- <div style="background-image:url(/live/image/gid/32/width/1600/height/300/crop/1/36399_header-volume-13.rev.1488992352.png)"/>
Micronutrients in IBD
Department of Biology
Lake Forest College
Lake Forest, Illinois 60045
Objective: To summarize rates of supplement adherence in youth with inflammatory bowel disease (IBD) and to identify predictors and outcomes of supplement adherence.
Methods: 49 adolescents (age 11-18) participated. Youth completed six monthly assessments of adherence and supplement knowledge. Youth and parents completed questionnaires. Medical record reviews provided medical and laboratory data.
Results: Mean supplement adherence rates ranged from 32% to 44%. Predictors of supplement adherence included higher knowledge (for multivitamin, iron, and calcium adherence) higher family involvement (for iron and calcium adherence), and greater inflammation (via a hematological index) for multivitamin adherence. Few relationships between supplement adherence and inflammation, growth, or nutritional outcomes were found.
Conclusion: Supplement adherence is problematic in pediatric IBD, and predictors of adherence were primarily psychosocial rather than biological variables. Future research should focus on larger sample sizes and assessments of supplement adherence that do not rely solely on patient self-report.
Overview of IBD
Inflammatory bowel disease (IBD) is an idiopathic autoimmune disorder that causes chronic inflammation in one or more sections of the gastrointestinal tract (Gore, Balthazar, Ghahremani, & Miller, 1996; Hommel & Baldassano, 2010). The location of the inflammation, as well as the depth of inflammation in the intestinal tissue can indicate one of three types of IBD: Chron’s disease (CD), ulcerative colitis (UC), or intermediate colitis (Eiden, 2003; Gore et al., 1996). In CD, inflammation may be present in any location from mouth to anus, yet in a majority of cases it is found in the terminal ileum. When diagnosing CD, the gastroenterologist will look for inflammation of the mucosa lining, epithelium invaded with neutrophils, and granuloma formation, which are indicative of CD (Eiden, 2003; Gore et al., 1996). This is unlike UC, in which patients experience inflammation localized mainly in the mucosa of the colon and rectum (Eiden, 2003; Gore et al., 1996). Indeterminate colitis is diagnosed when physicians cannot clearly distinguish between CD and UC. Guindi and Ridell (2004) found these patients are often later diagnosed with CD or UC.
Patients with IBD experience a number of primary symptoms caused by the disease and secondary symptoms caused by IBD medication. Primary symptoms include diarrhea, abdominal pain, fatigue, loss of appetite, weight loss, anemia, stunted growth, and delayed puberty. Secondary symptoms include: cushingoid appearance, weight gain, pancreatitis, and increased risk of cancer (Hommel, Denson, Crandall, & Mackner, 2008). Physicians apply a combination of symptoms, clinical examination, laboratory indices, radiology, endoscopy, and histology to make the diagnosis, to assess severity, and to predict the outcome of disease (Vermeire, Van Assche, & Rutgeerts, 2006). Unfortunately, there is not one nutritional formulation for disease management that works for all patients. Instead, physicians must closely observe changes in patients’ weight, eating habits, and gastrointestinal (GI) symptoms (Eiden, 2003). The annual incidence of CD in children is 48 out of 100,000. The annual incidence of UC in children is 29 out of 100,000 (Kappelman, Moore, Allen, & Cook, 2012). The large number of individuals afflicted by IBD demonstrates a necessity to explore the many challenges faced by IBD patients.
Medication Adherence in IBD
Medication adherence is defined as the extent to which a patient follows the guidelines given by their physician or health care team regarding their medical regimen (Stockwell-Morris & Schulz, 1992). Across multiple pediatric conditions, nonadherence rates of approximately 50% are the norm and nonadherence is considered the greatest cause of treatment failure among those with chronic medical conditions (Quittner, Modi, Lemanek, Landis, & Rapoff, 2008). Treatment regimens of IBD patients are often complex, sometimes involving multiple medications with varying dosing schedules and pill quantities, nutritional supplements, dietary modifications, medications delivered via infusions, clinic visits, and surgery (Hommel et al., 2008). Regimen complexity and other psychological factors likely combine to make treatment adherence challenging. Several studies have explored behavioral and family functioning in pediatric IBD, but few studies have focused on treatment adherence in IBD (Hommel et al., 2008). Existing studies of treatment adherence in pediatric IBD have focused primarily on adherence to oral IBD maintenance medications (i.e., medications taken to sustain disease remission). As summarized by Hommel et al., (2008), the prevalence of nonadherence to oral medication in IBD ranges from 38% to 66% depending on the type of medication and method of adherence assessment used.
Little is known about adherence to dietary supplements among youth with chronic illnesses, particularly those with IBD. Potential differences in supplement adherence and medication adherence may be attributed to a number of barriers unique to supplements including: larger pill size compared to IBD medications, bad taste of pill, lack of knowledge about the purpose of the supplement, disease activity, experience of symptoms, and differences in child and parent involvement in supplementation versus primary IBD medications (Greenley, Stephens, Doughty, Raboin, & Kugathasan, 2010; Hommel & Baldassano, 2010; Ingerski et al., 2010). Factors that influence IBD medication adherence may also influence adherence to supplements.
Kitney et al. (2009) investigated rates of supplement (multivitamins, vitamin D, iron, calcium, fish oil, and other herbal supplements) adherence and reasons given by patients for nonadherence to supplements. Adherence to supplements ranged from approximately 11% to 58 %, with fish oil having the lowest adherence rate and vitamin D having the highest adherence rate (Kitney et al., 2009). Reasons participants gave for nonadherence to their supplements included being too busy, it involved taking too many pills, and their supplements made them feel sick (Kitney et al., 2009). Additional research is necessary to validate these findings and to understand other factors that can affect supplement adherence
Relevance of Supplement Adherence in Pediatric IBD
It is important to look at supplement adherence in adolescents because disease and medication side effects associated with IBD can cause nutritional deficiencies. Because adolescents are still developing, nutritional deficiencies may have implications for short and long-term health including lack of attainment of full height potential and development of bone disease (Moeeni & Day, 2011). Thus, oral micronutrient supplementation has been one medical treatment option to prevent or treat nutritional deficiencies.
Nutritional Deficiencies in IBD
Patients with IBD experience many specific nutritional deficiencies including iron and calcium deficiencies. Many IBD patients have one or more episodes of anemia, resulting from iron deficiency. Patients can also experience more subtle iron deficiencies that are below the threshold for a diagnosis of anemia (Bager al., 2011). Across multiple studies it has been found that the approximate prevalence of anemia in IBD (30%) is only slightly lower than the approximate prevalence of iron deficiency (45%) (Munoz, Gomez-Ramirez, & Garcia-Erce, 2009). A lack of iron effects multiple essential physiological functions including oxygen transport. Iron deficiency can be exacerbated by chronic disease symptoms. IBD associated anemia is most often caused by blood loss from the bowel or decreased iron absorption due to inflammation of the digestive tract (Bager et al., 2011; Eiden, 2003). Hematological evaluations for anemia consist of a complete blood count, which includes hemoglobin (HGB), hematocrit (HCT), and iron indices such as mean corpuscular volume (MCV) or mean corpuscular hemoglobin (MCH) (Strople & Gold, 2008; Zemel, 2008). Physicians define anemia via low levels of hemoglobin or hematocrit (Strople & Gold, 2008).
Many physicians will delay iron therapy at the time of diagnosis because anemia can resolve concurrently with IBD symptoms (Gurram, Joeckel, & Stephens, 2012). However, it can be difficult to correct iron deficiency without iron supplementation or a high iron diet (Eiden, 2003). Due to the fact that iron deficiency and anemia can have a large impact on adolescent quality of life, motivation to adhere to supplements specific to this deficiency may be higher relative to youth with no symptoms (Rapoff, 2010). Additionally, adherence to iron supplements can result in a decrease in symptoms and fewer laboratory abnormalities (Eiden, 2003). Therefore, iron supplement adherence can have profound implications for improving quality of life. This is especially true for patients experiencing high disease activity.
Many patients with IBD also suffer from calcium deficiency. Inadequate intake and absorption of calcium is a mechanism by which skeletal disorders can develop. Skeletal disorders result from decreased bone density as well as chronic inflammatory and autoimmune diseases (Peterlik & Cross, 2005). Decreased bone mineral density results from multiple factors in patients with IBD. Medications such as corticosteroids, vitamin D and calcium deficiency, malabsorption, and inflammation are considered contributing factors to declining bone mass in IBD patients. (Eiden, 2003; Thayu, Semeao, & Leonard, 2008). Prednisone, a common drug for IBD patients, has been found to cause calcium loss because it leads to vitamin D resistance (Eiden, 2003). It is recommended that adolescents with IBD incorporate calcium and vitamin D supplementation as well as exercise into their regimens to prevent bone disorders (Thayu et al., 2008). Several studies have shown that calcium supplementation aids in child and adolescent bone formation (Thayu et al., 2008). In a small sample of patients diagnosed with multiple sclerosis, calcium, vitamin D, and magnesium supplementation decreased the relapse rate and had protective effects against colorectal cancer development (Peterlik & Cross, 2005). These results indicate adding calcium supplements into an IBD regimen is essential to overcoming the many barriers that patients face in calcium absorption in relation to bone health.
Disease Activity in IBD
Disease activity in the context of IBD is a multidimensional construct that can be assessed through serum biomarkers of inflammation, patient report, and physician evaluation. The presence of inflammation is one indicator of disease activity and is verified through serum biomarkers levels. Level of inflammation is used for diagnosis, determining disease activity, predicting relapse and assessing the effect of treatment (Lok et al., 2008). C-reactive protein (CRP) and erythrocytic sedimentary rate (ESR) elevate in the presence of inflammation, and are two commonly used measures (Rees & Gibson, 2011). CRP is the most commonly used biomarker for indication of acute and chronic inflammation (Rees & Gibson, 2011). CRP is an important protein produced by hepatocytes during an innate immune response, indicating inflammation or infection in various organs (Laihia, Koskinen, Waris, & Jansen, 2005; Vermeire et al., 2006). After detection of inflammation, hepatocytes are upregulated by three cytokines which increases the amount of CRP produced by the liver (Vermeire et al., 2006). Monitoring CRP levels is important for health providers to predict disease relapse. This can encourage physicians to stress to patients the importance of adherence to their medical regimens.
Another biomarker of inflammation is ESR. ESR is the rate at which erythrocytes migrate through the plasma (Vermeire et al., 2006). Normally, erythrocytes have a net negative charge and repel one another, which keeps the sedimentary rate high (Husain & Kim, 2002). Fibrinogen, a protein important for coagulation, forms bonds between platelets and becomes elevated during inflammation adding a positive charge to the serum (Husain & Kim, 2002). During times of inflammation, erythrocytes and fibrinogen attract one another and sedimentary rates decrease. An abnormal ESR can be indicative of non-specific inflammation (Vermeire et al., 2006). Unlike CRP, changes in ESR are prolonged rather than immediate (Vermeire et al., 2006). Because CRP levels can provide valuable information about current levels of inflammation and disease activity, and ESR levels provide information about chronic inflammation, both measures are important when assessing inflammation
Other methods of assessment of disease activity in the context of IBD are patient reports and physician report via physician global assessment (PGA). These assessments may affect adherence. Patients are accurate in identifying disease activity (Rapoff, 2010). Asthma and cystic fibrosis patients, who have more frequent and severe symptoms, and those with higher disease activity, are typically more willing to adhere to medication regimens than patients with fewer, less severe symptoms and lower disease activity (Rapoff, 2010). In addition, patient and parental perceptions of disease severity have proved to be more useful predictors of adherence than those of providers (Rapoff, 2010). Physician global assessment has been found to be very accurate in identifying levels of disease activity as indicated by high agreement between different physician raters (Hyams et al., 1991) PGAs have also been shown to have high agreement between physician ratings and other measures of disease activity that rely on laboratory values, medical exam results, and patient report. As found by Sewitch et al. (2003), disease activity predicts medication nonadherence in adult populations with IBD.
Nutritional & Growth Impairment in IBD
Growth impairment affects 15% to 40% of children with CD, while only 3% to 10% of adolescents with UC experience growth impairment (Gurram et al., 2012; Newby et al., 2008). Growth impairment at time of diagnosis or growth failure as an outcome of disease is more common in CD patients than in patients with UC (Moeeni & Day, 2011). This is mainly due to nutritional deprivation brought on by malabsorption and increased permeability of the ileum (Eiden, 2003; Moeeni & Day, 2011; Sturniolo, Di Leo, Ferronato, D’Odorico, & D’Inca, 2000; Vasseur et al., 2010). Paerregaard and Uldall-Urne (2005) found that the children with UC did not have any growth impairment when compared to controls. Additionally, they reported 50% of those with CD had a height below the 25th percentile for their age and sex. Growth impairment in IBD is often due to disease activity, including inflammation. In studies using mouse models, it was found that 40% of growth impairment resulted from gut-produced inflammatory protein molecules such as Interleukin-6 and tumor necrosis factor-alpha, both of which suppress levels of insulin-like growth factor (Moeeni & Day, 2011; Newby et al., 2008). Since IBD symptoms have severe implications regarding growth, it is vital to explore methods that can counteract growth impairment.
There are many causes of malnutrition and growth failure in pediatric IBD, including poor appetite, dietary restrictions, disease location, malabsorption, intestinal loss, drug-nutrient interaction, medication induced metabolic and hormonal change, and psychological disturbance (Eiden, 2003; Gurram et al., 2012). Nearly 90 percent of patients with CD experience malnutrition (Gurram et al., 2012). Poor nutrition and malabsorption can contribute to growth failure in IBD (Newby et al., 2008; Zemel, 2008). Growth status is an accurate indicator of well-being and nutritional status in adolescents (Zemel, 2008). Vasseur et al. (2010) investigated growth impairment and malnutrition in children with CD. They found that 9.5% of their sample experienced growth impairment and 32% experienced malnutrition. A follow up conducted six years after diagnosis found that 6.9% of the sample had growth impairment and 15% of the sample had malnutrition (Vasseur et al., 2010). Malnutrition and other morbidities provide reasoning for increased supplementation in adolescents with IBD
Predictors of Adherence
A number of studies have examined factors associated with higher adherence in IBD populations and among patients with other chronic medical conditions. Predictors can be grouped into two broad categories: biological or disease-related factors and psychosocial factors. With respect to biological factors, disease activity has been one factor that has been examined in relation to adherence. Stockwell-Morris and Schulz (1992) found that the presence of symptoms as reported by the patient predicted levels of nonadherence. Rapoff (2010) reviewed findings that indicated that higher disease activity was related to higher adherence among children with juvenile rheumatoid arthritis and asthma. Disease of longer duration also tended to be associated with lower adherence (Rapoff, 2010).
With respect to psychological factors, both knowledge and patterns of family involvement have been investigated as predictors of adherence. Alm-Roijer, Stagmo, Uden, and Erhardt (2004) found that patients diagnosed with coronary heart disease had better adherence to required lifestyle changes and treatment when they had a higher degree of knowledge regarding disease risk factors. Additionally, knowledge of one’s disease and regimen was found to be important, but not sufficient for medication adherence (Hommel et al., 2013). Anderson, Auslander, Jung, Miller, and Santiago (1990) found higher nonadherence in adolescents with diabetes mellitus when there was a misunderstanding or disagreement in medical regimen responsibilities between parents and youth. Decreased supervision of adherence, with assumption of increased adolescent responsibility and involvement, was related to increased age. Similarly, when expectations for self-administration of asthma medication were developmentally inappropriate, poorer adherence was a result (Walders, Drotar, & Kercsmar, 2000). Therefore, there are many factors that must be considered when attempts are made to improve nonadherence in adolescents with IBD.
Gaps in Current Research
Although substantial research addresses nonadherence to oral medication in pediatric IBD, little research exists examining rates of nonadherence to supplements in the same population. This is noteworthy, given that medical professionals and researchers have regularly discussed the necessity for supplementation in medical regimens of IBD. Furthermore, little is known about predictors of higher levels of supplement adherence. No research currently exists investigating potential beneficial outcomes of higher levels of supplement adherence such as growth, nutrition, and disease status. Thus, the focus of this study was on analyzing both biological and psychosocial predictors of adherence as well as biological outcomes of adherence relevant to adolescents with IBD.
Whenever possible, multiple measures of these predictors and outcomes were used to acquire the most comprehensive picture of the construct. A limitation of many past IBD studies is the use of a wide age range of participants. To address this limitation, the current study focused on adolescents, a group that has an increased risk for nonadherence, and therefore potentially long-term deleterious health effects. In addition, most past studies that have examined adherence, including the Kitney study of supplement adherence, have collected data at just one time point (Kitney et al., 2009). In contrast, this study is strengthened by its longitudinal approach, as it examines adherence over a relatively long period of time (i.e., 6 months), giving a more stable and comprehensive estimate of adherence, rather than the cross sectional approach as used by Kitney et al. (2009). This research also examines adherence to three of the most commonly used supplements in pediatric IBD separately, rather than examining supplements as a single category.
Current Study Aims and Hypotheses
Aim 1 was to summarize rates of multivitamin, iron, and calcium adherence within the sample of youth with IBD. It was hypothesized that rates of adherence to supplements would be similar or slightly lower than previously documented medication adherence rates in this sample population (i.e., 90% or higher; Greenley et al., 2012).
Aim 2 was to examine the role of specific biological and psychological factors in predicting adherence to supplements in youth with IBD (both individually and combined). Specific predictor domains of interest include: a) youth knowledge of the need for supplementation; b) baseline disease activity indices based on physician report, patient report, and laboratory hematological markers of inflammation; c) specific laboratory-based indices of nutritional deficiencies based on hematological markers of iron deficiency and low total calcium at baseline, and d) high levels of both parent and child involvement in disease management at baseline. It was hypothesized that 1) higher knowledge of reasons for taking the supplement would be associated with higher adherence during study participation; 2) higher levels of disease activity via PGA and patient report, and greater proportion of hematological lab abnormalities indicating inflammation at baseline would be associated with higher adherence during study participation; 3) a greater proportion of nutrient-specific hematological lab abnormalities in the year preceding study participation would be associated with higher adherence during study participation; and 4) high levels of both parent and child involvement in disease management at baseline would be associated with higher rates of supplement adherence during study participation.
Aim 3 was to examine relationships between supplement adherence during the study, disease, and linear growth outcomes during the 12 months following study participation. Specific outcomes of interest included: a) indices of disease activity based on patient report at 6 month follow up and laboratory hematological markers of inflammation in the 12 months after study participation; b) specific laboratory-based indices of nutritional deficiencies based on hematological markers of iron deficiency and low total calcium in the 12 months after study participation; and c) linear growth improvement in the 12 months following study participation. It was hypothesized that, 1) higher adherence during the study would be associated with lower levels of disease activity based on patient report of symptoms at the 6 month follow up, as well as a lower proportion of hematological lab abnormalities indicating inflammation in the 12 months following study participation; 2) higher adherence during the study would be associated with a lower proportion of hematological lab abnormalities in the year after study participation; and 3) higher adherence during the study would be associated with linear growth improvement in the 12 months after study participation.
All participants were part of a larger longitudinal study that examined adherence to thiopurine, an oral IBD maintenance medication, in English-speaking IBD patients between the ages of 11 and 18 years who had no other chronic medical conditions for which they needed to take daily medication. All participants included in the current analyses also took one or more supplements (i.e., multivitamin, iron, or calcium) daily. These three supplements were examined specifically because they are among the supplements that are most commonly utilized by patients with IBD and because they were the supplements most readily utilized by our sample. Patients who fit the above inclusion criteria were approached at the Children’s Hospital of Wisconsin Gastroenterology Clinic. When they agreed to participate, children and parents provided consent or assent and completed baseline questionnaires evaluating demographic information, disease symptoms, barriers to adherence, and youth and family functioning. At this time, youth also completed a validated, semi-structured interview to assess their knowledge of and adherence to medications and supplements via the Medication Adherence Measure (MAM) (Zelikovsky & Schast, 2008). This measure was also completed by participants every month via phone throughout the 6 months of study participation. Finally, patients were asked to complete questionnaires (similar to those completed at baseline) once more at the end of the six month study participation timeline.
For analyses, medical records were also reviewed to obtain information about patient disease activity, linear growth, nutrition and laboratory values for three time periods: one year prior to the larger longitudinal study, the 6 months during which time participants were enrolled in the study, and one year after the study.
Forty-nine of the sixty-eight participants enrolled in the larger longitudinal study were included in this analytic sample. Those that were not included either withdrew from the study prior to completing the six months of data collection, or they were not taking one of the three supplements of interest. Participants did not significantly differ from nonparticipants in age [t (66) = 0.50, p >. 05] or sex (Φ =0.09, p > .05).
Of the 49 participants, 29 were male (59%), 45 were Caucasian (92%), and 42 were diagnosed with Crohn’s disease (86%). Participants ranged from 11-18 years old [M (SD) = 14.63 years (2.03 years)]. Thirty seven (76%) patients in our sample had no disease activity at baseline based on PGA rating, while nine (18%) had mild disease activity, and three (6%) had moderate disease activity.
The majority of participants (92%; n = 45) were taking a multivitamin. Fewer were taking calcium (49%; n = 24) or iron (35% n = 17) supplements. Most subjects were prescribed two supplements (59%; n = 29); fewer were taking only one supplement (33%; n = 16). Only 8% of the sample (n = 4) was on three supplements. Among the 16 youth taking one supplement, the majority (n = 13; 81%) were prescribed a multivitamin, with fewer prescribed iron (n = 2; 13%) or calcium (n = 2; 13%). The majority of the 29 youth prescribed two supplements were taking a multivitamin and calcium (n = 18; 62%), while the remainder was prescribed either a multivitamin and iron (n = 10; 35%) or calcium and iron (n = 1; 3%).
Parents provided all demographic information at time of baseline assessment by completing a questionnaire. Information provided included adolescent’s age, gender, race/ethnicity, and socioeconomic status.
Adherence was assessed based on patient self-report on the MAM (Zelikovsky & Schast, 2008), a semi-structured adherence interview. Patients were asked to report all supplements they were prescribed as well as how many days in the past week a dose was missed for each supplement. This information was retrieved at baseline, as well as at each of the monthly follow-ups. Adherence to each supplement was calculated in the following manner: [(number of doses prescribed in a given week – number of doses reported missed in a given week) / number of doses prescribed in a given week] * 100. Each participant was given a total adherence score by averaging each monthly adherence score for each supplement individually.
Self report of symptoms. Self-reported symptoms were obtained from participants through baseline and 6 month follow up questionnaires. The IBD Symptoms Checklist was created for the larger longitudinal study based on the most common symptoms and medication side effects seen in IBD as reported by the Crohn’s and Colitis Foundation of America (2013). Based on symptoms commonly used in other measures of disease activity, a subset of IBD symptoms was chosen to create an index for the present investigation. The specific symptoms included in the patient-report of disease activity in the present study were: abdominal pain, nausea, vomiting, sleepiness, dizziness, diarrhea, loss of appetite, and fever. Patients reported experience any of the symptoms in the past four weeks, in a yes or no response format. Yes responses were coded as 1, while no responses were coded as 0. The total number of symptoms endorsed was summed to create a total disease activity score for each participant. Higher disease activity scores reflected an increased number of symptoms.
Physician global assessment of disease activity. At enrollment, a PGA rating of disease activity was acquired for each patient to determine disease activity. PGA ratings give an indication of disease activity based on a four-point rating scale where 0 is no symptoms, 1 is mild, 2 is moderate, and 3 is severe disease activity. PGA ratings have been shown to be consistent with the Pediatric Crohn’s Disease Activity Index (PCDAI) and other, more complex measures of disease activity (Hyams et al., 1991).
Hematological biomarkers of inflammation. All hematological values were retrieved from laboratory blood draws throughout the 12 months prior to the study (pre study) and during the 12 months after participants had completed the study (post study). Each biomarker was abstracted from the patient’s medical record and analyzed individually. Hematological biomarkers of inflammation included CRP and ESR. Because participants may have had more than one lab draw during the pre study or post study periods, the proportions of abnormal CRP and ESR values were computed for the pre study and post study periods. Abnormal CRP and ESR proportions were calculated identically, but separately. The total number of abnormal CRP or ESR lab values was summed for each participant during each time interval, either pre study or post study. The proportion reflected the number of abnormal lab values divided by the total number of lab values available during the given time interval. Thus, each participant was given an abnormal lab proportion for each biomarker in each time interval that was used for analyses, resulting in 4 proportions for each participant—Proportion of abnormal CRP values pre study, proportion of abnormal ESR values pre study, proportion of abnormal CRP values post study, and proportion of abnormal ESR values post study.
Hematological Indices of Nutritional Status
All nutritional outcomes were assessed via hematological lab values available in participant medical records. Lab values were retrieved from blood draws performed throughout the 12 months prior to the study (pre study) and during the 12 months after participants had completed the study (post study). Hematological indices included nutritional labs specific to calcium and iron supplementation and biomarkers indicative of inflammation. Baseline total calcium was the primary lab value used specific to calcium supplementation. Several baseline hematological indices were used for predicting iron deficiency: red blood cell count (RBC), HGB, HCT, MCV, and MCH levels. From these, a single iron deficiency score was calculated for each participant.
For both calcium and iron nutritional deficiencies, a proportion was calculated to reflect the number of abnormal lab values compared to the total number of lab values available during the either pre study or post study period. First, the total number of abnormal calcium or iron-related lab values was summed for each participant during each time interval (pre study or post study). Second, the number of total calcium or iron-related lab values was summed for each participant during each time interval (pre study or post study). The proportion reflected the number of abnormal lab values divided by the total number of lab values available during the given time interval. Thus, each participant was given an abnormal lab proportion for calcium and iron in each time interval that was used for analyses, resulting in 4 proportions for each participant: proportion of abnormal iron-related values pre study, proportion of abnormal calcium values pre study, proportion of abnormal iron-related values post study, and proportion of abnormal calcium values post study.
Supplement knowledge was assessed using the MAM at baseline and each monthly follow up (Zelikovsky & Schast, 2008). Participants were asked the following question: “What kind of medicine is this?” for each of the supplements that they were prescribed. Patients’ responses were recorded verbatim and later placed into one of three knowledge categories developed specifically for the present study. Categorization was completed for each supplement separately. Categories were developed based on degree of knowledge of their supplements, with Category 1 being the least sophisticated response and Category 3 reflecting the most sophisticated response. Children whose responses fell into Category 1 indicated that they did not know what the supplement was prescribed for (i.e. “I don’t know”). Children whose responses fell into Category 2 listed the name of supplement or described a nonspecific way the supplement was helpful to them (i.e. “Calcium” or “It makes me healthy”). Responses that fell into Category 3 reported a specific way that the supplement helped their health, body, or IBD symptoms (i.e. “Helps my inflammation”). Two raters coded all responses and inter-rater agreement for response categorization was 98%. Discrepancies were resolved by discussion. For the purposes of the analyses, raters used the most sophisticated response given by a given participant over the full study interval as their measure of knowledge in the analyses.
Parent and Child Involvement in Supplement Taking
Parent and child involvement was measured with the Inflammatory Bowel Disease Family Responsibility Questionnaire at time of enrollment (Greenley et al., 2010). Youth answered item 24 of the questionnaire: “How involved is each family member in making sure you are getting a daily multivitamin or nutritional supplement”. Children rated their own involvement as well as their perception of their mother’s involvement on a 0 to 3 scale. Families were placed into either high or low involvement groups. High involvement was defined as a score of 2 or 3 for both mother and child. In instances which either the mother or the child had a low involvement rating (0 or 1), the family would be included in the low involvement group.
Growth improvement of each participant during the 12 months following study participation was calculated via a standard means of computing growth velocity. To calculate growth velocity, each patient’s height (in centimeters) was recorded at the 6-month follow up and again at 12 months after study completion. Next, average height for each participant’s age and sex was abstracted from Center for Disease Control and Prevention (CDC) growth tables (CDC, 2009). Two z-scores were then computed: a) z1 = [(Height at 6 month follow up) – (Average height for age and sex)] / (Standard deviation of height from CDC); and b) z2 = [(Height 12 months after completion of study) – (Average height for age and sex)] / (Standard deviation of height from CDC). In accord with standard definitions used in other pediatric IBD research (Walters & Griffiths, 2008), if the difference between z1 and z2 scores was ≥ 0.5 standard deviations, that participant was considered to have growth improvement.
All analyses were performed using IBM’s Statistical Package for Social Sciences (SPSS), version 17.0.
Aim 1 Analyses
To summarize rates of supplement adherence within the sample of youth with IBD, mean levels of supplement adherence, standard deviations, and ranges of adherence scores were computed for each supplement.
Aim 2 Analyses
Aim two sought to examine the role of biological and psychological factors in predicting adherence to supplements in youth with IBD (both individually and combined). Specific predictors of interest included: a) knowledge of the need for supplementation; b) general disease activity indices; c) specific laboratory-based indices of nutritional deficiencies and d) family involvement.
To examine knowledge as a possible predictor of adherence, mean differences in levels of adherence across the three different knowledge categories were measured using an ANOVA. Effect size analyses were conducted to determine the magnitude of differences in adherence between the three knowledge categories due to the small sample size. To compute effect sizes, Cohen’s d was calculated. This statistic looks at the differences between two means divided by the standard deviation. A d value of .20 = a small effect size, a d value of .50 = a medium effect size, and a d value of .80 = a large effect size. In psychological research, most researchers agree that a medium effect size is one that suggests a meaningful real-world difference between groups. We conducted effect size analyses because of our small sample size, which made obtaining p values < .05 difficult.
The following analyses were conducted to examine disease activity as a predictor of adherence: a) correlations between patient report of symptoms at baseline and adherence; b) correlations between physician report of disease activity at baseline and adherence; c) correlations between proportion of hematologic markers of inflammation (CRP and ESR) in the year prior to study participation and adherence during the study for each supplement separately.
Examination of the possible association between proportion of nutrition related laboratory abnormalities in the year before study participation and adherence correlations were run to test for a significant relationship. Specific nutritional lab abnormalities investigated were: total calcium, HGB levels, RBC count, HCT levels, MCV, and MCH levels.
Additional correlations were run to examine the possible relationship between family involvement and supplement adherence during study participation, for each supplement separately. Participants were grouped into one of two categories based on family involvement.
Examination of the combined influence of multiple biological and psychological factors on supplement adherence was conducted via regression analyses. Regression models containing multiple independent variables were examined as predictors of supplement adherence. Predictor variables were included in these regression models if they were correlated at r = .30 or higher with adherence, the equivalent of a medium effect size. Separate regression analyses were conducted for each of the three supplements.
Aim 3 Analyses
The goal of Aim 3 was to examine relationships between supplement adherence during the study and disease and growth outcomes during the 12 months following study participation. Specific outcomes of interest were: a) indices of disease activity; b); presence of hematologic lab abnormalities associated with specific nutrient deficiencies; and c) growth improvement.
To examine the possible relationship between adherence during the study and disease activity after study participation, correlations were run between adherence during the study and youth reported symptoms at 6 month follow up. Correlations were also run between adherence during the study and proportion of hematologic markers of inflammation (CRP and ESR) in the year following study participation.
To study possible associations between adherence during study participation and proportion of nutritional lab abnormalities in the 12 months following study participation, two correlations were run. The first correlation examined associations between the proportion of calcium specific lab abnormalities and adherence rates. The second correlation examined associations between the proportion of iron specific lab abnormalities and adherence rates. Specific nutritional lab abnormalities investigated were: total calcium, HGB levels, RBC count, HCT levels, MCV, and MCH levels.
Finally, correlations were run between adherence during study participation and growth improvement in the 12 months following study participation to test for possible significant relationships. For these analyses, participants were coded as showing growth improvement versus no evidence of growth improvement.
For all analyses related to Aims 2 and 3, each supplement was examined separately. In addition, because of the small sample size, we did not rely solely on p values to determine significance of findings (since with a small sample size, we were at an increased risk of making a Type I Error). Rather, any findings meeting criteria for a medium effect size (e.g., bivariate or semi-partial correlations of .30 or higher) were considered to be meaningful findings.
Aim 1 Results
Aim 1 summarized rates of adherence to multivitamin, iron, and calcium supplements (Table 1). Mean adherence to multivitamin, iron, and calcium regimens ranged from 32% to 44%. Mean multivitamin adherence was 36.47% + 30.77 SD, mean iron adherence was 44.19% + 34.86 SD, and mean calcium adherence was 31.67% + 27.85 SD in the sample. Supplement adherence was well below average adherence of 93-98% to prescribed oral medication in a similar population (Greenley et al., 2012). Thus, our finding of mean supplement adherence rates of 32% to 44% was not consistent with the hypothesis that supplement adherence rates would be similar to or slightly lower than published oral medication adherence rates. Instead, they were much lower than oral medication rates.
Aim 2 Results
Aim 2 examined the role of specific biological and psychological factors in predicting adherence to supplements in youth with IBD. Relationships between predictors and adherence both individually and combined were examined. Specific predictors examined included: a) youth knowledge of the need for supplementation; b) baseline disease activity indices based on physician report, patient report, and laboratory hematological markers of inflammation; c) specific laboratory-based indices of nutritional deficiencies at baseline, and d) level of parent and child involvement in disease management at baseline (coded as high or low involvement). Because of the small sample size, analysis did not rely solely on p values to determine significance of findings (since with a small sample size, we were at an increased risk of making a Type I Error). Rather, any findings meeting criteria for a medium effect size (e.g., correlations of .30 or higher) were considered to be meaningful findings.
Knowledge and Adherence
One-way ANOVAS were conducted to examine the relationship between knowledge (categorized into low, moderate, and high groups) and adherence to multivitamin, iron, and calcium supplements (Figure 1). Knowledge had a significant main effect on adherence for all three supplements. The results suggested that knowledge had a main effect on multivitamin adherence [F (2, 42) = 21.29, p < .01]. Follow up analysis results showed that those with low knowledge had lower multivitamin adherence (M = 7.00% SD = 14.95), than those with moderate (M = 44.71% SD = 26.38; d = 1.64), or those with high knowledge (M = 70.91% SD = 14.79; d = 4.29). Additionally, those with moderate knowledge had significantly lower adherence than those with high knowledge (d = 1.06). All effect sizes that examined differences in adherence between the three knowledge groups were large in magnitude. These findings were consistent with the hypothesis that higher knowledge of multivitamin supplement would be related to better adherence to a multivitamin supplement.
Knowledge also had a main effect on adherence to iron supplements [F (2,14) = 5.48, p < .05] Results indicated that those with low knowledge had lower adherence (M = 15.36% SD = 22.93) than those with moderate (M = 45.71% SD = 32.95; d = 1.16) or high knowledge (M = 65.24% SD = 29.53; d = 1.85; Figure 1). The moderate knowledge group also had significantly lower adherence compared to the high knowledge group (d = 0.64). Effect sizes were found to be medium and large. These results support the hypothesis that higher knowledge of iron supplementation would be related to higher adherence to an iron supplement.
Finally, knowledge had a main effect on calcium supplement adherence (F (2,21) p < .05). Analyses indicated that those with low knowledge had lower adherence (M = 7.14% SD = 17.50), than those with moderate (M = 39.68% SD = 30.00; d = 1.26), or those with high knowledge (M = 40.02% SD = 23.16; d = 1.55; Figure 1). Effect sizes were large in magnitude, and therefore support the hypothesis stating higher calcium supplement knowledge is related to higher calcium supplement adherence. However, only a small effect size was found for the difference between the moderate and high knowledge group (d = 0.01) and therefore this comparison did not support the hypothesis.
Disease Activity and Adherence
With respect to disease activity as a predictor of adherence during the study, we used three different measures of disease activity: 1) baseline self-report of disease-related symptoms; 2) baseline physician’s global assessment of disease activity; and 3) baseline laboratory hematological indicators of inflammation. Only one meaningful correlation between disease activity and adherence was found. Specifically, a meaningful correlation was found between abnormal CRP levels and multivitamin adherence (r = .43, p > .05), such that higher inflammation prior to the study predicted better multivitamin adherence during the study.
Correlations between self-reported symptoms and multivitamin adherence were not significant (r = -.17, p > .05). Correlations between self-reported symptoms and iron adherence were not significant (r = .28, p > .05). Correlations between self-reported symptoms and calcium adherence were not significant (r = .24, p > .05). Correlations across all three supplements were small in magnitude. Thus, it was determined that the lack of significant findings was not necessarily due to the small sample size. The hypothesis that a higher number of self-reported symptoms of disease activity will predict higher adherence was not supported.
Additionally, no significant correlations were found between baseline physician’s global assessment of disease activity and supplement adherence for all three supplements (multivitamin adherence r = -.09, p > .05; iron adherence r = -.02, p > .05; and calcium adherence r = .24, p > .05). All three of these correlations failed to meet the cut off for a designation of a small effect size and thus were found to be equal to or less than small in magnitude. Therefore, the hypothesis that physician’s assessment of disease activity was associated with supplement adherence was not supported.
Regarding inflammation as a predictor of adherence to multivitamin, calcium, and iron adherence, two different baseline hematological indices of inflammation were used: CRP and ESR. No significant correlations were found between abnormal CRP levels and iron adherence (r = .24, p > .05), nor between abnormal CRP levels and calcium adherence (r = .01, p > .05). Correlations between abnormal CRP levels and iron adherence, and abnormal CRP levels and calcium adherence presented small effect sizes. No significant correlations were found between abnormal ESR and iron adherence (r = .24, p > .05), multivitamin adherence (r = .02 p > .05) or calcium adherence (r = -.04, p > .05). These correlations were either small or less than small in effect size. Taken together, results examining CRP and ESR as measures of inflammation offered little support for the hypothesis that with an increased proportion of abnormal inflammatory markers present at baseline, there would be increased adherence to the three supplement categories throughout the study. Specifically, in only one of six analyses this hypothesis was supported.
Laboratory-Based Nutritional Abnormalities and Adherence
The presence of nutritional abnormalities the year prior to study participation was used to predict adherence to iron and calcium supplements during study participation. Nutritional abnormalities included factors indicative of iron deficiency (correlated with iron adherence only) and total calcium (correlated with calcium adherence only). Regarding factors indicative of possible iron deficiency, the following hematological indices were used: HGB, RBC, HCT, MCV, and MCH. Correlations were run using the proportion of abnormal laboratory indices in the year prior to enrollment as the predictor. No significant correlation was found between presence of abnormal iron labs and iron adherence (r = -.12, p > .05). These findings did not support the hypothesis that those with a greater proportion of abnormal iron labs would have better adherence to iron supplements during the study.
Finally, patients’ proportion of abnormal calcium lab values during the one-year before enrollment was examined as a possible predictor of calcium adherence. It was found that those with low total calcium one year prior to the study were not more likely to have higher adherence to a calcium supplement during the study (r = .27, p > .05). This finding did not support the hypothesis that those with a greater proportion of abnormal calcium labs present in the year prior to the study would have better adherence to their calcium supplement during the study.
Family Involvement and Adherence
Correlations were examined between youth and mother involvement at baseline and adherence to multivitamin, iron, and calcium supplements, respectively. Meaningful correlations were found between involvement and both iron adherence (r = .59, p < .05) and calcium adherence (r = .33, p > .05), while no meaningful correlation was found between multivitamin adherence and involvement (r = .29, p > .05). The correlation between iron and involvement was medium in magnitude, and the correlation between calcium adherence and involvement was small in magnitude. This supported the hypothesis that the group of families with high youth and mother involvement in administration of supplements would have higher adherence to those supplements compared to families in which one or both family members had low levels of involvement.
Linear multiple regression analyses were run in order to examine the combined impact of several of the above independent variables on supplement adherence. Analyses were run with all independent variables entered using simultaneous entry. Only independent variables with correlations of .30 or larger (the equivalent of medium effect sizes) with the dependent variable were chosen for inclusion in the regression analyses. Three regressions were run in total, one for each supplement dependent variable (i.e., multivitamin, iron, and calcium).
Results of a regression analysis in which multivitamin adherence was the dependent variable showed that multivitamin knowledge and proportion of abnormal CRP levels combined contributed 48.4% of the variance to the predication of multivitamin adherence (Multiple R = .70; F (12, 11) = 5.16, p < .05; Table 2). After controlling for the impact of knowledge, proportion of abnormal CRP levels were not a statistically significant predictor of multivitamin adherence (β involvement = .35; t = 1.59, p > .05). However, when controlling for the effect of CRP abnormality, knowledge was a statistically significant predictor of multivitamin adherence (β involvement = .55; t = 2.52, p < .05). Examination of effect sizes by looking at the squared semi-partial correlations showed a medium effect size for multivitamin knowledge (sr2 = .30) and a small effect size for abnormal CRP (sr2 = .12). These results support the hypothesis that combining independent variables would account for even more variance in multivitamin adherence than each of the independent variables alone (Table 2).
For the regression analysis in which iron adherence was the dependent variable, iron knowledge and family involvement variables combined contributed 50.3% variance to the prediction of iron adherence, a statistically significant result (Multiple R = .71; F (2, 13) = 6.59, p < .05; Table 3). After controlling for the impact of family involvement, knowledge was not a statistically significant predictor of iron adherence (β knowledge = .48; t = 2.02, p > .05). After controlling for the impact of knowledge, family involvement was not a statistically significant predictor of iron adherence (β involvement = .32; t = 1.34, p > .05). Examination of effect sizes by looking at the squared semi-partial correlations presented a small effect size for knowledge (sr2 = .16) and a less than small effect size for involvement (sr2 = .07). These results support the hypothesis that combining independent variables would account for even more variance in the prediction of iron adherence than each of the independent variables alone.
In a regression analysis in which calcium adherence was the dependent variable, calcium knowledge and family involvement variables combined accounted for 18.9% of the variance in the prediction of calcium adherence. This finding was statistically significant (Multiple R = 0.55; F (2, 21) = 4.61, p < .05; Table 4). After controlling for the effect of knowledge, family involvement was not a statistically significant predictor of calcium adherence (β involvement = .34; t = 1.87, p > .05). However, after controlling for family involvement, knowledge was a statistically significant predictor of calcium adherence (β involvement = .44; t = 2.42, p < .05). Effect sizes, determined by examining squared semi-partial correlations, were small for both calcium knowledge (sr2 = .19) and family involvement (sr2 = .11). These results support the hypothesis that combining independent variables would account for even more variance in calcium adherence than the each of the independent variables alone.
Aim 3 Results
Aim 3 examined relationships between supplement adherence during the study and disease and growth outcomes during the 12 months following study participation. Specific outcomes of interest included: a) indices of disease activity; and b) presence of hematologic lab abnormalities associated with specific nutrient deficiencies. Because of the small sample size, we did not rely solely on p values to determine significance of findings (with a small sample size, we were at an increased risk of making a Type I Error). Rather, any findings meeting criteria for a medium effect size (e.g., correlations of .30 or higher) were considered to be meaningful findings.
Disease Activity and Adherence
When examining the relationship between supplement adherence during the study and disease activity outcomes, two different measures of disease activity were used: 1) presence of self-reported disease-related symptoms at the six month follow up; and 2) the proportion of abnormal hematological indicators of inflammation collected during the 12 months following study participation. No significant correlations were found between supplement adherence and self-reported disease symptoms for any of the three supplements (multivitamin adherence r = -.09, p > .05; iron adherence r = -.01, p > .05; and calcium adherence r = .16, p > .05). Correlations between multivitamin adherence and iron adherence with self-reported disease activity did not meet the cut off for a small effect size. The correlation between calcium adherence and self-report of disease activity was small in magnitude. These findings did not support the hypothesis that higher adherence would be related to the outcome of fewer self-reports of disease-related symptoms at the six-month follow up.
Correlations were also run to examine the relationship between supplement adherence during the study and indicators of disease inflammation as outcomes. The proportions of two different hematological indices of inflammation, CRP and ESR, collected during the year following study participation were used in analyses. A meaningful correlation was found between calcium adherence and proportion of abnormal CRP levels (r = .35, p > .05), yet no significant correlations were found between multivitamin adherence and proportion of abnormal CRP levels (r = -.03, p > .05), or between iron adherence and proportion of abnormal CRP levels (r = -.07, p > .05). The correlation between calcium adherence and proportion of abnormal CRP levels was found to be medium in magnitude. Correlations between multivitamin adherence and proportion of abnormal
Eukaryon is published by students at Lake Forest College, who are solely responsible for its content. The views expressed in Eukaryon do not necessarily reflect those of the College.
Articles published within Eukaryon should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.