The rational use of a CT scan in the diagnosis of appendicitis

Click on any heading or subheading to rapidly navigate to the relevant section of this manuscript

For printing purposes, an adobe pdf file version of this manuscript is available by clicking here

The basic purpose of this opinion piece is to offer interested clinicians common sense advice regarding the value of an abdominal CT scan in suspected appendicitis patients. This is a very divisive issue and many clinicians have strong opinions regarding this matter. Some surgeons think that there is only a limited role for a CT scan, and they think that it should only be used if the presentation of appendicitis is atypical or if the clinical likelihood of appendicitis is low. Other clinicians think that an appendiceal CT scan should be used routinely in all suspected appendicitis patients -- irrespective of the pre-test probability of appendicitis. I will discuss the rationale for routinely using a CT scan in patients with suspected appendicitis, and I will also describe a few decision-making problems that can result from the unnecessary use, or incorrect interpretation, of an appendiceal CT scan. Most of the controversial medical decision-making problems that occur secondary to the use of a CT scan in patients with suspected appendicitis are due to false-positive (FP) and false-negative (FN) CT scan results. The problem of too many FP (relative to TP) CT scan results will most likely occur in suspected appendicitis patients who have a very low clinical probability of appendicitis. However, decision-making problems can also occur secondary to a clinician's inability to discriminate between true-negative (TN) and false-negative (FN) CT scan results, and this problem is most likely to occur if a CT scan is indiscriminately used in patients with a high clinical probability of appendicitis. A major section of this manuscript will be devoted to a description of my prescribed methodology of how best to identify possible FP CT scan results. I believe that a clinician could likely identify many FN and FP CT scan results by actually reviewing the radiologist's interpretative findings in a structured manner. I will dissect this particular issue in great detail and offer clinician-readers a number of useful recommendations that may help them decrease the number of medical decision-making errors due to FP and FN CT scan results.

This manuscript is mainly written for an audience of medical students and/or medical residents-in-training (emergency medicine residents, family practice residents or surgical residents). However, I strongly suspect that many experienced clinicians will also find some of the explicatory information clinically useful.

The potential value of an appendiceal CT scan in the diagnosis of appendicitis

Rao took 100 consecutive suspected appendicitis patients, who were due to be admitted to Massachusetts General Hospital (MGH) because their surgeons thought that they may have appendicitis. All those suspected appendicitis patients had to have a >20% likelihood of appendicitis in order to be admitted to hospital. All those patients had an appendiceal CT scan performed after the surgeon's clinical evaluation. 53% of those patients had appendicitis confirmed, which means that the average prevalence of appendicitis was ~50% in that study population of suspected appendicitis patients.

The surgeons had to estimate the clinical likelihood of appendicitis according to the following classification scheme (prior to the CT scan being performed):-

These are the results of the MGH surgeons' clinical impressions and the CT scan results as interpreted by the MGH radiologists.

How accurately could those Harvard surgeons estimate the likelihood of appendicitis based on their initial clinical examination?

First consider the high probability appendicitis patient group. That patient group was labelled "definitely appendicitis" and it was more precisely defined as suspected appendicitis patients who had a 80-100% clinical probability of appendicitis. That is the type of patient who would immediately be taken to surgery for an appendectomy by most general surgeons in community practice. The general surgeons would have presumably stated "if it looks, acts and quacks like a duck, then it must be a duck." However, note that appendicitis was only confirmed in 78% (18/23) of those patients, which means that the negative appendectomy rate would have been 22% if the MGH surgeons operated solely on the basis of their strong initial clinical impression. That 22% negative appendectomy rate figure is not very different to the "average" US negative appendectomy rate of 15-20%.

If you look at the "probably appendicitis" group, which was defined as a 60-79% probability of appendicitis, note that appendicitis was only confirmed in 56% of those patients. That suggests that many MGH surgeons tended to over-estimate the likelihood of appendicitis at the time of their initial clinical evaluation if they thought that the patient "probably" had appendicitis.

Consider the "possibly appendicitis" group of suspected appendicitis patients, who were presumed to have a 20-39% probability of appendicitis. Note that 43% (7/16) of those patients had appendicitis. That suggests that the MGH surgeons tended to under-estimate the likelihood of appendicitis if they thought that the suspected appendicitis patient only "possibly" had appendicitis.

In summary, there is substantial reason to believe that experienced surgeons cannot estimate the clinical likelihood of appendicitis with sufficiently great precision, and there is a potential advantage to using a diagnostic test of high sensitivity and high specificity to decrease the negative appendectomy rate.

How accurately could appendiceal CT scans diagnose appendicitis in that study population of 100 suspected appendicitis patients?

The radiologists were extraordinarily accurate in diagnosing appendicitis in that particular study. If the radiologists thought that the CT scan was definitely positive for appendicitis, then they were very accurate and there was only one FP result and 50 TP results. That means that if the MGH surgeons operated on any patient labelled "definitely appendicitis" on the basis of a positive CT scan result, then they would have only had a negative appendectomy rate of 2% (1/51). That result is considerably better than the 22% negative appendectomy rate that would have occurred if the surgeons operated on "definite" appendicitis patients solely on the basis of their initial clinical impression.

The MGH radiologists were also very accurate when it came to excluding the presence of appendicitis. They were 100% accurate in excluding appendicitis if they were confident that the patient "definitely" did not have appendicitis and there were no FN CT scan results in that subgroup of patients.

Those excellent CT scan results suggest that a CT scan should be used routinely in all suspected appendicitis patients because the CT scan had a very high sensitivity and specificity for diagnosing appendicitis (>95% sensitivity and >95% specificity). Theoretically, routine use of a CT scan, which has a very high sensitivity and very high specificity, should enable a surgeon to decrease his negative appendectomy rate from an "US-average" of 15-20% to <5%. However, many surgeons in community practice believe that an appendiceal CT scan does not help them decrease their negative appendectomy rate. In fact, they consider a CT scan a hindrance, rather than a help, in most patients with suspected appendicitis. Why?

I think that the reason why many community surgeons don't like to routinely use an appendiceal CT scan is because they are frequently "burnt" by FP and FN CT scan results. Why does this occur? The main reason is that community radiologists cannot achieve a >95% sensitivity and >95% specificity result when performing a CT scan on suspected appendicitis patients in community hospitals. The reasons are multi-factorial and include less than optimal CT scan techniques, technically inadequate CT scan studies, and imperfect radiological interpretative skills.

Consider an example of the performance of an appendiceal CT scan in community practice.

Consider the following study performed at Stamford Hospital, Connecticut [2]. The surgical department at Stamford Hospital examined all their non-incidental appendectomy results for a 15 month period from April 1999 to June 2000, and they also checked to see how many patients had a CT scan prior to surgery. They then calculated the accuracy of the CT scan results by using the findings at appendectomy as the "gold standard" test. They noted that 57 patients had a CT scan prior to appendectomy and that 46 of those 57 patients had appendicitis confirmed at surgery. Of those 46 patients, 44 patients had a positive CT scan (and they represent TPs) and 2 patients had a negative CT scan (FNs). 11 patients had a normal appendix (negative appendectomy rate of 19% in patients who had a pre-operative CT scan) and 9 of those 11 patients had a positive CT scan (FPs). They therefore concluded that a CT scan was not helpful, because they calculated the sensitivity and specificity of the CT scan as follows:- 44 TP, 9 FP, 2 FN and 2 TN results is equal to a sensitivity of 95.7% and a specificity of 18.2%. Those sensitivity and specificity figures result in a likelihood ratio (LR+) figure of 1.16, which means that a positive CT scan result would not significantly change the probability of appendicitis. Note that if they had a patient with a 75% clinical (pre-test) probability of appendicitis, they estimated (using those sensitivity and specificity figures) that a positive CT scan would only drive the post-test probability of appendicitis to a figure of 77%. They therefore concluded that the use of an appendiceal CT scan in their hospital would not decrease their negative appendectomy rate (compared to an earlier pre-CT scan time period) and they suggested that a CT scan should not be used routinely in patients with suspected appendicitis. Are their conclusions valid?

I believe that their conclusions are not valid because they determined the value of an appendiceal CT scan retrospectively in a selected group of patients, and not prospectively in an unselected group of suspected appendicitis patients. They did not seemingly derive any value from performing a CT scan because the diagnostic value of a CT scan had been "used up" along the way when it was used to exclude appendicitis (all the TN results) in an unselected group of suspected appendicitis patients. In order for you to understand this critically important fact, you need to fully understand the following line of reasoning. It will provide you with the key to understanding how to accurately estimate the value of any diagnostic test by better appreciating how a diagnostic test performs differently in different clinical practice contexts.

Dr. James Wagner of the University of Texas reviewed diagnostic clinical data from over 2,000 suspected appendicitis patients in primary care and in inpatient clinical wards [3]. His study provides us with a great deal of insight as to how the sensitivity and specificity of a diagnostic test changes, depending on whether it is used in a primary care setting or a tertiary care setting.

First of all, let's consider the value of determining RLQ tenderness as a diagnostic sign of appendicitis. Do you think that the presence of RLQ tenderness is very predictive of the presence of appendicitis? In other words, do you think that it has a high likelihood ratio (LR+)? What you may not realise is that the presence of RLQ tenderness as a sign of appendicitis has different LR+ values depending on whether it is detected in unselected abdominal pain patients in a primary care setting compared to selected abdominal pain patients in a tertiary care setting.

Consider the following table of the accuracy of RLQ tenderness in the diagnosis of appendicitis.

	Primary care settings		Tertiary care settings
	Appendicitis		Appendicitis
	Yes (%)	No (%)	Yes (%)	No (%)
RLQ tenderness
Present	84	11	81	84
Absent	16	89	19	16
Total	100	100	100	100
Frequency of appendicitis	14%		63%
Frequency of positive sign	21%		82%
Sensitivity	84%		81%
Specificity	89%		16%
LR+	7.6		1
LR-	0.2		1

Note that in the primary care setting that RLQ tenderness was present in 21% of the patients with abdominal pain, and that 14% of those patients actually had appendicitis. Note that the presence of RLQ tenderness had a LR+ of 7.6 for the presence of appendicitis (sensitivity of 84% and specificity of 89%). In other words, the presence of RLQ tenderness in a population of unselected abdominal pain patients is very suggestive of the presence of appendicitis.

Note that in the tertiary care setting that RLQ tenderness only had a LR+ of 1.0 (sensitivity of 81% and specificity of 16%) for the presence of appendicitis, and that the presence of appendicitis was as likely to be present in patients with appendicitis as it was likely to be present in patients without appendicitis (81% versus 84%). Why is that? The reason has to do with referral bias. The patients seen by the surgeons in the tertiary care setting were pre-selected by the primary care physicians, and referred to the surgeons for secondary evaluation. It is obvious that the primary care physicians would refer most/all patients with RLQ tenderness to the surgeons for clinical evaluation, which accounts for the fact that 82% of the patients seen by the surgeons had RLQ tenderness. That means that by the time the surgeons saw the suspected appendicitis patients, the value of RLQ tenderness as a highly specific sign of appendicitis was devalued. Its diagnostic value had been "used up". In other words, one cannot rationally claim that RLQ tenderness has no value in the diagnosis of appendicitis by only considering the value of the presence of RLQ tenderness in selected patients seen in a tertiary care setting by surgeons. One has to assess the value of RLQ tenderness in its proper context -- an unselected population of suspected appendicitis patients in a primary care or an ED setting.

If you understood the above example of the value of RLQ tenderness in the clinical evaluation of appendicitis, then you will realise that you need to assess the value of an appendiceal CT scan by evaluating how it performs in an unselected population of abdominal pain patients with suspected appendicitis in an ED setting. So let's presume that appendiceal CT scans were performed in 100 suspected appendicitis patients who presented to the Stamford ED with abdominal pain, and let's presume that they had the following CT scan results: - 44 TPs, 9 FPs, 2 FNs, and 45 TNs. Using the online Bayesian calculator at http://www.intmed.mcw.edu/clincalc/bayes.html, and plugging in those figures, one would note that the prevalence of appendicitis in their 100 patients with suspected appendicitis was 46% (which was similar to Rao study's figure of 53%) and that the CT scan's estimated sensitivity would be 95% and the specificity would be 83%. Those CT scan sensitivity and specificity figures are much better than the figures quoted in the Stamford study (sensitivity 95% specificity 18%), but they are still not optimal -- especially the low specificity figure of 83% -- when compared to Rao's sensitivity (>95%) and specificity (>95%) figures. If a CT scan only has a specificity of 83%, then there will be 9 FP CT scan results for every 100 appendiceal CT scans performed if the "average" prevalence of appendicitis in the population of suspected appendicitis patients is approximately 50%. What happened to those 9 FP cases at Stamford Hospital. Note that the Stamford surgeons operated on all of those FP CT scan cases, because they routinely operated on all suspected appendicitis patients who had a positive CT scan result -- even if they were not convinced that the patients definitely had appendicitis. I contacted one of the study's authors (personal e-mail communication) to find out why they operated on all the suspected appendicitis patients who had a positive CT scan, and I discovered that the main driving force for the medical decision-making process was medicolegal in nature. In other words, the surgeons in that hospital operated on all the CT scan positive patients, because they feared being sued if they delayed (or didn't perform) surgery in any CT scan positive patient who actually turned out to have appendicitis. They preferred to take the chance of having a negative appendectomy result (due to an inability to recognise a FP CT scan result pre-operatively), rather than undertake the alternative option of delaying surgery until they were clinically more convinced that the positive CT scan result was a TP result (rather than a FP result). If that problematic situation of too many unrecognised FP CT scan results exists in your community hospital, then it is understandable why your surgeons would regard a pre-operative CT scan a hindrance, rather than a help.

Is there an alternative method of dealing with this problem of too many unrecognised FP CT scan results?

There are two medical decision-making approaches that one can adopt if the sensitivity and specificity of an appendiceal CT scan in your hospital is sub-optimal (compared to Rao's CT scan results) and there are a disproportionately large number of FP results. The one approach is to discourage the routine use of appendiceal CT scans in suspected appendicitis patients because the surgeons, who have the final responsibility of deciding whether to perform an appendectomy, cannot readily differentiate a TP from a FP CT scan result, and they may therefore feel obliged to operate on all patients with a positive CT scan result. This practice pattern of operating on all positive CT scan patients will not necessarily decrease the hospital's negative appendectomy rate, and it would not be a cost-effective practice because of the added cost of routine CT scans in many suspected appendicitis patients. The second approach is to increase one's ability to differentiate TP from FP CT scan results, so that the surgeons only operate on patients who have a TP CT scan result. It may not be possible to always differentiate a TP from a FP CT scan result pre-operatively, but if those Stamford surgeons could have identified 5 out of those 9 FP CT scan cases pre-operatively, then they could have significantly decreased their negative appendectomy rate. Could they have identified some of those FP CT scan cases as being FP rather than TP cases prior to surgery? I obtained brief radiological reports on those 9 FP cases (personal e-mail communication) and I discovered that they had certain vague features, which could allow them to be identified as possible FP results. If it is possible to identify a signficant number of FP CT scan results prior to surgery, then I suspect that many more surgeons would decide to routinely use a CT scan in patients who have moderate-high probability of appendicitis, because its high positive predictive value would significantly decrease their negative appendectomy rates (an increase in the detection rate of FP results would increase the specificity of the CT scan). This is a key point that I want to address in this manuscript -- how to identify likely FP CT scan results with a greater degree of reliability.

A suggested approach for optimising the radiological reporting of an appendiceal CT scan

Why can emergency physicians and surgeons not correctly identify likely FP CT scan results? It is my suspicion that the main reason why many clinicians cannot accurately gauge whether a positive CT scan result is likely to be a TP rather than a FP result is related to the fact that those clinicians adopt the wrong mindset when it comes to CT scan interpretations. Clinicians usually expect a radiologist to give them a definitive answer as to whether a suspected appendicitis patient actually has appendicitis -- yes or no! I think that clinicians need to change their approach and they should not expect the radiologist to provide definitive dichotomous (yes-no) answers. Rather, they should demand that radiologists change the way they report their interpretation of the appendiceal CT scan. Clinicians should not consider the radiologists' final interpretative conclusion in isolation. Rather, they should specifically identify all the positive CT scan findings that caused the radiologist to diagnose appendicitis, and they should categorize the strength of the radiologist's conviction with respect to each radiological sign. Why do I suggest this alternative approach?

Let's re-consider the results from Wagner's study of 2,000 suspected appendicitis patients [3]. I previously demonstrated that the diagnostic value of RLQ tenderness as a discriminatory sign of appendicitis was far more useful in a primary care setting (LR+ 7.6) than a tertiary care setting (LR+ 1.0), and that its diagnostic value got "used up" as one proceeded along the referral pathway. What about the value of the clinical sign of abdominal rigidity?

Consider the following table of the accuracy of RLQ abdominal rigidity in the diagnosis of appendicitis.

	Primary care settings		Tertiary care settings
	Appendicitis		Appendicitis
	Yes (%)	No (%)	Yes (%)	No (%)
Rigid abdomen
Present	40	26	23	6
Absent	60	74	77	94
Total	100	100	100	100
Frequency of appendicitis	14%		47%
Frequency of positive sign	28%		14%
Sensitivity	40%		24%
Specificity	74%		94%
LR+	1.5		5
LR-	0.8		0.8

Note a very interesting fact -- the LR+ of the clinical sign of abdominal rigidity as a predictor of the presence of appendicitis was much higher in the tertiary care setting (LR+ of 5.0) compared to a primary care setting (LR+ of 1.5) and that its presence was much more useful in the diagnosis of appendicitis when it was determined by experienced surgeons in a tertiary care setting compared to the situation when that same physical sign was determined by primary care physicians. In other words, the specificity of the physical sign of abdominal rigidity increased along the referral pathway. That result is the exact opposite of the RLQ tenderness sign (previously described) whose specificity decreased along the referral pathway. Why is that? The most likely reason is that primary care physicians "over-read" the physical sign of abdominal rigidity because they are too inexperienced and too anxious about not missing the diagnosis of appendicitis. Note that the primary care physicians diagnosed the presence of abdominal rigidity in 28% of their patients even though only 14% of their patients actually had appendicitis. Contrast that result to the tertiary care setting, where surgeons decided that abdominal rigidity was only present in 14% of their patients even though 47% of their patients had appendicitis. Experienced surgeons, who are usually more capable of a nuanced clinical evaluation of a suspected appendicitis patient, presumably placed the correct diagnostic weight on the presence of abdominal rigidity by not "over-reading" the clinical sign of abdominal rigidity. By using more stringent subjective criteria for deciding that abdominal rigidity was present, the specificity of that physical sign increased from 74% (in the primary care setting) to 94% (in the tertiary care setting) and the surgeons only had a 6% FP rate (abdominal rigidity present, but appendicitis absent). I believe that surgeons should likewise acquire the necessary skills to evaluate the results of an appendiceal CT scan in a similarly highly specific manner, so that they do not unnecessarily operate on FP CT scan patients (CT scan positive, but appendicitis absent).

Consider the radiographic interpretation of an appendiceal CT scan in community hospitals. Some radiologists do not want to miss the diagnosis of appendicitis and they may tend to "over-read" the presence of radiological signs of appendicitis. The process of "over-reading" a CT scan is equivalent to decreasing the specificity of the CT scan, which will result in more FP readings (relative to TP readings). Therefore, if a surgeon wants to more accurately assess the CT scan in such a way that he increases the specificity of the CT scan reading (which will decrease the FP rate), then he must study the CT scan findings in a more nuanced way. He should specifically attempt to identify "hard" radiological signs of appendicitis (high LR+ signs) , and not "over-read" the "soft" radiological signs of appendicitis (low LR+ signs). It is important to realise that while some radiologists consciously (or unconsciously) "over-read" a CT scan to minimise missing a case of appendicitis, surgeons need to minimise their negative appendectomy rate by not operating unnecessarily on FP CT scan cases. Surgeons therefore need to re-evaluate "soft" CT scan findings in a more judicious manner, so that they can more clearly discern which positive CT scan patients actually have appendicitis (TP CT scans), in order to decrease their negative appendectomy rates (rate of operating on suspected appendicitis patients with FP CT scans). I will later describe how surgeons can better evaluate the likely TP-validity of any positive CT scan finding by getting the radiologist to use a structured checklist. However, before I describe my structured checklist idea, I need to get across another point -- that the quality of a clinician's final diagnostic conclusion is only as good as the quality of each of its component parts.

To make this important point more easily understandable, consider a "hypothetical" suspected appendicitis patient who is examined by a neophyte surgical resident.

Presume that a neophyte surgical resident examined a suspected appendicitis patient, and that he then described his clinical findings to the attending surgeon as follows.

Would the attending surgeon be satisfied with this case presentation, and could he decide whether the patient definitely had appendicitis based on the presented clinical evidence? One could argue that the surgical resident's case presentation is deficient for a variety of reasons, but I will only discuss a limited number of pertinent issues.

Note that the surgical resident does not describe the character of the pain, site of the pain, constancy of the pain, migration of the pain, or factors that aggravate/relieve the pain. He mentions anorexia as if it is a clinically relevant symptom that makes the diagnosis of appendicitis more likely. His description of the RLQ tenderness is vague. What does "some" tenderness mean? Should one regard RLQ tenderness as being clinically significant if it is only present with vigorous palpation? He does not mention whether the patient has voluntary RLQ guarding, rebound tenderness, localised RLQ rigidity, or a positive Rovsig's sign. He does not mention other clinical tests such as the psoas or obturator tests.

From my perspective, a clinical history and examination of a suspected appendicitis patient should have a number of important features. First of all, it should include all pertinent symptoms and signs that have a high LR+ (symptoms or signs that significantly increase the likelihood of appendicitis) or a high LR- (symptoms or signs that significantly decrease the likelihood of appendicitis).

Consider the potential usefulness of different clinical symptoms and signs in accurately diagnosing appendicitis by studying this table from reference number [4].

The table is the result of a meta-analysis of a number of clinical studies that attempted to determine the usefulness of different clinical signs in accurately identifying the presence (or absence) of appendicitis.

Note that the presence of RLQ pain has a very high LR+ (7.31-8.46) and that its presence really favors the positive likelihood of appendicitis -- to a much greater degree than any of the other clinical signs. That is why experienced clinicians give special weight to the presence of localised RLQ pain and tenderness. Persistent, and intensifying, RLQ tenderness can be perceived of as being a "hard" clinical sign that strongly favors the clinical diagnosis of appendicitis.

Note that the presence of anorexia only has LR+ value of 1.27, which means that the presence (or absence) of anorexia should not change a clinician's impression as to the likely presence of appendicitis in a particular patient. It can be regarded as a "soft" symptom.

Note that rebound tenderness has a LR+ of 1.1-6.3. The wide confidence interval figures suggest that it is a clinical sign of variable usefulness. Many experienced clinicians think that the presence of rebound tenderness does not add additional diagnostic weight to the positive likelihood of appendicitis if the patient already has definite RLQ tenderness and voluntary guarding.

One does not only expect a neophyte clinician to seek all the relevant historical facts and clinical signs in a patient with suspected appendicitis, one also expects him to impress us with the "surety" of his clinical impression with respect to each clinical symptom or sign. For example, the presence of RLQ tenderness is an important physicial sign, and one naturally expects a clinician who examines a suspected appendicitis patient to be confident as to its presence (or absence) by being an "artful" examiner, and one also expects the clinician to be relatively precise when quantifying its degree of intensity and its degree of consistency. There are different methods of quantifying RLQ tenderness. One could describe it as mild, moderate or severe. Or, one could describe RLQ tenderess as 1+, 2+ or 3+. Some clinicians would argue that there is a large interobserver variability in detecting (and describing) the degree of RLQ tenderess, and that the large interobserver variability dilutes the usefulness of any attempt to quantify the degree of RLQ tenderness. That is obviously true. However, the following clinical description by a surgical resident (or emergency physician) could give an attending surgeon a distinct impression that the surgical resident (or emergency physician) is a careful and skilled clinical practioner:- "The patient has moderate tenderness in the RLQ on light palpation of the abdomen, with maximum tenderness at McBurneys point. The tenderness is consistently present, even when the patient's attention is distracted. Increased RLQ pain is also elicited when the abdomen is firmly palpated in the LLQ. The patient also expresses increased RLQ pain during the psoas test and when the pelvis is rocked from side-to-side. The patient starts to guard voluntarily when RLQ palpation pressure increases to a moderate level, and he consistently winces when light palpation pressure is suddenly released. I also noted that all these physical signs were consistently present when I repeated the physical examination one hour later."

The point that I am trying to make is that we do not judge the likelihood of a particular patient having appendicitis solely on the basis of any individual clinician's final clinical conclusion. Rather, we base our assessment of the validity of the clinician's clinical conclusion on the cumulative effect of three synergistic factors that form the basic foundation of any clinician's clinical evaluation of a suspected appendicitis patient:- i) The thoroughness of the clinician's relevant history-taking and the thoroughness of his physical examination search for all relevant clinical signs suggestive of appendicitis. ii) The "surety" of the clinician's impression that each relevant clinical symptom and sign is accurately identified and quantified. iii) The consistency of that clinician's clinical skills with respect to accurately identifying appendicitis (in different patients) over an extended period of time. In other words, we do not trust any clinician's final clinical conclusion if we doubt the quality of the clinician's clinical evaluation from any of those three perspectives -- its thoroughness, its surety and his personal record of diagnostic consistency over an extended period of time.

The same attitude should be adopted with respect to a radiologist's interpretation of an appendiceal CT scan. A clinician should ensure that the radiologist thoroughly seeks all the relevant radiological signs of appendicitis (both LR+ and LR-), and that the radiologist is confident that those radiological signs are definitely present.

Rao reviewed the individual radiological signs that were present in his MGH study of 200 consecutive CT scans for suspected appendicitis (100 CT scans of patients with proven appendicitis and 100 CT scan of patients without appendicitis) to determine which radiological signs were particularly useful for diagnosing, or excluding, appendicitis [5].

This table from reference number 5 summarises the final CT scan sensitivity/specificity results.

(see the appendix for more detailed information on radiological signs of appendicitis)

Note that certain radiologic signs are 100% sensitive but not 100% specific for appendicitis. For example, fat standing was present in 100% of patients with appendicitis, but it was only 80% specific for appendicitis. In other words, if fat stranding is the only abnormal radiological sign seen in the appendiceal CT scan, and the CT scan is read as positive because of the isolated presence of fat stranding around an ill-defined opacity, then the FP rate could be as high as 20% (presuming that the pre-test likelihood of appendicitis, based on the clinical evaluation, was approximately 50%). Fat stranding was seen in 20% of patients with a normal appendix in Rao's study. Fat stranding can also be seen in other RLQ inflammatory conditions, and an important clue that it is not due to appendicitis is that the fat stranding is not peri-appendiceal in location and the appendix is normal in appearance.

Note, by contrast, that the specificity of certain radiological signs was 100%. So, although an appendicolith was identified in only 44% of the patients with proven appendicitis, if an appendicolith was seen on the CT scan in Raos' study, then there was a 100% probability that the patient definitely had appendicitis. The same applies to other radiological signs such as the arrowhead sign, cecal bar sign, focal cecal apical thickening sign and the presence of an appendiceal abscess.

If a clinician wants to decrease his FP CT scan rate, then he needs to be sure that the CT scan was technically well perfomed and that adequate cecal distension with contrast was achieved, and that the radiologist specifically attempted to identify all the radiological signs that are highly specific for appendicitis. In other words, he wants to be certain that the radiologist was very thorough in his evaluation of the CT scan and that all the high yield LR+ radiological signs were sought and identified. Of course, it may not be possible for the radiologist to be 100% accurate in identifying all the high yield LR+ radiological signs, because of technical and interpretative problems. For example, the presence of contrast may obscure an appendicolith. Or incomplete filling of the cecum with contrast may decrease the likelihood of seeing a positive arrowhead sign. Another factor that may affect the diagnostic yield of a CT scan is the type of CT scan study performed. If the CT scan study was a non-contrast CT scan, then it is obvious that certain high yield LR+ radiological signs cannot be present, because they are dependent on the presence of orally or rectally administered contrast eg. arrowhead sign or cecal bar sign. A third factor that may decrease the detection of high yield LR+ radiological signs is the fact that the radiologist may be inexperienced in reading an appendiceal CT scan and he may not be aware of certain high yield LR+ radiological signs such as the cecal bar sign or the arrowhead sign. All these technical and radiologist-skill factors affect the reliability of the CT scan reading and the surety of the final CT scan interpretation.

Most true-positive CT scan cases have three or more radiological signs of appendicitis. The most common combination of radiological signs seen in TP appendicitis cases is the presence of a distended, non-opacified appendix centered within an area of peri-appendiceal fat stranding. The likelihood of the CT scan case being a TP case (rather than a FP case) is greater if the radiologist is very confident that the distended structure is definitely the appendix and not a loop of small bowel (or another anatomical structure), and if the fat stranding is mainly localised to the peri-appendiceal area. The presence of other highly specific radiological signs (appendicolith or arrowhead sign) makes it even more likely that one is dealing with a a TP rather than a FP CT scan case.

The major point that I am trying to make is that a clinician should be fully aware of the "likely" sensitivity and specificity values of each radiological sign that is suggestive of appendicitis, so that he can more accurately interpret the appendiceal CT scan's findings. He should also accurately ascertain that the radiologist searched for each radiological sign by ensuring that the radiologist completed a checklist for each appendiceal CT scan sign, and he should expect the radiologist to also assign a confidence level to the detection of each radiological sign. If the technical study is inadequate and it is difficult to identify certain radiological signs, then the radiologist should indicate that fact by assigning a lower confidence level to that specific radiological sign.

One way of achieving this checklist goal is to supply the radiologist with the following checklist form. If the radiologist completes the checklist form, it will enable the clinician to better decide whether the CT scan result is a FP rather than a TP result.

An adobe pdf version of this form (for printing purposes) is available by clicking here

By completing this checklist form, the radiologist will be supplying the clinician with much more information regarding the type of CT scan study, the technical quality of the study, and the likely presence of all (and not only some) of the radiological signs suggestive of appendicitis.

Armed with this detailed checklist information, a clinician will be be able to better decide whether a CT scan result is likely to be a TP rather than a FP result. For example, if the CT scan was reported as demonstrating a "definitely distended unfilled appendix, with definite peri-appendiceal fat stranding, and a definite appendicolith, and an equivocal arrowhead sign, and a level 4 likelihood of dependent fluid in the RLQ", then that CT scan is much more likely to be a TP case (rather than a FP case) than a CT scan that was reported as demonstrating an "equivocally present distended appendix, with probable fat stranding and a level 4 likelihood of extraluminal air". In other words, by studying a CT scan's interpretative results in a much more nuanced manner, one is less likely to "over-read" a positive scan, and this nuanced approach will likely decrease the number of FP relative to TP CT scan interpretations (and improve the specificity of the CT scan reading).

I think that a clinician makes the "best" diagnostic decisions when he accumulates all the evidence and weighs all the evidence (clinical and laboratory and radiological) together to come to a final conclusion. I previously argued that an astute clinician needs to weigh the clinical evidence acquired from a thorough and reliable clinical examination using his best clinical judgement. I do not believe that he should think differently when faced with the results of the CT scan. He should take all the diagnostic information from the CT scan -- all its positive findings and the degree of surety that those findings are definitely present -- and add all those radiological findings to the cumulative weight of his clinical impression and come to a revised clinical impression. The Stamford surgeons argued that they operated on all patients with a positive CT scan because they feared the medical malpractice consequences of not operating on any patient with a positive CT scan -- even if their clinical impression did not favor the likely diagnosis of appendicitis. I think that their medical malpractice concerns were not a justifiable reason for not attempting to differentiate FP from TP CT scan results prior to surgery. After looking at the CT scan reports from those 9 FP CT scans, it appeared to me that they could be interpreted as equivocally positive CT scan results. I think that if a surgeon has a low-moderate clinical suspicion of appendicitis after clinically evaluating a suspected appendicitis patient, then he should not necessarily feel obliged to operate on the suspected appendicitis patient just because the radiologist's final CT scan interpretative conclusion is positive for appendicitis. He should evaluate the CT scan findings independently and give the positive findings their proper evidentiary weight, depending on whether the findings are "hard" or "soft" and depending on whether the radiologist is absolutely confident regarding their true presence. Equivocal CT scan results are non-diagnostic, and they do not change the pre-test clinical likelihood of appendicitis. If the surgeon remains uncertain about the definite probability of appendicitis, then he could continue to serially observe the patient until he becomes more convinced that appendicitis is definitely present. From a medicolegal perspective, the surgeon could demonstrate how he rationally weighed all the evidence (clinical and radiological). If he had a CT scan report that included my recommended checklist, and all the positive CT scan findings were "soft" radiological findings with low confidence scores (equivocally positive), then he should have no/little problem defending the rationale of his clinical judgements in a court of law.

Should a CT scan be used in a suspected appendicitis patient who has a classical presentation?

Consider the following case scenario of a "hypothetical" suspected appendicitis patient.

Do you consider that the patient has a "classical" presentation of appendicitis? What do you estimate the clinical probability of appendicitis to be based on the above clinical findings?

I presume that all clinicians would agree that the clinical presentation is suggestive of the classical presentation of appendicitis in the sense that the patient first had generalised abdominal pain which subsequently localised predominantly to the RLQ. I also presume that all clinicians would agree that the patient probably has definite appendicitis (80-100% probability of appendicitis) if his RLQ tenderness was moderate-severe in degree, and he also had voluntary guarding, a positive Rovsig's sign and an elevated temperature and elevated WBC. However, some clinicians would think that this patient's probability of appendicitis -- in the absence of those additional elements -- should be classified as probable appendicitis (60-79% probability of appendicitis) rather than definite appendicitis (80-100% probability of appendictis). If those clinicians thought that the patient only had a 70% probability of appendicitis at the time of the initial clinical evaluation, then they would presumably not recommend immediate surgery because it would be associated with a 30% likelihood of a negative appendectomy if the initial clinical impression is accurate. They would likely recommend the traditional serial clinical observation approach as the optimum clinical strategy. As an alternative diagnostic approach, the clinician could order an immediate appendiceal CT scan. What is the value of an appendiceal CT scan in this clinical situation?

Obviously, the post-test likelihood of appendicitis after a positive CT scan depends on the CT scan's sensitivity and specificity. In Rao's study, the CT scan had a very high sensitivity (>95%) and very high specificity (>95%) for diagnosing appendicitis, but there is no guarantee that the radiologists in community hospitals could perform as well as Rao's MGH radiologists. Let's consider the likely results of a CT scan in community practice, as compared to Rao's results.

The following table presents the positive and negative predictive results of an appendiceal CT scan of varying sensitivity and specificity. PPV stands for the positive predictive value of a positive CT scan result and it is equivalent to the post-test probability of appendicitis if the pre-test probability of appendicitis is 70%. The FP rate is 100-PPV. NPV stands for the negative predictive value of a negative CT scan test, and it is the percentage likelihood of a TN result if the patient has a pre-test probability of appendicitis of 70%. The FN rate is 100-NPV.

Sensitivity/specificity of a CT scan (%)	PPV	FP rate	NPV	FN rate
95%/95%	98%	2%	89%	11%
90%/90%	95%	5%	79%	21%
85%/85%	93%	7%	71%	29%
80%/80%	90%	10%	63%	37%
75%/75%	87%	13%	56%	44%

Can an appendiceal CT scan be used to "rule-out" the presence of appendicitis in a patient who has a classical presentation of appendicitis, and a likely clinical probability of appendicitis of 70%?

Note that it is often not possible to definitely exclude the diagnostic possibility of appendicitis if the appendiceal CT scan is negative in a patient with a 70% clinical probability of appendicitis -- especially if the CT scan has a low sensitivity/specificity. The predicted FN rate of a negative CT scan is substantial and varies from 11-44% depending on the quality-performance (sensitivity) of the CT scan. However, remember that those FN rate figures apply to CT scans performed on a population of suspected appendicitis patients, who all have a 70% pre-test probability of appendicitis. They do not necessarily apply to an individual patient, and surgeons have to make decisions for one patient at a time. Therefore, the surgeon needs to carefully examine the specific radiological signs found in the CT scan. If the radiologist identifies a definitely normal appendix that fills with contrast and is not enlarged (<6mm in maximum diameter), and there are no other abnormal CT scan findings, then that CT scan result is likely to be a TN result. The likelihood of a TN CT scan result is even greater if other definite radiological signs of a normal appendix are present eg. appendiceal wall thickness <2mm and the presence of air in the distal appendix lumen. If the radiologist cannot see the appendix and/or there are other subtly abnormal CT scan findings in the RLQ, then the possibility of a FN CT scan result is substantial (due to an "over-reading" of the CT scan from a "rule-out" disease perspective, which markedly decreases the CT scan's sensitivity). A prudent surgeon should therefore not discharge a suspected appendicitis patient, who has a high likelihood of a FN CT scan result, and continued serial observation is the recommended approach. If the high probability appendicitis patient has florid clinical signs of appendicitis (marked RLQ tenderness and guarding, elevated temperature and elevated WBC) and a definitely normal appendix clearly visible in the appendiceal CT scan, then another RLQ inflammatory condition should be presumed to be present. The patient may have ruptured a Meckel's diverticulum or a right sided colitic diverticulum. A decision to perform a diagnostic laparoscopy or exploratory laparotomy should be based on the surgeon's overall clinical suspicion of a surgically remediable RLQ inflammatory process, and he should not be swayed by the fact that a normal appendix is visible in the CT scan (CT scan negative for appendicitis).

Some clinicians already realise that an appendiceal CT scan should not be used to "rule-out" appendicitis in patients with a high clinical probability of appendicitis (70% probability) because a negative CT scan has a poor negative predictive value in high probability patients. Rather, the main value of an appendiceal CT scan lies in its positive predictive value. If the pre-test clinical probability of appendicitis is 70%, and the PPV of a positive CT scan is estimated to be >95%, then the negative appendectomy rate will be <5% if surgeons operate on all patients with a positive CT scan result. Contrast that good outcome with the MGH surgeons, who would have had a 22% negative appendectomy rate if they operated immediately on all suspected appendicitis patients, who were deemed to definitely have appendicitis (80-90% clinical probability of appendicitis) based on their initial clinical evaluation.

Note from that table that if the appendiceal CT scan is positive in a suspected appendicitis patient with a 70% probability of appendicitis, that the possibility of a FP result varies from 2-13%, depending on the CT scan's sensitivity and specificity. It would make eminent sense for a surgeon to operate on all positive CT scan cases if he knew that his hospital's appendiceal CT scan performance was excellent (sensitivity >90% and specificity >90%), and he predicted that the likely negative appendectomy rate would be <5% if he operated on all positive CT scan cases. However, if a surgeon knew that his hospital's appendiceal CT scan sensitivity and specificity figures were around 75%, then he should anticipate an "average" FP rate of 13%. He could adopt two approaches in that situation. The one approach would be to not order an appendiceal CT scan if he had a relatively high clinical suspicion of appendicitis (70% probability of appendicitis) because the likely negative appendectomy rate would be 13% if he operated immediately on all patients with a positive CT scan result. He could use the traditional serial observation approach instead, especially if he knew that his personal negative appendectomy rate averaged <10% using the serial observation approach. The second (alternative) approach would be to perform routine CT scans, but to use discriminatory clinical judgement when faced with a positive CT scan result in order to maximise one's identification of FP CT scan results. If a surgeon can readily differentiate FP from TP CT scan results pre-operatively with a reasonable degree of accuracy, then this routine CT scan approach would be an acceptable approach that could result in a <5% negative appendectomy rate.

The value of a routine CT scan, to decrease the negative appendectomy rate, for a suspected appendicitis patient with a 70% probability of appendicitis, therefore depends on the sensitivity and specificity of a hospital's appendiceal CT scan and the surgeon's ability to differentiate a TP from a FP CT scan result prior to surgery.

The disadvantages and pitfalls of using a CT scan in a patient with a low probability of appendicitis

Many clinicians, especially emergency physicians working in a busy ED, have a low threshold for ordering a CT scan in patients with abdominal pain. What is the major downside of ordering a CT scan if the clinical likelihood of appendicitis is low.

Consider the NPV and PPV of a CT scan in a patient with a 15% pre-test probability of appendicitis.

Sensitivity/specificity of a CT scan (%)	PPV	FP rate	NPV	FN rate
95%/95%	77%	23%	99%	1%
90%/90%	61%	39%	98%	2%
85%/85%	50%	50%	97%	3%
80%/80%	41%	59%	96%	4%
75%/75%	36%	64%	94%	6%

Can a negative CT scan result exclude the presence of appendicitis if the patient has a 15% clinical probability of appendicitis?

Note that the CT scan has a good negative predictive value in this clinical situation, because one can be reasonably certain that the patient does not have appendicitis if the CT scan is negative (94-99% certainty depending on the CT scan's performance). However, a clinician should be aware of a potential pitfall that can frequently occur in clinical practice. A 4-6% FN rate figure presumes that one has an atypical abdominal patient who has a 15% probability of having well-established appendicitis, which can theoretically be detected by a CT scan of relatively high sensitivity (75-80% sensitivity). That clinical scenario situation does not apply to a different type of low probability clinical situation where the suspected appendicitis patient has early appendicitis (major spectrum bias problem). In other words, if the patient has very early appendicitis (distal appendicitis) then the sensitivity of an abdominal CT scan for isolated distal appendicitis may be considerably less than 75%. In a 15% (or greater) probability of early appendicitis clinical scenario, a negative CT scan result could more likely be a FN result, because of the limited sensitivity of a CT scan in detecting early appendicitis. Therefore, if one is dealing with a patient who only has a history of RLQ abdominal pain of a few hours duration, then one should either not order the CT scan in the first place, or at least warn the patient that a negative CT scan does not exclude the possibility of early appendicitis. The CT scan negative patient, who is discharged from the ED, should be carefully instructed to return to the ED if clinical signs suggestive of definite appendicitis subsequently develop.

Can a positive CT scan result prove that appendicitis is present if the patient has a 15% clinical probability of appendicitis?

Note how high the FP rate is when one orders a CT scan in a patient with a low probability of appendicitis (23-64% FP rate depending on the CT scan's performance). There is an especially high likelihood of a FP result (64%) if the CT scan has a low specificity of 75%. This situation is going to seriously challenge a surgeon's ability to differentiate a TP from a FP CT scan result. Most prudent surgeons would presume that the CT scan result could be a FP result if their clinical impression suggests that the patient has a low clinical likelihood of appendicitis. They would obviously not immediately operate on the patient, but they would be under considerable pressure to admit the patient to hospital for serial observation. The need to unnecessarily hospitalise patients for serial observation is a great disadvantage of the practice of liberally ordering CT scans in low probability appendicitis patients, because most cases of a positive CT scan in low probability patients are going to be FP CT cases.

To get some idea of the complications that can arise from FP CT scan results in patients with a low clinical probability of appendicitis, you should read the following journal article [6]. In that article, the authors describe 4 case reports that graphically demonstrate how an appendiceal CT scan led to increased confusion, rather than clarity, regarding appropriate medical managment, because of substantial discordance between the radiographic and clinical findings. I will describe just one of those cases to give you some idea of the consequences of liberally ordering an CT scan in a patient with a very low clinical probability of appendicitis.

Case scenario: A 13 year-old female was initially seen in the ED for a complaint of RLQ abdominal pain. The history was positive for 8 hours of RLQ pain, equivocal anorexia, mild nausea and no fever. The physical findings were apparently unremarkable. The WBC count was 10,900/cu.mm. A surgeon was consulted who recommended discharge and outpatient followup. Nonetheless, the EP ordered a CT scan which was reported as positive for appendicitis (it apparently demonstrated an enlarged, unfilled appendix with peri-appendiceal fat stranding and a small amount of pelvic fluid). The positive CT scan imaging findings led to a second surgical consultant becoming involved. Despite agreement between both surgeons' clinical impressions, the patient was admitted for overnight observation solely as a result of the radiographic findings. The patient was subsequently hospitalized for 2 days and had 2 ultrasound studies performed despite consistently benign physical findings and normal laboratory studies. Again, all interventions were based solely on the initial imaging results. The patient never required antibiotics or surgery and was well at the 3-week follow-up visit.

It is difficult to critique the quality of the radiological interpretation in that case scenario without access to the CT scan images, and we do not know how confident the radiologist was in the accuracy of his personal radiographic interpretation. What we do know is that there was discordance between the surgeon's clinical impression (low probability of appendicitis) and the radiological interpretation of the CT scan (high probability of appendicitis), and that hospitalisation and further diagnostic testing was required to determine whether the CT scan result was a TP or a FP result. This problem of a FP CT scan result can also occur in patients with a moderate or high clinical probability of appendicitis, but the problem is most acute in patients with a low clinical probability of appendicitis, because there are far fewer TP relative to FP CT scan results in low probability patients. A clinician should therefore always strongly suspect a FP result if the CT scan result is positive for appendicitis in a patient with a very low clinical probability of appendicitis (discordance between a low clinical impression and a positive CT scan result).

From a cost-effective practice perspective, it makes no sense to liberally order abdominal CT scans in an ED setting if the unselected abdominal pain patients only have a low clinical likelihood of appendicitis. Most of the CT scans are going to be TNs and there will only be a few positive CT scan results, and most of those will be FP results. Therefore, an EP should be very judicious when deciding whether to order a CT scan in low probability of appendicitis patients -- especially if the history of abdominal pain is short in time duration, and abdominal pain has only been present for a few hours. In that situation, if early appendicitis is even present, it will likely not be detected by a CT scan.

The following recommendations reflect my personal opinion, and they are for general guidance purposes only.

An appendiceal CT scan can be very valuable as a routine diagnostic test in patients with suspected appendicitis -- irrespective of the pre-test probability of appendicitis -- if the CT scan has a very high sensitivity (>95%) and specificity (>95%). However, an appendiceal CT scan should not be used a routine diagnostic test in patients with suspected appendicitis if it has suboptimal sensitivity and specificity, and the clinician taking care of the patient cannot accurately differentiate between FN and TN, and between FP and TP CT scan results.

An appendiceal CT scan can be used to decrease the negative appendectomy rate to <5% if it used routinely for suspected appendicitis patients who have probable appendicitis (60-80% clinical probability of appendicitis) -- but only if the appendiceal CT scan has a very high sensitivity (>95%) and specificity (>95%). However, it is important to realise that a negative CT scan does not necessarily exclude appendicitis in high probability appendicitis patients, because there is a high frequency of FN CT scan results in high probability patients (even if the CT scan has high sensitivity and high specificity). Careful re-examination of the negative CT scan is indicated. Serial observation, rather than hospital discharge, is recommended if there discordance between a high clinical probability of appendicitis and a negative CT scan result -- if a FN CT scan result is deemed possible. A high probability appendicitis patient should only be discharged if the CT scan is definitely a TN result, and there is no other reason to clinically suspect another RLQ inflammatory process mimicing appendicitis, which warrants hospitalisation.

An appendiceal CT scan of suboptimal specificity will result in a significant number of FP results if it is used in suspected appendicitis patients who have probable appendicitis (60-80% probability of appendicitis). Clinicians should not use suboptimal appendiceal CT scans routinely for this category of suspected appendicitis patient, unless they have a reasonably good ability to differentiate between TP and FP CT scan results prior to surgery. If a surgeon cannot differentiate a TP from a FP CT scan result in a suspected appendicitis patient, who has probable appendicitis, and he therefore decides to operate on all CT scan positive patients, then the surgeon's negative appendectomy rate may not be reduced, despite the additional burden and cost of a routine pre-operative CT scan. Radiologists should routinely complete a structured checklist form on all appendiceal CT scans, because it may enable surgeons to better identify likely FP CT scan results prior to surgery.

Routine use of an appendiceal CT scan in low probability appendicitis patients is not cost-effective practice and it can be problematic, because of the relatively high proportion of FP CT scan results. This will result in unnecessary hospital admissions (for serial observation) if a clinician cannot accurately differentiate a TP from a FP CT scan result. Immediate surgery is obviously not recommended if there is discordance between a low clinical probability of appendicitis and a positive CT scan result -- unless the surgeon is convinced that he is dealing with a TP CT scan result. Also, it is important to realise that a negative CT scan cannot confidently exclude early distal appendicitis, because a CT scan may have a markedly reduced sensitivity for detecting early appendicitis (spectrum bias problem).

If a patient has an atypical presentation of suspected appendicitis, a CT scan may also be useful in diagnosing other diseases if appendicitis is not present. Other diseases -- such as mesenteric adenitis, PID, ovarian cysts, right sided colitic diverticulitis, Meckel's diverticulitis, ileitis, ureteral stone, superior mesenteric vein thrombosis -- can be identified in ~25% of CT scan's that are performed in abdominal pain patients with a clinical presentation that is suggestive of an atypical presentation of appendicitis. Clinical judgement must be exercised when deciding whether to order a CT scan in atypical abdominal pain patients. A non-focused CT scan may have a better diagnostic yield than a CT scan that is too tightly focused on the RLQ in those atypical abdominal pain patients.

The accuracy of a CT scan to diagnose appendicitis depends on optimal ileocecal opacification and thin collimation helical scanning. The results will be less accurate if a contrast agent is not used, or if older generation CT scanners are used. Some radiologists achieve significant success using a non-contrast CT scan, but the degree of success apparently depends on the radiologist's experience and skill. Rao uses oral and rectal contrast in his CT scan studies, and he does not use IV contrast. Therefore, he does not mention radiological signs such as appendiceal wall thickening and appendiceal wall enhancement, which can only be readily seen when using IV contrast. Rao thinks that the risks of IV contrast (allergy) outweigh the small additional diagnostic benefits achieved by using IV contrast in addition to oral/rectal contrast.

CT scan signs suggestive of appendictis include an abnormal appendix (enlarged appendix and/or appendicolith); RLQ inflammatory changes (fat stranding, fluid, phlegmon, abscess, extraluminal air, adenopathy, adjacent bowel wall thickening) and cecal apical changes (focal cecal apical thickening, arrowhead sign, cecal bar).

An enlarged appendix is defined as >6mm in diameter. The distended appendix doesn't opacify with contrast in appendicitis because the appendiceal lumen is usually proximally occluded. Extensive fat stranding or fluid can occasionally obscure a distended appendix. A distended appendix may also not be seen if a phlegmon or abscess is present. Usually additional radiological signs of appendicitis are present when the phlegmon or abscess obscures the distended appendix eg. appendicolith or arrowhead sign. Remember that an inflammed appendix can also lose its rounded or distended appearance if it perforates, and it may therefore not be visisble on the CT scan.

Appendicoliths are visible in roughly 25-50% of CT scans that are positive for appendicitis. Although the presence of appendicoliths were 100% specific for appendicitis is Rao's study, other CT scan studies have shown a normal appendix in the presence of appendicoliths, so always look for additional radiological signs of appendicitis to confirm the presence of appendicitis.

RLQ adenopathy is defined as a cluster of three or more nodes of at least 5mm in diameter in the smallest dimension. Adenopathy is not a highly specific sign of appendicitis and it may also be seen in mesenteric adenitis, and occasionally in ileiltis or colitis.

Focal cecal wall thickening (not apical) is a rare, but highly specific radiological sign of appendicitis. Diffuse cecal wall thickening is not usually seen in patients with appendicitis. Diffuse cecal wall thickening was seen in 9% of Rao's patients who had a normal appendix, and was usually due to other RLQ diseases eg. colitis or right sided diverticulitis.

Selected CT scan images showing focal cecal apical thickening, the arrowhead sign and the cecal bar sign (from reference number [4]).

Image a = Focal cecal apical thickening (arrow) at the origin of the enlarged appendix.

Image b = Arrowhead sign (black arrow), formed by symmetric cecal apical thickening and funneling of contrast material to the point of appendiceal (white arrow) occlusion.

Image c = Cecal bar sign (black arrow) separates contrast material in the cecal lumen (C) from a proximal appendicolith (white arrow) (which incidentally contained matrix air).

Focal cecal apical thickening is a localised thickening of the cecal wall at the origin of the appendix, often seen with appendicitis at CT scan if good cecal distention is present.

The arrowhead sign is a triangle-shaped collection of contrast material seen at the cecal apex when contrast meaterial funnels symmetrically to the point of appendiceal occlusion.

The cecal bar is a straight or slightly curved band of inflammed soft tissue that can separate proximal calcified appendoliths from similarly dense contrast material.

1. Rao PM, Rhea JT, Novelline RA, et al. Effect of computed tomography of the appendix on treatment of patients and use of hospital resources. N Engl J Med 1998;338:141-6.

2. Perez J - Liberal use of computed tomography scanning does not improve diagnostic accuracy in appendicitis. Am J Surg - 01-March-2003; 185(3): 194-7.

3. The Evidence Base of Clinical Diagnosis. BMJ Books. First edition. Edited by J Andre Knotterus. Page 33.

4. Wagner JM. Does This Patient Have Appendicitis? [The Rational Clinical Examination]. JAMA. Volume 276 (19) 20 November 1996 pp 1589-1594.

5. Rao PM. Sensitivity and specificity of the individual CT signs of appendicitis: experience with 200 helical appendiceal CT examinations. J Comput Assist Tomogr. - 01-Sep-1997; 21(5): 686-92

6. Safran D. Is appendiceal CT scan overused for evaluating patients with right lower quadrant pain? American Journal of Emergency Medicine Volume 19. Number 3. May 2001.