Comorbidity-polypharmacy score predicts readmission in older trauma patients

B. Chaise Housley, BA,a Stanislaw P.A. Stawicki, MD, FACS,b,1 David C. Evans, MD,b Christian Jones, MDb

a The Ohio State University College of Medicine, 370 W 9th Avenue, Columbus, OH 43210 USA

b The Ohio State University College of Medicine, Department of Surgery, Division of Trauma, Critical Care and Burn, 395 W 12th Avenue Ste 634, Columbus, OH 43210 USA

1 Present Address: St. Luke’s University Health Network, Department of Research & Innovation, Bethlehem, PA 18015 USA

Contact information:B. Chaise Housley, BA - blain.housley@osumc.eduStanislaw P.A. Stawicki, MD, FACS - Stanislaw.Stawicki@sluhn.orgDavid C. Evans, MD - david.evans@osumc.eduChristian Jones, MD (Corresponding author) - christian.jones@osumc.edu

Tel.: +1 614 293 9348The Ohio State University College of MedicineDepartment of Surgery, Division of Trauma, Critical Care and Burn395 W 12th Avenue Ste 634Columbus, OH 43210 USA

Author contributions: Study design: BCH, SPAS, DCE, CJ; Data collection: BCH, CJ; Statistical analysis & interpretation: BCH, CJ; Manuscript authorship: BCH, CJ; Critical review & revision of manuscript: all authors

Accepted - May 2015

Abstract

Background: Hospital readmissions are considered to be a measure of quality of care, correlate with worse outcomes, and may soon lead to decreased reimbursement. The comorbidity-polypharmacy score (CPS) is the sum of the number of pre-injury medications and comorbidities, and may estimate patient frailty more effectively than patient age. This study evaluates the association between CPS and readmission.

Methods: Medical records for trauma patients ≥45 years old evaluated between January 1 and December 31, 2008, at our ACS-verified Level 1 trauma center were reviewed to obtain information on demographics, injuries, pre-injury comorbidities and medications, and occurrences of readmission to our facility within 30 days of discharge. Chi-square and Kruskal-Wallis testing was used to evaluate differences between readmitted and non-readmitted patients, with multiple logistic regression used to evaluate the contribution of independent risk factors for readmission.

Results: 879 patients were included; their ages ranged from 45-103 (median 58), injury severity scores (ISS) from 0-50 (median 5), and CPS from 0-39 (median 7). 76 patients (8.6%) were readmitted to our facility within 30 days of discharge. The readmitted cohort had higher CPS (median 9.5, range 0-32, p=0.031) and ISS (median 9, range 1-38, p=0.045), but no difference in age (median 59.5, range 47-99, p=0.646). Logistic regression demonstrated independent association of higher CPS with increased risk of readmission, with each CPS point increasing readmission likelihood by 3.5% (p=0.03).

Conclusion: CPS appears to correlate well with readmissions within 30 days. Frailty defined by CPS was a significantly stronger predictor of readmission than was patient age. Early recognition of elevated CPS may improve discharge planning and help guide interventions to decrease readmission rates in older trauma patients.

Keywords: trauma, readmission, frailty, elderly, quality

1. Introduction

Unplanned hospital readmission is considered an indicator of poor quality of care, has been found to correlate with worse outcomes, and may soon lead to decreased reimbursement from third party payors.1,2 Despite the importance of this quality marker, known risk factors for readmission are still poorly defined.3 One important suggested risk factor for readmission is patient age.3 As the current population ages, a larger number of older individuals will be injured and require hospital care for trauma.4 All “older” individuals are not identical, however, and “physiologic age” or “frailty” may be more important to outcomes than chronologic age. This “physiologic old age” brings with it an increased number of comorbidities, and long-term control of these chronic diseases necessitates the use of an increasing number of medications.5,6

The polypharmacy score is simply the number of drugs a patient takes. As the polypharmacy score in older individuals rises, an increased rate of complications following injury is seen.5,7 The polypharmacy score alone is a predictor for unfavorable outcomes in older trauma patients, but our group’s previous work has shown that the addition of comorbidities to the score may result in a more powerful tool for estimating morbidity and mortality in this population.8,9

The comorbidity-polypharmacy score (CPS) is the sum of the number of pre-injury medications and the number of pre-injury comorbidities and may estimate patient frailty more effectively than patient age does.5,8,9 Though CPS has previously been correlated with accuracy of triage from the emergency department and with clinical outcomes in older trauma patients8,9, no information is currently available regarding the association between CPS and hospital readmission. This study was designed to evaluate the association between hospital readmission and CPS in older trauma patients, with the secondary goal to compare CPS to other potential predictors for readmission. We hypothesized that there would be a positive correlation between CPS and hospital readmission in trauma patients 45 years of age and older.

2. Materials and methods

2.1 General information

Institutional Review Board approval was granted. We used our institutional trauma registry to identify all injured patients 45 years of age or older seen at our ACS-verified Level 1 trauma center over the 2008 calendar year. Per ACS standards, this registry includes patients with identified injuries at time of admission or at time of evaluation in the institution’s emergency department, and therefore includes patients who are admitted to the hospital and those discharged directly from the emergency department. Patients excluded from evaluation included inmates, those who died prior to discharge, and those discharged to hospice care. We retrospectively gathered data on each patient from the institutional trauma registry and from the institution’s electronic medical record.

2.2 Data collection

An allergy and medication history is taken on each patient seen at our facility and that information is entered into the institution’s electronic medical record; this process has been previously described.5 All patients similarly have chronic medical problems delineated and recorded. Exceptions to the gathering of these data occur in patients who present in conditions that do not allow them to communicate their history and when no alternative source of history is available. Institutional trauma registry data and electronic medical records were reviewed to obtain patient demographics, trauma activation level (1, 2, or not activated), mechanism of injury (blunt, penetrating, or burn), injury severity score (ISS), Glasgow coma score (GCS), the prospectively gathered pre-injury comorbidities and medications, lengths of stay (LOS), intensive care unit (ICU) LOS, and occurrences of readmission to our facility within 30 days of discharge.

2.3 Comorbidity-Polypharmacy Score

Each patient’s CPS was calculated retrospectively based on the prospectively gathered data by totaling the patient’s outpatient medications and pre-injury comorbidities. For example, a patient presenting with history of hypertension and gout and currently taking metoprolol and allopurinol is given a comorbidity score of 2, a polypharmacy score of 2, and a CPS of 4. Prior studies9 have arbitrarily defined “CPS groups” of level 1 “Minor” (CPS 0-7); 2, “Moderate” (8-14); 3, “Severe” (15-21); and 4, “Morbid” (≥22), and we additionally classified each patient into the appropriate CPS group.

2.4 Readmission criteria

Readmission was defined as being admitted to our institution within thirty days of the most recent discharge from our institution. We did not consider the participation of a patient in our medical center’s inpatient rehabilitation center as being “hospitalized”; readmissions were included if within thirty days of discharge from our institution’s non-rehabilitation units. Evaluations at or admissions to other medical centers were not obtained and were not included in tabulated readmissions.

2.4 Statistical analysis

Data analysis was performed to evaluate differences between readmitted patients and those who were not, including evaluation of patient age, gender, mechanism of injury, LOS, ICU LOS, trauma activation level, ISS, GCS, CPS, and CPS group. Appropriate descriptive statistics were evaluated for each variable. Chi-square testing was used to evaluate differences in categorical variables (gender, mechanism of injury, trauma activation level, and CPS group) and Kruskall-Wallis testing was used to evaluate differences in ordinal variables (age, LOS, ICU LOS, ISS, GCS, and CPS) between admitted and non-readmitted groups. Variables which approached statistically significant difference (p<0.2) between admitted and non-readmitted groups were included in multiple logistic regression analysis to evaluate the independence of the variables and the adjusted odds ratio (AOR) of independent risk factors for readmission. Because of the obvious interplay of CPS and CPS group, CPS group was not included in the multiple logistic regression model. However, additional predictive modeling based on rates of readmission of each

CPS group was evaluated using descriptive statistics and chi-square testing. Statistical significance of all results was set at α = 0.05 per convention.

3. Results

960 injured patients age 45 years and older were evaluated at our institution in 2008 (Figure 1); 79 patients were excluded per above criteria, including 48 who died, 29 prisoners, and 2 discharged to hospice care; 2 further were excluded due to unobtainable medical records. 879 patients were included in our final analysis; their ages ranged from 45-103 (median 58) years, injury severity scores (ISS) from 0-50 (median 5), and CPS from 0-39 (median 7). 76 patients (8.6%) were readmitted to our facility within 30 days of discharge. Detailed results of the overall patient group as well as the readmitted and non-readmitted cohorts are shown in Table 1. Of note, the readmitted cohort had a higher CPS (median 9.5 versus 7, p=0.031), and ISS (median 9 vs 5, p=0.045), but no difference in age (median 59.5 vs. 58, p=0.646). Statistically significant differences were additionally noted in both hospital and ICU LOS. CPS group differences were noted with p=0.1202; no other variables differed with p<0.2. Age, gender, mechanism of injury, trauma activation level, and GCS all failed to be associated with readmission risk.

Multiple logistic regression including LOS, ICU LOS, ISS, and CPS was performed on the 866 (98.5%) patients with all four data points available. This demonstrated independent association of all four with readmission (Table 2). Higher CPS was associated with increased risk of readmission, with each CPS point increasing likelihood of readmission by 3.6% (adjusted odds ratio 1.0364, 95% confidence interval 1.0059-1.0678, p=0.0191). ISS was also independently associated with increased risk of readmission (adjusted unit odds ratio 1.0504, 95% CI 1.013-1.0891, p=0.0078). Increased LOS additionally was associated with increased risk of readmission, while increased ICU LOS was associated with decreased readmission risk (AOR 0.9334, CI 0.8778-0.9926, p=0.0279).

Evaluation of CPS groups and their individual rates of readmission revealed a trend toward higher rates of readmission as group level increased (Table 3, Figure 2), but this was not statistically significant (p=0.1223).

4. Discussion

4.1 Background

As the general population ages, so too does the population of hospitalized patients and trauma patients. The variability in patient conditions even at similar ages is becoming more apparent, and “physiologic old age” is differentiated from “chronologic old age” with a range of objective and subjective observations. Defining “frailty” loosely as the consequence of “physiologic old age” leads to the recognition of frailty as a potential confounder in patient outcomes, potentially one more appropriate than a patient age alone. Objective evaluations of frailty have varied from limiting the definition of frailty to a structured syndrome10 to measuring as many as seventy variables including comorbidities, laboratory values, physical findings, and activity limitations11.

These measures have been correlated with patient outcomes, but simpler mechanisms for estimating patient frailty are sought.

Providers are beginning to see the effects of older individuals being prescribed an increasing number of pharmacological therapies6–8,12. Trauma patients may be more likely to experience adverse sequelae as compared to non-trauma patients8,13,14, and the risk for complications has already been shown to increase as the polypharmacy score rises8,13. The risk of poor outcomes in trauma patients specifically is known to increase with higher polypharmacy scores5. While a patient’s number of medical comorbidities (or the presence of specific comorbidities) has previously been described as only a partial contributor to the patient’s overall frailty, we hypothesized that the combination of comorbidities with the patient’s level of polypharmacy could function as an effective measure of patient frailty. Prior evaluations have demonstrated CPS accuracy in predicting discharge destination15 and in estimating risk of undertriage from the emergency department8, outcomes commonly measured by other frailty indices.

For hospitals and physicians, readmissions suggest the possibility that better care could have been given during the initial hospitalization1,2 Readmissions not only represent a surrogate for negative outcomes for patients but may also translate into “negatives” for physicians and health systems from standpoints of cost, reputation, and even reimbursement. As healthcare costs are further evaluated, readmission rates will likely decrease hospital reimbursement directly while additionally being linked with decreased patient satisfaction, potentially further lowering hospital and physician reimbursements.

Our study found a statistically significant positive correlation between CPS and hospital readmission in older trauma patients. ISS and CPS appear to have similar statistical power in independently predicting readmission, though ISS may not be known early in the patient’s hospitalization. Our results suggest CPS is better able to indicate the “frailty” of a person than age, especially with regard to their risk for readmission.

There are many studies that have identified risk factors for hospital readmissions, including increased hospital LOS, DVT development, number of prior admissions, discharge destination, number of comorbidities, age, and surgical site infection.16–19 While this information is helpful in evaluating readmission rates, the correlation between CPS and readmission has not been studied until now. The especially appealing quality of the CPS is that it appears to be a predictor of multiple complications that is readily available at the beginning of a patient’s hospitalization in the era of mandatory medication reconciliation.

4.2 Benefits and Limitations of CPS

Earlier risk factor scoring systems based on comorbidities have underscored particular diseases and downplayed or ignored the severity of individual diseases. The commonly evaluated Charlson comorbidity index was developed in 1987 and weighs several specific diseases based simply on their presence14. This unfortunately leads to “dating” of the scoring system as medicine advances and some diseases become more easily managed, as other illnesses are found

to be greater predictors of complications, and as patient populations change and comorbidities once common become less frequent. These potentially lead to the antiquation of such scoring systems.

As a general rule, however, we assume that a preexisting medical condition that requires fewer medications for treatment is in fact less severe than a case requiring many medications. For example, a patient who is taking one oral medication for diabetes mellitus is likely to have less-severe diabetes mellitus than a patient who requires multiple agents for glycemic control9. Similarly, we hypothesized that focusing exclusively on the pharmacologic profile of a patient without regard to the multiple pre-existing diagnoses that patient underestimates the importance of disease synchronicity and the synergistic fervor with which multiple diseases attack the physiologic reserve. The simple use of number of illnesses and number of medications additionally allows the scoring system to keep pace with medical progress. As a disease becomes easier to manage, the number of medications necessary to do so decreases. As populations change and different illnesses become more common, they are not ignored.

The Charlson index makes use of assigned weights to a specific list of diseases. Other frailty scores, while often useful, require online calculators or a plethora of data points. Such systems should become more automated and readily available as electronic medical record systems advance, but this has not yet materialized. However, CPS is easily evaluated by simply glancing at a patient’s medication and diagnosis lists. While physicians’ subjective ease of use of other indices could not be compared to CPS in this retrospective evaluation, until other scores are more automatically provided, CPS must be assumed at least as available as the others.

The corollary to the ease of calculating CPS must also be the assumption of missing information, of course. The Rockwood frailty index11 with more than 50 variables clearly captures more data, and we cannot discard the possibility that some of that data may be clinically relevant. With simplification comes the risk of oversimplification, and CPS has not been compared to other traditional frailty scores directly.

In other publications, CPS has demonstrated the aforementioned correlations with accuracy of triage from the emergency department and with clinical outcomes in older trauma patients8,9 CPS is easy to calculate; it can be done quickly as a patient arrives at the ED without the need to ask for any “extra” information from the patient. CPS has been shown to be comparable to the Charlson comorbidity index in predicting mortality.20

CPS is not a universally applicable tool. We chose to study patients 45 years and older based on research which shows that younger patients have an insufficient number of chronic comorbidities and too few long-term (“maintenance”) medications to make the use of CPS effective in that age group.5,6,8,21 Even “healthy” older patients, including the marathon runners in their sixties who take no medications and have no worrisome medical history should not be equated with the 20-year-old college student with excess physiologic reserve. Similarly, patients without regular medical care are likely to have undiagnosed and untreated illnesses, and must not be presumed

equivalent to individuals with an astute primary care physician who has identified no chronic medical problems.

Some limitations are mainly based upon the reliability of patient reporting. When a patient lacks a medication list or is unable to recall all of their medications, an artificially low CPS may result. An artificially high CPS may result if a patient reports medication usage when they are, in fact, non-adherent. The most extreme cases of artifactually incorrect CPS may result commonly in trauma patients, the absence of necessary data for patients who present to the medical center unable to relay their comorbidities and medications due to either prior incapacitation (e.g., dementia) or the traumatic condition (e.g., intubated status, acute intoxication, or unconsciousness).

Finally, it must be assumed that, much like other scoring systems which identify specific diagnoses, medications, or laboratory values, CPS also misses information by equating all diagnoses and medications. We reasonably assume that a patient with cerebral vascular disease who takes an irreversible platelet inhibitor is at a different risk level following injury than a patient who takes allopurinol for gout, though both would have a CPS of 2. While the former would likely have other comorbidities and medications as well, the rare patient who does fit this hypothetical mold is probably underserved by CPS.

4.3 Other Predictors

This study additionally identified hospital length of stay (LOS) and injury severity score (ISS) as independent predictors of increased readmission risk. These are unsurprising and consistent with prior studies. However, the failure of patient age to have significant association with readmission risk agrees with studies by Morris et al.15,17 while being at odds with a large study by Moore et al.16 None of these studies were limited to older patients, however.

Somewhat more surprising is the inverse association apparent between ICU LOS and readmission risk, especially as an independent variable not further confounded by comorbidities (CPS) and injury severity. Though need for discharge to a facility was not evaluated in this study, it is reasonable to consider that a patient with an extended ICU stay would be more likely to be discharged to a skilled facility, and potentially therefore less likely to return to the hospital. Alternatively, a patient’s increased ICU LOS may prompt a provider to be more conservative in discharge planning and thus decrease their readmission likelihood.

4.4 Limitations of the Study

Limitations of the study include the inability to determine causal relationships based on its retrospective design, as well as those limitations noted above inherent to the CPS itself. Our study is strengthened, however, by the high quality of pre-injury medication and comorbidity data reported during the patient’s hospital stay3,4; that no effort was needed to remove patients with such artificially depressed CPS in order to obtain clear statistical significance again speaks to the scoring system’s power.

Though the use of older (2008) data is a potential weakness of the study, the data set was specifically chosen for its well-reviewed validity after uses in multiple trauma patient retrospective evaluations. CPS again bolsters this use, as the scoring system represents appropriate medical care at the time of evaluation rather than limiting itself to specific diagnoses whose relative severity may have changed since the data collection for this study took place. CPS applicability to patients now should reasonably be expected to be equivalent to its use in 2008.

As noted above, this retrospective evaluation does not allow direct comparison to other frailty indices. Other predictors of readmission (including ISS and potentially age) are included, and other intended prognostic indicators such as GCS are indeed found to be suboptimal in this study. Another study has directly compared CPS with the Charlson comorbidity index for predicting patient mortality14, but further prospective analysis will be required to compare CPS to other frailty measurement systems in prediction of readmission risk.

The final potential frailty measurement not included here is the functional status of the older trauma patient. It is reasonable to consider a patient who is totally or partially dependent more frail than a fully independent patient regardless of the number of comorbidities and medications. The surrogate measures of admission from or discharge to a skilled nursing facility may allow further evaluation of this in the future.

The most glaring and presumably important underestimate of readmissions in our study is the inability to include re-evaluation or admission at a different medical center; this data is unfortunately unavailable to us except for the fraction of our patients served by the Centers for Medicare and Medicaid Services (CMS). However, it most certainly impacts the national quality measures of readmission that CMS is beginning to monitor, and will be an important part of most readmission studies going forward.

4.5 Recommendations & Future Directions

We see potential in using CPS to improve outcomes for trauma patients. Further studies are needed in order to expand the current understanding of the clinical utility of CPS, including its ability to predict outcomes prospectively and its usefulness to providers who are made aware of its value at the time of patient hospitalization. However, while increasing CPS does predict increasing risk of readmission, a particular threshold of concern is not yet known. It is tempting to identify the CPS group of a patient as a predictor of hospital readmission risk even without statistical significance. For instance, recognition that a patient is in group 4, the “Morbid” CPS group, would reasonably lead one to consider a significantly increased likelihood of readmission. However, these groups have previously only been arbitrarily defined, and optimization of the group inclusion parameters needs to be undertaken.

Prospective evaluation of CPS in clinical and readmission outcomes is ongoing, including determining appropriate “high” and “low” risk groups, as well as determining whether early recognition of a higher CPS leads to changes in discharge planning or other interventions to

reduce readmission risks. Finally, prospective comparisons of CPS to other frailty indices in a multicenter fashion are being considered.

5. Conclusion

CPS is simple to calculate and appears to correlate well with readmissions within 30 days. Indeed, CPS was a significantly stronger predictor of readmission than patient age was, and was equivalent to ISS. Early recognition of elevated CPS may improve discharge planning and help guide interventions aimed at decreasing readmission rates in older trauma patients, and prospective evaluations of this score are ongoing. Larger multicenter evaluations of CPS as an indicator for the frailty of older patients, especially in comparison to other frailty indices, are warranted.

AcknowledgementsThe authors appreciate the chart review and data entry work performed by Mr. Nicholas J. Kelly, BA, and Mr. Fady J. Baky, BS, of The Ohio State University College of Medicine.

Disclosure

This research was made possible in part by The Ohio State College of Medicine Roessler Medical Student Research Scholarship. The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article.

REFERENCES

1. Axon RN, Williams M V. Hospital readmission as an accountability measure. JAMA. 2011;305(5):504–5. doi:10.1001/jama.2011.72.

2. Holt PJE, Poloniecki JD, Hofman D, Hinchliffe RJ, Loftus IM, Thompson MM. Re-interventions, readmissions and discharge destination: modern metrics for the assessment of the quality of care. Eur J Vasc Endovasc Surg. 2010;39(1):49–54. doi:10.1016/j.ejvs.2009.10.003.

3. Hasan M. Readmission of patients to hospital: still ill defined and poorly understood. Int J Qual Health Care. 2001;13(3):177–9.

4. Fantus RJ. NTDB® data points. Older and thinner. Bull Am Coll Surg. 2010;95(9):59–60.

5. Evans DC, Gerlach AT, Christy JM, et al. Pre-injury polypharmacy as a predictor of outcomes in trauma patients. Int J Crit Illn Inj Sci. 2011;1(2):104–9. doi:10.4103/2229-5151.84793.

6. Stawicki SP, Gerlach AT. Polypharmacy and medication errors : Stop , Listen , Look , and Analyze …. OPUS 12 Sci. 2009;3(1):6–10.

7. Wisler JR, Springer AN, Hateley K, et al. Pre-injury neuro-psychiatric medication use, alone or in combination with cardiac medications, may affect outcomes in trauma patients. J Postgrad Med. 60(4):366–71. doi:10.4103/0022-3859.143957.

8. Justiniano CF, Evans DC, Cook CH, et al. Comorbidity-polypharmacy score: a novel adjunct in post-emergency department trauma triage. J Surg Res. 2013;181(1):16–9. doi:10.1016/j.jss.2012.05.042.

9. Evans DC, Cook CH, Christy JM, et al. Comorbidity-polypharmacy scoring facilitates outcome prediction in older trauma patients. J Am Geriatr Soc. 2012;60(8):1465–70. doi:10.1111/j.1532-5415.2012.04075.x.

10. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001;56(3):M146–M156. doi:10.1093/gerona/56.3.M146.

11. Rockwood K, Song X, MacKnight C, et al. A global clinical measure of fitness and frailty in elderly people. Cmaj. 2005;173(5):489–495. doi:10.1503/cmaj.050051.

12. Fastbom J. [Increased consumption of drugs among the elderly results in greater risk of problems]. Lakartidningen. 2001;98(14):1674–9.

13. Corbett SM, Rebuck JA. Medication-related complications in the trauma patient. J Intensive Care Med. 23(2):91–108. doi:10.1177/0885066607312966.

14. Lazarus HM, Fox J, Evans RS, et al. Adverse drug events in trauma patients. J Trauma. 2003;54(2):337–43. doi:10.1097/01.TA.0000051937.18848.68.

15. Justiniano CF, Coffey RA, Evans DC, et al. Comorbidity-polypharmacy score predicts in-hospital complications and the need for discharge to extended care facility in older burn patients. J Burn Care Res. 36(1):193–6. doi:10.1097/BCR.0000000000000094.

16. Morris DS, Rohrbach J, Rogers M, et al. The surgical revolving door: risk factors for hospital readmission. J Surg Res. 2011;170(2):297–301. doi:10.1016/j.jss.2011.04.049.

17. Moore L, Stelfox HT, Turgeon AF, et al. Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study. Ann Surg. 2014;259(2):374–80. doi:10.1097/SLA.0b013e31828b0fae.

18. Morris DS, Rohrbach J, Sundaram LMT, et al. Early hospital readmission in the trauma population: are the risk factors different? Injury. 2014;45(1):56–60. doi:10.1016/j.injury.2013.04.029.

19. Dharmarajan K, Hsieh AF, Lin Z, et al. Hospital readmission performance and patterns of readmission: retrospective cohort study of Medicare admissions. BMJ. 2013;347:f6571.

20. Holmes M, Garver M, Albrecht L, Arbabi S, Pham TN. Comparison of Two Comorbidity Scoring Systems for Older Adults with Traumatic Injuries. J Am Coll Surg. 2014. doi:10.1016/j.jamcollsurg.2014.05.014.

21. Christy JM, Stawicki SP, Jarvis AM, et al. The impact of antiplatelet therapy on pelvic fracture outcomes. J Emerg Trauma Shock. 2011;4(1):64–9. doi:10.4103/0974-2700.76841.

Figure 1. Patients enrolled and excluded out of all trauma patients ages 45 years and older evaluated in calendar year 2008.

Table 1. Univariate analyses of patient characteristics and their associations with readmission. Though 879 patients were evaluated, not all data were available for all patients; percentages are of patients with data available, and n values are given for variables which did not include all patients. Numbers are given as median (range) for ordinal data and percent (number/total available) for categorical data. Statistical significance is evaluated with chi-square tests for gender, mechanism of injury, trauma level, and CPS group; it is evaluated with Kruskal-Wallis tests for age, ISS, GCS, LOS, ICU LOS, and CPS. ISS: injury severity score; GCS: Glasgow coma score; LOS: length of stay; ICU: intensive care unit: CPS: comorbidity-polypharmacy score.

Overall Readmitted Nonreadmitted pTotal 8.6% (76) 91.4% (803)Age (years) 58 (45-103) 59.5 (47-99) 58 (45-103) 0.646Gender (n=659)

MaleFemale

58.3% (384/659)41.7% (275/659)

61.8% (34/55)38.1% (21/55)

57.9% (350/604)42.1% (254/604)

0.6801

Mechanism of injury (n=557)

BluntPenetratingBurn

86.0% (479/557)2.7% (15/557)11.3% (63/557)

81.8% (36/44)2.3% (1/44)15.9% (7/44)

86.4% (443/513)2.7% (14/513)10.9% (56/513)

0.6005

Trauma level12Not activated

6.5% (57/879)34.8% (306/879)58.7% (516/879)

10.5% (8/76)30.3% (23/76)59.2% (45/76)

6.1% (49/803)35.2% (283/803)58.7% (471/803)

0.2739

ISS (n=867) 5 (0-50) 9 (1-38) 5 (0-50) (n=791) 0.045GCS (n=797) 15 (3-15) 15 (3-15) (n=67) 15 (3-15) (n=730) 0.295LOS (days) (n=878) 3 (0-124) 5 (0-43) 3 (0-124) (n=802) 0.00268ICU LOS 0 (0-124) 0 (0-16) 0 (0-124) 0.05CPS 7 (0-39) 9.5 (0-32) 7 (0-39) 0.031CPS Group

1234

52.6% (462/879)27.9% (245/879)12.4% (109/879)7.2% (63/879)

42.1% (32/76)28.9% (22/76)17.1% (13/76)11.8% (9/76)

53.7% (431/803)27.8% (223/803)12.0% (96/803)6.7% (54/803)

0.1202

Table 2. Variables evaluated for independent contribution to risk of readmission. Multiple logistic regression was used to determine independence and adjusted unit odds ratio for each variable; all were statistically significant. ICU: intensive care unit; ISS: injury severity score; CPS: comorbidity-polypharmacy score; CI: confidence interval.

Adjusted unit odds ratio (95% CI)

p

Length of Stay (days) 1.0492 (1.0125-1.0872) 0.0081ICU Length of Stay (days) 0.9334 (0.8778-0.9926) 0.0279ISS 1.0504 (1.013-1.0891) 0.0078CPS 1.0364 (1.0059-1.0678) 0.0191

Table 3. Rates of readmission by CPS group. Numbers are total (percent of CPS group). Chi-square testing shows no statistical significance, p=0.1223.CPS Group 1

(CPS 0-7)2(CPS 8-14)

3(CPS 15-21)

4(CPS ≥22)

Total patients 462 245 109 63Readmitted 32 (6.9%) 22 (9.0%) 13 (11.9%) 9 (14.3%)Nonreadmitted 430 (93.1%) 223 (91.0%) 96 (88.1%) 54 (85.7%)

Figure 2. Readmission rates per CPS group. CPS groups are defined as 1, "Minor" (CPS 0-7); 2, "Moderate" (CPS 8-15); 3, "Severe" (CPS 16-21); and 4, "Morbid" (CPS ≥22).