Association of clinical and laboratory variables with ultrasound findings in right upper quadrant abdominal pain.

Background: The purpose of this study was to determine how laboratory values and physical examination findings correlate with ultrasound findings in the setting of right upper quadrant pain.

Methods: Patients undergoing emergent ultrasound for the evaluation of biliary disease between November 1999 and April 2000 were included. Physical examination findings, laboratory data, and ultrasound results were variables. Logistic regression was performed. Ultrasound diagnosis of acute cholecystitis, cholelithiasis, and normal biliary tract were end points. One hundred seventy-seven patients were enrolled.

Results: Forty-two percent were diagnosed with acute cholecystitis, 30.5% with cholelithiasis, and 27.1% with normal biliary tract. Alkaline phosphatase, Murphy sign, white blood cell count, and total bilirubin were statistically significant predictors of acute cholecystitis. A Murphy sign was defined as arrest of inspiration with pressure over the right upper quadrant.

Conclusions: The findings from this study allow clinicians to apply objective significance to laboratory data and physical examination findings in patients with suspected gallstone disease. The data can be applied to create a predictive model.

Key Words: abdominal pain, cholecystitis, cholelithiasis, ultrasound

**********

Gallbladder disease affects 500,000 people in the United States annually. (1-3) Approximately 20 million adults in the Unites States have cholelithiasis (CL). (4) Between 17 and 31% of people with cholelithiasis progress to acute cholecystitis (AC). (5-11)

Several studies have been performed to determine the predictive value of individual laboratory measures and physical examination findings in diagnosing AC. Most of the past studies used the hepatoiminodiacetic acid (HIDA) scan as the gold standard. (12-17) Although a hepatoiminodiacetic acid scan can differentiate between AC and a normal gallbladder, it cannot accurately identify gallstones. These previous studies either found limited use of individual predictors, such as temperature and white blood cell count, (12) or failed to show any predictive values of laboratory and physical examination findings. (14-16, 18, 19)

No recent predictive studies have used right upper quadrant (RUQ) ultrasonography as the gold standard for diagnosing acute cholecystitis. Ultrasonography is a noninvasive, rapid study to identify CL. (12-14) The sensitivity and specificity of ultrasonography for diagnosing CL and AC is 92 to 96% and 95 to 99%, respectively. (15-20)

The purpose of this study was to determine how laboratory values and physical examination findings correlate with ultrasound findings in the setting of right upper quadrant pain.

We hypothesized that various laboratory and physical examination measures may change in their predictive capability when evaluated against an outcome variable with less error. The object of our study was the predictive capability of these measures in differentiating AC and CL from each other as well as from a normal gallbladder.

Materials and Methods

Study design

Institutional review board approval was granted for this project. A retrospective chart review was conducted between the months of November 1999 to April 2000. Data collected by chart review included sex, age, medical history, diagnosis, physical examination, and presence of vomiting. Vital signs and laboratory values, including aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase, white blood cell count, amylase, lipase, and total bilirubin were obtained from a computerized laboratory database. Ultrasonographic findings were obtained from attending radiology dictation.

Study setting and population

The setting of the study was the emergency department of an inner-city teaching hospital with a volume of 120,000 patients per year. During the emergency department visit, all patients were evaluated by a second- or third-year emergency medicine resident and an attending emergency medicine faculty member.

All patients between 16 and 65 years of age who received emergent RUQ ultrasonography for the evaluation of suspected biliary disease were enrolled. Patients over the age of 65 were excluded to minimize confounding data caused by concurrent comorbid conditions. Patients with previously diagnosed cirrhosis, human immunodeficiency virus, or idiopathic or viral hepatitis were excluded. Patients with human immunodeficiency virus were excluded secondary to possibly confounding data caused by the possibilities of human immunodeficiency virus cholangiopathy, idiopathic pancreatitis, and hepatitis.

Emergent RUQ ultrasonographic imaging is available 24 hours per day, 7 days per week at this institution. Registered diagnostic medical sonographers perform the studies. Emergent studies are obtained at the discretion of emergency medicine faculty, based on clinical suspicion of acute cholecystitis. The diagnosis of AC was determined by the surgical attending physician.

Measures

Thirteen continuous physiologic variables were studied to determine which if any were potential predictors. These were AST, ALT, gamma glutamyl transferase (GGT), alkaline phosphatase (AlkP), amylase, lipase, total bilirubin (TB), white blood count (WBC), neutrophil count, band count, serum glucose, heart rate, temperature, and age. In addition, three naturally dichotomized variables were studied: clinical Murphy sign, nausea or vomiting (N/V), and sex. A clinical Murphy sign is defined as arrest of inspiration with pressure over the RUQ in the midclavicular line. Unavailable data were recorded as such. Missing data points were excluded from data analysis.

Subsequently, 11 continuous measures were dichotomized with cutoff points determined by laboratory normal values. Heart rate and temperature were not dichotomized. The median age was 36 years. This was selected as the point at which to dichotomize the age variable.

Data analysis

Three diagnostic categories were considered: AC, CL, and normal gallbladder. Biliary pancreatitis and choledocholithiasis were grouped with AC. These diagnoses were grouped because they all are acute, clinically significant complications of cholelithiasis that require admission to the hospital. These diagnoses were paired into three categories, AC versus CL, AC versus normal, and CL versus normal. Ultrasonographic diagnosis of AC, CL, or normal was used as the gold standard.

With each of the three categories, a univariate logistic regression was performed with each physiologic variable as the independent variable. Then, as previously mentioned, 11 continuous variables were dichotomized into high and normal levels, based on established laboratory values. Another logistic regression, including predictive power and test characteristics, was performed on the 15 dichotomized variables.

This analysis resulted in two data sets for each category, one of continuous variables and one of dichotomized variables. Ordinarily, a Bonferroni correction would be applied to these multiple comparisons. However, within the context of this study, a type II error was considered more serious than a type I error. If a measure were presented as being potentially important, further studies would validate or invalidate its importance. If a measure were presented as unrelated to the disease state, then further studies may never be performed to prove this wrong. As Tables 1 through 6 show, few measures have probability values between 0.05 and 0.10.

Results

All eligible patients were included in the study. One hundred seventy-seven patients were identified, 76.8% (136) women and 23.2% (41) men. Seventy-five (42.4%) had AC, 26.7% (20) men and 73.3% (55) women. Fifty-four had CL, 9% (5) men and 91% (49) women. Forty-eight (30.5%) had normal ultrasound studies, 33.3% (16) men and 66.7% (32) women. Twenty-two patients were excluded because of the presence of one or more of the exclusion criteria.

The results are presented in Tables 1 through 6. The odds ratios can be read as favoring the model for each unit increase in the continuous variable. For the dichotomized variables, the odds ratios are the odds of the event for the high end of a measure as compared with the low end. In the sex category, the odds are for females as compared with males. The logistic regressions for AC and normal are modeling the probability for AC and the logistic regressions for CL and normal are modeling the probability for CL. The logistic regressions for AC and CL are modeling the probability for AC.

Different variables were significant in the three groups. In differentiating AC from CL (Tables 1 and 2), the following variables were significant for predicting AC: AlkP, clinical Murphy sign, and TB. In this category, three dichotomized variables also predicted the presence of AC. These were AlkP, TB, and WBC.

In the category AC versus normal (Tables 3 and 4), the variables AlkP, AST, TB, and N/V were predictive of AC. The dichotomized variables for AlkP, ALT, AST, GGT, TB, and WBC also predicted the presence of AC.

To differentiate CL from normal (Tables 5 and 6), the variables ALT, AST, and female sex were statistically significant predictors. The statistically significant dichotomized variables were ALT, AST, GGT, WBC, and age.

When evaluating for missing data, in the group with AC, the following variables had no more than one missing data value: AST, ALT, AlkP, amylase, lipase, WBC, clinical Murphy sign, age, and sex. Two subjects were missing GGT and TB.

In the group with CL, the following variables had no more than one missing data value: TB, AST, ALT, AlkP, WBC, clinical Murphy sign, age, and sex. Three subjects were missing GGT.

In the group of normal gallbladders, the following variables had no more than one missing data value: TB, AST, ALT, AlkP, WBC, clinical Murphy sign, age, and sex. Two subjects were missing GGT and amylase.

Discussion

Gallbladder disease has been studied with oral cholecystogram, hepatoiminodiacetic acid scan, CT scan, and plain radiographs. (13,14) Although the aforementioned studies can make the diagnosis of AC, ultrasonography allows the accurate distinction of AC from CL. (15,16) The ability to definitively distinguish between AC, CL, and a normal biliary tract enables a new statistical significance to be derived for each physical examination finding and laboratory variable. Understanding the significance of these predictive values should enhance clinic decision-making in the setting of RUQ pain. (15,18,19) Better knowledge of the significance of each predictive measure may obviate the need for emergent imaging studies in the evaluation of some patients with RUQ pain.

In our analysis of the predictive measures, each of three categories had variables that showed statistically significant predictive power in distinguishing the three conditions.

The first category, AC versus CL, clinically poses the most difficult diagnostic challenge. For predicting AC as opposed to CL, the predictors AlkP, TB, clinical Murphy sign, and ultrasonographic Murphy sign were statistically significant. For each incremental increase in the variables AlkP and TB, the odds of AC, in relation to the odds of CL, increase by a fixed amount. Each 1-U/L increase in AlkP increases the odds of AC by 1% over the odds of CL. Every 1-mg/dL increase in TB increases the likelihood of AC by 230% over the odds of CL.

When the variables in the AC versus CL group are dichotomized (Tables 1 and 2), AlkP, TB, and WBC remain statistically significant predictors of AC. A WBC greater than 10,000 doubles the odds of having AC over the odds of CL. AlkP greater than 128 U/L triples the odds of AC over the odds of CL. TB greater than 1.3 mg/dL increases the odds four-fold. The presence of a clinical Murphy sign doubles the odds ratio for AC. Female sex decreased the odds ratio of AC by 28%. AlkP, WBC, TB, clinical Murphy sign, and male sex individually contribute to the distinction between CL and AC. In the setting of cholelithiasis, elevations of AlkP, WBC, TB, or the presence of a Murphy sign should appropriately increase clinical suspicion of AC.

The analysis of AC versus normals reveals several statistically significant variables. On the continuous scale, AlkP, AST, N/V, and TB were statistically significant (Table 3 and 4) predictors. TB is most strongly indicative of AC, doubling the odds ratio of AC for every increase of 1 mg/dL in TB. When the variables are dichotomized, AlkP, ALT, AST, and TB remain statistically significant. GGT and WBC are also statistically significant when dichotomized. Elevated AlkP, AST, ALT, and TB increase the odds ratio of AC by six times. GGT and TB increase the odds ratio by four and five, respectively, when elevated. Sex does not reliably distinguish AC from normal gallbladders. Elevations in AlkP, ALT, AST, N/V, TB, GGT, and WBC individually may help to distinguish between AC and a normal gallbladder (Tables 3 and 4).

When contrasting symptomatic CL with normal gallbladders, the individual variables AST, ALT, and sex are statistically significant predictors of the presence of CL (Tables 5 and 6). Female sex increases the odds ratio of CL by five. The dichotomized variables of ALT, AST, GGT, and WBC show statistical significance in this category. High GGT and AST increase the odds ratio of CL by approximately three. High ALT increases the odds ratio of CL by four. High WBC increases the odds of CL by two. Age less than 36 years decreased the odds ratio of CL by 42%. AST, ALT, GGT, WBC, and sex may individually be used in distinguishing CL from normal in the setting of RUQ pain.

Conclusion

Gallbladder disease has long caused a diagnostic dilemma. Emergent imaging has often been required to distinguish CL from AC. (17,20) This study evaluates a number of laboratory and clinical findings in patients with RUQ pain for the purpose of differentiating among AC, CL, and a normal gallbladder, using ultrasound as the definitive diagnostic tool.

By capitalizing on ultrasound results, predictors are identified by which clinicians may be able to anticipate the disposition of patients before imaging. This will enable earlier notification of surgeon/admitting physician and earlier administration of appropriate antibiotics. Clinicians can apply these findings to guide their diagnostic decision making when evaluating RUQ pain in the nonicteric patient. In the future, these powerful data can be used to derive an algorithm, evaluating each value in conjunction with the other values. With a data set of values with known predictive power, the presence of AC can be accurately predicted. The development of this algorithm may decrease the need for emergent imaging in the setting of RUQ pain. The algorithm will allow patients with suspected biliary disease to be grouped into those at low risk or high risk for AC, without emergent imaging.

Limitations and future directions

This study was limited by its retrospective design; therefore, missing laboratory values may limit the power of some of the potential predictors. Only patients undergoing emergent RUQ ultrasonography were included in this study. Patients scheduled for outpatient imaging secondary to low suspicion for acute cholecystitis were not included.

The retrospective design did not allow for an assessment of interrater reliability of Murphy sign or the ultrasonographic interpretations. However, only five attending radiologists read ultrasounds at this institution, which decreases the variability in the interpretation of the ultrasonographic studies. The presence of a Murphy sign was determined by emergency medicine faculty, decreasing the likelihood of inaccuracy caused by inexperience.

The study did not include patients over 65 years of age. In this paper, potential predictors have been studied individually. In the future, a set of predictors may be derived that would provide adequate discrimination among the three conditions. If such a set is found, then prospective validation will also be required.

Table 1. Acute cholecystitis and cholelithiasis: Univariate predictors
modeling acute cholecystitis

Variable                    n      Odds   95% CI       P       [R.sup.2]

Age                         75/54  1.026  0.998-1.055  0.0704   0.04
Alk P                       75/54  1.009  1.003-1.016  0.003    0.12
ALT                         74/54  1.002  1.000-1.004  0.0465   0.05
AST                         75/54  1.002  1.000-1.004  0.0735   0.04
GGT                         73/51  1.003  1.001-1.006  0.0049   0.012
T Bili                      73/54  2.307  1.044-3.686  0.005    0.24
Amylase                     74/49  1.001  1.000-1.002  0.0709   0.06
Lipase                      74/50  1      1.000-1.001  0.1288   0.04
WBC                         75/54  1.022  0.996-1.048  0.0994   0.04
Neutrophil                  68/54  1.025  0.998-1.054  0.071    0.04
Bands                       68/45  0.977  0.834-1.015  0.7749  <0.01
Pulse                       64/43  1.016  0.991-1.041  0.2046   0.02
Temp                        66/43  1.423  0.825-2.453  0.2048   0.02
Dichotomized variables (a)
  Sex                       75/54  0.281  0.098-0.804  0.018    0.07
  N/V                       63/40  0.782  0.069-8.918  0.8431  <0.01
  Murphy +                  75/54  2.344  1.140-4.821  0.0206   0.06
  Age                       75/54  2.01   0.97-4.18    0.0616   0.04
  Alk P                     75/54  3.14   1.48-6.63    0.0028   0.1
  ALT                       74/54  1.39   0.68-2.85    0.3844  <0.01
  AST                       75/54  2      0.98-4.10    0.0584   0.04
  GGT                       73/51  1.93   0.92-4.05    0.0806   0.03
  T Bili                    73/54  4.28   1.88-9.78    0.0006   0.13
  Amylase                   74/49  1.33   0.54-3.27    0.541   <0.01
  Lipase                    68/42  1.4    0.67-3.32    0.3319   0.01
  WBC                       75/54  2.26   1.11-4.62    0.025    0.05
  Neutrophil                68/42  1.4    0.67-3.32    0.3319   0.01
  Bands                     68/45  1.56   0.45-5.42    0.4812  <0.01

Variable                          [beta]          Standard error

Age                                0.026           0.14
Alk P                              0.0009          0.003
ALT                                0.002           0.001
AST                                0.002           0.001
GGT                                0.003           0.001
T Bili                             0.836           0.239
Amylase                            0.001          <0.01
Lipase                            <0.001          <0.001
WBC                                0.021           0.013
Neutrophil                         0.025           0.014
Bands                             -0.023           0.081
Pulse                              0.016           0.012
Temp                               0.352           0.278
Dichotomized variables (a)
  Sex                             -1.271           0.537
  N/V                             -0.246           1.242
  Murphy +                         0.852           0.368
  Age                              0.698           0.373
  Alk P                            1.143           0.382
  ALT                              0.331           0.365
  AST                              0.693           0.366
  GGT                              0.659           0.377
  T Bili                           1.454           0.421
  Amylase                          0.282           0.461
  Lipase                          -0.242           0.429
  WBC                              0.816           0.364
  Neutrophil                       0.397           0.409
  Bands                            0.447           0.634

Alk P, alkaline phosphatase: Murphy +, clinical Murphy sign present:
N/V, nausea/vomiting present: Temp, temperature (T > 37.9[degrees]C);
T Bili, total bilirubin; WBC, white blood cells; bands, bands,
present.
(a) Continuous variables were dichotomized such that 1 indicates the
high values and 0 indicates the normal values, except age, where 1
indicates the high and 0 indicates the low values and gender, where 1
indicates female. Age (<36 years vs >36 years); alkaline phosphatase
(<129 vs >128 U/L); ALT (<41 vs >40 U/L); amylase (<109 vs >108 U/L);
AST (<41 vs >40 U/L), band (0 vs >0 [mm.sup.3]), GGT (<79 vs >78 U/L),
lipase (<60 vs >59 U/L); neuts (<81 vs >80%); T Bili (<1.4 vs >1.3
mg/dL); WBC (<11 vs >109/L).

Table 2. Acute cholecystitis and cholelithiasis: Univariate logistic
results for dichotomous variables

Variable    n      Sensitivity  Specificity  PPV  P     [chi square]

Sex         75/54  73%           9%          53%  0.01   6.09
N/V         63/40  97%          54%          61%  1      0.04
Murphy +    75/54  56%           3%          69%  0.02   5.46
Alk P       75/54  55%          72%          73%  0      9.24
ALT         74/54  64%          44%          61%  0.36   3.62
AST         75/54  67%          50%          65%  0.06   0.82
GGT         73/51  68%          47%          65%  0.08   3.09
T Bili      73/54  49%          81%          78%  0     12.7
Amylase     74/49  23%          82%          65%  0.54   0.38
Lipase      74/50  22%          74%          55%  0.57   0.39
WBC         75/54  63%          57%          67%  0.02   5.1
Neutrophil  68/42  43%          67%          67%  0.33   0.95
Bands       68/45  13%          91%          69%  0.48   0.5

Abbreviations and dichotomization per Table 1.

Table 3. Acute cholecystitis and normal biliary tract: Univariate
predictors modeling acute cholecystitis

Variable                  n        Odds      95% CI          P

Age                       75/48    1.002      0.973-1.031     0.9053
Alk P                     75/47    1.018      1.008-1.027     0.0002
ALT                       74/47    1.018      1.008-1.027     0.046
AST                       75/47    1.016      1.007-1.024     0.0002
GGT                       73/46    1          0.999-1.002     0.4
T Bili                    73/47    1.919      1.255-2.936     0.0026
Amylase                   74/46    1.002      1.000-1.003     0.07
Lipase                    74/45    1          1.000-1.001     1
WBC                       75/46    1.019      0.993-1.046     0.1479
Neutrophil                68/38    1.022      0.994-1.051     0.1308
Bands                     68/35    1.126      0.830-1.528     0.4441
Pulse                     65/39    0.998      0.975-1.021     0.8608
Temp                      66/40    1.274      0.788-2.059     0.3232
Dichotomized variables
  Sex                      1.375   0.402      0.4287          0.318
  N/V                      7.907   0.834      0.0132          2.068
  Murphy +                 3.427   0.4        0.0021          1.232
  Age                      0.85    0.37       0.6587         -0.163
  Alk P                    5.83    0.452     <0.0001          1.771
  ALT                      5.7     0.421     <0.0001          1.74
  AST                      5.83    0.415     <0.0001          1.764
  GGT                      4.97    0.408     <0.0001          1.603
  T Bili                   5.56    0.472      0.0003          1.716
  Amylase                  2       0.518      0.1843          0.687
  Lipase                   0.85    0.447      0.7216         -0.159
  WBC                      3.47    0.395      0.0016          1.244
  Neutrophil               1.61    0.426      0.2635          0.477
  Bands                    1.18    0.641      0.7939          0.174

Variable                  [R.sup.2]       [beta]    Standard error

Age                       <0.01            0.002     0.015
Alk P                      0.24            0.018     0.005
ALT                        0.38            0.002     0.001
AST                        0.035           0.016     0.004
GGT                        0.01           <0.001     0.001
T Bili                     0.2             0.652     0.217
Amylase                    0.09            0.002     0.001
Lipase                     0.05           <0.001    <0.001
WBC                        0.03            0.0193    0.013
Neutrophil                 0.03            0.022     0.014
Bands                      0.01            0.119     0.156
Pulse                     <0.01           -0.002     0.012
Temp                       0.01            0.242     0.245
Dichotomized variables
  Sex                      0.625-3.026    75/48     <0.01
  N/V                      1.540-40.58    63/34      0.1
  Murphy +                 1.566-7.498    75/48      0.11

  Age                      0.41-1.75      75/48     <0.01
  Alk P                    2.42-14.3      75/47      0.19
  ALT                      2.50-13.0      74/47      0.2
  AST                      2.59-13.15     75/47      0.21
  GGT                      2.24-11.0      73/46      0.18
  T Bili                   2.20-14.0      73/47      0.17
  Amylase                  0.72-5.48      74/46      0.02
  Lipase                   0.36-2.05      74/45     <0.01
  WBC                      1.60-7.52      75/46      0.11
  Neutrophil               0.70-3.72      68/38      0.02
  Bands                    0.34-4.15      68/35     <0.01

Abbreviations and dichotomization per Table 1.

Table 4. Acute cholecystitis and normal biliary tract: Univariate
logistic results for dichotomous variables

Variable    n      Sensitivity  Specificity  PPV  [chi square]  P

Sex         75/48  73%          33%          63%   0.63         0.43
N/V         63/34  97%          21%          69%   7.96         0.01
Murphy +    75/48  56%          73%          76%   9.9          0
Alk P       75/47  55%          83%          84%  17            0.001
ALT         74/47  64%          77%          81%  18.5          0.001
AST         75/47  67%          74%          81%  19.6          0.001
GGT         73/46  68%          70%          78%  16.4          0.001
T Bili      73/47  49%          85%          84%  14.7          0
Amylase     74/76  23%          87%          74%   1.81         0.18
Lipase      74/45  78%          24%          59%   0.13         0.72
WBC         75/46  63%          67%          76%  10.3          0
Neutrophil  68/38  43%          68%          71%   1.26         0.26
Bands       68/35  13%          89%          69%   0.07         0.79

Abbreviations and dichotomization per Table 1.

Table 5. Cholelithiasis and normal biliary tract: Univariate predictors
modeling acute cholecystitis

Variable         n         Odds       95% CI          P

Age              54/48      0.976     0.946-1.006     0.1185
Alk P            54/47      1.006     0.998-1.013     0.1634
ALT              54/47      1.013     1.004-1.022     0.0036
AST              54/47      1.01      1.003-1.017     0.006
GGT              51/46      1         0.999-1.001     0.6214
T Bili           54/47      0.987     0.644-1.512     0.9517
Amylase          49/46      1.001     0.998-1.004     0.387
Lipase           50/45      1         1.000-1.001     0.7033
WBC              54/46      1         0.977-1.023     0.9843
Neutrophil       42/38      0.996     0.967-1.025     0.7785
Bands            45/35      1.136     0.855-1.510     0.3789
Pulse            43/39      0.976     0.946-1.008     0.139
Temp             43/40      0.968     0.541-1.730     0.9122
Dichotomized variables
  Sex            54/48      4.9       1.633-14.69     0.0046
  N/V            40/34     10.11      1.178-86.97     0.0351
  Murphy +       42/38     10.46      0.627-3.409     0.3798
  Age            54/48      0.42      0.19-0.95       0.0364
  Alk P          54/47      1.88      0.71-4.93       0.2022
  ALT            54/47      4.09      1.73-9.69       0.0014
  AST            54/47      2.92      1.25-6.79       0.0131
  GGT            51/46      2.57      1.12-5.92       0.0266
  T Bili         54/47      1.3       0.45-3.74       0.6278
  Amylase        49/46      1.5       0.49-4.61       0.4788
  Lipase         50/45      1.09      0.43-2.75       0.8819
  WBC            54/46      1.53      0.68-3.48       0.3064
  Neutrophil     42/38      1.08      0.42-2.77       0.8673
  Bands          45/35      0.76      0.18-3.26       0.7079

Variable         [R.sup.2]     [beta]       Standard error

Age               0.03          -0.025       0.016
Alk P             0.03           0.006       0.004
ALT               0.023          0.013       0.005
AST               0.18           0.01        0.004
GGT               0.004        <-0.01        0
T Bili           <0.01          -0.013       0.218
Amylase           0.01           0.001       0.001
Lipase           <0.01          <0.01       <0.01
WBC              <0.01         <-0.01        0.012
Neutrophil       <0.01          -0.004       0.015
Bands             0.02           0.128       0.145
Pulse             0.04          -0.24        0.016
Temp             <0.1           -0.033       0.296
Dichotomized variables
  Sex             0.012          1.589       0.561
  N/V             0.012          2.314       1.098
  Murphy +       <0.01           0.38        0.432
  Age             0.06          -0.861       0.412
  Alk P           0.02           0.629       0.493
  ALT             0.14           1.409       0.44
  AST             0.14           1.07        0.431
  GGT             0.07           0.944       0.426
  T Bili         <0.01           0.261       0.539
  Amylase        <0.01           0.406       0.572
  Lipase         <0.01           0.082       0.474
  WBC             0.01           0.427       0.418
  Neutrophil     <0.01           0.08        0.478
  Bands          <0.01          -0.28        0.746

Abbreviations and dichotomization per Table 1.

Table 6. Cholelithiasis and normal biliary tract: Univariate predictors
modeling cholelithiasis

Variable    n      Sensitivity  Specificity  PPV  [chi square]  P

Sex         54/48  91%          33%          60%   9.01         0
N/V         63/34  97%          21%          69%   7.96         0.01
Murphy +    75/48  56%          73%          76%   9.9          0
Alk P       75/47  55%          83%          84%  17            0.001
ALT         74/47  64%          77%          81%  18.5          0.001
AST         75/47  67%          74%          81%  19.6          0.001
GGT         73/46  68%          70%          78%  16.4          0.001
T Bili      73/47  49%          85%          84%  14.7          0
Amylase     74/46  23%          87%          74%   1.81         0.18
Lipase      74/45  78%          24%          59%   0.13         0.72
WBC         75/46  63%          67%          76%  10.3          0
Neutrophil  68/38  43%          68%          71%   1.26         0.26
Bands       68/38  43%          68%          71%   1.26         0.79

Abbreviations and dichotomization per Table 1.

Accepted March 11, 2004

Please see Atul K. Madan's editorial on page 138 of this issue.

References

1. Friedman GD, Kannel WB, Dawber TR. The epidemiology of gallbladder disease: observations in the Framingham study. J Chron Dis 1966; 19:292-293.

2. Moscati RM. Cholelithiasis, cholecystitis, and pancreatitis. Emerg Clin North Am 1996; 14:719-737.

3. Glasgow RE, Cho M, Hutter MM, Mulvihill S. The spectrum and cost of complicated gallstone disease in California. Arch Surg 2000; 135:1021-1027.

4. Friedman GD. Natural history of asymptomatic and symptomatic gallstones. Am J Surg 1993; 165:399-404.

5. Gracie WA, Ransohoff DF. The innocent gallstone is not a myth. N Engl J Med 1982; 307:798-800.

6. McSherry, Ferstenburg H, Calhoun WF. The natural history of diagnosed gallstone disease in symptomatic and asymptomatic patients. Ann Surg 1985; 202:59-63.

7. Thistle JL, Cleary PA, Lachin JM, et al. The natural history of cholelithiasis. Ann Intern Med 1984; 101:171-175.

8. Toouli J, Wright TA. Gallstones. Med J Aust 1998;169:166-171.

9. Rosenthal RJ, Rossi RL, Martin RF. Options and strategies for the management of choledocholithiasis. World J Surg 1998; 22:1125-1132.

10. Attili AF, DeSantis A, Capri R. The natural history of gallstones: the CREPCO experience Hepatology 1995;21:656-660.

11. Barbara L, Sama C, Labate AMM, et al. A population study on the prevalence of gallstone disease: the Sirmione study. Hepatology 1987; 7:913-917.

12. Kim CK, Yun M, Lim JK, et al. Refinement of the positive predictive value of gall bladder non-visualization after morphine administration for acute cholecystitis based on the temporal pattern of gall bladder duct activity. Clin Nucl Med 2000; 25:603-607.

13. Barr RG, Agnesi JN, Schaub CR. Acalculous gall bladder disease: a controversial entity and imaging dilemma revisited. Radiology 1997; 204:105-111.

14. Shea JA, Berlin JA, Escarce JJ, et al. Revised estimates of diagnostic test sensitivity and specificity in suspected biliary tract disease. Arch Intern Med 1994; 154:2573-2581.

15. Gallagher EJ. Evidence-based evaluation of diagnostic tests. Ann of Emerg Med 1996; 28:347-349.

16. Engel JM, Deitch EA, Sekkema W. Gall bladder wall thickness: sonographic accuracy and relation to disease. Am J Radiol 1980;134:907-909.

17. Marton KI, Doubiliet P. How to image the gall bladder in suspected cholecystitis. Ann Intern Med 1988; 109:722-729.

18. Menezes N, Narson LP, deBeaux AC, et al. Prospective analysis of a scoring system to predict choledocholithiasis. Br J Surg 2000; 87:1176-1181.

19. Addison NV, Finan PJ. Urgent and early cholecystectomy for acute gallbladder disease. Br J Surg 1988; 75:141-143.

20. Lainge FC. Diagnostic evaluation of patients with suspected acute cholecystitis. Radiol Clin North Am 1983; 21:471-493.

RELATED ARTICLE: Key Points

* Alkaline phosphatase, a Murphy sign, an elevated white blood cell count, and an elevated bilirubin are statistically significant predictors of acute cholecystitis.

* With an accurate predictive model, patient disposition can be determined without emergent imaging.

* This facilitates the clinician in determining early disposition of patient and early notification of consultants.

Lisa D. Mills, MD, Trevor Mills, MD, and Barbara Foster, PHD

From the Division of Emergency Medicine, Louisiana State University at New Orleans, New Orleans, LA, and Academic Computing, University of Texas Southwestern at Dallas, Dallas, TX.

Research was performed at the Section of Emergency Medicine, Department of Surgery, University of Texas Southwestern at Dallas, Dallas, TX. This study received IRB approval from the UT Southwestern IRB.

Reprint requests to Dr. Lisa D. Mills, Division of Emergency Medicine, Louisiana State University at New Orleans, Louisiana, 1532 Tulane Ave, Room 1351, New Orleans, LA 70112. Email: [e-mail]