![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
,**,2
* Department of Medicine, Division of Endocrinology,
Center for Human Nutrition, and
** Department of Preventive Medicine and Biometrics, University of Colorado Health Sciences Center, Denver, CO
1To whom correspondence should be addressed. E-mail: Janine.higgins{at}uchsc.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT SUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
KEY WORDS: • dietary compliance • lithium • adherence
With long-term dietary intervention studies, it is important to ensure that free-living subjects adhere to the prescribed diet. However, current methods to monitor adherence in free-living settings, including random 24-h food recall questionnaires, FFQs, weighed food records, and maintenance of food diaries are notoriously problematic. Studies using doubly labeled water showed that underreporting of food intake is a common problem for all self-report methods and subjects routinely underreport their energy intake by 20–50% (1–5). Although self-report methods are useful in determining the subject’s habitual dietary intake, they are less desirable as dietary compliance tools. Therefore, a tool that does not rely on self-reported food intake would be useful to assess dietary compliance in any dietary intervention study.
There are several biochemical markers that have been examined for use in measuring dietary compliance. These include riboflavin, inulin, para-aminobenzoic acid, low-dose phenobarbitone, and lithium (Li). In a long-term feeding trial, it would be desirable to use a marker that is inexpensive, has very low toxicity, a long half-life, can be measured in blood (so that subjects do not have to undertake 24-h urine collections), and is relatively easy to assay. Of the compounds that can be used to measure compliance, Li possesses the additional advantage of having a high melting point, which allows it to be baked directly into foods prepared for subjects. Traditionally, Li estimates were performed on 24-h urine samples. However, it was recently reported that trace amounts of Li can be measured in serum samples (6).
Lithium is also a good candidate as a marker of compliance due to extensive knowledge of its pharmacokinetics. Serum Li concentrations were demonstrated to peak 1–3 h after ingestion and steady-state Li concentrations are achieved after 3–4 d of administration (7). The therapeutic dose of Li for treatment of bipolar disease is 10–50 mmol/d, whereas the tracer dose given to measure dietary compliance is 0.25 mmol/d (6). The baseline level of Li (reflecting Li in foods) is 
0.6 µmol/L; the tracer dose results in serum Li concentrations of 4–10 µmol/L, and therapeutic doses give concentrations of 0.4–1.0 mmol/L (6,7). Toxicity was reported to occur at concentrations >1.5 mmol/L (7). Therefore, Li appears to be minimally present in usual foods, and measurement of its concentration in serum after administration of tracer doses is a safe and stable method with which to assess dietary compliance.
However, the one study to evaluate serum Li as a measure of compliance by deRoos et al. (6) suggested that, due to large interindividual variation, Li was not a good measure of compliance in individual subjects. The investigators came to this conclusion by initially determining steady-state Li at one time point (the morning before the next dose) after 14 d of supplementation. Subsequently, the dose was cut in half for 2 d and serum Li concentrations were determined. Given that the time for Li to reach steady state is 3–4 d and that Li peaks 1–3 h after the last dose, it is unclear whether subjects were in steady state at the time of these measurements. Thus, it is possible that the conclusions of deRoos et al. (6) might not be fully justified. This study was therefore designed to test the hypothesis that a dose-response curve for serum Li could be determined under free-living conditions and that this could be used as a test for dietary compliance.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Each subject consumed 3 diets of different Li concentration thereby acting as his/her own control. Subjects were admitted to the GCRC for a basal blood draw on d 1 to determine serum Li and creatinine concentrations. Creatinine clearance was calculated using the Cockcroft and Gault formula (8) to assess kidney function. Subjects were then asked to consume a breakfast with a single food supplemented with Li citrate. Subjects were required to consume, under supervision, all of the food and beverages provided. They were advised to eat their normal diet for the remainder of the day.
Subjects received 3 different doses of Li, in random order, over the 3 wk of the study. These doses were 250, 213, and 175 µmol Li/meal (i.e., per day) designed to approximate compliance levels of 100, 85, and 70%, respectively. Each week, on d 1, a blood sample was taken to ensure that serum Li levels did not change compared with prestudy values. Subjects then received a Li-supplemented breakfast, under supervision, for 4 d (Li contained in omelet, pancake, oatmeal, or muffin on Monday, Tuesday, Wednesday, or Thursday, respectively). On d 4, blood samples were taken before (time = 0 h) and at 1, 2, 3, 5, 7, 9, and 24 h after ingestion of the test meal.
Serum creatinine was measured with the Jaffé reaction using a Beckman Coulter Synchron LXi 725 Clinical System analyzer (Beckman Coulter). Serum Li concentration was determined by inductively coupled plasma MS (PlasmaQuad 3, VG Elemental). Beryllium (Be, Aldrich Chemical) was used as an internal standard. Serum samples and Li standards were diluted using 4.4 µmol/L Besolution in 2% (v:v) HNO3 (Optima, Fisher Scientific). To verify the accuracy of the measurement, Standard Reference Material 909b, Lyophilized Human Serum (U.S. Department of Commerce National Institute of Standards and Technology), was inserted between samples in each run. The mean recovery of lithium in SRM was 102% with a relative SD of 2.18%.
Data analysis. Participants’ characteristics are presented as means ± SEM with ranges. Serum Li concentrations are summarized using means, SD, and CV at each time and for each dose. Decreases in serum Li from the 100% dose vs. the 85% dose and from the 100% dose vs. the 70% dose were estimated and tested using a linear mixed model to account for repeated measurements on subjects. Analyses were done separately at each of the 6 measurement times because this gave the clearest answers to the questions of interest. A factor for order of dose administration was included but was nonsignificant except at 24 h (P = 0.04). Because this factor had no substantive effects on the results, the results from the model omitting the period factor are presented for simpler interpretation.
We also evaluated the potential for using serum Li as a test for noncompliance of individual subjects during a study. We assumed a design similar to the present study, i.e., before the study begins, the subject will be given a 100% dose of Li as described above and serum Li will be measured at a given time later (e.g., 3 h). Then, on a study day, serum Li is measured at approximately the same time after the last Li-containing food was consumed. If the subject’s serum Li on the study day is more than C units (see online supplemental file for calculation)3 below their serum Li on the 100% dose at that time, the subject would be declared noncompliant on that study day (i.e., they have a positive diagnostic test results for noncompliance). We used data from the present study to estimate C so that the probability of a subject being declared noncompliant by the test, when in fact they were compliant (i.e., did eat all of the required foods), is 0.05 (i.e., the specificity of the diagnostic test is set to 95%). Using these values of C, we evaluated the ability of the test to detect noncompliant subjects (probability of a subject being declared noncompliant by the test when in fact they were noncompliant, or sensitivity). Separate tests were developed at each time (0, 1, 3, 6, 9, 24 h) and each dose (70 and 85%). Details of these calculations are given in the online supplemental information.3 Differences were considered significant at P 
0.05.
|
RESULTS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
85% were significantly different from the 100% dose in the 7 subjects. We also evaluated the potential use of Li as a test of individual subject compliance in a dietary study. The cutoff value, C, was derived as the probability of having serum Li low enough to be declared noncompliant (i.e., serum Li on the study day is more than C µmol/L below that after the 100% dose; Table 2) when in fact the individual was compliant is 5% (i.e., specificity of the diagnostic test is 95%). Using these values, it appears that the most effective times to test for compliance are 3, 6, and 9 h after ingestion of Li because those times gave the highest probabilities of subjects being declared noncompliant when in fact they were not compliant (i.e., sensitivity) at both the 70 and 85% doses (Table 2). At those times, the compliance criteria resulted in tests that detected 90–95% of subjects noncompliant at the 70% dose. The detection rates at the 85% dose were 34–52%. Thus, Li as used in our study is likely capable of detecting noncompliant individual subjects at doses 70%.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
70%. These results suggest that a Li marker can be used effectively to test for even moderate levels of noncompliance of individual subjects. These conclusions are in contrast to those of deRoos et al. (7), who stated that Li would not be a good marker for small lapses in compliance but would be valuable to test for gross noncompliance. There are several reasons for such disparate conclusions. Foremost, this study design allowed a direct test of compliance at the 85 and 70% levels over several time points and at steady-state concentrations, whereas the subjects in the deRoos study might not have been at steady state on the "noncompliant" 50% dose because measures were made after only 2 d of a dosage change. Additionally, all serum Li measurements in the deRoos study were made just before the next dose (i.e., the nadir of Li concentration), which proved to be a time of high variation in the current study, thus limiting the ability to detect differences between a full dose and a noncompliant dose. Of note, the findings of our study were in a subject population with a relatively wide range of weight, further supporting such a method for use in dietary intervention studies.
As previously reported, using Li as a marker of compliance and monitoring compliance with serum Li levels proved to be a safe and inexpensive method. Pharmaceutical grade Li for dosing is readily available. The use of serum samples obviates the need for 24-h urine collections, a procedure fraught with its own difficulties in compliance. Additionally, Li is stable both in cooked foods and in stored serum awaiting analysis. Finally, the use of MS to measure Li is very sensitive and allows for a relatively high throughput.
Several modifications could be made to the simple compliance testing procedure we described. For example, measuring a time course at the 100% dose as we have done for each subject at the start of the study would allow future, participant-specific measurements and compliance tests during the study to be made at any one of the times (3, 6, or 9 h), thus providing greater flexibility. In longer-term trials, it may not be reasonable to omit subjects who fail a single compliance test, which suggests using multiple tests during the study as a criterion for inclusion or omission of subjects’ data. Because Li is cleared through the kidneys and creatinine clearance provides a measure of kidney function, correcting serum Li for creatinine clearance might further decrease the variability and increase the sensitivity of this test.
There are several limitations to the present study. Foremost, the use of Li as a marker for compliance is applicable only as a measure of adherence to a diet provided by the study. This method is not able to detect additional consumption of either that macronutrient or of other foods outside of the prescribed diet. Second, although Li was administered in the context of a free-living setting with a diet that was controlled only during Li administration, the use of these methods in truly free-living subjects might add variability. Extrapolation of these results and methods to subjects outside the confines of an experimental setting would require future studies.
In summary, this study shows that Li can be used to detect moderate levels of noncompliance to a dietary component or macronutrient. From these data, the ideal means to use Li as a marker would be to establish a steady-state serum Li concentration in subjects (4 d of observed therapy of 100% dose) using the macronutrient of interest, and then measuring serum Li concentration. Subsequently, in long-term dietary studies, the macronutrient of interest can be supplemented with Li, and serum Li measured 4 d after dosing. These data suggest that the best times for measuring Li would be 3–9 h after the last dose, and the results shown in Table 2 for those times indicate that this protocol and methods could provide an acceptable and clinically useful test for individual noncompliance.
|
FOOTNOTES |
|---|
3 Calculations for testing compliance of individuals are available with the online posting of this paper at www.nutrition.org.
Manuscript received 29 April 2004. Initial review completed 24 May 2004. Revision accepted 2 September 2004.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1. Goris, A., Meijer, E., Kester, A. & Westerterp, K. (2001) Use of a triaxial accelerometer to validate reported food intakes. Am. J. Clin. Nutr. 73:549-553.
2. Black, A. & Cole, T. (2001) Biased over- and under-reporting is characteristic of individuals whether over time or by different assessment methods. J. Am. Diet. Assoc. 101:70-80.[Medline]
3. Tran, K., Johnson, R., Soultanakis, R. & Matthews, D. (2000) In-person vs telephone administered multiple-pass 24-hour recalls in women: validation with doubly labeled water. J. Am. Diet. Assoc. 100:777-783.[Medline]
4. Kaczkowski, C., Jones, J., Feng, J. & Bayley, H. (2000) Four-day multimedia diet records underestimate energy needs in middle-aged women as determined by doubly-labeled water. J. Nutr. 130:802-805.
5. Kroke, A., Klipstein-Grobusch, K., Voss, S., Moseneder, J., Thielecke, F., Noack, R. & Boeing, H. (1999) Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods. Am. J. Clin. Nutr. 70:439-447.
6. De Roos, N., de Vries, J. & Katan, M. (2001) Serum lithium as a compliance marker for food and supplement intake. Am. J. Clin. Nutr. 73:75-79.
7. Parfitt, K. eds. Martindale: The Complete Drug Reference 32nd ed. 1999:290-296 Pharmaceutical Press London, UK. .
8. Cockcroft, D. W. & Gault, M. H. (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16:31-41.[Medline]
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
