Methods

We prospectively acquired a quality improvement database tracking patient flow through our TTE laboratory for 3 consecutive months. For this database, all fellows and sonographers recorded timestamps for each step of a TTE examination ( Figure 1 ). The electronic reporting system was customized to include the following timestamps: (1) begin review of requisition and previous data, (2) begin review of study priorities with the attending echocardiographer and set up equipment, (3) begin scanning, (4) finish scanning, (5) begin review of scan findings with the attending echocardiographer and clean equipment, and (6) complete entry of measurements and a preliminary report into the reporting system. At the conclusion of 3 months, data for all TTEs (n = 3,386) were downloaded to an Excel spreadsheet. For each study, TT was calculated as the time from the beginning of requisition review to completion of the preliminary report; studies without TT data (n = 853) were excluded. Interpretation time was not recorded; however, when an attending echocardiographer acquired images, that time was recorded as part of the scanning time (ST).

Components of TT. Diagrammatic representation of the steps involved in performing a complete examination in our laboratory. TT comprises the time required from obtaining the requisition to entering the data and releasing the study for staff review.

For this study, data from the quality improvement database were combined with patient and study characteristics that were abstracted from the electronic medical record. Patient factors included age, gender, weight, referral diagnosis, indication for study, new patient (first-time TTE), genetic syndrome, and hemodynamic instability. Presence of restricted acoustic windows was recorded in the echocardiogram report. Scan characteristics included study location, study under sedation, teaching examination and personnel involved (cardiology fellow vs sonographer-in-training), level of sonographer experience, preintervention or postintervention study, acquisition of tissue Doppler or 3-dimensional imaging, number of measurements entered into the report, and study limitations such as excessive patient motion or equipment malfunction. Technical charges for each examination were obtained from the hospital’s financial database. Professional charges were not included.

Anatomic lesions were assigned a complexity score using the Delphi method on the basis of input from 7 attending echocardiographers. Examples of low anatomic complexity included patients with structurally normal hearts and elevated right ventricular pressures or with lesions such as isolated atrial or ventricular septal defects. Examples of high anatomic complexity included patients with single ventricle physiology after total cavopulmonary anastomosis or those undergoing first-time evaluation for CHD such as hypoplastic left heart syndrome.

The primary outcome variable was TT, a surrogate for resource utilization. Patient and study characteristics are presented as number (percentage) for categorical variables and median (twenty-fifth to seventy-fifth percentile; range) for continuous variables. The Spearman correlation coefficient was used to examine the relation between TT and hospital charges. High resource utilization was defined as the top quartile of TT for subsequent analyses. To develop a risk prediction model for high TT, studies (n = 2,533) were randomly divided into a development sample (n = 1,686) and a validation sample (n = 847). In the development sample, Fisher’s exact and the Wilcoxon rank sum tests were conducted to evaluate the relations between candidate variables and high TT. Variables significant at the 0.20 level were considered for inclusion in a multivariate logistic regression model. A p-value <0.05 was required for retention in the final model. The discrimination of the model in the development and validation samples was quantified using the area under the receiver operating characteristic (ROC) curve (C-statistic). Data from the 2 samples were combined to create a risk score. Regression coefficients were used to assign either 1 or 2 points for each risk factor included in the final multivariate model; the points for each study were summed. The discrimination of this risk score to predict high TT was assessed using the area under the ROC curve, first treating risk score as a continuous variable and then categorizing it into low-, medium-, and high-risk groups.

Two additional analyses were conducted. In the first, candidate variables were restricted to those that are known before beginning the TTE, including patient demographics, new patient (first-time TTE), study location, study under sedation, teaching study, and preintervention or postintervention study. In the second, top-quartile ST was used as the outcome variable. As with previous analyses, a multivariate logistic regression model was created for the entire sample using all candidate variables.

The study methods were approved by the Institutional Review Board at Boston Children’s Hospital. Statistical analyses were performed using commercially available statistical analysis software (Stata version 12, College Station, Texas).

Results

Table 1 summarizes patient and scan characteristics. Median TT was 65 minutes (twenty-fifth to seventy-fifth percentile: 54, 81; range 14 to 370 minutes) with the top quartile of TT ≥85 minutes comprising the high TT group ( Figure 2 ). Median time in the room was 45 minutes (twenty-fifth to seventy-fifth percentile: 35, 60; range 4 to 310 minutes) with a median ST of 30 minutes (twenty-fifth to seventy-fifth percentile: 21, 44; range 0 to 221), and median data entry time was 13 minutes (twenty-fifth to seventy-fifth percentile: 10, 17; range 0 to 45 minutes). Common reasons for very short ST included lack of patient co-operation or extremely poor acoustic windows. A typical scenario for a very long ST comprised a new patient with highly complex anatomy such as heterotaxy syndrome with multiple cardiovascular anomalies.

Distribution of TT. Graph depicting the distribution of TT in the study cohort. Median TT was 65 minutes (twenty-fifth to seventy-fifth percentile: 54, 81; range 14 to 370 minutes). The top quartile (high TT) comprised those studies with TT ≥85 minutes. Studies with a TT >200 minutes (n = 5) are not shown on the graph.

Hospital technical charges correlated poorly with TT ( r = 0.2, Figure 3 ). In subgroup analyses, there was a similarly poor correlation between charges and TT in patients aged <4 years ( r = 0.3) and in those ≥4 years ( r = 0.12). There was also a poor correlation between charges and TT in children aged <18 years ( r = 0.22) and no correlation in those aged ≥18 years ( r = 0.12). Likewise, poor correlation was seen when restricting the analysis to outpatients ( r = 0.18) and to inpatients on cardiac floors ( r = 0.27) and noncardiac floors ( r = 0.23). The correlation coefficients were similar for new patients and for those with low anatomic complexity lesions.

Correlation between TT and technical charge. Graph showing a weak correlation between TT and technical charges.

As reported in Table 2 , the development sample identified several variables to be independently associated with TT ≥85 minutes. Age, gender, anatomic complexity score, and presence of other factors affecting imaging were not independently associated with high TT. The final multivariate model had an area under the curve (AUC) of 0.78 for predicting TT ≥85 minutes in the development sample and an AUC of 0.75 when applied to the validation sample.

Stay updated, free articles. Join our Telegram channel

Join