The cost-effectiveness of targeted or universal screening for vasa praevia at 18–20 weeks of gestation in Ontario

Authors


Dr GS Zaric, Richard Ivey School of Business, University of Western Ontario, 1151 Richmond Street North, London, ON, Canada, N6A 3K7. Emailgzaric@ivey.uwo.ca

Abstract

Please cite this paper as: Cipriano L, Barth W, Zaric G. The cost-effectiveness of targeted or universal screening for vasa praevia at 18–20 weeks of gestation in Ontario. BJOG 2010;117:1108–1118.

Objectives  To estimate the cost-effectiveness of targeted and universal screening for vasa praevia at 18–20 weeks of gestation in singleton and twin pregnancies.

Design  Cost–utility analysis based on a decision-analytic model comparing relevant strategies and life-long outcomes for mother and infant(s).

Setting  Ontario, Canada.

Population  A cohort of pregnant women in 1 year.

Methods  We constructed a decision-analytic model to estimate the lifetime incremental costs and benefits of screening for vasa praevia. Inputs were estimated from the literature. Costs were collected from the London Health Sciences Centre, the Ontario Health Insurance Program, and other sources. We used one-way, scenario and probabilistic sensitivity analysis to determine the robustness of the results.

Main outcome measures  Incremental costs, life expectancy, quality-adjusted life-years (QALY) and incremental cost-effectiveness ratio (ICER).

Results  Universal transvaginal ultrasound screening of twin pregnancies has an ICER of $5488 per QALY-gained. Screening all singleton pregnancies with the risk factors low-lying placentas, in vitro fertilisation (IVF) conception, accessory placental lobes, or velamentous cord insertion has an ICER of $15,764 per QALY-gained even though identifying some of these risk factors requires routine use of colour Doppler during transabdominal examinations. Screening women with a marginal cord insertion costs an additional $27,603 per QALY-gained. Universal transvaginal screening for vasa praevia in singleton pregnancies costs $579,164 per QALY compared with targeted screening.

Conclusions  Compared with current practice, screening all twin pregnancies for vasa praevia with transvaginal ultrasound is cost-effective. Among the alternatives considered, the use of colour Doppler at all transabdominal ultrasound examinations of singleton pregnancies and targeted use of transvaginal ultrasound for IVF pregnancies or when the placenta has been found to be associated with one or more risk factors is cost-effective. Universal screening of singleton pregnancies is not cost-effective compared with targeted screening.

Introduction

Vasa praevia is a condition of pregnancy in which unprotected fetal vessels traverse the membranes in the lower uterine segment.1 These unprotected vessels may be subject to compression by the presenting part during labour and may tear at the time of membrane rupture causing serious fetal haemorrhage. Reports of perinatal mortality for intrapartum diagnosis range from 22.5 to 100%, with the largest series to date reporting a mortality rate of 56.4%.2,3

The incidence of vasa praevia is uncertain, with reported rates ranging from 1 : 6080 to 1 : 1340.4,5 Vasa praevia is almost exclusively caused by velamentous cord insertion (Type I vasa praevia) and bilobed or succenturiate lobed placenta (Type II vasa praevia) in the lower uterine segment. In addition, membranous vessels may develop from a marginally inserted cord resulting in vasa praevia when those vessels traverse the internal os;1 however, there are few reports of marginal cord insertion as the sole aetiology for vasa praevia.5–7 Other pregnancy factors have been correlated to incidence of vasa praevia including second-trimester low-lying placenta (regardless of whether placenta praevia resolves in the third trimester), in vitro fertilisation (IVF) pregnancies and multiple pregnancies.5

Despite its relatively low incidence, vasa praevia is an important public-health issue because the consequences of the condition are so severe. Feasibility of routine prenatal diagnosis has been demonstrated in several large observational studies using transabdominal ultrasound with colour Doppler in combination with transvaginal ultrasound imaging at 18–20 weeks of gestation.5,6,8 Prenatal diagnosis permits delivery by planned caesarean section sufficiently early to avoid membrane rupture. A review of 155 cases found that prenatal diagnosis results in a 95% reduction in late fetal and neonatal mortality.3 Such observations have inspired several clinical groups2,3,5,8–12 and parents13 to recommend routine identification of umbilical cord insertion at 20 weeks of gestation and targeted screening for vasa praevia in women identified as at high risk for the condition.

Currently, Canadian Medicare provides all pregnant women with the opportunity to access a second-trimester transabdominal ultrasound scan to evaluate fetal anatomy, placental position, the amount of amniotic fluid and soft-markers for fetal aneuploidies,14,15 at which time screening for additional risk factors could occur. However, screening for some vasa praevia risk factors, such as velamentous cord insertion, requires the routine use of colour Doppler, which is not currently a standard part of prenatal ultrasound care and may not be ubiquitously available. In August 2009, the Society of Obstetricians and Gynaecologists of Canada released guidelines for the management of vasa praevia that state that the cord insertion site should be identified at second-trimester ultrasound when the placenta is low lying and that transvaginal ultrasound may be considered for all women at high risk for vasa praevia including those with low or velamentous cord insertion, bilobed or succenturiate placenta, or those with vaginal bleeding.16 The guidelines also recommend hospitalisation at 30–32 weeks of gestation for all women whose pregnancies are affected by vasa praevia.16 The budget impact and the cost-effectiveness of these recommendations have not been evaluated.

Given constraints on public-health resources, it is important to know whether the incremental costs of newly proposed initiatives are affordable and of good value relative to other proposed public-health programmes. To determine value, policy-makers must compare alternative policy choices based on their long-term costs and benefits to society. The purpose of the study was to evaluate the cost-effectiveness of universal or targeted screening for vasa praevia in singleton and twin pregnancies.

Methods

Model overview

We developed a mathematical model to evaluate the cost and health consequences of the status quo (no routine screening for vasa praevia) and several strategies for the routine screening of women at 18–20 weeks of gestation. We evaluated the total costs and total benefits [measured in life-years gained and quality-adjusted life-years (QALYs) gained; see Supplementary Appendix S1 for definition] for cohorts of pregnant women in Ontario in 1 year (average age of 30 years; 130,000 who were pregnant with a singleton pregnancy and 2150 who were pregnant with a twin pregnancy17). We performed the cost–utility analysis (see Supplementary Appendix S1 for definition) from the perspective of the Canadian publicly funded healthcare system, which includes direct healthcare costs paid by the public payer, private insurers, women and their families and caregivers, as well as time costs incurred by women and their families and caregivers.18 Total costs and benefits are the sum of those accrued by mothers and infants. Costs and benefits were discounted (see Supplementary Appendix S1 for explanation) at 5% annually. We followed the Canadian Guidelines for the Economic Evaluation of Health Technologies throughout the development of the model and the analysis.18 A glossary of health economic terms used and a detailed explanation of the model assumptions and analysis are available in Supplementary Appendices S1 and S2.

For singleton pregnancies, we considered nine strategies; there included seven primary vasa praevia screening strategies:

  • 1 no routine transvaginal ultrasound screening for vasa praevia (the status quo) and targeted screening of all pregnancies with
  • 2 IVF
  • 3 low-lying placenta
  • 4 a bilobed or succenturiate lobed placenta
  • 5 a velamentous cord insertion
  • 6 a marginal cord insertion
  • 7 universal screening of all pregnancies;

as well as two scenarios in which individuals with one or more risk indicator would be referred to transvaginal ultrasound screening for vasa praevia:

  • 8 screening of high-risk indicators defined as IVF, low-lying placenta, accessory placental lobes, or velamentous cord insertion
  • 9 screening of any risk indicator defined as IVF, low-lying placenta, accessory placental lobes, velamentous cord insertion, or marginal cord insertion.

For twin pregnancies, we considered three vasa praevia screening strategy options:

  • 1 no routine screening (the status quo)
  • 2 screening all pregnancies resulting from IVF
  • 3 universal screening of all twin pregnancies.

Descriptions of each risk factor, its prevalence in the pregnant population, and the frequency of vasa praevia in women with the risk factor are given in Table 1.

Table 1.   Risk factors for vasa praevia
Risk factorDescriptionPrevalence of risk factor at 18–20 weeks of gestationProportion of women with this indicator that have vasa praeviaProportion of vasa praevia cases with this risk indicatorRelative risk of vasa praevia given the presence of the risk factor*
  1. *Calculated: Relative risk = Proportion of vasa praevia cases who have this indicator/Proportion of vasa praevia cases who do not have this indicator.

  2. **Estimated using the number of reported cases in which velamentous or marginal cord insertion site was observed and the total number of cases described in which cord insertion site was described.2,5,6,28,32

Singleton pregnancies  0.04%8,27 (0.024,6–0.085,28)100% 
Singleton pregnancy with…
 IVFConception of pregnancy aided with IVF techniques0.50%17,29 (0.05–1.0)0.49%5 (0.06–1.2)6.1%14.3
 Low-lying placentaPlacenta within 2 cm of internal os1.15%30 (0.60–2.0)1.32%5 (0.6–2.6)37.9%52.3
 Bilobed or succenturiate lobed placentaPresence of two or more accessory placental lobes, of equal (bilobed) or unequal (succenturiate) size3.2%31 (2.0–4.2)0.67% (0.42–1.0)53.3%3 (33.36–81.18)34.5
 Velamentous cord insertionUnprotected fetal vessels traverse the chorioamniotic membranes before inserting into the placenta1.11%1 (0.30–1.23)1.76% (1.0–2.4)48.9%** (27.38–66.76)85.3
 Marginal cord insertionUmbilical cord inserts on the edge of the placenta rather than the centre of the placenta6.90%1 (2.70–9.00)0.06% (0.03–0.08)11%5 (6.0**–13.16)1.7
 Any high-risk indicatorIVF, low-lying placenta, accessory placental lobes, velamentous cord insertion5.85% (2.9–5.6)0.59% (0.23–2.6)95%306.2
 Any risk indicatorIVF, low-lying placenta, accessory placental lobes, velamentous cord insertion, marginal cord insertion12.34% (8.2–16.5)0.29% (0.12–1.4)100% 
Twin pregnancies  0.43%5 (0.2–0.8)100% 
Twin pregnancy with…
 IVFConception of pregnancy aided with IVF techniques13.81%17,29 (7.0–20.0)0.49%5 (0.1–1.0)15.6%1.2

A schematic of the model is presented in Figure 1. In step 1, pregnant women are divided into four groups based on whether they have the high-risk indicator that classifies them as part of the targeted screening population in the scenario and whether their pregnancy is affected by vasa praevia. In step 2, women who attend their routine transabdominal ultrasound may be identified as being in the high-risk group for the scenario. Those women are referred to receive a follow-up transvaginal ultrasound to screen for vasa praevia. Some women not identified as being high risk for the scenario will be referred to transvaginal ultrasound for an indication unrelated to suspicion of vasa praevia. In step 3, independent of the presence or absence of vasa praevia, some fetuses will not survive to 28 weeks of gestation. Pregnancies diagnosed with vasa praevia are all planned to be delivered early by caesarean section (at 35 weeks of gestation for singletons and 34 weeks of gestation for twins). In the model, onset of labour before the planned delivery results in an emergency caesarean section. Some pregnancies without a diagnosis will have a planned caesarean delivery (at 39 weeks of gestation for singletons and 37 weeks of gestation for twins). In step 4, infant outcomes are determined. Late fetal and neonatal death depends on the presence of vasa praevia and the timing and mode of delivery. In the case of twins, we made the simplifying assumption that each twin’s outcome is not affected by the outcome of the other twin. In step 5, maternal outcomes are determined. Maternal mortality and morbidity are not affected by the presence of vasa praevia but are affected by the mode of delivery.

Figure 1.

 Schematic of the decision-analytic model. The small square represents the decision to implement a policy of screening some or all pregnant women for vasa praevia or to continue the status quo of not screening. Circles represent chance events. Health states in quotations refer to states of diagnosis rather than true health states. U/S = ultrasound.

Data sources

A detailed description of parameter assumptions is given in Supplementary Appendix S2. The probabilities associated with each event, as well as the costs and benefits associated with each outcome in the model, including test characteristics and outcomes, are given in Supplementary Table S1. We estimated base-case values for model variables based on the medical literature and expert opinion. We also considered alternative values for inputs based on low and high values observed in the literature to determine how sensitive model results were to our choice of base case parameters.

Costs

Costs are presented in 2008 Canadian dollars. When necessary, costs were converted to 2008 Canadian dollars using the Canadian Consumer Price Index19 and appropriate exchange rates. Costs included direct healthcare costs accrued by the Ontario Ministry of Health and Long-Term Care, private healthcare expenditures, and costs of patient time.

The Ontario Ministry of Health and Long-Term Care Schedule of Benefits for Physician Services was used to calculate the reimbursement for transabdominal and transvaginal ultrasound imaging and for other physician services included in the model, such as delivery.20 Currently there is no additional reimbursement for the use of colour Doppler and it is not standard practice to identify cord insertion site. We estimated the cost of the additional resources necessary to provide these services as $12.28 (see Supplementary Appendix S2 for details). We estimated the hospital component of the cost of vaginal delivery, caesarean delivery, emergency caesarean delivery and the costs of the initial hospitalisation of the neonates using costs observed in 2005/2006 at the London Health Sciences Centre.21

Benefits

Benefits in the model were calculated as late fetal and neonatal deaths averted, life-years gained, and QALYs gained. Age-adjusted baseline quality-of-life weights for women and for infants were estimated from Canadian averages.22 Quality of life of mothers was reduced if delivery-associated morbidity occurred, or in the instance of a late fetal or neonatal death.

Cost-effectiveness

Cost-effectiveness results are presented using the incremental cost-effectiveness ratio (ICER) in units of dollars per additional QALY ($/QALY-gained; see Supplementary Appendix S1 for definition). In the Canadian context, Laupacis et al. suggested that, in general, there is ‘strong evidence for adoption and appropriate utilisation’ of health programmes with ICERs less than $20,000 per QALY-gained and that health programmes with ICERs greater than $100,000 per QALY-gained are unlikely to be an appropriate use of shared resources.23

Results

Base-case analysis, singletons

We estimated the net discounted costs and benefits accrued by both mothers and infants under each of the hypothetical screening scenarios (Table 2). The increased number of caesarean deliveries (2.5–50 additional deliveries by caesarean section depending on the screening scenario) results in an increase in maternal deaths during labour in all scenarios (between 0.00007 and 0.00198 additional maternal deaths per year associated with childbirth). On average, women experience an increase in quality-adjusted life expectancy in all scenarios because of the decreased number of late fetal and neonatal deaths.

Table 2.   Costs, life-years, and QALYs for various vasa praevia screening strategies (costs and benefits are presented per pregnancy and discounted at 5% annually)
StrategyPregnancies ≥ 1 TV ultrasoundCostsLife-yearsQALYsCompared with next-best alternative
  Screening programme costMother’s lifetimeInfant’s lifetimeTotal costInfant’s life-yearsTotal life-yearsInfant’s QALYsTotal QALYs$/LY gained$/QALY gained
  1. *0.7370 in additional costs = the cost of 587 additional transvaginal ultrasounds and appropriate follow up from those ultrasounds (including patient time costs) divided over the entire singleton pregnancy cohort (130,000 pregnant women).

  2. **118,560 = (130,000 × 96% who use any prenatal ultrasound services) – 6240 referred to transvaginal ultrasound for an indication other than suspicion of vasa praevia in the no screening scenario.

Singletons
No routine TV screening6240    $161   $77,529   $51,946$129,63720.148739.156819.073336.0503  
Additional cost and health impact of policies below compared to no routine screening when screening all pregnancies with:
 IVF5930.74410.51820.35561.61780.00020.00020.00020.0002DominatedDominated
 Low lying placenta25493.22793.19072.20538.62390.00140.00140.00130.0014     6320    6145
 Bilobed or succenturiate lobed placenta40255.04933.41782.343810.81090.00140.00140.00140.0015DominatedDominated
 Velamentous cord insertion171714.02804.05882.827120.91390.00180.00170.00170.0018DominatedDominated
 Marginal cord insertion848922.26491.08860.643623.99710.00040.00040.00040.0004DominatedDominated
 Any high-risk indicator  (All listed above except  marginal cord insertion)872622.73716.99304.788434.51850.00300.00300.00280.003016,21215,764
 Any risk indicator (All listed above)16,64832.51378.09525.449746.05870.00340.00340.00320.003528,38927,603
Universal screening118,560**146.112210.70445.8266162.64330.00360.00360.00340.0037595,214579,159
Twins
No routine TV screening103  $161   $78,505 $102,780$181,44739.605158.612937.491354.4021  
Additional cost and health impact of policies below compared to no routine screening when screening all pregnancies with …
 IVF27120.549019.648219.696459.89370.01150.01150.01080.011552285213
Universal screening1961148.4623126.3147126.2714401.04910.07340.07340.06960.073655045489

To assess the cost-effectiveness of screening for vasa praevia in singleton pregnancies, we compared no routine transvaginal ultrasound screening for vasa praevia (the status quo) to each of the other strategies. Using our base-case parameter values, we estimated that screening pregnancies identified as having a low-lying placenta would cost $6145 per QALY-gained compared with no routine screening (Table 2, Figure 2A). Screening only those pregnancies identified as having a marginal cord insertion is never the optimal strategy because it costs more and shortens quality-adjusted life expectancy compared with screening only pregnancies with a low-lying placenta. Screening only IVF pregnancies, only pregnancies with accessory placental lobes, or only pregnancies with velamentous cord insertion is never the optimal strategy because a combination of other strategies could provide more QALYs at a lower cost. Screening pregnancies identified to have any of the four high-risk indicators (IVF, low-lying placenta, accessory placental lobes, or velamentous cord insertion) costs $15,764 per QALY-gained compared with screening individuals identified as having a low-lying placenta. The incremental cost of screening for any of the five possible risk indicators (including marginal cord insertion) compared with screening for only high-risk indicators is $27,603 per QALY-gained. Compared with targeted screening, universal screening costs $579,164 per QALY-gained.

Figure 2.

 Cost-effectiveness plane. (A) Average incremental lifetime costs and QALYs for mother and expected infant for various vasa praevia screening strategies per singleton pregnancy. (B) Average incremental lifetime costs and QALYs for mother and expected infants for various vasa praevia screening strategies per twin pregnancy.

Compared to the status quo, screening singleton pregnancies affected by a high-risk indicator will result in approximately 8726 more women receiving at least one transvaginal examination during their pregnancy, 33 more caesarean sections in total but 24 fewer emergency caesarean sections. In addition, we estimate that screening targeted at high-risk singleton pregnancies results in 19 fewer late fetal or neonatal deaths each year, compared with 23 fewer late neonatal or fetal deaths if universal screening were to be adopted.

Compared with the status quo, screening all singleton pregnancies affected by any high-risk indicator will have a lifetime discounted cost to the Ontario Ministry of Health and Long Term Care of $3.4 million per cohort of singleton pregnancies, 80% of which will be incurred in the first year as costs of screening ($2.2 million) and delivery. The remaining 20% of which will be the cost associated with the lifetime health care of infants that would otherwise have died. If marginal cord insertion is included, the Ontario Ministry of Health and Long Term Care cost per cohort increases to $4.2 million. Universal screening of singleton pregnancies will increase costs by $10.6 million per cohort.

Base-case analysis, twins

For twin pregnancies, we estimated that screening IVF pregnancies would cost $5213 per QALY-gained compared with no routine screening (Table 2, Figure 2B). Universal screening of twin pregnancies costs $5489 per QALY-gained compared with only screening twin pregnancies conceived through IVF. We estimate that universal screening for vasa praevia in twin pregnancies will result in 7.8 fewer late fetal and neonatal deaths each year.

The Ontario Ministry of Health and Long-Term Care will incur an additional lifetime discounted cost of $573,000 per cohort of universally screened twins, of which $148,100 will be the cost of screening.

Sensitivity analysis

We performed many additional analyses using alternative input parameters based on the low and high values reflected in the literature to assess how the results were influenced by our choice of base-case parameters (see Supplementary Figure S1 and Table S2). In scenarios in which the frequency of risk indicators increased and the total incidence of vasa praevia in the population decreased, the incremental cost of screening individuals identified as having any risk indicator increased from $27,603 per QALY-gained to $64,508 per QALY-gained. If transvaginal ultrasound is less accurate at diagnosing vasa praevia than we assumed in our base case (both in its ability to accurately identify cases of vasa praevia and in its ability to appropriately rule out vasa praevia), the cost of screening for any risk indicator increased to $101,337 per QALY gained. However, if marginal cord insertion is excluded as an indication for vasa praevia screening, the cost-effectiveness of screening all individuals with a high-risk indicator for vasa praevia is only $25,174 per QALY-gained even with fairly poor diagnostic accuracy. We also considered a scenario in which only 1% of women with vasa praevia had marginal cord insertion as their only risk indicator; in this case, the incremental cost of including marginal cord insertion as an indicator for referral to transvaginal examination increased from $27,603 per QALY-gained in the base case to $37,902 per QALY-gained.

In our base-case analysis, we did not include the one-time cost of training obstetricians and sonographers in the use of colour Doppler, the detection of velamentous vessels or vasa praevia, or in the appropriate management of women with vasa praevia. We performed a threshold analysis to determine the maximum amount of annual training investment that would be required before targeted screening for vasa praevia cost more than $100,000 per QALY-gained. We calculate that $3.9 million annually could be invested in training ($30 per singleton pregnancy) before the ICER of screening for any risk indicator exceeded $100,000 per QALY-gained. In addition, investment in clinical skills training in vasa praevia would have to exceed $18 million ($138 per singleton pregnancy) before screening targeted at only the high-risk indicators would have an ICER exceeding $100,000 per QALY-gained.

We constructed a ‘worst-case scenario’ (described in Supplementary Table S2). In this case, screening for low-lying placenta has an ICER of $9256 per QALY-gained. Compared with that strategy, screening for any high-risk indicator cost $108,907 per QALY-gained and a screening strategy including marginal cord insertion was dominated (see Supplementary Appendix S1 for definition).

Probabilistic sensitivity analysis (see Supplementary Appendix S1 for definition) indicates that for singleton pregnancies, screening for any risk indicator has a 51% probability of being cost-effective at a willingness-to-pay threshold (see Supplementary Appendix S1 for definition) of $50,000 per QALY-gained and a 75% probability of being cost-effective at a willingness-to-pay threshold of $100,000 per QALY-gained. The probability that no screening (i.e. maintain the status quo) is the most cost-efficient choice is less than 1% at any willingness to pay greater than $16,000 per QALY-gained (see Supplementary Figure S1A,B).

The incremental cost of screening all twin pregnancies for vasa praevia using transvaginal ultrasound at 18–20 weeks of gestation was below $20,000 per QALY-gained for all deterministic scenarios considered. In probabilistic sensitivity analysis, we identified that at a willingness-to-pay threshold of $20,000 per QALY-gained, there is a 99.6% probability that universal screening is the optimal choice. The probability that not screening is the most cost-efficient choice is <1% at any willingness-to-pay threshold greater than $10,000 per QALY-gained (see Supplementary Figure S1C,D).

Discussion

We conducted a cost–utility analysis of the addition of targeted and universal screening strategies for vasa praevia in singleton and twin pregnancies and found that screening for vasa praevia decreases late fetal and neonatal mortality but costs more than the status quo of not screening. We are not aware of any other study of the cost-effectiveness of screening for vasa praevia.

Based on the ranges for defining ‘cost-effective’ in Canada suggested by Laupacis et al.,23 the optimal strategy is one in which all routine transabdominal ultrasounds at 18–20 weeks of gestation were augmented with colour Doppler and that all pregnancies identified as having any risk indicator for vasa praevia (IVF conception, low-lying placenta, accessory placental lobes, or velamentous or marginal cord insertion) were then screened for vasa praevia by transvaginal ultrasound imaging. Universal transvaginal ultrasound screening of singleton pregnancies for vasa praevia consistently costs more than $100,000 per QALY-gained across all analyses and therefore is not likely to be cost-effective compared with targeted screening.

For twin pregnancies, universal screening with transvaginal ultrasound at 18–20 weeks of gestation consistently costs <$20,000 per QALY-gained and so is almost certainly cost-effective. Sensitivity analysis of the singleton pregnancy screening strategies indicated that in some scenarios transvaginal ultrasound screening of individuals with any of the risk indicators or even just high-risk indicators may not be cost-effective at a willingness-to-pay of $100,000 per QALY-gained. In probabilistic sensitivity analysis, we found that the probability that targeted screening for any risk indicators is cost-effective is 49% at a willingness-to-pay of $100,000 per QALY-gained. The literature is least clear on the role of marginal cord insertion, which is relatively common, as an aetiology for vasa praevia. More information about this potential risk factor may be necessary before routine transvaginal screening of women with marginal cord insertion should begin. The conclusion that universal screening of twin pregnancies is cost-effective was robust in sensitivity analysis.

The ability of transvaginal ultrasound screening to accurately rule out vasa praevia is critically important to the cost-effectiveness of a screening programme because of the risks to both the mother and the infant inherent in unnecessary early delivery by caesarean section. At specificity <99.95%, referring all women with any risk indicator to transvaginal ultrasound has an ICER > $100,000 per QALY-gained. Current reports indicate that the specificity of transvaginal ultrasound is >99.95%.5,6,8 Although we have not explicitly modelled it, in real clinical practice the sonographer and clinician would have several follow-up examinations at which they could identify instances in which the unprotected vessels have moved and no longer cross-over the internal os to prevent women from undergoing unnecessary hospitalisation or caesarean section.6

A recent survey of obstetricians in the UK found that only 80% would recommend caesarean section for vasa praevia, one-third could not name even one risk factor for the condition, and only 60% who perform obstetrical scanning consider themselves able to identify vasa praevia on transvaginal scan.24 Hence, substantial education may be required to ensure the high diagnostic accuracy that is possible5,6,8,25 and to ensure the appropriate patient management essential to minimising late fetal and neonatal deaths from vasa praevia.3 Our analysis indicates that even with a substantive one-time or ongoing investment in training and education, targeted screening for singleton pregnancies and universal screening for twin pregnancies is very likely to be cost-effective.

Even with a screening programme, not all cases of vasa praevia will be detected. Abdominal wall scarring, maternal body habitus and fetal position may prevent optimal visualisation of the cord insertion site. Vessels that course over the cervix in a transverse rather than anteroposterior direction may also be difficult to detect.8 Our analysis indicates that 7.5 affected singleton pregnancies and 0.9 affected twin pregnancies would remain undetected each year even if all risk indicators were used to indicate screening in singleton pregnancies and twin pregnancies compared with universal screening.

New advancements in imaging that improve the sensitivity and specificity for detecting vasa praevia and its risk factors will probably improve the cost-effectiveness of screening over time. Furthermore, advancements, such as the recent demonstration of identifying velamentous cord insertion using colour Doppler imaging at first-trimester ultrasound,26 may provide new screening strategies not considered in this study.

Limitations

The decision tree used in this analysis is large and complex involving many uncertainties. Because vasa praevia is a rare condition, not all inputs, in particular quality-of-life weights, are specific to the condition. We based inputs on the best data available, and when necessary on comparable health states and our personal judgments. To quantify the effects of the uncertainty around those judgments we performed extensive sensitivity analysis on all variables. We used Canada-specific data whenever they were available and therefore our results may not be generalisable to the health systems of other countries.

We did not model multiple pregnancies other than twins because they occur in very small numbers. We assumed that early fetal mortality was independent of vasa praevia in all pregnancies. Twin outcomes are not independent; however, in this model, we assumed that the fetal and neonatal mortality of the second twin was independent of the outcome of the first twin, which overestimated the number of mothers affected by the death of one infant. We did not consider costs or quality of life for fathers.

Colour Doppler may not be available in all centres; however, we have assumed an incrementally higher reimbursement that should fairly compensate centres for investing in the technology and for the longer examination time.

Pre-eclampsia and congenital heart defects can be identified using colour Doppler and cervical length can be measured at transvaginal ultrasound. We do not include the costs or benefits that may accrue because of these findings.

We modelled only some potential screening strategies based on six risk indicators. In theory, all 64 strategies based on the presence or absence of any combination of these factors could be modelled. The optimal screening strategy may involve a combination of risk indicators that we did not assess. Modelling all possible combinations would require knowledge about the correlations between all risk-factor combinations that are not known. Including more screening scenarios is unlikely to change the overall conclusion of the study that targeted screening of at-risk singleton pregnancies and universal screening of twin pregnancies is very likely to be cost-effective.

Conclusion

Initiation of expanded prenatal screening to include all twin pregnancies and at-risk singleton pregnancies would result in approximately 27.5 fewer late fetal and neonatal deaths each year in a population with approximately 132,000 pregnancies. Despite uncertainty as to which targeted screening strategy is optimal (specifically whether to include screening for marginal cord insertion), the probability that the status quo (not screening for vasa praevia) is the cost-effective choice is negligible. Among women with singleton pregnancies who already access ultrasound screening services, identifying those at risk for vasa praevia and referring them to transvaginal ultrasound screening has a cost-effectiveness ratio within the range considered ‘cost-effective’ in Canada.23 Therefore, a policy in which women with multiple gestation, low-lying placentas, IVF conception, accessory placental lobes, velamentous cord insertion, or marginal cord insertion are referred to transvaginal ultrasound to screen for vasa praevia should be considered for adoption.

Disclosure of interests

The authors report no conflicts of interest.

Contribution to authorship

Lauren E. Cipriano had the original idea for the research and contributed to study design, model construction, data acquisition, results analysis and interpretation, and drafting and revising the manuscript. William H. Barth Jr contributed to study design, data acquisition, results analysis and interpretation, and revising the manuscript. Gregory S. Zaric contributed to study design, data acquisition, results analysis and interpretation, and revising the manuscript. All authors have approved the final version.

Details of ethics approval

Not required.

Funding

This work was funded by the Natural Science and Engineering Research Council of Canada (NSERC).

Acknowledgements

The authors thank Dr Charles Botz and Randy Welch, London Health Sciences Centre, for their assistance in estimating costs and caesarean delivery rates. We also thank Dr Jenna Rawlins, Department of Obstetrics and Gynaecology, Windsor Regional Hospital, and Drs. Elizabeth Randle, Brenda Sohn and Cynthia Chan, from the Schulich School of Medicine and Dentistry at the University of Western Ontario, for their constructive feedback.

References

Journal Club

Discussion points

1. Background: Describe and discuss the current policy for screening cord abnormalities, including vasa praevia, in your area and the evidence, if any, to support it; compare with the policy in the authors’ area.

Discuss the rationale for economic evaluations and possible sources of data to inform them.

2. Methods: Evaluate the methods of this study using a published guideline and checklist.1 The same guideline explains the different types of economic analysis: which one have the authors used for this study?

Compare the methods of this study with those in a previously published economic evaluation, with particular reference to the source, accuracy and generalisability of the data.2 Also, discuss which other costs the authors of this current study might have included to better capture a societal (as opposed to health sector) perspective.

Discuss the use of assumptions for economic modelling and debate the ones in this manuscript, based on your experience. Is the sensitivity analysis (shown in Table S2) reassuring?

Have the authors considered all screening options, based on the possible combinations of risk factors? Discuss any alternative screening options that you might consider more appropriate/feasible for your area and explain why to your colleagues.

Debate the use of Quality-Adjusted Life-Years (QALYs, a measure that combines mortality and quality of life gains/outcome of a treatment, measured as the number of years of life saved, adjusted for quality) as outcome measure. What do you think about the fact that $50,000 is the (willingness-to-pay per QALY) threshold for considering a programme cost-effective?

3. Findings: Discuss the results of this study, the degree of uncertainty in the conclusions, and the possible implications for policy, practice and research, with particular reference to reproducibility and feasibility of both the intervention and the analysis.

DSiassakos
Southmead Hospital, Bristol, UK
Email jsiasakos@gmail.com

References

Ancillary