Sage Journals: Discover world-class research

Abstract

Comorbidity network analysis (CNA) is a technique in which mathematical graphs encode correlations (edges) among diseases (nodes) inferred from the disease co-occurrence data of a patient group. The present study applied this network-based approach to identifying comorbidity patterns in older patients undergoing hip fracture surgery. This was a retrospective observational cohort study using electronic health records (EHR). EHR data were extracted from the one University Health System in the southeast United States. The cohort included patients aged 65 and above who had a first-time low-energy traumatic hip fracture treated surgically between October 1, 2015 and December 31, 2018 (n = 1,171). Comorbidity includes 17 diagnoses classified by the Charlson Comorbidity Index. The CNA investigated the comorbid associations among 17 diagnoses. The association strength was quantified using the observed-to-expected ratio (OER). Several network centrality measures were used to examine the importance of nodes, namely degree, strength, closeness, and betweenness centrality. A cluster detection algorithm was employed to determine specific clusters of comorbidities. Twelve diseases were significantly interconnected in the network (OER > 1, p-value < .05). The most robust associations were between metastatic carcinoma and mild liver disease, myocardial infarction and congestive heart failure, and hemi/paraplegia and cerebrovascular disease (OER > 2.5). Cerebrovascular disease, congestive heart failure, and myocardial infarction were identified as the central diseases that co-occurred with numerous other diseases. Two distinct clusters were noted, and the largest cluster comprised 10 diseases, primarily encompassing cardiometabolic and cognitive disorders. The results highlight specific patient comorbidities that could be used to guide clinical assessment, management, and targeted interventions that improve hip fracture outcomes in this patient group.

Keywords

aged comorbidity hip fractures comorbidity network analysis

Introduction

Hip fracture is an ongoing health issue among older adults. More than 340,000 hospitalizations for hip fractures occur annually among those ages 65 years or older in the United States (Wu et al., 2021). With the population aging, the cases are projected to double from 2.6 million by 2025, further increasing to 4.5 million by 2050, imposing even greater societal burdens (Penrod et al., 2008).

Hip fracture shares risk factors with several chronic conditions, such as chronic pulmonary disease, diabetes mellitus, dementia, and cancer, which explains why patients with hip fractures exhibit a high incidence of comorbidity (Holroyd et al., 2008; Jørgensen & Schwarz, 2008; Kanis et al., 2001; Sennerby et al., 2009). Especially for older patients undergoing hip fracture surgery, comorbidity has a critical role in the course of treatment because this is associated with postoperative complications, higher cost of hospitalization, longer length of hospital stay, decreased quality of life, and lower likelihood of survival (Hershkovitz et al., 2010; Öztürk et al., 2019; Nikkel et al., 2012; Peeters et al., 2016; Sheehan et al., 2018). Thus, examining comorbidity is of vital clinical importance.

Beyond evaluating the presence or severity of comorbidity with simple descriptive statistics or a composite score, identifying the specific comorbidity patterns—that is, understanding the integrated associations among different health conditions—has attracted many researchers’ attention (Fotouhi et al., 2018). Such an understanding could help elucidate high-impact comorbidities or patterns with meaningful clustering of disorders, which could guide specific screening or intervention procedures among inpatients. Furthermore, the comorbidity patterns may reveal clinically relevant differences in response to treatment or prognosis (Brunson et al., 2020).

Using network analysis to assess the associations among diseases is an emerging research strategy. In brief, network analysis is a graph theory-based methodology that can enhance our understanding of the complex relations among the entities investigated, where entities are represented as “nodes” and associations among entities are represented as “edges” between nodes in the network (Borsboom & Cramer, 2013). In recent decades, network analysis has become a powerful method for studying phenomena as diverse as interpersonal interaction, connections among genes, and individual symptom association, or biological mechanisms (Pavlopoulos et al., 2011). It is increasingly being used in clinical science and medical informatics to assess the “diseasome” characterized by comorbid associations among disorders (Brunson et al., 2020).

In particular, comorbidity network analysis (CNA) is an approach that assesses networks aggregated from measures of co-occurrence between pairs of disorders (Hidalgo et al., 2009; Fotouhi et al., 2018). In the comorbidity network, an individual disease is represented by a “node,” and the significant co-occurrence of two diseases is represented by an “edge.” Specifically, CNA can help identify influential disease associations and specific sets of comorbidities for further investigation by examining strongly connected individual nodes (i.e., diseases) and subnetworks of strongly connected nodes (i.e., clusters of comorbidities) in the network model (Brunson & Laubenbacher, 2018). In doing so, this approach may not only confirm existing knowledge on disease comorbidity but also reveal novel clinical comorbidity relationships that other approaches have not reported (Mu et al., 2020). Thus far, CNA has been applied to study comorbidity patterns among inpatient samples, including patients with depression (Qiu et al., 2022), ischemic heart disease (Zhou et al., 2022), and colorectal cancer (Ljubic et al., 2020). Nevertheless, to our knowledge, no studies have comprehensively analyzed the comorbidity patterns among older patients undergoing hip fracture surgery, nor have they employed CNA in such a context.

Objective

Using CNA, the present study aimed to investigate the comorbidity patterns among older patients undergoing hip fracture surgery. Specific aims were to (a) identify the complex associations among different diseases through network-based measures and (b) uncover the specific clusters of comorbidities through the cluster detection method. Particularly, we constructed comorbidity networks based on the standardized International Classification of Diseases (ICD) diagnosis codes included in the Charlson Comorbidity Index (CCI), which takes into consideration major medical comorbidities.

Methods

Design and Setting

This retrospective observational cohort study relied on electronic health records (EHR) data from Duke University Health System. EHR data offer rich and valuable insights into the nature of comorbidities. In particular, standardized ICD codes found in EHR comprise the principal method of recording diseases with the aid of national variants and exhibit significant promise for assessing, at the same time, the whole range of diagnoses, both physical and mental, in the process of evaluation of the patterns of comorbidity. Data were extracted through the Duke Enterprise Data Unified Content Explorer (DEDUCE), a web-based clinical research query tool that enables researchers to access administrative and clinical information such as diagnostic codes generated during patient care (Horvath et al., 2011). DEDUCE employs tools of business intelligence to provide insights into Duke Medicine’s enterprise data warehouse, offering a Guided Query functionality with the aid of a wizard-like interface enabling users to sift through millions of clinical records, delve into aggregate reports, and export extracts (Horvath et al., 2011).

Cohort Selection

The study comprised an initial cohort of older patients aged 65 and above (at the admission of the index hospitalization) who underwent surgery for the first time for a low-energy traumatic fracture of the hip between October 1, 2015 and December 30, 2018 (n = 1,171). We identified the records of patients with hip fractures through the ICD-10-CM diagnosis code of proximal femur fracture S72.0 (femoral neck fracture), S72.1 (pertrochanteric fracture), or S72.2 (subtrochanteric fracture) (see Supplemental Table S1). As ICD-10-CM replaced ICD-9-CM in 2015 and our period of interest relating to index hospitalization was after October 1, 2015, only ICD-10-CM codes were used. Surgery procedures were selected from ICD-10th procedure codes combined with the Current Procedural Terminology codes (see Supplemental Tables S2 and S3). Hip fracture surgery was defined as open reduction internal fixation, hemiarthroplasty, or total hip arthroplasty. The codes were selected based on consultations with two surgeons and an insurance coding expert. Two authors performed the cohort selection independently and discussed and resolved any discrepancies by going through the patients’ charts. Further details of cohort selection in this study are shown in Figure 1 and previously reported (Lee et al., 2022).

Figure 1.

Process of cohort selection method.

Measures

In the current study, comorbidity includes 17 chronic conditions classified by the CCI. It includes myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease/rheumatic disease, peptic ulcer disease, mild liver disease, diabetes without complications, diabetes with complications, hemi/paraplegia, renal disease, cancer, moderate or severe liver disease, metastatic carcinoma, and human immunodeficiency virus/acquired immunodeficiency syndrome. These conditions were identified based on the ICD diagnosis from encounters before and at the index hospitalization considering both ICD-9-CM and ICD-10-CM codes by converting ICD-9-CM to ICD-10-CM (Glasheen et al., 2019) (see Supplemental Table S4 for more details on diagnosis codes).

Statistical Analysis

All analyses were performed with R software 4.2.2 (The R Foundation, Vienna, Austria) and its associated packages. Descriptive and univariate statistics were used for the initial characterization of sample populations.

Comorbidity Identification (Aim 1)

A disease network model represents prevalent diseases as nodes in a graph, with pairwise co-occurrence relationships between diseases as edges. The purpose of the CNA is to assess whether diseases tend to co-occur more often than expected. To quantify the strength of such comorbid relationships, we calculated the observed-to-expected ratio (OER) of all possible disease pairs using the statistical approach described by Fotouhi et al. (2018).

OER_ij, the comorbidity strength of the disease pair i and j, is defined as the ratio of the observed prevalence of disease pair i and j to the expected prevalence that is estimated as the product of the observed disease i prevalence and the observed disease j prevalence under the assumption that the diseases are independent:

{OER}_{i j} = \frac{{\tilde{p}}_{i, j}}{\tilde{p_{i}} {\tilde{p}}_{j}}

OER is closely related to the relative risk in the previous studies of disease networks (Hidalgo et al. 2009; Park et al. 2009; Jeong et al. 2017).

Generally, OER greater than 1 signifies that pairwise diseases co-occurred more often in the same individual than expected, and OER less than 1 means that pairwise diseases tend to be mutually exclusive. Statistical tests were performed to examine the proportional differences between the denominator and the numerator, and p-values were obtained based on continuity correction. Only disease pairs with OERs > 1 and p-value < .05 were included in a network model, the associations were visualized using the R-package visNetwork. The diameter of each node reflects the prevalence of each disease, whereas the thickness of the edges reflects the OER value. The edges have no directionality due to the cross-sectional nature of the study design. For clarity, diseases with a prevalence of less than 1.0% of the total sample were excluded from the analysis (e.g., AIDS/HIV).

Several network centrality measures were used to evaluate the importance of nodes, namely degree, strength, closeness, and betweenness centrality (Hevey, 2018). Degree centrality is the simplest property of the nodes, estimated as the number of direct connections from each node in the network to other nodes. In other words, a disease with a higher degree centrality has more co-occurrence relationships with other diseases. Strength centrality is calculated by the sum of the strengths of the directly connected edges of a node, representing how strongly a disease is connected to other diseases. Closeness centrality gives the average shortest distance of a node to all other nodes, reflecting how fast a node can be approached from other nodes; a disease with a higher closeness centrality has a higher risk of being diagnosed with other diseases in fewer steps. Lastly, betweenness centrality denotes the number of times that the node acts as the bridge along the shortest paths, indicating how important a node is in the average path between other pairs of nodes.

Cluster Detection (Aim 2)

Various clustering algorithms implemented in the R-package igraph enable distinguishing the subsets of nodes in a network model that have a greater density of connections within their subset compared to across subsets. In the present study, following the study of Lee & Park (2021), we used the “infomap” algorithm available in igraph to distinguish those subsets of potential comorbidities that are more closely connected in a graph. The clusters and their respective nodes are illustrated in different colors.

Ethical Considerations

This was a retrospective observational cohort study using existing EHR data, and therefore it was exempted by the Duke university institutional review board (protocol number Pro0010528). The safety and security of all study data were ensured by storing, managing, and analyzing it within the highly protected virtual network space of our institution.

Results

Table 1 provides the sociodemographic and clinical characteristics of patients. The mean age of the patients (n = 1,171) was 81.5 years [standard deviation = 8.73], and the median was 82 years. Females accounted for 71.2% of this total, and the majority were White (82.7%). More than 60.0% of the patients had at least 2 of the 17 diagnoses. Furthermore, 22.8% of the patients had at least 5 of the 17 diagnoses. The most common were cerebrovascular (34.2%), chronic pulmonary (34.0%), and peripheral vascular (30.3%) diseases. More details are shown in Table 2.

Table 1.

Sociodemographic and Clinical Characteristics of Patients (n = 1,171).

Variables	Mean ± standard deviation or n (%)
Age (years, n = 1,171)	81.5 ± 8.73
65–74	283 (24.2%)
75-	888 (75.8%)
Sex (n = 1,171)
Male	337 (28.8%)
Female	834 (71.2%)
Race (n = 1,165)
White	964 (82.7%)
Non-white	201 (17.3%)
Current marital status (n = 1,151)
Married	446 (38.8%)
Not married	705 (61.2%)
Insurance (n = 1,171)
Medicare advantage and others	244 (20.8%)
Medicare	927 (79.2%)
Smoke (n = 1,137)
No	1,048 (92.2%)
Yes	89 (7.8%)
Alcohol (n = 1,114)
No	852 (76.5%)
Yes	262 (23.5%)
Procedure (n = 1,171)
Open reduction and internal fixation	687 (58.7%)
Hemiarthroplasty	431 (36.8%)
Total hip replacement	53 (4.5%)
Admission source (n = 1,171)
Home	868 (74.1%)
Another hospital	171 (14.6%)
Skilled nursing facility	104 (8.9%)
Others	28 (2.4%)
Hospital length of stay (days, n = 1,171)	6.6 ± 4.26

Table 2.

Comorbidities Between 17 Diseases, Ranked by Frequency.

No. of Comorbidities
0		271 (23.1%)
1		141 (12.0%)
2		207 (17.7%)
3		169 (14.4%)
4		116 (9.9%)
≥5		267 (22.8%)
Diseases ranked by frequency
Cerebrovascular disease	400 (34.2%)	Diabetes with complications	129 (11.0%)
Chronic pulmonary disease	398 (34.0%)	Metastatic carcinoma	87 (7.4%)
Peripheral vascular disease	355 (30.3%)	Connective tissue disease/rheumatic disease	86 (7.3%)
Renal disease	321 (27.4%)	Mild liver disease	72 (6.1%)
Congestive heart failure	309 (26.4%)	Peptic ulcer disease	65 (5.6%)
Dementia	269 (23.0%)	Hemi/paraplegia	46 (3.9%)
Cancer	218 (18.6%)	Moderate or severe liver disease	12 (1.02%)
Myocardial infarction	184 (15.7%)	AIDS/HIV^a	2 (0.2%)
Diabetes without complications	182 (15.5%)

Note. AIDS/HIV = acquired immunodeficiency syndrome/human immunodeficiency virus.

Diseases with a prevalence of less than 1.0% of the total sample were excluded from the network analysis.

Twelve diseases were observed via visNetwork within the comorbidity network of the 17 diseases analyzed (OER > 1 with p-values < .05; Figure 2). Table 3 lists the strengths of the disease associations represented by OER values and the corresponding p-values. Specific examples indicate more robust associations than others, such as metastatic carcinoma and mild liver disease, myocardial infarction and congestive heart failure, and hemi/paraplegia and cerebrovascular disease. The probability of co-occurrence for each of these disease pairs was more than 2.5 times higher than expected (OER > 2.5). Table 4 details the results on the node centrality. Cerebrovascular disease and congestive heart failure were the most connected (central) diseases in the network based on the calculated node degree, strength, closeness, and betweenness values. Myocardial infarction also demonstrated higher strength and closeness values than other diseases in the network.

Figure 2.

Comorbidity network among older patients undergoing hip fracture surgery.

Table 3.

Association Strength Between Diseases (largest to smallest).

From	To	OER	p-Value
Metastatic carcinoma	Mild liver disease	3.55	.010
Myocardial infarction	Congestive heart failure	2.55	.000
Hemi/paraplegia	Cerebrovascular disease	2.55	.000
Diabetes with complications	Renal disease	2.46	.002
Myocardial infarction	Diabetes with complications	2.12	.006
Diabetes with complications	Congestive heart failure	2.12	.000
Congestive heart failure	Peptic ulcer disease	2.04	.018
Myocardial infarction	Renal disease	1.94	.000
Congestive heart failure	Peripheral vascular disease	1.88	.000
Renal disease	Congestive heart failure	1.81	.000
Myocardial infarction	Peripheral vascular disease	1.78	.000
Diabetes with complications	Peripheral vascular disease	1.76	.004
Myocardial infarction	Cerebrovascular disease	1.66	.001
Peripheral vascular disease	Cerebrovascular disease	1.62	.000
Congestive heart failure	Chronic pulmonary disease	1.59	.000
Myocardial infarction	Chronic pulmonary disease	1.57	.005
Cerebrovascular disease	Dementia	1.55	.001
Diabetes with complications	Cerebrovascular disease	1.54	.001
Renal disease	Peripheral vascular disease	1.51	.026
Congestive heart failure	Cerebrovascular disease	1.48	.001
Peripheral vascular disease	Chronic pulmonary disease	1.42	.008
Renal disease	Cerebrovascular disease	1.38	.002
Cerebrovascular disease	Chronic pulmonary disease	1.30	.018

Note. OER = observed-to-expected ratio.

Table 4.

Results on the Centrality of Nodes in the Comorbidity Network.

Diseases	Degree	Strength	Closeness	Betweenness
Cerebrovascular disease	8	13.06	0.54	17
Congestive heart failure	7	13.46	0.44	8
Myocardial infarction	6	11.62	0.38	0
Diabetes with complications	5	10.00	0.37	0
Peripheral vascular disease	6	9.96	0.42	0
Renal disease	5	9.10	0.40	0
Chronic pulmonary disease	4	5.87	0.42	0
Mild liver disease	1	3.55	0.28	0
Metastatic carcinoma	1	3.55	0.28	0
Hemi/paraplegia	1	2.55	0.24	0
Peptic ulcer disease	1	2.04	0.24	0
Dementia	1	1.55	0.31	0

Using igraph, the network was separated into two clusters. The largest cluster had 10 nodes (i.e., congestive heart failure, cerebrovascular disease, myocardial infarction, diabetes with complications, peripheral vascular disease, renal disease, chronic pulmonary disease, hemi/paraplegia, peptic ulcer disease, and dementia). Metastatic carcinoma and mild liver disease were also clustered into a group. The clusters and nodes are color-coded in Figure 3.

Figure 3.

Results of a clustering algorithm.

Discussion

This study applied a network-based approach to analyzing and visualizing the comorbidity patterns of older patients undergoing hip fracture surgery. Most patients had more than two comorbidities, confirming observations from previous studies (Edelmuth et al., 2018; Nikkel et al., 2012). The diseases most prevalent among patients were cerebrovascular disease, chronic pulmonary disease, and peripheral vascular disease. Indeed, these diseases are the common risk factors for fractures. Mineralization and vascular calcification share common pathological pathways and medications, including cholesterol-lowering statins and bisphosphonates, which affect atherosclerosis and osteoporotic fractures (Holroyd et al., 2008; Kanis et al., 2001; Sennerby et al., 2009). Furthermore, patients with chronic pulmonary disease have an increased risk of osteoporosis and fractures; smoking, physical inactivity, malnutrition, and several pathophysiological explanations (e.g., a chronic inflammatory state) may be responsible for this association (Jørgensen & Schwarz, 2008). Chronic pulmonary disease has been consistently reported as one of the most prevalent diseases in older patients with hip fractures (Nikkel et al., 2012; Padron-Monedero et al., 2017; Wei et al., 2019).

Notably, cerebrovascular disease, congestive heart failure, and myocardial infarction were identified as central diseases that co-occur and significantly interlink with several diseases in the network. Cerebrovascular disease emerged as the most central disease in the patient group, possibly due to its high prevalence, shared pathogenic processes, and common risk factors with other diseases. Such findings support the necessity of a comprehensive approach to treating patients diagnosed with these central diseases to ensure optimal postoperative care and effectively manage the concurrent health conditions they may have. For optimal recovery and to prevent postoperative complications in patients with hip fractures, clinical guidelines or treatment strategies oriented to individual diseases might be ineffective and incomplete.

Critically, the strongest association involving cerebrovascular disease was with hemi/paraplegia. The literature supports that patients with hip fractures and post-stroke hemiplegia history benefit less from rehabilitation, resulting in increased chronic disability following a hip fracture, increasing the risk of a poor long-term prognosis, and including higher mortality (Nho et al., 2014; Wang et al., 2022). Feng et al. (2009) also highlighted that these patients typically have more severe health conditions, alongside multiple comorbidities, diminished cognitive and walking abilities, and longer hospital stays, all of which emphasize the need for the assessment of these concurrent conditions in orthogeriatric care.

The results revealed two distinct comorbidity clusters of highly interlinked diseases. Notably, patients exhibited co-occurrence with 10 illnesses, including diseases of the cardiovascular/circulatory system (e.g., heart failure), renal disease, pulmonary disease, peptic ulcer disease, other metabolic diseases (e.g., diabetes mellitus), and dementia. The findings highlight the co-occurrence of both physical and cognitive comorbidity among hip fracture patients. In addition, metastatic carcinoma and mild liver disease, each of which is well-known to independently heighten the risk of surgical complications (Forssten et al., 2022; Frost et al., 2011; Kannegaard et al., 2010), were most densely connected out of all disease pairs and also clustered together. Identifying such clusters can help providers take preemptive actions related to the health of a patient.

Importantly, the occurrence of the diseases in a distinct cluster implies that the patient population may be markedly prone to serious complications associated with these conditions. Because a cluster has the maximum possible interactions between nodes, its presence in a patient can be critical due to the high risk (Kalgotra, 2017). Such a cluster indicates a trap state in a patient where escape is challenging (Kalgotra, 2017; Van Borkulo et al., 2015). An important research question is related to the influence of these clusters on health outcomes: “Can we identify a cluster as a trap state, where its presence in a patient increases mortality risk, hospital stay duration, or other critical outcomes?” The influence of multiple concurrent diseases and their underlying interactions can be significantly more intricate and synergistic than the sum of their effects for each disease.

Furthermore, the results based on the clustering algorithm underscore a need for a multidisciplinary orthogeriatric team approach to patient care for patients undergoing hip fracture surgery, which has recently gained greater literature support (Henderson et al., 2017; Pfeufer et al., 2020; Santy-Tomlinson et al., 2021). A multidisciplinary approach to identifying and optimizing comorbidities perioperatively enables better perioperative periods and long-term functional outcomes.

Limitations

This study is subject to several limitations. First, the cohort in the current study was based on a single hospital system, which can result in a lack of generalization of the study findings and selection bias. Thus, data interpretation should be treated with caution. Second, the undirected graph provides a limited insight into the causality of the observed interactions. It should be noted that to use a CNA appropriately to guide intervention strategies and understand disease progression, longitudinal designs in which we can model changes in variables across time are needed. Third, we acknowledge that critical risk factors that influence comorbidities patterns such as age, sex, race, or socioeconomic factors have not adequately been accounted for in the networks, such as performing analyses by stratifying the cohort by these factors. Indeed, prior studies based on CNA have demonstrated differences in the strength of comorbidities in patients of different sexes and races (Qiu et al., 2022; Zhou et al., 2022), possibly reflecting variations in biological processes, environmental factors, or the quality of healthcare provided to each population. Although the overall sample size was not small, the dataset displays an imbalanced distribution of age, sex, race, and socioeconomic factors (e.g., insurance status), and therefore, conducting rigorous stratification analyses was not feasible in the current study. In particular, due to the small sample sizes for some diseases, using these imbalanced stratified samples might lead to the non-detection of multimorbid disease pairs that are disproportionally represented in the dataset, which could result in an underestimation of the complexity of the comorbidity network. With our study as a starting point for the application of the CNA approach to older patients undergoing hip fracture surgery, it is our hope that future studies integrate more samples and investigate how diseases co-occur differently and form distinct network structures across diverse patient populations. Finally, psychiatric illnesses (e.g., depression and anxiety disorders) were not considered in this study. The extension of network models to include these diseases could significantly contribute to deepening our understanding of patient comorbidities. For instance, depressive disorders were reported to be prevalent among older hip fracture surgery patients (Nikkel et al., 2012) and even lead to an increase in postoperative complications after hip fracture surgery (Broggi et al., 2021), but are often overlooked and not incorporated into the clinical care of patients with hip fractures.

Conclusions

Using a network approach and diagnostic codes from EHR data, the complex relationships between chronic conditions were explored in depth, which has important implications for the integrated management of patients undergoing hip fracture surgery. We hope this study can serve as a preliminary guideline for many other nursing studies; thus, we share the R code in Supplemental Table S5.

Supplemental Material

sj-docx-1-cnr-10.1177_10547738231209367 – Supplemental material for Comorbidity Patterns in Older Patients Undergoing Hip Fracture Surgery: A Comorbidity Network Analysis Study

Supplemental material, sj-docx-1-cnr-10.1177_10547738231209367 for Comorbidity Patterns in Older Patients Undergoing Hip Fracture Surgery: A Comorbidity Network Analysis Study by Chiyoung Lee, Sijia Wei, Eleanor S. McConnell, Hideyo Tsumura, Tingzhong (Michelle) Xue and Wei Pan in Clinical Nursing Research

Footnotes

Acknowledgements

None.

Credit Authorship Contribution Statement

Chiyoung Lee: Conceptualization, Data curation, Formal analysis, Investigation, Writing-original draft

Sijia Wei: Conceptualization, Data curation, Investigation, Writing-original draft, Writing-review & editing

Eleanor S. McConnell: Investigation, Supervision, and Writing-review & editing

Hideyo Tsumura: Investigation, Data curation, and Review & Editing

Tingzhong (Michelle) Xue: Investigation, Data curation, and Review & Editing

Wei Pan: Conceptualization, Investigation, Methodology, Supervision, Writing-review & editing

Availability of data,code,and materials

Deidentified participant data is available upon reasonable request from the corresponding author. The R code is shared in the Supplementary file.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by Sigma Theta Tau International’s Psi-at-Large Chapter.

ORCID iDs

Chiyoung Lee

Tingzhong (Michelle) Xue

Supplemental Material

Supplemental material for this article is available online.

Author Biographies

Dr. Chiyoung Lee is an Assistant Professor at the School of Nursing & Health Studies, University of Washington Bothell. She obtained her BSN and MSN from Seoul National University in Seoul, Korea and Ph.D. from Duke University School of Nursing. Dr. Lee has published many papers using various network analysis techniques and has provided extensive consulting on network analysis to the researchers in the nursing field.

Dr. Sijia Wei is a postdoctoral research fellow at Northwestern University Feinberg School of Medicine. She received a BA in Nursing with a concentration in Biomedical Studies from St. Olaf College and Ph.D. from Duke University School of Nursing. Her research focuses on care coordination in care transitions for people with chronic conditions and disabilities using social network analysis and large datasets.

Dr. Eleannor S. McConnell is a tenured Associate Professor of Duke University School of Nursing. Her program of research focuses on factors that influence functional decline in very frail older adults. She has been funded by the National Institute of Nursing Research and the Department of Veterans Affairs to conduct a series of studies designed to identify modifiable risk factors for worsening self-care disability in long-stay nursing home residents with chronic cognitive impairment.

Dr. Hideyo Tsumura is a practicing certified registered nurse anesthetist (CRNA). She received her Ph.D. from Duke University School of Nursing. Her program of research focuses on improving nurse anesthesia care to reduce incidence and severity of postoperative pulmonary complications and to optimize surgical patient outcomes.

Ms. Tingzhong (Michelle) Xue is a graduate student at Duke University School of Nursing. She studied Actuarial Science as an undergraduate at University of Minnesota, Twin Cities. She then found her passion in healthcare and earned her MS degree in Nursing from DePaul University in 2016. Her research interest is in older adults with chronic cognitive impairment.

Dr. Wei Pan is a tenured Associate Professor of Health Measurement, Biostatistics, and Data Science in the Division of Clinical Health Systems and Analytics in the School of Nursing and in the Department of Population Health Sciences in the School of Medicine at Duke University. He received his Ph.D. in Measurement & Quantitative Methods from Michigan State University and M.S. in Statistics from Fuzhou University, China. His research interests are causal inference, advanced statistical modeling, data analytics, meta-analysis, and psychometrics; and their applications in the social, behavioral, and health sciences.

References

Borsboom

Cramer

A. O.

(2013). Network analysis: An integrative approach to the structure of psychopathology. Annual Review of Clinical Psychology, 9, 91–121. https://doi.org/10.1146/annurev-clinpsy-050212-185608

Broggi

M. S.

Oladeji

P. O.

Tahmid

Hernandez-Irizarry

Allen

(2021). Depressive disorders lead to increased complications after geriatric hip fractures. Geriatric Orthopaedic Surgery & Rehabilitation, 12, 21514593211016252. https://doi.org/10.1177/21514593211016252

Brunson

J. C.

Laubenbacher

R. C.

(2018). Applications of network analysis to routinely collected health care data: A systematic review. Journal of the American Medical Informatics Association, 25(2), 210–221. https://doi.org/10.1093/jamia/ocx052

Brunson

J. C.

Agresta

T. P.

Laubenbacher

R. C.

(2020). Sensitivity of comorbidity network analysis. JAMIA Open, 3(1), 94–103. https://doi.org/10.1093/jamiaopen/ooz067

Edelmuth

S. V. C. L.

Sorio

G. N.

Sprovieri

F. A. A.

Gali

J. C.

Peron

S. F.

(2018). Comorbidities, clinical intercurrences, and factors associated with mortality in elderly patients admitted for a hip fracture. Revista brasileira de ortopedia, 53, 543–551.

Feng

Zhang

Shen

Cao

(2009). Predictors of prognosis for elderly patients with poststroke hemiplegia experiencing hip fractures. Clinical Orthopaedics and Related Research®, 467, 2970–2978. https://doi.org/10.1007/s11999-009-0835-5

Forssten

M. P.

Cao

Trivedi

D. J.

Ekestubbe

Borg

Bass

G. A.

Ismail

A. M.

Mohseni

(2022). Developing and validating a scoring system for measuring frailty in patients with hip fracture: A novel model for predicting short-term postoperative mortality. Trauma Surgery & Acute Care Open, 7(1), e000962. https://doi.org/10.1136/tsaco-2022-000962

Fotouhi

Momeni

Riolo

M. A.

Buckeridge

D. L.

(2018). Statistical methods for constructing disease comorbidity networks from longitudinal inpatient data. Applied Network Science, 3, 1–34. https://doi.org/10.1007/s41109-018-0101-4

Frost

S. A.

Nguyen

N. D.

Black

D. A.

Eisman

J. A.

Nguyen

T. V.

(2011). Risk factors for in-hospital post-hip fracture mortality. Bone, 49(3), 553–558. https://doi.org/10.1016/j.bone.2011.06.002

10.

Glasheen

W. P.

Cordier

Gumpina

Haugh

Davis

Renda

(2019). Charlson Comorbidity Index: ICD-9 update and ICD-10 translation. American Health & Drug Benefits, 12(4), 188–197.

11.

Henderson

C. Y.

Shanahan

Butler

Lenehan

O’Connor

Lyons

Ryan

J. P.

(2017). Dedicated orthogeriatric service reduces hip fracture mortality. Irish Journal of Medical Science (1971-), 186, 179–184. https://doi.org/10.1007/s11845-016-1453-3

12.

Hershkovitz

Polatov

Beloosesky

Brill

(2010). Factors affecting mortality of frail hip-fractured elderly patients. Archives of Gerontology and Geriatrics, 51(2), 113–116. https://doi.org/10.1016/j.archger.2009.09.003

13.

Hevey

(2018). Network analysis: A brief overview and tutorial. Health Psychology and Behavioral Medicine, 6(1), 301–328. https://doi.org/10.1080/21642850.2018.1521283

14.

Hidalgo

C. A.

Blumm

Barabási

A. L.

Christakis

N. A.

(2009). A dynamic network approach for the study of human phenotypes. PLoS Computational Biology, 5(4), e1000353. https://doi.org/10.1371/journal.pcbi.1000353

15.

Holroyd

Cooper

Dennison

(2008). Epidemiology of osteoporosis. Best Practice & Research Clinical Endocrinology & Metabolism, 22(5), 671–685. https://doi.org/10.1016/j.beem.2008.06.001

16.

Horvath

M. M.

Winfield

Evans

Slopek

Shang

Ferranti

(2011). The DEDUCE Guided Query tool: providing simplified access to clinical data for research and quality improvement. Journal of Biomedical Informatics, 44(2), 266–276.

17.

Jeong

Han

H. W.

(2017). Network-based analysis of diagnosis progression patterns using claims data. Scientific Reports, 7(1), 15561. https://doi.org/10.1038/s41598-017-15647-4

18.

Jørgensen

N. R.

Schwarz

(2008). Osteoporosis in chronic obstructive pulmonary disease patients. Current Opinion in Pulmonary Medicine, 14(2), 122–127. https://doi.org/10.1097/MCP.0b013e3282f4efb6

19.

Kalgotra

(2017). Impact of inferred comorbidity networks on health outcomes [Doctoral dissertation, Oklahoma State University].

20.

Kanis

Oden

Johnell

(2001). Acute and long-term increase in fracture risk after hospitalization for stroke. Stroke, 32(3), 702–706. https://doi.org/10.1161/01.STR.32.3.702

21.

Kannegaard

P. N.

Van Der Mark

Eiken

Abrahamsen

B. O.

(2010). Excess mortality in men compared with women following a hip fracture. National analysis of comedications, comorbidity and survival. Age and Ageing, 39(2), 203–209. https://doi.org/10.1093/ageing/afp221

22.

Lee

H. A.

Park

(2021). Comorbidity network analysis related to obesity in middle-aged and older adults: Findings from Korean population-based survey data. Epidemiology and Health, 43, e2021018. https://doi.org/10.4178/epih.e2021018

23.

Lee

McConnell

E. S.

Wei

Xue

Tsumura

Pan

(2022). Effect of race/ethnicity, insurance status, and area deprivation on hip fracture outcomes among older adults in the United States. Clinical Nursing Research, 31(3), 541–552.

24.

Ljubic

Pavlovski

Alshehri

Roychoudhury

Bajic

Van Neste

Obradovic

(2020). Comorbidity network analysis and genetics of colorectal cancer. Informatics in Medicine Unlocked, 21, 100492. https://doi.org/10.1016/j.imu.2020.100492

25.

X. M.

Wang

Jiang

Y. Y.

Feng

(2020). Patterns of comorbidity in hepatocellular carcinoma: A network perspective. International Journal of Environmental Research and Public Health, 17(9), 3108. https://doi.org/10.3390/ijerph17093108

26.

Nikkel

L. E.

Fox

E. J.

Black

K. P.

Davis

Andersen

Hollenbeak

C. S.

(2012). Impact of comorbidities on hospitalization costs following hip fracture. Jbjs, 94(1), 9–17.

27.

Nho

J. H.

Lee

Y. K.

Kim

Y. S.

Y. C.

Suh

Y. S.

Koo

K. H.

(2014). Mobility and one-year mortality of stroke patients after hip-fracture surgery. Journal of Orthopaedic Science, 19, 756–761. https://doi.org/10.1007/s00776-014-0593-4

28.

Öztürk

Johnsen

S. P.

Röck

N. D.

Pedersen

A. B.

(2019). Impact of comorbidity on the association between surgery delay and mortality in hip fracture patients: A Danish nationwide cohort study. Injury, 50(2), 424–431. https://doi.org/10.1016/j.injury.2018.12.032

29.

Padron-Monedero

López-Cuadrado

Galán

Martinez-Sanchez

E. V.

Martin

Fernandez-Cuenca

(2017). Effect of comorbidities on the association between age and hospital mortality after fall-related hip fracture in elderly patients. Osteoporosis International, 28, 1559–1568. https://doi.org/10.1007/s00198-017-3926-2

30.

Park

Lee

D. S.

Christakis

N. A.

Barabási

A. L.

(2009). The impact of cellular networks on disease comorbidity. Molecular Systems Biology, 5(1), 262. https://doi.org/10.1038/msb.2009.16

31.

Pavlopoulos

G. A.

Secrier

Moschopoulos

C. N.

Soldatos

T. G.

Kossida

Aerts

Schneider

Bagos

P. G.

(2011). Using graph theory to analyze biological networks. BioData Mining, 4, 1–27.

32.

Peeters

C. M.

Visser

Van de Ree

C. L.

Gosens

Den Oudsten

B. L.

De Vries

(2016). Quality of life after hip fracture in the elderly: A systematic literature review. Injury, 47(7), 1369–1382. https://doi.org/10.1016/j.injury.2016.04.018

33.

Penrod

Smith

Terwilliger

Gueldner

. (2008). Demographic perspective: The magnitude of concern, In Gueldner

Grabo

Newman

Cooper

(Eds.), Osteoporosis: Clinical guidelines for prevention, diagnosis and management (pp. 9–18), Springer Publishing Company LLC.

34.

Pfeufer

Kammerlander

Stadler

Roth

Blauth

Neuerburg

Böcker

Zeckey

Lechleitner

Gosch

(2020). Multidisciplinary inpatient rehabilitation improves the long-term functional status of geriatric hip-fracture patients. European Journal of Medical Research, 25(1), 1–8.

35.

Qiu

Wang

Zeng

Pan

(2022). Comorbidity patterns in depression: A disease network analysis using regional hospital discharge records. Journal of Affective Disorders, 296, 418–427. https://doi.org/10.1186/s40001-020-00433-2

36.

Santy-Tomlinson

Hertz

Myhre-Jensen

Brent

(2021). Nursing in the Orthogeriatric setting. Orthogeriatrics: The Management of Older Patients with Fragility Fractures, 293–310.

37.

Sennerby

Melhus

Gedeborg

Byberg

Garmo

Ahlbom

Pedersen

N. L.

Michaëlsson

(2009). Cardiovascular diseases and risk of hip fracture. JAMA, 302(15), 1666–1673. https://doi.org/10.1001/jama.2009.1463

38.

Sheehan

K. J.

Williamson

Alexander

Filliter

Sobolev

Guy

Bearne

L. M.

Sackley

(2018). Prognostic factors of functional outcome after hip fracture surgery: A systematic review. Age and Ageing, 47(5), 661–670. https://doi.org/10.1093/ageing/afy057

39.

Van Borkulo

Boschloo

Borsboom

Penninx

B. W.

Waldorp

L. J.

Schoevers

R. A

. (2015). Association of symptom network structure with the course of depression. JAMA Psychiatry, 72, 1219–1226. https://doi.org/10.1001/jamapsychiatry.2015.2079

40.

Wang

Luo

Wang

Zhu

(2022). Analysis of the surgical outcomes in elderly patients with hip fractures combined with hemiplegia. Clinical Interventions in Aging, 17, 1093–1098.

41.

Wei

Zeng

Luo

Xiang

Ding

(2019). Relationship between comorbidities and treatment decision-making in elderly hip fracture patients. Aging Clinical and Experimental Research, 31, 1735–1741. https://doi.org/10.1007/s40520-019-01134-5

42.

Tong

Wang

Sun

(2021). Worldwide research tendency and hotspots on hip fracture: A 20-year bibliometric analysis. Archives of Osteoporosis, 16, 1–14. https://doi.org/10.1007/s11657-021-00929-2

43.

Zhou

Wang

Ding

Shen

Qiu

(2022). Phenotypic disease network analysis to identify comorbidity patterns in hospitalized patients with ischemic heart disease using large-scale administrative data. Healthcare, 10, 80. https://doi.org/10.3390/healthcare10010080

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB