Abstract
Purpose
Type 2 diabetes mellitus (T2DM) is the subject of numerous randomized controlled trials (RCTs). The validity of RCTs may be threatened by attrition bias due to the discontinuation of the study. The aim of this systematic review is to evaluate the reasons of patient’s withdrawal from these RCTs.
Methods
A systematic literature search on PubMed, Cochrane Library, Web of Science, and Scopus databases was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram. The aim was to obtain all relevant blinded RCTs published before January 2017 in which the effectiveness of synthetic drugs, vitamins/minerals were compared to that of placebo or active control in T2DM. The quality of RCTs was assessed using the Jadad score. The frequency of withdrawal reasons was presented based on treatments with placebo/active control, national/international level of the studies, and publication year. Meta-analysis was not performed due to the heterogeneity.
Results
Overall, 1368 articles comprising of 640,780 subjects were included. In the majority of the RCTs (75.0%), the intervention and the placebo arms were compared. Most of the included studies (96%) were classified in the high-quality category (Jadad score≥3). The highest proportion of reported withdrawal cases was found in international studies, national RCTs conducted in Japan, and RCTs published in 2011. The withdrawal reasons were reported for 91,669 (63.75%) of the total 143,794 participants who had withdrawn from these studies. The main reported reasons were “adverse effects” (24.04%), “withdraw consent” (16.10%), and “missing data” (11.08%). Variations in the reported withdrawal reasons were based on the country or published year. RCTs with triple blinded design as well as those in which anti-hyperlipidemia and anti-obesity medications were applied, showed significantly higher probability of reported the withdrawal.
Conclusion
High proportion of reported discontinuation in blinded RCTs on patients with T2DM was related to drug adverse effects. Overall, the total number and reason of drop out were unsatisfactory.
Supplementary Information
The online version contains supplementary material available at 10.1007/s40199-020-00380-7.
Keywords: Diabetes mellitus, Randomized controlled trials, Bias, Withdrawal, Discontinuing, Reasons, Systematic review
Introduction
Diabetes mellitus is a multifactorial disease defined by chronic hyperglycemia secondary to defect in insulin secretion, insulin function, or both. It is classified into two main forms, type 1 and type 2, with the majority of the cases belonging to the type 2 [1]. The key role for oxidative stress in the pathophysiology of type 2 diabetes mellitus (T2DM) through several mechanisms is well established [2, 3]. Regardless of the mechanism, T2DM and its serious comorbidities, including cardiovascular disease, neuropathy and nephropathy are among the most growing challenges of the global health [4]. According to the International Diabetes Federation (IDF) report, the worldwide prevalence of T2DM will increase from 463 million in 2019 to 700 million in 2045 [4]. The highest and lowest age-adjusted prevalence of diabetes were reported in the Middle East and North Africa (MENA) and the African (AFR) Regions. These numbers are estimated to change from 12.2%, and 4.7% in 2019 to 13.9% and 5.2% in 2045, correspondingly [4]. In 2019, the prevalence of diabetes in adult women and men was 9.0% and 9.6%, respectively. These numbers will reach 10.8% and 11.1% in 2045, respectively [4]. Tight glycemic control is believed to be the best strategy for the prevention of long-term complications. Available anti-diabetic synthetic drugs include Metformin, Sulfonylureas, insulin, Thiazolidinediones, Glucagon-like peptide-1 (GLP-1) receptor agonists, α-Glucosidase inhibitors, Dipeptidyl peptidase-4 (DPP-4) inhibitors, and Sodium-glucose-co-transporter 2 inhibitors [5, 6]. Currently, several randomized clinical trials (RCTs) are assessing the different therapeutic modalities for proper diabetes control. However, adherence to therapy is an important factor in achieving the optimal control of diabetes and preventing mortality [7].
In evidence-based medicine, RCTs are the cornerstone of determining the efficacy and safety of new treatments or medical approaches [8, 9]. In clinical research, RCTs are the best tools to answer the patient-related questions of health care providers as they produce important and substantial evidence [10, 11]. Furthermore, evidences for the approval of novel therapies by authorities are mostly based on the results of such RCTs [12]. Hence, conducting well-designed RCTs are essential for achieving the above goals.
Despite the emphasis on the importance of reporting the discontinuation of RCTs in the trials’ checklists [13, 14], some researchers fail to mention the reasons behind the patients’ withdrawal from the trial [15]. In a systematic review conducted in, out of 29 conducted RCTs, only 13 studies (44.82%) had reported the rate and reasons of withdrawals [15]. The attrition bias caused by loss of patient’s participation in clinical trials threatens the internal and external validity of results [16]. Since there is a dearth of research in the literature on the status and frequency of withdrawal from blinded RCTs, the current systematic review was conducted to assess the withdrawal reporting rate and reasons in blinded RCTs on T2DM.
Methods
Data searches
All relevant available RCTs that assessed the effectiveness of synthetic drugs, vitamins and minerals in patients with T2DM and were published before January 2017 were included. To obtain all relevant studies, we searched all web databases, including PubMed, Cochrane Library, Web of Science, and Scopus using a combination of search keywords and Medical Subject Headings (MeSH) terms “type 2 diabetes mellitus”, and “randomized controlled trials” (Table 1).
Table 1.
Search strategy utilized in studied field in PubMed
| Subject domains | Search strategy |
|---|---|
| Type 2 diabetes mellitus | ((“Diabetes Mellitus, Type 2”[Mesh] OR NIDDM [tiab] OR T2DM[tiab] OR T2D[tiab] OR ((DM[tiab] OR diabet*[tiab]) AND (“Type-II” [tiab] OR “Type II” [tiab] OR “TYPE-2” [tiab] OR “TYPE 2” [tiab] OR TYPE2[tiab] OR T2[tiab])) OR “Non Insulin” [tiab] OR “Non-Insulin” [tiab] OR Noninsulin[tiab]) |
| AND | |
| Randomized controlled trials | (“Clinical Trials as Topic”[Mesh] OR “Controlled Trials as Topic”[Mesh] OR “Clinical Trial” [Publication Type] OR RCT[TIAB] OR RANDOM*[TIAB] OR TRIAL*[TIAB]) |
In case additional information was needed, maximum three e-mails were sent to the corresponding authors of each of the retrieved paper. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart [17] was used to assess these papers. All retrieved papers were imported into the Endnote software. After the exclusion of duplicated articles, the title and abstract of the papers were assessed. Then, the full text of eligible studies was evaluated along with the hand searching of reference lists. All these steps were performed by at least two researchers independently. Any disagreement was resolved by a third researcher. In case of multiple publications from the same study, only the study with the largest sample size was included.
Study selection
All blinded RCTs that met the following criteria were included: (i) studies conducted on T2DM patients; (ii) those comparing the effects of synthetic drugs, vitamins or minerals; (iii) those comparing the pharmacological interventions with placebo.
Exclusion criteria were as following: (i) animal studies, (ii) studies on healthy population, type 1 diabetics, children or pregnant women, (iii) RCTs on herbal medicines or non-drug trials, (iv) observational studies, (v) review articles, letters to the editor, conference abstracts and theses. To narrow down the search results, English language restriction was considered.
Data extraction and quality assessment
The following data were extracted: first author, published year, study design, participants’ characteristics separately for each intervention, and control groups (total sample size before and after intervention, age and gender), dosage of the medication, study duration, frequency and reasons of withdrawal.
The Jadad score was used for the quality assessment of the included studies [13]. The quality score < 3 points was classified as low quality or high risk of bias. The recorded range was between 0 and 5.
The study was reviewed and approved by the Ethics Committee of Endocrinology and Metabolism Research Institute, Tehran University of Medical Sciences, Tehran, Iran (IR.TUMS.EMRI.REC.1396.00177).
Study analysis
Descriptive analysis of data was conducted using absolute and relative frequencies for categorical variables, and mean and standard deviations for continuous variables. In addition, univariate and multivariate binary regression was used to examine probable predictors of withdrawal. Independent variables were gender, age, national/international type, setting, design and duration of study, baseline number of participants, and classification of interventional drugs. The dependent variable was “reported withdrawal”. Data analysis was performed using SPSS version 15, and P value ≤ 0.05 was considered as statistically significance. Meta-analysis was not possible due to the heterogeneity of the included studies.
Results
According to the PRISMA flowchart (Fig. 1), 1368 articles with 640,780 participants were included in the systematic review. The majority of the studies were published in 2014 (n=155 blinded RCTs). The time trend of the publications was illustrated in Fig. 2. The R-squared value between the total number of published documents and the publication year was 0.751 (P value >0.5).
Fig. 1.
Flow diagram of the study selection process
Fig. 2.
Time trend of publications of blinded RCTs in the field
In most of the blinded RCTs (1026 studies, 75.0%), the intervention and placebo arms were compared. More than 90% of the studies were conducted on both genders. Anti-diabetic agents had the highest share of the studied medication (57.5%). Maximum duration of the studies was 552 weeks. Details on the other participants’ characteristics were presented in Table 2.
Table 2.
General characteristics of participants
| Variable | Total RCTs (n=1368) | RCTs compared intervention with placebo (n=1026) | RCTs compared intervention with an active control (n=342) |
|---|---|---|---|
| Design (n/%) | |||
| SBRCT | 69 (5.0) | 43 (4.2) | 26 (7.6) |
| DBRCT | 1139 (83.3) | 856 (83.4) | 283 (82.7) |
| TBRCT | 6 (0.4) | 5 (0.5) | 1 (0.3) |
| CSBRCT | 5 (0.4) | 5 (0.5) | – |
| CDBRCT | 149 (10.9) | 117 (11.4) | 32 (9.4) |
| Setting (n/%) | |||
| Clinic | 803 (58.7) | 601 (58.6) | 202 (59.1) |
| Hospital | 96 (7.0) | 74 (7.2) | 22 (6.4) |
| Community | 35 (2.6) | 28 (2.7) | 7 (2.0) |
| Unknown | 434 (31.7) | 323 (31.5) | 111 (32.5) |
| National studies (n/%) | 972 (71.1) | 743 (72.4) | 229 (67.0) |
| Baseline population (n) | 640,780 | 459,086 | 181,694 |
| End study population (n) | 496,986 | 368,002 | 128,984 |
| Baseline age (yr) (mean±SD) | 57.08±5.39 | 56.99±5.53 | 57.35±5.0 |
| Sex (n/%) | |||
| Female | 25 (1.8) | 21 (2.0) | 4 (1.2) |
| Male | 63 (4.6) | 56 (5.5) | 6 (1.8) |
| Both | 1280 (93.6) | 949 (92.5) | 332 (97.0) |
| Baseline female population (n/%) | 290,129 (47.45) | 213,164 (49.48) | 76,965 (26.53) |
| Intervention drugs (n/%) | |||
| Anti-diabetic | 787 (57.5) | 568 (55.4) | 219 (64.0) |
| Anti-hyperlipidemia | 69 (5.0) | 51 (5.0) | 18 (5.3) |
| Anti-hypertensive | 143 (10.5) | 96 (9.4) | 47 (13.7) |
| Anti-obesity | 30 (2.2) | 25 (2.4) | 5 (1.5) |
| Vitamins/Minerals | 91 (6.7) | 89 (8.7) | 2 (0.6) |
| Hormone/Others | 248 (18.1) | 197 (19.2) | 51 (14.9) |
| Duration of intervention (%) | |||
| <20 weeks | 655 (47.9) | 543 (52.9) | 112 (32.8) |
| ≥20 weeks | 713 (52.1) | 483 (47.1) | 230 (67.2) |
| RCTs reported withdrawal (n/%) | 1273 (93.1) | 949 (92.5) | 324 (94.7) |
| Quality assessment (n/%) | |||
| Low quality (Jadad<3) | 58 (4.2) | 45 (4.4) | 13 (3.8) |
| High quality (Jadad≥3) | 1310 (95.8) | 981 (95.6) | 329 (96.2) |
SBRCT, Single blind randomized controlled trials; DBRCT, Double blind randomized controlled trials; TBRCT, Triple blind randomized controlled trials; CSBRCT, Cross-over single blind randomized controlled trials; CDBRCT, Cross-over double blind randomized controlled trials; n, number; yr, year; SD, Standard deviation
The reasons of withdrawal were reported for 91,669 of the 143,794 participants. In detail, 59,401 (65.22%) subjects were withdrawn from blinded RCTs comparing interventions with placebo, and 32,268 (61.22%) from those comparing interventions with active control. The frequency and classification of the reasons for reported drop out from studies were presented in Table 3. The major reasons for withdrawal were “adverse effects” (n= 22,033, 24.04%), “withdraw consent” (n= 14,763, 16.10%), and “missing data” (n= 10,153, 11.08%).
Table 3.
Comparison of frequency of drop out reasons in placebo controlled with active controlled trials
| Reasons of withdrawal | Frequency of reported drop out subjects | |
|---|---|---|
| Participants in RCTs with placebo (n= 59,401) (n/%) | Participants in RCTs with an active control (n=32,268) (n/%) | |
| Adverse effects | 14,994 (25.24) | 7039 (21.81) |
| Withdraw consent | 9866 (16.61) | 4897 (15.18) |
| Loss to follow | 4215 (7.10) | 2474 (7.67) |
| Drop out | 1523 (2.56) | 647 (2.01) |
| Lack Efficiency | 2439 (4.11) | 3837 (11.89) |
| Poor compliance | 2231 (3.76) | 354 (1.10) |
| Change medication/Admin decision | 2229 (3.75) | 989 (3.06) |
| Protocol deviation | 1271 (2.14) | 1486 (4.61) |
| Personal reasons | 3595 (6.05) | 2152 (6.67) |
| Other drugs | 242 (0.41) | 139 (0.43) |
| Other study | 92 (0.15) | 2 (0.01) |
| Other illness | 159 (0.27) | 27 (0.08) |
| Travel/Migration | 93 (0.16) | 39 (0.12) |
| Abuse drug | 1 (0.002) | 6 (0.02) |
| Death | 5582 (9.40) | 358 (1.11) |
| Missing data/Others | 6169 (10.39) | 3984 (12.35) |
RCTs, Randomized controlled trials; n, number
The frequency and reasons for drop out from blinded RCTs in different countries were shown in Table 4. Overall, withdrawal was reported in more than 90% of the national and international RCTs (56.43% and 68.56%, respectively). Top three reported withdrawal reasons in national blinded RCTs were “adverse effects” (21.62%), “withdraw consent” (15.82%) and “loss to follow” (10.55%).
Table 4.
Distribution of reported frequency and reasons of drop out in blinded RCTs among countries
| Variable | National blinded RCTs (n=972) | Top five numbers of national blinded RCTs | International blinded RCTs (n=396) | ||||
|---|---|---|---|---|---|---|---|
| US (n=293) | Japan (n=90) | Italy (n=88) | UK (n=63) | Iran (n=58) | |||
| Baseline population (n) | 231,531 | 113,944 | 19,973 | 17,148 | 19,866 | 4367 | 409,249 |
| End population (n) | 174,509 | 84,014 | 17,807 | 13,094 | 14,776 | 3662 | 322,477 |
| Reported withdrawal RCTs (n/%) | 897 (92.3) | 262 (89.4) | 87 (97.8) | 85 (96.6) | 57 (90.5) | 49 (84.5) | 376 (94.9) |
| Reported withdrawn participants (n/%) | 32,176 (56.43) | 16,042 (53.60) | 1519 (70.13) | 3842 (94.77) | 2244 (44.09) | 283 (40.14) | 59,493 (68.56) |
| Reasons of withdrawal (n/%) | |||||||
| Adverse effects | 6955 (21.62) | 3660 (22.82) | 456 (30.02) | 1123 (29.23) | 285 (12.70) | 24 (8.48) | 15,078 (25.34) |
| Withdraw consent | 5090 (15.82) | 2966 (18.49) | 161 (10.60) | 926 | 260 | 38 (13.43) | 9673 (16.26) |
| Loss to follow | 2675 (8.31) | 1516 (9.45) | 53 (3.49) | (24.10) 195 | (11.59) 205 | 40 (14.13) | 4014 (6.75) |
| Drop out | 561 (1.74) | 322 (2.01) | 22 | (5.08) 41 | (9.14) 25 | 11 (3.89) | 1609 (2.70) |
| Lack Efficiency | 3395 (10.55) | 1487 (9.27) | (1.45) | (1.07) 1102 | (1.11) | 4 (1.41) | 2881 (4.85) |
| Poor compliance | 713 (2.22) | 401 (2.50) | 137 (9.02) | (28.68) 15 | 160 (7.13) | 1872 (3.15) | |
| Change medication/Admin decision | 1200 (3.73) | 745 (4.64) | 22 (1.45) | (0.39) 122 (3.18) | 28 (1.25) | 17 (6.01) | 2018 (3.39) |
| Protocol deviation | 1380 (4.29) | 912 (5.69) | 95 (6.25) | 67 (1.74) | 16 (0.71) | 7 (2.47) | 1377 (2.31) |
| Personal reasons | 946 (2.94) | 340 (2.12) | 29 (1.91) | 14 (0.36) | 92 | 0 (0) | 4801 (8.07) |
| Other drugs | 105 (0.33) | 19 (0.12) | 61 (4.02) | 4 (0.10) 0 (0) | (4.10) 26 (1.16) | 29 (10.25) 5 (1.77) | 276 (0.46) |
| Other study | 4 (0.01) | 2 (0.01) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 90 (0.15) |
| Other illness | 158 (0.49) | 22 (0.14) | 6 (0.39) | 7 (0.18) | 8 (0.36) | 29 (10.25) | 28 (0.05) |
| Travel/Migration | 87 (0.27) | 44 (0.27) | 3 (0.20) | 1 (0.03) | 1 (0.04) | 9 (3.18) | 45 (0.08) |
| Abuse drug | 1 (0.003) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 6 (0.01) |
| Death | 1919 (5.96) | 495 (3.09) | 2 (0.13) | 31 (0.81) | 26 (1.16) | 6 (2.12) | 4021 (6.76) |
| Missing data/Others | 2853 (8.87) | 1190 (7.42) | 190 (12.51) | 119 (3.10) | 70 (3.12) | 28 (9.89) | 7300 (12.27) |
RCTs, Randomized controlled trials; n, number; US, United States; UK, United Kingdom
Top five countries with the highest numbers of blinded RCTs were the United States (US), Japan, Italy, the United Kingdom (UK) and Iran, respectively. The reported withdrawal reasons differed between various countries. Adverse effects were the main withdrawal reason in the US (22.82%), Japan (30.02%), Italy (29.23%), and the UK (12.70%). The highest reason for withdrawal in Iran was “loss to follow” (14.13%). The next commonly reported reasons were “withdraw consent” and “loss to follow” in the US and the UK, “missing data” and “withdraw consent” in Japan, “lack of efficiency” and “withdraw consent” in Italy, “withdraw consent” and “personal reasons”/“other illness” in Iran. Japan reported the highest number of withdrawals from RCTs (97.8%). This figure, however, only represented 70.13% of those who had withdrawn. The highest and lowest numbers of reported withdrawal cases to total number of participants were in Italy (94.77%) and Iran (40.14%), respectively. Top three reported withdrawal reasons in international blinded RCTs were “adverse effects” (25.34%), “withdraw consent” (16.26%) and “missing data” (12.27%). Details of the above-mentioned findings were shown in Table 4.
The results of withdrawal reasons and the frequency of drop out in blinded RCTs based on the publication year were shown in Table 5. Overall, withdrawal was reported in more than 80% of RCTs. The highest and lowest percentage of reported withdrawal reasons were in RCTs published in 2011 (99.2%) and 2001 (81.4%), respectively. Top three reported withdrawal reasons among blinded RCTs published in 2011 were “adverse effects” (26.71%), “withdraw consent” (24.20%) and “loss to follow” (11.08%), correspondingly. The highest and lowest coverage of reported withdrawal reasons were found in RCTs published in 2014 (97.40%) and 2002 (27.41%), respectively. Top three reported withdrawal reasons for RCTs published in the same year were “adverse effects” (24.57%), “withdraw consent” (19.51%) and “lack of efficiency” (15.24%). Overall, “adverse effects” was the main withdrawal reason among all trials. Other common withdrawal reasons based on the publication years were “lack efficiency”, “missing data”, “poor compliance”, “loss to follow”, “personal reasons”, and “death”. Details of the above-mentioned findings were shown in Table 5.
Table 5.
Distribution of reported frequency and reasons of drop out in blinded RCTs in different published years
| Variable | Published year | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2000 (n=27) | 2001 (n=43) | 2002 (n=50) | 2003 (n=52) | 2004 (n=60) | 2005 (n=57) | 2006 (n=59) | 2007 (n=54) | 2008 (n=54) | 2009 (n=82) | 2010 (n=80) | 2011 (n=118) | 2012 (n=133) | 2013 (n=130) | 2014 (n=155) | 2015 (n=120) | 2016 (n=94) | |
| Baseline population (n) | 4598 | 11,593 | 45,196 | 16,915 | 23,592 | 29,560 | 39,637 | 26,714 | 20,482 | 53,334 | 36,438 | 48,394 | 50,236 | 58,771 | 39,766 | 90,991 | 44,563 |
| End population (n) | 3354 | 8179 | 23,343 | 14,435 | 18,391 | 24,713 | 32,614 | 21,616 | 16,036 | 32,004 | 28,865 | 40,200 | 41,222 | 48,372 | 33,247 | 76,122 | 34,273 |
| Reported withdrawal RCTs (n/%) | 26 (96.3) | 35 (81.4) | 46 (92.0) | 45 (86.5) | 53 (88.3) | 52 (91.2) | 52 (88.1) | 48 (88.9) | 50 (92.6) | 79 (96.3) | 79 (98.8) | 117 (99.2) | 121 (91.0) | 114 (87.7) | 150 (96.8) | 116 (96.7) | 90 (95.7) |
| Reported withdrawn participants (n/%) | 597 (47.99) | 3024 (88.58) | 5989 (27.41) | 1907 (76.90) | 3779 (72.66) | 4490 (92.63) | 3993 (56.86) | 3065 (60.12) | 4140 (93.12) | 10,647 (49.92) | 5415 (71.50) | 6860 (83.72) | 6394 (70.93) | 6459 (62.11) | 6350 (97.40) | 13,633 (91.69) | 4927 (47.88) |
| Reasons of withdrawal (n/%) | |||||||||||||||||
| Adverse effects | 195 (32.66) | 819 (27.08) | 1156 (19.30) | 238 (12.48) | 897 (23.74) | 1104 (24.59) | 1096 (27.45) | 599 (19.54) | 468 (11.30) | 1836 (17.24) | 1028 (18.98) | 1832 (26.71) | 1617 (25.29) | 1849 (28.63) | 1560 (24.57) | 4867 (35.70) | 872 (17.70) |
| Withdraw consent | 61 (10.22) | 294 (9.72) | 142 (2.37) | 87 (4.56) | 570 (15.08) | 878 (19.55) | 777 (19.46) | 476 (15.53) | 379 (9.15) | 1734 (16.29) | 718 (13.26) | 1660 (24.20) | 1015 (15.87) | 1322 (20.47) | 1239 (19.51) | 2681 (19.67) | 730 (14.82) |
| Loss to follow | 13 (2.18) | 41 (1.36) | 188 (3.14) | 58 (3.04) | 408 (10.80) | 206 (4.59) | 431 (10.79) | 378 (12.33) | 185 (4.47) | 813 (7.64) | 357 (6.59) | 760 (11.08) | 728 (11.39) | 506 (7.83) | 644 (10.14) | 738 (5.41) | 235 (4.77) |
| Drop out | 0 (0) | 0 (0) | 7 (0.12) | 0 (0) | 693 (18.34) | 0 (0) | 0 (0) | 13 (0.42) | 101 (2.44) | 0 (0) | 70 (1.29) | 1 (0.01) | 606 (9.48) | 219 (3.39) | 108 (1.70) | 225 (1.65) | 127 (2.58) |
| Lack Efficiency | 164 (27.47) | 195 (6.45) | 195 (3.26) | 117 (6.14) | 244(6.46) | 338 (7.53) | 269 (6.74) | 507 (16.54) | 161 (3.89) | 990 (9.30) | 618 (11.41) | 243 (3.54) | 168 (2.63) | 562 (8.70) | 968 (15.24) | 504 (3.70) | 33 (0.67) |
| Poor compliance | 36 (6.03) | 1 (0.03) | 30 (0.50) | 0 (0) | 2 (0.05) | 915 (20.38) | 94 (2.35) | 35 (1.14) | 8 (0.19) | 366 (3.44) | 44 (0.81) | 162 (2.36) | 107 (1.67) | 224 (3.47) | 211 (3.32) | 287 (2.11) | 63 (1.28) |
| Change medication/ Admin decision | 2 (0.34) | 42 (1.39) | 223 (3.72) | 27 (1.42) | 53 (1.40) | 79 (1.76) | 17 (0.43) | 72 (2.35) | 165 (3.99) | 573 (5.38) | 530 (9.79) | 197 (2.87) | 26 (0.41) | 105 (1.63) | 379 (5.97) | 551 (4.04) | 177 (3.59) |
| Protocol deviation | 71 (11.89) | 53 (1.75) | 67 (1.12) | 32 (1.68) | 200 (5.29) | 174 (3.88) | 281 (7.04) | 223 (7.28) | 64 (1.55) | 222 (2.09) | 181 (3.34) | 254 (3.70)3 | 219 (3.43)3 | 137 (2.12) | 162 (2.55) | 173 (1.27) | 244 (4.95) |
| Personal reasons | 4 (0.67) | 26 (0.86) | 2010 (33.56) | 24 (1.26) | 49 (1.30) | 71 (1.58) | 164 (4.11) | 190 (6.20) | 220 (5.31) | 1299 (12.20) | 71 (1.31) | 30 (0.44) | 5 (0.55) | 452 (7.00) | 236 (3.72) | 742 (5.44) | 124 (2.52) |
| Other drugs | 0 (0) | 2 (0.07) | 17 (0.28) | 1 (0.05) | 4 (0.11) | 0 (0) | 7 (0.18) | 52 (1.70) | 40 (0.97) | 3 (0.03) | 13 (0.24) | 117 (1.71) | 0 (0) | 1 (0.02) | 32 (0.50) | 87 (0.64) | 5 (0.10) |
| Other study | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 2 (0.07) | 0 (0) | 0 (0) | 90 (1.66) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 2 (0.04) |
| Other illness | 9 (1.51) | 4 (0.13) | 9 (0.15) | 5 (0.26) | 0 (0) | 5 (0.11) | 8 (0.20) | 0 (0) | 0 (0) | 4 (0.04) | 9 (0.17) | 50 (0.73) | 26 (0.41) | 16 (0.25) | 2 (0.03) | 24 (0.18) | 15 (0.30) |
| Travel/Migration | 0 (0) | 2 (0.07) | 0 (0) | 0 (0) | 1 (0.03) | 0 (0) | 7 (0.18) | 14 (0.46) | 1 (0.02) | 17 (0.16) | 17 (0.31) | 23 (0.34) | 0 (0) | 11 (0.17) | 33 (0.52) | 2 (0.01) | 4 (0.08) |
| Abuse drug | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 6 (0.09) | 0 (0) | 1 (0.01) | 0 (0) |
| Death | 5(0.84) | 22 (0.73) | 34 (0.57) | 157 (8.23) | 30 (0.79) | 379 (8.44) | 89 (2.23) | 904 (29.49) | 254 (6.14) | 943 (8.86) | 56 (1.03) | 210 (3.06) | 807 (12.62) | 145 (2.24) | 4 (0.06) | 707 (5.19) | 1194 (24.23) |
| Missing data/Others | 125 (20.94) | 51 (1.69) | 1958 (32.69) | 26 (1.36) | 171 (4.53) | 345 (7.68) | 419 (10.49) | 172 (5.61) | 136 (3.29) | 728 (6.84) | 379 (7.00) | 561 (8.18) | 934 (14.61) | 889 (13.76) | 773 (12.17) | 1862 (13.66) | 624 (12.66) |
n, number
Coefficients of univariate and multivariate regression models between the independent and dependent variables were shown in Table 6.
Table 6.
Coefficients of univariate and multivariate regression models between independent variables and reported drop out
| Independent variables | Total RCTs | Placebo for comparison intervention | Active control for comparison intervention | |||
|---|---|---|---|---|---|---|
| Univaraite | Multivariate | Univaraite | Multivariate | Univaraite | Multivariate | |
| OR (CI95%) | OR (CI95%) | OR (CI95%) | OR (CI95%) | OR (CI95%) | OR (CI95%) | |
| Age (year) | 1.02 (0.98–1.07) | – | 1.01 (0.97–1.06) | – | 1.12* (1.00–1.25) | 1.10 (0.98–1.23) |
| Sex | ||||||
| Female | Reference | Reference | Reference | Reference | Reference | – |
| Male | 1.38 (0.32–6.00) | 0.69 (0.14–3.29) | 1.55 (0.35–6.81) | 0.73 (0.13–3.97) | 0.00 0.00 | |
| Both | 3.05* (1.49–6.24) | 2.16 (0.99–4.75) | 3.16* (1.52–6.58) | 3.03** (1.33–6.94) | 0.00 0.00 | |
| Type of the study | ||||||
| National | Reference | Reference | Reference | Reference | Reference | – |
| International | 1.57* (0.95–2.61) | 1.01 (0.57–1.78) | 1.49* (0.85–2.64) | 1.06 (0.54–2.07) | 1.77 (0.57–5.52) | |
| Design of RCTs | ||||||
| SBRCTs | Reference | Reference | Reference | Reference | Reference | Reference |
| DBRCTs | 0.877 (0.35–2.22) | 0.71 (0.25–2.04) | 0.82 (0.25–2.67) | 0.78 (0.22–2.83) | 091 (0.19–4.50) | 0.34 (0.05–2.41) |
| TBRCTs | 0.478* (0.27–0.84) | 0.49** (0.26–0.94) | 0.57* (0.30–1.08) | 0.63 (0.30–1.31) | 0.23* (0.07–0.80) | 0.38 (0.09–1.51) |
| CSBRCTs | 1.553 (0.17–14.09) | 2.45 (0.26–23.16) | 2.00 (0.21–19.28) | 3.54 (0.34–36.46) | – | – |
| CDBRCTs | 0.00 (0.00) | 0.00 0.00 | 0.00 0.00 | 0.00 0.00 | – | – |
| Setting | ||||||
| Clinic | Reference | Reference | Reference | Reference | Reference | – |
| Non-clinic | 1.493* (0.98–2.29) | 0.75 (0.48–1.17) | 0.67* (0.42–1.07) | 0.76 (0.46–1.25) | 0.68 (0.25–1.86) | |
| Interventional drugs | ||||||
| Anti-diabetic | Reference | Reference | Reference | Reference | Reference | Reference |
| Anti-hyperlipidemia | 0.44* (0.25–0.79) | 0.48** (0.26–0.88) | 0.54* (0.28–1.04) | 0.58 (0.28–1.19) | 0.25 (0.08–0.77) | 0.35 (0.10–1 .26) |
| Anti-hypertensive | 2.28* (1.06–4.90) | 2.19 (0.96–5.02) | 3.34* (1.43–7.81) | 3.67** (1.43–9.41) | 0.44 (0.05–3.94) | 0.51 (0.05–5.02) |
| Anti-obesity | 2.13* (1.05–4.35) | 2.73** (1.28–5.83) | 2.46* (1.14–5.29) | 3.36** (1.46–7.71) | 0.00 0.00 | 0.00 0.00 |
| Vitamins/Minerals | 0.90 (0.42–1.93) | 0.81 (0.37–1.81) | 1.26 (0.53–2.98) | 1.19 (0.47–3.00) | 0.33* (0.06–1.74) | 0.32 (0.06–1.77) |
| Hormone/Others | 0.37 (0.05–2.88) | 0.47 (0.06–3.72) | 0.51 (0.06–4.00) | 0.73 (0.09–5.90) | 0.00 0.00 | 0.00 0.00 |
| Type of comparison intervention | ||||||
| Placebo | Reference | Reference | – | – | – | – |
| Active comparison | 1.61* (0.92–2.80) | 1.34 (0.75–2.40) | ||||
| Baseline number of participants (n) | 1.00 | – | 1.00* | 1.00 | 0.99* | 0.99 |
| (1.00-1.00) | (1.00–1.00) | (1.00–1.00) | (0.99–1.00) | (0.99–1.00) | ||
| Duration of intervention (w) | 1.003* | 1.003 | 1.00* | 1.00 | 1.00 | – |
| (0.99–1.01) | (0.99–1.01) | (0.99–1.01) | (0.99–1.01) | (0.99–1.01) | ||
SBRCT, Single blind randomized controlled trials; DBRCT, Double blind randomized controlled trials; TBRCT, Triple blind randomized controlled trials; CSBRCT, Cross-over single blind randomized controlled trials; CDBRCT, Cross-over double blind randomized controlled trials; OR, Odds ratio; CI, Confidence interval; W, Week
*P < 0.2 in binary regression with univariate model is considered to perform analysis by binary regression with multivariate model
* * P ≤ 0.05 in binary regression with multivariate model is considered as statistically significant
From among the included RCTs, triple blind studies as well as those on anti-hyperlipidemia and anti-obesity classes of interventional drugs were associated significantly higher number of reported withdrawals. The probability of withdrawal was highest in RCTs which a placebo group compared with anti-hypertensive or anti-obesity interventional drugs. No significant effect of independent variables was found in RCTs with active control group. Details of these results were shown in Table 6.
Discussion
In this systematic review, we identified a high proportion of reported drop out in blinded RCTs on T2DM. The “adverse effects”, “withdraw consent”, and “missing data” were the main reasons for drop out from the studies. Regardless of the national/international nature of the RCTs or their publication year, “adverse effects” was the most common reported withdrawal reason. Moreover, a significant association was observed between the reported drop out and the design of blinded RCTs as triple blind study and medication (anti-hyperlipidemia, anti-hypertensive or anti-obesity).
The investigator and patient have certain preconceived expectations that could influence the report of disadvantages and advantages of drugs in RCTs. Designing the study as double blinded is an easy solution to overcome this bias. Adding a comparison group with the aim of comparing the treatment effect in the intervention and placebo subjects is another solution in this regard [18]. These strategies could assist with a proper estimation of both favorable and unfavorable treatment responses [18]. In our study, all included RCTs were designed as blinded studies in which the intervention group was compared with a placebo or an active control group.
One of the major challenges in RCTs is recruiting and retaining the eligible subjects as dropping out negatively affects the external validity of the study [19]. This also results in high attrition rate that attenuates the generalizability of the results and increases the costs due to the need to increase the sample size to maintain the study power [20]. Attrition rate ranges between 5 and 50% with an acceptable rate of 20–30% [21], although this value could vary among different studies [21]. On the other hand, reporting withdrawal from study has a critical role in the interpretation of the results, especially in trials conducted to assess drug safety [20]. In other words, under-recording and under-reporting of withdrawal reasons and the frequency of participants discontinuing the study has a negative impact on estimating the true treatment response [22], resulting in overestimation of the intervention advantages and underestimation of its harms.
Drop out from study might be caused by reasons such as being unsatisfied with the therapeutic effects, occurrence of adverse events, or even death [11]. Regardless of the withdrawal reason, power of the study might not be sufficient to properly meet the trial objectives if a large number of participants withdraw [20]. One of the suggested solutions to avoid such bias is recruiting more than the required sample size [23].
While the attrition rate in our included blinded RCTs was acceptable (22.44%), the withdrawal reasons were reported for only 63.75% of the dropped out subjects (91669). While the coverage of reported withdrawal reasons in international blinded RCTs was higher than that of the national trials (68.56% vs. 56.43%), suggesting suboptimal reporting of withdrawal reasons in both national and international RCTs. This finding shows that reporting of withdrawal cases should be mandatory in interventional trials. A multi-center RCT designed to optimize the secondary prevention of coronary heart disease had a high retention rate of 85% [24]. The authors suggested strategies such as “applying the study design such as pilot studies”, “improving the knowledge of the setting” “shortening the gap between researcher and practice staff”, and continually “reviewing and monitoring the trial execution” to help maximize the recruitment and retention rate. Based on a systematic review of 68 studies with more than 74,000 participants, Cochrane Library suggested two effective methods to improve the recruitment process, “telling people what they are receiving in the trial” and “phoning people who do not respond to a postal invitation” [25]. Other suggested strategies were frequent assessment of completion rates, overcoming the time and administrative barriers and taking into account cultural issues in the study design [26].
Scirica et al. in their multi-center double blind RCTs conducted in 25 countries assessed vascular outcomes of Saxagliptin, a Dipeptidyl Peptidase 4 (DPP-4) inhibitor, in approximately 16,500 T2DM patients [27]. They offered several strategies to minimize missing data, including training investigators and monitors on the importance of study retention, appropriate classification of the patients, and encouraging them to be involved more actively. Other strategies proposed to reduce the withdrawal rate includes emphasizing on the benefits of participating in the study, regular communication between the research team and the participants, benefiting from skilled research staff that the participants can contact, and giving monetary incentives [28].
Recently, the quality of reporting of abstracts of RCTs published in the emergency medicine journals were systematically reviewed [29]. A small but non-significant improvement in the quality was noted after the publication of CONSORT extension of the abstracts (CONSORT-EA). RCTs from journals endorsing CONSORT, multi-center studies, those with large sample size, involving pharmacological intervention, reporting positive results for their primary outcome, and receiving funds from the industry were more prone to fit the CONSORT-EA. Moreover, the number of reported items in RCT abstracts with pharmacological interventions was significantly larger than that of those with non-pharmacological interventions [29]. The authors of RCT abstracts suggested strategies to improve the quality of reporting through collaboration between journal editors, reviewers and authors by endorsing the CONSORT statement and the CONSORT-EA.
Studies showed that the withdrawal rate is higher in trials in which new treatment is compared with placebo in contrast to trials with an active control group. This observation might be related to lack of the efficiency of placebo [30]. Our results on the withdrawal rate were consistence with the above-mentioned findings; 91,084 vs. 52,710 reported withdrawn participants in trials with placebo or active control groups, respectively. In addition, the frequency of reported withdrawal reasons was higher in trials comparing an intervention with placebo (59,401 (65.22%) of withdrawn subjects) than those comparing that with an active control group (32,268 (61.22%) of total withdrawn subjects). This difference might be due to the fact the researchers are more concerns when a placebo group is involved. Three common reasons for withdrawal from RCTs includes not meeting inclusion/ exclusion criteria, deviation from the protocol after randomization before receiving medication, and missing data [31]. Our results regarding higher proportion of “adverse effects”, “withdraw consent” and “missing data” as the main withdrawal reasons are in line with these findings [31]. The ranking of these reasons in national RCTs was different between countries. Main reasons were “adverse effects” for the US, Japan, Italy, and the UK, and “loss to follow” for Iran. This could be contributed to the fact that most of the RCTs evaluating newer hypoglycemic medications such as SGLT-2 inhibitors are conducted in the US, Japan, Italy, and the UK rather than Iran (data not shown). Perhaps both investigators and researchers in these countries are more focused on the adverse effects of the new medications. As a result, “adverse effects” was reported more frequently in RCTs from these countries then Iran. Reporting “loss to follow” in RCTs from Iran indicates a gap in the collaboration between the researchers and participants and lack of continuous monitoring process in these trials. Informing the patients on the trial details would help increase the recruitment, reduce attrition, and decrease missing data [19]. On the other word, it is essential to consider the participants’ culture in the study design to achieve success.
Assessment of the withdrawal rate in RCTs published in different years showed variations in frequency of reporting withdrawals and their reasons. The highest number of withdrawals was reported in 2011, and the lowest in 2001. This could be contributed to the increased number of RCTs on anti-diabetics over time from 21 RCTs in 2001 to 71 in 2011.
The highest number of withdrawal cases to total RCT population was found in 2014. The reasons for this finding might be partly due to the high number of published international RCTs (39.6%) and Japanese RCTs at the national level in that year (19.8%) (data not shown).
The prevalence of diabetes, its complications and comorbidities such as hyperlipidemia and obesity are increasing worldwide [4] and consequently the trials assessing these medications are increasing. Our univariate and multivariate regression analysis showed the high probability of reporting withdrawals in patients being treated with anti-hyperlipidemia or anti-obesity medications compared to anti-diabetic medications.
Our study had some strengths and limitations. The main strength of our study was the inclusion of only blinded RCTs. Due to the low probability of bias in blinded RCTs, these types of trials are more appropriate for assessing the withdrawal reasons. In addition, most of our included trials were assessed to have a high quality. The main limitation of the study was the heterogeneity in the definition of withdrawal reasons that could result in under- or over-estimation of some reasons. “Drop out” was considered as one of the withdrawal reasons in some studies, whereas in others it was used as a general term for anyone leaving study. Not being able to perform a meta-analysis was another limitation of the study.
Conclusion
In conclusion, main reported withdrawal reasons in blinded RCTs on T2DM patients are secondary to drug adverse effects. There are, however, challenges with reporting the drop out causes as transient non-threatening and even non drug-related side effects leading to consent withdrawal may have been reported as adverse events. Hence, new definitions of adverse events must be developed. Despite the emphasis of RCTs’ checklists on reporting withdrawal and its reasons, this is not considered properly in all trials. Efforts to endorse journal editors and reviewers to check the trial checklists when submitting RCTs can improve the quality of reports.
Supplementary Information
(PDF 221 kb)
Authors’ contributions
SHM did literature bibliography, and drafted some parts of the paper. OTM did literature bibliography, reviewed data, wrote draft and conceived the paper. MP, HSE, MKH, EN, ZN, KKH, MA extracted data, and drafted some parts of the paper. BL conceived, and edited the paper. All authors read and approved the final manuscript.
Funding
This study has been supported by “Diabetes Research Center, Endocrinology and Metabolism Clinical Sciences Institute, Tehran University of Medical Sciences grant numbered 1396-02-97-2161. All of authors thank the Diabetes Research Center for its financial support.
Compliance with ethical standards
Conflict of interests
All authors declare that there is no conflict of interests. In addition, all authors declare that the supporting fund has not been influenced on submitted manuscript.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Shahrzad Mohseni, Email: shmohseni58@gmail.com.
Ozra Tabatabaei-Malazy, Email: tabatabaeiml@sina.tums.ac.ir.
Maryam Peimani, Email: m_peimani@razi.tums.ac.ir.
Hanieh-Sadat Ejtahed, Email: haniejtahed@yahoo.com.
Mehrnoosh Khodaeian, Email: mehrnoosh657@gmail.com.
Elahe Nazeri, Email: Nazeri_elahe@yahoo.com.
Zahra Nouhi, Email: Zahra_nouhi@yahoo.com.
Kajal Khodamoradi, Email: k.khodamoradi@gmail.com.
Maryam Aboeerad, Email: aboeerad@yahoo.com.
Bagher Larijani, Email: emrc@tums.ac.ir.
References
- 1.WHO. Definition, diagnosis and classification of diabetes mellitus and its complications: report of a WHO consultation, Part 1: diagnosis and classification of diabetes mellitus. World Health Organization 1999, Geneva.
- 2.Rahimi R, Nikfar S, Larijani B, Abdollahi M. A review on the role of antioxidants in the management of diabetes and its complications. Biomed Pharmacother. 2005;59:365–373. doi: 10.1016/j.biopha.2005.07.002. [DOI] [PubMed] [Google Scholar]
- 3.Tabatabaei-Malazy O, Larijani B, Abdollahi M. Targeting metabolic disorders by natural products. J Diabetes Metab Disord. 2015;14:57. doi: 10.1186/s40200-015-0184-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.IDF Diabetes atlas, 9th edition 2019. Available at https://www.diabetesatlas.org. . [DOI] [PubMed]
- 5.Tabatabaei-Malazy O, Nikfar S, Larijani B, Abdollahi M. Drugs for the treatment of pediatric type 2 diabetes mellitus and related co-morbidities. Expert Opin Pharmacother. 2016;17(18):2449–2460. doi: 10.1080/14656566.2016.1258057. [DOI] [PubMed] [Google Scholar]
- 6.Tootee A, Esfahani EN, Larijani B. Diabetes management during Ramadan amid Covid-19 pandemic. Daru. 2020;28:1–4. doi: 10.1007/s40199-020-00357-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.DiMatteo MR. Variations in patients' adherence to medical recommendations: a quantitative review of 50 years of research. Med Care. 2004;42(3):200–209. doi: 10.1097/01.mlr.0000114908.90348.f9. [DOI] [PubMed] [Google Scholar]
- 8.Kabisch M, Ruckes C, Seibert-Grafe M, Blettner M. Randomized controlled trials: part 17 of a series on evaluation of scientific publications. Dtsch Arztebl Int. 2011;108(39):663–668. doi: 10.3238/arztebl.2011.0663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P, Schünemann HJ, GRADE Working Group GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924–926. doi: 10.1136/bmj.39489.470347.AD. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Manchikanti L, Hirsch JA, Smith HS. Evidence-based medicine, systematic reviews, and guidelines in interventional pain management: part 2: randomized controlled trials. Pain Physician. 2008;11(6):717–773. doi: 10.36076/ppj.2008/11/717. [DOI] [PubMed] [Google Scholar]
- 11.Stang A. Randomized controlled trials-an indispensible part of clinical research. Dtsch Arztebl Int. 2011;108(39):661–662. doi: 10.3238/arztebl.2011.0661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Harbour R, Miller J. A new system for grading recommendations in evidence based guidelines. BMJ. 2001;323(7308):334–336. doi: 10.1136/bmj.323.7308.334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials. 1996;17:1–12. doi: 10.1016/0197-2456(95)00134-4. [DOI] [PubMed] [Google Scholar]
- 14.Pandis N, Chung B, Scherer RW, Elbourne D, Altman DG. CONSORT 2010 statement: extension checklist for reporting within person randomised trials. BMJ. 2017;357:j2835. doi: 10.1136/bmj.j2835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hasanshiri F, Moosavi S, Atashzadeh-Shoorideh F. Self-care programs in diabetic patients: quality of methodological report in randomized controlled trials published in Iranian journals between 2010-2016. Iranian Journal of Endocrinology and Metabolism. 2017;18(6):446–454. [Google Scholar]
- 16.Evans SR. Fundamentals of clinical trial design. J Exp Stroke Transl Med. 2010;3(1):19–27. doi: 10.6030/1939-067X-3.1.19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535. doi: 10.1136/bmj.b2535. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Furberg BD. Furberg CD. What are the strengths ofrandomized controlled clinical trials? In: Furberg BD, Furberg CD editors. Evaluating clinical research, the second edition. Springer; 2007, pp: 11–16.
- 19.Bornhoft G, Maxion-Bergemann S, Wolf U, Kienle GS, Michalsen A, Vollmar HC, et al. Checkist for the qualitative evaluation of clinical studies with particular focus on external validity and model validity. BioMed Central Med Res Methodol. 2006;6:56. doi: 10.1186/1471-2288-6-56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nüesch E, Trelle S, Reichenbach S, Rutjes AW, Bürgi E, Scherer M, et al. The effects of excluding patients from the analysis in randomised controlled trials: meta-epidemiological study. BMJ. 2009;339:b3244. doi: 10.1136/bmj.b3244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Fisher L, Hesler D, Naranjo D, Polonsky W. AASAP: a program to increase recruitment and retention in clinical trials. Patient Educ Couns. 2012;86:372–377. doi: 10.1016/j.pec.2011.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Joseph SW. Problems in dealing with missing data and informative censoring in clinical trials. Curr Control Trials Cardiovasc Med. 2002;3(1):4. doi: 10.1186/1468-6708-3-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sakpal T. Sample size estimation in clinical trial. Perspect Clin Res. 2010;1(2):67–69. [PMC free article] [PubMed] [Google Scholar]
- 24.Leathem CS, Cupples ME, Byrne MC, O'Malley M, Houlihan A, Murphy AW, Smith SM. Identifying strategies to maximise recruitment and retention of practices and patients in a multicentre randomised controlled trial of an intervention to optimise secondary prevention for coronary heart disease in primary care. BMC Med Res Methodol. 2009;9:40. doi: 10.1186/1471-2288-9-40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Treweek S, Pitkethly M, Cook J, Fraser C, Mitchell E, Sullivan F, Jackson C, Taskila TK, Gardner H, Cochrane Methodology Review Group Strategies to improve recruitment to randomised trials. Cochrane Database Syst Rev. 2018;2:MR000013. doi: 10.1002/14651858.MR000013.pub6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.McCullagh MC, Sanon MA, Cohen MA. Strategies to enhance participant recruitment and retention in research involving a community-based population. Appl Nurs Res. 2014;27(4):249–253. doi: 10.1016/j.apnr.2014.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Scirica BM, Bhatt DL, Braunwald E, Steg PG, Davidson J, Hirshberg B, et al. The design and rationale of the saxagliptin assessment of vascular outcomes recorded in patients with diabetes mellitus-thrombolysis in myocardial infarction (SAVOR-TIMI) 53 study. Am Heart J. 2011;162(5):818–25.e6. doi: 10.1016/j.ahj.2011.08.006. [DOI] [PubMed] [Google Scholar]
- 28.Brueton VC, Tierney JF, Stenning S, Meredith S, Harding S, Nazareth I, Rait G. Strategies to improve retention in randomised trials: a Cochrane systematic review and meta-analysis. BMJ Open. 2014;4(2):e003821. doi: 10.1136/bmjopen-2013-003821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Germini F, Marcucci M, Fedele M, Galli MG, Heath T, Mbuagbaw L, et al. Quality of reporting in abstracts of RCTs published in emergency medicine journals: a systematic survey of the literature suggests we can do better. Emerg Med J. 2019. pii: emermed-2019-208629. 10.1136/emermed-2019-208629. [DOI] [PubMed]
- 30.Rabinowitz J, Levine SZ, Barkai O, Davidov O. Dropout rates in randomized clinical trials of antipsychotics: a meta-analysis comparing first- and second-generation drugs and an examination of the role of trial design features. Schizophr Bull. 2009;35(4):775–788. doi: 10.1093/schbul/sbn005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Furberg BD, Furberg CD. What happened to the study subjects who disappeared from the analysis? In: Furberg BD, Furberg CD, editors. Evaluating clinical research; the second edition. Springer; 2007, pp: 76–81.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
(PDF 221 kb)