Motivation: After even a few months in the wards, I think that all of us learn that coumadin/warfarin inspires mixed emotions. With too little drug, you don't get the proven benefits. And, with too much, there is chance of bleeding - even to death. This summer, while in the neuro service, I saw a man on coumadin die from intraventricular hemorrhage. The problem with coumadin is that its metabolism is affected by so many environmental variables that steady anti-coagulation is hard to maintain. Recently, the FDA approved dabigatran, which is a direct inhibitor of thrombin, for anti-coagulation in atrial fibrillation. It is expected that in the future, dabigatran can be substituted for most cases of anticoagulation. But, what is the data to justify replacing good old coumadin?
Paper(s): Two primary studies for dabigatran: (1) Trial RE-COVER: Dabigatran versus Warfarin in the Treatment of Acute Venous Thromboembolism. Schulman, S. et. al. NEJM, 2009 (361): 2342-2352. (2) Trial RE-LY: Dabigatran versus Warfarin in Patients with Atrial Fibrillation. Stuart, J. et. al. NEJM, 2009 (361): 1139-1151.
Links: RE-COVER: http://www.nejm.org/doi/full/10.1056/NEJMoa0906598#t=article; RE-LY: http://www.nejm.org/doi/full/10.1056/NEJMoa0905561#t=article
Methods: For the RE-COVER study, 2,539 patients were recruited from 228 clinical centers in 29 countries and, in a blinded way, randomized to a fixed dose of dabigatran (150 mg bid) or warfarin adjusted to INR of 2.0-3.0. Patients were essentially adults with symptomatic deep vein thrombosis who did not have pulmonary embolism with hemodynamic instability. The primary end-point was symptomatic venous thromboembolism (VTE) or death from VTE. For the RE-LY study, 18,113 patients were recruited from 951 clinical centers in 44 countries and randomized to a fixed dose of dabigatran (110 mg bid or 150 mg bid) or warfarin adjusted to INR of 2.0-3.0. In the study, patients essentially had to have atrial fibrillation plus one more condition: previous stroke/TIA, heart failure, or older age (65-74) with diabetes, HTN, or CAD. The primary end-point was stroke or systemic embolism.
Results:
RE-COVER: In the primary end-point of symptomatic or deadly VTE, 2.4% in the dabigatran group had such events while 2.1% in the warfarin group (difference not significant). There was no difference in overall mortality. In the warfarin group, INR was therapeutic for only 66% of the time. In terms of safety, there were equivalent rates for any bleeding events or for major bleeds (1.6% in dabigatran vs. 1.9% in coumadin). Patients in dabigatran were marginally more likely to experience any adverse event (9.0% vs. 6.8%, p=0.05). Follow-up was for six months and essentially complete (>98%).
RE-LY: For the primary end-point of stroke or systemic embolism, the 150 mg dose of dabigatran was superior to coumadin (rate of 1.11% vs. 1.69% per year, p<0.001). The 110 mg dose of dabigatran was equivalent to coumadin (rate of stroke or embolism: 1.53% vs. 1.69%, p=0.34). Rates of hemorrhagic stroke were lower in the 150 mg group compared to warfarin (0.10% vs. 0.38%, p<0.001). Similar trend was found in the 110 mg group. Overall rates of mortality were not significantly different though there was a slight trend of lower mortality in the 150 mg group (3.64% vs. 4.13%, p=0.051).
Follow-up was essentially complete in all patients for two years. In the warfarin group, INR was therapeutic only 64% of the time. In terms of safety, rate of MI was higher in the 150 mg group compared to warfarin (0.74% vs. 0.53%, p = 0.048). Same trend held for 110 mg group. Overall, life-threatening bleeds were lower in both dabigatran group compared to warfarin, but rates of major GI bleeds were higher in the 150 mg dabigatran group compared to warfarin (1.51% vs. 1.02%, p<0.001). In the 110 mg group, rates of GI bleeding were comparable.
Conclusion: Overall, both studies were well done with good follow-up and clinically relevant end-points chosen. The RE-COVER trial satisfactorily showed, I think, that dabigatran was equivalent to warfarin in terms of preventing complications of thromboembolism. Another interesting result from this trial was that even though warfarin was not therapeutic 1/3 of the time in the trial, the missed target did not seem to make any difference for thromboembolic events or bleeding events when compared with dabigatran, which presumably has pretty constant anti-coagulation all of the time. The RE-LY also provided many interesting results. On one hand, 150 mg dose of dabigatran proved to be superior to coumadin (and 110 mg dose) for preventing embolic events from atrial-fibrillation. On the other hand, the side-effect of dabigatran is a mixed bag. While it has lower rates of potentially deadly intraventricular hemorrhage and life-threatening bleeds, dabigatran had higher rates of MI and GI bleeds - not altogether benign either. There is no good explanation for why dabigatran has higher rates of MI. Finally, dabigatran is estimated to cost about five times more than coumadin. What this does for our health care spending with 30 million prescriptions for warfarin each year is an open issue. In conclusion, I think dabigatran is a definite first choice for patients with atrial-fibrillation or DVT with poor follow-up in INR clinics and good first-choice alternative in other patients with atrial fibrillation, who can afford the drug and understand the varying side-effect profile.
Wednesday, October 20, 2010
Tuesday, October 12, 2010
Giving an infant candy - does it work?
Motivation: In pediatrics clerkship, I have come across babies who just howl after blood draws, and one of the techniques we use is coating their pacifier with a sweet liquid. I guess the theory is that sugar makes babies happy and must therefore counteract the pain. But, does the sugar work? Or, should we provide babies with analgesics instead?
Turns out that this topic has been extensively studied in newborns (with 44 RCTs!) stemming from concerns about initial pain experience on subsequent neurodevelopment, and based on subjective behavioral changes, sugar pacifies babies. But, recent evidence suggests that neonates are perhaps too immature to link pain experience and behavior reliably. So, does sugar really help?
Paper: Oral Sucrose as an Analgesic Drug for Procedural Pain in Newborn Infants: A Randomised Controlled Trial. Slater, R., Cornellisen, L., et. al. Lancet, 2010 (376): 1225-1232. http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(10)61303-7/abstract?rss=yes#
Method: At the University College Hospital in London, 59 full-term healthy neonates were randomized to either 0.5 mL 24% sucrose solution or 0.5 mL sterile water before receiving a heel-stick. Before the babies received the heel-stick, they were subjected to a control non-painful stimulus by just touching the blunt end of the lancet to their heels. The cortical response to both the control and noxious stimuli was recorded via an EEG. The primary outcome was difference in pain-specific EEG activity in response to treatment with sugar. Secondary outcomes were observation of facial expression and physiological changes. Both parents and clinicians were entirely blinded to the identity of the administered solution.
Results: Because of technical difficulties of performing EEG on squirming babies and other measurement barriers, the sucrose group ultimately had 20 neonates and the control group had 24 neonates. The two groups were not substantially different in baseline characteristics after the dropout.
There was no difference in cortical pain-specific EEG changes between the sucrose group and control group. Physiologically, there were no differences in heart rate or oxygen saturation. However, based on facial expression, babies who received sucrose were less likely to alter facial expression to pain (35% in the sucrose group did not show any facial changes to heel-stick while all babies in the control group had facial changes like crying).
Conclusion: After this study, my conclusion is that we really don't know what babies are feeling. The key fact that we want to understand is the subjective experience of pain in babies. Since neonates do not talk, there is no definite gold standard. Previously, the facial expression was used as a measure of the subjective experience. This paper challenges that viewpoint. There are two alternative explanations for the paper's results. One is that the EEG is simply measuring nonciceptive input to the brain, and the facial expression is showing how the baby is processing that input. From this perspective, relying on the baby's facial expression makes more sense. Alternatively, the sucrose may be reflexively changing the baby's facial expression, but the pain is felt in equal intensity in the cortex. From this viewpoint, relying on the EEG cortical response makes more sense. Right now, I think that the only way to separate these alternative explanations is to do longer term studies to see which pain response is more closely linked with aberrant neurodevelopment.
Turns out that this topic has been extensively studied in newborns (with 44 RCTs!) stemming from concerns about initial pain experience on subsequent neurodevelopment, and based on subjective behavioral changes, sugar pacifies babies. But, recent evidence suggests that neonates are perhaps too immature to link pain experience and behavior reliably. So, does sugar really help?
Paper: Oral Sucrose as an Analgesic Drug for Procedural Pain in Newborn Infants: A Randomised Controlled Trial. Slater, R., Cornellisen, L., et. al. Lancet, 2010 (376): 1225-1232. http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(10)61303-7/abstract?rss=yes#
Method: At the University College Hospital in London, 59 full-term healthy neonates were randomized to either 0.5 mL 24% sucrose solution or 0.5 mL sterile water before receiving a heel-stick. Before the babies received the heel-stick, they were subjected to a control non-painful stimulus by just touching the blunt end of the lancet to their heels. The cortical response to both the control and noxious stimuli was recorded via an EEG. The primary outcome was difference in pain-specific EEG activity in response to treatment with sugar. Secondary outcomes were observation of facial expression and physiological changes. Both parents and clinicians were entirely blinded to the identity of the administered solution.
Results: Because of technical difficulties of performing EEG on squirming babies and other measurement barriers, the sucrose group ultimately had 20 neonates and the control group had 24 neonates. The two groups were not substantially different in baseline characteristics after the dropout.
There was no difference in cortical pain-specific EEG changes between the sucrose group and control group. Physiologically, there were no differences in heart rate or oxygen saturation. However, based on facial expression, babies who received sucrose were less likely to alter facial expression to pain (35% in the sucrose group did not show any facial changes to heel-stick while all babies in the control group had facial changes like crying).
Conclusion: After this study, my conclusion is that we really don't know what babies are feeling. The key fact that we want to understand is the subjective experience of pain in babies. Since neonates do not talk, there is no definite gold standard. Previously, the facial expression was used as a measure of the subjective experience. This paper challenges that viewpoint. There are two alternative explanations for the paper's results. One is that the EEG is simply measuring nonciceptive input to the brain, and the facial expression is showing how the baby is processing that input. From this perspective, relying on the baby's facial expression makes more sense. Alternatively, the sucrose may be reflexively changing the baby's facial expression, but the pain is felt in equal intensity in the cortex. From this viewpoint, relying on the EEG cortical response makes more sense. Right now, I think that the only way to separate these alternative explanations is to do longer term studies to see which pain response is more closely linked with aberrant neurodevelopment.
Tuesday, September 21, 2010
Update on Cocaine and Beta-Blockers
Motivation: Cocaine use in Baltimore is unfortunately common. During my sub-I in medicine, I met more than a couple of patients presenting with chest-pain after cocaine use. If I did not know that the patient had been using cocaine, I would have suggested beta-blocker for its anti-arrythmic and anti-hypertensive effects. But, the traditional teaching is that beta-blockers are contraindicated after cocaine use. Just searching Google for "cocaine beta blocker" pulls up many sites warning against beta-blocker use - including trusty Wikipedia.
The traditional teaching is that since cocaine is a norepinephrine reuptake inhibitor, blocking the beta-receptor sites would lead to "unblocked" alpha-adrenergic effect of increased hypertension. Beta-receptors (esp. Beta2 receptor) have some vasodilatory effect. But, is this all theory or are there trials? Despite decades of official warning against beta-blocker use, the following paper is the first paper to assess use of beta-blockers in clinical chest pain.
Paper: Beta-blockers for Chest Pain Associated with Recent Cocaine Use. Rangel, C. et. al. Arch. Intern. Med. 2010 (170): 874-879. http://archinte.ama-assn.org/cgi/content/full/170/10/874
Method: In a retrospective study, authors looked at patients admitted to San Francisco General Hospital with chest pain and U-tox positive for cocaine. The authors primarily examined association between cocaine use and death. Secondary outcomes were blood pressure levels, troponin levels, occurrence of v-fib/v-tach, intubation, or need for vasopressors. Patients with clearly documented pulmonary etiologies such as pneumonia or pulmonary embolus were excluded. 331 patients met criteria of chest pain with positive urine toxicology.
Results:
Characteristics: Of 331 patients with chest pain and cocaine use, 46% got beta-blocker in the ED - mostly IV metoprolol. Patients who got beta-blockers tended to be a little bit older (51 years versus 49 years) and likely to have higher blood pressure (SBP of 159 versus 141), history of HTN (70% vs 58%) and coronary bypass grafting (6% vs 1%), and have concurrent use of ace inhibitor (42% vs 29%) and statin (17% vs 8%).
Death: 45 patients died during follow-up after hospitalization. 12% of those who received beta-blocker died compared to 15% of those not getting beta-blockers (p = 0.38). After adjusting for confounding variables, being discharged on a beta-blocker was associated with 70% reduction in risk of cardiovascular death (HR: 0.29 CI: 0.09-0.98).
Secondary outcomes: After adjusting for other medications received, patients on beta-blockers had a mean 8 mmHg greater decrease in systolic blood pressure compared to patients who did not get beta-blockers. Receiving beta-blocker did not result in meaningful ECG differences, differences in peak troponin levels or incidence of malignant ventricular arrythmias.
Conclusion: Beta-blockers did not seem to harm patients with positive cocaine use history. In particular, beta-blocker administration in the ED resulted in lower rather than the hypothesized higher blood pressure! Also, being discharged on beta-blockers significantly decreased risk of cardiovascular death. What I found remarkable was that in general, the patients given beta-blockers might have been unhealthier in terms of age, blood pressure, and bypass history.
Being a retrospective study, of course, imposes some significant limitations on the study. The group getting beta-blocker and the group not getting beta-blocker were different, and these differences may have influenced the results in unforseen ways not easily corrected by statistical adjusting. Also, some of the confidence intervals were rather large. The confidence interval showing 70% risk reduction in CVD death had an upper limit confidence interval of the hazard ratio at 0.98.
The traditional teaching is that since cocaine is a norepinephrine reuptake inhibitor, blocking the beta-receptor sites would lead to "unblocked" alpha-adrenergic effect of increased hypertension. Beta-receptors (esp. Beta2 receptor) have some vasodilatory effect. But, is this all theory or are there trials? Despite decades of official warning against beta-blocker use, the following paper is the first paper to assess use of beta-blockers in clinical chest pain.
Paper: Beta-blockers for Chest Pain Associated with Recent Cocaine Use. Rangel, C. et. al. Arch. Intern. Med. 2010 (170): 874-879. http://archinte.ama-assn.org/cgi/content/full/170/10/874
Method: In a retrospective study, authors looked at patients admitted to San Francisco General Hospital with chest pain and U-tox positive for cocaine. The authors primarily examined association between cocaine use and death. Secondary outcomes were blood pressure levels, troponin levels, occurrence of v-fib/v-tach, intubation, or need for vasopressors. Patients with clearly documented pulmonary etiologies such as pneumonia or pulmonary embolus were excluded. 331 patients met criteria of chest pain with positive urine toxicology.
Results:
Characteristics: Of 331 patients with chest pain and cocaine use, 46% got beta-blocker in the ED - mostly IV metoprolol. Patients who got beta-blockers tended to be a little bit older (51 years versus 49 years) and likely to have higher blood pressure (SBP of 159 versus 141), history of HTN (70% vs 58%) and coronary bypass grafting (6% vs 1%), and have concurrent use of ace inhibitor (42% vs 29%) and statin (17% vs 8%).
Death: 45 patients died during follow-up after hospitalization. 12% of those who received beta-blocker died compared to 15% of those not getting beta-blockers (p = 0.38). After adjusting for confounding variables, being discharged on a beta-blocker was associated with 70% reduction in risk of cardiovascular death (HR: 0.29 CI: 0.09-0.98).
Secondary outcomes: After adjusting for other medications received, patients on beta-blockers had a mean 8 mmHg greater decrease in systolic blood pressure compared to patients who did not get beta-blockers. Receiving beta-blocker did not result in meaningful ECG differences, differences in peak troponin levels or incidence of malignant ventricular arrythmias.
Conclusion: Beta-blockers did not seem to harm patients with positive cocaine use history. In particular, beta-blocker administration in the ED resulted in lower rather than the hypothesized higher blood pressure! Also, being discharged on beta-blockers significantly decreased risk of cardiovascular death. What I found remarkable was that in general, the patients given beta-blockers might have been unhealthier in terms of age, blood pressure, and bypass history.
Being a retrospective study, of course, imposes some significant limitations on the study. The group getting beta-blocker and the group not getting beta-blocker were different, and these differences may have influenced the results in unforseen ways not easily corrected by statistical adjusting. Also, some of the confidence intervals were rather large. The confidence interval showing 70% risk reduction in CVD death had an upper limit confidence interval of the hazard ratio at 0.98.
Friday, September 17, 2010
Is There Evidence for Lung Cancer Screening?
Is There Evidence for Mass Lung Cancer Screening?
Mr. JF is a 52 year old man with hypertension and a 30 pack year smoking history. In addition to smoking cessation is there anyway to decrease his mortality from lung cancer through screening?
Lung cancer is:
Can mass screening lower fatality through earlier detection of localized disease?
National Guidelines Clearinghouse:
“Screening for Lung Cancer” (2010)
Results
Critiques of Methodology
Discussion
More frequent CXR vs. Less frequent CXR
Annual CXR/4m Sputum vs. Annual CXR Alone
Definitions:
No study addressed whether screening is better than no screening
Survival results were most heterogeneous. Survival can be confounded by lead-time, length time and overdiagnosis bias.
More frequent CXR leading to both increase in disease specific mortality and disease specific survival in pooled data further suggest unreliability of survival as outcome.
Increased CXR was shown to actually increase cancer mortality in several studies.
CXR unlikely to cause increased mortality per se due to low radiation dose but may lead to unnecessary surgery and early diagnosis that can lead to depression.
Several studies had methodological flaws such as baseline differences b/w groups and poor randomization/masking
Contamination (control group pts received intervention) and compliance (intervention group pts not receiving intervention) decrease effect of screening
CXR does not detect small tumors whose removal may have the most benefit to pts.
Recent large uncontrolled trial of spiral CT showed 92% of lung cancers dx were stage I, with those undergoing resection having a 10yr survival of 85%.
CT Lung screening associated with 3x increase in lung cancer dx and 10x increase in surgery.
Current Undergoing Studies
Summary
Current ACCP guidelines do not recommend routine screening with sputum, CXR or CT for lung cancer
A recent Cochrane meta-analysis shows that most trials did not compare screening vs. no screening but only the type/frequency of screening.
More frequent CXR screening and addition of sputum did not improve all cause mortality but may improve lung cancer specific survival at 5 yrs.
Several large RCT’s are underway that compare screening with CT to no screening
Mr. JF is a 52 year old man with hypertension and a 30 pack year smoking history. In addition to smoking cessation is there anyway to decrease his mortality from lung cancer through screening?
Lung cancer is:
- #1 cancer killer in men and women
- Poor prognosis of 85-90% case fatality rate
- Most present with advanced stage disease
Can mass screening lower fatality through earlier detection of localized disease?
National Guidelines Clearinghouse:
- “Screening for lung cancer: ACCP evidence-based clinical practice guidelines” (2003).
- We do not recommend that low-dose helical CT be used […]except in the context of a well-designed clinical trial. Grade of recommendation, 2C
- We recommend against the use of serial chest radiographs[...]. Grade of recommendations, 1A
- We recommend against the use of single or serial sputum cytologic […]. Grade of recommendation, 1A
“Screening for Lung Cancer” (2010)
- Analyzed 7 major trials
- Conducted in 1970’s-1980s worldwide
- Population: mixed but most male smokers>45yo
- Intervention: frequent CXR, sputum cytology
- Comparison: less frequent CXR +/- sputum
- Outcomes:
- 1.lung cancer specific survival
- 2.lung cancer specific mortality
- 3.overall survival
- Czech Study, RCT, 1976-1982
- Erfurt (German) Study, controlled-non randomized, 1972-1977
- JHH Study, RCT, 1973-1978
- Kaiser Study, RCT, 1964-1980
- Mayo Study, RCT, 1971-1976
- Sloan Kettering Study, RCT, 1974-1978
- North London Study, Cluster Randomized Trial, 1960-1964
- Czech: Males 40-64, current smokers with greater than 20 pack-years hx. Expected to live and functionally participate for 5 yrs.
- Erfurt: Males 40-65 living in Erfurt. 41k in intervention and 102K in control.
- JHH: Males >45, smokers (>1pack/day) near Baltimore, recruited through mail ads.
- Kaiser: M&F 35-54, of which only ~17% smoke, members of Kaiser Permanente Health Plan.
- Mayo: Males >45 recruited from Mayo Outpatient practice.
- MSKCC: Male smokers >45
- N.London: Males>40, working in industrial firms in N.London
| Name | Control Arm | Intervention Arm | Screening Duration |
| N. London | CXR before and after study | CXR before and after study and CXR q 6 ms | 3 Yrs |
| MSKCC | Annual CXR | Annual CXR + Sputum q4 ms | 5 yrs |
| Mayo | Annual CXR/Sputum | CXR/Sputum q 4ms | 6 yrs |
| Kaiser | Routine Care (Annual Physical+ CXR) | Additional Encouragement to undergo routine care | ? |
| JHH | Annual CXR | Annual CXR + Sputum q4 ms | 5yrs |
| Erfurt | CXR q 18 month | CXR q 6 month | 5ys |
| Czech | One CXR/Sputum at study termination | CXR/Sputum q 6 month | 3yrs |
| Czech | After initial 3 yrs, another 3 years of CXR for both | 3yrs | |
Results
Critiques of Methodology
| Name | Assignment Random | Allocation Concealed | Blinding of Death Assessment | Incomplete Data Addressed | No Other Bias |
| N. London | Y | ? | ? | Y | Baseline differences b/w Pt groups |
| MSKCC | Y | Y | Y | Y | Y |
| Mayo | Y | ? | Y | Y | Y |
| Kaiser | N | ? | Y | N | Baseline differences b/w Pt groups |
| JHH | Y | ? | Y | ? | Y |
| Erfurt | N | N | ? | Y | Y |
| Czech | Y | ? | ? | N | Pt Baseline data not fully provided |
Discussion
More frequent CXR vs. Less frequent CXR
- 5 yr lung cancer survival. Small benefit
- 5 yr lung cancer mortality. Same/?Harm
- 5yr all cause mortality. Same
Annual CXR/4m Sputum vs. Annual CXR Alone
- 5 yr lung cancer survival. Small benefit
- 5 yr lung cancer mortality. Small benefit
- 5yr all cause mortality. Same
Definitions:
- Lung cancer survival: alive or died from non-lung cancer cause
- Lung cancer mortality: died from lung cancer
- All cause mortality: died for any reason
No study addressed whether screening is better than no screening
Survival results were most heterogeneous. Survival can be confounded by lead-time, length time and overdiagnosis bias.
More frequent CXR leading to both increase in disease specific mortality and disease specific survival in pooled data further suggest unreliability of survival as outcome.
Increased CXR was shown to actually increase cancer mortality in several studies.
CXR unlikely to cause increased mortality per se due to low radiation dose but may lead to unnecessary surgery and early diagnosis that can lead to depression.
o
Contamination (control group pts received intervention) and compliance (intervention group pts not receiving intervention) decrease effect of screening
CXR does not detect small tumors whose removal may have the most benefit to pts.
Recent large uncontrolled trial of spiral CT showed 92% of lung cancers dx were stage I, with those undergoing resection having a 10yr survival of 85%.
CT Lung screening associated with 3x increase in lung cancer dx and 10x increase in surgery.
Current Undergoing Studies
| Name | Type | Population | Control Arm | Intervention Arm | Start date |
| NELSON (Dutch) | Multicenter RCT, parallel grp, no blinding | 47-75 current smokers or quit <10yrs goal="15K | Smoking cessation advice | Chest CT at year 4, sputum, blood tests, PFTs, smoking cessation | 2003 |
| NLST (US) | Multicenter RCT, parallel grp | Current or former smokers 55-74 goal=50K | Annual CXR for 3 yrs | Annual Chest CT for 3 yrs | 2002 |
| PLCO (US) | Multicenter RCT, parallel grp | Males and females 55-74 | ? | “Annual Chest Radiography” * | 1992 |
Summary
Current ACCP guidelines do not recommend routine screening with sputum, CXR or CT for lung cancer
A recent Cochrane meta-analysis shows that most trials did not compare screening vs. no screening but only the type/frequency of screening.
More frequent CXR screening and addition of sputum did not improve all cause mortality but may improve lung cancer specific survival at 5 yrs.
Several large RCT’s are underway that compare screening with CT to no screening
Thursday, September 16, 2010
Arcus Senilis - What does it mean?
Motivation: A few years ago, before I entered medical school, I was volunteering at an inner-city health clinic for the homeless and thought that everyone over fifty was developing cataracts. So many people had this white ring around their iris. Since then, I have been disillusioned and learnt about arcus senilis. Last month, however, arcus senilis again came to my attention. A resident and I were evaluating a patient in ED with suspicious chest pain, and the resident talked about how the arcus in the context of the patient's history suggested underlying vascular disease. So, I wondered, how predictive is arcus for atherosclerosis?
As introduction, corneal arcus is a lipid-rich deposit at the junction of the cornea and sclera. The corneal arcus lipid deposition is thought to share some similarity with lipid deposition in atherosclerosis.
Paper: Relation of Corneal Arcus to Cardiovascular Disease (from the Framingham Heart Study Data Set). Fernandez, A. B., et. al. The American Journal of Cardiology 2008 (103): 64-66.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2636700/?tool=pubmed
Method: The Framingham study is a prospective study initiated in 1948 to study factors influencing incidence of cardiovascular disease (CVD). Based on initial evaluation of arcus, the paper determined the predictive value of arcus for CVD after four and eight years. The total cohort examined in the paper consisted of 23,376 patients for four year prediction study and 13,469 for eight year study.
Results:
Unadjusted: Just the presence of arcus was predictive of first cardiovascular event with a hazard ratio of 2.28 (2.02-2.57) at 4 years and 2.52 (2.15-2.95) at 8 years.
Age and gender adjusted: Since increased age and male gender correlate with CVD, the authors adjusted for age and gender. After adjustment, arcus was predictive of events with hazard ratio of 1.07 (0.95-1.22) at 4 years and 1.18 (0.99-1.39) at 8 years.
Multivariate adjustment: Of course, there are more known factors influencing CVD than just age and gender. The authors also modeled the data after adjusting for age, gender, cholesterol, blood pressure, diabetes, smoking, and BMI. With multivariable adjustment, arcus was predictive of events with hazard ration of 1.04 (0.92-1.18) at four years and 1.14 (0.96-1.35) at eight years.
Conclusion: After adjusting for age and gender, arcus lost independent predictive value for CVD. The most likely explanation for the unadjusted association of arcus with CVD is that older people are more likely to have both arcus and vascular disease. This was the largest study examining the association of arcus with CVD.
Other studies in the past have linked arcus as an independent predictor of CVD. A major difference in the Framingham study is that detailed ophthalmological exams were not done, and arcus was assigned by visual inspection. So, while detailed examination of arcus may have some underlying predictive value, just visually finding arcus is not more predictive of CVD than just age and gender!
As introduction, corneal arcus is a lipid-rich deposit at the junction of the cornea and sclera. The corneal arcus lipid deposition is thought to share some similarity with lipid deposition in atherosclerosis.
Paper: Relation of Corneal Arcus to Cardiovascular Disease (from the Framingham Heart Study Data Set). Fernandez, A. B., et. al. The American Journal of Cardiology 2008 (103): 64-66.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2636700/?tool=pubmed
Method: The Framingham study is a prospective study initiated in 1948 to study factors influencing incidence of cardiovascular disease (CVD). Based on initial evaluation of arcus, the paper determined the predictive value of arcus for CVD after four and eight years. The total cohort examined in the paper consisted of 23,376 patients for four year prediction study and 13,469 for eight year study.
Results:
Unadjusted: Just the presence of arcus was predictive of first cardiovascular event with a hazard ratio of 2.28 (2.02-2.57) at 4 years and 2.52 (2.15-2.95) at 8 years.
Age and gender adjusted: Since increased age and male gender correlate with CVD, the authors adjusted for age and gender. After adjustment, arcus was predictive of events with hazard ratio of 1.07 (0.95-1.22) at 4 years and 1.18 (0.99-1.39) at 8 years.
Multivariate adjustment: Of course, there are more known factors influencing CVD than just age and gender. The authors also modeled the data after adjusting for age, gender, cholesterol, blood pressure, diabetes, smoking, and BMI. With multivariable adjustment, arcus was predictive of events with hazard ration of 1.04 (0.92-1.18) at four years and 1.14 (0.96-1.35) at eight years.
Conclusion: After adjusting for age and gender, arcus lost independent predictive value for CVD. The most likely explanation for the unadjusted association of arcus with CVD is that older people are more likely to have both arcus and vascular disease. This was the largest study examining the association of arcus with CVD.
Other studies in the past have linked arcus as an independent predictor of CVD. A major difference in the Framingham study is that detailed ophthalmological exams were not done, and arcus was assigned by visual inspection. So, while detailed examination of arcus may have some underlying predictive value, just visually finding arcus is not more predictive of CVD than just age and gender!
Friday, September 3, 2010
Temperature Measurement - Right way?
Motivation: In the clinic, many patients come with "high fever" - one patient insisted that at home he consistently measured his temperature at 105.0 although we found his fever to be no higher than 100.0. What happened? I wonder now whether we were using the same measurement techniques. Talking to attendings, I have found that almost all favor or dislike one or more methods of temperature measurement. The following paper compares some commonly used measurement techniques of core body temperature to the gold standard - pulmonary artery catheter temperature measurement.
Paper: Accuracy and Precision of Noninvasive Temperature Measurement in Adult Intensive Care Patients. Lawson, L. et. al. Am. J. Crit. Care, 2007 (16): 485-496. http://ajcc.aacnjournals.org/cgi/content/abstract/16/5/485
Methodology: The authors collected temperature by pulmonary artery catheter (PAC), axillary, temporal artery, tympanic membrane, and oral techniques. The four external measurements and PAC temperature were collected within a minute of each other. Sequential temperature measurements using all techniques were taken three times at twenty minute intervals to analyze intra and inter-method variability and concordance.
Patient Selection: Sixty adults in ICU (40 male and 20 female) with cardiopulmonary disease and pulmonary artery catheter. Patients were excluded if they had oral pathology, head trauma, or not visible tympanic membrane.
Results:
PAC vs oral - On average, oral measurement underestimated PAC temperature by about 0.09°C (0.16°F). The precision (reproducibility) was 0.43°C (0.77°F). 19% of measurements were more than 0.5°C (0.9°F) different from PAC. Oxygen delivery via nasal cannula did not make a clinical difference in temperature measure, but intubated patients consistently had higher oral measurements.
PAC vs tympanic membrane - On average, tympanic membrane overestimated PAC temperature by 0.36°C (0.65°F). Precision was 0.56°C (1.0°F). 49% of measurements were more than 0.5°C (0.9°F) different from PAC.
PAC vs temporal artery - On average, temporal artery overestimated PAC temperature by 0.02°C (0.04°F). Precision was 0.47°C (0.85°F). Administration of vasopressor did not significantly alter concordance. 20% of measurements were more than 0.5°C (0.9°F) different from PAC.
PAC vs axillary - On average, axillary underestimated PAC temperature by 0.23°C (0.41°F). Precision was 0.44°C (0.79°F). 27% of measurements were more than 0.5°C (0.9°F) different from PAC.
Conclusion: On average, oral and temporal artery measurements are likely good estimates of core body temperature. Axillary temperature is probably next on the list followed last by tympanic membrane measure, in which 49% of measurements differed by more than 0.5°C from core body temperature. Another point to take home is that for any technique, changes of about 0.5°C - the precision level of almost all the techniques - can be explained simply by measurement variability. A final point is that even for the best non-invasive techniques like oral measurement, about 20% of the time, the temperature will be off by 0.5°C or higher.
Limitations: The major limitation in this paper is that only three patients were actually febrile. The concordance rates may differ with febrile patients. Also, all of the patients were in the ICU. Perhaps, in an outpatient setting, the results may vary. Finally, the measurements were taken by experienced ICU nurses. The accuracy and precision of measurements by medical students or by patients may be a whole different story.
Paper: Accuracy and Precision of Noninvasive Temperature Measurement in Adult Intensive Care Patients. Lawson, L. et. al. Am. J. Crit. Care, 2007 (16): 485-496. http://ajcc.aacnjournals.org/cgi/content/abstract/16/5/485
Methodology: The authors collected temperature by pulmonary artery catheter (PAC), axillary, temporal artery, tympanic membrane, and oral techniques. The four external measurements and PAC temperature were collected within a minute of each other. Sequential temperature measurements using all techniques were taken three times at twenty minute intervals to analyze intra and inter-method variability and concordance.
Patient Selection: Sixty adults in ICU (40 male and 20 female) with cardiopulmonary disease and pulmonary artery catheter. Patients were excluded if they had oral pathology, head trauma, or not visible tympanic membrane.
Results:
PAC vs oral - On average, oral measurement underestimated PAC temperature by about 0.09°C (0.16°F). The precision (reproducibility) was 0.43°C (0.77°F). 19% of measurements were more than 0.5°C (0.9°F) different from PAC. Oxygen delivery via nasal cannula did not make a clinical difference in temperature measure, but intubated patients consistently had higher oral measurements.
PAC vs tympanic membrane - On average, tympanic membrane overestimated PAC temperature by 0.36°C (0.65°F). Precision was 0.56°C (1.0°F). 49% of measurements were more than 0.5°C (0.9°F) different from PAC.
PAC vs temporal artery - On average, temporal artery overestimated PAC temperature by 0.02°C (0.04°F). Precision was 0.47°C (0.85°F). Administration of vasopressor did not significantly alter concordance. 20% of measurements were more than 0.5°C (0.9°F) different from PAC.
PAC vs axillary - On average, axillary underestimated PAC temperature by 0.23°C (0.41°F). Precision was 0.44°C (0.79°F). 27% of measurements were more than 0.5°C (0.9°F) different from PAC.
Conclusion: On average, oral and temporal artery measurements are likely good estimates of core body temperature. Axillary temperature is probably next on the list followed last by tympanic membrane measure, in which 49% of measurements differed by more than 0.5°C from core body temperature. Another point to take home is that for any technique, changes of about 0.5°C - the precision level of almost all the techniques - can be explained simply by measurement variability. A final point is that even for the best non-invasive techniques like oral measurement, about 20% of the time, the temperature will be off by 0.5°C or higher.
Limitations: The major limitation in this paper is that only three patients were actually febrile. The concordance rates may differ with febrile patients. Also, all of the patients were in the ICU. Perhaps, in an outpatient setting, the results may vary. Finally, the measurements were taken by experienced ICU nurses. The accuracy and precision of measurements by medical students or by patients may be a whole different story.
Monday, August 23, 2010
Ibuprofen or Tylenol for Fever - Surprise
Sorry for the long delay in writing a blog post. From now on, expect a weekly edition of Siriasis .
Motivation: On the inpatient service, when someone has fever, the first line of symptomatic treatment is administration of acetaminophen. This summer, however, when I was febrile for a few days with a viral illness, I found that I had better control of fever with ibuprofen than with acetaminophen. I wondered is acetaminophen really better than ibuprofen? What are the data?
Paper: Efficacy and Safety of Ibuprofen and Acetaminophen in Children and Adults: A Meta-Analysis and Qualitative Review. Pierce, C.A., The Annals of Pharmacotherapy (2010) Vol. 44: 489-506.
Type of Study: A meta-analysis containing randomized controlled trials studies that directly compared ibuprofen to acetaminophen and provided comparative safety data. The authors analyzed pediatric and adult populations separately. While analyzing adverse events, the authors excluded expected side-effects such as GI disturbance for ibuprofen and mild liver enzyme abnormalities for tylenol from "serious" side effects.
Results:
Analgesic efficacy in adults: Out of 36 studies included, 26 concluded that ibuprofen was superior to acetaminophen. No study showed acetaminophen superiority. The overall effect size was medium (standardized mean difference of 0.69(CI: 0.57 to 0.81)).
Antipyretic efficacy in adults: Of the five studies, three concluded that ibuprofen was superior while two found no difference. Of note, not all of these studied infectious fevers - some studied fever caused by iatrogenic sources like interferon injections.
Antipyretic efficacy in children: Meta-analysis of seven trials concluded that fever control at four hours is significantly better with ibuprofen compared to acetaminophen - the effect size is relatively small (standardized mean difference of 0.26 (0.10 to 0.41)).
Adverse events: When taken as directed in adults and children, the odds of suffering at least one adverse event is not significantly different between ibuprofen to acetaminophen. When taken as directed, very few serious adverse events occur in either arm.
Limitations of Data: There are a number of limitations in the meta-analysis. To me, the chief one is that no single model of fever is considered. Rather, the efficacy of treating fever caused by diverse causes is lumped together. While potentially powerful in one sense since you can apply the data to fever from any source, we don't know if subgroups benefit differently. Also, for adverse events, excluding GI side-effects from ibuprofen group may have changed the results.
Conclusion: For fever, good data do not exist for adults, but in the pediatric population and extrapolating from this population, ibuprofen is likely a better anti-pyretic compared to acetaminophen. As an analgesic, ibuprofen is better for adults. Also, excluding expected side effects for the drugs, acetaminophen and ibuprofen do not carry additional toxicity. My conclusion is that in the outpatient setting if patient does not have history of serious risk factors for GI bleeding, better fever control and analgesia are likely achieved by prescribing time limited doses of ibuprofen.
Link to paper: http://www.theannals.com/cgi/content/full/44/3/489
Motivation: On the inpatient service, when someone has fever, the first line of symptomatic treatment is administration of acetaminophen. This summer, however, when I was febrile for a few days with a viral illness, I found that I had better control of fever with ibuprofen than with acetaminophen. I wondered is acetaminophen really better than ibuprofen? What are the data?
Paper: Efficacy and Safety of Ibuprofen and Acetaminophen in Children and Adults: A Meta-Analysis and Qualitative Review. Pierce, C.A., The Annals of Pharmacotherapy (2010) Vol. 44: 489-506.
Type of Study: A meta-analysis containing randomized controlled trials studies that directly compared ibuprofen to acetaminophen and provided comparative safety data. The authors analyzed pediatric and adult populations separately. While analyzing adverse events, the authors excluded expected side-effects such as GI disturbance for ibuprofen and mild liver enzyme abnormalities for tylenol from "serious" side effects.
Results:
Analgesic efficacy in adults: Out of 36 studies included, 26 concluded that ibuprofen was superior to acetaminophen. No study showed acetaminophen superiority. The overall effect size was medium (standardized mean difference of 0.69(CI: 0.57 to 0.81)).
Antipyretic efficacy in adults: Of the five studies, three concluded that ibuprofen was superior while two found no difference. Of note, not all of these studied infectious fevers - some studied fever caused by iatrogenic sources like interferon injections.
Antipyretic efficacy in children: Meta-analysis of seven trials concluded that fever control at four hours is significantly better with ibuprofen compared to acetaminophen - the effect size is relatively small (standardized mean difference of 0.26 (0.10 to 0.41)).
Adverse events: When taken as directed in adults and children, the odds of suffering at least one adverse event is not significantly different between ibuprofen to acetaminophen. When taken as directed, very few serious adverse events occur in either arm.
Limitations of Data: There are a number of limitations in the meta-analysis. To me, the chief one is that no single model of fever is considered. Rather, the efficacy of treating fever caused by diverse causes is lumped together. While potentially powerful in one sense since you can apply the data to fever from any source, we don't know if subgroups benefit differently. Also, for adverse events, excluding GI side-effects from ibuprofen group may have changed the results.
Conclusion: For fever, good data do not exist for adults, but in the pediatric population and extrapolating from this population, ibuprofen is likely a better anti-pyretic compared to acetaminophen. As an analgesic, ibuprofen is better for adults. Also, excluding expected side effects for the drugs, acetaminophen and ibuprofen do not carry additional toxicity. My conclusion is that in the outpatient setting if patient does not have history of serious risk factors for GI bleeding, better fever control and analgesia are likely achieved by prescribing time limited doses of ibuprofen.
Link to paper: http://www.theannals.com/cgi/content/full/44/3/489
Subscribe to:
Posts (Atom)