Improving Hip Fractures Outcomes for COPD Patients

Abstract

Hip fractures in the elderly have high rates of mortality and perioperative complications. Both men and COPD patients have worse mortality and complications but this may be due to more co-morbid disease. We assessed mortality and complications in a large cohort (n = 12,646) of men undergoing hip fracture surgery within the Veteran's Health Affairs (VHA) to define the association of COPD to these outcomes after adjusting for other key factors. We looked for opportunities to improve outcomes for COPD patients. Methods: Using the VA Surgical Quality Improvement Program (VASQIP), and administrative databases, we determined COPD status, types of co-morbid conditions and surgical factors, and compared these to outcomes of surgical complications, 30-day and one-year mortality for patients who underwent hip fracture repair during 1998 to 2005. Results: COPD was noted in 47% of the hip fracture patients studied. In 3,261 (26%) cases, the COPD was “severe: (indicated by functional disability, previous hospitalization for exacerbation, chronic drug treatment or record of FEV₁ <75% predicted), and in 2,736 (21%) cases it was considered “mild” (any previous outpatient visit or hospitalization with a coded diagnosis of COPD). Severe COPD patients had one year mortality of 40.2% compared to 31.0% in mild COPD and 28.8% in non-COPD subjects. Current smoking, use of general anesthesia and delays to surgery were significant modifiable risk factors identified in adjusted models. Osteoporosis was known pre-fracture in only 3% of subjects. Conclusions: COPD was very common in male veterans with hip fractures and was associated with increased risk of death and complications. Increased use of regional anesthesia and urgent scheduling of hip fracture surgery may improve outcomes for patients with COPD. Osteoporosis was rarely identified preoperatively. Improving diagnosis and treatment of osteoporosis in COPD patients could reduce the incidence of hip fractures.

Keywords :

• COPD
• Hip fracture
• Mortality
• Osteoporosis
• Steroid use
• Surgical omplications
• Pneumonia
• Spinal anesthesia
• General anesthesia
• Current smoking.

Introduction

Hip fractures are common in aging populations, largely due to the combined effect of reduced bone strength and an increased propensity to fall. The occurrence of a hip fracture often precipitates death, both in the immediate perioperative period and during the following year (1–3). Specific factors associated with mortality include advancing age, male gender (2, 7), co-morbid diseases (4), delays getting patients into surgery (3) and choice of anesthetic (5, 6).

Co-morbid chronic obstructive pulmonary disease (COPD) appears to be associated with worse 30-day and 1-year mortality after hip fracture. A large population based study from Denmark found COPD patients had 60–70% increase in mortality at 1 year after hip fracture but did not study surgical factors or adjust for other factors such as co-morbid disease or baseline functional status (8). Previous work with VA datasets found increased perioperative complications in COPD subjects after adjustment for demographics and other co-morbid conditions, but found adjusted 30 day mortality was not increased in COPD (6). Other publications have identified the negative impact of COPD on hip fracture outcomes but have not controlled for surgical and perioperative patient factors (8–10). Although the effect of COPD and other co-morbid conditions cannot be eliminated at the time of hip fracture, perioperative management strategies may improve surgical outcomes. Furthermore, given the association of osteoporosis to both COPD (11) and hip fractures (12), primary prevention of hip fractures may also improve population outcomes.

The present study primarily focused on COPD as it pertains to hip fracture repair outcomes. It included identifying cases with mild COPD and also two additional years of data as well as 1-year mortality outcomes for the cohort. We postulated that men with COPD would have worse mortality and higher complications following hip fractures, and that pre-operative diagnosis of osteoporosis would be low. We were particularly interested in identifying surgical aspects of hip fracture care in the context of co-existent COPD that might reduce risks for this subset of high-risk patients.

Methods

Study design

This was a retrospective cohort study using data from the U.S. Department of Veterans Affairs Surgical Quality Improvement Program (VASQIP) from 1998–2005. Investigators received institutional review board approval, a waiver of informed consent, and approval from the VA Surgical Quality Data Use Group prior to accessing or analyzing any data.

Cohort

We selected male patients who underwent acute hip fracture repair surgery during fiscal years 1998 to 2005 in any VA hospital. Patients were identified by CPT codes 27235, 27236, 27244, or 27245 indicating open or closed treatment of a hip fracture. We also included CPT codes 27125 (hemiarthroplasty) and 27130 (total hip arthroplasty) when the patient had a concurrent diagnosis code consistent with an acute hip fracture (ICD-9 codes 820.x, 820.2x, or 820.8) (Table 1). We excluded patients with disseminated cancer because the etiology and prognosis for these cases is likely to differ from other hip fractures.

Table 1.  Description of CPT codes.

	Description
CPT code 27125	Hemiarthroplasty not femoral neck fracture
CPT code 27235	Percutaneous fixation of femoral neck fracture
CPT code 27236	Femoral Neck fracture treated with prosthesis or internal fixation
CPT code 27244	Intertrochanteric fractures and sliding hip screw
CPT code 27245	Intertrochanteric fracture treated with Intramedullary rod
CPT code 27130	Total hip arthroplasty

Data collection

The data collection methods for the VASQIP have been described in detail previously (13). Nearly all patients who are scheduled to undergo major non-cardiac procedures with use of general, spinal, or epidural anesthesia in VA hospitals are included in the database.

Outcome variables

Thirty-day and 1-year mortality and 30-day perioperative complications were the outcome measures. Mortality was ascertained through a combination of the VASQIP data and the VA Vital Status file.

Assessing diagnosis of Chronic Obstructive Pulmonary Disease

History of severe COPD is defined in VASQIP as “emphysema and/or chronic bronchitis resulting in any one or more of the following: functional disability (e.g., dyspnea, inability to perform activities of daily living), hospitalization in the past for treatment of COPD, need for chronic bronchodilator therapy with oral or inhaled agents, or a recorded FEV₁ of <75% of predicted on pulmonary function testing.” (excluding asthma, interstitial fibrosis and sarcoidosis). We defined a second group of subjects as having mild COPD; these subjects had COPD recorded as a diagnosis during a previous clinic visit or hospitalization but did not meet the VASQIP criteria for severe COPD. We postulated that this group would have increased perioperative risks compared to those without any previous diagnosis of COPD. It was not possible to utilize the GOLD criteria (14) in this study of administrative data for COPD severity.

Covariates

VASQIP provided the following covariates: age, gender, and race; preoperative co-morbidities; pre-fracture risk factors (smoking, alcohol use, recent significant weight loss, steroid use), key laboratory values; pre-fracture functional status (independent, partially dependent, or completely dependent); surgical Current Procedural Terminology (CPT) codes; operative times; anesthesia technique; American Society of Anesthesiologists (ASA) class. The Elixhauser co-morbidity index was calculated from ICD-9 diagnosis codes within 5 years of surgery from the VA's PTF inpatient and outpatient files (15).

Statistical analysis

We compared demographic and clinical variables using Chi-square or Fisher exact tests for categorical variables, as appropriate, and a t-test or analysis of variance for continuous variables between subjects with and without COPD. Guided by these statistical results, we constructed logistic regression models to predict outcomes of mortality and the occurrence of at least one complication. Data were compiled using the Statistical Analysis Software (SAS) (Version 9.1 SAS Corporation, Cary, North Carolina), and analyzed using Stata (release 10.2 SE, StataCorpLP, College Station, Texas).

Results

A total of 12,646 hip fracture repair records were analyzed from VASQIP from 1998 to 2005. Nearly half (47.6%) of patients with hip fractures had COPD. Approximately 26% (3,261 cases) were classified as severe COPD and an additional 2,736 (21.6%) were classified as mild COPD. The annual overall proportion of hip fracture subjects with either mild or severe COPD remained fairly constant over the study period.

Cohort characteristics and surgical details

A description of the entire cohort and the COPD subgroups is presented in Table 2. There were no significant age differences between groups.

Table 2.  Patient characteristics by COPD status** (1998–2005).

	No COPD	Mild COPD	Severe COPD	P-value*
Number of Cases (n)	6,649	2,736	3,261
Age, years (mean)1	72.8	73.7	74.4	0.11
Age <65 (% of sample)	24	21	15	0.40
Age 65–74 (% of sample)	24	26	30
Age 75–84 (% of sample)	42	44	47
Age 85+ (% of sample)	11	10	8
Race Category1
White (% of sample)	72	76	81	<0.001
Black (% of sample)	12	11	7
Hispanic (% of sample)	5	4	3
Other (% of sample)	1	0	0
Unknown (% of sample)	10	8	8
Current Smoker* (% of sample)1	24	35	43	<0.001
Frequent Alcohol* (% of sample)1	13	15	18	<0.001
Steroid Use (% of sample)1	2	2	10	<0.001
Premorbid Functional Status1
Independent (% of sample)	59	59	51	<0.001
Partial Dependence (% of sample)	32	33	40
Dependent (% of sample)	9	8	9
Other Pre-Fracture Diagnoses:
Count of Co-morbidities (Elixhauser)2	4.3	5.7	5.5	<0.001
Osteoporosis (% of sample)2	2.7	2.9	2.6	0.3
Dementia2	19	18	14	<0.001
Recent Significant weight loss1	3	4	8	<0.001
Diabetes1	22	20	18	<0.001
CHF1	4	6	11	<0.001


*P-values reflect results of Chi-Square tests of differences across categories for discrete measures or ANOVA with post-estimation tests across COPD categories for continuous measures.

** Severe COPD is identified by VASQIP interview as: “emphysema and/or chronic bronchitis resulting in any one or more of the following: functional disability (e.g., dyspnea, inability to perform activities of daily living), hospitalization in the past for treatment of COPD, need for chronic bronchodilator therapy with oral or inhaled agents, or an FEV1 of <75% of predicted on pulmonary function testing”.

Mild COPD are those subjects who had a previous clinic visit or hospitalization with an ICD-9 code for COPD but did not meet the VASQIP criteria.

1 Information derived from the VASQIP data

2 Information derived from patient medical records.

There were progressively more White subjects in the mild and severe COPD groups compared to the group without COPD. Pre-morbid functional status was significantly worse only in the severe COPD group. Both current smoking status (24%, 35%, and 43%) and frequent alcohol use (13%, 15%, 18%) were progressively more common in the mild and severe COPD groups compared to no COPD. Ten percent of the severe COPD patients used steroids for a chronic condition compared to 2% in the mild COPD and non-COPD groups. Both mild and severe COPD subjects had significantly more co-morbidities (15, 16) with mean Elixhauser scores of 5.7 and 5.5 respectively compared to 4.3 (p < 0.001) for patients without a history of COPD. Of interest, the severe COPD subjects had significantly less dementia and diabetes but more congestive heart failure (CHF) and more recent weight loss compared to the other two groups. Prior knowledge of osteoporosis in all of the groups was infrequent, with less than 3% of the subjects having a recorded diagnosis in their medical records.

Details of the surgical procedures were informative (Table 3). ASA class was incrementally higher (implying sicker patients) in both the mild and severe COPD subjects. Both mild and severe COPD subjects were significantly less likely to have general anesthesia (GA) compared to the group without COPD (No COPD - 68% had GA; mild COPD - 63% had GA; severe COPD - 55% had GA, p < 0.001). The mean operating time was significantly lower in both the mild and severe COPD subjects (p < 0.001). However, there were no differences identified between relative work units (RVU) reported per case, nor pre-op wound classifications.

Table 3.  Operative characteristics by COPD status** (1998–2005).

	No COPD	Mild COPD	Severe COPD	P-value*
Number of Cases (n)	6,649	2,736	3,261
Operating Time (mean hours)	1.6	1.5	1.5	<0.001
Work RVU's/complexity (higher -> more complex)	15.8	15.8	15.8	0.795
Wound Class (% of cases)
clean	97.6	97.7	96.7
clean contaminated	2.2	2.1	3.2	0.068
contaminated	0.1	0.1	0.1
infected	0.1	0.1	0.0
general anesthesia (vs. other method) (% of cases)	68	63	55	<0.001
Staff MD (vs. resident) (% of cases)	22	29	25	<0.001
Emergency case	19	20	23	<0.001
ASA Class
1 - Healthy patient	1	0	0	<0.001
2 - mild systemic disease	14	7	3
3 - severe systemic disease	67	66	60
4 - constant threat to life	18	26	36
5 - moribund patient	0	0	1
One or more RBC's given (%)	11.7	11.7	13.2	0.082
Operation Delayed 4+ days (%)	25.8	29.4	31.7	<0.001
Days from Admit to Surgery (avg)	5.1	5.0	4.1	0.85


*P-values reflect results of Chi-Square tests of differences across categories for discrete measures or ANOVA with post-estimation tests across COPD categories for continuous measures.

** Severe COPD is identified by VASQIP interview as: “emphysema and/or chronic bronchitis resulting in any one or more of the following: functional disability (e.g., dyspnea, inability to perform activities of daily living), hospitalization in the past for treatment of COPD, need for chronic bronchodilator therapy with oral or inhaled agents, or an FEV1 of <75% of predicted on pulmonary function testing”.

Mild COPD are those subjects who had a previous clinic visit or hospitalization with an ICD-9 code for COPD but did not meet the VASQIP criteria.

All data for Table 3 was derived from the VASQIP non-cardiac file.

Mortality and complications

Mortality at 30 days and 1-year as well as post-operative complications for the cohort are shown in Figure 1. Subjects with COPD had nearly double (6.8% to 11.7%, p < .001) the 30-day mortality compared to those without COPD; 1-year mortality rose from 24.5% to 37.6% between subjects without COPD and those with severe COPD. Perioperative complications were similarly increased in COPD subjects compared to those without. COPD subjects had more cardiopulmonary related complications –pneumonia, failure to wean, reintubation for respiratory and cardiac complications and cardiac arrest (See Table 4). Severe COPD patients had more wound infections and were significantly more likely to develop sepsis than those with mild or no COPD (1.4%, 1.3% vs. 3.3%, p < 0.001). There was no significant difference between the two groups in myocardial infarction (MI), stroke, deep vein thrombosis/pulmonary embolism (DVT/PE), major blood loss nor peripheral nerve injury (data not shown).

Figure 1.  Outcomes of hip fractures in men. Three major outcomes of hip fractures in men treated within the VAH system were studied in detail: 30 day mortality, one or more perioperative complications and 1 year mortality. Unadjusted percentages for 30 day mortality (A), complications (B) and 1 year mortality (C) are shown. Severe COPD (that required chronic medication or impacted function) was associated with greater mortality and more complications (p<0.0001 for each outcome in both the unadjusted comparisons and the adjusted logistic regression model). In adjusted models there was no significant difference for mild COPD compared to no COPD. Unadjusted comparisons of mild COPD compared to no COPD indicated no significant difference in the 30-day rates of mortality or complications, but a higher 1-year mortality rate for patients with mild COPD (t-test p<0.001).

Table 4.  Complications by COPD status** (1998–2005).

	No COPD	Mild COPD	Severe COPD	P-value*
Number of Cases (n)	6,649	2,736	3,261
Reintubation (% of cases)	2.4	2.9	4.9	<0.001
Failure to Wean (% of cases)	1.3	1.35	3.4	<0.001
Pneumonia (% of cases)	5.5	5.7	9.4	<0.001
Wound Infection (% of cases)	2.4	2.3	3.3	0.009
Sepsis (% of cases)	1.4	1.6	2.8	<0.001
Bleeding requiring >4 units transfusion (% of cases)	0.66	0.44	0.58	0.44
DVT (% of cases)	0.9	1.2	0.6	0.05
CVA (% of cases)	0.5	0.6	0.6	0.85
Cardiac arrest (% of cases)	1.7	1.7	2.4	0.027
MI (% of cases)	0.99	0.91	1.38	0.14


*P-values reflect results of Chi-Square tests of differences across COPD categories.

** Severe COPD is identified by VASQIP interview as: “emphysema and/or chronic bronchitis resulting in any one or more of the following: functional disability (e.g., dyspnea, inability to perform activities of daily living), hospitalization in the past for treatment of COPD, need for chronic bronchodilator therapy with oral or inhaled agents, or an FEV1 of <75% of predicted on pulmonary function testing”.

Mild COPD are those subjects who had a previous clinic visit or hospitalization with an ICD-9 code for COPD but did not meet the VASQIP criteria.

Presence of these complications within 30-days of surgery was derived from the VASQIP non-cardiac data file.

The logistic regression models (see Table 5) confirmed that severe COPD vs. no COPD was a significant risk factor for 30-day and 1-year mortality after adjusting for the covariates shown. C-statistics for these models suggested good model fit. Age, Caucasian race, steroid use, dementia, lack of functional independence, use of general anesthesia, ASA class 3 and 4 and recent weight loss were also significant predictors of 30-day and 1 year mortality. A delay in surgery of greater than 4 days was strongly associated with 30-day mortality, 1 year mortality and increased complications. Early discharge after surgery and heavy alcohol use were associated with reduced odds ratios for mortality. Dementia, steroid use and co morbidity count were significantly related to 30-day and 1-year mortality. Variables predicting 1-year mortality were similar to 30-day, except that current smoking, race, early discharge and use of general anesthesia were not predictive. Interestingly, current smoking was only significant as a predictor of 30-daymortality but not complications or 1-year mortality.

Table 5.  Logistic regression models predicting outcomes (entire sample).

	Mortality within 30 Days		Mortality within 1 year		Complication(s) within 30 Days
	Odds Ratio	p-value*	Odds Ratio	p-value*	Odds Ratio	p-value*
Mild COPD (vs no COPD)	0.95	0.59	0.98	0.80	0.95	0.40
Severe COPD (vs no COPD)	1.34	<0.001	1.43	<0.001	1.25	<0.001
Caucasian Race (vs. other)	1.26	0.01	1.07	0.20	1.26	<0.001
Age 65–74 (vs. under age 65)	1.43	0.01	1.49	<0.001	1.54	<0.001
Age 75–84 (vs. under age 65)	2.26	<0.001	2.18	<0.001	2.11	<0.001
Age 85+ (vs. under age 65)	4.14	<0.001	3.89	<0.001	2.48	<0.001
Current smoker (vs. non-smoker)	1.22	0.03	1.05	0.39	1.00	0.99
Heavy Alcohol use (vs. not)	0.83	0.15	0.85	0.03	1.25	<0.001
Steroid use for Chronic condition (vs. not)	1.47	0.01	1.61	<0.001	1.00	0.97
Dementia (vs. not)1	1.47	<0.001	1.64	<0.001	1.04	0.53
Osteoporosis diagnosis (vs. not)1	0.73	0.24	0.80	0.15	0.66	0.03
Recent Significant Weight Loss (vs. not)	1.93	<0.001	2.45	<0.001	1.30	0.01
Elevated Hematocrit	1.46	<0.001	1.50	<0.001	1.18	0.01
Elevated Sodium	3.43	<0.001	1.76	0.06	1.31	0.38
Poor Renal Function	1.41	<0.001	1.30	<0.001	1.25	<0.001
Partial Functional Dependence (vs. independent)	1.55	<0.001	1.34	<0.001	1.35	<0.001
Functional Dependence (vs. independent)	2.04	<0.001	1.99	<0.001	1.58	<0.001
History of CHF	1.71	<0.001	1.18	0.06	1.31	<0.001
History of Stroke	0.84	0.26	0.94	0.52	1.00	0.98
Count of non-COPD Co-morbidities (Elixhauser)1	1.03	0.03	1.08	<0.001	1.04	<0.001
General anesthesia (vs. other method)	1.22	0.01	0.96	0.33	1.35	<0.001
ASA class 3 (vs. 1 or 2)	1.75	0.01	1.40	<0.001	1.51	<0.001
ASA class 4 (vs. 1 or 2)	3.11	<0.001	2.21	<0.001	1.99	<0.001
Emergency Admission	1.14	0.14	1.01	0.82	1.09	0.18
Attending Surgeon (vs. resident)	0.97	0.73	0.91	0.09	1.01	0.80
Intraoperative Blood transfusions	1.07	0.06	1.08	0.01	1.18	<0.001
Early alive discharge (<2 days after surgery)	0.42	<0.001	0.88	0.17	0.58	<0.001
Operation 4 or more days from admission	1.33	<0.001	1.28	<0.001	1.12	0.04
CPT code 27125 (vs. 27130)	1.12	0.66	1.23	0.19	1.18	0.32
CPT code 27235 (vs. 27130)	0.89	0.68	1.08	0.64	0.82	0.29
CPT code 27236 (vs. 27130)	1.26	0.36	1.34	0.05	1.12	0.49
CPT code 27244 (vs. 27130)	1.12	0.65	1.16	0.32	1.02	0.91
CPT code 27245 (vs. 27130)	1.07	0.82	1.28	0.13	1.10	0.58
C-statistic	0.76		0.74		0.68
Number of Observations	11,798		11,798		11,724


*p-values are based on 95% confidence intervals for odds ratios. Standard errors were calculated using robust regression.

** “Severe” COPD and “Mild” COPD are defined in text but are not based on GOLD criteria. Bolded values highlight variables that are significant with a p-value <0.05.

1 Variable derived from the patient medical record (vs. VASQIP).

Perioperative complications showed many of the same predictive factors as the mortality models. Severe COPD was a significant predictor of complications within 30 days after adjusting for other variables. Age, race, ASA class, functional pre-morbid status, recent weight loss, history of congestive heart failure and the co-morbidities count were also important predictors for post operative complications in the model. Heavy alcohol use and choice of general anesthesia were both associated with peri-operative complications.

Discussion

In this analysis of 12,646 men with hip fractures treated surgically within the VA hospital system we found that COPD was strongly associated with increased mortality after hip fracture surgery. The association of COPD with death persisted after adjusting for other factors including lower baseline physical function. Males with hip fracture have previously been reported to have worse outcomes than females (21–23), and our cohort's 1-year mortality rates (24–38%) are similar to those previously reported for men. This study confirms previous findings that COPD is a major factor predicting mortality and complications.

Beyond the patient characteristics of age, co-morbid disease and worse ASA class, important predictors of hip fracture outcomes are: current smoking, pre-morbid functional status and pre-existing dementia. These patient characteristics are not modifiable in the perioperative period. However, after adjustment for these factors, the effect of operative delays greater than 4 days on both short term and long term mortality remained significant. Likewise, use of general anesthesia remained a predictor of short term mortality and complications after adjustment, and 55% of severe COPD subjects had a general anesthetic. Thus, operative delays and anesthetic choice appeared to represent opportunities for improved management of hip fracture patients and especially those with COPD.

Reasons for surgical delays included acute medical problems that require treatment and lack of operating room/staff availability. Recent work suggests that lack of standardization in preparing patients for surgery plays a role in the variation related to delays from medical problems (17). Quality improvement projects and further research are needed to establish critical factors for hip fracture patients that must be rapidly corrected (severe anemia, untreated infections and metabolic abnormalities) and those which cannot be corrected and must be accepted and managed during the perioperative period (underlying heart or lung disease). Addressing these issues may allow guidelines to be established for ensuring that patients are treated quickly and appropriately. Individual hospitals should monitor and address the delays related to lack of timely operating room capacity for hip fractures.

A critical difference between the COPD and non-COPD groups was in anesthesia choice. General anesthesia remains a significant predictor of increased mortality and greater perioperative complications in our adjusted models. The non-COPD group was much more likely to have received general anesthesia (68% compared to 55%). Previous work from the VASQIP project had already suggested that general anesthesia was associated with higher mortality and complication rates in hip fractures, which may be reflected in the lower rates we see in the COPD patients (6).

In the United Kingdom more than 70% of hip fractures surgeries are performed with spinal or other regional anesthesia techniques (18) and a study from Spain reports greater than 90% of hip fracture patients receiving regional anesthesia (17), suggesting randomized trials of anesthetic choice for hip fracture patients with and without COPD would be helpful to determine if decreased use of general anesthesia would improve hip fracture outcomes.

Classifying VA hip fracture cases with COPD as mild or severe is an innovation of this study. The mild subjects were identified from within the “control group” of subjects without clearcut COPD. We found that mild disease that has not resulted in functional limitations or required chronic medication is still associated with more co-morbid conditions and non-significant increases in mortality and complications after adjusting for other covariates. Because our analysis was based on administrative data we are not able to define the degree of obstruction in these patients but rather rely on coded diagnoses where presumably the physician was aware of the diagnosis.

The group of mild COPD may represent early or mild disease that may not be clearly characterized in a usual or standard pre-op risk assessment. More widespread screening of former smokers might improve risk assessment for COPD and lead to greater selection of spinal anesthesia, good post-operative pulmonary toilet and other measures to prevent complications in this group. Although up to half of the population in the United States (and an even larger percentage of veterans) are current or former smokers with potential risk of COPD (19, 20), screening for occult lung disease lags behind screening for cardiovascular disease (21, 22).

Primary prevention of deaths from hip fracture may be effected by early identification of osteoporosis in high-risk groups — because treatment of osteoporosis could reduce the risk of and complications following hip fractures. Osteoporosis was infrequently (<3%) identified prior to hip fracture in this population. We would advocate for more intensive screening and treatment for patients with current smoking history or significant past history of tobacco use, as well as in known or suspected COPD.

Diagnosis and treatment of osteoporosis has been shown to reduce fracture risk by up to 25% (23). Several studies have shown high rates of osteoporosis in COPD subjects, including men (24–26). Men with COPD have high rates of fractures, but are not commonly screened for osteoporosis (27). Other recent studies suggest that diagnosis and treatment for osteoporosis in COPD patients is poor (28), making them an important group for prevention strategies.

Strengths of this study include the large number of subjects and the extensive perioperative data available in the VASQIP data collection that allowed more detailed understanding of surgical details than is typically offered in observational studies. The study was limited to men because of the small numbers of female VA patients identified with hip fracture surgeries. However, since other studies have found that men typically have worse outcomes after hip fracture, the sample for this study may present the best opportunity for improving outcomes in this understudied group. A longitudinal study of hip fractures in New York City during a similar time frame reported only 8.5% of their subjects with COPD and a marginally significant hazard ratio of COPD for mortality (29). Our larger sample size and higher prevalence of disease may have improved our ability to detect this effect in our cohort.

Our findings indicate several opportunities for future research. The presence and impact of pre-existing lung disease in hip fracture subjects needs further exploration in a prospective study. Formally testing the impact of spinal and general anesthesia on mortality and the incidence of pulmonary and infectious complications in a randomized clinical trial would address what is now a largely empiric choice by surgeons and anesthesiologists. Hip fracture is a sentinel event that poses many subsequent health challenges for patients. Results of our study indicate that hip fracture patients with COPD face even more challenges to recovery.

Acknowledgments

The authors also acknowledge the VA Surgical Quality Data Use Group (SQDUG) for its role as scientific advisors and for the critical review of data use and analysis presented in this manuscript. The opinions expressed are those of the authors and not necessarily those of the Department of Veterans Affairs or the United States Government.

Declaration of Interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

This research was supported by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development Service through the Research Enhancement Award Program and an investigator-initiated grant (IIR 04–173). Drs. Radcliff, Cowper Ripley, and Hutt received salary support from the VA.