Saturday, September 6, 2008

August Journal Club and Farewell Dr. Diner

What a terrific turn-out for the August Journal Club and farewell fiesta for Dr. Diner. Great to see so many folks there to wish off one of our own. Thanks to Dr. Ander for hosting, and subjecting his house to all of us.

On to the articles:

1) "Utility of Routine Testing for Patients With Asymptomatic Severe Blood Pressure Elevation in the Emergency Department," Karras, D, et. al. Annals of Emergency Medicine 51(3) March 2008

A prospective, observational study examining the benefit of standard recommended testing on asymptomatic patients with hypertension. The authors for this study include my mentor from residency, as well as our own current department chair and program director!
Design: A prospective, observational study of patients at three urban EDs with sustained blood pressures greater than or equal to 180 mmHg systolic or 110 mmHg diastolic. Any patient with symptoms which could even remotely be tied to their hypertension were excluded - the authors wanted truly asymptomatic patients. A panel of tests including basic metabolic panel, urinalysis, ECG, CBC count, and chest radiograph were obtained, though not every patient got every test. Treating physicians were then surveyed as to their reasons for performing the tests, and as to whether the results were "clinically meaningful." Clinically meaningful results were defined as those leading to unanticipated hospitalization, medication modification, or further immediate evaluation. The primary outcome was the prevalence of unanticipated clinically meaningful test abnormalities.
Results: 109 patients were enrolled (out of 409 screened). Approximately 50% of these patients had "unexpected" test results, though the article did not characterize these results. Only 7 (6%) had "clinically meaningful" abnormal test results, 2 of which were thought to be very unlikely to be related to the patients' blood pressure. The clinically meaningful outcomes were as follows:

click image to enlarge

Discussion points: We REALLY wanted this article to give us an answer as to how to work-up the patient with severe, but asymptomatic, hypertension. Unfortunately, the group felt that it fell short in this goal. While the authors calculated that they would need 300 patients to obtain adequate power, and that they would need to screen 400 patients for this, they were only ably to enroll 109 subjects. While this low number was likely due to their rigorous definition of "asymptomatic" it lead to a much smaller study than anticipated. This small size may have limited the applicability of the results. Also, 1 in 20 patients was found to have an abnormality requiring alteration in management, with no easily identifiable way to recognize them up front (eg, there was no correlation with underlying cardio- or renal-vascular disease). It was discussed that given the nature of the population studied (largely medically under served) these abnormalities might represent "baseline noise" rather than consequences of hypertension. In other words, if we were to screen every patient with theses tests, we would likely find a not insignificant number with similar abnormalities.
Overall, this article will not change any of our practice. Most faculty agreed that there is a blood pressure cut-off, above which they will screen for end-organ damage, but there was no consensus as to what tests to order. This magic number seemed to be about 180-200/110-120 for most people, but a very few voiced the opinion that in truly asymptomatic patients, they would do nothing. Generally, it was felt that these patients should have their renal function checked, but this still remains controversial.
2) "The Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Compared With Stress Nuclear Imaging in Emergency Department Low-Risk Chest Pain Patients," Gallagher, M,. et. al. Annals of EMergencncy Medicine, 2006.
Yet another study with an Emory author, this article compared standard stress tests to CT angiography for low-risk chest pain patients in an ED observation unit.
Design: Prospective study of patients with chest pain felt to be low-risk by the Reilly/Goldman criteria who had negative serial ECGs and cardiac markers. The goal was to identify significant coronary artery disease (> 70% stenosis) or adverse cardiac event within 30-days of evaluation. Patients recieved both nuclear sestamibi stress testing and CT coronary angiography. Abnormal stress results (perfusion defects) or CT results (> 50% stenosis or calcium scopre > 100) were considered for cardiac catheterization, and all patients had 30-day follow-up for adverse cardiac events. Primary outcomes were the accuracy of multidetector CT and myocardial perfusion imaging for the detection of an acute coronary syndrome and 30-day major adverse cardiac events (defined as an acute coronary syndrome, [STEMI, non-STEMI, or unstable angina], the development of new Q waves on subsequent ECGs, new congestive heart failure or cardiogenic shock, major dysrhythmias (high-grade atrioventricular block, ventricular tachycardia, ventricular fibrillation), cardiac arrest, or death from an acute coronary syndrome.)
Results: 96 patients were enrolled who met the following enrollment criteria:
- no ECG evidence of acute infarction or ischemia (including new left bundle branch block)
- no pain that was worse than usual angina or like a previous myocardial infarction
- no recent revascularization
- no rales above both bases
- systolic blood pressure that was greater than 110 mm Hg
- negative initial chest radiograph
- negative initial set of cardiac markers (CK-MB and troponin I)
Of these patients, 4 were later excluded to to protocol violations, and 7 had uninterpretable CT results due to poor image quality. 85 patients were included in the study, 7 (8%) of whom were identified to have significant coronary disease, but none of whom had an MI or adverse cardiac event on 30-day follow-up. The investigators were able to perform 30-day follow-up on 83 of the enrolled patients,
The statistical calculations in this paper were complex, and I am not going to reproduce the number here. Generally, CTA had a higher sensitivity, specificity, and predictive value than stress testing, though 13 (15%) of patient had discordant results. 5 of these received a cardiac catheterization, 3 with significant stenosis. None of the other 8 (all with positive CTA and negative stress tests) had a major cardiac adverse event in the 30 day period.
Discussion points: This was a very complex paper which was chosen to spark conversation as to the role of CT coronary angiography in the ED.
The biggest flaw in the paper is the number of patients with the "target condition": 7! It is very difficult to conclusively evaluate any test when only 7 patients actually have the disease being studied. When you apply a test to a population with such a low disease prevalence you are very unlikely to get meaningful results. The group felt that there might in fact be a role for CTA in the future, but that more literature was necessary.
Other points:
1) Calcium scoring is a measure of calcification of the coronary vessels on CT scan, thought to be a surrogate measure of coronary artery disease. This is only useful in older patients, as young people typically have non-calcified CAD.
2) The size of a coronary lesion does NOT predict its propensity for rupture with thrombus formation, and therefore is not indicative of the likelihood for causing acute MI. More important is the stability of the plaque, determined by the amount of lipid core versus fibrous cap. The thinner the cap, the more unstable, and the higher likelihood of rupture. No current risk stratification tool, including cardiac catheterization, can identify this. Experimental procedures including intra-coronary ultrasound may provide this information, but we are not there yet.
Overall, we are unsure as to the role of coronary CTA in the ED. This controversial topic will likely get a lot more study, and hopefully a reasonable role will be found.

Monday, August 4, 2008

July Journal Club

Wow. July already. The new Interns have jumped in full speed, and the 2's and 3's are making smooth transitions to their new roles.

Since I hosted this month, I make no comments on the quality of the food, though it seemed to disappear fairly quickly, so I assume it was palatable.

Two interesting, and controversial, topics this time around.

1) "Safety and Clinical Effectiveness of Midazolam versus Propofol for Procedural Sedation in the Emergency Department: A systematic Review," Hohl, CM, et. al. Academic Emergency Medicine 15(1) January 2008

A meta-analysis comparing the safety and efficacy of two commonly used medications for emergency department procedural sedation.

Design:
A systematic review of randomized controlled trials (RCTs) as well as observational studies that either directly compared or used either propofol or midazolam for emergency department procedural sedation. Articles were found by searching multiple medical databases as well as the indices of three major emergency medicine journals. All abstracts were manually reviewed by one of the investigators, and the following were excluded: commentaries, reviews, surveys, editorials, case reports, letters, and pediatric studies. The investigators identified 4 head-to-head RCTs, 9 RCTs using either agent, and 15 observational studies which were analyzed. The primary outcome was defined as the proportion of major adverse events (death, disability, hospital admission, prolonged ED stay, intubation, and vomiting with aspiration pneumonitis) and the secondary outcome was the mean difference in the proportion of patients successfully sedated so the procedure could be completed. The studies were evaluated for quality and heterogeneity by the investigators.

Results:
Overall, the quality of the included studies was not great. There was poor reporting of methods, and no consistent dose of the study drugs. Minor adverse events were not constantly reported by the studies, and could not be easily analyzed. Overall estimates of major adverse events were as follows:
Midazolam: 0/847 patient exposures (CI 0% - 0.45%)
Propofol: 1/2,453 patient exposures (CI 0.01% - 0.23%)
Regarding the secondary outcome of procedural sedation success, combined data indicated a non-significant trend towards greater success with propofol. The authors conclude that based on the limited available published data, both agents appear safe and effective for procedural sedation, but there is definitely room for a large, well designed randomized controlled trial.

Discussion points:
Overall, the group agreed that multiple agents are reasonable choices for emergency department procedural sedation. We discussed the two different types of sedation: moderate (where airway reflexes are preserved) and deep (where airway reflexes are lost) and how different situations may cause for different levels of sedation. We discussed both the advantages and disadvantages of the commonly used versed/fentanyl regimen, including the need for escalating doses and the easy availability of reversal agents. We also discussed some of the benefits and pitfalls of propofol, including its brief duration of action and its potential to cause hypotension. Finally, we discussed the use of etomidate (at half-dose for sedation) and ketamine for select populations. It is vitally important to know what is available in you clinical setting, and to be comfortable with any drug you want to use.

The upshot:
In the right hands, it is likely that many of the available procedural sedation agents are safe and effective. Knowledge and preparation are key, and practitioners must be familiar with the dose, time of onset, and duration of activity of any medications they are going to uses, as well as with the use of applicable reversal agents (thanks to Dr. Berger for this excellent point). In other words, know what you are going to use and how to use it before you push your meds!

Also, remember that the quality of studies which go into a meta-analysis determine the qulaity of the results that come out.


2) The CORTICUS study: "Hydrocortisone Therapy for Patients with Septic Shock," Sprung, CL, et. al. NEJM 358(2) January 10, 2008

A large, randomized study designed to identify the role of IV steroids in patients with septic shock. The issues raised by this study are EXTREMELY complex and controversial, and I will do my best to summarize them here.

Design:
Randomized, double-blind, placebo-controlled study at 52 participating ICUs comparing IV dexamethasone 50 mg every 6 hours for 5 days followed by a taper to placebo. Inclusion criteria "were clinical evidence of infection, evidence of a systemic response to infection, and the onset of shock within the previous 72 hours (as defined by a systolic blood pressure of <>a need for vasopressors for at least 1 hour) and hypoperfusion or organ dysfunction attributable to sepsis." Listed "exclusion criteria included underlying disease with a poor prognosis, a life expectancy of less than 24 hours, immunosuppression, and treatment with long-term corticosteroids within the past 6 months or short-term corticosteroids within the past 4 weeks." (Emphasis added by me - you'll see why later) The primary outcome measure was defined as death at 28 days in corticotropin non-responders. Secondary outcomes included "rates of death at 28 days in patients who had a response to corticotropin and in all patients, the rates of death in the ICU and in the hospital, the rates of death at 1 year after randomization, a reversal of organ system failure (including shock), and the duration of the stay in the ICU and the hospital."

Results:
499 patients were enrolled as compared to the estimated 800 needed to detect a statistically significant difference in the primary outcome. (Enrollment was topped early. Is this large enough?) There was no significant difference in the primary outcome of death at 28-days in corticotrpoin non-responders (39.2% with hydrocortisone, 36.1% in placebo). No significant difference was identified in mortality rate at any other measured time point in the study.

A number of interesting findings were identified in post-hoc analysis of the study data, some of which are listed below. Remember that post-hoc analyses are always somewhat suspect as these were NOT outcomes the study was designed to detect, but rather results derived by manipulating the patient groups and data variables:
1) Increased 28 day death rate for those given etomidate in both the steroid and placebo group.
2) Earlier reversal of shock in the hydrocortisone group
3) 11.2% reduction in mortality from steroids in patients with hypotension despite 24 hours of vasopressor therapy

Discussion points:
A number of potential problems were identified with this study, including substantial differences between the design and population of this and earlier steroid studies. First, the mortality rate and disease severity of the patients in CORTICUS was substantially lower than that in earlier studies which showed a significant mortality benefit. Second, in the present study, patients could receive steroids up to 72 hours after identification of septic shock, as opposed to 8 hours in the earlier study. Also, patients in the previous studies, patient needed to be in shock refractory to vasopressors to receive steroids, which was not a requirement in CORTICUS (and this sub-group had improved mortality in CORTICUS). Finally, patients who received steroids within the last 4-8 weeks were excluded in CORTICUS. One wonders whether this included patients who received them as a part of EGDT in the ER prior to admission.

The point about etomidate increasing mortality is something that has been discussed extensively in the literature recently. Please read the recent Clinical Controversies the July Annals of Emergency Medicine for two well presented perspectives. I think this question has yet to be answered.

The upshot:
The role of steroids in the treatment of septic shock remains as controversial as ever. There is likely a role for patients who remain hypotensive despite vasopressor therapy, but we have likely not yet identified the ideal target population. Keep your eyes peeled for upcoming studies re0evaluating the role of EGDT which may give us some better answers.

Thursday, July 10, 2008

June Journal Club

First JC of the new academic year with a STRONG showing by the interns! Some great burgers put together by our own residency director, and terrific conversation. Let's get to the articles.

1) "A Clinical Decision Rule for Cranial Computed Tomography in Minor Pediatric Head Trauma," Atabaki, SM, et. al. Archives of Pediatric and Adolescent Medicine. 162(5) May 2008

Yet another attempt to derive a reliable prediction rule for what kids bonked on the head need CT scans.

Design:
Prospective, observational study of a convenience sample of patients aged 0-21 years at 4 pediatric trauma centers. Patients were included if they had a closed head injury and a GCS greater than 13, and if the treating physician elected to obtain a head CT. Patients with intracranial injury (defined as subdural, epidural, subarachnoid, intrparenchymal or intraventricular hemorrhage, as well as contusion and cerebral edema) were identified and the study attempted to identify high risk clinical features. The secondary outcome measure was defined as the performance of any neurosurgical procedure.

Results:
1000 patients meeting the study criteria were enrolled. 65 (6.5%) of these had a positive finding on CT, and 6 of the 65 (9.2%, but only 0.6% of enrolled patients) required neurosurgical intervention.

The study identified eight features which were strongly associated with the presence of intracranial imaging(the primary outcome measure) through recursive partitioning. These are as follows:

1) GCS < class="blsp-spelling-error" id="SPELLING_ERROR_15">hematoma were also statistically associated with the primary outcome measure, but fell out on recursive partitioning.

Discussion points:
Overall, while this study raised some good points, the group did not feel it provided a reliable decision rule for pediatric head injury. One of the biggest flaws is in the number of patients. While 1000 patients were enrolled, there was a very low incidence of injury. Only 6.5% of the patients had an injury, and only 6 required neurosurgical intervention. This is a very small number of patients on which to base a decision rule.

Another flaw is the fact that only patients receiving CTs at the discretion of the treating physician were enrolled, raising the possibility of missing injuries in those not scanned. An alternative strategy would have been to CT all kids with minor CHI and use this population to define the decision rule.

The upshot:
We all agreed that deciding which kids to image is challenging, but that this study does not provide a reliable guide. While it did identify some important high risk features, especially age < 2 years, we do not feel comfortable relying on the decision rule.


2) "Efficacy and Safety of Recombinant Activated Factor VII for Acute Intracerebral Hemorrhage," Mayer, SA, et. al. The New England Journal of Medicine. 358(20) May 15, 2008

An attempt to identify a useful treatment for the otherwise vexing problem of intracranial hemorrhage.

Design:
Randomized, placebo-controlled trial where patients with intracranial hemorrhage were given either placebo, 20 mcg/kg, or 80 mcg/kg of rFVIIa within 4 hours after the onset of the bleeding. 819 patients at 122 sites around he world were analyzed, The primary outcome measure was defined as severe disability or death as determined by the modified Rankin scale at 90 days. Other outcome measures were also identified, including rate of hemorrhage growth and occurrence of thromboembolic events.

Results:
No statistically significant difference was found in any of the primary outcome measures.

Death rates: 18% in 20 mcg/kg group; 21% in 80 mcg/kg group; 19% in placebo group.
Poor outcome (as defined by a Rankin score of 5 or 6) rates were as follows:
26% in the 20 mcg/kg group, 30% in the 80 mcg/kg group, and 24% in the placebo group.

There was a decrease in the rate of hemorrhage expansion in the 80 mcg/kg group which reached statistical significance, but this had no effect on the primary outcome measures. Additionally, there as a significant increase in the number of arterial thrombo-embolic events (MI and CVA) in this group, but the total number of these events was low.

Discussion points:
Unfortunately, the study did not demonstrate any meaningful clinical improvement in patient outcomes. While there was some effect on the size of the bleeding, this did not translate into any improvement in death rate or Rankin score. This points out an important difference between "disease oriented outcomes" and "patient oriented outcomes." The former refers to changes in disease parameters; in this case, the rate of increase in patients' bleeding. The later refers to factored that effect patient functional or symptomatic outcome; in this case mortality and poor neurologic outcome. Patent oriented outcomes are typically more clinically important than disease oriented outcomes when considering the value of an intervention.

Interestingly, the results of this study are at odds with the authors' previously published phase 2b study which showed an overall reduction in mortality of 38% in patients receiving rFVIIa. While the authors attempted to identify reasons for this difference, we did not believe that they adequately explained this. It is important to remember that phase 2 studies are safety trials, and should not necessarily be relied on to provide definitive outcome results.

The upshot:
Clearly, this study did not demonstrate a meaningful clinical benefit from the administration of rFVIIa to patients with intracranial hemorrhage. I suspect we will see more studies in the future, but for now the role of rFVIIa is unclear.

Thursday, June 5, 2008

May 2008 Journal Club

A bit late, but here's the details from the May Journal Club. Dr. Hirsh blew us away with her culinary skills, and her awesome new kitchen. The articles were great too, but clearly took a back seat to the food!



1) Byyny, RL, et. al. Sensitivity of Noncontrast Cranial Computed Tomography for the Emergency Department of Subarachnoid Hemorrhage. Annals of Emergency Medicine 2008 (Article in press)

The article asks whether new multi-detector CT scanners are sensitive enough to rule out SAH without performing an LP.

Design:
Retrospective chart review of patient diagnosed in the ED with SAH at a tertiary care academic medical center with an annual volume of 40,000. Patients were identified by ICD-9 codes, and charts were reviewed to see whether they had a CT or LP in the ED. All patients with SAH were scanned with a 4-detector CT scanner.

Results:
149 patients with ED diagnosis of SAH were identified. Non-contrast CT identified 139 of these, yielding a sensitivity of 93% (95% CI 88% to 97%). Of patients with diagnosed SAH and a normal mental status, CT was 91% sensitive, whereas it missed only 1/61 in patients with an abnormal mental status. The authors conclude that a normal non-contrast CT scan is insufficient for ruling out SAH in the ED population.

Discussion points:
The group agreed that while this studies has some flaws, it does demonstrate that CT scanning alone is not sufficient to rule out SAH. The overall sensitivity of 90-95% is similar to the performance of early generation CT scanners, and we agreed that lumbar puncture still needs to be performed when evaluating for subarachnoid bleeding.

One concern with this paper is that it enrolled only patients with diagnosed SAH, which likely missed a substantial number of people with headache discharged after a normal head CT. It is possible, then, that there were patients discharged from the ED with a missed SAH as they did not have further testing performed after a negative CT. This would lower the sensitivity of the scan even further.

Another question raised by the group was whether the sensitivity would improve with a 16-, 32-, or 64-slice scanner. I suspect we will see more studies coming out addressing this question. Until then, we'll all still be doing LPs.

Article number 2 coming soon....

Thursday, April 24, 2008

April Articles

Another GREAT Journal Club last night at Brit's house, with terrific food and lively conversation. Two articles on HIV testing in the ED were explored.

I) Haukoos, JS, et. al. Patient acceptance of rapid HIV testing practices in an urban emergency department: Assessment of the 2006 CDC recommendations for HIV screening in health care settings. Annals of Emergency Medicine. 51(3) March, 2008.

This article sought to ask the question "Will patients in the ED accept the CDC recommendation for routine HIV screening in the emergency department?"

Design:
Cross-sectional survey of a convenience sample of patients at an urban ED in Denver. Patients were given a 6-item questionnaire, which attempted to assess their willingness to accept HIV screening in the ED. No actual HIV testing was performed. Specific questions:
1) Opt-out (test UNLESS patient declines) vs Opt-in (test ONLY if patient agrees)
2) Does physician recommendation influence decision?
3) Do patients want pre- and post-test counseling?
4) Do patients want a separate consent?

Results: 529 (8% of eligible patients) enrolled.
1) 81% would accept testing with either the opt-out or opt-in strategy, but 11% of patients needed an explanation of the meaning of "opt-out."
2) 93% would accept testing if recommended by their physician.
3) Roughly 35% felt pre-test counseling was necessary, and a similar number would want counseling after a NEGATIVE test. They did not ask patients whether they would want post-test counseling for a POSITIVE test result.
4) 50% felt that consent for HIV testing should be separate from general medical consent.

Discussion points:
Some major flaws were identified in this paper.

1) First, the patient enrollment process seems problematic. While the authors report that 8% of eligible patients were enrolled, they DO NOT tell us how many potential patients were approached and refused enrollment. If a large number of patients refused enrollment, this might skew the data by only including patients willing to participate. Would the patients who refused to enroll actually agree to testing? Who knows...

The authors also do not explain how patients were chosen. In fact, the demographics of their selected patients was substantially different from their ED population, with far fewer Hispanics. This raised concerns over whether investigators might have preferentially selected patients where there would be no perceived language barrier, and whether there might have been other biases in the selected patients.

2) No HIV testing was performed; study participants were simply asked whether they would agree to be tested without the actual potential consequences of being tested. It is possible that fewer patients would agree if they were actually going to be tested.

3) The questionnaire was unclear in its wording. Specifically, the question concerning the "opt-out" design did not actually present a true opt-out option, as it stated that testing would be "offered" unless patients declined. A true opt-out structure means that testing WILL be performed, unless the patient declines. While seemingly a small semantic difference, it was pointed out that surveys are all about semantics, and that any small inaccuracy could dramatically change results.

The upshot: The group agreed that while the study was deeply flawed, a majority of patients will probably be willing to accept HIV testing in the ED. Concerns were raised about funding and the cost-benefit of screening asymptomatic patients, as well as the ability to obtain follow-up for patients for either counseling or confirmatory testing. We all agreed that these were important issues, and it will be interesting to see how they play out.

Finally, we had a discussion on how to interpret statistical sensitivity and specificity, as well as positive and negative predictive values.

II) Lyss, SB, et. al. Detecting unsuspected HIV infection with a rapid whole-blood HIV test in an urban emergency department. Journal of Acquired Immune Deficiency Syndrome. 44(4) April 1, 2007.

This article sought to ask not only whether patients would agree to testing in the ED, but whether HIV infection rates would differ in patients who were enrolled by investigators versus those referred by a clinician.

Design:
Cross-sectional study of a convenience sample of patients at an urban emergency department in Chicago. Patients could be enrolled directly by research associates or by ED clinicians who felt they should be tested.

Results: 2824 (58%) eligible screened patients agreed to be tested. 414 (95%) provider-referred patients agreed to be tested.
35 (1.2%) screened patients and 48 (11.6%) provider-referred patients were HIV positive.
Of these, 18 (51%) screened patients and 24 (50%) referred patients reported no traditional risk factors.
27 (77%) screened patients and 38 (79%) referred patients entered HIV care.
Of those with CD4 cell counts available, 14 (45%) of 31 screened patients and 37 (82%) of 45 provider-referred patients had counts <200.

Discussion points:
A few concerns were raised about this study which were similar to those of the previous article. We discussed the advantages and disadvantages of a convenience sample, and it was pointed out that for an epidemiological study, convenience samples are poor ways of obtaining a representative sample. Concerns were raised as to whether patient seen at different times might have different responses to being tested. Specific variables mentioned included patient financial and insurance status, wit time prior to being seen by a physician, and acuity level, all of which might differ depending on when the patient presented to the ED. We were not able to answer whether this might skew the data, but the possibility was raised.

While the follow-up rates seem quite high, the article does not specify what percentage of patients who followed-up were admitted to the hospital versus discharged from the ED. It was speculated by the group that a large percentage of those who followed-up may have been admitted, especially considering how similar the number of admitted patients was to the number who made their follow-up appointments.

The upshot:
The group agreed that having rapid HIV testing as an option would be beneficial, and might alter management on a number of patients we see in the ED. Again, significant questions still exist as to the funding for and cost-benefit of routine screening, as well as to the ability for patients to get appropriate follow-up care.

Great job to all who attended! Who knows what next month will bring.....

Thursday, March 27, 2008

Welcome to the new Emory Journal CLub Blog

This blog will be used to provide the results of the monthly Emory Department of Emergency Medicine journal club discussions. Feel free to leave comments discussing the findings.