satisfaction with FMT, we also asked patients if they would undergo another FMT in case of RCDI and if they would prefer this to antibiotics as an initial treatment option.

Data were analyzed using SAS version 9.4 (SAS institute, Cary, North Carolina). Differences in proportions were tested with χ² and Fisher exact tests as appropriate. Differences in means were tested with 2-sample t tests. Logistic regression was used to estimate etiologic associations and odds ratios (ORs) between potential risk and protective factors for CDI recurrence post-FMT. A P value <.05 was considered statistically significant.

RESULTS

Patient Characteristics

Of 232 patients who received FMT, 26 were deceased and 15 were unable to be reached with available contact information. Of the remaining 191 eligible patients, 137 (71.7%) responded to the survey and completed a majority of the questions (Table 1). Review of the patients who died after FMT suggested that none of the deaths were attributable to FMT (Table 2). Median age of participants was 66 years (range, 19–91); 93/137 participants (68%) were women. The median time from last FMT to telephone follow-up was 22 months (range, 3–51). The majority of patients (116/137, 85%) had received 1 FMT at the time of follow-up; 16/137 (12%) had 2 FMTs, 2/137 (1%) had 3 FMTs, and 3/137 (2%) had 4 FMTs. At the time of FMT, 112/137 patients (82%) were treated in an outpatient setting, 23/137 (17%) were inpatients, and 2/137 (1%) were in a nursing home facility.

Table 1.

Characteristics of Patients Who Underwent Fecal Microbiota Transplant for Recurrent Clostridium difficile Infection from July 2012 to December 2017, Responding to Follow-up Survey

Characteristic	N (total n = 137)
Women, n (%)	93/137 (68)
Median age, in years, at follow-up (range)	66 (19–91)
Race
White	90/137 (66%)
Black	19/137 (14%)
Other	3/137 (2%)
Unspecified	25/137 (18%)
Median follow-up time, in months (range)	22 (3–51)
Admission status at time of FMT
Hospitalized	23/137 (17%)
Nursing home facility	2/137 (1%)
Outpatient	112/137 (82%)
Median number antibiotic courses for CDI pre-FMT	4
Antibiotics received for CDI pre-FMT
Fidaxomicin	53/137 (39%)
Metronidazole	108/137 (79%)
Oral vancomycin	133/137 (97%)
Rifaximin	2/137 (1%)
Route of FMT administration
Colonoscopy	133/137 (97%)
Nasogastric tube	2/137 (1.5%)
Percutaneous endoscopic gastrostomy tube	2/137 (1.5%)
Recurrent diarrhea after FMT	66/137 (48%)
Recurrent Clostridium difficile–positive diarrhea after FMT	24/137 (18%)

Abbreviations: CDI, Clostridium difficile infection; FMT, fecal microbiota transplant.

Table 2.

Characteristics of Patients Who Died in the Post-Fecal Microbiota Transplant Period

Characteristic	N (total n = 26)
Median age, in years, at time of death (range)	72 (22–95)
Median time, in months, from fecal microbiota transplant to time of death (range)	10 (1–33)
Cause of deatha	Time in months post-FMT
Metastatic cancer	9
Chronic kidney disease	2
Heart failure	2
Respiratory failure due to graft rejection	2
Sepsis due to bacteremia	2
Recurrent Clostridium difficile infection	2
Unspecified	2
Bowel obstruction due to anal stricture	1
Myocardial infarction	1
Acute renal failure	1
Ischemic stroke	1
Seizure	1

Abbreviation: FMT, fecal microbiota transplant.

^a Cause of death listed in death summary and/or post-mortem clinician’s communication note.

Donor Characteristics

All but 1 of the eligible patients who underwent FMT at Emory University received screened material from unrelated donors who were unknown to the recipient but known to the author who processed the donated feces (C. K.). These unrelated donors were screened by the consensus donor screening recommendations at the time [13]. In 8 patients who reported receiving additional FMT outside the study institution, 6 reported having a spouse or relative as a donor and 2 had unrelated donors.

Pre-FMT Data

The median number of antibiotic treatment courses that failed before FMT was 4 (range, 1–12). Oral vancomycin was used in 133 of 137 patients (97%), 108 (79%) received metronidazole, 53 (39%) received fidaxomicin, and 2 (1%) received rifaximin (Table 1).

Post-FMT Data

Sixty-six of 137 patients (48%) experienced recurrence of diarrhea between the time of the last FMT and follow-up. Of these patients, 42/66 (64%) had diarrhea that was self-limited or C. difficile PCR negative and not attributed to RCDI. The most commonly reported causes of recurrent diarrhea after FMT not related to RCDI were Crohn’s disease flares, microscopic or ulcerative colitis (6/66, 9%), surgery (3/66, 5%), irritable bowel syndrome (IBS; 2/66, 3%), and self-limited gastroenteritis (2/66, 3%). The remaining 24/66 patients (36%) with diarrhea who had positive PCR testing for C. difficile after FMT were classified as having RCDI. Table 3 shows post-FMT data from all participants at time of follow-up as well as participants stratified by recurrence and nonrecurrence of CDI.

Table 3.

Crude and Stratified Frequencies of Recurrent Clostridium difficile Infection Risk Factors and Clinical Outcomes in Patients Who Underwent Fecal Microbiota Transplant for Recurrent C. difficile Infection July 2012 to December 2017 and Responded to Follow-up Survey

Characteristic	Overall (n = 137)	Non-RCDI (n = 113)	RCDI (n = 24)	P Valuea
Antibiotic use after last FMT	61/137 (45%)	43/113 (38%)	18/24 (75%)	.0009
Antibiotic use by class after FMT
Cephalosporin	18/61 (30%)	9/34 (21%)	9/9 (50%)	.03
Clindamycin	7/61 (11%)	6/43 (14%)	1/18 (6%)	.66
Fluoroquinolone	25/61 (41%)	15/43 (35%)	10/18 (56%)	.16
Penicillin	13/61 (21%)	9/34 (21%)	4/18 (30%)	1.00
Probiotic use after last FMT	61/137 (45%)	46/113 (41%)	15/24 (62%)	.05
Surgeryb	41/130 (32%)	29/108 (27%)	12/22 (55%)	.01
Hospitalizationb	54/131 (41%)	36/108 (33%)	18/23 (78%)	<.0001
New symptom/diagnosisb	45/134 (34%)	38/73 (34%)	7/16 (30%)	.81
Improved symptoms/diagnosisb	15/135 (11%)	13/111 (11%)	2/24 (8%)	1.00
Median weight change, in pounds (interquartile range)	5 (−5, 10)	5 (−3, 10)	0 (−9, 8)	.18

Abbreviations: FMT, fecal microbiota transplant; RCDI, recurrent Clostridium difficile infection.

^aFor χ² test, Fisher exact tests for low frequency values, and t tests for difference of means.

^bNot all respondents completed this question.

Overall antibiotic use post-FMT for indications other than C. difficile was reported in 61/137 (45%) respondents. Stratified by recurrence, non-CDI antibiotic use after FMT was more common in the RCDI group compared to the non-RCDI group (75% and 38%, respectively; P = .0009). The median number of antibiotic courses post-FMT was 2 (range, 1–10). Stratified by antibiotic use, RCDI was found in 18/61 patients who used antibiotics in the post-FMT period (29.5%) compared to 6/76 patients who did not use antibiotics (7.9%). The most frequently prescribed antibiotics post-FMT overall and in the RCDI group were quinolones, cephalosporins, macrolides, and nitrofurantoin. In the non-RCDI group, quinolones, macrolides, penicillins, and cephalosporins were the most frequently prescribed antibiotics. Receipt of antibiotics previously reported to be associated with CDI (cephalosporins, clindamycin, and fluoroquinolones) had an estimated OR of 3.78 (95% confidence interval [CI], 1.16–12.30) for RCDI after FMT [14]. ORs for RCDI after use of these antibiotics were similar after controlling for receipt of proton pump inhibitors or probiotics. There was no clear positive association with increasing count of antibiotics after controlling for receipt of an antibiotic with higher risk of CDI (cephalosporin, clindamycin, fluoroquinolone), as shown in Table 4.

Table 4.

Number of Antibiotic Courses After Fecal Microbiota Transplant (FMT) and Crude and Adjusted Odds Ratios for Recurrent Clostridium difficile Infection in Patients Who Underwent FMT From July 2012 to December 2016 and Responded to the Follow-up Survey

Antibiotic Courses After Fecal Microbiota Transplant	Overall (n = 137)	Non-RCDI (n = 113)	RCDI (n = 24)	Crude OR, (95% CI)	Adjusted OR,a (95% CI)
0	76	70	6	—	—
1	19	16	3	2.19 (0.49–9.69)	1.82 (0.40–8.39)
2	17	11	6	6.36 (1.74–23.30)	5.00 (1.30–19.27)
3	7	5	2	4.67 (0.74–39.37)	2.10 (0.25–17.67)
4	10	8	2	2.92 (0.50–16.95)	1.58 (0.22–11.38)
>4	8	3	5	19.44 (3.71–101.90)	9.23 (1.41–60.98)

Abbreviations: CI, confidence interval; OR, odds ratio; RCDI, recurrent Clostridium difficile infection.

^aControlling for receipt of cephalosporin, fluoroquinolone, or clindamycin.

Regarding medical complications post-FMT, 54/137 (39%) of respondents reported hospitalization after FMT; however, no patients were hospitalized for reasons directly related to FMT. Patients with RCDI were more likely to be hospitalized (18/24, 75%) than patients without RCDI (36/113, 32%). Nine hospitalizations in the RCDI group were related to C. difficile complications, while 1 hospitalization in the non-RCDI group was related to chronic complications of a previous CDI. Overall, 41/137 patients (30%) had surgery post-FMT for medical conditions not related to the FMT procedure. Specifically, 50% in the RCDI group (12 of 24 patients) had surgery and 26% in the non-RCDI group (29 of 113 patients, P = .024) had surgery; 3 of the surgeries in the RCDI group were due to CDI complications (colectomy in all 3 cases); 2 of the surgeries in the non-RCDI group were due to chronic complications of prior CDI (abdominal wall repair due to toxic megacolon and partial colectomy). One patient was noted to have an episode of hemoptysis 1 year after FMT administered via gastrostomy tube; results of Helicobacter pylori testing were not available.

Overall, 12/137 patients (9%) reported resolution or improvement of a preexisting medical condition other than CDI post-FMT; these medical conditions included Crohn’s disease, IBS, diverticulitis, and common variable immune deficiency (CVID)–associated infections (Table 5). After further review of the medical record of the patient with CVID, the patient had been experiencing intermittent infections and was still receiving intravenous immunoglobulin, so this may have primarily been an improvement in the patient’s perceived degree of illness. Forty-three of 137 patients (31%) reported diagnosis of a new medical condition or development of new symptoms post-FMT, as shown in Table 6. Interestingly, Crohn’s disease was one of the medical conditions reported to have worsened post-FMT; 2 patients experienced a Crohn’s flare-up, including 1 patient who developed pyoderma gangrenosum. Patients reported a median weight gain of 5 pounds (range, −53–60) from time of FMT to time of follow-up. Stratifying by RCDI, the RCDI group had a median 0-pound change in weight, and the non-RCDI group had a median increase of 5 pounds from FMT to time of follow-up.

Table 5.

Improvement in Preexisting Medical Conditions in Respondents to Survey of Patients Who Underwent Fecal Microbiota Transplant From July 2012 to December 2016

Improved Medical Condition	N (total n = 137)a
Conditions previously or potentially associated with intestinal microbiotab
Irritable bowel syndrome	3
Crohn’s disease	2
Diverticulosis	2
Hypercholesterolemia	1
Toxic megacolon	1
Diabetes mellitus	1
Rheumatoid arthritis	1
Conditions not previously associated with intestinal microbiota
Common variable immunodeficiency–associated infections	1
Low back pain	1
Hypertension	1

^aPercentages not reported, as denominator not well defined for these conditions in this population.

^bSee [15–22].

Table 6.

New Diagnoses or Symptoms in Respondents to Survey of Patients Who Underwent Fecal Microbiota Transplant From July 2012 to December 2016

New Diagnosis or Symptom After FMT	N (total n = 137)a
Conditions previously or potentially associated with intestinal microbiotab
Weight gain	76
Constipation	3
Diabetes mellitus	2
Microscopic colitis	1
Gastric ulcer	1
Osteoporosis	1
Nonalcoholic fatty liver disease	1
Femoral avascular necrosis	1
Crohn’s disease flare-up	1
Colon cancer	1
Conditions not previously associated with intestinal microbiota
Kidney disease/end-stage renal disease	5
Hypothyroidism	2
Back pain	2
Stroke	2
Memory loss	2
Arrhythmias	2
Hemoptysis	1
Pyoderma	1
Ménière’s disease exacerbation	1
Unilateral vision loss	1
Sustained ventricular tachycardia	1
Peeling skin	1
Arthritis	1
Peanut allergy	1
Narcolepsy	1
Aspiration	1
Dysautonomia	1
Pleural effusion	1
Chronic pancreatitis	1
Carpal tunnel syndrome	1
Benign prostatic hyperplasia	1
Anxiety	1
Tonic-clonic seizure	1

Abbreviation: FMT, fecal microbiota transplant.

^aPercentages not reported, as denominator not well defined for these conditions in this population.

^bSee [15–19, 23, 24].

Patients reported high levels of satisfaction with FMT for RCDI. Of all respondents, 95% indicated willingness to repeat the procedure in case of a future C. difficile episode. Of those who indicated willingness to repeat the procedure, 70% reported a preference for FMT as the initial treatment option over antibiotic therapy, 26% preferred initial retreatment with antibiotics, and 4% were undecided between the 2 treatment options.

DISCUSSION

In this telephone follow-up survey of durability of FMT, 82% of 137 respondents were free from CDI recurrence at a median of 22 months follow-up. Given that FMT has been reported in other studies to exhibit short-term cure for RCDI of approximately 90%, these data suggest that FMT may also have long-term effectiveness in breaking the cycle of RCDI [10, 25].

These findings underscore the value of FMT in avoiding repeated antibiotic administration that can perpetuate intestinal dysbiosis. However, these data also highlight the limitations of FMT and other factors that modify risk of RCDI. Forty-five percent of all participants received antibiotics for non-CDI indications after FMT. Receipt of cephalosporin, clindamycin, or fluoroquinolone after FMT compared to receipt of other antibiotics had an estimated OR of 3.78 (95% CI, 1.16–12.30) for CDI recurrence. Despite the increased odds of RCDI with these antibiotics, increasing numbers of antibiotic courses did not have a clear positive linear association with odds of RCDI. This suggests that attention to antibiotic selection in this patient group may pay dividends in mitigating RCDI after FMT [26]. Probiotics have been variably shown to be protective, and use of proton pump inhibitors is a known risk factor for CDI; however, these effects were not seen in this retrospective series [27, 28]. Probiotic use was more frequent in patients with RCDI compared to patients without RCDI (62% vs 41%, respectively; P = .05). However, we suspect that this represents bias related to a higher rate of probiotic use either for higher perceived risk of RCDI among these patients due to underlying illnesses, antibiotic exposures, or some other cause rather than a causal effect of probiotics in increasing risk of RCDI. Though improved microbiome diversity has been well demonstrated after FMT, the durability of this effect when faced with disruption from repeat antibiotic challenges has been poorly defined. The potential benefits of prophylactic FMT or other microbial therapeutics to improve microbiome diversity after exposure to these antibiotic classes should be further studied. Creation of microbiome diversity indices to estimate the health of a specific patient’s intestinal microbiome may also assist in better identifying patients who would benefit most from such interventions [29].

Beyond interrupting cycles of RCDI, FMT is being increasingly studied for other indications. As noted in other follow-up studies of FMT for RCDI [10], patients in our study reported improvement in preexisting medical conditions as well as development of new medical conditions in the post-FMT period. Five patients reported improvement in their preexisting inflammatory bowel disease (IBD) or IBS, and 2 patients reported worsening of IBD. This inconsistent response echoes the muddy signal from multiple recent clinical trials of FMT for IBD and IBS that have yet to demonstrate consistent therapeutic benefit, though many maintain hope that the intestinal microbiome will be an amenable therapeutic target for these debilitating diseases [30–32]. One patient was diagnosed with nonalcoholic fatty liver disease (NAFLD) after FMT. There are a number of hypothesis-generating studies that have suggested associations between the intestinal microbiota and NAFLD. However, without a control group, we were not able to further evaluate a potential causal association with FMT in this case [15].

One patient in our study was diagnosed with colon cancer in the post-FMT period. Though there is increasing recognition of associations between the intestinal microbiome and colorectal cancer, identification of 1 case of colorectal cancer in this series does not dramatically exceed the expected incidence for patients in this age group [33, 34]. The rarity of this event and expected event rate make it unlikely that there was a causal association between FMT and new diagnosis of colorectal cancer. More study is needed to better evaluate the relative contributions of all components of fecal material, including bacteria, archaea, viruses, fungi, protists, human cells, and metabolites, to the safety and efficacy of microbial therapeutics for non-RCDI indications.

The vast majority of patients in our study (97%) received oral vancomycin prior to undergoing FMT but had sufficient symptoms to warrant further consideration of FMT. Though there is significant referral bias in this statistic given the requirement at our institution that patients fail oral vancomycin prior to undergoing FMT, this emphasizes the limitations of this therapeutic approach in treatment of RCDI. Clinicians would do well by their patients to refer them for consideration of FMT after their second recurrence of severe disease or hospitalization for CDI, and hospitals and regulatory agencies should work to ensure wide availability of microbial therapeutics. With regard to the number of antibiotic courses tried before initiation of FMT, the vast majority of patients in our study underwent the procedure after failure of at least 3 courses of CDI antibiotic treatment. However, 1 patient with severe CDI underwent the procedure after failure of 1 antibiotic course due to history of allergic reaction to antibiotics, and 4 patients had FMT after 2 courses of CDI antibiotic treatment due to a previous history of refractory RCDI.

This study’s limitations include its retrospective design and potential recall bias within the phone survey. The majority of FMT recipients at our institution were treated in the outpatient setting (82%), which may have contributed to a referral bias toward the null of post-FMT complications given the anticipated better overall health of patients who undergo an outpatient vs inpatient procedure. Surgery and hospitalization were more common in the RCDI group, but this is likely a reflection of disease severity and the comorbidities commonly associated with RCDI. The retrospective design of the study limited our capacity to collect samples for microbiome analyses to test microbial differences in patients with improved or worsening medical conditions after FMT. The single-center design of this study could raise questions of external validity and generalizability, but baseline patient characteristics are similar to those from a multi-center study from 2012 [10]. Strengths of our study include a larger sample size than prior studies, longer duration of follow-up and use of universal fecal donors who were screened by consensus donor screening recommendations at the time [13]. Patient registries as announced by the American Gastroenterological Association and the Infectious Diseases Society of America will likely further advance our understanding of long-term safety and efficacy of FMT in the treatment of RCDI [35].

Notably, 95% of patients indicated that they would undergo FMT again if necessary, with 70% indicating that it would be their first-line treatment of choice. The “yuck factor” associated with FMT is an often-referenced barrier to more widespread use [36]. However, data suggest that patients who undergo FMT for RCDI are highly satisfied with the procedure and testify to the significant patient-level burden of RCDI.

In summary, this follow-up study has demonstrated that FMT was an effective treatment option in our cohort of 137 RCDI patients. The durability of the procedure was supported by a maximal post-FMT disease-free interval of 51 months with low rates of FMT-related complications and highly positive patient perception of the procedure.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. Y. M. is the primary author of this manuscript. Y. M. planned the study, collected and interpreted the data for the study, and drafted and revised the manuscript. M. H. W. conducted data analysis and revised the manuscript. C. S. K. supervised study planning, data collection, and analysis; searched the literature; and revised the manuscript. T. D. supervised study planning and revised the manuscript. T. W. revised the manuscript. All authors agreed to submit the manuscript and approved the final version of the manuscript.

Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health (NIH).

Financial support. M. H. W. is supported by the National Center for Advancing Translational Sciences of the NIH (UL1TR000454) and by the National Institute of Allergy and Infectious Diseases of the NIH (UM1AI104681).

Potential conflicts of interest. C. S. K. serves as a scientific board advisory member for Luminex Corporation, Inc. All remaining authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.