Abstract
Background:
High loss-to-follow-up rates are a risk in even the most rigorously designed randomized controlled trials (RCTs). Consequently, predicting and preventing loss to follow-up are important methodological considerations. We hypothesized that certain baseline characteristics are associated with a greater likelihood of patients being lost to follow-up. Our primary objective was to determine which baseline characteristics are associated with loss to follow-up within 12 months after an open fracture in adult patients participating in the Fluid Lavage of Open Wounds (FLOW) trial. We also present strategies to reduce loss to follow-up in trauma trials.
Methods:
Data for this study were derived from the FLOW trial, a funded trial in which payments to clinical sites were tied to participant retention. We conducted a binary logistic regression analysis with loss to follow-up as the dependent variable to determine participant characteristics associated with a higher risk of loss to follow-up.
Results:
Complete data were available for 2,381 of 2,447 participants. One hundred and sixty-three participants (6.7%) were lost to follow-up. Participants who received treatment in the U.S. were more likely to be lost to follow-up than those who received treatment in other countries (odds ratio [OR] = 3.56, 95% confidence interval [CI]: 2.46 to 5.17, p < 0.001). Male sex (OR = 1.75, 95% CI: 1.15 to 2.67, p = 0.009), current smoking (OR = 1.82, 95% CI: 1.28 to 2.58, p = 0.001), high-risk alcohol consumption (OR = 1.88, 95% CI: 1.16 to 3.05, p = 0.010), and an age of <30 years (OR = 2.16, 95% CI: 1.19 to 3.95, p = 0.012) all significantly increased the odds of a patient being lost to follow-up. Conversely, participants who had sustained polytrauma (OR = 0.52, 95% CI: 0.37 to 0.73, p < 0.001) or had a Gustilo-Anderson type-IIIA, B, or C fracture (OR = 0.60, 95% CI: 0.38 to 0.94, p = 0.024) had lower odds of being lost to follow-up.
Conclusions:
Using a number of strategies, we were able to reduce the loss-to-follow-up rate to <7%. Males, current smokers, young participants, participants who consumed a high-risk amount of alcohol, and participants in the U.S. were more likely to be lost to follow-up even after these strategies had been employed; therefore, additional strategies should be developed to target these high-risk participants.
Clinical Relevance:
This study highlights an important need to develop additional strategies to minimize loss to follow-up, including targeted participant-retention strategies. Male sex, an age of <30 years, current smoking, high-risk alcohol consumption, and treatment in a developed country with a predominantly privately funded health-care system increased the likelihood of participants being lost to follow-up. Therefore, strategies should be targeted to these participants. Use of the planning and prevention strategies outlined in the current study can minimize loss to follow-up in orthopaedic trials.
High loss-to-follow-up rates are a risk in even the most rigorously designed randomized controlled trials (RCTs). When a large number of participants are lost to follow-up the study power is reduced and substantial bias can occur when the participants who are lost to follow-up differ from those who remain in the study in terms of outcome variables1.
In the past, researchers have typically considered a 20% loss to follow-up to be the threshold for an increased risk of bias due to loss to follow-up2, but this threshold was not evidence-based. More recently, Zelle et al.3 determined that, in more than one-quarter of 50 simulated orthopaedic trials, the results changed from significant to nonsignificant when the authors simulated a loss of follow-up of 20%. This indicates that the 20% threshold may not be strict enough to mitigate the risk of bias from loss to follow-up. The authors recommended that researchers aim to achieve a follow-up rate of >80%3. Strategies that have previously been used to reduce loss to follow-up from clinical trials include regularly searching for alternative contact information, searching death registries and other national registries, setting follow-up-rate goals, calling at different hours of the day, leaving voice messages, and repeated contact4-7.
While RCTs in all surgical specialties are susceptible to loss to follow-up, orthopaedic trials have a particularly poor record in terms of both reporting loss-to-follow-up rates8,9 and achieving adequate follow-up rates10. Consequently, minimizing loss to follow-up has become an important methodological consideration in trauma trials11. However, even when participant-retention strategies are employed, it is likely that some participants will still be lost to follow-up. Young age, male sex, smoking, substance abuse, an isolated orthopaedic injury, and a lack of insurance were linked to loss to follow-up of trauma patients from an orthopaedic practice10. However, that study was limited by the fact that it was a retrospective review conducted at only 1 center in the U.S. and included a small number of participants (n = 309). A systematic review of loss-to-follow-up rates in orthopaedic trials demonstrated that studies conducted in the U.S. had higher rates and that trauma trials did not have significantly worse rates than those in other orthopaedic subspecialties9. The pooled loss-to-follow-up rate in this systematic review was 10.4%.
We hypothesize that not all trauma trial participants have the same risk of being lost to follow-up and that certain baseline characteristics are associated with a greater likelihood of being lost to follow-up. The objectives of this study were (1) to determine which baseline characteristics were associated with loss to follow-up within 12 months after an open fracture in an adult patient participating in the Fluid Lavage of Open Wounds (FLOW) trial, in which general strategies for reducing loss to follow-up were implemented; and (2) to describe the strategies that we used in the FLOW trial to achieve a lower-than-average loss-to-follow-up rate. Identifying which trauma trial participants are at an increased risk of being lost to follow-up would allow future research to focus on developing and evaluating retention strategies aimed specifically at participants identified as having a high risk of being lost to follow-up. Our experiences with reducing loss to follow-up in the FLOW trial may be beneficial to other investigators wishing to reduce their loss-to-follow-up rates as well.
Materials and Methods
Data for this study were derived from the FLOW trial, a large multicenter initiative that was conducted at 41 clinical sites in Canada, the U.S., Norway, Australia, and India (ClinicalTrials.gov: NCT00788398). The FLOW trial was a 2 × 3 factorial RCT evaluating the impact of irrigation solution (soap versus saline solution) and irrigation pressure (high versus low versus very low) on the risk of a reoperation to promote wound or bone healing or to treat infection in adult patients with an open fracture of an extremity. The FLOW trial recruited 2,447 participants, who were scheduled to be followed for 12 months with follow-up appointments at 1, 2, and 6 weeks and 3, 6, 9, and 12 months postsurgery. A detailed description of the trial methodology12 and results13 have been published elsewhere.
Definition of Loss to Follow-up in the FLOW Trial
In the FLOW trial, participants were considered lost to follow-up when they had been withdrawn from the study before their final 12-month follow-up visit because they could not be located by study personnel. Participants who were withdrawn early for other reasons such as death or because they withdrew their consent were not classified as lost to follow-up because these reasons for withdrawal were deemed to be beyond the study team’s control. We determined that a participant’s withdrawal of consent was beyond our control because we had already implemented strategies to minimize factors that might cause a participant to withdraw consent; these strategies included allowing participants who felt overburdened by the questionnaires to fill out fewer questionnaires and allowing telephone visits for those who did not want to return to the clinic. However, it should be noted that, since good clinical practice guidelines14 and the Canadian Tri-Council Policy Statement15 dictate that research participants must be free to withdraw consent at any time, strategies aimed at reducing early withdrawal must be carefully considered to ensure that they are not coercive.
Strategies to Minimize Loss to Follow-up in the FLOW Trial
We implemented an intensive strategy to minimize loss to follow-up in the FLOW trial, as shown in Table I. Some main features of the strategy included designing the follow-up schedule to align with standard-of-care visits, selection of a primary outcome that could be obtained without an in-person visit, selection of clinical sites with experienced research coordinators, efforts to be flexible with scheduling and not overburden participants, multiple attempts to contact participants, and routine verification of contact information. When a participant had missed visits and was at risk of being lost to follow-up, we implemented specific procedures, including searching medical records for new admissions, searching obituaries and death registries, attempting to contact participants in the evening and on weekends, and reducing the questionnaire burden. The methods center also placed a query on the data of participants who were presumed lost to follow-up as a reminder to continue to attempt to locate them. Participants were not compensated for attending follow-up visits, and none of their medical costs were covered by the study. Clinical sites were paid an average of $500 CAD outside of the U.S. and $750 USD within the U.S. for complete follow-up of participants. Payments were issued in installments and tied to data submission to incentivize sites to maximize follow-up.
TABLE I.
Strategies to Prevent Loss to Follow-up in the FLOW Trial
| Strategy | |
| Design phase |
|
| At baseline | |
| Methods-center level |
|
| Site level |
|
| During trial | |
| Methods-center level |
|
| Site level |
|
| For patients lost, or at high risk of being lost, to follow-up | |
| Methods-center level |
|
| Site level |
|
Statistical Methods
Demographics
The characteristics of the participants who were lost to follow-up are reported separately from those of participants who were not. Means and standard deviations (SDs) are presented for continuous variables, and frequencies and percentages are presented for categorical variables. All analyses were conducted using SPSS version-23.0 software (IBM).
Primary Analysis
We conducted a binary logistic regression analysis with loss to follow-up as the dependent variable16. In the binary logic regression model, we included independent variables (i.e., age category, sex, smoking status, intravenous [IV] drug abuse, high-risk alcohol consumption, work-related injury, employment status, location of fracture, polytrauma, Gustilo-Anderson fracture type, and health-care system) selected on the basis of clinical relevance and previous research10,17-19. We defined “high-risk alcohol consumption” as >7 drinks per week for women and >14 drinks per week for men according to National Institutes of Health (NIH) guidelines20. Authors of previous studies have observed greater challenges with follow-up of trauma patients who are younger males18,19. Because this greater challenge has also been reported for uninsured participants10, we hypothesized that participants who receive treatment in health-care systems that are not publicly funded (e.g., those in India and the U.S.) would be more likely to be lost to follow-up. We tested for interactions between key variables. We present results as odds ratios (ORs) with 95% confidence intervals (CIs) and associated p values. We assessed multicollinearity using the variance inflation factor for all variables included in the model. We decided a priori that variables with a variance inflation factor of ≥10 would be considered multicollinear and removed. Finally, we used leverage and deviance statistics to identify influential outliers. As there were few cases of missing data (<3%), we did not complete any statistical imputations.
Sensitivity Analysis
It was decided post hoc to conduct sensitivity analyses in which a binary logistic regression was conducted without influential outliers.
Results
Participant Characteristics
This study included 2,447 participants, and complete data were available for 2,381 (97.3%) of them. One hundred and sixty-three participants (6.7%) were lost to follow-up. Table II presents demographic and injury characteristics of the participants who were lost to follow-up, those who were not, and overall. Most participants were male (69%), and the mean age in the cohort was 45 years. Seven hundred and seventy-seven (32%) were current smokers, 31 (1.3%) reported IV drug abuse, and 260 (11%) reported high-risk alcohol consumption. Three hundred and forty-nine (14%) had a work-related injury. Most fractures (68%) were of the lower extremity, and most participants (59%) had several injuries.
TABLE II.
Participant Characteristics by Loss-to-Follow-up Status
| No. (%)* |
|||
| Characteristic | Not Lost to Follow-up (N = 2,284) | Lost to Follow-up (N = 163) | Total (N = 2,447) |
| Mean age ± SD (yr) | 45.1 ± 17.8 | 38.6 ± 15.9 | 44.7 ± 17.8 |
| Sex | |||
| Male | 1,552 (68.0) | 128 (78.5) | 1680 (68.7) |
| Female | 713 (31.2) | 35 (21.5) | 748 (30.6) |
| Missing | 19 (0.8) | 0 (0.0) | 19 (0.8) |
| Current smoker | |||
| Yes | 696 (30.5) | 81 (49.7) | 777 (31.8) |
| No | 1,549 (67.8) | 79 (48.5) | 1,628 (66.5) |
| Missing | 39 (1.7) | 3 (1.8) | 42 (1.7) |
| IV drug abuse | |||
| Yes | 30 (1.3) | 1 (0.6) | 31 (1.3) |
| No | 2,214 (96.9) | 157 (96.3) | 2,371 (96.9) |
| Missing | 40 (1.8) | 5 (3.1) | 45 (1.8) |
| High-risk alcohol consumption | |||
| Yes | 234 (10.2) | 26 (16.0) | 260 (10.6) |
| No | 1,993 (87.3) | 133 (81.6) | 2,126 (86.9) |
| Missing | 57 (2.5) | 4 (2.5) | 61 (2.5) |
| Work-related injury | |||
| Yes | 331 (14.5) | 18 (11.0) | 349 (14.3) |
| No | 1,925 (84.3) | 145 (89.0) | 2,070 (84.6) |
| Missing | 28 (1.2) | 0 (0.0) | 28 (1.1) |
| Employment status | |||
| Unemployed/disabled | 277 (12.1) | 36 (22.1) | 313 (12.8) |
| Retired/student/employed/homemaker | 1,970 (86.3) | 125 (76.7) | 2,095 (85.6) |
| Missing | 37 (1.6) | 2 (1.2) | 39 (1.6) |
| Health-care system | |||
| Industrialized/publicly funded | 1,369 (59.9) | 58 (35.6) | 1,427 (58.3) |
| Industrialized/privately funded | 687 (30.1) | 93 (57.1) | 780 (31.9) |
| Industrializing/privately funded | 228 (10.0) | 12 (7.4) | 240 (9.8) |
| Missing | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Location of fracture | |||
| Upper extremity | 705 (30.9) | 53 (32.5) | 758 (31.0) |
| Lower extremity | 1,560 (68.3) | 110 (67.5) | 1,670 (68.2) |
| Missing | 19 (0.8) | 0 (0.0) | 19 (0.8) |
| Polytrauma | |||
| Yes | 1,380 (60.4) | 72 (44.2) | 1,452 (59.3) |
| No | 886 (38.8) | 91 (55.8) | 977 (39.9) |
| Missing | 18 (0.8) | 0 (0.0) | 18 (0.7) |
| Gustilo-Anderson fracture type | |||
| I | 585 (25.6) | 54 (33.1) | 639 (26.1) |
| II | 842 (36.9) | 57 (35.0) | 899 (36.7) |
| IIIA/B/C | 829 (36.3) | 52 (31.9) | 881 (36.0) |
| Missing | 28 (1.2) | 0 (0.0) | 28 (1.1) |
For all variables except age.
Loss to Follow-up
The final model included the following variables: age category, sex, current smoking, high-risk alcohol consumption, work-related injury, polytrauma, Gustilo-Anderson fracture type, IV drug abuse, employment status, fracture location, and type of health-care system in which the treatment was received (Table III). No variables were removed from the model as a result of multicollinearity as the variance inflation factor was low (<1.2) for all variables. There were 100 participants with moderate deviance values, and 8 with high leverage values.
TABLE III.
Binary Logistic Regression Model for Loss to Follow-up*
| Covariate | OR | 95% CI | P Value |
| Age | |||
| <30 yr | 2.16 | 1.19-3.95 | 0.012 |
| 30-60 yr | 1.31 | 0.72-2.37 | 0.371 |
| >60 yr | — | — | — |
| Male | 1.75 | 1.15-2.67 | 0.009 |
| Current smoker | 1.82 | 1.28-2.58 | 0.001 |
| High-risk alcohol consumption | 1.88 | 1.16-3.05 | 0.010 |
| Polytrauma | 0.52 | 0.37-0.73 | <0.001 |
| Gustilo-Anderson fracture type | |||
| I | — | — | — |
| II | 0.66 | 0.43-1.02 | 0.059 |
| IIIA/B/C | 0.60 | 0.38-0.94 | 0.024 |
| Health-care system | |||
| Industrialized/publicly funded | — | — | — |
| Industrialized/privately funded | 3.56 | 2.46-5.17 | <0.001 |
| Industrializing/privately funded | 0.92 | 0.49-1.91 | 0.966 |
Adjusted for work-related injury, IV drug abuse, employment status, and fracture location (upper versus lower extremity).
The odds of being lost to follow-up were 3.56 times greater (95% CI: 2.46 to 5.17, p < 0.001) for participants who received treatment in the U.S. (i.e., in a predominantly privately funded health-care system) compared with those who received treatment in Canada, Europe, or Australia (which have publicly funded health-care systems). Receiving treatment in India (a developing country with a predominantly privately funded health-care system) was not significantly associated with loss to follow-up (OR = 0.92, 95% CI: 0.49 to 1.91, p = 0.966). Male sex (OR = 1.75, 95% CI: 1.15 to 2.67, p = 0.009), current smoking (OR = 1.82, 95% CI: 1.28 to 2.58, p = 0.001), high-risk alcohol consumption (OR = 1.88, 95% CI: 1.16 to 3.05, p = 0.010), and an age of <30 years (OR = 2.16, 95% CI: 1.19 to 3.95, p = 0.012) all significantly increased the odds of the participant being lost to follow-up. Conversely, participants who had sustained polytrauma (OR = 0.52, 95% CI: 0.37 to 0.73, p < 0.001) or a Gustilo-Anderson type-IIIA, B, or C fracture (OR = 0.60, 95% CI: 0.38 to 0.94, p = 0.024) had lower odds of being lost to follow-up. None of the interaction terms were significant (all p > 0.13).
Sensitivity Analyses
The 108 participants identified as influential outliers were removed for binary logistic regression sensitivity analysis, which revealed findings similar to those of the primary analysis. The 1 exception was work-related injury, which the sensitivity analysis showed to be significantly associated with a decreased odds of being lost to follow-up (OR = 0.10, 95% CI: 0.01 to 0.81, p = 0.031).
Discussion
The results of this study show that the odds of being lost to follow-up are greater for participants who are male, currently smoke, are younger than the age of 30 years, consume a high-risk amount of alcohol, or received treatment in the U.S. Conversely, the odds of being lost to follow-up are lower for participants who sustained polytrauma or have a Gustilo-Anderson type-IIIA, B, or C fracture.
These findings are congruent with our past clinical experiences conducting trauma trials and also with previous research. For example, previous studies have shown that women tend to visit their doctors more than men21, which may suggest that they would be more likely to attend medical follow-up appointments. Smoking or consuming more alcoholic drinks may be indicative of poor health behavior attitudes, which may make participants less likely to return for medical follow-up appointments18. Participants who are older are less transient, and it may be easier for study personnel to locate them for follow-up visits. Participants with more severe injuries (a higher fracture type or polytrauma) may be more likely to return for study follow-up visits because of longer recovery times, need for ongoing care, and ongoing limitations to their functional abilities. Finally, participants who receive care in privately funded health-care systems may face barriers to attending scheduled follow-up visits if they are uninsured19. In agreement with our findings, a systematic review of loss-to-follow-up rates in orthopaedic trials also showed higher rates in trials completed in the U.S. than in those carried out in other countries9. We did not compensate participants for their time, travel costs, or medical care in this trial. Other studies have shown that providing monetary incentives to participants could improve follow-up rates, so this may be a strategy to consider in the future.
Work-related injury was not significantly associated with the odds of being lost to follow-up in the primary or secondary analysis. This finding is surprising given that treatment for work-related injuries is often covered through workplace insurance. Additionally, participants who are receiving Workers’ Compensation have been shown to have worse outcomes after orthopaedic trauma17,22. However, when the sensitivity analyses were conducted with the influential outliers removed, having a work-related injury was significantly associated with lower odds of loss to follow-up. Future research should include an investigation of the association between work-related injuries and loss to follow-up given that our findings were inconclusive.
There have been previous studies on strategies to prevent loss to follow-up in clinical trials, several of which were conducted in an orthopaedic setting. A recent Cochrane review of 38 trials showed that 6 broad types of strategies have been used to improve retention in clinical trials, including incentives, communication strategies, new questionnaire strategies, behavioral/motivational strategies, case management, and methodological strategies23. The only intervention that consistently improved retention was adding a monetary incentive23. We were fortunate to be able conduct the FLOW trial at sites with experienced research coordinators and to have funding to pay those sites for recruiting and following participants, thereby motivating them, through financial incentives, to follow participants for the full study period. However, a number of the strategies (e.g., all of the design-phase elements) that we implemented in the FLOW trial to reduce loss to follow-up (Table I) are not associated with any additional costs and some may help to improve efficiency and thus reduce costs. For example, conducting a telephone visit often takes less time than completing an in-clinic visit. Even strategies that have associated costs such as repeatedly contacting participants and calling them on weekends may be more cost-effective in the long term because preventing loss to follow-up helps to maintain study power, thereby mitigating the need to increase the study sample size. Future studies could focus on the cost-effectiveness of these strategies. It is also interesting to note that participants in the U.S. were more likely to be lost to follow-up even though U.S. sites were paid more per participant than those in other countries.
It is possible that retention strategies would be more effective if they were targeted toward participants at high risk of being lost to follow-up. The results of our study will allow clinical trial personnel at methods centers and at recruiting sites to be more aware of which participants are at risk of being lost to follow-up. We are unaware of any trials evaluating targeted strategies to enhance retention, which is an area of future study.
This study has several limitations. Given that we obtained our data from a completed RCT, we could include only previously collected variables in the model. For example, we hypothesized that socioeconomic status was an important predictor of loss to follow-up, but such data were not available. It is also likely that some of the variables collected in this study (e.g., substance abuse, employment, and smoking status) change over time and consequently so does the participant’s risk of being lost to follow-up. While this is an important area of future research, assessing risk over time is beyond the scope of this study. It is also possible that, because these data came from a single trial, studies of different orthopaedic populations may reveal different risk factors for loss to follow-up. However, the risk factors that we found in the current study are in line with hypothesized risk factors and previous research from other specialties10,17-19, suggesting that our study is generalizable.
The strengths of this study include its large sample size, minimal amount of missing data (<3%), and diverse population. While we used data from a specific orthopaedic trial, the baseline variables included in the model represent data that are routinely collected in most trauma trials, enhancing the generalizability to other trauma trials.
The FLOW trial was a funded trial in which rigorous participant-retention strategies were employed in the design phase, at the start of the trial, and throughout the trial. However, despite these efforts, nearly 7% of the participants were lost to follow-up. This highlights an important need for additional strategies to minimize loss to follow-up, including targeted participant-retention strategies. Male sex, an age younger than 30 years, current smoking, high-risk alcohol consumption, and treatment in a developed country with a predominantly privately funded health-care system increased the likelihood of participants being lost to follow-up. Therefore, strategies should be targeted to these participants. Use of the planning and prevention strategies outlined in the current study can minimize loss to follow-up in orthopaedic trials.
Appendix
A list of the FLOW investigators is available with the online version of this article as a data supplement at jbjs.org (http://links.lww.com/JBJS/D156).
Footnotes
Investigation performed at McMaster University, Hamilton, Ontario, Canada
A commentary by Douglas R. Dirschl, MD, is linked to the online version of this article at jbjs.org.
Disclosure: The FLOW (Fluid Lavage of Open Wounds) trial was supported by research grants from the Canadian Institutes of Health Research (MCT-93173), U.S. Army Institute of Surgical Research Orthopedic Trauma Research Program (W81XWH-08-1-0473), U.S. Army Institute of Surgical Research Peer Reviewed Orthopedic Research Program (W81XWH-12-1-0530), and Association Internationale pour l’Ostéosynthèse Dynamique. Stryker donated Surgilav irrigators for the trial for clinical sites in Asia. Zimmer provided the Pulsavac irrigator at a reduced cost to selected clinical sites in North America. Triad Medical donated castile soap; castile soap from Aplicare was purchased at full cost. Ms. Madden is funded by a Canadian Institutes of Health Research (CIHR) Doctoral award (GSD-134929). Dr. Bhandari is funded, in part, by a Canada Research Chair. No donor or funder had a role in the design or conduct of the study, the collection or analyses of the data, or the preparation of the manuscript. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work (http://links.lww.com/JBJS/D155).
References
- 1.Murray DW, Britton AR, Bulstrode CJ. Loss to follow-up matters. J Bone Joint Surg Br. 1997. March;79(2):254-7. [DOI] [PubMed] [Google Scholar]
- 2.Sackett DL, Straus SE, Richardson WS, Rosenberg W, Haynes RB. Evidence-based medicine: how to practice and teach EBM. New York: Churchill Livingstone; 1997. [Google Scholar]
- 3.Zelle BA, Bhandari M, Sanchez AI, Probst C, Pape HC. Loss of follow-up in orthopaedic trauma: is 80% follow-up still acceptable? J Orthop Trauma. 2013. March;27(3):177-81. [DOI] [PubMed] [Google Scholar]
- 4.Ezell JM, Saltzgaber J, Peterson E, Joseph CLM. Reconnecting with urban youth enrolled in a randomized controlled trial and overdue for a 12-month follow-up survey. Clin Trials. 2013. October;10(5):775-82. Epub 2013 Aug 27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Froelicher ES, Miller NH, Buzaitis A, Pfenninger P, Misuraco A, Jordan S, Ginter S, Robinson E, Sherwood J, Wadley V. The Enhancing Recovery in Coronary Heart Disease Trial (ENRICHD): strategies and techniques for enhancing retention of patients with acute myocardial infarction and depression or social isolation. J Cardiopulm Rehabil. 2003. Jul-Aug;23(4):269-80. [DOI] [PubMed] [Google Scholar]
- 6.David MC, Alati R, Ware RS, Kinner SA. Attrition in a longitudinal study with hard-to-reach participants was reduced by ongoing contact. J Clin Epidemiol. 2013. May;66(5):575-81. Epub 2013 Feb 4. [DOI] [PubMed] [Google Scholar]
- 7.Morrison TC, Wahlgren DR, Hovell MF, Zakarian J, Burkham-Kreitner S, Hofstetter CR, Slymen DJ, Keating K, Russos S, Jones JA. Tracking and follow-up of 16,915 adolescents: minimizing attrition bias. Control Clin Trials. 1997. October;18(5):383-96. [DOI] [PubMed] [Google Scholar]
- 8.Bhandari M, Richards RR, Sprague S, Schemitsch EH. The quality of reporting of randomized trials in the Journal of Bone and Joint Surgery from 1988 through 2000. J Bone Joint Surg Am. 2002. March;84(3):388-96. [DOI] [PubMed] [Google Scholar]
- 9.Somerson JS, Bartush KC, Shroff JB, Bhandari M, Zelle BA. Loss to follow-up in orthopaedic clinical trials: a systematic review. Int Orthop. 2016. November;40(11):2213-9. Epub 2016 May 3. [DOI] [PubMed] [Google Scholar]
- 10.Zelle BA, Buttacavoli FA, Shroff JB, Stirton JB. Loss of follow-up in orthopaedic trauma: who is getting lost to follow-up? J Orthop Trauma. 2015. November;29(11):510-5. [DOI] [PubMed] [Google Scholar]
- 11.Sprague S, Leece P, Bhandari M, Tornetta P 3rd, Schemitsch E, Swiontkowski MF. S.P.R.I.N.T. Investigators. Limiting loss to follow-up in a multicenter randomized trial in orthopedic surgery. Control Clin Trials. 2003. December;24(6):719-25. [DOI] [PubMed] [Google Scholar]
- 12.FLOW Investigators. Fluid lavage of open wounds (FLOW): design and rationale for a large, multicenter collaborative 2 x 3 factorial trial of irrigating pressures and solutions in patients with open fractures. BMC Musculoskelet Disord. 2010. May 6;11:85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Bhandari M, Jeray KJ, Petrisor BA, Devereaux PJ, Heels-Ansdell D, Schemitsch EH, Anglen J, Della Rocca GJ, Jones C, Kreder H, Liew S, McKay P, Papp S, Sancheti P, Sprague S, Stone TB, Sun X, Tanner SL, Tornetta P 3rd, Tufescu T, Walter S, Guyatt GH. FLOW Investigators. A trial of wound irrigation in the initial management of open fracture wounds. N Engl J Med. 2015. December 31;373(27):2629-41. Epub 2015 Oct 8. [DOI] [PubMed] [Google Scholar]
- 14.International Conference on Harmonisation. ICH harmonised tripartite guideline: guideline for good clinical practice. 1996. June http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Efficacy/E6/E6_R1_Guideline.pdf. Accessed 2017 Feb 16.
- 15.Canadian Institutes of Health Research, Natural Science and Engineering Research Council of Canada, Social Sciences and Humanitities Research Council of Canada. Tri-council policy statement: ethical conduct for research involving humans. 2010. December http://www.pre.ethics.gc.ca/pdf/eng/tcps2/TCPS_2_FINAL_Web.pdf. Accessed 2017 Feb 16. [Google Scholar]
- 16.Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, Vickers AJ, Ransohoff DF, Collins GS. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015. January 6;162(1):W1-73. [DOI] [PubMed] [Google Scholar]
- 17.de Moraes VY, Godin K, Tamaoki MJ, Faloppa F, Bhandari M, Belloti JC. Workers’ Compensation status: does it affect orthopaedic surgery outcomes? A meta-analysis. PLoS One. 2012;7(12):e50251 Epub 2012 Dec 5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Neuner B, Fleming M, Born R, Weiss-Gerlach E, Neumann T, Rettig J, Lau A, Schoenfeld H, Kallischnigg G, Spies C. Predictors of loss to follow-up in young patients with minor trauma after screening and written intervention for alcohol in an urban emergency department. J Stud Alcohol Drugs. 2007. January;68(1):133-40. [DOI] [PubMed] [Google Scholar]
- 19.Whiting PS, Greenberg SE, Thakore RV, Alamanda VK, Ehrenfeld JM, Obremskey WT, Jahangir A, Sethi MK. What factors influence follow-up in orthopedic trauma surgery? Arch Orthop Trauma Surg. 2015. March;135(3):321-7. Epub 2015 Jan 24. [DOI] [PubMed] [Google Scholar]
- 20.National Institute for Alcohol Abuse and Alcoholism. Drinking levels defined. https://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/moderate-binge-drinking. Accessed 2016 Oct 27.
- 21.Bertakis KD, Azari R, Helms LJ, Callahan EJ, Robbins JA. Gender differences in the utilization of health care services. J Fam Pract. 2000. February;49(2):147-52. [PubMed] [Google Scholar]
- 22.de Moraes VY, Godin K, Dos Santos JB, Faloppa F, Bhandari M, Belloti JC. Influence of compensation status on time off work after carpal tunnel release and rotator cuff surgery: a meta-analysis. Patient Saf Surg. 2013. January 2;7(1):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Brueton VC, Tierney J, Stenning S, Harding S, Meredith S, Nazareth I, Rait G. Strategies to improve retention in randomised trials. Cochrane Database Syst Rev. 2013. December 3;12:MR000032. [DOI] [PMC free article] [PubMed] [Google Scholar]