Skip to main content
NCBI home page
Journal of Clinical Orthopaedics and Trauma logoLink to Journal of Clinical Orthopaedics and Trauma
. 2020 Dec 3;16:27–34. doi: 10.1016/j.jcot.2020.12.001

Differences in the Predictive value of Elixhauser Comorbidity Index and the Charlson Comorbidity indices in patients with hand infections

Dominick V Congiusta a,∗, Kamil M Amer a, Katie Otero a, Michael Metrione a, Aziz M Merchant b, Michael Vosbikian a, Ifran Ahmed a
PMCID: PMC7919929  PMID: 33680828

Abstract

Purpose

Hand infections are a common source of potentially debilitating morbidity, particularly in patients with comorbid disease. We hypothesize that there is a difference in predictive value between two commonly used comorbidity indices for the prognosis of hand infections, which may have clinical implications in the management of these conditions.

Methods

The Nationwide Inpatient Sample 2001–2013 database was queried for hand infections using International Classification of Diseases, Ninth Revision codes. The Elixhauser (ECI) and Charlson (CCI) comorbidity scores were calculated based on validated sets of ICD-9 codes. Primary outcomes included mortality, prolonged length of stay (LOS, defined as >95 percentile), discharge destination, and postoperative complications. Indices were compared using receiver operating characteristic (ROC) curves and the areas under the curve (AUC). If confidence intervals overlapped, significance was determined using the DeLong method for correlated ROC curves. This is a validated, non-parametric comparison used for the calculation of the difference between two AUCs.

Results

A weighted total of 1,511,057 patients were included in this study. The majority were Caucasian (57.1%) males (61.4%). Complication rates included 0.9% mortality, 5.3% prolonged length of stay, 25.3% discharges to non-home destinations, and 5.3% post-operative complications. The ECI and CCI each demonstrated good predictive value for mortality, but poor predictive value for non-routine discharge, prolonged LOS, and post-operative complications. There was a significantly increased likelihood of each complication with increasing comorbidity score for both indices, with the greatest odds ratio in the ECI ≥4 cohort.

Conclusions

The CCI was superior in predicting mortality while the ECI was superior in predicting non-routine discharge, prolonged length of stay, and postoperative complications, but these indices may not be clinically relevant. While both represent good predictive models, a score specifically designed for patients with hand infections may have superior prognostic value.

Level of evidence

Level IV.

Keywords: Elixhauser comorbidity index, Charlson comorbidity index, Hand infections, National inpatient sample

1. Introduction

The surgical treatment of hand infections is complex and challenging due to the intricate anatomy involved, the low tolerance for adhesions, and the potential for long-term morbidity. They serve as a common etiology for emergency department admissions1 and tend to occur in younger males (median age 36–40), with as much as 89% male predominance.1, 2, 3 Because of the impact on patient lifestyle, executive function, and survival, appropriate treatment must be given in a timely manner.4 Delays and inadequate treatment can result in complication rates ranging from 6% to 70%.5,6 A combination of pathogen virulence, location of the infection, and host defense mechanisms, which are influenced by patient comorbid conditions, influence treatment outcomes.7 Studies have shown that complications in these populations are indeed linked with concurrent hypertension, diabetes, alcohol use, hepatitis, HIV, smoking, cancer, corticosteroid use, and mental illness.3,8, 9, 10, 11, 12, 13, 14, 15

Physicians account for increased risk of these conditions through the use of risk-assessment tools, such as comorbidity scales. A feared consequence of inappropriate risk-adjustment application is financial and administrative penalization16, 17, 18 and underscores the importance of appropriate risk assessment tools. Several models, including the Elixhauser (ECI) and Charlson (CCI) Comorbidity Indices, provide physicians a rapid way to measure risk based off of patient history. These are two of the most widely used risk assessment tools with large, administrative data19,20 and have recently been compared in a variety of subspecialties, including orthopedics.20,21

The Charlson measure was initially developed as a weighted index to be used for predicting mortality using 19 comorbidities,22 and has since been adapted for use with International Classification of Diseases, 9th revision (ICD-9)23 and 10th revision (ICD-10)24 codes in larger populations. By comparison, the Elixhauser measure incorporates 31 comorbidities25 and has been shown to have superior predictive value in some disease states and surgeries.20,26, 27, 28 In attempt to improve accuracy, remove discrepancies, and maintain consistency between the previously established list of ICD codes, enhanced algorithms have been developed that produce similar or superior predictive value compared to their predecessors.29 The objectives of this study was to compare the predictive value of the enhanced ECI and CCI in the treatment of hand infections and to determine the best available index for appropriate risk stratification in this population. We hypothesize that both measures will demonstrate, at minimum, good predictive value for complication rate. Our null hypothesis states that there will be no predictive value of either index.

2. Methods

2.1. Data source and patient population

The Nationwide Inpatient Sample (NIS) is a national database maintained by the Healthcare Cost and Utilization Project that approximates a 20% sample of all discharges from U.S. community hospitals.30 It reports up to 15–25 possible diagnoses per patient and includes demographic, socioeconomic, and hospital administrative data. The NIS 2001–2013 dataset was queried for hand infections using ICD-9 codes for hand infections (Table 1). This particular list has been used in other studies to identify relevant cases.31,32 Data were weighted using HCUP provided trend weights, which allowed the generation of national estimates.33

Table 1.

ICD-9a codes for hand infections.

Description Code
Cellulitis and abscess of finger, unspecified 681.00
Felon 681.01
Onychia and paronychia of finger 681.02
Cellulitis and abscess of hand, except fingers and thumb 682.4
Other tenosynovitis of hand and wrist 727.05
Unspecified disorder of synovium, tendon, and bursa 727.9
Open wound of finger(s), without mention of complication 883.0
Open wound of finger(s), complicated 883.1
Open wound of hand except finger(s) alone, complicated 882.1
Open wound of hand except finger(s) alone, without mention of complication 882.0
Open in a new tab
a

International Classification of Diseases, 9th Revision.

The ECI and CCI scores were determined based on defined algorithms of ICD-9 codes that have been previously validated (Table 3, Table 4).29 Dichotomous variables based on these codes in any of the 15–25 diagnosis categories were created (see Table 2). This ensured that we did not erroneously exclude patients who have relevant comorbidities listed in categories other than the “principle diagnosis.” The final scores were calculated by summing the weights of all variables for each patient (Table 5). As our study does not involve human subjects, Institutional Review Board (IRB) approval was not needed, as per our institution’s policy.

Table 3.

Enhanced elixhauser comorbidity index coding algorithm.

Comorbidities ICD-9 CM Codes Weight
Congestive Heart Failure 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, 425.4–425.9, 428.x 1
Cardiac arrhythmias 426.0, 426.13, 426.7, 426.9, 426.10, 426.12, 427.0–427.4, 427.6–427.9, 785.0, 996.01, 996.04, V45.0, V53.3 1
Valvular disease 093.2, 394.x–397.x, 424.x, 746.3–746.6, V42.2, V43.3 1
Pulmonary circulation disorders 415.0, 415.1, 416.x, 417.0, 417.8, 417.9 1
Peripheral vascular disorders 093.0, 437.3, 440.x, 441.x, 443.1–443.9, 447.1, 557.1, 557.9, V43.4 1
Hypertension, uncomplicated 401.x 1
Hypertension, complicated 402.x–405.x 1
Paralysis 334.1, 342.x, 343.x, 344.0–344.6, 344.9 1
Other neurological disorders 331.9, 332.0, 332.1, 333.4, 333.5, 333.92, 334.x–335.x, 336.2, 340.x, 341.x, 345.x, 348.1, 348.3, 780.3, 784.3 1
Chronic pulmonary disease 416.8, 416.9, 490.x −505.x, 506.4, 508.1, 508.8 1
Diabetes, uncomplicated 250.0–250.3 1
Diabetes, complicated 250.4–250.9 1
Hypothyroidism 240.9, 243.x, 244.x, 246.1, 246.8 1
Renal failure 403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 585.x, 586.x, 588.0, V42.0, V45.1, V56.x 1
Liver disease 070.22, 070.23, 070.32, 070.33, 070.44, 070.54, 070.6, 070.9, 456.0–456.2, 570.x, 571.x, 572.2–572.8, 573.3, 573.4, 573.8, 573.9, V42.7 1
Peptic ulcer disease excluding bleeding 531.7, 531.9, 532.7, 532.9, 533.7, 533.9, 534.7, 534.9 1
AIDS/HIV 042.x–044.x 1
Lymphoma 200.x–202.x, 203.0, 238.6 1
Metastatic cancer 196.x–199.x 1
Solid tumor without metastasis 140.x–172.x, 174.x– 195.x 1
Rheumatoid arthritis/collagen vascular diseases 446.x, 701.0, 710.0–710.4, 710.8, 710.9, 711.2, 714.x, 719.3, 720.x, 725.x, 728.5, 728.89, 729.30 1
Coagulopathy 286.x, 287.1, 287.3–287.5 1
Obesity 278 1
Weight loss 260.x–263.x, 783.2, 799.4 1
Fluid and electrolyte disorders 253.6, 276.x 1
Blood loss anemia 280 1
Deficiency anemia 280.1–280.9, 281.x 1
Alcohol abuse 265.2, 291.1–291.3, 291.5–291.9, 303.0, 303.9, 305.0, 357.5, 425.5, 535.3, 571.0–571.3, 980.x, V11.3 1
Drug abuse 292.x, 304.x, 305.2–305.9, V65.42 1
Psychoses 293.8, 295.x, 296.04, 296.14, 296.44, 296.54, 297.x, 298.x 1
Depression 296.2, 296.3, 296.5, 300.4, 309.x, 311 1
Open in a new tab

Table 4.

Descriptive statistics.

 Frequency (%)
Total 1,511,057
Demographics
 Age ≥ 60 483,818 (32.0)
 Female 582,578 (38.6)
Race
 Caucasian 862,688 (57.1)
 African American 169,080 (11.2)
 Hispanic 147,611 (9.8)
 Other 66,443 (4.4)
Complications
 Mortality 12,332 (0.9)
 Non-routine Discharge 382,825 (25.3)
 Prolonged Length of Stay 80,813 (5.3)
 Postoperative Complications 80,167 (5.3)
Open in a new tab

Table 2.

Enhanced charlson comorbidity index coding algorithm.

Comorbidities ICD-9 CM Codes Weight
Myocardial Infarction 410.x, 412.x 1
Congestive Heart failure 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, 425.4–425.9, 428.x 1
PVD 093.0, 437.3, 440.x, 441.x, 443.1–443.9, 47.1, 557.1, 557.9, V43.4 1
Cerebrovascular Disease 362.34, 430.x–438.x 1
Dementia 290.x, 294.1, 331.2 1
Chronic Pulmonary Disease 416.8, 416.9, 490.x–505.x, 506.4, 508.1, 508.8 1
Rheumatic Disease 446.5, 710.0–710.4, 714.0–714.2, 714.8, 725.x 1
Peptic Ulcer Disease 531.x–534.x 1
Mild Liver Disease 070.22, 070.23, 070.32, 070.33, 070.44, 070.54, 070.6, 070.9, 570.x, 571.x, 573.3, 573.4, 573.8, 573.9, V42.7 1
Diabetes without chronic complication 250.0–250.3, 250.8, 250.9 1
Diabetes with chronic complication 250.4–250.7 2
Hemiplegia or paraplegia 334.1, 342.x, 343.x, 344.0–344.6, 344.9 2
Renal disease 403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 582.x, 583.0–583.7, 585.x, 586.x, 588.0, V42.0, V45.1, V56.x 2
Any malignancy, including lymphoma and leukemia, except malignant neoplasm of skin 140.x–172.x, 174.x–195.8, 200.x–208.x, 238.6 2
Moderate or severe liver disease 456.0–456.2, 572.2–572.8 3
Metastatic solid tumor 196.x–199.x 6
AIDS/HIV 042.x–044.x 6
Open in a new tab

Table 5.

Frequency of comorbidity index scores.

Charlson Comorbidity Index Elixhauser Comorbidity Index
Score Frequency (%) Score Frequency (%)
0 303,805 (20.1%) 0 367,354 (24.3%)
1 138,895 (9.2%) 1 269,852 (17.9%)
2 68,810 (4.6%) 2 163,319 (10.8%)
3 39,181 (2.6%) 3 87,685 (5.8%)
4 19,704 (1.3%) 4 38,870 (2.6%)
5 15,196 (1%) 5 14,667 (1.0%)
6 5942 (0.4%) 6 4332 (0.3%)
7 5879 (0.4%) 7+ 1342 (0.1%)
9+ 2908 (0.2%)
Open in a new tab

2.2. Outcomes

The primary outcome of interest was mortality. Secondary outcomes included prolonged length of stay, discharge destination, and postoperative complications. Prolonged length of stay was defined as >95th percentile, which was 14 days in our population. Postoperative complications were defined by ICD-9 diagnosis codes (Appendix). Non-routine discharge destination was defined according to the NIS variable DISPUNIFORM, which specifies routine discharge. Other types of discharge, excluding “mortality,” were included in the non-routine discharge variable (i.e., to a short-term hospital, skilled nursing facility, intermediate care, and another type of facility, home health care services, against medical advice, or alive with unknown destination).

2.3. Statistical analysis

Univariate analysis with Pearson’s chi square was conducted to determine whether the ECI and CCI were associated with each outcome of interest. For each outcome (dependent) variable, multivariate logistic regression models were conducted to confirm the relationship between each index and outcomes. Results were reported as odds ratio (OR) with 95% confidence interval (95% CI).

Discriminative ability of the indices was compared using receiver operating characteristic (ROC) curves. In each model, the area under the curve (AUC) and 95% CI was used to assess predictive value. Statistical significance between the AUCs was determined using the DeLong method for correlated ROC curves.34 This is a previously validated, non-parametric comparison used for the calculation of the difference between two AUCs. It is a helpful method of evaluating the performance of diagnostic tests or indices by calculating a significance metric for each AUC (i.e., each index measured).34 Significance was defined as p < 0.01, which would indicate a statistically significant difference in the predictive capabilities of each index. Data were analyzed using Statistical Package for the Social Science (SPSS) (International Business Machines, Corp., Armonk, NY).

3. Results

3.1. Demographics

A weighted total of 1,511,057 patients were included in this study. The majority were Caucasian (57.1%) and male (61.1%). Mean age was 49.0 years old (standard deviation 22.4 years). Of note, 65.0% of patients who had CCI ≥3 were aged 60 or older, while 66.4% of patients with an ECI ≥4 were aged 60 or older (p < 0.01). The most common complication in our cohort was non-routine discharge (25.3%).

3.2. Chi square and logistic regression

Complication rates included 0.9% mortality (n = 12,332), 5.3% prolonged lengths of stay (n = 80,813), 25.3% non-routine discharges (n = 382,825), and 5.3% postoperative complications (n = 80,167) (Table 4). Chi square analysis revealed significant associations with both the CCI and ECI for all outcomes (p < 0.01) (Table 6). Both the CCI and ECI were independently predictive of outcomes after accounting for demographic factors in multivariate analysis (p < 0.01) (Table 7). As expected, the odds of suffering each complication increased with increasing comorbidity score. The greatest predictor of each complication was the highest comorbidity score recorded (≥3 for CCI, ≥4 for ECI), with the greatest odds found with an ECI ≥4 for prolonged length of stay (OR 5.97, p < 0.01).

Table 6.

Bivariate analysis of complications and comorbidity indices.

Number of Comorbidities Mortality (%) P value Non-routine Discharge (%) P value Prolonged Length of Stay (%) P value Postoperative Complications (%) P value
Elixhauser Score
0 (Ref∗) <0.01 <0.01 <0.01 <0.01
1 2131 (0.6) 87,260 (23.9) 16,172 (4.4) 17,289 (4.7)
2 2357 (0.9) 85,527 (32.0) 18,033 (6.7) 16,485 (6.1)
3 2553 (1.6) 66,149 (41.2) 15,382 (9.4) 12,892 (7.9)
4+ 3685 (2.5) 73,507 (51.4) 18,609 (12.7) 14,633 (1.0)
Charlson Score
0 (Ref∗) <0.01 <0.01 <0.01 <0.01
1 2537 (0.008) 91,777 (30.5) 16,684 (5.5) 18,587 (6.1)
2 2283 (0.016) 57,958 (42.5) 14,180 (10.2) 11,576 (8.3)
3+ 4528 (0.028) 77,270 (49.9) 21,184 (13.3) 15,080 (9.4)
Open in a new tab

Table 7.

Binary logistic regression for complications, ORa (95% CIb).

 Mortality
 Non-routine Discharge
 Prolonged LOSc
 Postoperative Complications
CCId ECIe CCId ECIe CCId ECIe CCId ECIe
Age ≥ 60 3.51 (3.35–3.68)f 3.87 (3.69–4.06)f 2.9 (2.87–2.92)f 2.69 (2.66–2.71)f 1.21 (1.19–1.23)f 1.18 (1.16–1.2)f 1.35 (1.32–1.37)f 1.32 (1.29–1.34)f
Female 0.79 (0.76–0.82)f 0.75 (0.72–0.78)f 1.15 (1.14–1.16)f 1.11 (1.1–1.12)f 1 (0.99–1.02) 0.96 (0.94–0.97)f 1.02 (1.01–1.04) 1 (0.99–1.02)
Race
Caucasian Ref Ref Ref Ref Ref Ref Ref Ref
African American 0.98 (0.92–1.05) 1.04 (0.98–1.11) 0.96 (0.95–0.98)f 0.97 (0.96–0.98)f 1.31 (1.28–1.34)f 1.34 (1.31–1.37)f 0.84 (0.82–0.86)f 0.85 (0.83–0.87)f
Hispanic 0.95 (0.89–1.02) 1.02 (0.95–1.09) 0.77 (0.76–0.79)f 0.82 (0.81–0.83)f 1.27 (1.24–1.3)f 1.37 (1.34–1.41)f 0.86 (0.83–0.88)f 0.89 (0.86–0.91)f
Other 1.12 (1.03–1.22) 1.19 (1.1–1.3)f 0.81 (0.8–0.83)f 0.85 (0.84–0.87)f 1.4 (1.36–1.45)f 1.49 (1.44–1.54)f 1.01 (0.97–1.04) 1.04 (1–1.07)
Comorbidity Scores
1 1.83 (1.72–1.94)f 1.36 (1.26–1.46)f 1.57 (1.56–1.59)f 1.74 (1.71–1.76)f 1.74 (1.7–1.78)f 1.98 (1.93–2.03)f 1.49 (1.46–1.52)f 1.31 (1.28–1.34)f
2 2.97 (2.78–3.16)f 1.8 (1.67–1.94)f 2.3 (2.26–2.33)f 2.25 (2.22–2.28)f 3.27 (3.19–3.34)f 3.05 (2.97–3.13)f 1.99 (1.94–2.04)f 1.66 (1.62–1.7)f
3 (ECId), ≥3 (CCIe) 5.09 (4.82–5.38)f 3.09 (2.88–3.32)f 3.04 (3–3.08)f 3.15 (3.1–3.19)f 4.3 (4.21–4.39)f 4.37 (4.25–4.49)f 2.28 (2.23–2.34)f 2.18 (2.13–2.24)f
≥ 4 (ECI) – 4.25 (3.96–4.56)f – 4.37 (4.31–4.44)f – 5.97 (5.81–6.13)f – 2.73 (2.66–2.8)f
Open in a new tab
a

Odds Ratio.

b

Confidence Interval.

c

Length of Stay.

d

Charlson Comorbidity Index Score.

e

Elixhauser Comorbidity Index Score.

f

Significance defined as p < 0.01.

3.3. Receiver Operating Curve, AUC

The AUC (95% CI) for the ECI was 0.708 (0.698–0.718) for mortality, 0.683 (0.681–0.695) for non-routine discharge, 0.678 (0.674–0.682) for prolonged length of stay, and 0.612 (0.608–0.617) for postoperative complications. These data demonstrate that the ECI model was a good fit for predicting mortality, while it was a poor fit for predicting secondary outcomes. The AUC (95% CI) for the CCI was 0.725 (0.714–0.735) for mortality, 0.651 (0.649–0.654) for non-routine discharge, 0.660 (0.656–0.665)for prolonged length of stay, and 0.600 (0.596–0.605) for postoperative complications. Similar to the ECI, the CCI model was a good fit for predicting mortality and a poor fit for predicting secondary outcomes. The differences between AUCs were significant for each outcome (Table 8) (p < 0.01). The AUCs for each outcome are shown in Fig. 1.

Table 8.

Areas Under Curve (95% Confidence Interval) for Outcomes after Applying Comorbidity Indices, p < 0.01

Death Non-routine Discharge Prolonged Length of Stay (>95 Percentile) Post-operative Complications
ECIa 0.708 (0.698-0.718) 0.683 (0.681–0.695)c 0.678 (0.674–0.682)c 0.612 (0.608–0.617)c
CCIb 0.725 (0.714–0.735)c 0.651 (0.649-0.654) 0.660 (0.656-0.665) 0.600 (0.596-0.605)
Open in a new tab
a

Elixhauser Comorbidity Index.

b

Charlson Comorbidity Index.

c

Significantly greater area under curve.

Fig. 1.

Fig. 1

Open in a new tab

Receiver Operating Curve for Complications by Comorbidity Indices, p < 0.05.

4. Discussion

Patients presenting with hand infections often have comorbidities that affect rates of healing, pathogen clearance, the need for surgery, and survival. Inpatient treatment may involve adjustment of management depending on these factors, as understanding the difference in risk between patients is necessary for optimizing care. Due to the absence of a validated risk assessment index for adult hand infections, our objective was to compare predictive value of complications between two commonly used comorbidity indices.

It is well known that comorbid disease complicates the treatment of infectious etiologies. Vasculopathies, for example, are known to impede wound healing and prolong clearance of infections35, 36, 37, 38 and chronic disease, such as chronic pulmonary, hepatic, or renal disease, may tax organ systems and result in impaired immune and healing responses.39, 40, 41 Validated comorbidity scores offer the means to estimate cumulative disease burden on patients, and often show greater predictive value compared to individual comorbidities. Our data shows that after controlling for age, gender, and race, the likelihood of suffering each complication, including mortality, increased significantly with each increase in comorbidity score. Those with an ECI score of ≥4, for example, had 5.97 times increased likelihood of having a prolonged length of stay (greater than 2 weeks) in our cohort by multivariable analysis.

We also demonstrate that predictive values for both the ECI and CCI are not uniform across all studied complications, which may suggest that appropriate use of these measures may involve both, rather than one over the other, depending on outcome of interest. The magnitudes of each AUC we obtained from our primary analysis, however, show that the indices performed only at an “acceptable” level. In general, AUC values less than 0.70 are considered “poor” models of predictability. Values between 0.70 and 0.80 are regarded as “acceptable,” and higher than 0.80 are considered “excellent.“42,43 The only measures above 0.70 that we report were those for mortality (ECI = 0.708 and CCI = 0.725, p < 0.01). The remainder of measures were between 0.600 and 0.699, indicating poor discriminative ability. This is unsurprising as both indices were not developed specifically for patients with hand infections and may highlight a need for development of a more pathology-specific index. As all AUCs are greater than 0.50, however, the null hypothesis suggesting no discriminative power is still rejected.

Regarding mortality, the CCI demonstrated superior discriminative ability compared to the ECI. Some studies report similar outcomes in a variety of specialties44,45 while other studies report the opposite.20,28,46,47 Due to this lack of consensus, a single, unifying opinion about superiority of one index over another seems to be impractical. Instead, it seems that the indices may have to be applied in specific situations, limiting their generalizability. As suggested by Menendez et al. there may be some unmeasured comorbidities that are yet unaccounted for in orthopedic inpatient morbidity.20 Another possible explanation is that outcomes are particularly difficult to accurately record as they are subject to clinical interpretation or overlap. Included in the definition of “postoperative complications,” for example, were ICD-9 codes 998.8–998.9, defined as “other complications of procedure (subcutaneous emphysema, non-healing wound, complications not elsewhere classified)” (Appendix). Clinicians recording data with this ICD-9 code may not record ICD-9 code 998.3, defined as “wound disruption.” In this example, wound complications may not be recorded accurately due to the limitations of the ICD-9 coding system. While we would have chosen outcomes involving functional status, which is of critical importance in hand surgery, the ICD-9 limits us from extracting such data. Furthermore, errors in the accuracy of recorded data due to human error are an inherent drawback of large database studies.

We used the NIS database to obtain our data. Due to the retrospective nature of our analysis, we cannot unequivocally establish a causal relationship between either comorbidity index and our outcome variables. The database also has inherent selection bias, since data only comes from patients who have been treated as inpatients. Cases treated as outpatients, such as those with less severe infections (small abscesses, paronychia, or felons), were not studied, and so our conclusions may not be wholly applicable for every hospital or provider. Furthermore, patients that left against medical advice or were discharged to an unknown destination were included in our definition of “non-routine discharge.” These may bias our data, particularly in urban populations, where rates of discharge against medical advice and to homeless shelters may be higher.48 Lastly, the lack of temporal data on the timing of each diagnosis (preoperative vs. postoperative status) is recognized in the literature with administrative datasets similar to the NIS.20,49,50 We mitigated this potential limitation by including only those diagnoses that most likely occurred after surgery. Therefore, the reported complication rate may be slightly lower than in truth.

5. Conclusions

There is a statistically significant difference in the discriminative performance of the Elixhauser and Charlson Comorbidity scores for morality, non-routine discharge, prolonged length of stay, and postoperative complications in the treatment of inpatient hand infections. While both show predictive value, the Charlson Comorbidity Index was superior in predicting mortality rate in the treatment of hand infections. The Elixhauser Comorbidity Index was superior in predicting non-routine discharge, prolonged length of stay, and postoperative complications. Development of a new index specific to hand infections may be warranted.

Sources of funding

None.

Statement of human and animal rights

This article does not contain any studies with human or animal subjects.

Declaration of competing interest

All authors have declared that they have no conflicts of interest for this project. This manuscript has not been submitted to, nor is under review at, another journal or other publishing venue. The authors have no affiliation with any organization with a direct or indirect financial interest in the subject matter discussed in the manuscript.

Acknowledgements

None.

Contributor Information

Dominick V. Congiusta, Email: dvc33@njms.rutgers.edu.

Kamil M. Amer, Email: kamil.amer@rutgers.edu.

Katie Otero, Email: ko247@njms.rutgers.edu.

Michael Metrione, Email: michael.metrione@rutgers.edu.

Aziz M. Merchant, Email: am1771@njms.rutgers.edu.

Michael Vosbikian, Email: vosbikmm@njms.rutgers.edu.

Ifran Ahmed, Email: ahmedi2@njms.rutgers.edu.

Appendix.

List of ICD-9 Codes of Postoperative Complications
ICD-9 Code Description
415.11, 415.12, 415.13, 415.19 Pulmonary embolism or infarction
518.4 Acute lung edema
518.51, 518.52, 518.53 Pulmonary insufficiency
518.81 Acute respiratory failure
996.40, 996.41, 996.42, 996.43, 996.44, 996.45, 996.46, 996.47, 996.49, 996.66, 996.67, 996.69, 996.77, 996.78, 996.79, 996.90, 996.91, 996.92, 996.93, 996.94, 996.95, 996.96 Mechanical, infective, inflammatory, or other adverse event due to internal device or reattachment
998.00, 998.01, 998.02, 998.09 Postoperative shock
998.11, 998.12, 998.13 Hematoma or seroma
998.30, 998.31, 998.32, 998.33 Wound disruption
998.51, 998.59 Postoperative infection
998.6 Postoperative Fistula
998.81, 998.82, 998.83, 998.89, 998.9 Other complications of procedure (subcutaneous emphysema, non-healing wound, complications not elsewhere classified)
999.1, 999.2, 999.31, 999.33, 999.34, 999.39, 999.51, 999.52, 999.59 Medical complications, not elsewhere classified
Open in a new tab

References

  • 1.Brown T.S., Hinojosa L.S., Cline K.E., Black S.R., Jamieson M.D., Starr A.J. Lack of acute hand care in the southwest United States. J Orthop Trauma. 2016;30(4):e129–e131. doi: 10.1097/BOT.0000000000000487. [DOI] [PubMed] [Google Scholar]
  • 2.Chong C.W., Ormston V.E., Tan A.B. Epidemiology of hand infection–a comparative study between year 2000 and 2009. Hand Surg. 2013;18:307–312. doi: 10.1142/S0218810413500317. [DOI] [PubMed] [Google Scholar]
  • 3.Houshian S., Seyedipour S., Wedderkopp N. Epidemiology of bacterial hand infections. Int J Infect Dis. 2006;10(4):315–319. doi: 10.1016/j.ijid.2005.06.009. Epub 2006 Feb 17. [DOI] [PubMed] [Google Scholar]
  • 4.Türker T., Capdarest-Arest N., Bertoch S.T., Bakken E.C., Hoover S.E., Zou J. Hand infections: a retrospective analysis. PeerJ. 2014;2:e513. doi: 10.7717/peerj.513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Dreyfuss U., Singer M. Human bites of the hand: a study of one hundred six patients. J Hand Surg. 1985;10A:884–889. doi: 10.1016/s0363-5023(85)80167-2. [DOI] [PubMed] [Google Scholar]
  • 6.Spiegel J.D., Szabo R.M. A protocol for the treatment of severe infections of the hand. J Hand Surg. 1988;13A:254–259. doi: 10.1016/s0363-5023(88)80060-1. [DOI] [PubMed] [Google Scholar]
  • 7.Gunther S.F. Introduction and overview (hand infections) Instr Course Lect. 1990;39:539–546. [PubMed] [Google Scholar]
  • 8.Tosti R., Trionfo A., Gaughan J., Ilyas A.M. Risk factors associated with clindamycin-resistant, methicillin-resistant Staphylococcus aureus in hand abscesses. J Hand Surg Am. 2015;40(4):673–676. doi: 10.1016/j.jhsa.2014.12.044. [DOI] [PubMed] [Google Scholar]
  • 9.Kiran R.V., McCampbell B., Angeles A.P. Increased prevalence of community-acquired methicillin-resistant Staphylococcus aureus in hand infections at an urban medical center. Plast Reconstr Surg. 2006;118(1):161–166. doi: 10.1097/01.prs.0000221078.63879.66. discussion 167-9. [DOI] [PubMed] [Google Scholar]
  • 10.Imahara S.D., Friedrich J.B. Community-acquired methicillin-resistant Staphylococcus aureus in surgically treated hand infections. J Hand Surg Am. 2010;35(1):97–103. doi: 10.1016/j.jhsa.2009.09.004. [DOI] [PubMed] [Google Scholar]
  • 11.Fowler J.R., Ilyas A.M. Epidemiology of adult acute hand infections at an urban medical center. J Hand Surg Am. 2013;38(6):1189–1193. doi: 10.1016/j.jhsa.2013.03.013. [DOI] [PubMed] [Google Scholar]
  • 12.Akdemir O., Lineaweaver W. Methicillin-resistant Staphylococcus aureus hand infections in a suburban community hospital. Ann Plast Surg. 2011;66(5):486–487. doi: 10.1097/SAP.0b013e318205734a. [DOI] [PubMed] [Google Scholar]
  • 13.Anwar M.U., Tzafetta K., Southern S.J. Review of community-referred hand infections. Surg Infect. 2008;9(3):357–366. doi: 10.1089/sur.2007.031. [DOI] [PubMed] [Google Scholar]
  • 14.Bach H.G., Steffin B., Chhadia A.M., Kovachevich R., Gonzalez M.H. Community-associated methicillin-resistant Staphylococcus aureus hand infections in an urban setting. J Hand Surg Am. 2007;32(3):380–383. doi: 10.1016/j.jhsa.2007.01.006. [DOI] [PubMed] [Google Scholar]
  • 15.Kapellen T.M., Galler A., Kiess W. Higher frequency of paronychia (nail bed infections) in pediatric and adolescent patients with type 1 diabetes mellitus than in non-diabetic peers. J Pediatr Endocrinol Metab. 2003;16(5):751–758. doi: 10.1515/jpem.2003.16.5.751. [DOI] [PubMed] [Google Scholar]
  • 16.Schneider E.C., Epstein A.M. Influence of cardiac-surgery performance reports on referral practices and access to care. A survey of cardiovascular specialists. N Engl J Med. 1996;335(4):251–256. doi: 10.1056/NEJM199607253350406. [DOI] [PubMed] [Google Scholar]
  • 17.Narins C.R., Dozier A.M., Ling F.S., Zareba W. The influence of public reporting of outcome data on medical decision making by physicians. Arch Intern Med. 2005;165:83–87. doi: 10.1001/archinte.165.1.83. [DOI] [PubMed] [Google Scholar]
  • 18.Pine M.J.H., Elixhauser A. Enhancement of claims data to improve risk adjustment of hospital mortality. J Am Med Assoc. 2007;297(1):71–76. doi: 10.1001/jama.297.1.71. [DOI] [PubMed] [Google Scholar]
  • 19.Austin S.R., Wong Y.-N., Uzzo R.G., Beck J.R., Egleston B.L. Why summary comorbidity measures such as the Charlson Comorbidity Index and Elixhauser score work. Med Care. 2005;53(9):e65–e72. doi: 10.1097/MLR.0b013e318297429c. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Menendez M.E., Neuhaus V., van Dijk C.N., Ring D. The elixhauser comorbidity method outperforms the Charlson index in predicting inpatient death after orthopaedic surgery. Clin Orthop Relat Res. 2014;472(9):2878–2886. doi: 10.1007/s11999-014-3686-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Kim C.Y., Sivasundaram L., LaBelle M.W., Trivedi N.N., Liu R.W., Gillespie R.J. Predicting adverse events, length of stay, and discharge disposition following shoulder arthroplasty: a comparison of the Elixhauser Comorbidity Measure and Charlson Comorbidity Index. J Shoulder Elbow Surg. 2018;27(10):1748–1755. doi: 10.1016/j.jse.2018.03.001. [DOI] [PubMed] [Google Scholar]
  • 22.Charlson M.E., Pompei P., Ales K.L., MacKenzie C.R. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40(5):373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
  • 23.Deyo R.A., Cherkin D.C., Ciol M.A. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619. doi: 10.1016/0895-4356(92)90133-8. [DOI] [PubMed] [Google Scholar]
  • 24.Sundararajan V., Henderson T., Perry C., Muggivan A., Quan H., Ghali W.A. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol. 2004;57(12):1288–1294. doi: 10.1016/j.jclinepi.2004.03.012. [DOI] [PubMed] [Google Scholar]
  • 25.Elixhauser A., Steiner C., Harris D.R., Coffey R.M. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27. doi: 10.1097/00005650-199801000-00004. [DOI] [PubMed] [Google Scholar]
  • 26.Grendar J., Shaheen A.A., Myers R.P. Predicting in-hospital mortality in patients undergoing complex gastrointestinal surgery: determining the optimal risk adjustment method. Arch Surg. 2012;147(2):126–135. doi: 10.1001/archsurg.2011.296. [DOI] [PubMed] [Google Scholar]
  • 27.Chu Y.T., Ng Y.Y., Wu S.C. Comparison of different comorbidity measures for use with administrative data in predicting short- and long-term mortality. BMC Health Serv Res. 2010;10:140. doi: 10.1186/1472-6963-10-140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Lieffers J.R., Baracos V.E., Winget M., Fassbender K. A comparison of Charlson and Elixhauser comorbidity measures to predict colorectal cancer survival using administrative health data. Cancer. 2011;117(9):1957–1965. doi: 10.1002/cncr.25653. [DOI] [PubMed] [Google Scholar]
  • 29.Quan H., Sundararajan V., Halfon P. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–1139. doi: 10.1097/01.mlr.0000182534.19832.83. [DOI] [PubMed] [Google Scholar]
  • 30.Healthcare Cost and Utilization Project. Overview of the National (Nationwide) Inpatient Sample (NIS) 2018. [Google Scholar]
  • 31.Fowler J.R., Greenhill D., Schaffer A.A., Thoder J.J., Ilyas A.M. Evolving incidence of MRSA in urban hand infections. Orthopedics. 2013;36(6):796–800. doi: 10.3928/01477447-20130523-27. [DOI] [PubMed] [Google Scholar]
  • 32.Kistler J.M., Thoder J.J., Ilyas A.M. MRSA incidence and antibiotic trends in urban hand infections: a 10-year longitudinal study. Hand (N Y). 2018;1558944717750921 doi: 10.1177/1558944717750921. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.HCUP NIS Trend Weights . 2018. Healthcare Cost and Utilization Project (HCUP) [Google Scholar]
  • 34.DeLong E.R., DeLong D.M., Clarke-Pearson D.L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–845. [PubMed] [Google Scholar]
  • 35.Bao P., Kodra A., Tomic-Canic M., Golinko M.S., Ehrlich H.P., Brem H. The role of vascular endothelial growth factor in wound healing. J Surg Res. 2009;153(2):347–358. doi: 10.1016/j.jss.2008.04.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Reinke J.M., Sorg H. Wound repair and regeneration. Eur Surg Res. 2012;49(1):35–43. doi: 10.1159/000339613. [DOI] [PubMed] [Google Scholar]
  • 37.Sahin H., Tholema N., Petersen W., Raschke M.J., Stange R. Impaired biomechanical properties correlate with neoangiogenesis as well as VEGF and MMP-3 expression during rat patellar tendon healing. J Orthop Res. 2012;30(12):1952–1957. doi: 10.1002/jor.22147. [DOI] [PubMed] [Google Scholar]
  • 38.Ackermann P.W., Hart D.A. Influence of comorbidities: neuropathy, vasculopathy, and diabetes on healing response quality. Adv Wound Care. 2013;2(8):410–421. doi: 10.1089/wound.2012.0437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Pradhan L., Nabzdyk C., Andersen N.D., LoGerfo F.W., Veves A. Inflammation and neuropeptides: the connection in diabetic wound healing. Expet Rev Mol Med. 2009;11:e2. doi: 10.1017/S1462399409000945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Edmonds M. Body of knowledge around the diabetic foot and limb salvage. J Cardiovasc Surg. 2012;53(5):605–616. [PubMed] [Google Scholar]
  • 41.Armstrong D.G., Nguyen H.C. Improvement in healing with aggressive edema reduction after debridement of foot infection in persons with diabetes. Arch Surg. 2000;135:1405. doi: 10.1001/archsurg.135.12.1405. [DOI] [PubMed] [Google Scholar]
  • 42.Greiner M., Pfeiffer D., Smith R.D. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med. 2000;45:23–41. doi: 10.1016/s0167-5877(00)00115-x. [DOI] [PubMed] [Google Scholar]
  • 43.Swets J.A. Measuring the accuracy of diagnostic systems. Science. 1988;240:285–1293. doi: 10.1126/science.3287615. [DOI] [PubMed] [Google Scholar]
  • 44.Weir R.E., Jr., Lyttle C.S., Meltzer D.O., Dong T.S., Ruhnke G.W. The relative ability of comorbidity ascertainment methodologies to predict in-hospital mortality among hospitalized community-acquired pneumonia patients. Med Care. 2018;56(11):950–955. doi: 10.1097/MLR.0000000000000989. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Ladha K.S., Zhao K., Quraishi S.A. The Deyo-Charlson and Elixhauser-van Walraven Comorbidity Indices as predictors of mortality in critically ill patients. BMJ Open. 2015;5(9) doi: 10.1136/bmjopen-2015-008990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Chang H.J., Chen P.C., Yang C.C., Su Y.C., Lee C.C. Comparison of elixhauser and Charlson methods for predicting oral cancer survival. Medicine (Baltim) 2016;95(7) doi: 10.1097/MD.0000000000002861. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Southern D.A., Quan H., Ghali W.A. Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administrative data. Med Care. 2004;42(4):355–360. doi: 10.1097/01.mlr.0000118861.56848.ee. [DOI] [PubMed] [Google Scholar]
  • 48.Menendez M.E., van Dijk C.N., Ring D. Who leaves the hospital against medical advice in the orthopaedic setting? Clin Orthop Relat Res. 2015;473(3):1140–1149. doi: 10.1007/s11999-014-3924-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Ghali W.A., Quan H., Brant R. Risk adjustment using administrative data: impact of a diagnosis-type indicator. J Gen Intern Med. 2001;16(8):519–524. doi: 10.1046/j.1525-1497.2001.016008519.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Myers R.P., Quan H., Hubbard J.N., Shaheen A.A., Kaplan G.G. Predicting in-hospital mortality in patients with cirrhosis: results differ across risk adjustment methods. Hepatology. 2009;49(2):568–577. doi: 10.1002/hep.22676. [DOI] [PubMed] [Google Scholar]

Articles from Journal of Clinical Orthopaedics and Trauma are provided here courtesy of Elsevier

RESOURCES