Letters
in community-dwelling older adults: results of a randomized controlled trial. J Am Geriatr Soc. 2009;57(11):2020-2028.
HEALTH CARE REFORM
Effect of Public Disclosure on Antibiotic Prescription Rate for Upper Respiratory Tract Infections Although antibiotics are not required for treating uncomplicated upper respiratory tract infection (URTI),1 which is mostly viral, they are often prescribed, fueling antibiotic resistance
and loss of protective flora. Accordingly, many studies worldwide have tried to decrease inappropriate antibiotic prescribing behavior.2 In South Korea, where the National Health Insurance provides universal coverage, the Health Insurance Review & Assessment Service (HIRA) oversees claims reviews, quality assessment, and benefits and coverage standards. Since 2001, HIRA has used claims data to assess the appropriateness of care based on various quality indicators, including the antibiotic
Table. Comparison of Antibiotic Prescription Rates for URTIs by Intervention Intervention Beforea
Afterb
Visits for URTI, No.
Antibiotic Prescription Rate [A], % (95% CI)
Visits for URTI, No.
Antibiotic Prescription Rate [B], % (95% CI)
Difference in Rate (A − B), % (95% CI)
Ratio of Rate (B:A) (95% CI)
4 651 905
58.8 (58.7-58.8)
7 013 624
53.0 (53.0-53.1)
5.7 (5.7-5.8)c
0.90 (0.90-0.90)d
4 495 231
58.9 (58.9-58.9)
6 668 069
53.3 (53.3-53.4)
5.6 (5.5-5.6)c
0.91 (0.90-0.91)d
Secondary care hospital
87 559
54.6 (54.3-54.9)
215 387
46.6 (46.4-46.8)
8.0 (7.7-8.4)c
0.85 (0.85-0.86)d
Tertiary care hospital
69 115
56.2 (55.8-56.6)
130 168
49.7 (49.4-49.9)
6.5 (5.2-5.2)c
0.88 (0.88-0.89)d
Hospital Type Total Primary clinic
Abbreviation: URTI, upper respiratory tract infection.
c
P < .001, t test.
a
January 1, 2003, through January 31, 2006.
d
P < .001, χ2 test.
b
February 1, 2006, through December 31, 2010.
Figure. Segmented Linear Regression of the Rate of Antibiotic Prescriptions and Predicted Rate of Antibiotic Prescriptions With ARIMA Model A
Rate of Antibiotic Prescriptions
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Jan 2003
Jul 2004
Jan 2006
Jul 2007
Jan 2009
Jul 2010
Date B
Rate of Antibiotic Prescriptions
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Jan 2003
Jul 2004
Jan 2006
Jul 2007
Jan 2009
Jul 2010
Date
jamainternalmedicine.com
A, Segmented linear regression is shown. B, Autoregressive integrated moving average (ARIMA) model: (p, d, q) = (2, 0, 0); seasonal components, (p, d, q, s) = (1, 0, 0, 12). Vertical line indicates February 2006; dots, observed rates; solid line and curve, predicted rates; dashed line (A), trend of rates; dotted curves (B), 95% CIs of predicted rates.
(Reprinted) JAMA Internal Medicine March 2015 Volume 175, Number 3
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/14/2015
445
Letters
prescription rate for URTIs,3 and has used them for monitoring and giving feedback to physicians. Because it is a prerequisite that the public perceive low antibiotic prescription rates as good quality of care, starting in February 2006, HIRA began warning the public of the potential harms of antibiotic overuse through television and radio campaigns. In addition, HIRA has disclosed physicians’ antibiotic prescription rates to the general public via its website. We evaluated the effect of such public disclosure using nationally representative data.
of physicians were aware, and the abrupt change might be explained by observer effect.6 Our study should be interpreted within the context of South Korea’s medical care, in which medical services are provided mostly by private providers who might be motivated to maintain a good reputation for economic reasons. In addition, the preexisting National Health Insurance database for fee-for-service reimbursement makes individual medical institutions’ antibiotic prescription rate easily available and implementation of the program affordable.
Methods | From the National Health Insurance claims data, a random sample of 3% of the adult population was selected (1 162 354 persons who were 20 years or older on December 31, 2002). The claims data of medical services, which include all prescriptions written from January 1, 2003, through December 31, 2010, were used. From these data, we defined the cases of URTI based on the International Classification of Diseases, Tenth Revision (ICD-10) codes J00 through J06, and the antibiotics according to their Anatomical Therapeutic Chemical classification (J01 category). To perform segmented regression, we totaled the visits each month and calculated the rate of antibiotic use. We performed segmented linear regression to compare the trend of antibiotic use before and after February 1, 2006. We analyzed the trend using the autoregressive integrated moving average model, which included the seasonal effect. We also compared the rates by intervention at 3 hospital levels: primary clinic, secondary care hospital, and tertiary care hospital. Analyses were conducted using STATA, version 13.1 (StataCorp).
Jae Moon Yun, MPH Dong Wook Shin, MD, DrPH, MBA Seung-sik Hwang, MD, PhD Juhee Cho, MA, PhD You Seon Nam, MD Jung Hoe Kim, DrPH Be Long Cho, MD, MPH, PhD
Results | From the predefined cohort, 938 118 persons with URTI (80.7%) had visited a clinic at least once between January 1, 2003, and December 31, 2010. The number of visits was 11 665 529, and 95.7% of the visits were to primary clinics. The rates of antibiotic prescriptions before and after public disclosure were 58.8% and 53.0%, respectively, and the decrease in rates by intervention were consistent regardless of the hospital level (Table). Segmented linear regression showed that antibiotic prescription rates abruptly decreased by February 2006, and differences between actual rates and predicted rates persisted until the end of the follow-up period. Autoregressive integrated moving average models showed that all actual rates after the intervention were below the predicted values, and most rates were observed below the lower limits of the 95% CIs (Figure). Discussion | Our data show that public disclosure was effective in lowering antibiotic use for URTIs. Based on the findings in a previous study4 on the effect of publicly disclosing quality improvement in other clinical areas, 2 primary mechanisms can be suggested: selection of high-quality providers by patients and provider response to report cards. A survey5 in 2007 found that 21.5% of health consumers were aware of the disclosure of antibiotic prescription rates and 40.3% of them had changed their health care provider. The survey also found that 95.0% 446
Author Affiliations: Health Promotion Center, Department of Family Medicine, Seoul National University Hospital, Seoul, Korea (Yun, Shin, Nam, B. L. Cho); Department of Social and Preventive Medicine, Inha University School of Medicine, Incheon, Korea (Hwang); Department of Health Sciences and Technology, Samsung Advanced Institute of Health Science and Technology, Sungkyunkwan University, Seoul, Korea (J. Cho); Health Insurance Review & Assessment Service, Seoul, Korea (Kim). Corresponding Author: Dong Wook Shin, MD, DrPH, MBA, Health Promotion Center, Department of Family Medicine Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea ([email protected]). Published Online: December 15, 2014. doi:10.1001/jamainternmed.2014.6569. Author Contributions: Drs Shin and Yun had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Yun, Shin, Hwang, B. L. Cho. Acquisition, analysis, or interpretation of data: Yun, Shin, Hwang, J. Cho, Nam. Drafting of the manuscript: All authors. Critical revision of the manuscript for important intellectual content: Yun, Shin, J. Cho, Nam, B. L. Cho. Statistical analysis: Yun, Hwang. Obtained funding: B. L. Cho. Administrative, technical, or material support: Kim. Study supervision: Shin, J. Cho, Nam, B. L. Cho. Conflict of Interest Disclosures: Dr Kim reports being affiliated with the Health Insurance Review & Assessment Service, but she did not participate in data collection and analysis. No other disclosures were reported. Funding/Support: This research was supported by grant 800-20130091 from the Korea Centers for Disease Control and Prevention. Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Additional Contributions: Sang Hyuck Kim, MD, Yun Hee Chung, MD, Hyejin Lee, MD, MPH, and Eunmi Ahn, MD, MPH, Health Promotion Center, Department of Family Medicine, Seoul National 11 University Hospital, contributed to the critical revision of the manuscript. They were not financially compensated. 1. Kenealy T, Arroll B. Antibiotics for the common cold and acute purulent rhinitis. Cochrane Database Syst Rev. 2013;6:CD000247. 2. Ranji SR, Steinman MA, Shojania KG, et al. Closing the Quality Gap: A Critical Analysis of Quality Improvement Strategies. Vol 4. Rockville, MD: Agency for Healthcare Research and Quality; 2006. 3. Cho W, Lee S, Kang HY, Kang M. Setting national priorities for quality assessment of health care services in Korea. Int J Qual Health Care. 2005;17(2): 157-165. 4. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA. 2005;293(10):1239-1244.
JAMA Internal Medicine March 2015 Volume 175, Number 3 (Reprinted)
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/14/2015
jamainternalmedicine.com
Letters
5. Joong Ang Economy Research Institute. A survey report of perception of disclosure of antibiotic prescription rate for acute upper respiratory infection [in Korean]. http://www.mw.go.kr/front_new/al/sal0301vw.jsp?PAR_MENU_ID =04&MENU_ID=0403&CONT_SEQ=40112. Accessed December 11, 2013. 6. Mangione-Smith R, Elliott MN, McDonald L, McGlynn EA. An observational study of antibiotic prescribing behavior and the Hawthorne effect. Health Serv Res. 2002;37(6):1603-1623.
LESS IS MORE
Computed Tomographic Pulmonary Angiography: Clinical Implications of a Limited Negative Result A substantial proportion of computed tomographic pulmonary angiography (CTPA) results are degraded by patientrelated and technical factors, yielding a limited negative interpretation, which qualifies the definitive exclusion of pulmonary embolism. This leads to diagnostic uncertainty and may affect patient management. A multicenter randomized clinical trial of CTPA vs ventilation-perfusion scintigraphy considered CTPA results negative if no pulmonary embolism was demonstrated to the lobar level.1 Investigators have explicitly questioned the clinical relevance of small pulmonary emboli,2,3 and there is growing concern that pulmonary embolism is overdiagnosed on CTPA.4,5 We hypothesized that a limited negative CTPA result portends similar clinical outcomes to a definitively negative result; thus, emphasis on examination limitations may, in fact, be detrimental.
Methods | This retrospective cohort study was approved by the institutional review board of the Montefiore Medical Center, which waived written informed consent. Included were all adults with suspected pulmonary embolism who underwent CTPA from July 1, 2011, through June 30, 2012, at the 3 emergency departments or inpatient sites of Montefiore Medical Center, Bronx, New York. We defined a limited negative CTPA result as one that reported no pulmonary embolism but also explicitly stated a limitation or qualification in the report impression. The primary outcome was 90-day venous thromboembolism incidence, a measure of the false-negative rate. Secondary outcomes were additional imaging, initiation of anticoagulation treatment, and bleeding complications. Results | The study population comprised 2553 patients with mean (SD) age 55 (18) years, 66% women and 85% ethnic minorities. A total of 264 CTPA results (10%) were positive; 1663 (65%) were definitively negative; 569 (22%) were limited negative; and 57 (2%) were nondiagnostic for pulmonary embolism. In all, 25% of negative CTPA results (569 of 2232) were limited negative. Patients with limited negative CTPA results were more similar to patients with definitely negative results than they were to patients with positive results in their baseline characteristics and outcomes (Table 1). The false-negative rate (90-day venous thromboembolism incidence) for patients with a limited negative CTPA re-
Table 1. Clinical Characteristics of Study Population CTPA Resulta Negative Positive (n = 264)
Characteristic
Limited (n = 569)
Definitively (n = 1663)
P Valueb
Patient demographics Age, median (IQR), y
65 (48-74)
Female sex
54 (42-68)
54 (41-69)
.80
159 (60)
337 (59)
1155 (69)
.99
Primary language Spanish Patient comorbidities
Acute myocardial infarction
8 (3)
12 (2)
35 (2)
Congestive heart failure
16 (6)
76 (13)
193 (12)
.30
Atrial fibrillation
23 (9)
58 (10)
135 (8)
.14
187 (71)
169 (30)
445 (27)
.10
Malignant condition
46 (17)
58 (10)
178 (11)
.90
Metastatic malignant condition
26 (10)
23 (4)
76 (5)
.70
Chronic pulmonary disease
Abbreviations: CT, computed tomography; CTPA, computed tomographic pulmonary angiography; DVT, deep venous thrombosis; IQR, interquartile range; LE, lower extremity; PE, pulmonary embolism; V/Q, ventilation perfusion scan. a
Unless otherwise indicated data are reported as number (percentage) of study participants.
b
P values indicate comparison between limited negative and definitively negative categories.
c
Diagnosis of PE and/or DVT within 90 days after index CT scan.
d
Additional imaging test performed within 5 days after index CT scan.
e
Anticoagulation (either dalteparin or IV heparin) administered within 5 days after index CT scan.
f
Any bleeding complication occurring within 90 d after index CT scan.
Outcomes 90-Day venous thromboembolism incidencec
30 (11)
8 (1)
30 (2)
.70
Additional CT of thorax, V/Q, or LE Doppler studyd
18 (7)
20 (45)
37 (2)
.12
11 (4)
6 (1)
19 (1)
>.99
Additional V/Q
0
13 (2)
Additional LE Doppler study
7 (3)
Additional CT of thorax
Anticoagulation, dalteparin or intravenous heparine Bleeding complicationsf
jamainternalmedicine.com
1 (
in community-dwelling older adults: results of a randomized controlled trial. J Am Geriatr Soc. 2009;57(11):2020-2028.
HEALTH CARE REFORM
Effect of Public Disclosure on Antibiotic Prescription Rate for Upper Respiratory Tract Infections Although antibiotics are not required for treating uncomplicated upper respiratory tract infection (URTI),1 which is mostly viral, they are often prescribed, fueling antibiotic resistance
and loss of protective flora. Accordingly, many studies worldwide have tried to decrease inappropriate antibiotic prescribing behavior.2 In South Korea, where the National Health Insurance provides universal coverage, the Health Insurance Review & Assessment Service (HIRA) oversees claims reviews, quality assessment, and benefits and coverage standards. Since 2001, HIRA has used claims data to assess the appropriateness of care based on various quality indicators, including the antibiotic
Table. Comparison of Antibiotic Prescription Rates for URTIs by Intervention Intervention Beforea
Afterb
Visits for URTI, No.
Antibiotic Prescription Rate [A], % (95% CI)
Visits for URTI, No.
Antibiotic Prescription Rate [B], % (95% CI)
Difference in Rate (A − B), % (95% CI)
Ratio of Rate (B:A) (95% CI)
4 651 905
58.8 (58.7-58.8)
7 013 624
53.0 (53.0-53.1)
5.7 (5.7-5.8)c
0.90 (0.90-0.90)d
4 495 231
58.9 (58.9-58.9)
6 668 069
53.3 (53.3-53.4)
5.6 (5.5-5.6)c
0.91 (0.90-0.91)d
Secondary care hospital
87 559
54.6 (54.3-54.9)
215 387
46.6 (46.4-46.8)
8.0 (7.7-8.4)c
0.85 (0.85-0.86)d
Tertiary care hospital
69 115
56.2 (55.8-56.6)
130 168
49.7 (49.4-49.9)
6.5 (5.2-5.2)c
0.88 (0.88-0.89)d
Hospital Type Total Primary clinic
Abbreviation: URTI, upper respiratory tract infection.
c
P < .001, t test.
a
January 1, 2003, through January 31, 2006.
d
P < .001, χ2 test.
b
February 1, 2006, through December 31, 2010.
Figure. Segmented Linear Regression of the Rate of Antibiotic Prescriptions and Predicted Rate of Antibiotic Prescriptions With ARIMA Model A
Rate of Antibiotic Prescriptions
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Jan 2003
Jul 2004
Jan 2006
Jul 2007
Jan 2009
Jul 2010
Date B
Rate of Antibiotic Prescriptions
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Jan 2003
Jul 2004
Jan 2006
Jul 2007
Jan 2009
Jul 2010
Date
jamainternalmedicine.com
A, Segmented linear regression is shown. B, Autoregressive integrated moving average (ARIMA) model: (p, d, q) = (2, 0, 0); seasonal components, (p, d, q, s) = (1, 0, 0, 12). Vertical line indicates February 2006; dots, observed rates; solid line and curve, predicted rates; dashed line (A), trend of rates; dotted curves (B), 95% CIs of predicted rates.
(Reprinted) JAMA Internal Medicine March 2015 Volume 175, Number 3
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/14/2015
445
Letters
prescription rate for URTIs,3 and has used them for monitoring and giving feedback to physicians. Because it is a prerequisite that the public perceive low antibiotic prescription rates as good quality of care, starting in February 2006, HIRA began warning the public of the potential harms of antibiotic overuse through television and radio campaigns. In addition, HIRA has disclosed physicians’ antibiotic prescription rates to the general public via its website. We evaluated the effect of such public disclosure using nationally representative data.
of physicians were aware, and the abrupt change might be explained by observer effect.6 Our study should be interpreted within the context of South Korea’s medical care, in which medical services are provided mostly by private providers who might be motivated to maintain a good reputation for economic reasons. In addition, the preexisting National Health Insurance database for fee-for-service reimbursement makes individual medical institutions’ antibiotic prescription rate easily available and implementation of the program affordable.
Methods | From the National Health Insurance claims data, a random sample of 3% of the adult population was selected (1 162 354 persons who were 20 years or older on December 31, 2002). The claims data of medical services, which include all prescriptions written from January 1, 2003, through December 31, 2010, were used. From these data, we defined the cases of URTI based on the International Classification of Diseases, Tenth Revision (ICD-10) codes J00 through J06, and the antibiotics according to their Anatomical Therapeutic Chemical classification (J01 category). To perform segmented regression, we totaled the visits each month and calculated the rate of antibiotic use. We performed segmented linear regression to compare the trend of antibiotic use before and after February 1, 2006. We analyzed the trend using the autoregressive integrated moving average model, which included the seasonal effect. We also compared the rates by intervention at 3 hospital levels: primary clinic, secondary care hospital, and tertiary care hospital. Analyses were conducted using STATA, version 13.1 (StataCorp).
Jae Moon Yun, MPH Dong Wook Shin, MD, DrPH, MBA Seung-sik Hwang, MD, PhD Juhee Cho, MA, PhD You Seon Nam, MD Jung Hoe Kim, DrPH Be Long Cho, MD, MPH, PhD
Results | From the predefined cohort, 938 118 persons with URTI (80.7%) had visited a clinic at least once between January 1, 2003, and December 31, 2010. The number of visits was 11 665 529, and 95.7% of the visits were to primary clinics. The rates of antibiotic prescriptions before and after public disclosure were 58.8% and 53.0%, respectively, and the decrease in rates by intervention were consistent regardless of the hospital level (Table). Segmented linear regression showed that antibiotic prescription rates abruptly decreased by February 2006, and differences between actual rates and predicted rates persisted until the end of the follow-up period. Autoregressive integrated moving average models showed that all actual rates after the intervention were below the predicted values, and most rates were observed below the lower limits of the 95% CIs (Figure). Discussion | Our data show that public disclosure was effective in lowering antibiotic use for URTIs. Based on the findings in a previous study4 on the effect of publicly disclosing quality improvement in other clinical areas, 2 primary mechanisms can be suggested: selection of high-quality providers by patients and provider response to report cards. A survey5 in 2007 found that 21.5% of health consumers were aware of the disclosure of antibiotic prescription rates and 40.3% of them had changed their health care provider. The survey also found that 95.0% 446
Author Affiliations: Health Promotion Center, Department of Family Medicine, Seoul National University Hospital, Seoul, Korea (Yun, Shin, Nam, B. L. Cho); Department of Social and Preventive Medicine, Inha University School of Medicine, Incheon, Korea (Hwang); Department of Health Sciences and Technology, Samsung Advanced Institute of Health Science and Technology, Sungkyunkwan University, Seoul, Korea (J. Cho); Health Insurance Review & Assessment Service, Seoul, Korea (Kim). Corresponding Author: Dong Wook Shin, MD, DrPH, MBA, Health Promotion Center, Department of Family Medicine Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea ([email protected]). Published Online: December 15, 2014. doi:10.1001/jamainternmed.2014.6569. Author Contributions: Drs Shin and Yun had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Yun, Shin, Hwang, B. L. Cho. Acquisition, analysis, or interpretation of data: Yun, Shin, Hwang, J. Cho, Nam. Drafting of the manuscript: All authors. Critical revision of the manuscript for important intellectual content: Yun, Shin, J. Cho, Nam, B. L. Cho. Statistical analysis: Yun, Hwang. Obtained funding: B. L. Cho. Administrative, technical, or material support: Kim. Study supervision: Shin, J. Cho, Nam, B. L. Cho. Conflict of Interest Disclosures: Dr Kim reports being affiliated with the Health Insurance Review & Assessment Service, but she did not participate in data collection and analysis. No other disclosures were reported. Funding/Support: This research was supported by grant 800-20130091 from the Korea Centers for Disease Control and Prevention. Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Additional Contributions: Sang Hyuck Kim, MD, Yun Hee Chung, MD, Hyejin Lee, MD, MPH, and Eunmi Ahn, MD, MPH, Health Promotion Center, Department of Family Medicine, Seoul National 11 University Hospital, contributed to the critical revision of the manuscript. They were not financially compensated. 1. Kenealy T, Arroll B. Antibiotics for the common cold and acute purulent rhinitis. Cochrane Database Syst Rev. 2013;6:CD000247. 2. Ranji SR, Steinman MA, Shojania KG, et al. Closing the Quality Gap: A Critical Analysis of Quality Improvement Strategies. Vol 4. Rockville, MD: Agency for Healthcare Research and Quality; 2006. 3. Cho W, Lee S, Kang HY, Kang M. Setting national priorities for quality assessment of health care services in Korea. Int J Qual Health Care. 2005;17(2): 157-165. 4. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA. 2005;293(10):1239-1244.
JAMA Internal Medicine March 2015 Volume 175, Number 3 (Reprinted)
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/14/2015
jamainternalmedicine.com
Letters
5. Joong Ang Economy Research Institute. A survey report of perception of disclosure of antibiotic prescription rate for acute upper respiratory infection [in Korean]. http://www.mw.go.kr/front_new/al/sal0301vw.jsp?PAR_MENU_ID =04&MENU_ID=0403&CONT_SEQ=40112. Accessed December 11, 2013. 6. Mangione-Smith R, Elliott MN, McDonald L, McGlynn EA. An observational study of antibiotic prescribing behavior and the Hawthorne effect. Health Serv Res. 2002;37(6):1603-1623.
LESS IS MORE
Computed Tomographic Pulmonary Angiography: Clinical Implications of a Limited Negative Result A substantial proportion of computed tomographic pulmonary angiography (CTPA) results are degraded by patientrelated and technical factors, yielding a limited negative interpretation, which qualifies the definitive exclusion of pulmonary embolism. This leads to diagnostic uncertainty and may affect patient management. A multicenter randomized clinical trial of CTPA vs ventilation-perfusion scintigraphy considered CTPA results negative if no pulmonary embolism was demonstrated to the lobar level.1 Investigators have explicitly questioned the clinical relevance of small pulmonary emboli,2,3 and there is growing concern that pulmonary embolism is overdiagnosed on CTPA.4,5 We hypothesized that a limited negative CTPA result portends similar clinical outcomes to a definitively negative result; thus, emphasis on examination limitations may, in fact, be detrimental.
Methods | This retrospective cohort study was approved by the institutional review board of the Montefiore Medical Center, which waived written informed consent. Included were all adults with suspected pulmonary embolism who underwent CTPA from July 1, 2011, through June 30, 2012, at the 3 emergency departments or inpatient sites of Montefiore Medical Center, Bronx, New York. We defined a limited negative CTPA result as one that reported no pulmonary embolism but also explicitly stated a limitation or qualification in the report impression. The primary outcome was 90-day venous thromboembolism incidence, a measure of the false-negative rate. Secondary outcomes were additional imaging, initiation of anticoagulation treatment, and bleeding complications. Results | The study population comprised 2553 patients with mean (SD) age 55 (18) years, 66% women and 85% ethnic minorities. A total of 264 CTPA results (10%) were positive; 1663 (65%) were definitively negative; 569 (22%) were limited negative; and 57 (2%) were nondiagnostic for pulmonary embolism. In all, 25% of negative CTPA results (569 of 2232) were limited negative. Patients with limited negative CTPA results were more similar to patients with definitely negative results than they were to patients with positive results in their baseline characteristics and outcomes (Table 1). The false-negative rate (90-day venous thromboembolism incidence) for patients with a limited negative CTPA re-
Table 1. Clinical Characteristics of Study Population CTPA Resulta Negative Positive (n = 264)
Characteristic
Limited (n = 569)
Definitively (n = 1663)
P Valueb
Patient demographics Age, median (IQR), y
65 (48-74)
Female sex
54 (42-68)
54 (41-69)
.80
159 (60)
337 (59)
1155 (69)
.99
Primary language Spanish Patient comorbidities
Acute myocardial infarction
8 (3)
12 (2)
35 (2)
Congestive heart failure
16 (6)
76 (13)
193 (12)
.30
Atrial fibrillation
23 (9)
58 (10)
135 (8)
.14
187 (71)
169 (30)
445 (27)
.10
Malignant condition
46 (17)
58 (10)
178 (11)
.90
Metastatic malignant condition
26 (10)
23 (4)
76 (5)
.70
Chronic pulmonary disease
Abbreviations: CT, computed tomography; CTPA, computed tomographic pulmonary angiography; DVT, deep venous thrombosis; IQR, interquartile range; LE, lower extremity; PE, pulmonary embolism; V/Q, ventilation perfusion scan. a
Unless otherwise indicated data are reported as number (percentage) of study participants.
b
P values indicate comparison between limited negative and definitively negative categories.
c
Diagnosis of PE and/or DVT within 90 days after index CT scan.
d
Additional imaging test performed within 5 days after index CT scan.
e
Anticoagulation (either dalteparin or IV heparin) administered within 5 days after index CT scan.
f
Any bleeding complication occurring within 90 d after index CT scan.
Outcomes 90-Day venous thromboembolism incidencec
30 (11)
8 (1)
30 (2)
.70
Additional CT of thorax, V/Q, or LE Doppler studyd
18 (7)
20 (45)
37 (2)
.12
11 (4)
6 (1)
19 (1)
>.99
Additional V/Q
0
13 (2)
Additional LE Doppler study
7 (3)
Additional CT of thorax
Anticoagulation, dalteparin or intravenous heparine Bleeding complicationsf
jamainternalmedicine.com
1 (
