|Year : 2021 | Volume
| Issue : 4 | Page : 229-235
Surgeons' re-operative valve replacement practices in patients with endocarditis due to drug use
Julie M Aultman1, Oliwier Dziadkowiec2, Dianne McCallister3, Michael S Firstenberg4
1 Department of Family and Community Medicine, Northeast Ohio Medical University, Rootstown, Ohio, USA
2 Graduate Medical Education, HCA Healthcare, Physician Services Group, Denver, USA
3 President, Diagnosis Well, Greenwood Village, CO, Memphis TN, USA
4 Director of Research and Special Projects, William Novick Global Cardiac Alliance, Memphis TN, USA
|Date of Submission||12-Dec-2020|
|Date of Acceptance||01-Jan-2021|
|Date of Web Publication||18-Dec-2021|
Dr. Julie M Aultman
Department of Family and Community Medicine,Northeast Ohio Medical University, Rootstown, Ohio 44272
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background: This study discerns surgeons' attitudes and practices in the determination of heart valve replacement for patients with infectious endocarditis (IE) due to intravenous drug use (IVDU). We aimed to identify the factors contributing to surgeons' decision-making process for initial and recurrent surgical heart valves and the availability of institutional guidance.
Methods: An IRB-approved, anonymous mixed-methods, open survey instrument was designed and validated with 24 questions. A convenience sample of cardiothoracic surgeons in the United States and globally resulted in a total of 220 study participants with 176 completing every question on the survey.
Results: A cluster analysis revealed that although surgeons can be divided into subgroups based on their previous experience with valve replacements, these groups are not perfectly homogenous, and the number of identified clusters is dependent on technique used. Analysis of variance revealed the variables that most clearly divided the surgeons into subgroups were, in order of importance, years of practice, number of valve replacements, and geography.
Conclusions: Our analysis showed heterogeneity among cardiothoracic surgeons regarding how they make clinical decisions regarding re-operative valve replacement related to IE-IVDU. Therefore, an opportunity exists for interprofessional teams to develop comprehensive guidelines to decrease variability in surgical decision-making regarding valve replacement associated with IE-IVDU.
Keywords: Cardiac surgical procedures, endocarditis, heart valve replacement, substance-related disorders
|How to cite this article:|
Aultman JM, Dziadkowiec O, McCallister D, Firstenberg MS. Surgeons' re-operative valve replacement practices in patients with endocarditis due to drug use. Int J Crit Illn Inj Sci 2021;11:229-35
|How to cite this URL:|
Aultman JM, Dziadkowiec O, McCallister D, Firstenberg MS. Surgeons' re-operative valve replacement practices in patients with endocarditis due to drug use. Int J Crit Illn Inj Sci [serial online] 2021 [cited 2022 Jan 26];11:229-35. Available from: https://www.ijciis.org/text.asp?2021/11/4/229/332867
| Introduction|| |
The treatment and management of infectious endocarditis (IE) due to intravenous drug use (IVDU) are largely centered on eradicating infection and reversing the intracardiac and systemic manifestations of infection. Treatment begins with prolonged parenteral antibiotic therapy on admission to the hospital and completed in the outpatient setting. The delivery of parenteral antibiotics via central intravenous access creates concerns that IE-IVDU patients will misuse their intravenous access and create further infection. This is one of the several examples of the intersections of clinical decision-making and moral and social considerations in the treatment and management of endocarditis.
Identifying which patients would benefit from surgery is an important feature of the management of IE; the decision to operate is made using case-by-case surgical judgment and personal experience. The number of IE-IVDU patients requiring surgery increased from 14.8% in 2002 to 26% in 2012. Indications for valve replacement surgery, supported by observational studies, include heart failure, severe regurgitation, persistent infection, infection caused by fungi or highly resistant organisms, and large, embolic vegetations. Recommendations for early surgery in patients with IE-IVDU, along with integrated, comprehensive addiction treatment, have been made in addition to the use of existing general United States (U.S.) and European surgical intervention guidelines for IE.,,,
Despite risks of surgery, patients with IE-IVDU who receive first-time surgery were found to have lower 1-year mortality than patients who received medical treatment only, indicating a survival benefit to patients with surgical indications. Compared to other patients with IE, the 30-day postoperative mortality for patients with IE-IVDU is similar, with some studies estimating mortality at 10% for both groups., Although surgery is initially successful, some studies suggest that long-term outcomes are worse for patients with IE-IVDU, in part because patients' drug addiction was not properly addressed or not at all.,,
For example, in a retrospective study of 180 patients (54 IE-IVDU) from 2011 to 2016, the leading cause of death at mid-term (median equals 20.7 months) was continued drug use (69.2% of patients). Studies have found that 60%–81% of IE-IVDU patients return to drug use following surgery.,, The 1-year survival of patients with recurrent IE was 63.3% compared to 95.4% in patients with a single episode of IE-IVDU, suggesting that recurrent IE-IVDU patients are critically ill with a more complicated course.
The decision for second operations on IE-IVDU patients after relapse of IVDU and reinfection is highly contested and physician dependent. IE-IVDU patients differ from traditional IE patients in terms of characteristics, demographics, diagnosis, and outcomes, which complicate clinical and ethical decision-making for physicians. Published recommendations and consensus guidelines are not specific to IVDU and the sequelae of substance use disorder (SUD), especially in patients with recurrent IE. It is unknown if repeated surgical interventions will benefit patients given their poor long-term outcomes. Some surgeons believe that surgical treatment of repeat IE-IVDU is futile or intervening on an end-stage disease, despite the fact that IE-IVDU patients are younger and could benefit from added life expectancy if they manage to cease their drug use.,
This study discerns surgeons' attitudes and practices in the determination of heart valve replacement for patients with IE-IVDU. There is a paucity of data on the treatment of IE-IVDU along with the layers of social, moral, and medical issues that complicate treatment decisions. Thus, we aim to identify the factors contributing to surgeons' decision-making process for initial and recurrent surgical artificial valves and the availability of institutional guidance.
| Methods|| |
This prospective, descriptive study utilized a mixed-methods, open survey to assess cardiothoracic surgeons' demographics and experience, quantify practices specific to IE-IVDU, and determine the influence of various medical and social factors on decision-making regarding surgical intervention in patients with IE-IVDU. This method was chosen due to the ability to disseminate the survey among a convenience sample of cardiothoracic surgeons in the U.S. and internationally and the ease of completion. The e-mail invitation was sent out to 1000 cardiothoracic surgeons through e-mail lists provided by our cardiothoracic surgeon co-investigator, who has access to two major listservs for purposes of recruitment, including “The Heart Surgery Forum Discussion” and The Society for Thoracic Surgeons.
The study was approved under an expedited review process by the institutional review board at Northeast Ohio Medical University. Because of the anonymous nature of the open survey and minimal risks to human subjects, the protocol was approved under an expedited review process (IRB#18-008). Because of the anonymous nature of this study, a waiver of signed informed consent was approved by the IRB and an informational sheet was given to the participants in a recruitment e-mail and with the survey link describing the purpose, benefits, risks, duration, data storage and maintenance, and contact information for the PI and the Human Subjects Protections Administrator, while describing and emphasizing the voluntary nature of the study. Human subject participants reviewed the informational sheet and participated without compensation.
The open survey was developed by the first author, senior author, and two, dual-enrolled medical graduate students. To improve validity and reliability of the survey tool, face validity with five surgical experts was first established to identify the ease of use, whether we were asking the right questions, and if questions could measure what we aimed to measure. This process was followed by a pilot test with 20 individuals, resulting in minor adjustments to improve clarity and reduce misinterpretations of survey questions. The survey was then hosted on SurveyMonkey.com (Gold Edition) consisting of 28 questions, including demographic information. Participants were given options to use the survey on their laptops, tablets, or smartphones and had the option of viewing one question per screen or one page (total 5) at a time. A screening question was used to determine eligibility (during your career, have you surgically treated patients with IE-IVDU with a valve replacement?). The first 13 questions are discussed in this paper, beginning with background information regarding the surgeon's background information, demographics, and medical and/or surgical interventions. Surgeon participants were permitted to go back to a previous question and revise; however, there was no evidence of this occurring. Survey data were collected over the course of 2 months, and the duration of the entire study from initial development to analysis was from July 2018 to June 2020.
Background information included years in practice in ranges of five, and the total cases per year performed as the primary surgeon in ranges of 50 cases. Demographic information included the choice between urban, suburban, or rural, with the option of selecting all that apply. Region of practice was presented using a map of the U.S. divided between five regions as well as an international option: Pacific, West, Midwest, South, Northwest, and International. Surgeons were asked the greatest number of valve replacements done for a single patient with the option of including a description of the circumstances. The utility of guidelines was assessed by asking surgeons if their institution has specific policies or guidelines in place regarding recurrent IE-IVDU (yes or no options) and whether they feel that guidelines would benefit their practice (yes, no, or undecided options). Surgeons were asked how they gained their current practices regarding recurrent IE-IVDU with the option of choosing textbooks/guidelines, taught by another physician or mentor, gained from personal experience, or other with the option to further specify; surgeons could choose more than one option.
Finally, surgeons were asked to choose which descriptions aligned with their personal belief of addiction, including moral problem, medical problem, social problem, or other with the option to specify; again, surgeons could choose more than one option. Surgeons were asked to consider the percent change in consultations per year for IE-IVDU in the past 5 years, the percent change in cases per year in the past 5 years for first-episode IE-IVDU, and the percent change in cases per year in the past 5 years for recurrent IE-IVDU. Options were presented in 9-point Likert scale and ranged in increments of 25% from “−100% or more,” to “no change,” to “+100% or more.”
Cluster analysis is a multivariate method to classify subjects (or objects), based on a set of variables, into different subgroups such that similar subjects are placed in the same subgroup. In the current study, cluster analysis was performed to explore subgroups of cardiothoracic surgeons based on their responses to the survey questions, i.e., their practices and beliefs surrounding IE-IVDU and valve replacement surgeries.
To identify the optimal number of clusters to extract, we used Gaussian finite mixture models (GFMMs), which fits a variety of models to the data and then compares their fit for the data using Bayesian information criterion (BIC) statistics. The quantitative analysis was performed in R Studio (https://rstudio.com), (R Foundation for Statistical Computing, Vienna, Austria) specifically, package “cluster” was used for cluster analysis., In addition, fixed–effects analysis of variance was used to determine the importance of variables in determining cluster membership.
Thematic analysis was used to analyze all open-ended responses that either clarified or expanded upon surgeons' descriptive survey responses, providing context for their answer choices within the survey.
| Results|| |
Out of the 220 surveys, 175 surveys were returned with complete data fields needed for the analysis [Figure 1]. A total of 224 individuals viewed the first question of the survey following the informational form, and due to our screening question, 220 were eligible and participated. We had a completion rate of 79.5% (175:220), and the average time it took the survey was 12 min. Of the 175 surgeons who completed the survey, 36.6% reported that they have been in practice for 25 years or more followed by 21.1% reporting that they have been in practice for 16–20 years [Table 1]. Of the 175 surgeons who responded, 38.9% reported that they perform more than 200 cases per year as the primary surgeon followed by 29.7% who reported that they perform 151–200 cases per year. When we examined the practice demographics, 64% of surgeons reported the demographics of their practice as an urban setting, followed by 36.5% suburban and 14.9% rural [Table 1].
Twelve percent of the surgeons reported that they are from the South, 39% reported being from the Midwest, 28% reported being from the Northeast, and 21% reported being from the West. Cardiothoracic surgeons most commonly reported that their personal belief of addiction aligns with the idea that it is a social problem (151 responses) followed by medical (115 responses) and moral (63 responses). When asked about their experience with valve replacements, 103 of 175 cardiothoracic surgeons (58.9%) reported that they have performed a maximum of two valve replacements, followed by 44 (25.1%) reporting a maximum of one valve replacement and 26 (14.9%) reporting a maximum of three or more valve replacements in a single patient. In the free response section, one surgeon responded, “only one replacement no matter the circumstances,” and another stated, “denied replacement if still using drugs.” Conversely, one surgeon responded that they would, “will give [a] second or third chance,” another responded that they “will give 3 + replacements if young, healthy, good social support, and sober,” and another indicated that “3 + replacements if duration is long between surgeries.”
When asked whether their institutions have policies or guidelines regarding surgery in patients with recurrent IE-IVDU, 87.5% (154/175) reported “no” with only 7.4% (13/175) reporting “yes.” In the free response section, one surgeon stated that their hospital does have policies, but they make exceptions, another responded that their policy does not recommend reoperations, and a third stated that their institution is currently establishing policy due to the increase in cases and controversial decision-making. When asked if guidelines would be beneficial in their decision making, 50% (88/175) reported “yes,” 19.9% (35/175) reported “no,” and 14.8% (26/175) reported “undecided.” When asked where cardiothoracic surgeons learned their current practices regarding the treatment of patients with recurrent IE-IVDU, surgeons most commonly chose “personal experience” (156 responses), followed by “another physician or mentor” (64 responses) and textbook guidelines (44 responses).
Cardiothoracic surgeons were asked what they perceived to be the percent change in consultations and cases of IE-IVDU in the past 5 years. The majority reported a percent increase in consultations and first-episode, with 24% and 24.6% reporting a 50% increase, respectively. Twenty-one percent reported no change in the cases of recurrent IE-IVDU, although a cumulative 75.9% reported a 25%–100% increase.
To identify the optimal number of clusters to extract, we used GFMM, with the Expectation-Maximization (EM) Algorithm. GFMM method fits a variety of models to the data and then compares their fit for the data using statistics such as BIC. When the models were compared to each other (the colored lines) based on their fit to the data (BIC = −6225.7), the best practically fitting and useful solution consisted of three clusters.
Variable influence on cluster
According to [Table 2], having a practice in a rural demographic, endorsing that addiction is a social problem, performing either 1 or more than 3 valve replacements, and using personal knowledge as source of current practice for subsequent valve replacement were the most influential variables in dividing surgeons into clusters (subgroups).
The first two clusters had a substantial percentage of surgeons with more than 25 years of experience, but Cluster 3 displayed a lower proportion: Cluster 1 having the highest proportion of rural demographic, Cluster 2 having highest proportion of urban demographic, Cluster 1 and 2 about equal proportion of suburban demographic [Table 3]. Higher percentages of surgeons indicated that addiction is a medical problem in Cluster 3, then in Cluster 1 and 2; in addition, Cluster 3 had the lowest percentage of surgeons who indicated that addiction is a moral problem. Interestingly, only Cluster 1 had percentage of surgeons who did not indicate that addiction was a social problem.
Surgeons in Cluster 1, which is the largest cluster, have not performed more than one valve replacement as compared to surgeons in Cluster 3, who all completed more the 3. Surgeons in Cluster 2 performed about two replacements. Related, only in Cluster 1, there was a group of surgeons who did not use personal experience to guide their practice surgeons. In addition, despite reporting their lack of value and use in our qualitative data, surgeons in Cluster 1 and 3 endorsed textbooks as a source of their practice guide more frequently than surgeons in Cluster 2. Surgeons in Cluster 1 and 2 seem to endorse having a mentor or physician/colleague guide their practice more often than surgeons in Cluster 3 [Table 3], which is consistent with the qualitative data, i.e., open commentary from of our respondents who expressed “I learn from mentors or colleagues” and “I have relied on my training and from each patient.”
| Discussion|| |
Surgeons in Cluster 1 (the largest cluster) perform fewer subsequent valve replacement surgeries compared to surgeons in Clusters 2 and 3, although they have been in practice, on average, about the same amount of time. Relevant to the number of valve replacement surgeries is that over 2/3 of the surgeons in Cluster 1 practice in suburban and rural environments where resources, including surgical personnel and administrative support, could be barriers to multiple valve replacement surgeries, as well as increased risks, for a single patient. Surgeons in this cluster seem to utilize policies or establish their own guidelines that limit valve surgeries to 1 or 2, except in rare circumstances (e.g., multiple valves needing replaced in a single surgery due to extensive vegetation).
Surgeons in Cluster 3 (the smallest cluster) performed many more subsequent valve replacement surgeries compared to the surgeons in Clusters 1 and 2 (100% of the Cluster 3 group). A smaller percentage of Cluster 3 surgeons reported addiction is a moral problem, though 100% reported that addiction was a social problem and 85% reported that it was a medical problem. Cluster 3 surgeons also endorsed personal knowledge to guide valve replacement decisions, thus suggesting that when they are doing three or more valve replacement surgeries, they are not heavily relying on textbooks or mentors to make these surgical decisions; 40% look to the literature for guidance, although have expressed “not much data are out there to guide decisions.” Data also suggest that surgeons in Cluster 3 are less likely to support policies that impose on decision-making and of treating each patient as an individual.
More surgeons in Cluster 2-endorsed addiction is a moral problem embedded among social and medical factors, yet more likely to place blame on patients or society, resulting in potentially more stringent rules imposed by the practicing surgeon (e.g., “no more than 2 replacement surgeries per patient”). Experience with subsequent valve replacements might be a contributor to how surgeons conceptualize addiction and how current practices emerged regarding subsequent valve replacement.
The majority of surgeons (71.1%) in our study do not have hospital policies or specific guidelines to guide decisions for recurrent valve replacement surgeries; they greatly rely on personal knowledge and experience, which was also documented in their written responses, particularly when determining whether to perform recurrent valve replacement surgery. Similarly, in the literature, we find that some surgeons create care plan agreements stating that noncompliance and reinfection might result in withholding redo surgeries and are using these care plans for guiding their future surgical decisions, i.e., to not conduct recurrent valve replacement surgeries. The use of patient contracts or care plans was identified among two surgeons in our qualitative data as a way to “fill the gaps of not having policies in place” or as another wrote, “I have done four reoperations on a single patient until we came up with a contract to not operate on recurrent IVDA [abuse] patients.” Our findings further suggest that if surgeons believe recurrent valve replacements for IE-IVDU are not just medical and social issues but also moral issues, they are less likely to do more than 2 valve replacements for a given patient with IE-IVDU. Nevertheless, regardless of age, years of practice, or location of surgical practice (e.g., rural), nearly 60% (59.6%) of our surgeons are willing to give this patient population a second chance and recognize the medical, social, and ethical complexities associated with surgical or medical interventions, addiction, and of the circumstances surrounding each patient.
Our mixed-methods findings suggest that surgeons are more likely to perform recurrent valve replacement surgeries, giving patients a second chance at life, and would value concrete guidance in decision-making when patients are identified on a case-by-case basis, that surgeon autonomy is respected (i.e., not having inflexible or stringent policies that dictate practice, “they cannot be that rigid. Every situation is not the same”), and that other healthcare professionals are involved in the care of this patient population to address addiction and the need for social support toward recovery. There is a sense of recognition of individual patients' needs and the importance of ethical decision-making (e.g., “guidance is needed to do the right thing,” “each patient and [their] need is different”). Besides the need for further research examining this patient population and the effectiveness of the few policies that are being utilized among surgeons, there is a need for comprehensive, integrated care guidelines for IE-IVDU patients with recurrent valve replacements that simultaneously acknowledge provider experience, the clinical environment, and the complex ethical, social, and clinical issues that are presented with each case.
This study attempts to understand the physician-dependent practices in treating IE-IVDU, particularly in cases of recurrent disease. Published recommendations and consensus guidelines are not specific to IE-IVDU. In an effort to generalize the results of this study, participants of this study were recruited from and represent a broad geographic and demographic area to capture the diversity of IE-IVDU and the differences in clinical judgment and treatment in the greater population of cardiothoracic surgeons. However, the study was limited by sample size. The results of this study may not capture the thoughts and opinions of physicians practicing in rural areas and other demographic regions that received lower response rates. It is notoriously difficult to attain high response rates among surgeons.
While surgeons appeared forthcoming in their responses, written views and expressed attitudes may not translate to their actual clinical practice due to social desirability, recall, and response bias. The findings of this study, such as the number of physicians who indicate they are likely to initiate treatment for substance use disorder, would need to be compared to retrospective clinical data to eliminate bias.
Research quality and ethics statement
This study was approved via expedited review by the Institutional Review Board / Ethics Committee at Northeast Ohio Medical University (Approval # 18-011; Approval date July 17, 2018). The authors followed the applicable EQUATOR Network (http://www.equator-network.org/) guidelines, specifically the STROBE Guidelines, during the conduct of this research project.
We would like to acknowledge Emanuela Peshel, MD, who contributed to original survey data collection.
Financial support and sponsorship
This research was supported in part by HCA Healthcare and/or an HCA Healthcare affiliated entity. The views expressed in this publication represent those of the authors and do not necessarily represent the official views of HCA Healthcare or any of its affiliated entities.
Conflicts of interest
There are no conflicts of interest.
| References|| |
Colville T, Sharma V, Albouaini K. Infective endocarditis in intravenous drug users: A review article. Postgrad Med J 2016;92:105-11.
Mylonakis E, Calderwood SB. Infective endocarditis in adults. N Engl J Med 2001;345:1318-30.
Rodger L, Glockler-Lauf SD, Shojaei E, Sherazi A, Hallam B, Koivu S, et al
. Clinical characteristics and factors associated with mortality in first-episode infective endocarditis among persons who inject drugs. JAMA Netw Open 2018;1:e185220.
Njoroge LW, Al-Kindi SG, Koromia GA, ElAmm CA, Oliveira GH. Changes in the association of rising infective endocarditis with mortality in people who inject drugs. JAMA Cardiol 2018;3:779-80.
Kim JB, Ejiofor JI, Yammine M, Ando M, Camuso JM, Youngster I, et al
. Surgical outcomes of infective endocarditis among intravenous drug users. J Thorac Cardiovasc Surg 2016;152:832-410.
Aultman JM, Peshel E, Harfouche C, Firstenberg MS. The ethics in repeat heart valve replacement surgery. In: Firstenberg MS, editor. Advanced Concepts in Endocarditis. London: InTechOpen; 2018. p. 133-147.
Pettersson GB, Coselli JS, Hussain ST, Griffin B, Blackstone EH, Gordon SM, et al
. 2016 The American Association for Thoracic Surgery (AATS) consensus guidelines: surgical treatment of infective endocarditis: Executive summary. J Thorac Cardiovasc Surg 2017; 153:1241-58.
Habib G, Lancellotti P, Antunes MJ, Bongiorni MG, Casalta JP, Del Zotti F, et al
. 2015 ESC Guidelines for the management of infective endocarditis: The Task Force for the Management of Infective Endocarditis of the European Society of Cardiology (ESC). Endorsed by: European Association for Cardio-Thoracic Surgery (EACTS), the European Association of Nuclear Medicine (EANM). Eur Heart J 2015;36:3075-128.
Baddour LM, Wilson WR, Bayer AS, Fowler VG Jr., Tleyjeh IM, Rybak MJ, et al
. Infective endocarditis in adults: diagnosis, antimicrobial therapy, and management of complications: A scientific statement for healthcare professionals from the American Heart Association. Circulation 2105;132:1435-86.
Carozza A, De Santo LS, Romano G, Della Corte A, Ursomando F, Scardone M, et al
. Infective endocarditis in intravenous drug abusers: Patterns of presentation and long-term outcomes of surgical treatment. J Heart Valve Dis 2006;15:125-31.
Rabkin DG, Mokadam NA, Miller DW, Goetz RR, Verrier ED, Aldea GS. Long-term outcome for the surgical treatment of infective endocarditis with a focus on intravenous drug users. Ann Thorac Surg 2012;93:51-7.
Rosenthal ES, Karchmer AW, Theisen-Toupal J, Castillo RA, Rowley CF. Suboptimal addiction interventions for patients hospitalized with injection drug use-associated infective endocarditis. Am J Med 2016;129:481-5.
Nguemeni Tiako MJ, Mori M, Bin Mahmood SU, Shioda K, Mangi A, Yun J, et al
. Recidivism is the leading cause of death among intravenous drug users who underwent cardiac surgery for infective endocarditis. Semin Thorac Cardiovasc Surg 2019;31:40-5.
Huang G, Barnes EW, Peacock JE Jr. Repeat infective endocarditis in persons who inject drugs: “Take Another Little Piece of my Heart”. Open Forum Infect Dis 2018;5:ofy304.
Østerdal OB, Salminen PR, Jordal S, Sjursen H, Wendelbo Ø, Haaverstad R. Cardiac surgery for infective endocarditis in patients with intravenous drug use. Interact Cardiovasc Thorac Surg 2016;22:633-40.
Schranz AJ, Fleischauer A, Chu VH, Wu LT, Rosen DL. Trends in drug use-associated infective endocarditis and heart valve surgery, 2007 to 2017: A study of statewide discharge Data. Ann Intern Med 2019;170:31-40.
RStudio Team. RStudio: Integrated Development for R. Available from: http://www.rstudio.com/
. [Last accessed on 2020 Dec 27].
Scrucca L, Fop M, Murphy TB, Raftery AE. mclust 5: Clustering, classification and density estimation using gaussian finite mixture models. R J 2016;8:289-317.
Struyf A, Hubert M, Rousseeuw P. Clustering in an object-oriented environment. J Stat Softw 1997;1:1-30.
[Table 1], [Table 2], [Table 3]