Abstract
Rationale: Randomized trials have shown that noninvasive ventilation (NIV) can reduce the need for intubation and improve the survival of patients with severe exacerbations of chronic obstructive pulmonary disease (COPD); however, it is not known whether hospitals with greater use of NIV achieve lower rates of intubation and better patient outcomes.
Objectives: To describe patterns of mechanical ventilation use for patients with COPD across a large sample of hospitals, and to analyze the relationship between use of NIV and other outcomes.
Methods: Cross-sectional analysis of 77,576 patients hospitalized for COPD between June 2009 and June 2011 at 386 U.S. hospitals.
Measurements and Main Results: Using hierarchical modeling, we estimated hospital risk-standardized percentages of ventilator starts that were noninvasive (RS-NIV%). We examined the association between RS-NIV% and other outcomes, including risk-standardized rates of invasive ventilation and NIV failure, total ventilation, in-hospital mortality, length of stay, and costs. At the hospital level, the median RS-NIV% was 75.1% (range: 9.2–94.1%). Smaller hospitals and those located in rural areas had higher RS-NIV%. When stratified into quartiles on the basis of the RS-NIV%, hospitals in the highest quartile had lower risk-standardized rates of invasive mechanical ventilation (Q4 vs. Q1: 4.0% vs. 13.3%, P < 0.01) and modestly higher risk-standardized total rates of ventilation (Q4 vs. Q1: 23.9% vs. 22.0%, P = 0.03). Hospitals with the highest RS-NIV% had lower risk-standardized mortality among ventilated patients who received ventilation (Q4 vs. Q1: 8.5% vs. 9.0%, P = 0.01) and marginally lower mortality rates among all patients with COPD (Q4 vs. Q1: 2.2% vs. 2.3%, P = 0.03) compared with hospitals with the lowest RS-NIV%. Higher RS-NIV% was associated with lower hospital costs (Q4 vs. Q1: $11,148 vs. $14,032, P < 0.001), shorter length of stay (Q4 vs. Q1: 5.5 vs. 6.8 d, P < 0.001), and lower NIV failure rates (Q4 vs. Q1: 12.8 vs. 32.5%, P < 0.001).
Conclusions: Use of NIV as the initial ventilation strategy for patients with COPD varies considerably across hospitals. Institutions with greater use of NIV have lower rates of invasive mechanical ventilation and better patient outcomes.
Keywords: COPD, costs and cost analysis, cross-sectional analysis, length of stay, outcomes research
Chronic obstructive pulmonary disease (COPD) affects as many as 24 million individuals in the United States, is responsible for more than 800,000 hospitalizations annually, and is the nation’s third leading cause of death (1–3). Inpatient mortality has been estimated at approximately 3% (4, 5); the 30-day mortality rate among Medicare beneficiaries is 8.6%; and approximately one in five patients are rehospitalized within 30 days of discharge (6, 7). COPD accounts for one-fourth of all cases of respiratory failure, following pneumonia and congestive heart failure (8).
Standard therapy for those requiring hospitalization includes supplemental oxygen, short-acting bronchodilators, systemic corticosteroids, and, frequently, broad-spectrum antibiotics (9). Beginning in the 1990s, a series of small, randomized trials demonstrated that, when applied to carefully selected patients with COPD with severe exacerbation, noninvasive ventilation (NIV) could reduce the need for invasive mechanical ventilation (IMV) and resulted in better short-term survival (10–13). In, meta-analyses, authors have estimated that, as an alternative to IMV, treatment of these patients with NIV is associated with a 40–55% reduction in mortality (14, 15).
Although use of NIV in the management of patients with COPD and other causes of respiratory failure has increased steadily over the past decade (8, 16, 17), little is known about the uniformity of its adoption across hospitals or about the association between hospital patterns of mechanical ventilation and other hospital outcomes. Using data from a large sample of U.S. hospitals, we studied the use of IMV and NIV in the management of patients hospitalized for COPD. On the basis of the results of randomized trials, we hypothesized that NIV would be used primarily as an alternative to IMV and that hospitals with higher rates of NIV would have better patient outcomes, including mortality, length of stay, and costs.
Methods
Design, Setting, and Subjects
We conducted a cross-sectional study of patients hospitalized for exacerbations of COPD between June 2009 and June 2011 at a geographically and structurally diverse set of 386 U.S. hospitals that participate in a voluntary, fee-supported database used for quality improvement (Premier Healthcare Informatics, Charlotte, NC). Data are collected electronically from these sites, are audited regularly to ensure validity, and the database has been used extensively in outcomes research (18–20). In addition to the information contained in the standard hospital discharge abstract (i.e., UB-04), the database contains a date-stamped log of all items and services charged to patients or their insurers, including medications, laboratory and other diagnostic tests, and therapeutic services, such as respiratory and physical therapy.
Patients were included in the analysis if they were 40 years of age or older, received a principal discharge diagnosis (coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM]) consistent with acute exacerbation of COPD (491.21, 492.22, 491.8, 491.9, 492.8, or 496) or a secondary diagnosis of COPD when accompanied by a principal diagnosis of acute respiratory failure (518.81, 518.82, or 518.84) (21), and were treated with short-acting bronchodilators and systemic corticosteroids. We excluded patients if they were transferred from or to another acute care facility, if their length of stay was less than 2 days, if they were cared for by an attending physician whose specialty would not be expected to care for patients with COPD (e.g., psychiatry), or if they were classified in a Medicare severity diagnosis-related group inconsistent with COPD. We also excluded patients with sleep apnea because we could not differentiate chronic use of NIV (i.e., positive airway pressure devices) from treatment for acute respiratory failure. Finally, to obtain stable estimates of mechanical ventilation rates, we excluded patients cared for at hospitals with fewer than 20 eligible COPD admissions during the 2-year study period.
Patient, Physician, and Hospital Information
In addition to patient age, sex, race, and ethnicity, we computed a comorbidity score using diagnosis codes as described by Gagne and colleagues (22). To better assess the severity of the patient’s lung disease and its impact on health, we calculated the number of hospitalizations for COPD in the year prior to the index admission. Given the high incidence of pneumonia and its association with NIV failure, we also recorded whether the patient received a diagnosis of pneumonia at the time of hospital admission by using present on admission indicator codes (23).
For each hospital, we recorded the number of beds, the annual number of admissions for COPD, teaching status, geographic region, and whether the hospital served an urban or rural population. Using data from the American Hospital Association annual survey, we computed the ratios of full-time equivalent hospitalists, intensivists, nurses, and respiratory therapists to beds for each hospital (24).
Noninvasive and Invasive Ventilation
For each patient, we examined standardized charge codes generated daily by respiratory therapists, as well as dated ICD-9-CM procedure codes, to determine whether or not the patient was treated with assisted ventilation and, if so, whether it was NIV or IMV. We considered a patient to have received ventilation if either the respiratory therapist charge or ICD-9-CM procedure code (93.90 for NIV; 96.04, 96.70–96.72 for IMV) indicated treatment. We defined the primary method of ventilation as the first method by date, and we noted any changes that occurred throughout the hospitalization. In those cases where both NIV and IMV were initially recorded on the same date, we used service charges on subsequent days to infer the initial therapy. We validated the NIV procedure codes and respiratory therapy charge codes by retrospective medical chart review of 200 patients who had been admitted to Baystate Medical Center with an ICD-9-CM code for respiratory failure at any time from 2010 through 2011. Using ICD-9-CM codes alone yielded a sensitivity of 86% (95% confidence interval [CI], 81–92%) and specificity of 92% (95% CI, 84–98%) for NIV. The approach of using ICD-9-CM procedure codes and/or respiratory therapist charges increased sensitivity to 99% (95% CI, 98–100%) without reducing specificity (92%; 95% CI, 84–99%).
Outcomes
The primary study outcomes were the hospital rate of IMV, total mechanical ventilation, NIV failure, in-hospital mortality, median length of stay, and median costs. The NIV failure rate was defined as the percentage of patients who were treated with IMV after an initial exposure to NIV.
Analysis
For each hospital, we calculated the percentage of patients treated according to each of the primary ventilatory strategies: no assisted ventilation, NIV, and IMV. We then calculated the percentage of patients initially treated with NIV among those who received assisted ventilation. We used hierarchical logistic regression with a random hospital effect to model initial use of NIV among patients started on ventilation, adjusting for demographics, comorbidity score, prior COPD admissions, and the presence of comorbid pneumonia (25). From the model, a predicted NIV percentage for each hospital was computed as the NIV percentage that would be anticipated at a particular hospital by using its hospital random effect, given the patient case mix.
The expected NIV percentage was computed as the rate that would be expected if the same patient mix were treated at an “average” hospital, using the average hospital effect. A hospital risk-standardized NIV percentage was computed as the ratio of predicted to expected NIV percentage standardized by the overall unadjusted mean NIV percentage for all admissions in our model. We used analogous methods to compute risk-standardized rates of IMV, total ventilation, NIV failure, and mortality. For patients with multiple admissions during the study period, mortality models were restricted to a randomly selected admission to address the threat of survival bias associated with high admission rates. We evaluated the Spearman correlation between the hospital risk-standardized NIV percentage and risk-standardized total IMV rates. We then evaluated variation in hospital characteristics across quartiles of risk-standardized NIV percentage using Kruskal-Wallis tests. Finally, we compared risk-standardized rates of IMV, total ventilation, NIV failure, and mortality, as well as hospital median length of stay and hospital median costs, across quartiles based on the risk-standardized NIV percentage by using Kruskal-Wallis analysis of variance, adjusting for hospital characteristics.
All analyses were performed using SAS version 9.3 software (SAS Institute, Cary, NC). The Institutional Review Board at Baystate Medical Center approved the study and waived the requirement for explicit participant consent.
Results
Patient and Hospital Characteristics
We identified 102,654 admissions among 77,576 patients for COPD between June 2009 and June 2011 at 386 U.S. hospitals (Figure 1). The patients’ median age was 70 years, and 58.6% were female. Heart failure, anemia, and diabetes were the most common comorbidities, and the median comorbidity score was 2 (range, 1–4). A total of 33,245 (32.4%) had at least one admission for COPD in the year prior to the index admission, and pneumonia was present at the time of admission in 18.9% of cases. The median length of stay in the hospital was 4 days, and 2,023 (2.0%) of admissions resulted in death before discharge. Approximately 38% of hospitals included in the study operated fewer than 200 beds, and 24% had more than 400 beds, 76% served an urban population, and 25% were engaged in house staff training.
Figure 1.
Study cohort flowchart. COPD = chronic obstructive pulmonary disease; DRG = Medicare Severity-Diagnosis Related Group; ICD-9 = International Classification of Diseases, Ninth Revision, Clinical Modification.
Use of Noninvasive and Invasive Ventilation
A total of 15,448 (15.1%) patients admitted for exacerbations of COPD were initially treated with NIV, and 5,824 (5.7%) received initial treatment with IMV. In 2,339 (15.5%) of the NIV-treated admissions, patients were later treated with IMV (due to NIV failure). Additionally, in 1,631 (28.0%) admissions, patients initially treated with IMV were later treated with NIV.
We observed substantial variation in the risk-standardized NIV percentage across the hospitals included in the analysis, varying from 47.4% at the 10th percentile to 84.7% at the 90th percentile (Figure 2). We found modest association between the hospital risk-standardized NIV percentage and other hospital characteristics. Hospitals that were smaller, had lower annual COPD case volumes, and those located in rural areas had slightly higher risk-standardized NIV percentages; however, physician, nurse, and respiratory therapist staffing ratios were not associated with ventilation patterns (Table 1).
Figure 2.
Distribution of percentage of ventilator starts that were noninvasive at hospitals included in the analysis. NIV = noninvasive ventilation.
Table 1.
Risk-standardized percentage of ventilator starts that were noninvasive by characteristics of hospitals included in the analysis
| Number | Risk-standardized NIV percentage, Median % (Q1–Q4) | P Value* | |
|---|---|---|---|
| Overall |
386 |
75.1 (66.1–80.7) |
|
| Hospital characteristics | |||
| Teaching status | 0.143 | ||
| Nonteaching | 289 | 75.3 (66.9–81.0) | |
| Teaching | 97 | 73.3 (63.4–79.7) | |
| Rural/urban status | 0.0037 | ||
| Rural | 94 | 77.8 (70.1–81.8) | |
| Urban | 292 | 74.2 (63.7–79.9) | |
| Region | 0.073 | ||
| Midwest | 71 | 74.8 (63.8–81.3) | |
| Northeast | 56 | 72.3 (64.2–77.7) | |
| South | 177 | 76.2 (67.0–80.9) | |
| West | 82 | 76.7 (66.1–81.7) | |
| Beds | 0.064 | ||
| Small, <200 | 148 | 76.7 (67.1–81.6) | |
| Medium, 200–400 | 146 | 74.8 (64.9–80.1) | |
| Large, >400 | 92 | 74.0 (63.2–79.5) | |
| Annual COPD case volume | 0.063 | ||
| 10–64 admissions | 96 | 77.4 (71.1–81.4) | |
| 65–117 | 97 | 73.2 (63.8–79.9) | |
| 118–184 | 95 | 74.2 (65.3–80.5) | |
| 185+ | 98 | 74.5 (62.8–79.7) | |
| Staffing ratios† | |||
| Hospitalists | 0.127 | ||
| Missing | 134 | 73.5 (61.6–79.9) | |
| <1.42 | 86 | 74.9 (66.3–80.7) | |
| 1.42–3.65 | 85 | 76.2 (68.5–79.6) | |
| 3.66+ | 81 | 77.1 (68.9–82.6) | |
| Respiratory therapists | 0.629 | ||
| Missing | 102 | 74.5 (62.8–80.1) | |
| <5.36 | 90 | 75.3 (65.1–79.6) | |
| 5.36–8.45 | 93 | 77.0 (66.9–81.4) | |
| 8.46+ | 103 | 74.6 (66.9–81.0) | |
| Nurses | 0.860 | ||
| Missing | 134 | 74.7 (63.5–80.2) | |
| <107.68 | 80 | 75.9 (67.7–80.7) | |
| 107.68–150.47 | 82 | 76.0 (66.3–82.1) | |
| 150.48+ | 90 | 74.9 (63.6–80.6) | |
| Intensivists | 0.0056 | ||
| Missing | 140 | 72.2 (59.0–79.7) | |
| None | 124 | 76.4 (69.0–81.3) | |
| >0–1.91 | 64 | 74.0 (59.4–80.1) | |
| 1.92+ | 58 | 77.4 (71.6–81.4) |
Definition of abbreviations: COPD = chronic obstructive pulmonary disease; NIV = noninvasive ventilation.
Calculated using the Kruskal-Wallis test.
Each staff member is FTE to 100 beds.
Association between Use of Noninvasive Ventilation and Outcomes
We observed a strong negative correlation (−0.70) between the risk-standardized percentage of ventilator starts that were noninvasive and risk-standardized rates of IMV. (Figure 3). When grouped into quartiles, the median percentage of ventilator starts that were noninvasive ranged from 54.5% in the first quartile (Q1) to 83.7% in Q4. We observed strong associations between the risk-standardized NIV percentage and other hospital outcomes (Figure 4 and Table 2). As the hospital risk-standardized NIV percentage increased, there was a large decline in the risk-standardized rate of IMV (rate in Q1 vs. Q4: 13.3% vs. 4.0%, P < 0.01) and a modest increase in the risk-standardized total rate of mechanical ventilation (rate in Q1 vs. Q4: 22.0% vs. 23.9%, P = 0.03). Across quartiles of risk-standardized NIV percentage, the overall percentage of patients experiencing NIV failure remained relatively constant; however, when viewed as the percentage of patients initially treated with NIV, the rate of failure declined dramatically (rate in Q1 vs. Q4: 32.5% vs. 12.8%, P < 0.01) (Figure 4 and Table 2). A higher percentage of ventilator starts that were noninvasive was associated with reduced risk-standardized mortality among patients who received ventilation (rate in Q1 vs. Q4: 9.0% vs. 8.5%, P = 0.01). We found that hospitals with the highest risk-standardized NIV percentage had costs for patients who received ventilation that were approximately $3,000 lower and length of stay that was 1 day shorter (costs in Q1 vs. Q4: $14,032 vs. $11,148 and 6.8 vs. 5.5, respectively; P < 0.001 for both).
Figure 3.
Correlation between risk-standardized percentage of ventilator starts that were noninvasive and risk-standardized rates of invasive ventilation. NIV = noninvasive ventilation.
Figure 4.
Use of invasive and noninvasive ventilation (NIV) by quartile of risk-standardized NIV percentage. IMV = invasive mechanical ventilation.
Table 2.
Hospital risk-standardized rates of mortality, length of stay, and costs and ventilation by quartiles of risk-standardized percentage of ventilated patients started on noninvasive ventilation
| Risk-Standardized Hospital Rates | Quartiles of Risk-Standardized NIV Percentage |
P Value* | Q4 vs. Q1 P Value* | |||
|---|---|---|---|---|---|---|
| Q1, <66.1% | Q2, 66.1–75.0% | Q3, 75.1–80.7% | Q4, >80.7% | |||
| Any ventilation | 22.0 | 21.6 | 22.3 | 23.9 | 0.121 | 0.026 |
| Primary invasive ventilation | 13.3 | 6.8 | 5.1 | 4.0 | <0.0001 | <0.0001 |
| Any invasive ventilation | 16.1 | 9.8 | 7.5 | 6.4 | <0.0001 | <0.0001 |
| NIV failure | 32.5 | 20.2 | 15.2 | 12.8 | <0.0001 | <0.0001 |
| Mortality among patients who received ventilation | 8.96 | 9.02 | 8.46 | 8.48 | 0.0009 | 0.006 |
| Mortality among all patients | 2.34 | 2.31 | 2.21 | 2.22 | 0.038 | 0.034 |
| Length of stay (median days) among patients who received ventilation | 6.8 | 6.0 | 5.8 | 5.5 | 0.0007 | <0.0001 |
| Length of stay (median days) among all patients | 4.1 | 4.1 | 4.1 | 4.0 | 0.400 | 0.176 |
| Cost (median $) among patients who received ventilation | $14,032 | $13,295 | $10,985 | $11,148 | <0.0001 | <0.0001 |
| Cost (median $) among all patients | $7,169 | $7,435 | $6,892 | $6,653 | 0.002 | 0.044 |
Definition of abbreviations: NIV = noninvasive ventilation.
Kruskal-Wallis P value adjusted for hospital region, size and urban/rural location, teaching status, staffing ratios, and annual COPD patient volume.
In a secondary analysis that included all patients with COPD, and was thus not limited to those requiring ventilation, we found that hospitals with the highest risk-standardized NIV percentage had marginally lower risk-standardized mortality (rate in Q1 vs. Q4: 2.3% vs. 2.2%, P = 0.03) (Table 2). Hospitals with the highest risk-standardized NIV percentage also had lower median costs (median costs in Q1 vs. Q4: $7,169 vs. $6,653, P = 0.04) but similar length of stay compared with hospitals in the lowest quartile.
Discussion
In this large observational study, we found that the use of NIV among patients hospitalized with COPD varied widely across hospitals. The percentage of ventilator starts that were noninvasive ranged from 9% to 94%, even after adjusting for differences in patient case mix. This suggests that whether a patient with a severe exacerbation of COPD will be intubated or receive NIV is highly dependent on which hospital they are admitted to. When compared with hospitals with a lower percentage of NIV starts, those with a higher percentage had substantially lower rates of IMV and similar rates of NIV failure. Further, hospitals with higher rates of NIV use had modestly lower mortality, lower costs, and shorter length of stay among patients who received ventilation.
Numerous factors are probably responsible for the variation we observed in the use and outcomes of NIV across hospitals. First, providers must be familiar with the selection of appropriate patients, as well as with proper application, to achieve optimal outcomes. Successful implementation of NIV also requires trained personnel, appropriate technology, and interdisciplinary coordination. A survey of hospitals in Massachusetts and Rhode Island suggested that factors that may lead to variation in use of NIV across hospitals include lack of knowledge, inadequate training of staff, and insufficient equipment (26). If NIV is viewed as similar to other complex sociotechnical undertakings, such as ensuring that patients with acute coronary syndrome are brought to the cardiac catheterization laboratory in a timely manner, then the variation we observed should not be surprising.
The most significant finding of our study was the relationship that we observed between the use of NIV and other outcomes, including intubation rates, mortality, and costs. We hypothesized that NIV would be used primarily as an alternative to intubation. Our finding that greater use of NIV was associated with lower rates of IMV supports this view. For example, compared with hospitals in the lowest quartile of NIV use, those in the highest quartile intubated 30 fewer patients per 100 admissions requiring mechanical ventilation. However, this was not a foregone conclusion. An alternative possibility was that greater use of NIV might have reflected a lower threshold for initiating assisted ventilation among patients with exacerbations of COPD. Had that been the case, we should have found that hospitals with higher NIV use had similar rates of IMV and higher overall rates of ventilation. Although we observed modestly higher rates of ventilation at the hospitals with the greatest use of NIV, almost all of the increase in NIV was offset by a decrease in IMV. It is important to emphasize, though, that using NIV as an alternative to IMV does not mean that it can be used as a substitute. Typically, NIV is used early in the course of respiratory failure to avert the need for intubation, but use of NIV should not delay a needed intubation.
Our finding that higher NIV use is associated with lower risk-standardized mortality rates, as well as shorter length of stay, in the hospital suggests that the benefits observed in clinical trials are translating into better outcomes for patients. Interestingly, the NIV failure rate remained stable in the face of higher NIV rates and declined as a percentage of all noninvasive starts. Although our study was not designed to address this directly, it is possible that as hospitals become more experienced with using NIV, they may become better at selecting patients likely to tolerate it and be treated successfully. Similarly, hospitals may develop the skills and confidence that enable them to continue the use of NIV in a patient whose treatment might be considered a failure and switched to IMV at another site without increasing the mortality rate.
Our findings should be interpreted in light of several recent studies. Chandra and colleagues reported on trends in the use of invasive and NIV based on 1998–2008 data from the Nationwide Inpatient Sample (16). They found a dramatic increase in the use of NIV over the period, with a concomitant reduction in the use of IMV and improved survival. Our analysis, conducted at the hospital level, corroborates those aggregate findings by demonstrating that institutions with higher use of NIV have lower rates of intubation. At the same time, our analysis identified wide variation in the use of NIV that suggests opportunities for quality improvement. Using the Nationwide Emergency Department Sample, Tsai and colleagues compared the effectiveness of invasive and noninvasive forms of ventilation in patients with COPD and respiratory failure (27). They reported that rates of NIV varied from 0 to 100% across hospitals included in the sample and found that hospitals in the Northeast, hospitals in nonmetropolitan areas, and hospitals with higher annual case volumes had higher rates of NIV. Our results extend the findings of that study by suggesting that hospitals with a preference for NIV achieve better outcomes, including lower rates of intubation, mortality, and costs. The 2008 UK national audit of COPD care (28) included data from 9,716 patients at 232 hospitals. It found that 12% of hospitalized patients received ventilatory support, including 11% treated with NIV and only 1% with IMV. Our results suggest that the provision of ventilatory assistance is far more common in the United States than in the United Kingdom, whereas use of NIV, as a proportion of all ventilator starts, lags our UK counterparts. Interestingly, whereas NIV was reportedly unsuccessful in 27% of cases in the UK audit, only 2% of those patients went on to receive intubation, a far lower percentage than the 15% found in our analysis.
Making causal inferences about the relationship between hospital patterns of ventilation and patient outcomes is challenging. On the one hand, a hospital level analysis is appealing because it seemingly overcomes serious concern about confounding by indication that arises in patient-level comparisons of individuals treated with invasive and noninvasive forms of ventilation. To invalidate the findings of a hospital level analysis, one would have to posit large and unobservable differences in the severity of illness of patients with COPD at some hospitals compared with others. On the other hand, hospital-level analyses could be subject to a stage migration problem whereby a reduction in the threshold to initiate NIV would reduce the average severity of illness of the ventilated cohort. Our observation of proportionately lower rates of IMV at hospitals with the strongest preference for NIV reduces this threat. In addition, our secondary analyses that included all patients with COPD, and was not limited to those undergoing ventilation, provides another check against this form of bias. In those analyses, we found smaller, though statistically significant, benefits associated with a preference for NIV.
Our study has a number of strengths. We included a large number of patients cared for within a geographically and structurally diverse set of U.S. hospitals. We used hierarchical modeling to estimate risk-standardized rates of IMV and NIV, as well as mortality, allowing us to make comparisons across hospitals that took into account differences in case mix and in hospital structural characteristics. We grouped hospitals according to the percentage of ventilator starts that were NIV, overcoming the bias that could be introduced at hospitals with low overall rates of ventilation but with a strong preference for NIV.
Although our results were striking, our findings should be interpreted in light of several limitations. First, we used ICD-9-CM codes to identify patients with COPD, and these codes do not have perfect accuracy (29). However, we took advantage of hospital pharmacy data to limit the analysis to patients with COPD who were treated with bronchodilators and systemic steroids. Second, although our analyses were adjusted for differences in patient demographics, comorbidities, and prior hospitalizations, including receipt of mechanical ventilation, we did not have direct physiologic measures to assess disease severity. Thus, some of the variation in risk-standardized rates across hospitals may reflect residual unmeasured factors. Third, although we adjusted for differences in patient and hospital characteristics across the hospitals, stratified by the percentage of ventilator starts that were noninvasive, our findings could reflect confounding by other hospital factors. Fourth, our analyses were not intended to define an optimal percentage of NIV starts, although when we stratified hospitals into quartiles on the basis of NIV use, higher rates were associated with better outcomes without increased rate of NIV failure. Fifth, our study was limited to outcomes occurring during hospitalization, and it is possible that longer-term hospital outcomes (such as 30-d mortality rate) may also relate to patterns of mechanical ventilation.
In conclusion, we found that use of NIV in the management of COPD varied widely across hospitals. Institutions with greater use of NIV used less IMV and generally achieved better patient outcomes. Future research should explore the contextual factors and strategies associated with successful implementation of NIV.
Footnotes
Supported by Grant 1R18HL108810-01 from the National Heart, Lung, and Blood Institute of the National Institutes of Health.
Author Contributions: P.K.L., M.S.S., M.B.R., P.S.P., N.S.H.: conceived of and designed the study. P.K.L. acquired the data used in the analysis. P.K.L., M.S.S., P.S.P., M.-S.S., M.B.R., N.S.H.: performed data analysis and interpretation. P.K.L.: drafted the manuscript. M.S.S., P.S.P., M.-S.S., M.B.R., N.S.H.: reviewed the manuscript and contributed to revisions prior to manuscript submission.
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1.Centers for Disease Control and PreventionFastStats: chronic obstructive pulmonary disease (COPD) includes: chronic bronchitis and emphysema [accessed 2013 Dec 9]. Available fromhttp://www.cdc.gov/nchs/fastats/copd.htm
- 2.National Heart, Lung, and Blood Institute. Bethesda, MD: National Institutes of Health; 2009. Morbidity and mortality: 2009 chart book on cardiovascular, lung, and blood diseases. [Google Scholar]
- 3.Wier LM, Elixhauser A, Pfuntner A, Au DH.Overview of hospitalizations among patients with COPD, 2008Healthcare cost and utilization project (HCUP) statistical briefsRockville, MD: Agency for Health Care Policy and Research; 2006. Statistical Brief 106 [accessed 2013 Nov 27]. Available fromhttp://www.ncbi.nlm.nih.gov/books/NBK53969/ [Google Scholar]
- 4.Patil SP, Krishnan JA, Lechtzin N, Diette GB. In-hospital mortality following acute exacerbations of chronic obstructive pulmonary disease. Arch Intern Med. 2003;163:1180–1186. doi: 10.1001/archinte.163.10.1180. [DOI] [PubMed] [Google Scholar]
- 5.Tabak YP, Sun X, Johannes RS, Hyde L, Shorr AF, Lindenauer PK. Development and validation of a mortality risk-adjustment model for patients hospitalized for exacerbations of chronic obstructive pulmonary disease. Med Care. 2013;51:597–605. doi: 10.1097/MLR.0b013e3182901982. [DOI] [PubMed] [Google Scholar]
- 6.Lindenauer PK, Grosso LM, Wang C, Wang Y, Krishnan JA, Lee TA, Au DH, Mularski RA, Bernheim SM, Drye EE. Development, validation, and results of a risk-standardized measure of hospital 30-day mortality for patients with exacerbation of chronic obstructive pulmonary disease. J Hosp Med. 2013;8:428–435. doi: 10.1002/jhm.2066. [DOI] [PubMed] [Google Scholar]
- 7.Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360:1418–1428. doi: 10.1056/NEJMsa0803563. [DOI] [PubMed] [Google Scholar]
- 8.Stefan MS, Shieh M-S, Pekow PS, Rothberg MB, Steingrub JS, Lagu T, Lindenauer PK. Epidemiology and outcomes of acute respiratory failure in the United States, 2001 to 2009: a national survey. J Hosp Med. 2013;8:76–82. doi: 10.1002/jhm.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Global Initiative for Chronic Obstructive Lung DiseaseGlobal strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease [accessed 2015 Feb 17]. Available from: http://www.goldcopd.org/uploads/users/files/GOLD_Report_2015.pdf
- 10.Bott J, Carroll MP, Conway JH, Keilty SE, Ward EM, Brown AM, Paul EA, Elliott MW, Godfrey RC, Wedzicha JA, et al. Randomised controlled trial of nasal ventilation in acute ventilatory failure due to chronic obstructive airways disease. Lancet. 1993;341:1555–1557. doi: 10.1016/0140-6736(93)90696-e. [DOI] [PubMed] [Google Scholar]
- 11.Brochard L, Mancebo J, Wysocki M, Lofaso F, Conti G, Rauss A, Simonneau G, Benito S, Gasparetto A, Lemaire F, et al. Noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease. N Engl J Med. 1995;333:817–822. doi: 10.1056/NEJM199509283331301. [DOI] [PubMed] [Google Scholar]
- 12.Kramer N, Meyer TJ, Meharg J, Cece RD, Hill NS. Randomized, prospective trial of noninvasive positive pressure ventilation in acute respiratory failure. Am J Respir Crit Care Med. 1995;151:1799–1806. doi: 10.1164/ajrccm.151.6.7767523. [DOI] [PubMed] [Google Scholar]
- 13.Plant PK, Owen JL, Elliott MW. Early use of non-invasive ventilation for acute exacerbations of chronic obstructive pulmonary disease on general respiratory wards: a multicentre randomised controlled trial. Lancet. 2000;355:1931–1935. doi: 10.1016/s0140-6736(00)02323-0. [DOI] [PubMed] [Google Scholar]
- 14.Lightowler JV, Wedzicha JA, Elliott MW, Ram FSF. Non-invasive positive pressure ventilation to treat respiratory failure resulting from exacerbations of chronic obstructive pulmonary disease: Cochrane systematic review and meta-analysis. BMJ. 2003;326:185. doi: 10.1136/bmj.326.7382.185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Quon BS, Gan WQ, Sin DD. Contemporary management of acute exacerbations of COPD: a systematic review and metaanalysis. Chest. 2008;133:756–766. doi: 10.1378/chest.07-1207. [DOI] [PubMed] [Google Scholar]
- 16.Chandra D, Stamm JA, Taylor B, Ramos RM, Satterwhite L, Krishnan JA, Mannino D, Sciurba FC, Holguín F. Outcomes of noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease in the United States, 1998-2008. Am J Respir Crit Care Med. 2012;185:152–159. doi: 10.1164/rccm.201106-1094OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Walkey AJ, Wiener RS. Use of noninvasive ventilation in patients with acute respiratory failure, 2000-2009: a population-based study. Ann Am Thorac Soc. 2013;10:10–17. doi: 10.1513/AnnalsATS.201206-034OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lindenauer PK, Pekow P, Gao S, Crawford AS, Gutierrez B, Benjamin EM. Quality of care for patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 2006;144:894–903. doi: 10.7326/0003-4819-144-12-200606200-00006. [DOI] [PubMed] [Google Scholar]
- 19.Rothberg MB, Pekow PS, Lahti M, Brody O, Skiest DJ, Lindenauer PK. Antibiotic therapy and treatment failure in patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. JAMA. 2010;303:2035–2042. doi: 10.1001/jama.2010.672. [DOI] [PubMed] [Google Scholar]
- 20.Auerbach AD, Hilton JF, Maselli J, Pekow PS, Rothberg MB, Lindenauer PK. Shop for quality or volume? Volume, quality, and outcomes of coronary artery bypass surgery. Ann Intern Med. 2009;150:696–704. doi: 10.7326/0003-4819-150-10-200905190-00007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Stein BD, Bautista A, Schumock GT, Lee TA, Meltzer DO, Naureckas ET, Krishnan JA. Specificity of ICD-9 diagnosis codes for identifying patients hospitalized for COPD. Am J Respir Crit Care Med. 2009;179:A2159. [Google Scholar]
- 22.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64:749–759. doi: 10.1016/j.jclinepi.2010.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Antonelli M, Conti G, Moro ML, Esquinas A, Gonzalez-Diaz G, Confalonieri M, Pelaia P, Principi T, Gregoretti C, Beltrame F, et al. Predictors of failure of noninvasive positive pressure ventilation in patients with acute hypoxemic respiratory failure: a multi-center study. Intensive Care Med. 2001;27:1718–1728. doi: 10.1007/s00134-001-1114-4. [DOI] [PubMed] [Google Scholar]
- 24.American Hospital AssociationAHA annual survey database fiscal year 2011 [accessed 2015 Feb 17]. Available from: http://www.healthforum.com/datacoding/surveys.dhtml#2
- 25.Normand ST, Shahian DM. Statistical and clinical aspects of hospital outcomes profiling. Stat Sci. 2007;22:206–226. [Google Scholar]
- 26.Maheshwari V, Paioli D, Rothaar R, Hill NS. Utilization of noninvasive ventilation in acute care hospitals: a regional survey. Chest. 2006;129:1226–1233. doi: 10.1378/chest.129.5.1226. [DOI] [PubMed] [Google Scholar]
- 27.Tsai C-L, Lee W-Y, Delclos GL, Hanania NA, Camargo CA., Jr Comparative effectiveness of noninvasive ventilation vs invasive mechanical ventilation in chronic obstructive pulmonary disease patients with acute respiratory failure. J Hosp Med. 2013;8:165–172. doi: 10.1002/jhm.2014. [DOI] [PubMed] [Google Scholar]
- 28.Royal College of Physicians of London, British Thoracic Society, and British Lung FoundationReport of the National Chronic Obstructive Pulmonary Disease Audit 2008: clinical audit of COPD exacerbations admitted to acute NHS units across the UKLondon: Royal College of Physicians of London; November 2008[accessed 2013 Nov 30]. Available from: http://www.rcplondon.ac.uk/sites/default/files/report-of-the-national-copd-audit-2008-clinical-audit-of-copd-exacerbations-admitted-to-acute-nhs-units-across-the-uk.pdf [Google Scholar]
- 29.Stein BD, Bautista A, Schumock GT, Lee TA, Charbeneau JT, Lauderdale DS, Naureckas ET, Meltzer DO, Krishnan JA. The validity of International Classification of Diseases, Ninth Revision, Clinical Modification diagnosis codes for identifying patients hospitalized for COPD exacerbations. Chest. 2012;141:87–93. doi: 10.1378/chest.11-0024. [DOI] [PMC free article] [PubMed] [Google Scholar]