You are here

  • Home
  • Archive
  • Volume 78, Issue 10
  • Emulated trial investigating effects of multiple treatments: estimating combined effects of mucoactive nebulisers in cystic fibrosis using registry data

Article Text

Article menu
Cystic fibrosis
Original research
Emulated trial investigating effects of multiple treatments: estimating combined effects of mucoactive nebulisers in cystic fibrosis using registry data
  1. Emily Granger1,
  2. Gwyneth Davies2,3,
  3. Ruth H Keogh1
  1. 1 Department of Medical Statistics, London School of Hygiene & Tropical Medicine, London, UK
  2. 2 UCL Great Ormond Street Institute of Child Health, UCL, London, UK
  3. 3 Respiratory Medicine, Great Ormond Street Hospital For Children NHS Foundation Trust, London, UK
  1. Correspondence to Dr Emily Granger, Department of Medical Statistics, London School of Hygiene & Tropical Medicine, London, WC1E 7HT, UK; emily.granger{at}lshtm.ac.uk

Abstract

Introduction People with cystic fibrosis (CF) are often on multiple long-term treatments, including mucoactive nebulisers. In the UK, the most common mucoactive nebuliser is dornase alfa (DNase). A common therapeutic approach for people already on DNase is to add hypertonic saline (HS). The effects of DNase and HS used alone have been studied in randomised trials, but their effects in combination have not. This study investigates whether, for people already prescribed DNase, adding HS has additional benefit for lung function or use of intravenous antibiotics.

Methods Using UK CF Registry data from 2007 to 2018, we emulated a target trial. We included people aged 6 years and over who were prescribed DNase without HS for 2 years. We investigated the effects of combinations of DNase and HS over 5 years of follow-up. Inverse-probability-of-treatment weighting was used to control confounding. The period predated triple combination CF transmembrane conductance regulator modulators in routine care.

Results 4498 individuals were included. At baseline, average age and forced expiratory volume in 1 s (FEV1%) predicted were 21.1 years and 69.7 respectively. During first year of follow-up, 3799 individuals were prescribed DNase alone; 426 added HS; 57 switched to HS alone and 216 were prescribed neither. We found no evidence that adding HS improved FEV1% at 1–5 years, or use of intravenous antibiotics at 1–4 years, compared with DNase alone.

Conclusion For individuals with CF prescribed DNase, we found no evidence that adding HS had an effect on FEV1% or prescription of intravenous antibiotics. Our study illustrates the emulated target trial approach using CF Registry data.

  • Cystic Fibrosis
  • Nebuliser therapy

Data availability statement

Data may be obtained from a third party and are not publicly available. To access the data, an application must be made to the UK CF Registry Research Committee. https://www.cysticfibrosis.org.uk/the-work-we-do/uk-cf-registry/apply-for-data-from-the-uk-cf-registry.

https://creativecommons.org/licenses/by/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

https://doi.org/10.1136/thorax-2023-220031

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • People with cystic fibrosis (CF) are often prescribed multiple long-term treatments, including mucoactive nebulisers such as DNase and hypertonic saline. The effects of DNase and hypertonic saline used alone on health outcomes have been studied in randomised trials, but the combined effects of both treatments have not been studied.

    WHAT THIS STUDY ADDS

    • We emulated a hypothetical target trial using UK CF Registry data to compare multiple treatment strategies involving DNase and hypertonic saline. The primary interest was in estimating the effect of adding hypertonic saline to DNase after 2 years, compared with continued use of DNase alone, on long-term clinical health outcomes. We found no evidence that adding hypertonic saline has an effect on forced expiratory volume in 1 s or prescription of intravenous antibiotics.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • We provide an example of target trial emulation to answer a clinical question for which there is no evidence from randomised controlled trials. This approach may be used to address clinical questions that are unlikely to be answered in randomised trials in the field of CF and more widely.

    Background

    Randomised controlled trials (RCTs) are the gold standard for evaluating the effects of treatments. However, there are many more questions relating to treatments than can reasonably be evaluated within RCTs. When there are multiple potential treatment strategies, for example, it can be challenging to recruit enough individuals in each treatment arm for sufficient power. An alternative is to use observational data to study treatment effects, and it is increasingly recognised that emulating a target trial when using observational data helps to clarify the research question and avoid common biases.1–5 In this study, we use registry data to emulate a target trial designed to compare multiple treatment strategies on health outcomes in people with cystic fibrosis (CF).

    CF is an inherited condition caused by mutations in the CF transmembrane conductance regulator (CFTR) gene.6 This mutation leads to an abnormal movement of chloride and sodium across the airway epithelium. Thickened secretions and a cycle of inflammation and infection in the lungs result in significant morbidity.6 Many people with CF take mucoactive nebulisers which aim to alleviate the downstream consequences of CFTR dysfunction. The most commonly prescribed mucoactive nebuliser is dornase alfa (DNase),7 which is recommended as the first choice of mucoactive agent if there is clinical evidence of lung disease.8 DNase works by reducing viscosity in the lungs, which helps to clear lung secretions.9 In contrast, mucoactive nebulisers such as hypertonic saline (HS), help to clear lung secretions by rehydrating the airway surface liquid.10 Within clinical practice, depending on clinical status and clinician or patient preference, individuals already prescribed DNase may subsequently follow several different treatment strategies. These may include continuing on DNase alone, adding HS to their daily regimen, or, more rarely, a decision may be made to stop DNase and switch to HS.

    RCTs have investigated the effects of DNase or HS alone in CF. A recent Cochrane systematic review9 found evidence to show that DNase alone may improve lung function and decrease pulmonary exacerbations in people with CF. Another Cochrane systematic review10 found evidence that HS alone can reduce pulmonary exacerbations and improve lung function, although the evidence for lung function was deemed low quality. Although DNase and HS are often prescribed in combination in clinical practice, their effects in combination have not been studied. Answering questions about the effects of different combinations of these two treatments using an RCT would be challenging, particularly if we are interested in studying long-term effects.

    We use UK CF Registry data to emulate a target trial designed to compare multiple treatment strategies involving DNase and HS and assess their long-term effects on health outcomes in people with CF. The focus is on patients already established on DNase as defined by CF Registry documentation of current prescription, and the primary aim is to investigate the causal effect of adding and continuing HS versus continuing to use DNase only on two health outcomes: lung function (measured using forced expiratory volume in 1 s, FEV1%) and prescription of intravenous antibiotics. We compare outcomes measured at 1, 2, 3, 4 and 5 years of follow-up under each treatment strategy, where each treatment strategy is to be sustained up to the outcome measurement time. Previous studies have investigated the long-term effects of DNase using registry data and the findings suggest that DNase may improve the rate of decline in lung function11 12 and that it may be more beneficial for people with lower lung function.13 However, HS has not been previously studied using registry data either alone or in combination with DNase.

    Our study is undertaken using data which predated the widespread introduction of triple combination CFTR modulator therapies into routine clinical care. The question we address is relevant to the CF population overall, although results from the premodulator period may not translate to a modulator-treated population. This will be able to be investigated using the same framework once more years of data are available. Furthermore, access to modulators is not universal globally, and in those countries with access, patients ineligible or unable to tolerate them represent an important minority in whom this question is particularly relevant.

    There are several examples of using the target trial framework to compare treatment strategies across disease areas,14–16 including CF,17 however, they have tended to focus on treatment strategies involving a single treatment. In this study, we use this approach to compare treatment strategies that involve a combination of two treatments.

    Methods

    Study design and data source

    Our study was designed to emulate a hypothetical RCT (ie, the ‘target trial’). The target trial framework involves describing the protocol for a randomised trial we would like to conduct if it were feasible, and then emulating that trial using the available observational data.1 A key element of the emulation of the trial involves controlling for confounding of the treatment-outcome association, as treatments are not randomly assigned in the observational data.

    The key components of the protocol for our target trial are outlined in table 1. We emulate the target trial using data from the UK Cystic Fibrosis Registry.18 This is a national database managed by the Cystic Fibrosis Trust. Data are collected on time-invariant variables, such as sex, ethnicity and genotype, and variables that change over time. Longitudinal data are collected at approximately annual visits on over 250 variables covering several domains, including hospital admissions, pulmonary function, chronic medications, health complications, and these data have been recorded in a centralised database since 2007. For this study, data were available from 2007 to 2018. Further details on the registry are provided elsewhere.18

    View this table:
    Table 1

    Components of the target trial we aim to emulate in this study

    Table 1 specifies the components of the target trial and summarises how the registry data are used to emulate the target trial. Individuals meeting the inclusion criteria are those aged 6 years or older, who have been prescribed DNase but not HS for two consecutive years between 1 January 2007 and 31 December 2017. The baseline year is defined as the first year the inclusion criteria were met, so the earliest possible baseline year was 2008 and the latest was 2017. It is possible for individuals to meet the inclusion criteria in more than 1 year. When this was the case, we defined the baseline year to be the year from 2008 to 2017 which was most recent, but which allowed for the maximum possible follow-up time up to 5 years. For example, a person meeting the inclusion criteria in 2012 and 2013 has 6 and 5 years of potential follow-up, respectively, so we would choose 2013 as their baseline year. For a person meeting the inclusion criteria in 2013 and 2014, we would choose 2013. Exclusion criteria are listed in table 1.

    Our treatments of interest are DNase and HS. At each annual review visit it is recorded whether individuals have been prescribed these treatments over the past year. Start and stop dates for long-term treatments are also available and these were used to impute any missing data on the treatments from the annual review visits. The two outcomes of interest are lung function and receipt of intravenous antibiotics. Both outcomes are recorded annually. Lung function is measured at annual visits by spirometry, with %-predicted values for FEV1% calculated using the Global Lung Initiative equations.19 At each annual visit the number of days on intravenous antibiotics (at home or in hospital) is recorded. In this study, use of intravenous antibiotics was treated as a binary variable indicating whether the individual has any recorded days on intravenous antibiotics since the last annual visit.

    The emulation of the target trial uses existing observational data, reflecting data collected during routine clinical care and without any randomisation to treatment strategy. It is important that the analysis accounts for the lack of randomisation, as far as possible. In the observational data, the association between treatment and the outcome is suspected to be confounded by several factors. We used directed acyclic graphs to show the assumed relationships between the relevant variables in our data and to inform which variables should be considered as confounders (see online supplemental figures S.1 and S.2) in our analyses. The following variables were considered as potential confounders: sex, CF genotype, ethnicity, age, respiratory infections, intravenous hospital admissions, body mass index z-score, pancreatic insufficiency, use of CFTR modulators, past FEV1%, past intravenous antibiotic use and past rate of decline in FEV1%. Except for sex, genotype and ethnicity, which we take to be fixed over time, these covariates are recorded annually. Further details on how these covariates were defined are provided in online supplemental file.

    Supplemental material

    Treatment effect estimands and statistical analysis

    The target trial specifies four longitudinal treatment strategies involving our two treatments of interest (table 1). Each treatment strategy involves beginning a particular combination of DNase and HS and sustaining that combination throughout follow-up. Our primary interest was in comparing the strategies of continuing DNase and adding HS (DN&HS) and continuing DNase only (DN). For the FEV1% outcome, the main estimands of interest were the mean differences in FEV1% at times 1–5 years had all individuals been following treatment strategy DN&HS, vs had all individuals been following treatment strategy DN. For intravenous antibiotics, the main estimands of interest were the corresponding ORs at times 1–4 years. Note that FEV1% is measured on the day of the annual review, whereas intravenous antibiotic use over the past year is recorded. To estimate 1-year, 2-year, 3-year, 4-year and 5- year treatment effects on FEV1%, we use FEV1% recorded at the first, second, third, fourth and fifth follow-up visit, respectively. To estimate the 1-year, 2-year, 3-year and 4- year treatment effects on intravenous antibiotic use, we use information recorded at the second, third, fourth and fifth follow-up visit. Comparisons between other treatment combinations were of secondary interest. We also compared the strategy of switching to and then continuing HS (HS) and the strategy of dropping DNase (Nil) with the strategy of continuing DNase only (DN).

    The treatment effect estimands specified above were estimated using marginal structural models (MSM) estimated using inverse-probability-of-treatment weighting (IPTW).20 An MSM specifies how the outcome at a given time depends on treatment history up to that time, and in our case also on time and baseline covariates. The MSM cannot be fitted directly due to time-dependent confounding. IPTW involves estimating the probability of individuals receiving the treatment they received at each time point conditional on their treatment and covariate history up to that time. Multinomial regression was used to estimate the probability of having a given treatment combination (DN, DN&HS, HS, Nil). The IPTW at a given time is inverse of the product of the probabilities up to that time. Stabilised weights were used to avoid extreme weights.21

    We assumed that consecutive visits were approximately 1 year apart. Some individuals had less than five follow-up visits after their baseline visit, due to the administrative end of follow-up, death or organ transplant. These individuals were censored at the time of death, transplant or end of follow-up. We did not use data from visits at which an individual had missing data in the outcome, or from visits at which individuals were using certain treatments (mannitol, lumacaftor/ivacaftor or tezacaftor/ivacaftor). Inverse-probability-of-censoring weights were used to address censoring, and inverse-probability-of-observation weights were used to handle exclusions at a given visit due to missing outcome data or use of certain treatments. Each individual had a combined weight at each time point which combines the IPTW with these other weights. Further details on the weights are provided in online supplemental file (see ‘Inverse-probability-of-treatment weighted estimation of marginal structural models’). As well as missing outcome data, there were missing data in some of the confounding variables. Full details on the amount of missing data and our approach to handling it are given in the online supplemental file (see ‘Missing data’). The MSM was fitted using the combined weights. For the FEV1% outcome the MSM is a linear regression model and for the intravenous antibiotic use outcome the MSM is a logistic regression model. The MSMs were fitted for all follow-up times combined with follow-up visit included as a covariate. The analysis for each outcome was conducted with and without interaction terms between treatment use and FEV1% at baseline in the MSM. The 95% CIs for the interaction terms were estimated to assess the evidence for treatment effect heterogeneity and we present treatment effects in individuals with low, moderate or high FEV1% at baseline by setting FEV1% to 40, 75 and 100, respectively. Full specification of the MSMs is provided in online supplemental file (see ‘Inverse-probability-of-treatment weighted estimation of marginal structural models’). All SEs and 95% CIs were estimated using the non-parametric bootstrap approach.

    Results

    Study population and descriptive statistics

    We identified 5836 individuals in the UK CF Registry who had been documented as having been prescribed DNase and not prescribed HS for at least two consecutive years between 2007 and 2017. Of these, 4498 individuals met our other inclusion and exclusion criteria for the emulated trial. Figure 1 describes the study sample derivation.

    Figure 1
    Figure 1

    Flowchart of participant selection into the study sample. CF, cystic fibrosis; DNase, dornase alfa; DN&HS, DNase and adding hypertonic saline.

    Table 2 summarises the characteristics, measured at baseline, of the study population, by the treatment combination they were observed to be using in the first year of follow-up. During the first year, 3799 individuals were prescribed DNase alone and 426 were prescribed both DNase and HS. Far fewer people were prescribed HS alone, or neither treatment (57 and 216, respectively). On average, individuals prescribed both treatments were the youngest (mean age: 18.7 years) and individuals taking neither treatment were the oldest (mean age: 24.2 years). Individuals prescribed HS alone had the lowest lung function (mean FEV1%: 66.2), whereas those prescribed DNase alone had the highest (mean FEV1% 70.1). The proportion of people who were recorded as taking no intravenous antibiotics in the year prior to baseline was highest for people prescribed neither treatment (46.3%) and lowest for individuals prescribed DNase and HS (35.4%). Individuals were observed to switch between treatment combination during the follow-up. Online supplemental figure S.3 shows the number of individuals prescribed each treatment combination by year, and online supplemental figure S.4 describes the flow of individuals between different treatment combinations by year. Of the 2521 individuals who were prescribed DNase alone in the first year and had 5 years of follow-up, 1615 (64.1%) remained on DNase alone for 5 years. Of the 260 individuals prescribed DNase and HS in the first year and who had 5 years of follow-up, 185 (71.2%) remained on DNase and HS for 5 years.

    View this table:
    Table 2

    Summary of characteristics at baseline overall and by treatment combination observed in the first year of follow-up

    Details on the amount of missing data by follow-up year, and on the number of people who were censored each year (due to loss to follow-up, death or transplant), or temporarily excluded due to missing outcome data, are provided in online supplemental tables S.1 and S.2. Outcome trajectories by follow-up year and the distribution of weights used in the analysis are also provided in online supplemental figures S.5–S.7.

    Estimates of the effects of treatment combinations

    Figure 2 shows the expected mean differences in FEV1% (at times 1–5 years) and the odds of intravenous antibiotics versus no intravenous antibiotics (at times 1–4 years) between the two treatment strategies DN&HS vs DN (ie, the effect of adding HS on FEV1% and odds of intravenous antibiotics within an annual review year). Corresponding tabulated values are provided in online supplemental tables S.3 and S.4. For FEV1%, the mean differences are close to 0 at times 1–5, and all corresponding 95% CI contain 0. Similarly for intravenous antibiotics, the ORs are close to 1 at times 1–4, and all corresponding 95% CI contain 1. In other words, we found no evidence that adding HS would result in a different mean FEV1% or different odds of intravenous antibiotics among individuals who are already established on DNase, compared with continuing to use DNase only. For both outcomes, CIs, particularly for the later time points, were wide reflecting the uncertainty in our estimates.

    Figure 2
    Figure 2

    Estimated effects of adding hypertonic saline to DNase compared with continuing DNase alone on FEV1% and prescription of intravenous antibiotics. Mean differences are presented for FEV1% and ORs are presented for intravenous antibiotics. DNase, dornase alfa; FEV1%, forced expiratory volume in 1 s.

    Although not our focus, we also considered the effect of switching from DNase to HS (HS vs DN) and the effect of dropping DNase (Nil vs DNase). Figures for these additional comparisons are provided in the online supplemental figures S.8 and S.9.

    Treatment effects by baseline FEV1%

    We found no evidence of treatment effect heterogeneity by FEV1% at baseline. Figure 3 shows the expected mean differences in FEV1% at times 1–5 between DN&HS and DN, by FEV1% at baseline (low, moderate, high). Figure 4 shows the corresponding ORs at times 1–4.

    Figure 3
    Figure 3

    Estimated effects of adding hypertonic saline to DNase compared with continuing DNase alone on FEV1%. Estimated effects are mean differences and are presented for people with high (100) moderate (75) or low (40) FEV1% at baseline. DNase, dornase alfa; FEV1%, forced expiratory volume in 1 s.

    Figure 4
    Figure 4

    Estimated effects of adding hypertonic saline to DNase compared with continuing DNase alone on prescription of intravenous antibiotics. Estimated effects are ORs and are presented for people with high (100) moderate (75) or low (40) FEV1% at baseline. The upper limit of the 95% CI for the effect estimate at year 4 in the low FEV1% group is 4.55. DNase, dornase alfa; FEV1%, forced expiratory volume in 1 s.

    For FEV1%, the estimated mean differences increase as baseline FEV1% decreases, suggesting that adding HS is more beneficial for individuals with lower FEV1%. However, the corresponding 95% CI all contain 0. Similarly, the results for the intravenous antibiotics outcome provide no evidence of an effect of adding HS to DNase at any level of FEV1%.

    Discussion

    We used UK CF Registry data to emulate a hypothetical RCT designed to investigate the effects of multiple treatment strategies for mucoactive nebulised treatments on lung function and prescription of intravenous antibiotics in people with CF. Our primary interest was to investigate whether, for individuals already treated with DNase, adding HS has any additional benefit for these clinical health outcomes. We found no evidence of an effect (harmful or beneficial) of adding HS to DNase, either in change of FEV1% or prescription of intravenous antibiotics. We found some suggestion that adding HS may benefit lung function for people with lower initial FEV1%, although the results were not statistically significant. This is in line with results of a previous study based on UK CF Registry data, which suggested that the use of DNase alone could be more beneficial for FEV1% for people with lower initial FEV1%.13 People on both DNase and HS in the first year of follow-up tended to have lower lung function and more intravenous days, reflecting clinical practice where HS may be added to DNase when there has been clinical deterioration. This was addressed in the analysis by using IPTW to account for potential confounders.

    Despite DNase and HS being commonly prescribed together in clinical practice, there have been no RCTs investigating the effects of these treatments used in combination. Hence, we demonstrate the application of target trial emulation to address a clinical question in CF for which there is no RCT evidence. The target trial framework applies the study design principles of RCTs to observational studies which helps to minimise biases that can arise due to study design and analysis choices.1–5 Target trial emulation has been used successfully in other disease areas to replicate the results from existing RCTs, for example, cardiovascular disease22 23 and diabetes.24 The UK CF Registry is cited as an exemplar patient registry in the NICE real-world evidence framework,25 holds pharmacovigilance credentials and hosts post-authorisation phase IV pharmacovigilance studies.26 It is the largest national CF registry outside of the USA and captures data on almost all the UK CF population.18 These data, coupled with target trial methodology, provide an opportunity for researchers in CF to address important questions for the CF community. We did not perform sample size calculations and there is some debate as to whether such calculations are needed in observational studies such as this.27–29 We used all the available data in the UK CF Registry, giving the biggest possible sample size for the study.

    Although our primary interest was to investigate the effect of adding HS when established on DNase, we also investigated the effect of dropping DNase after 2 years and switching to HS after 2 years. Results suggested poorer outcomes in terms of FEV1% and intravenous antibiotics when dropping DNase, although results were largely non-significant. We found no evidence of an effect (in either direction) of switching to HS on intravenous days, although some evidence that switching to HS after 2 years can improve FEV1%. This result is not clinically plausible and may be impacted by unmeasured confounding, which we discuss further below. Switching to HS is a rare decision in clinical practice. The number of people prescribed HS alone reflects this (table 2).

    There are several limitations to this study. A key limitation in analyses that use observational data to study treatment effects is the possibility of bias due to uncontrolled confounding. Our analyses crucially assume that we have captured all the reasons for prescribing different treatment combinations that are associated with the outcome. While our analyses have controlled for several factors considered as potential confounders, including indicators of disease severity, it is important to note the possibility of residual confounding due to factors we did not control for. There could be, for example, biological or socioeconomic factors30 that influence both the treatment strategy and the outcome but are not collected by the registry. A further limitation is that our analyses rely on accurate treatment data being entered into the registry by clinical teams. Recording of long-term treatments within the CF Registry captures whether the treatment has been prescribed over the past year, but there is no information on adherence to treatment or dosing regimen. It is, therefore, possible that some individuals did not take or were poorly adherent to their prescribed medicine, which could bias our results. Additionally, we were interested in the health outcomes FEV1% and pulmonary exacerbations (since previous trials have shown that DNase and HS can independently improve these outcomes), but we used intravenous antibiotic days as a proxy for exacerbations. The information on intravenous antibiotic days included both planned and unplanned intravenous and is not a direct marker of exacerbations. However, our approach is in line with previous studies using intravenous antibiotic days data from the UK CF Registry.31 32 Finally, the data used are from 2007 to 2018, and outcomes within the UK CF population are evolving rapidly with the introduction of CFTR modulators into routine care.7 This time frame, therefore, predates the widespread introduction of both dual and triple combination CFTR modulator therapy in the UK. It is unknown whether similar results would be obtained in patients established on highly effective modulator therapy. However, by predating the widespread introduction of CFTR modulators, we were, however, able to address this question without potential confounding by modulator status.

    Conclusions

    In an emulated trial using observational UK CF Registry data, we saw no additional benefit to lung function or use of intravenous antibiotics when HS was added to DNase.

    Our findings show that the UK CF Registry can support methodology for emulated trials. Although RCTs remain the gold standard, this methodology has the potential to address questions relevant to the CF community and could be particularly useful for assessing the long-term clinical effectiveness of multiple treatment strategies, since such questions are difficult to answer using a clinical trial. In future work, UK CF Registry data combined with the target trial framework could be used to repeat our study in a post-modulator population, including in groups with and without access to, or intolerant of, CFTR modulator treatments, and to answer related questions about discontinuing treatment in those using CFTR modulators.

    Data availability statement

    Data may be obtained from a third party and are not publicly available. To access the data, an application must be made to the UK CF Registry Research Committee. https://www.cysticfibrosis.org.uk/the-work-we-do/uk-cf-registry/apply-for-data-from-the-uk-cf-registry.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and this work used anonymised data from the UK CF Registry, which has Research Ethics Approval (Reference: 07/Q0104/2). The use of the data for this study was approved by the Registry Research Committee (data request reference 375). This study was also approved by the London School of Hygiene & Tropical Medicine Research Ethics Committee (Reference: 21390). Participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    The authors thank people with CF and their families for consenting to their data being held in the UK CF Registry, and NHS teams in CF centres and clinics for the input of data into the Registry. We also thank the UK Cystic Fibrosis Trust and the Registry Steering Committee for access to anonymised UK CF Registry data.

    References

    1. ↵
      2. Hernán MA ,
      3. Robins JM
      . Using big data to emulate a target trial when a randomized trial is not available. Am J Epidemiol 2016;183:758–64. doi:10.1093/aje/kwv254
      OpenUrlCrossRefPubMed
    2. ↵
      2. Hernán MA ,
      3. Sauer BC ,
      4. Hernández-Díaz S , et al
      . Specifying a target trial prevents immortal time bias and other self-inflicted injuries in observational analyses. J Clin Epidemiol 2016;79:70–5. doi:10.1016/j.jclinepi.2016.04.014
      OpenUrlCrossRefPubMed
    3. ↵
      2. Hernán MA ,
      3. Alonso A ,
      4. Logan R , et al
      . Observational studies analyzed like randomized experiments: an application to postmenopausal hormone therapy and coronary heart disease. Epidemiology 2008;19:766–79. doi:10.1097/EDE.0b013e3181875e61
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      2. Dickerman BA ,
      3. García-Albéniz X ,
      4. Logan RW , et al
      . Avoidable flaws in observational analyses: an application to Statins and cancer. Nat Med 2019;25:1601–6. doi:10.1038/s41591-019-0597-x
      OpenUrlCrossRefPubMed
    5. ↵
      2. Bakker LJ ,
      3. Goossens LMA ,
      4. O’Kane MJ , et al
      . Analysing electronic health records: the benefits of target trial emulation. Health Policy and Technology 2021;10:100545. doi:10.1016/j.hlpt.2021.100545
      OpenUrl
    6. ↵
      2. Shteinberg M ,
      3. Haq IJ ,
      4. Polineni D , et al
      . Cystic fibrosis. Lancet 2021;397:2195–211. doi:10.1016/S0140-6736(20)32542-3
      OpenUrlCrossRefPubMed
    7. ↵
      1. Cystic Fibrosis Trust
      . UK cystic fibrosis registry 2021 annual data report. 2021. Available: https://www.cysticfibrosis.org.uk/sites/default/files/2022-12/CF%20Trust%20Annual%20Data%20Report%202021%20-%20WEB.pdf [Accessed 12 Jan 2023].
    8. ↵
      1. National Institute for Health and Care Excellence
      . Cystic fibrosis: diagnoses and management. 2017. Available: https://www.nice.org.uk/guidance/ng78/resources/cystic-fibrosis-diagnosis-and-management-pdf-1837640946373 [Accessed 07 Dec 2022].
    9. ↵
      2. Yang C ,
      3. Montgomery M
      . Dornase alfa for cystic fibrosis. Cochrane Database Syst Rev 2018;9:CD001127. doi:10.1002/14651858.CD001127.pub4
    10. ↵
      2. Wark P ,
      3. McDonald VM
      . Nebulised hypertonic saline for cystic fibrosis. Cochrane Database Syst Rev 2018;9:CD001506. doi:10.1002/14651858.CD001506.pub4
    11. ↵
      2. Konstan MW ,
      3. Wagener JS ,
      4. Pasta DJ , et al
      . Clinical use of dornase alfa is associated with a slower rate of FEV1 decline in cystic fibrosis. Pediatr Pulmonol 2011;46:545–53. doi:10.1002/ppul.21388
      OpenUrlCrossRefPubMed
    12. ↵
      2. McKone EF ,
      3. Jackson AD ,
      4. Fletcher G , et al
      . Dornase alfa and rate of lung function decline in European patients with cystic fibrosis: a retrospective registry cohort study. J Cyst Fibros 2021;20:552–4. doi:10.1016/j.jcf.2020.08.004
      OpenUrl
    13. ↵
      2. Newsome SJ ,
      3. Daniel RM ,
      4. Carr SB , et al
      . Investigating the effects of long-term dornase Alfa use on lung function using registry data. J Cyst Fibros 2019;18:110–7. doi:10.1016/j.jcf.2018.08.004
      OpenUrlPubMed
    14. ↵
      2. García-Albéniz X ,
      3. Hsu J ,
      4. Hernán MA
      . The value of explicitly emulating a target trial when using real world evidence: an application to colorectal cancer screening. Eur J Epidemiol 2017;32:495–500. doi:10.1007/s10654-017-0287-2
      OpenUrlCrossRefPubMed
    15. ↵
      2. Admon AJ ,
      3. Donnelly JP ,
      4. Casey JD , et al
      . Emulating a novel clinical trial using existing observational data. predicting results of the prevent study. Ann Am Thorac Soc 2019;16:998–1007. doi:10.1513/AnnalsATS.201903-241OC
      OpenUrlPubMed
    16. ↵
      2. Ioannou GN ,
      3. Locke ER ,
      4. O’Hare AM , et al
      . COVID-19 vaccination effectiveness against infection or death in a national U.S. health care system: a target trial emulation study. Ann Intern Med 2022;175:352–61. doi:10.7326/M21-3256
      OpenUrlPubMed
    17. ↵
      2. Granger E ,
      3. Keogh RH ,
      4. Frost F
      . The long-term effects of insulin use in incident cystic fibrosis-related diabetes: a target trial emulated using longitudinal national registry data. ERJ Open Res 2022;8:00170-2022. doi:10.1183/23120541.00170-2022
    18. ↵
      2. Taylor-Robinson D ,
      3. Archangelidi O ,
      4. Carr SB , et al
      . Data resource profile: the UK cystic fibrosis registry. Int J Epidemiol 2018;47:9–10e. doi:10.1093/ije/dyx196
      OpenUrlCrossRefPubMed
    19. ↵
      2. Hall GL ,
      3. Stanojevic S
      . The global lung function initiative (GLI) network ERS clinical research collaboration: how International collaboration can shape clinical practice. Eur Respir J 2019;53:1802277. doi:10.1183/13993003.02277-2018
      OpenUrlAbstract/FREE Full Text
    20. ↵
      2. Hernán MA ,
      3. Brumback B ,
      4. Robins JM
      . Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men. Epidemiology 2000;11:561–70. doi:10.1097/00001648-200009000-00012
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      2. Cole SR ,
      3. Hernán MA
      . Constructing inverse probability weights for marginal structural models. Am J Epidemiol 2008;168:656–64. doi:10.1093/aje/kwn164
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      2. Franklin JM ,
      3. Patorno E ,
      4. Desai RJ , et al
      . Emulating randomized clinical trials with nonrandomized real-world evidence studies: first results from the RCT DUPLICATE initiative. Circulation 2021;143:1002–13. doi:10.1161/CIRCULATIONAHA.120.051718
      OpenUrlCrossRefPubMed
    23. ↵
      2. Matthews AA ,
      3. Dahabreh IJ ,
      4. Fröbert O , et al
      . Benchmarking observational analyses before using them to address questions trials do not answer: an application to coronary thrombus aspiration. Am J Epidemiol 2022;191:1652–65. doi:10.1093/aje/kwac098
      OpenUrl
    24. ↵
      2. Crown W ,
      3. Dahabreh IJ ,
      4. Li X , et al
      . Can observational analyses of routinely collected data emulate randomized trials? Design and feasibility of the observational patient evidence for regulatory approval science and understanding disease project. Value Health 2023;26:176–84. doi:10.1016/j.jval.2022.07.003
      OpenUrl
    25. ↵
      1. National Institute for Health Research
      . NICE real world evidence framework (ECD9). 2022. Available: https://www.nice.org.uk/corporate/ecd9/resources/nice-realworld-evidence-framework-pdf-1124020816837 [Accessed 07 Dec 2022].
    26. ↵
      1. Cystic Fibrosis Trust
      . Working with industries to make medicines safer. 2018. Available: https://www.cysticfibrosis.org.uk/sites/default/files/2020-11/CC32%20%20Working%20with%20Pharmas%20A4%20v4.pdf [Accessed 07 Dec 2022].
    27. ↵
      2. Hernán MA
      . Causal analyses of existing databases: no power calculations required. J Clin Epidemiol 2022;144:203–5. doi:10.1016/j.jclinepi.2021.08.028
      OpenUrlPubMed
    28. ↵
      2. Morris TP ,
      3. van Smeden M
      . Causal analyses of existing databases: the importance of understanding what can be achieved with your data before analysis (commentary on Hernán). J Clin Epidemiol 2022;142:261–3. doi:10.1016/j.jclinepi.2021.09.026
      OpenUrl
    29. ↵
      2. Mansournia MA
      . Sample size considerations are needed for the causal analyses of existing databases. J Clin Epidemiol 2022;141:212. doi:10.1016/j.jclinepi.2021.09.024
      OpenUrl
    30. ↵
      2. Taylor-Robinson DC ,
      3. Smyth RL ,
      4. Diggle PJ , et al
      . The effect of social deprivation on clinical outcomes and the use of treatments in the UK cystic fibrosis population: a longitudinal study. Lancet Respir Med 2013;1:121–8. doi:10.1016/S2213-2600(13)70002-X
      OpenUrl
    31. ↵
      2. Newsome SJ ,
      3. Keogh RH ,
      4. Daniel RM
      . Estimating long-term treatment effects in observational data: a comparison of the performance of different methods under real-world uncertainty. Stat Med 2018;37:2367–90. doi:10.1002/sim.7664
      OpenUrl
    32. ↵
      2. Newsome SJ ,
      3. Daniel RM ,
      4. Carr SB , et al
      . Using negative control outcomes and difference-in-differences analysis to estimate treatment effects in an entirely treated cohort: the effect of Ivacaftor in cystic fibrosis. Am J Epidemiol 2022;191:505–15. doi:10.1093/aje/kwab263
      OpenUrl

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Twitter @Egranger90, @daviesgwyneth, @RuthHKeogh

    • Contributors Authorship CRediT statement: EG: methodology, formal analysis, visualisation, writing—original draft. GD: conceptualisation, methodology, writing—review and editing, supervision. RHK: conceptualisation, methodology, writing—review and editing, supervision, funding acquisition. EG is responsible for the overall content as the guarantor.

    • Funding GD is supported by a UKRI Future Leaders fellowships (MR/T041285/1). RHK and EG are supported by a UKRI Future Leaders fellowship (MR/S017968/1) awarded to RHK. All research at UCL Great Ormond Street Institute of Child Health is made possible by the NIHR Great Ormond Street Hospital Biomedical Research Centre.

    • Disclaimer The views expressed are those of the author and not necessarily those of the NHS, the NIHR or the Department of Health.

    • Competing interests GD reports speaker honoraria from Chiesi Ltd and Vertex Pharmaceuticals. RHK reports a speaker honorarium from Vertex Pharmaceuticals.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.