USF Libraries
USF Digital Collections

A comparison of community-based versus university-based centers in clinical trial performance

MISSING IMAGE

Material Information

Title:
A comparison of community-based versus university-based centers in clinical trial performance
Physical Description:
Book
Language:
English
Creator:
Stockddale, Cynthia R
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Clinical center
Quality assurance
Monitoring
Private practice
Multi-center research
Dissertations, Academic -- Epidemiology & Biostatistics -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: The success of a clinical trial is largely dependent on the clinical sites that enroll the subjects, complete the follow-up visits, and collect the data. Many clinical trials are conducted using multiple site locations. Choosing such sites to participate in a clinical trial is an important aspect of study implementation. In the past, multi-center clinical trials were conducted mainly using university-based centers. In the last few decades, private practice, or community-based, centers have been included more often in clinical trial research. As more community-based centers participate in clinical trials, it is crucial to examine how these centers might differ from university-based centers. The purpose of this project was to compare community-based and university-based centers participating in a multi-center randomized trial evaluating treatments for diabetic macular edema. Aspects of recruitment, retention, protocol adherence, data collection, and observance of study required procedures were compared. Data from 102 participating centers were examined with 40 centers categorized as university-based and 62 centers categorized as community-based. Various measures of trial performance were compared using Wilcoxon rank-sum test, repeated measures logistic regression, and repeated measures analysis of variance (ANOVA), depending on the variable being compared. Characteristics of the centers and baseline subject characteristics were compared to evaluate for possible confounding. We found that university-based and community-based centers performed similarly in almost all performance aspects compared. Notable differences included community-based centers becoming certified for participation in the study 90 days sooner on average and university-based centers having half the percentage of ungradable fundus photographs. Overall, it is recommended that community-based centers be included more often in multi-center clinical trials.
Thesis:
Thesis (M.S.P.H.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Cynthia R. Stockddale.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 26 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001999361
oclc - 318357484
usfldc doi - E14-SFE0002472
usfldc handle - e14.2472
System ID:
SFS0026789:00001


This item is only available as the following downloads:


Full Text

PAGE 1

A Comparison of Community-Based Centers versus University-Based Centers in Clinical Trial Performance by Cynthia R. Stockdale A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Public Health Department of Epidemiology and Biostatitics College of Public Health University of South Florida Co-Major Professor: Aurora Sanchez-Anguiano, M.D., Ph.D. Co-Major Professor: Roy W. Beck, M.D., Ph.D. Craig Kollman, Ph.D. Date of Approval: April 4, 2008 Keywords: Clinical Center, Quality Assurance, Monitoring, Private Practice, Multi-Center Research Copyright 2008 Cynthia R. Stockdale

PAGE 2

i TABLE OF CONTENTS LIST OF TABLES ...........................................................................................................iii ABSTRACT ......................................................................................................................iv CHAPTER 1. INTRODUCTION ....................................................................................1 Background .....................................................................................................................1 Diabetic Retinopathy Clinical Research Network ..........................................................1 Purpose of Current Study ................................................................................................3 CHAPTER 2. LITERATURE REVIEW ........................................................................5 Oncologic Clinical Trials ................................................................................................5 Ophthalmic Clinical Trials ..............................................................................................6 Summary .........................................................................................................................7 CHAPTER 3. METHODS ...............................................................................................8 Study Design ...................................................................................................................8 Eligibility Criteria .......................................................................................................8 Data Collection ...........................................................................................................8 Exposure Variable ......................................................................................................9 Potential Confounders ................................................................................................9 Outcome Variables ......................................................................................................9 Statistical Analyses .......................................................................................................12 Wilcoxon Test ............................................................................................................12 Repeated Measures ANOVA .....................................................................................12 Repeated Measures Logistic Regression ..................................................................13 CHAPTER 4. RESULTS ................................................................................................14 Center and Subject Characteristics ...............................................................................14 Outcome Measures ........................................................................................................17 CHAPTER 5. DISCUSSION .........................................................................................21 Summary of Findings ....................................................................................................21 Confounding ..............................................................................................................21 Bias ...........................................................................................................................22 Chance ......................................................................................................................22 Statistical vs. Practical Significance .........................................................................22 External Validity .......................................................................................................23 Comparison to Other Studies ........................................................................................23

PAGE 3

ii Conclusion ....................................................................................................................24 LIST OF REFERENCES ...............................................................................................26

PAGE 4

iii LIST OF TABLES Table 1. Center Characteristics According to Center Type..15 Table 2. Baseline Subject Characteristics According to Center Type..16 Table 3. Study Recruitment and Rete ntion According to Center Type Table 4. Protocol Adherence and Data Collection According to Center Type.18 Table 5. Center and Personnel Perfor mance According to Center Type..20

PAGE 5

iv A COMPARISON OF COMMUNITY-BASED CENTERS VERSUS UNIVERSITY-BASED CENTERS IN CLINICAL TRIAL PERFORMANCE Cynthia R. Stockdale ABSTRACT The success of a clinical tria l is largely dependent on the clinical sites that enroll the subjects, complete the follow-up visits, and collect the data. Many clinical trials are conducted using multiple site locations. Choosi ng such sites to partic ipate in a clinical trial is an important aspect of study impleme ntation. In the past, multi-center clinical trials were conducted mainly using university -based centers. In the last few decades, private practice, or community-based, centers have been included more often in clinical trial research. As more community-based centers participate in clinical trials, it is crucial to examine how these centers might di ffer from university-based centers. The purpose of this project was to compare community-based and universitybased centers participating in a multi-center randomized trial evaluating treatments for diabetic macular edema. Asp ects of recruitment, retention, protocol adherence, data collection, and observance of study re quired procedures were compared. Data from 102 participating centers were examined with 40 cen ters categorized as university-based and 62 centers categorized as community -based. Various measures of trial performance were compared using Wilcoxon rank-sum test, repeated measures logistic regression, and repeated measures analysis of variance (ANOVA), depending on

PAGE 6

v the variable being compared. Characterist ics of the centers a nd baseline subject characteristics were compared to evaluate for possible confounding. We found that university-ba sed and community-based cen ters performed similarly in almost all performance aspects compare d. Notable differences included communitybased centers becoming certified for participa tion in the study 90 days sooner on average and university-based centers having half the percenta ge of ungradable fundus photographs. Overall, it is recommended th at community-based centers be included more often in multi-center clinical trials.

PAGE 7

1 CHAPTER 1. INTRODUCTION Background The success of a clinical tria l is largely dependent on clin ical centers that enroll the subjects in a reasonable time, complete the follow-up visits within stated windows, and collect the data according to the protocol. Ma ny clinical trials are conducted using multiple center locations. Choosing such centers to participate in a clinical trial is an important aspect of study implementation. In the past, multi-center clinical trials have been conducted mainly at non-profit university-based centers, with clinician-inves tigators whose research interest and purpose for seeing patients within the university setting was clinical trials. In the last few decades, community-based centers (sometimes called private practice sites) have been included more often in clinical trials in an effort to increase subject accrual rates, broaden the pool of potential subjects, stream line contractual arrangements w ith a smaller entity than a university, and centralize institutional review bo ard activities, an option not available for many university-based centers. As more co mmunity-based centers become involved in clinical trials, it is crucial to examine how these sites might differ from university-based centers in trial performance. Diabetic Retinopathy Clin ical Research Network The Diabetic Retinopathy Clinical Research Network (DRCR.net) is a collaborative group dedicated to conducting multi-center clini cal trials of diabetic retinopathy and its

PAGE 8

2 associated conditions. DRCR.net is funded by the National Institutes of Health (specifically the National Eye Institute), which is a branch of the federal government. The Network was created in order to facilitate evaluation of new treatments for diabetic retinopathy by developing an infrastructure of participating cl inical centers organize d and prepared to study new treatment approaches as soon as they b ecome available.(Diabetic Retinopathy Clinical Research Network) From its inception in 2002, the DRCR.net encouraged all clinical centers with access to the necessary equipment to conduct diabetic retinopathy clinical trials to apply for participation. Other requirements for a clinical center to participate in the Network include a qualified investigator, coordina tor, photographer, visual acu ity technician, and optical coherence tomography (OCT) technician. A quali fied investigator has either completed a one year retina fellowship or has completed three years in clinical practice with at least 50% retinal patients. Currently, the DRCR.net consists of 112 active centers and 329 active investigators from 38 states throughout the Unit ed States. The open participation concept of the DRCR.net allowed a multi-center networ k consisting of both university-based and community-based centers to be created, provi ding a unique opportunity to evaluate how center type affects the conduc t of clinical trials. Summary of DRCR.net Randomized Trial Currently, the DRCR.net has five completed studies, three studies currently in the follow-up phase, and two studies currently recrui ting. The data used for the current study are from the second randomized trial initiated by the Network entitled, A Randomized Trial Comparing Intravitreal Triamcinolone Acetonide and Laser Photocoagulation for Diabetic Macular Edema. The trial is currently in its follow-up phase and the data for this analysis

PAGE 9

3 are current as of February 1, 2008. The purpose of the trial is to compare injections into the eye of triamcinolone acetonide with laser trea tment for diabetic macular edema. The study involves required follow-up visits at 4-month inte rvals with additional visits in between when necessary for care of the subject. The primary outcome visit is at 2 years, and at the time of this analysis, approximately tw o-thirds of subjects had reached this time point. Data collected at each follow-up visit includes an oc ular examination, visual acuity testing, and optical coherence tomography (referred to as OCT) which uses a dim beam of light to measure the thickness of the retina. Special photog raphs of the retina and lens (referred to as fundus photographs) are also taken annually. The majority of the data is entered at the time of the visit directly on the DRCR.net study website using elec tronic case report forms. Any edits to the case report forms during the course of the trial are tracked. The data are then monitored by the Coordinating Center for any devi ations from protocol. At regular intervals, investigators are required to sign-off on the case report form data that are entered, any edits made, and any protocol deviati ons the Coordinating Center has identified. Centers are also required to ship the OCT images and fundus photographs to a Fundus Photograph Reading Center (FPRC) within 28 days of obtaining the image or photograph. Purpose of Current Study The purpose of this study is to compare clin ical trial performanc e of university-based centers and community-based centers particip ating in a phase III clinical trial being completed by the Diabetic Retinopathy Clinical Research Network en titled, A Randomized Trial Comparing Intravitreal Triamcinolone Acetonide and Laser Photocoagulation for Diabetic Macular Edema. Aspects of partic ipation, recruitment, re tention, and protocol

PAGE 10

4 adherence in the two groups of centers were compared. It is important to note that the majority of centers in this an alysis also participated in th e Networks inaugural study, which was aimed to evaluate different types of laser treatments. This allowed centers to become familiar with the Network procedures such as electronic case report form entry, measurement of visual acuity with the Electronic Visual Acuity Tester, transmission of photographs and optical coherence tomography images (OCTs) to the Fundus Photograph Reading Center before beginning the phase 3 drug trial. Theref ore, the centers being compared in the current study had prior opportunity to adapt to the Network specific study procedures being evaluated. The purpose of this project was not to de termine if one type of center should be included in clinical trials over another. The project was aimed to determine which aspects of trial conduct might be deficient in one type of center so that thes e deficiencies can be addressed and improved upon in the future. Furt hermore, if it is found that community-based centers perform as well or better than univers ity-based centers, the use of these types of centers in government and industry-sponsored large clinical trials may be increased.

PAGE 11

5 CHAPTER 2. LITERATURE REVIEW Oncologic Clinical Trials There is limited published literature evalua ting the conduct of clinical trials at community-based centers compared with university-based centers. The first published evaluations of the clinical trial performance of community-based centers were in multi-center oncology trials. In the late seventies, community hospitals and community-based centers were first being included in large cooper ative oncology groups under direction from the National Cancer Institute, which emphasized inclusion of all potential subjects in their studies.(Koretz, Jackson, Torti, & Carter, 1983) The Eastern Cooperative Oncology Group (ECOG), which consisted only of university hospitals and large treatment cente rs, began involving community affiliates in 1976.(Begg, Carbone, Elson, & Zelen, 1982) Commun ity affiliates were either smaller hospitals or community-based centers. Sin ce these community affiliates had little or no experience conducting clinical trials previous ly, their compliance with the protocols and study outcome data were compared with the member institutions to determine whether the objectives of the studies were being met at these new centers Begg, et al, found that the member institutions had significantly lower ineligibility and protocol-violation rates than the community affiliates. However, the authors believed the difference was not enough to make any practical impact. Inadequate data s ubmission was slightly lower in the community affiliates compared with the member institut ions. No difference was found in survival,

PAGE 12

6 response to treatment, or toxic ity. Overall, the authors conc luded that community hospitals should continue to be included in th eir clinical trials.(Begg et al., 1982) A similar analysis by the Northern Calif ornia Oncology group found that community affiliates performed at least equally to the universities in all but one of the compared aspects.(Koretz et al., 1983) The affiliates had a significantly lower proportion of evaluable subjects based on secondary review of eligibil ity and treatment. This was mostly due to differences in eligibility determination be tween the community phys ician and the central study pathologist, which could be a result of the inexperience of the physicians in determining eligibility. On the other hand, prot ocol adherence and data completeness rates were higher for the community a ffiliates compared with the uni versities.(Koretz et al., 1983) Ophthalmic Clinical Trials A more recent publication compared co mmunity and university-based centers conducting ophthalmic clinical trials.(Bressler et al., 2004) The data were from the Submacular Surgery Trials Research Gr oup, a multi-center research group conducting clinical trials funded by the National Eye Institu te. Out of 27 particip ating centers, 17 were community-based and 10 were university-based. Percentage of total completed exams, completed outcome exams, completed questionnai res as well as time to submit data to the Coordinating Center and images to the Phot ograph Reading Center we re compared. Using only descriptive statistics due to the small nu mber of centers, the authors found that overall community-based centers performed approximate ly equally to that of university based centers in trial performance with the majority of the centers performing at a high level.

PAGE 13

7 However, the few centers that performed inferi orly to the others tended to be communitybased centers. (Bre ssler et al., 2004) Summary Due to the limited available data on the subject, especially recently, it is anticipated that this study will contribute greatly to the knowledge of clinical trial implementation at university-based centers and community-based cent ers. As it is becoming increasingly more common to include community-based centers in la rge clinical trial networks, it is crucial to determine how the performance of these centers compares with university-based centers.

PAGE 14

8 CHAPTER 3. METHODS Study Design The current analysis is a retrospective ex amination of center-specific data collected during a multi-center randomized trial being co mpleted by the Diabetic Retinopathy Clinical Research Network (DRCR.net) eval uating treatments for diabetic macular edema. The data was prospectively collected as part of rou tine DRCR.net procedures. Outcome measures used to evaluate clinical tria l performance were based on real -time electronic data entry of case report forms, edits, protocol deviat ions, and tracking of image shipments. Eligibility Criteria To be included in the analysis, centers ha d to have been certified in the randomized trial of interest. Of 140 centers that expresse d interest in particip ation, 102 centers were certified. Certain analyses fu rther excluded centers that were certified but never recruited subjects (N = 14) leavin g 88 centers remaining. Data Collection The trial began with the first center being certified in June 2004 and the first subject randomization in July 2004. A total of 693 s ubjects were randomized between July 2004 and May 2006. Data collection for these analyses spanned from July 2004 until February 2008. For the baseline subject characteristics, dem ographic data and a complete medical history was collected from each participant. A glyc osylated hemoglobin level and visual acuity

PAGE 15

9 testing results were also record ed. For the site characteristic s, data from the 2000 US Census were used to determine region, population and median annual household income for the city in which the center was located. Exposure Variable The center type was self-reported by each center upon joining the Network and verified by the Coordinating Center In general, centers with institutional review boards are categorized as university-based, as these centers are research oriented in purpose. These centers also have a university official who completes the cont ract with the Coordinating Center for performance of the c linical trial. Centers without institutional review boards are generally categorized as commun ity-based, as these centers are private practice oriented in purpose. Potential Confounders Measures of site characteri stics and baseline subject ch aracteristics were compared and evaluated for possible confounding. The two cen ter types were also divided into low and high subject recruitment categories to explor e whether number of subjects was a confounding factor affecting clinical trial performance. Outcome Variables Outcome variables were defined as follows: Number of protocol deviations: Count of protocol deviations entered by the Coordinating center per subject.

PAGE 16

10 Number of case report form edits: C ount of changes made to electronic case report forms per subject. Number of data queries: Count of electronic queries from the Coordinating Center to the center regarding data issues per subject. Number of adverse events: Count of adve rse event forms entered per subject. Ungradable photographs and OCTs: Whether each photograph and OCT was categorized by the Photograph Reading Center as ungradable or not. Number of recruited subject s: Count of subjects enrolled and randomized into the trial at each center. Percentage of completed visits: Percen tage of follow-up visits required per protocol that were completed; not incl uding visits completed as part of the subjects usual care or visits completed following an injection to assess for safety concerns. Percentage of visits in-window: Completed protocol visits were categorized as inwindow or out-of-window based on whethe r they were completed during the protocol-specified time period or w indow for each particular visit. Primary outcome visit completion: Activ e subjects who were past the visit window for the 2-year visit were cate gorized as having completed or not completed the primary outcome visit. Subjects who were dropped prior to the primary outcome visit were automatically categ orized as not comple ting that visit. Primary outcome visit in-window: 2-year visits that were completed were categorized as being completed in-window or out-of-window.

PAGE 17

11 Dropped subjects: Subjects were categorized as dropped if a final status form was completed by the site to disconti nue participation in the study. Number of days to become certified: Difference between the date the center expressed interest in the pr otocol and the date the Coor dinating Center certified the center to begin re cruiting subjects. Number of investigators per center: Count of inve stigators who completed Network requirements to be certif ied for participation in the study. Number of conference calls attended by the Principa l Investigator : Count of monthly investigator conference calls fo r which the primary investigator is required to attend at least a majority. Number of conference ca lls attended by the Primary Coordinator: Count of monthly coordinator conference calls fo r which the primary coordinator is required to attend at least a majority. Days until sign-off of case report forms: Difference between the date the case report form was entered on the study we bsite and the date the investigator approved the data entry. Days until sign-off of edits: Difference between the date the edit was made and the date the investigator approved the edit. Days until sign-off of deviations: Difference between the date the deviation was entered by the Coordinating Center and the date the primary investigator approved the deviation.

PAGE 18

12 Days until submission of photographs and OCTs: Difference between the time the image was taken and the time the item was logged as shipped on the study website by the center. Statistical Analyses Statistical analyses were completed usi ng SAS version 9.1. Su mmary statistics only are presented for center and baseline subjec t characteristics, which were evaluated for potential confounding. For outcome variables, th e statistical test was chosen based on the type of variable, and all p-values presented are 2-tailed. Normality of distributions was evaluated and non-parametric tests were used where appropriate. Medians and interquartile ranges are reported for all continuous variables to provide information on the distribution of the data. Because of the limitations of mu ltiple comparisons, only p-values <0.01 were considered statistically significant. Wilcoxon Test Continuous variables that c ontained only one result per center were compared using the nonparametric Wilcoxon test. These incl uded days to become certified, number of investigators per center, number of conference calls attended by the principal investigator and primary coordinator, and number of recruited subjects per center. Repeated Measures ANOVA Subject-level continuous variables were compared using repeated measures ANOVA to account for potentially correlate d data from the same center. This included number of protocol deviations, case report forms, data queries, and adverse events per subject;

PAGE 19

13 percentage of completed protocol visits per subject; percentage of visits completed inwindow per subject; days until investigator sign-o ff of case report forms, protocol deviations, case report form edits; and days until shipme nt of OCTs and photographs to the Fundus Photography Reading Center. Repeated Measures Logistic Regression Binary variables with multiple results per center were compared using repeated measures logistic regression with generalized estimating equations (GEE). This included modeling the probability that the visit is comple ted and if completed, probability that the visit is in-window. The probability that a subjec t is dropped was also modeled using repeated measures logistic regression. Photograph and OCT quality were compared by modeling the probability that an OCT or photograph was deemed ungradable by the Photograph Reading Center.

PAGE 20

14 CHAPTER 4. RESULTS Center and Subject Characteristics The study included 102 certified centers. Forty centers were categorized as university-based, and sixty-tw o centers were categorized as community-based. The participating centers were located in 38 states with 42 centers from Southern states, 25 from Midwestern states, 17 from Pacific states, 15 fr om Northeastern states, and 3 from Mountain states (Table 1). The largest proportion of university-based centers was from the Midwest (35%) whereas the largest pr oportion of community-based centers was from the South (52%). According to 2000 U.S. Census data, th e population of the cities in which the centers were located ranged from 4,081 to 8,008,278 with the median population in cities of university-based cen ters being 434,205 and the median pop ulation in cities of communitybased centers being 135,466 (Table 1). However, the recruitment pool for these centers is not necessarily confined to the city limits so it is not clea r whether this difference would affect trial performance. The median annual household income for the cities in which the centers were located was similar between university and community-based centers ($38,459 and $37,426 respectively).

PAGE 21

15 TABLE 1. CENTER CHARACTERISTICS ACCORDING TO CENTER TYPE* CENTER CHARACTERISTICS University Based Community Based N= 40 N = 62 Region: n (%) Midwest 14 (35) 11 (18) Mountain 1 (3) 2 (3) Northeast 6 (15) 9 (15) Pacific 9 (23) 8 (13) South 10 (25) 32 (52) Population: Median (25 th 75 th percentile) 434,205 (184,455, 624,064) 135,466 (42,068, 337,977) [ range] [9,019-8,008,278] [4,0811,953,631] Population Income: Median (25 th 75 th percentile) $38,459 ($31,481, $40,653) $37,426 ($31,141, $47,498) [ range] [$25,928-$72,057] [$25,000-$91,162] *Percentages do not equal 100 due to rounding A total of 693 subjects were recruited at 88 of the participating cen ters with 207 recruited at university-based centers a nd 486 recruited at community-bas ed centers. Subjects at the two types of centers were similar in age, gende r, race, diabetes type, duration of diabetes, baseline HbA1c, and baseline visual acuity (Table 2).

PAGE 22

16 TABLE 2. BASELINE SUBJECT CHARACTERISTICS ACCORDING TO CENTER TYPE SUBJECT CHARACTERISTICS University Based Community Based N centers = 35 N subjects = 207 N centers = 53 N subjects = 486 Age (yrs) Median (25 th 75 th percentile) 63 (57, 68) 63 (57, 70) [ range] [38 84] [30-86] Gender : Female n (%) 105 (51) 232 (48) Race: n (%) White 139 (67) 361 (74) Hispanic or Latino 24 (12) 64 (13) African-American 28 (14) 41 (8) Asian 9 (4) 10 (2) American Indian/ Alaskan Native 2 (<1) 3 (<1) Native Hawaiian/Other Pacific Islander 0 (0) 1 (<1) More than one race 1 (<1) 0 (0) Unknown/not reported 4 (2) 6 (1) Diabetes Type: n (%) Type 1 9 (4) 20 (4) Type 2 188 (91) 422 (87) Uncertain 10 (5) 44 (9) Duration of Diabetes (years): Median (25 th 75 th percentile) 16 (11, 23) 15 (10, 22) [ range] [0.7-56] [<.1-59] HbA1c a : Median (25 th 75 th percentile) 7.7 (6.7,8.7) 7.5 (6.7, 8.6) [ range] [5.1-14.4] [4.1-16.3] Study Eye Visual AcuityRight Eye: (N=414) Median (25 th 75 th percentile) 61 (50, 67) 59 (50, 66) [ range] [27-73] [24-73] Study Eye Visual AcuityLeft Eye: (N=426) Median (25 th 75 th percentile) 63 (54, 68) 62 (55, 67) [ range] [24-73] [25-73] only includes sites with randomized subjects (N=88) Percentages do not equal 100 due to rounding a50 subjects are missing a baseline HbA1c

PAGE 23

17 Outcome Measures Variables related to recruitment and retenti on were compared according to center type (Table 3). Community-based centers recruite d more subjects on average than universitybased centers (p = 0.05), but the difference di d not meet the p <0.01 level for statistical significance. There was no signifi cant difference in percentage of completed protocol visits per center or percentage of visits completed in-window per center. There was also no significant difference in probability that a subject completes the primary outcome visit, that the primary outcome visit is in window, or th at the subject is droppe d according to center type. TABLE 3. STUDY RECRUITMENT AND RETENTION ACCORDING TO CENTER TYPE VARIABLE University Based Community Based P-Value N centers = 40 N subjects = 207 N centers = 62 N subjects = 486 Recruited Subjects per Center: Median (25 th 75 th percentile) 4 (2, 9) 7 (3, 10) 0.05 a [ range] [0-18] [0-31] Completed Protocol Visits per Center (%) : Median (25 th 75 th percentile) 95 (89, 99) 95 (93, 100) 0.17 b [ range] [50-100] [78-100] Protocol Visits Completed in Window per Center (%) *: Median (25 th 75 th percentile) 86 (76, 93) 88 (82, 94) 0.38 b [ range] [50-100] [66-100] Completed 1 Outcome Visits ** : n (%) 127/165 (77) 290/387 (75) 0.57 c 1 Outcome Visits Completed in Window *: n (%) 121/127 (95) 285/290 (98) 0.08 c Dropped Subjects : n (%) 30/207 (14) 90/486 (19) 0.40 c only includes sites with randomized subjects (N=88) *of protocol visits that were completed **includes subjects who dropped prior to the primary outcome visit as not completed aWilcoxon rank-sum test brepeated measures ANOVA crepeated measures logistic regression (GEE)

PAGE 24

18 Comparisons of additional variables related to protocol adherence and variables related to data collection are reported in Ta ble 4. There was no significant difference in number of protocol deviations, suggesting that protocol adheren ce in general is similar at the two types of centers. In terms of data coll ection, there was no difference in number of adverse events reported per subject, suggesti ng that reporting guideli nes are being followed similarly by the two types of centers. There wa s also no significant difference in number of case report form edits, data queries from the Coordinating Center, or ungradable OCT images. However, there was a significant difference in number of ungradable photographs (p = 0.002) with community-based centers collec ting a higher proportion of these poor quality images. TABLE 4. PROTOCOL ADHERENCE AND DATA COLLECTION ACCORDING TO CENTER TYPE VARIABLE University Based Community Based P-Value N centers = 35 N subjects = 207 N centers = 53 N subjects = 486 Median (25 th 75 th percentile) Median (25 th 75 th percentile) Protocol Adherence Protocol Deviations per Subject 3 (1, 5)* 2 (1, 4)* 0.79 a Data Collection Case Report Form (CRF) Edits per Subject 23 (13, 40) 26 (14, 45) 0.79 a Data Queries per Subject 7 (5, 12) 8 (5, 13) 0.44 a Adverse Events per Subject 8 (4, 13) 7 (4, 12) 0.81 a N (%) N (%) Ungradable Photographs 45 (3) 197 (6) 0.002 b Ungradable OCTs 22 (1) 46 (1) 0.8 b All variables in table only include sites with randomized subjects (N = 88) 31 subjects at university-based centers and 62 subj ects at community-based centers had 0 deviations arepeated measures ANOVA brepeated measures logistic regression (GEE)

PAGE 25

19 Variables related to center and perso nnel performance in completing Network procedural requirements were examined to dete rmine if there were differences by type of center (Table 5). The time to become certified at community-based centers was significantly lower than the time to become certified at univ ersity-based centers (p < 0.0001). The days until investigator sign-off of case report forms was significantly higher for community-based centers compared with university-based cen ters (p = 0.001). This difference remained significant when centers were divided into lo w and high recruiters and center type was compared within recruitment group. There was also a difference in days until sign-off of protocol deviations ( p =0.02) and submission of OCTs (p=0.02), however these did not meet the p <0.01 criterion used for st atistical significance. Ther e was no significant difference in number of investigators who completed certification requir ements per center, number of conference calls attended per primary investigat or, number of confer ence calls attended per primary coordinator by center type, days until investigator sign-off of edits, or days until submission of photographs (Table 5).

PAGE 26

20 TABLE 5. CENTER AND PERSONNEL PERFORMANCE ACCORDING TO CENTER TYPE VARIABLE University Based Community Based P-Value N= 40 N = 62 Days to Become Certified: Median (25 th 75 th percentile) 208 (128, 247) 117 (77, 171) <.0001 a [ range] [36-574] [16-326] Certified Investigators per Center: Median (25 th 75 th percentile) 2 (2, 3) 3 (2,3) 0.59 a [ range] [1-7] [1-11] Conference Calls Attended per Primary Investigator: Median (25 th 75 th percentile) 7 (0, 15) 8 (2, 11) 0.77 a [ range] [0-27] [0-22] Conference Calls Attended per Primary Coordinator: Median (25 th 75 th percentile) 7 (1, 18) 7 (2, 15) 0.76 a [ range] [0-33] [0-28] Days until Investigator Sign-Off Of Case Report Forms 5 (0, 14) 7 (1, 19) 0.001 b Of Protocol Deviations 17 (7, 27) 19 (7, 40) 0.02 b Of CRF Edits 13 (4, 23) 12 (5, 27) 0.11 b Days until Submission of OCTs 18 (8, 29) 12 (6, 26) 0.02 b Days until Submission of Photographs 18 (11, 29) 12 (6, 25) 0.13 b only includes sites with randomized subjects (N=88) aWilcoxon rank-sum test brepeated measures ANOVA

PAGE 27

21 CHAPTER 5. DISCUSSION Summary of Findings In general, university and community-based centers were similar in clinical trial performance. One outcome measure that show ed a statistically si gnificant difference for which community-based centers were superior was the number of days to become certified. Outcome measures that showed a statistically significant difference for which universitybased centers were superior in cluded days until sign-off of forms and number of ungradable photographs. Although the difference in days unt il sign-off of deviati ons was statistically significant, it is not believed th at a difference of 2 days is pr actically important in terms of clinical trial performance. Confounding Summary statistics for center and subject characteristics were evaluated for signs of possible confounding. All subject ch aracteristics were similar for the two center types. The two center types varied by site region and popul ation of the city in which the center is located. However, outcome measures did not appear to differ by site region (data not shown). City population was not evaluated fu rther for confounding since it was unclear how wide the recruitment pool for centers spread bey ond city limits. It is likely that centers in smaller cities see patients from nearby cities as well. Outcome measures also did not differ by low and high recruitment. There are other possible confounders for which data were not

PAGE 28

22 available for evaluation in this study. These in clude but are not limited to experience of the investigators and coordinators in previous clinical trials, monetary resources, or other unknown factors. Bias Whenever the exposure variable is self-repo rted, as in this case, there is potential for misclassification bias. Howeve r, the Coordinating Center re viewed the classification of center type and it is unlikely that misclassificat ion occurred. One factor to note is that centers were self-selected for participation in the trial. These centers were interested in conducting research and were confident that they could adhere to the necessary trial policies and procedures. If instead, community-based ce nters were chosen at random by the sponsor to participate, the trial performance of these centers may have been different. Chance Because of the large number of outcome variables being examined, multiple comparisons in this study could have led to fals e positives. This is likely for variables that were borderline significant including number of subjects recruited, days until sign-off of protocol deviations, and days until submission of photographs (p = 0.05, 0.02, and 0.02 respectively). Statistical vs. Practical Significance Despite the limitations of multiple comparisons it is believed that the differences in a few variables are valid including days to beco me certified (p < 0.001), days until sign-off of case report forms, (p = 0.001), and number of ungradable photographs (p = 0.002). Even though these all meet the criterion for statisti cal significance, it is noted that only the

PAGE 29

23 difference in days to become certified and num ber of ungradable photog raphs have practical significance for the success of the clinical trial. An average difference of two days in signoff of case report forms would not likely affect timely dissemination of the trial results. External Validity The results from the current study could likely be applied to other multi-center networks in which the same centers participate in several studies, particularly other multicenter ophthalmology research groups. In this study, the centers had experience with the DRCR.nets procedures prior to beginning enrollment in the tria l used in the analysis. The results also may be less generalizable to a si ngle study where multiple centers are recruited to participate, as it is not clear if the two types of centers would have the same learning curve for study-specific procedures. It is not clear whether the same conclusions would be found in studies of other diseases. However, the same issues of trial performan ce affect studies of all disease types. Comparison to Other Studies The most comparable study performed by Bre ssler, et al, also f ound that universitybased and community-based centers performe d similarly when completing an ophthalmic surgery clinical trial. The study included a smaller num ber of centers and therefore only presented summary statistics. The current st udy further confirms that the two types of centers are approximately equal in most aspe cts of clinical trial performance.

PAGE 30

24 Conclusion Since community-based centers performed as well or better than university-based centers in almost all measures of clinical trial performance, it is recommended that these types of centers be considered more often for pa rticipation in clinical trial research. The two aspects for which one type was deficient compar ed with the other shou ld be addressed for future studies. For instance, university-base d centers typically take more time to become certified because of obstacles with obtaining approval from the institutional review board and negotiating detailed contracts between the unive rsity official and the sponsor for completion of the trial. Community-based centers, on the other hand, are able to use the central IRB, which typically has a quicker turn around time, and have less difficulty promptly setting up contracts with a sponsor for completing trials. The lengthy process for obtaining IRB approval and negotiating c ontracts at university-based centers needs to be addressed if these centers want to stay involved in multi-cente r trial research. For community-based centers, the fact that photograph quality was poorer than univers ity-based centers s uggests that further training in certain data colle ction methods may be needed for these types of centers, particularly if research experience is limited. Public Health Implications Allowing community-based centers to particip ate more often in clinical trials will impact those patients who may not have had access to newer, experimental therapies previously or would have had to change doctors to access such therap ies. This will also broaden the pool of study subjects in clinical trials, making st udy results more generalizable. Since the time to become certified in the st udy was significantly lower for community-based

PAGE 31

25 centers than university-b ased centers, involving these centers may also be particularly useful when studies need to be completed in a short time frame so that results can be disseminated in a timely manner.

PAGE 32

26 LIST OF REFERENCES Begg, C. B., Carbone, P. P., Elson, P. J., & Ze len, M. (1982). Participation of community hospitals in clinical trials: Analysis of five years of experience in the Eastern Cooperative Oncology Group. N Engl J Med, 306(18), 1076-1080. Bressler, N. M., Hawkins, B. S., Bressler, S. B., Miskala, P. H., Marsh, M. J., & Submacular Surgery Trials Research Group. (2004). Clini cal trial performance of communityvs university-based practices in the submacular surgery trials (SST): SST report no. 2. Arch Ophthalmol, 122 (6), 857-863. Diabetic Retinopathy Clinical Research Netw ork. General Information about the Diabetic Retinopathy Clinical Research Network (DRCR.net) Retrieved February 2008, from http://public.drcr.net/ Koretz, M. M., Jackson, P. M., Torti, F. M ., & Carter, S. K. (1983). A comparison of the quality of participation of community aff iliates and that of universities in the Northern California Oncology Group. J Clin Oncol, 1 (10), 640-644.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001999361
003 fts
005 20090413132230.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 090413s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002472
035
(OCoLC)318357484
040
FHM
c FHM
049
FHMM
090
RA425 (Online)
1 100
Stockddale, Cynthia R.
2 245
A comparison of community-based versus university-based centers in clinical trial performance
h [electronic resource] /
by Cynthia R. Stockddale.
260
[Tampa, Fla] :
b University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 26 pages.
502
Thesis (M.S.P.H.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
3 520
ABSTRACT: The success of a clinical trial is largely dependent on the clinical sites that enroll the subjects, complete the follow-up visits, and collect the data. Many clinical trials are conducted using multiple site locations. Choosing such sites to participate in a clinical trial is an important aspect of study implementation. In the past, multi-center clinical trials were conducted mainly using university-based centers. In the last few decades, private practice, or community-based, centers have been included more often in clinical trial research. As more community-based centers participate in clinical trials, it is crucial to examine how these centers might differ from university-based centers. The purpose of this project was to compare community-based and university-based centers participating in a multi-center randomized trial evaluating treatments for diabetic macular edema. Aspects of recruitment, retention, protocol adherence, data collection, and observance of study required procedures were compared. Data from 102 participating centers were examined with 40 centers categorized as university-based and 62 centers categorized as community-based. Various measures of trial performance were compared using Wilcoxon rank-sum test, repeated measures logistic regression, and repeated measures analysis of variance (ANOVA), depending on the variable being compared. Characteristics of the centers and baseline subject characteristics were compared to evaluate for possible confounding. We found that university-based and community-based centers performed similarly in almost all performance aspects compared. Notable differences included community-based centers becoming certified for participation in the study 90 days sooner on average and university-based centers having half the percentage of ungradable fundus photographs. Overall, it is recommended that community-based centers be included more often in multi-center clinical trials.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Co-advisor: Aurora Sanchez-Anguiano, Ph.D.
Co-advisor: Roy W. Beck, M.D., Ph.D.
653
Clinical center
Quality assurance
Monitoring
Private practice
Multi-center research
0 690
Dissertations, Academic
z USF
x Epidemiology & Biostatistics
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2472