Statement of principal findings
Diagnostic accuracy showed high specificities for all selected pathologies, with the majority having specificity of ≥ 0.8, meaning a low proportion of FPs. Specificity was lower for aortic dissection and pulmonary disease than for other pathologies. For most pathologies there was also quite high sensitivity, with the majority having sensitivity of ≥ 0.6, with the exceptions of atrial thrombi, atrial septal defect and PE, for which sensitivity was lower. There was a high prevalence (around 25–30%) of ischaemic heart disease, valvular heart disease and heart failure in patients with AF in the included prevalence studies. TTE seems to be a sufficient diagnostic tool for most pathologies included here, but there may need to be extra screening for PE by lung scan, and atrial thrombi and atrial septal hypertrophy by TOE, to avoid FNs for these pathologies.
The results of the mathematical model indicated that in newly diagnosed patients with a CHADS2 score of 1, who are not already receiving warfarin, rivaroxaban, or dabigatran, it may be cost-effective to use TTE to help inform the decision whether to prescribe warfarin. In newly diagnosed patients aged ≥ 65 years it may be cost-effective to use TTE to help inform the decision about whether to prescribe dabigatran. A simplified approach indicated that only a small number of QALYs (0.0033) was required to deem a TTE to be cost-effective, and that incidental benefits may provide more than this number of QALYs.
Strengths and limitations of the assessment
A range of studies were identified that were of good quality and of relevance to UK populations.
It is possible that some studies were missed owing to limiting to studies published in the English language, and only one database being searched for diagnostic accuracy studies.
Data here are not a substitute for a trial of routine screening. In practice, the many different pathologies that could be identified may lead to many different treatment strategies. Patients may have more than one pathology in addition to AF, and may have been diagnosed with other conditions prior to AF diagnosis. It is also important that personnel performing these examinations receive adequate training to minimise bias and improve the quality of screening procedures. The outcome of screening in terms of treatment modification and subsequent prognostic impact will be complex. Receiving diagnoses may result in the patient making lifestyle changes as well as being provided with more appropriate medical treatment. In addition, there may be an emotional impact on patients in terms of undergoing testing, receiving additional diagnoses or being reassured where comorbidities are not diagnosed. Patients need to be provided with information about screening – including implications and limitations – before deciding whether to consent to testing. A trial of routine TTE screening in patients with newly diagnosed AF could address the impact on patients, which may go beyond simple changes in medical treatment, although any such trial would be costly owing to the large sample size and long length of follow-up needed to investigate outcomes including mortality; however, given the benefits and lack of adverse effects of TTE, it is unclear how useful additional evidence from a trial would be.
Our literature review identified no economic evaluations of TTE in patients with AF so it is believed that this is the first. One strength of the modelling is that it uses recent data assessing the sensitivity and specificity of TTE in identifying LA abnormality in patients categorised by CHADS2 score with confirmation provided by TOE. A limitation is that this study had a relatively small data set (n = 405) and was undertaken in Portugal, with the population not necessarily representative of a UK population.
A further strength is the simplified approach that was also undertaken. This showed that the QALYs required for TTE to be cost-effective were very small (< 0.005). Such values could be provided by many factors not incorporated into the mathematical model, and if clinicians believe that benefits other than those associated with reduced stroke rates (albeit at an increased risk of bleeding) are likely then it is probable that TTE is cost-effective.
Analyses have also been undertaken using different OACs with the conclusions remaining constant.
Uncertainties
There are a number of uncertainties within the economic evaluation of TTE. We elected to model the problem using data provided in Providencia et al.,13 as this was a recent, internally consistent study using the CHADS2 tool, and it had also conducted TOE. However, the study was not large (n = 405). This meant that data on the specificity and sensitivity of TTE in identifying LA abnormality were sparse. Using TTE could only affect the decisions in the low risk categories: CHADS2 = 0, and CHADS2 = 1. The number of patients in these categories was small, each fewer than 80 patients, with three values of fewer than 25 patients
A key uncertainty is whether there are other benefits that are accrued from a TTE other than identifying LA abnormality. If these exist, and produce even small QALY gains (> 0.0033) then TTE would be cost-effective in all scenarios.
Other relevant factors
As TTE is relatively easily available, and is a safe and non-invasive diagnostic, no other relevant factors were identified.
Implications for service provision
Our conclusions have few implications for service provision. Should TTE be recommended for those patients with CHADS2 scores of 0 or 1 then this is unlikely to place a great burden on hospitals who are likely to have staff trained in the use of TTE machines. Capacity will depend on scheduling the use of existing TTE equipment and extra staff time needed.
Suggested research priorities
Following up patients with newly diagnosed AF who have undergone TTE to study treatments given as a result of TTE diagnoses and subsequent cardiovascular events could identify potential benefits of routine testing, beyond stroke prevention.
Our conclusions regarding the cost-effectiveness of TTE have been limited by the available data relating to the proportion of people with CHADS2 scores of 0 or 1 who have LA abnormality. These proportions have been shown to markedly affect the cost per QALY and there are few data available. In obtaining such data more accurate estimates of the sensitivity and specificity of TTE in identifying LA abnormality should be collected.
Any additional benefit of TTE beyond those associated with treatment for stroke prevention also needs to be researched. Even small gains would equate to TTE being perceived as cost-effective.
