By Lisa Yelland, Paulo Ferreira, Lucas Ferreira, Katrina Scurrah, Monica Rankin, Jane Denton, Merryl Harvey, Katherine J Lee, Evie Kendal, Jeffrey M Craig
- Conducting clinical trials with twins has substantial benefits over trials with singletons.
- Trials that include twins pose practical, methodological and ethical difficulties, but these can be overcome with careful design and implementation.
- We need more clinical trials involving only twins.
This article discusses the design and implementation of randomised controlled trials which involve only twins, of which very few have been conducted. We outline specific designs that can be used, and the strengths, limitations and ethical issues arise in such trials, and provide examples. We conclude that clinical trials including only twins should be more widely.
Suggested citation: Yelland, L., P. Ferreira, L. Ferreira, K. Scurrah, M. Rankin, J. Denton, M. Harvey, K.J. Lee, E. Kendal, and J. M. Craig, Conducting clinical trials in twin populations: working with twins and their families. Conversations in Twins Research, Twins Research Australia, Melbourne, 2018, https://www.twins.org.au/2019/03/01/conducting-clinical-trials-in-twin-populations
Twins and medical research
Twins have participated in medical research for over a century (1-3). Statistical models used to analyse twin data include “classical” analysis of variance components, and “twin differences” models that compare twins within monozygotic (MZ – identical) or dizygotic (DZ – fraternal) pairs. The former models allow the variance of health and other outcomes to be partitioned into components due to genetics, shared (familial) environment and non-shared (twin-specific) environment. The latter models control for maternal and paternal factors, gestational age, geographical location and season of birth, age, ethnicity, sex (in MZ and same-sex DZ twins), and genetics (100% in MZ twins and 50% in DZ twins).
Studies of twins can also address questions of gene-environment interactions (4), cause vs association (5), and are used in more recent branches of medical science such as epigenetics (6), stem cells (7), and microbiome research (8).
Randomised controlled trials (RCTs) are the gold standard for assessing intervention effectiveness, because random assignment to treatment groups means differences in outcomes can be attributed to the intervention. However, only six of 186,027 US-registered clinical trials used only twins as participants (9), suggesting that there is scope for increasing the use this potentially valuable resource.
When twins participate in an RCT, they can be randomised to the same treatment group (a form of cluster randomisation), or one twin can be assigned to the intervention group and the other to the control group (a “co-twin intervention” design), or randomised independently in which case they may can be randomised to the same or different treatment groups. Cluster randomisation is useful when the risk of treatment group contamination is high (10), such as when participants know which treatment they are receiving (unblinded trials) and may discuss it with their twin. The co-twin intervention design is the most powerful, requiring the smallest sample size (11), so is useful for expensive trials or when few twins meet inclusion criteria. This design allows treatments to be compared within twin pairs, which enables near-perfect (for MZ twins) or partial (for DZ twins) control for genetics and shared environment, removing noise from the treatment effect estimate. Individual randomisation is useful when the risk of treatment group contamination and trial cost must be balanced. Under this design, approximately half the twin pairs will be assigned to the same treatment group, avoiding contamination, and the remainder to different groups; gains in power (compared with assigning all twin pairs to the same group) will be achieved by comparing treatments within twin pairs assigned to differing interventions.
The preference of twins and their caregivers is another important factor to consider in selecting a method of randomisation for twins in a trial. Parents of multiples and adult multiples have a strong preference for cluster randomisation (12), suggesting recruitment to such trials will be easier.
In some trials, only cluster randomisation is possible due to the nature of the intervention. For example, when interventions are delivered to the family unit, both twins will necessarily receive the same intervention. This can occur if, for example, the mother is treated during the pregnancy or a parental education program is delivered during childhood.
Power and sample size estimation
When designing an RCT, sample size calculations are performed to estimate the number of participants required to detect a particular treatment effect at specified levels of power and significance (13). Because responses from twins are more alike than responses from two randomly selected individuals, sample size calculations and analytical methods (see below) for trials involving twins need to recognise the paired nature of the data (14). If both twins will receive the same treatment and the correlation between outcomes of twins is not accounted for in sample size calculations, the trial may be underpowered to detect an effect. Similarly, if each twin will receive a different treatment and the correlation between outcomes of twins is ignored when calculating the sample size, power may be unnecessarily high.
When analysing an RCT, the main goal is to estimate the effect of the intervention on the trial outcomes. Most standard methods of analysis that can be used to compare outcomes between treatment groups are based on the assumption that the outcomes of all trial participants are independent. This assumption is violated in trials conducted in twin populations, since twins will have similar outcomes due to shared genetic and environmental factors. Thus, it is important to choose analytical methods that allow for the similarity between outcomes of twins; failure to do so can lead to false conclusions about intervention effectiveness (15).
Utility of MZ, same-sex DZ and opposite-sex DZ twins
lthough clinical trials tend to focus on comparing average outcomes between treatment groups, the variation in outcomes can also provide useful information. Conducting RCTs in twin populations permits the variation to be studied. MZ pairs provide the most information about random variation in outcomes. If the twins are assigned to the same treatment group, they can provide information about variation in outcomes between treatment occasions in individuals sharing genes and environment. If assigned to different treatments, twins can provide information about variation in outcomes between near-identical individuals when treated and untreated.
Correlation between outcomes between MZ and DZ twin pairs can also provide potentially useful information. Higher observed correlations for MZ pairs than DZ pairs would be consistent with genetic effects on response to treatment. If equal environments are assumed this would suggest a gene–environment interaction, or if there are unequal environments this would suggest a treatment–environment interaction. Again in this context, variance components models could be fitted to assess the relative contributions of shared genetic effects, shared environmental effects and unshared effects to variation in outcome.
Inclusion of opposite-sex pairs in an RCT allows assessment of whether outcome varies by sex. If the correlation in outcomes between same-sex DZ twins is higher than between opposite-sex DZ twins, an interaction between sex and treatment may be present.
Examples of clinical trials involving only twins
In one of the earliest examples of a co-twin intervention design, a group of twenty MZ twin girls ages 9-13 were randomised to a 6-month daily calcium and vitamin D supplement vs placebo (16). Across all pairs, there was evidence of improvements (p = 0.001) in bone density and bone strength in twins in the intervention arm. This strengthened the existing evidence for the importance of micronutrients on bone growth in puberty.
A recent co-twin RCT randomised 44 MZ and DZ twin pairs between an eight-week low- or high-fat diet (17). The study investigated the genetic and environmental factors influencing sensitivity to fatty acid taste, which influences dietary fat intake. Comparing data from MZ and DZ twins showed that environment, in the form of intake of dietary fat, is the sole influencer of fatty acid sensitivity; no evidence for a genetic influence was found. This study has important implications for the fight against obesity.
A current trial aims to assess whether response to exercise has a genetic component, and whether this depends on the type of exercise (18). Healthy MZ and DZ twin pairs aged 15–30 years are being randomised to resistance or endurance training, and train together for three months. After a three-month washout period, they perform the other training regime for three months. Preliminary results suggest that even genetically-identical individuals respond differently to endurance and resistance training, and that those who are relatively insensitive to one form will respond to the other. This has important implications for personalising exercise interventions.
An ongoing trial using the co-twin intervention design is testing a sleep intervention to improve symptoms in patients with low back pain (19). MZ and DZ twin pairs are being randomised to an online sleep quality intervention based on a cognitive behavioural approach involving sleep hygiene, or a control group, involving provision of online educational material. The main outcomes are pain self-efficacy, function, and disability associated with pain.
Advantages and disadvantages of clinical trials involving only twins
Advantages of trials involving only twins include:
- it is faster and cheaper to recruit and characterise a pair of twins than two unrelated individuals;
- twins are generally representative of the broader population (20);
- compliance may be better in twins, since they may encourage each other to follow treatment schedules;
- with the co-twin intervention design, twins provide greater statistical power than unrelated individuals is required (11); and
- twins provide information about gene–environment interactions in addition to treatment group effects.
Potential disadvantages of trials involving only twins include:
- it may be that only one of a pair of twins meet inclusion criteria and hence the twin pair is ineligible to participate;
- the withdrawal of one twin could lead to the withdrawal of the other, especially when the decision is made by a caregiver rather than the participant;
- treatment group contamination may be more likely when twins are assigned to different groups using either the co-twin control design or individual randomisation;
- randomising twins to the same treatment group provides less power than the same number of unrelated individuals (11); and
- for twin only designs, recruitment may be difficult due to the low prevalence of twins in most disease groups. This may be overcome by engaging with twin registries such as Twins Research Australia or multiple birth associations.
Ethical considerations for conducting clinical trials with twins can be broadly separated into those involving recruitment procedures and general research methods. Principles of informed consent and justice must guide the recruitment of all participants into clinical studies, but twins pose unique issues.
Researchers must ensure that potential study participants are not unduly influenced in their decision to enrol. In a twin-only study, one twin wishing to enrol might exert pressure on the other, since the refusal of one necessarily excludes the other – likewise, if one twin wishes to withdraw while the other wishes to remain. Researchers must also consider that parents and guardians may feel pressure to enrol their twins because the benefits of involving twins in research in general are well known. For studies that recruit twins in utero , any vulnerability of the pregnant woman to coercion should be addressed.
It is generally considered unethical for one population to bear the burdens of research that primarily benefits another, such as using low-income countries to test drugs ultimately sold in high-income countries (21). Hence, studies intended to benefit the general population should not unfairly burden twin populations. Twins could be financially or otherwise compensated for assuming the risks of research on behalf of others, and this might reduce the perception that their special genetic similarity is being exploited.
As discussed, the inclusion of twins above, twin is a study requires a decision about how they are randomised. There is an ethical obligation to choose a method that is acceptable to participants, and the psychological implications of assigning twins to different treatment groups should not be overlooked (12). Another ethical consideration is the need to deal with any differences in outcomes between twins; if one treatment is associated with higher morbidity, parents who allowed just their children to be randomised such that one received this treatment may feel distressed. This could have a long term impact on the parents and children so should be considered from the outset.
Factors influencing twins’ involvement in trials
Many factors affect twins’ involvement in clinical trials – such as ethical issues, as noted above, and logistics, particularly when infants or children are involved. If the twins were born prematurely, as is common, the family may decline to subject them to further treatments. Conversely, some families will welcome involvement in a trial that might prevent another family experiencing the trauma of a premature birth or its sequelae. When twins receive different treatments and one appears to be benefiting or suffering, the parents may be more likely to withdraw from the trial.
Having two children involved in a trial can be daunting, especially for families struggling to cope. Participation may involve remembering which child requires what treatment, documenting evidence for two children, and transporting children to research facilities or clinics on a regular basis. Many families need to be convinced of the potential benefits of the treatment – and that twins are an essential part of the process – to participate in a trial. They need to feel appreciated, that they are meaningfully involved in the trial/treatment, and that they will receive useful feedback.
Conducting clinical trials in twin populations either in trials restricted to twin or those also involving singletons raise special issues for the design, analysis and ethics of the trial but can add important insights into the action of specific interventions for the benefit of all society. We would suggest that the inclusion of twins be considered in RCTs where possible.
About the authors
Dr Lisa Yelland is an NHMRC Postdoctoral Research Fellow in the Healthy Mothers, Babies and Children Theme at the South Australian Health and Medical Research Institute and in the School of Public Health at The University of Adelaide.
Associate Professor Paulo Ferreira is a researcher and lecturer in the Discipline of Physiotherapy, Faculty of Health Sciences, the University of Sydney.
Lucas Ferreira is a PhD student and Research Assistant in the Melbourne School of Population and Global Health, University of Melbourne.
Dr Katrina Scurrah is a researcher and lecturer in Biostatistics and Genetic Epidemiology at the Melbourne School of Population and Global. Health, University of Melbourne.
Monica Rankin is the Chairperson of the International Council of Multiple Birth Organisations.
Jane Denton is the Director of the Multiple Births Foundation, Queen Charlotte’s & Chelsea Hospital, London and co lead of the Elizabeth Bryan Multiple Births Centre.
Professor Merryl Harvey works in the School of Nursing and Midwifery at Birmingham City University and co lead of the Elizabeth Bryan Multiple Births Centre
Associate Professor Katherine Lee is the Associate Director: Biostatistics of the Melbourne Children’s Trials Centre at the Murdoch Children’s Research Institute.
Evie Kendal is a lecturer in Bioethics and Health Humanities at the Deakin University School of Medicine.
Associate Professor Jeffrey Craig works in the Centre for Molecular and Medical Research at the Deakin University School of Medicine.
1. Hrubec Z, Robinette CD. The study of human twins in medical research. N Engl J Med. 1984;310(7):435-41.
2. Martin N, Boomsma D, Machin G. A twin-pronged attack on complex traits. Nat Genet. 1997;17(4):387-92.
3. Boomsma D, Busjahn A, Peltonen L. Classical twin studies and beyond. Nat Rev Genet. 2002;3(11):872-82.
4. Buil A, Brown AA, Lappalainen T, Vinuela A, Davies MN, Zheng HF, et al. Gene-gene and gene-environment interactions detected by transcriptome sequence analysis in twins. Nat Genet. 2015;47(1):88-91.
5. Sjölander A, Frisell T, Öberg S. Causal Interpretation of Between-Within Models for Twin Research. Epidemiologic Methods2012. p. 217.
6. Bell JT, Saffery R. The value of twins in epigenetic epidemiology. International journal of epidemiology. 2012;41(1):140-50.
7. Hibaoui Y, Grad I, Letourneau A, Sailani MR, Dahoun S, Santoni FA, et al. Modelling and rescuing neurodevelopmental defect of Down syndrome using induced pluripotent stem cells from monozygotic twins discordant for trisomy 21. EMBO Mol Med. 2014;6(2):259-77.
8. Smith MI, Yatsunenko T, Manary MJ, Trehan I, Mkakosya R, Cheng J, et al. Gut microbiomes of Malawian twin pairs discordant for kwashiorkor. Science. 2013;339(6119):548-54.
9. Sumathipala A, Yelland L, Green D, Shepherd T, Jayaweera K, Ferreira P, et al. Twins as Participants in Randomized Controlled Trials: A Review of Published Literature. Twin research and human genetics : the official journal of the International Society for Twin Studies. 2018;21(1):51-6.
10. Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. London: Arnold; 2000.
11. Yelland LN, Sullivan TR, Price DJ, Lee KJ. Sample size calculations for randomised trials including both independent and paired data. Stat Med. 2017;36(8):1227-39.
12. Bernardo J, Nowacki A, Martin R, Fanaroff JM, Hibbs AM. Multiples and parents of multiples prefer same arm randomization of siblings in neonatal trials. J Perinatol. 2015;35(3):208-13.
13. Kirby A, Gebski V, Keech AC. Determining the sample size in a clinical trial. Med J Aust. 2002;177(5):256-7.
14. Carlin JB, Doyle LW. Sample size. J Paediatr Child Health. 2002;38(3):300-4.
15. Yelland LN, Salter AB, Ryan P, Makrides M. Analysis of binary outcomes from randomised trials including multiple births: when should clustering be taken into account? Paediatr Perinat Epidemiol. 2011;25(3):283-97.
16. Greene DA, Naughton GA. Calcium and vitamin-D supplementation on bone structural properties in peripubertal female identical twins: a randomised controlled trial. Osteoporos Int. 2011;22(2):489-98.
17. Costanzo A, Nowson C, Orellana L, Bolhuis D, Duesing K, Keast R. Effect of dietary fat intake and genetics on fat taste sensitivity: a co-twin randomized controlled trial. Am J Clin Nutr. 2018;107(5):683-94.
18. Marsh CE, Thomas HJ, Scurrah KJ, Naylor LH, Green DJ. Cardiovascular responses to distinct exercise modalities in monozygotic twins European College of Sport Science 2018; Dublin2018.
19. Pinheiro MB, Ho KK, Ferreira ML, Refshauge KM, Grunstein R, Hopper JL, et al. Efficacy of a Sleep Quality Intervention in People With Low Back Pain: Protocol for a Feasibility Randomized Co-Twin Controlled Trial. Twin Research & Human Genetics. 2016;19(5):492-501.
20. Andrew T, Hart DJ, Snieder H, de Lange M, Spector TD, MacGregor AJ. Are twins and singletons comparable? A study of disease-related and lifestyle characteristics in adult women. Twin Res. 2001;4(6):464-77.
21. Zumla A, Costello A. Ethics of healthcare research in developing countries. J R Soc Med. 2002;95(6):275-6.