Comparison of calculations to estimate gastric emptying half-time of solids in humans


  • A. R. Zinsmeister,

    1. Department of Health Sciences Research, Division of Biomedical Statistics and Informatics, Mayo Clinic, College of Medicine, Rochester, MN, USA
    Search for more papers by this author
  • A. E. Bharucha,

    1. Clinical Enteric Neuroscience Translational and Epidemiological Research (CENTER), Mayo Clinic, College of Medicine, Rochester, MN, USA
    Search for more papers by this author
  • M. Camilleri

    1. Clinical Enteric Neuroscience Translational and Epidemiological Research (CENTER), Mayo Clinic, College of Medicine, Rochester, MN, USA
    Search for more papers by this author

  • Correction added after online publication 16 August 2012: ‘colonic compliance’ have been replaced by ‘gastric emptying’ in the running head.

Address for Correspondence
Michael Camilleri, Mayo Clinic, Charlton 8-110, 200 First St. S.W, Rochester, MN 55905, USA.
Tel: 507-266-2305; e-mail:


Background  Measuring solid gastric emptying (GE) at 4 h is used to identify gastroparesis. GE half-time (GE T1/2) is useful to assess overall and early GE.

Aim  To examine the validity of hourly imaging as a measurement of GE T1/2 compared with estimates from more detailed imaging.

Methods  155 human subjects (99 female, 56 male) underwent scintigraphic GE of a solid–liquid meal. We calculated the GE T1/2 using linear interpolation based on a full set of abdominal images obtained over 4 h, and the GE T1/2 based on images at 1, 2, 3, and 4 h after the meal with interpolation of data.

Key Results  Differences in GE T1/2 values (entire set of scan times compared with just the hourly scans) were small [overall median (5th, 95th percentiles) = −0.2[−7.5, 4.6] min] with slightly greater differences in males compared with females. The agreement between the two methods was very high [concordance correlation coefficient (CCC) (95% CI) = 0.993 (0.990, 0.995)] and a Bland–Altman plot indicated the variation in the results between the two methods did not change appreciably across the range of GE studied (within ±10 min for all but four subjects). Calculated GE T1/2 values, omitting the 3-h data from the hourly measurements, were associated with similar high accuracy overall and for fast GE, but were less accurate with slow GE.

Conclusions & Inferences  Results of GE T1/2 solids, using hourly imaging over 4 h, are accurate in the range 75–235 min which reflects the typical range of GE of solids in health and disease.


The conventional test for measurement of gastric emptying is scintigraphy.1,2 Using a solid-phase meal is recommended, as it quantifies the emptying of a physiologic caloric meal. Most centers use a 99mTc sulfur colloid–labeled egg-containing meal with standard imaging at 0, 1, 2, and 4 h. In fact, a 4-h gastric emptying test, using radiolabeled EggBeaters® (ConAgra Foods, Omaha, NE, USA) with jam, toast, and water and measured by external scintigraphy, is advocated by the Society of Nuclear Medicine and The American Neurogastroenterology and Motility Society as the recommended standard.2 Assessment of gastric emptying over 4 h is necessary, as extrapolation to measure T1/2 may be erroneous.3

Some regard the gastric retention of solids at 4 h as the most reliable parameter for identifying gastroparesis. Gastric emptying T1/2 (GE T1/2) is also acceptable if imaging has been performed to 4 h or at least to 50% emptying. However, it is also important to assess emptying at least 1 and 2 h after radiolabeled meal ingestion, since prolongation of the early phases of emptying may also be associated with symptoms of gastroparesis, even though the gastric retention at 4 h is normal or mildly delayed. Pharmacological studies have traditionally used GE T1/2 as an endpoint of significant interest. Traditionally, estimation of GE T1/2 has utilized detailed imaging over the 4-h period or a power exponential analysis4,5 which increases the cost of the procedure through the need for repeated imaging and special calculations. In the past, we have championed the use of hourly imaging for estimation of gastric emptying.6 The emptying phase of solids is generally linear after the initial lag phase, and gastroparesis due to neuropathic or myopathic motility disorders retards GE T1/2.5,7,8 Therefore, measurement of GE T1/2 is a relevant clinical and pharmacodynamic measurement, and given the approximately linear postlag phase, it is conceivable that imaging may be spaced out in time and still give accurate measurements of GE T1/2.

We aimed to examine the validity of the use of hourly imaging (0–4 h) as a measurement of GE T1/2 compared with T1/2 estimates obtained from more detailed imaging, every 15 min for the first 2 h and every 30 min for the subsequent 2 h.

Materials and methods

Study design

The data used to develop these simpler mathematical models were obtained as part of studies previously performed in our laboratory in healthy human volunteers and patients with upper gastrointestinal symptoms. The study was approved by the Mayo Clinic Institutional Review Board prior to initiation. All included subjects met general research authorization requirements per Minnesota state law.

Method to measure gastric emptying

The method, radiolabeled fat-containing, 296 kcal solid–liquid meal [2 scrambled eggs, 1 slice of bread and 240 mL of 1% milk (32% protein, 35% fat, 33% carbohydrate)], and conduct of the studies were described extensively in prior studies.9 One additional study10 that was included to increase the number of patients with delayed gastric emptying had used a 200 kcal, low-fat meal (4 oz. Eggbeaters®, two slices of white bread, jam, and 120 mL water). This meal is recommended by national societies of nuclear medicine and neurogastroenterology and motility.2 In summary, images of the abdomen were obtained every 15 min for the first 2 h and every 30 min in the subsequent 2 h.

Analysis of GE T1/2

Three estimates of GE T1/2, based on the proportions remaining in the stomach and obtained from the scintigraphically measured activity at each image scan time, were computed: firstly, linear interpolation using all the scan times over the 4-h imaging sequence; secondly, linear interpolation using only imaging results at 1, 2, 3, and 4 h; and thirdly, linear interpolation of scans done at 1, 2, and 4 h. This method linearly interpolates between the last point with ≥50% remaining vs the first time point with <50% remaining.


The three estimates of GE T1/2 computed for each subject were summarized for the entire group, as well as for males and females separately. The agreement of the reduced scan time estimates (i.e. second and third versions) with the estimate using the full set of scan times was assessed using Lin’s concordance correlation coefficient [CCC11] as well as visually using Bland–Altman plots. Separate analyses were also performed for patients with fast (GE T1/2 value <90 min based on full set of scan times) and slow gastric emptying GE (T1/2 value ≥180 min based on full set of scan times).


The results for the three estimates of GE T1/2 over the entire group and males and females separately are presented in Table 1: 21 participants had accelerated gastric emptying and nine had delayed gastric emptying. There was excellent agreement of the GE T1/2 estimates using the 4 hourly scans [CCC (95% CI) = 0.993 (0.990, 0.995) Fig. 1, upper panel] and using just hours 1, 2, and 4 [CCC (95% CI) = 0.974 (0.965, 0.981) Fig. 1, lower panel], with the GE T1/2 estimate using all scan time points. Similar agreement was obtained for males and females separately (Table 1).

Table 1.   Demographics of participants and gastric emptying data
  1. Summary values are mean (±SD) unless otherwise noted. CCC [95% CI] for slow and fast GE in each gender was not reported because of relatively small numbers of each gender.

  2. *GE T1/2 value <90 min (full set of scan times).

  3. GE T1/2 value ≥180 min (full set of scan times).

N 15599 (63.9%)56 (36.1%)
Age, years38 (±13)39 (±12)34 (±13)
BMI, kg m−227.6 (±5.1)27.2 (±5.1)28.4 (±5.0)
Proportion emptied at 4 h0.92 (±0.10)0.91 (±0.10)0.94 (±0.09)
GE T1/2 (min): all scan times120.6 (±31.4)126.8 (±32.0)109.6 (±27.3)
GE T1/2 (min): hourly scan times121.4 (±30.9)127.3 (±30.8)111.0 (±27.6)
GE T1/2 (min): hours 1, 2, and 4122.1 (±31.1)127.9 (±30.8)111.9 (±29.5)
Delta T1/2 : all vs Hourly−0.80 (±3.73)−0.47 (±4.09)−1.38 (±2.92)
Delta T1/2 : all vs 1, 2, and 4 h−1.50 (±6.96)−01.03 (±7.13)−2.32 (±6.61)
CCC [95% CI] hourly vs all0.993 [0.990,0.995]0.992 [0.987,0.994]0.993 [0.988,0.996]
CCC [95% CI] 1, 2, and 4 h vs all0.974 [0.965,0.981]0.974 [0.961,0.982]0.970 [0.950,0.982]
CCC [95% CI] Fast GE* (N = 21; 12 males) hourly vs all0.963 [0.915,0.984]  
CCC [95% CI] Fast GE* (N = 21) 1, 2, and 4 h vs all0.963 [0.915,0.984]  
CCC [95% CI] Slow GE (N = 9; 2 males) hourly vs all0.978 [0.915,0.995]  
CCC [95% CI] Slow GE (= 9; 2 males) 1, 2, and 4 h vs all0.520 [−0.136,0.859]  
Figure 1.

 Linear correlation of the gastric emptying T1/2 using 4 hourly images (total 5 images, Fig. 1, upper panel) or images at 1, 2, and 4 h (Fig. 1, lower panel) and detailed imaging (13 total images) over 4 h. Note the high degree of agreement between the estimates of T1/2 in 155 individuals.

The corresponding Bland–Altman plots (Fig. 2, upper and lower panels), indicating the variation in the results between the full scan time estimate and each of the reduced scan time estimates, did not change appreciably across the range of GE studied (deltas within ± 10 min for all but four subjects). For participants with slow GE (T1/2 ≥ 180 min), the CCC (Table 1) for the hourly scans was excellent compared with the full complement of scan data; in contrast (Fig. 3), there was a lower CCC for the calculated GE T1/2 based on the scans at 1, 2, and 4 h, suggesting that the hour 3 scan enhances accuracy for patients with delayed GE T1/2. In Fig. 3, note that the data for three of nine participants are not as well matched as the other 6, which are close to the line of identity.

Figure 2.

 Bland–Altman plot showing the mean difference in calculated gastric emptying T1/2 across the range of gastric emptying values when using either hourly imaging at 0–4 h (Fig. 2, upper panel) or images at 1, 2, and 4 h (Fig. 2, lower panel) compared with detailed imaging throughout the 4 h (13 images). Note the y axes are different in Fig. 2, upper and lower panels.

Figure 3.

 Linear correlation of the gastric emptying T1/2 using zero and 4 hourly images (total 5 images) or images at 0, 1, 2, and 4 h (total 4 images) and detailed imaging (13 total images) over 4 h in patients with slow gastric emptying. The interrupted line shows the line of identity. Note that three patients’ data are not optimally estimated by the hourly scan calculation of T1/2.


In this study, we wished to validate an easier method to calculate GE T1/2 of solids during scintigraphic gastric emptying studies. We demonstrated virtually identical estimates based on reduced numbers of scan times. Avoiding fitting of power exponential models and reducing the number of images while retaining accuracy over a wide spectrum of gastric emptying values from 75 to 235 min are relevant to clinical practice and pharmacodynamic studies.

In summary, our data suggest that simple linear interpolation of the gastric emptying data obtained hourly over 4 h after ingestion provides an accurate estimate of GE T1/2 This has the potential to reduce costs and to obtain a relevant parameter that appraises the entire postprandial period, not only the proportion of the meal remaining in the stomach at 4 h.


This work was supported in part by RO1-DK67071 to Dr. Camilleri from the National Institutes of Health.


The authors have no conflicts of interest.

Author contribution

ARZ was involved in concept, data analysis, authorship of paper; AEB was involved in acquisition of study data and authorship of paper; MC was the principal investigator, and was involved in study concept and authorship of paper.