Article Text
PDF
Abstract
Background Ambulatory management of primary spontaneous pneumothorax has been shown to reduce initial hospitalisation, but at the expense of increase adverse events. As a result, questions remain about the cost-effectiveness of this option.
Objectives A within-trial economic evaluation alongside a randomised controlled trial was performed to assess the cost-effectiveness of ambulatory care when compared with standard guideline-based management.
Methods Patients were randomly assigned to treatment with either an ambulatory device or standard guideline-based management (aspiration, standard chest tube insertion or both). Follow-up was 12 months. Outcomes included healthcare resource use and costs, quality of life, quality-adjusted life-years (QALYs) and cost-effectiveness.
Results 236 patients were recruited and randomly assigned to ambulatory care (n=117) and standard care (n=119). After multiple imputation for missing data, patients in the ambulatory care group had significantly lower National Health Service healthcare costs (−£788, 95% CI difference: −1527 to −50; p=0.037) than those in the standard care group. There were no differences in the number of QALYs gained (mean difference: −0.001, 95% CI difference: −0.032 to 0.030; p=0.95). When standard care was compared with ambulatory care, the incremental cost-effectiveness ratio was £799 066 per QALY gained, well above current thresholds of cost-effectiveness. As a result, the probability of ambulatory care being cost-effective was 0.93.
Conclusion Outpatient ambulatory management is highly likely to be a cost-effective option in the management of primary pneumothorax.
Trial registration number ISRCTN79151659.
- health economist
Data availability statement
Data are available upon reasonable request. All data requests should be submitted to the chief Investigator (NMR) for consideration. Access to anonymised data may be granted for non-commercial research at the discretion of NMR.
Statistics from Altmetric.com
Key messages
What is already known on this topic
Ambulatory management of primary spontaneous pneumothorax (PSP) has been shown to reduce initial hospital stay over aspiration and/or drainage at the expense of increased adverse events. This raises concerns on whether the benefits of ambulatory care are offset by increases in healthcare use over the longer term.
What this study adds
This is the first randomised controlled trial that evaluates the cost-effectiveness of ambulatory care management when compared with standard guideline treatment showing that ambulatory management is cost-effective.
How this study might affect research, practice or policy
These data suggest that PSP can be cost-effectively managed for outpatients, using ambulatory devices in those who require intervention.
Introduction
Pneumothorax is a common clinical problem, occurring in approximately 3000 patients per year in the UK.1 2 Most patients require an intervention to re-expand the lung and more than half of patients subsequently requiring the insertion of a chest tube.3 This results in hospital stays of up to 8 days4 and so the cost of treating pneumothorax can be considerable.
Therefore, interventions that remove the need for long hospital admissions, such as ambulatory management, could provide a highly cost-effective alternative to hospital treatment and admission. The recent Randomised Ambulatory Management of Primary Pneumothorax (RAMPP) trial showed that ambulatory care resulted in a significant decrease in hospital stays (median 0 vs 4 days) when compared with standard guideline-based management (ie, aspiration, drainage or both).5 However, this positive finding was at the expense of increased adverse events, some of which were serious, and required regular review of patients. This, coupled with the fact that pleural vent kits for use in ambulatory care are relatively costly (£212 each),6 raises concerns on whether ambulatory care represents a cost-effective intervention in the treatment of primary spontaneous pneumothorax (PSP).
Using the RAMPP trial data, a prospective, within-trial economic analysis was done to determine whether ambulatory management of patients with PSP was cost-effective when compared with standard care.
Methods
Trial design
Details of the RAMPP trial have been reported in detail elsewhere.5 Briefly, RAMPP was a multicentre, open-label, randomised controlled trial comparing ambulatory management of primary pneumothorax with standard care based on national guidelines.7
Participants
Eligible patients had presented with a symptomatic spontaneous pneumothorax (confirmed by a chest radiograph or CT scan) and were aged 16–55 years. Patients requiring intervention could be enrolled and randomly assigned up to 24 hours after presentation, provided that they remained hospitalised with an ongoing symptomatic pneumothorax despite initial intervention (eg, patients treated initially with aspiration and observed overnight) requiring chest tube insertion.
Interventions
Patients were randomly assigned (1:1) to either insertion of an ambulatory device (Rocket Pleural Vent, Rocket Medical, UK) or standard care (aspiration, standard chest tube insertion or both). All patients were followed up at 1 week after the completion of treatment and 30 days, 6 months and 12 months after randomisation. Hospital admissions and pneumothorax recurrence were measured at each follow-up point. Data were collected on further hospital admissions, wider healthcare resource use, pain and breathlessness Visual Analogue Scale (VAS) scores, quality of life, days off work (if applicable) and smoking status. Patients who were not able to attend face-to-face follow-up appointments were contacted by telephone to check for complications or recurrence, and if this was not possible, data on recurrence were collected through hospital medical records.
Follow-up and outcomes
The primary outcome used in the trial was the total length of hospital stay up to 30 days after randomisation, including initial hospital stay and readmissions.5
For this analysis, the perspective adopted in the economic evaluation was that of the National Health Service (NHS). We included the costs associated with the following healthcare resource use categories from randomisation to 12-month follow-up:
Initial procedures for the treatment of spontaneous pneumothorax (including insertion of an ambulatory device, aspiration, standard chest tube insertion or a combination of the latter two).
Initial length of stay following spontaneous pneumothorax.
Subsequent procedures for pneumothorax.
Subsequent stays in hospital or day cases due to any reason.
Accident and emergency visits.
Secondary outpatient care visits.
Primary care visits, including visits to a general practitioner (GP) at home, doctor’s office or via telephone or to nurse.
Visits for physiotherapy.
Costs of performing the initial procedures to treat pneumothorax were obtained through a micro-costing study in which a number of interventions performed in RAMPP were observed. Staff time included the consultant physician, specialist registrar, nurse and healthcare assistant and was valued using average salaries for that position.8 Unit costs of disposables were valued using prices obtained from the NHS Supply Chain.6 Costs of medications and sedation drugs were obtained from the British National Formulary (https://bnf.nice.org.uk/). Finally, costs of investigations undertaken as part of the intervention, including chest X-rays, were obtained from NHS reference costs.9
All other follow-up resource use was collected using patient questionnaires administered at treatment completion, 1, 6 and 12 months. As an aide memoire, patients were also provided with a resource-use log designed for them to fill in every time they had a contact with the healthcare system.
Unit costs for consultations with GPs and nurses were obtained from the Personal Social Services Research Unit’s Unit Costs of Health and Social Care publications for 2019.8 For all other contacts, unit costs were derived from the NHS National Schedule of Reference Costs 2018/19.8 Using the reasons for hospitalisation reported by patients in the resource-use questionnaires, we obtained diagnosis and procedure codes. These were then translated into a Healthcare Resource Group (HRG) using the HRG4+ 2018/19 Reference Costs Grouper (NHS Digital). Each HRG was then linked to a series of elective, non-elective and day-case reference costs obtained from the 2018/19 schedule of reference costs. All costs are reported in 2019 prices.
Although not relevant to the NHS perspective adopted, additional analyses was performed including the amount of days off work (if applicable) that patients had to take during the 12-month follow-up. Days off work was valued using UK mean earnings.10
Generic health-related quality of life (HRQoL) was measured using the EuroQol-5 Dimensions-5 Levels (EQ-5D-5L) questionnaire and EQ-VAS at time of randomisation (ie, baseline) and at each follow-up point.11 EQ-5D-5L responses were converted into utilities using the validated mapping function to derive utility values for the EQ-5D-5L from the existing EQ-5D-3L version.12
Statistical analysis
Utility and VAS scores at each follow-up point were presented as means (SD). A quality-adjusted survival curve was generated by plotting, against time, the product of the mean utility of patients living at time t and the probability of surviving to time t, in this case 1 as no deaths were observed during 12-month follow-up, in order to create three periods (ie, randomisation to 1-month follow-up, 1–6-month follow-up and 6–12-month follow-up). The area under this quality-adjusted survival curve then gave the mean quality-adjusted survival in each treatment group or quality-adjusted life-years (QALYs) gained over the 12-month follow-up. Utility was assumed to change linearly between each follow-up, rather than changing at the mid-point between follow-ups or being maintained from one follow-up to another.
For the base case, given we had missing data for follow-up resource use and EQ-5D responses, multiple imputation was used to impute missing cost and utility values (please see online supplemental table1).13–15 As per recommended best practice, imputation was implemented separately by randomised treatment allocation.16 Costs were imputed at the most disaggregated level at which the model would converge. As a result, we imputed values for general practice consultation costs (at practice, home and telephone); hospitalisation readmission costs; outpatient care costs (outpatient visits and transportation to visits); emergency care costs (visits to emergency departments and transport by ambulance) and other healthcare costs (physiotherapy and nurse visits). There were no missing data for treatment and initial hospitalisation costs. Rather than imputing missing responses for each of the five domains in the EQ-5D-5L, we imputed the overall EQ-5D-5L utility.17 The imputation of costs and utility was conducted using predictive mean matching (ie, imputes data from similar patients with complete data). Imputation was conducted using age, sex, ethnicity, tobacco and marijuana use history, history of pneumothorax and pneumothorax recurrence. We generated 40 replacement values for each missing case, generating 40 imputed data sets using the Stata ‘mi estimate’ command. We obtained mean estimates of cost and utility and SD. Differences across patients groups were obtained using ordinary least squares regression using the ‘mi estimate: reg’ command.
Supplemental material
In a sensitivity analysis, we also performed an available case analysis. For each treatment group, average resource use/costs were summed over the follow-up periods. Results are then presented as means together with 95% CIs, generated through 10 000 bootstrap estimates. Mean differences in QALYs, resource use and costs between the two treatment groups were also estimated, as well as the 95% CI of the difference using bootstrapping.
Cost-effectiveness was assessed through estimation of the incremental cost per QALY gained. This is based on the difference in mean total cost per patient between the intervention and control group (ambulatory and standard care, respectively), divided by the difference in mean QALYs in the two groups. The resulting incremental cost-effectiveness ratio (ICER) was compared with reference to what the NHS is willing to pay (WTP) for an additional QALY; this currently being ≤£20 000 per QALY gained.18 Results based on the 10 000 bootstrap replications of QALYs and total care costs at 12 months are displayed on a cost-effectiveness plane. The cost-effectiveness plane is used to visually represent the differences in costs and health outcomes between the two study groups in two dimensions. A cost-effectiveness acceptability curve was drawn to represent the probability of cost-effectiveness for different values of WTP.19
Subgroup cost-effectiveness analysis was done for previous pneumothorax.1
All analyses were performed using STATA MP V.15 (64-bit). Statistical significance was set at p<0.05.
Results
Study participants
We recruited patients between 27 August 2015 and 12 March 2019, reaching our target of 236 participants, of whom 117 were assigned to ambulatory care and 119 to standard care. For the primary outcome, we analysed 114 (97%) of 117 patients who received ambulatory care and 113 (95%) of 119 who received standard care, because data were not available for the remaining patients. These patients then formed the basis of the cost-effectiveness analysis (table 1).
Baseline characteristics by treatment allocation
Quality of life
Responses to the EQ-5D-5L at baseline, treatment completion, 1 week and 1, 6 and 12 months, as well as numbers responding, are reported in online supplemental tables 2 and 3. We found that patients attending their 12-month follow-up were more likely to be female and have no history of smoking tobacco or marijuana (online supplemental table 4), but these did not differ by treatment group (online supplemental table 5). EQ-5D responses were then converted into utilities (table 2). There were no statistically significant differences in utilities or VAS scores at 1 week and 1, 6 and 12 months.
EQ-5D utility and EQ-VAS scores at follow-up by treatment allocation
Given there were no observed deaths, follow-up duration (ie, 12 months) was combined with EQ-5D-5L utilities to estimate QALYs at 12 months after randomisation. After imputing for missing EQ-5D data, patients receiving ambulatory care had a non-significant decrease of 0.001 (95% CI difference: –0.032 to 0.030) QALYs when compared with those receiving standard care. In the available case analysis, ambulatory care patients also had a non-significant decrease of 0.005 (95% CI difference: −0.037 to 0.025).
Resource use and costs
For patients receiving ambulatory care, the initial mean hospital stay was of 1.33 (SD 3.24) days compared with 4.11 (SD 4.91) days in the standard care group, a statistically significant reduction of 2.78 (95% CI: −3.87 to −1.69; p<0.0001) days in hospital. There were no statistically significant differences between the two groups in hospital days due to readmissions up to the 1-month follow-up, with patients receiving ambulatory care having stays in hospital due to readmissions of 0.81 (SD 2.49) days versus 1.06 (SD 3.04) days in the standard care group (p=0.49). Except for outpatient visits, where patients receiving ambulatory care had, on average, 0.98 (95% CI difference: 0.21 to 1.71; p=0.012) more visits than those receiving standard care over the course of 1 year, there were no other statistically significant differences in healthcare resource use (table 3). Resource use by follow-up, as well the number of patients for whom we had data for, is reported in online supplemental table 6.
Twelve-month mean healthcare resource use (SD) by treatment allocation in the available case analysis
There were no missing data regarding the costs of initial hospital stays, trial treatments and 1-month readmissions. Given the significant reduction in initial hospital stay, mean costs of the initial hospital stay were significantly lower in the ambulatory care group than the standard care group (−£1036, 95% CI difference: −1487 to −583, p<0.0001 – table 4). However, costs of pneumothorax treatment were significantly higher for the ambulatory care group (£329) than the standard care group (£172; p<0.0001). There were no significant differences in 1-month readmissions or subsequent hospital admissions over the 1 year of follow-up.
One-year mean cost differences between patients randomised to standard care and ambulatory care
For follow-up costs, results of the multiple imputation showed that, except for outpatient care costs, which were higher in the ambulatory care group (£188, 95% CI difference: 91 to 285, p<0.001), there were no other significant differences in cost categories (table 4). Overall, during the 1-year follow-up, patients in the ambulatory care group had significantly lower NHS healthcare costs than those in the standard care group (−£788, 95% CI difference: −1527 to −50, p=0.037). Online supplemental figure 1 presents how mean QALY and cost differences between the two patient groups varied across the 40 imputed datasets used in the multiple imputation analysis.
Over the 12 months, mean number of days off work was 26 in patients receiving ambulatory care compared with 24 in those receiving standard care, a non-significant increase of 2 (95% CI difference: −14 to 25; p=0.81) days. This resulted in non-significantly higher productivity losses in the ambulatory care group (table 4). Overall, when NHS and productivity losses were combined together, there were no overall significant differences across groups.
Cost-effectiveness
Cost-effectiveness was assessed using an NHS perspective. Given that standard care resulted in non-significantly higher QALYs and significantly higher NHS costs than ambulatory care, cost-effectiveness was assessed by comparing the additional costs of standard care versus ambulatory care, divided by the additional QALY gains when standard care was compared with ambulatory care.
In the multiple imputation analysis, the incremental cost per QALY gained when standard care was compared with ambulatory care was £799 066, well above the current £20 000 cost-effectiveness threshold. At this cost-effectiveness threshold, the average, across the 40 imputed datasets, probability that ambulatory care was cost-effective over standard care was 0.93 (figure 1). Across the 40 imputed datasets this probability ranged between 0.71 and 0.99 (online supplemental figure 2). Repeating the analysis, using a wider perspective (ie, including NHS costs and productivity losses), the ICER when standard care was compared with ambulatory care was £696 986 per QALY gained, again above current thresholds of cost-effectiveness. From an NHS perspective, results of the available case analysis showed that the ICER of standard care versus ambulatory care was £151 438 (table 4, online supplemental table 7, well above conventional thresholds of cost-effectiveness, with the probability that ambulatory care was cost-effective at a £20 000 per QALY threshold being 0.87 (figure 2). From a wider perspective, the ICER was £89 904 per QALY gained when standard care was compared with ambulatory care.
Cost-effectiveness acceptability curve—probability that ambulatory care was cost-effective when compared standard care after multiple imputation analysis. QALY, quality-adjusted life-year.
Cost-effectiveness plane—cost-effectiveness of standard care when compared with ambulatory care in the available case analysis y-axis: Incremental costs when standard care was compared with ambulatory care x-axis: Incremental quality-adjusted life-years when standard care was compared with ambulatory care.
When the multiple imputation analysis was restricted to patients with previous history of pneumothorax, patients receiving ambulatory care had significantly lower NHS costs than those receiving standard care (−£1960, 95% CI difference: −3267 to −654) and non-significantly lower QALYS (−0.010, 95% CI difference: −0.070 to 0.051). In this subgroup, therefore, the incremental cost per QALY gained when standard care was compared with ambulatory care was £201 140.
Discussion
This multicentre, open-label, randomised controlled trial assessed the cost-effectiveness of an ambulatory treatment strategy for PSP with standard care using evidence-based guidelines (aspiration, chest drain insertion or both). The results show the cost-efficiency of the ambulatory strategy, resulting in significantly lower NHS costs during the 1 year of follow-up, with no difference in health outcome.
Given that the difference in QALYs between the two study groups was very small and non-significant, our results of cost-effectiveness were driven by differences in cost. We found that the additional cost per QALY gained when standard care was compared with ambulatory care was £799 066 after imputing for missing data. This high ICER, which is well above current UK cost-effectiveness thresholds, was obtained by dividing the difference in costs (standard care generated significant additional costs per patient when compared with ambulatory care) by the very small difference in QALYs.
Serious adverse events, defined as those needing admission to hospital, occurred exclusively in the ambulatory care arm.1 Serious adverse events related to treatment included enlargement of pneumothorax despite the ambulatory device being in place and device blockage and kinking, requiring chest tube insertion and hospitalisation. In the ambulatory care group, three patients had unrecognised haemopneumothoraces, which at review were not considered related to the intervention. Although this higher rate of serious adverse events is clearly important, it did not translate into higher costs of hospital readmissions in the ambulatory care group or negative health outcomes, as QALY gains were similar in both treatment groups.
Despite patients in the ambulatory care group spending significantly less time in hospital than those receiving standard care, we saw no differences, over the course of the year, in the number of days patients had to take off work. Although patients with primary pneumothorax being relatively young and otherwise healthy, on average, over the course of the year following pneumothorax, patients lost approximately 25 days off work, considerably more than the average 4.68 days lost due to sickness per employee in the UK.20
The burden of pneumothorax on patients was also observed in low levels of self-reported HRQoL in patients, particularly at time of pneumothorax and shortly after. At time of randomisation, patients reported utilities of 0.57 (out of a maximum of 1), with 59% reporting moderate to extreme pain. However, by end of follow-up mean EQ-5D utility was 0.92, with 62% of patients reporting perfect health. Our results also show that most of the care for pneumothorax at follow-up was in hospital. For example, irrespective of treatment group, patients made less than 1.5 visits to their GP over the course of the year. In addition, virtually all (98%) the NHS costs incurred over 1-year follow-up were incurred in hospital (treatment, hospitalisations, outpatient specialist care and emergency care).
Limitations
In addition to the trial limitations already reported,1 this economic evaluation alongside RAMPP also had some shortcomings. First, despite best efforts, we had considerable missing quality of life and resource use data at the latter follow-ups, especially at the 1-year follow-up. Patients who suffer with PSP are usually young with minimal comorbidity, and after resolution of their pneumothorax, their quality of life quickly returns to premorbid levels (see high EQ-5D scores at follow-up). As such, we presume that they no longer feel the need to attend follow-up, even as part of a clinical trial. This missing data can not only lead to loss of statistical power but also to bias.21 We, therefore, used multiple-imputation methods. Both after multiple imputation and when using available information only, we found that our conclusions that ambulatory care was cost-effective remained unaltered. If anything, the likelihood of ambulatory care being cost-effective was increased.
Quality of life was assessed at defined fixed time points over follow-up, rather than regularly (eg, weekly), or when an adverse or subsequent event occurred. As a result, the true impact of the interventions on quality of life might not have been captured in full. For example, the benefits of shorter admissions might not have been captured. Equally, the impact of adverse events or subsequent pneumothoraces happening half-way through follow-up might not have been captured at the follow-up visit. Therefore, the impact of an adverse event, whose consequences could last for a number of days and result in a reduction in quality of life during that time, but had resolved at the time of the follow-up visit would not have been captured.
In a bid to avoid patient overburden, follow-up questionnaires were kept to a minimum. As a result we did not include questions on informal care, more in depth questions about return to work and other NHS resource use.
There was a slight difference in proportion of patients with previous pneumothorax between the two groups. However, in this pragmatic study, this information would not have impacted on the initial treatment options for the patients assessed by the primary outcome. In the longer term, the patients with recurrent episodes may have been more likely to be referred for elective thoracic surgery (for recurrence prevention). However, the authors’ feel that this would not have a significant impact on the health economic (HE) analysis.
In conclusion, in patients with PSP, ambulatory management not only reduces initial hospital stay but also overall NHS costs over 12-month follow-up, with no decrease in health outcomes as measured by quality of life. Outpatient ambulatory management is highly likely to be a cost-effective option in the management of this condition.
Data availability statement
Data are available upon reasonable request. All data requests should be submitted to the chief Investigator (NMR) for consideration. Access to anonymised data may be granted for non-commercial research at the discretion of NMR.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by UK National Research Ethics Service Committee (15/SC/0240). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors thank and acknowledge the contribution of all the patients and trial team members at each recruitment site.
References
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 @ramonluengofern, @DrHallifax
Contributors RH and NMR: conceived study design, trial management and oversight, patient recruitment and data collection. RL-F and FL: statistical analysis. RL-F: literature search and initial manuscript preparation. RL-F is the guarantor of the data. All authors reviewed and approved the final manuscript.
Funding This paper presents independent research funded by the National Institute for Health Research (NIHR) under its Research for Patient Benefit Programme (PB-PG-0213-30098). RH was funded by a Medical Research Council Clinical Research Training Fellowship (MR/L017091/1). NMR was funded by the NIHR Oxford Biomedical Research Centre. Rocket Medical UK provided the Pleural Vent devices and consumables for the trial. The trial was sponsored by the University of Oxford and managed by the Oxford Respiratory Trials Unit, who also undertook data management. The views expressed are those of the authors and not necessarily those of the National Institute for Health Research or the UK Department of Health and Social Care.
Competing interests NMR reports consultancy fees from Rocket Medical, during the conduct of the study, and Lung Therapeutics and grants from BD Biosciences, outside the submitted work. All other authors declare no competing interests.
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.