|
Neonatal hearing screening: modelling cost and effectiveness of hospital- and community-based screening |
Grill E, Uus K, Hessel F, Davies L, Taylor R S, Wasem J, Bamford J |
|
|
Record Status This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn. Health technology The study examined two strategies for universal neonatal hearing screening. Under community-based screening, newborns were seen by health visitors at home, usually at 10 days of age. Under hospital-based screening, babies were screened in the maternity hospital prior to discharge, with follow-up of missing cases in a variety of ways.
Economic study type Cost-effectiveness analysis and cost-utility analysis.
Study population The study population comprised a hypothetical cohort of newborns.
Setting The settings were the community and a hospital. The economic study was carried out in the UK.
Dates to which data relate Some effectiveness and resource use data were derived from studies published between 1986 and 2005. The price year was not stated.
Source of effectiveness data The effectiveness evidence was derived from a synthesis of published studies and authors' opinions.
Modelling A modified version of a published Markov model was used to model the clinical and economic outcomes in a hypothetical cohort of 100,000 newborns. The end points were assessed at 6 and 12 months. The time horizon of the model was 120 months with monthly cycles. The health states considered were:
unknown status;
healthy (hearing) confirmed by diagnostic test or screening (true negative);
healthy (hearing) not confirmed by diagnostic test;
hearing impaired confirmed by diagnostic test or screening (true positive);
thought to be healthy (hearing) but hearing impaired (false negative);
thought to be hearing impaired but healthy (false positive); and
not followed up or not compliant.
Outcomes assessed in the review The outcomes estimated from the literature were:
the prevalence of newborn hearing impairment,
the sensitivity and specificity of screening,
screening coverage, and
the probability of "natural" discovery without systematic screening.
Study designs and other criteria for inclusion in the review The clinical data came from multiples sources, including a systematic review of the literature (details of which can be found at the following url: http://gripsdb.dimdi.de/de/hta/hta_ berichte/hta063_ bericht_ de.pdf& lt;/ url>) and empirical data derived from the English sites that participated in the NHSP in 2001. Limited information on the other sources of clinical data was provided.
Sources searched to identify primary studies Criteria used to ensure the validity of primary studies Methods used to judge relevance and validity, and for extracting data Number of primary studies included The clinical estimates used in the decision model appear to have been derived from 8 studies.
Methods of combining primary studies Some data, namely positive predictive values, were pooled using a random-effects model because of the heterogeneity of study sites. The approach used to combine the other data was not described.
Investigation of differences between primary studies The authors stated that some data were heterogeneous, thus a heterogeneity test was presumably carried out.
Results of the review The prevalence of newborn hearing impairment was 0.15 (range: 0.09 to 0.3).
The sensitivity of screening was 96% (range: 96 to 99) and the specificity was 99%.
The rate of screening coverage was 97%.
The median age of "natural" discovery without systematic screening was 18 months.
Methods used to derive estimates of effectiveness The authors made some assumptions, based on expert opinion, that were used in the decision model.
Estimates of effectiveness and key assumptions Each month detected before the age of 6 months was assigned a weight of 1.
If not detected within the first 12 months, profoundly and severely impaired children were assigned a weight of 0.85 and moderately impaired children a weight of 0.90.
Assuming that 50% of the children with permanent congenital hearing disorders were moderately impaired gave a weight of 0.875 for every month detected after the first birthday.
The incidence and prevalence of congenital permanent bilateral hearing loss were assumed to be equal.
Measure of benefits used in the economic analysis The summary benefit measures used were the true positives and the quality-weighted detected child months (QCM). These were estimated using the decision model. An annual discount rate of 1.5% was applied to future months gained. The false positives after screening were also reported.
Direct costs The perspective taken in the study was not explicitly reported, but it appears to have been that of the health service payer. The cost categories included in the cost analysis were staff, consumables, overheads and equipment, staff travel and staff training. Details on the cost calculation were reported, but the unit costs were generally not presented separately from the quantities of resources used. Most of the data on costs and resource use were derived from the English sites participating in the 2001 NHSP. Additional cost information was obtained from NHS salary scales, the NHS Rehabilitation Services Catalogue (screening equipment and consumables) and the Medical Research Council Institute of Hearing Research (other costs). Other information on resource consumption was based on authors' assumptions, which were explicitly stated. Discounting was relevant and an annual rate of 6% was applied. The price year was not reported.
Statistical analysis of costs The costs were treated deterministically in the base-case.
Indirect Costs The indirect costs were not included in the economic evaluation.
Sensitivity analysis Univariate and multivariate sensitivity analyses were carried out on all model inputs to assess the robustness of the costs and benefits to variations in clinical and economic data. The ranges used were derived either empirically or from the literature. A Monte Carlo simulation was also performed and stochastic distributions were given to each model input. Given that, in the base-case, test accuracy and other efficacy data were assumed to be identical for hospital- and community-based screening, an analysis of extremes was performed in which prevalence of hearing impairment, and the sensitivity and specificity of screening were assumed to be different between the two settings. This analysis was carried out to show the probability that one setting is more cost-effective than the other, and to indicate which difference in a certain parameter between sites might result in substantial differences in the costs.
Estimated benefits used in the economic analysis In a hypothetical cohort of 100,000 babies, there were 134 true positives at 6 months and 150 at 120 months with both screening options. The number of false-positives after screening was 12 for both screening options.
In a hypothetical cohort of 100,000 babies, the number of QCM was 794 at 6 months, 1,536 at 12 months and 13,751 at 120 months with both strategies.
Thus, in the base-case, the two strategies were equally effective in terms of cases detected and QCM.
Cost results The total costs in a hypothetical cohort of 100,000 children at 120 months were 3,690,022 with hospital-based screening and 3,343,572 with community-based screening. The cost-difference (346,450) was not statistically significant.
Synthesis of costs and benefits Average cost-effectiveness and cost-utility ratios were calculated to combine the costs and benefits of the two screening strategies. An incremental analysis was not performed since no differences between the groups were observed in any of the benefit measures.
The average cost per detected child was 25,813 with hospital-based screening and 23,390 with community-based screening. The average cost per QCM was 268 with hospital-based screening and 243 with community-based screening.
The univariate sensitivity analysis showed that the costs were sensitive mainly to changes in the prevalence rate. At rates higher than those used in the base-case, the cost-savings associated with community screening increased.
The Monte Carlo simulation suggested that the costs would be lower in hospital sites in 48% of the simulations, while the costs would be lower in community sites in 52% of simulations. Thus, the results were quite similar.
The extremes analysis showed that the results of the analysis were highly dependent on the base-case assumptions. For example, if prevalence in hospital sites was higher than in community sites, the probability of hospital being more cost-effective would be 95% if decision-makers were willing to pay at least 500 per QCM gained. If the willingness to pay was below 30 per QCM, community sites would be more cost-effective with a probability of 95%. However, if prevalence in community sites was higher than in hospital sites, community sites were more cost-effective for any ceiling ratio. Similarly, if decision-makers were willing to pay at least 300 per QCM gained, the probability of the hospital being more cost-effective under the assumption of higher sensitivity in the hospital would be 95%. If the willingness to pay was below 150 per QCM, community sites would be more cost-effective with a probability of 95%.
Authors' conclusions The two screening strategies were equally effective and similarly expensive. The sensitivity analysis showed that any statistically significant difference in prevalence, sensitivity, specificity and costs would result in significant differences in cost-effectiveness between the settings.
CRD COMMENTARY - Selection of comparators The rationale for the selection of the comparators was clear. The two screening options were selected because they were actually implemented in the NHSP. You should decide whether they are valid comparators in your own setting.
Validity of estimate of measure of effectiveness The effectiveness evidence was derived from published sources. The authors stated that details of the review were reported in another publication and limited information on the primary studies was reported in this paper. Full details of the literature review can be found at: http://gripsdb.dimdi.de/de/hta/hta_ berichte/hta063_ bericht_ de.pdf.
The issue of homogeneity of the primary studies was explicitly addressed with respect to some estimates. Some assumptions were also made, which might have introduced further uncertainty in the model. Extensive sensitivity analyses were, however, performed.
Validity of estimate of measure of benefit The summary benefit measure was specific to the disease considered in the study, thus it would be difficult to compare it with the benefits of other health care interventions. The impact of the screening strategy on quality of life was considered, which strengthens the validity of the benefit measure. The authors noted that empirical data on quality adjustments required to assess quality-adjusted life-years were not available.
Validity of estimate of costs The perspective adopted in the study was not stated clearly, but it might have been that of the NHS. Only the direct medical costs were considered in the analysis. The source of the data was reported for most items. A breakdown of the costs was not provided, thus limiting the possibility of replicating the results of the analysis in other settings. Further, the unit costs were not provided separately from the quantities of resources used. No statistical analyses of the costs were carried out in the base-case, but probabilistic distributions were given to economic inputs. All economic inputs were varied in the sensitivity analysis. The price year was unclear, which makes reflation exercises in other time periods difficult.
Other issues The authors stated that comparisons with other studies were difficult, but the projected costs per detected child were comparable with the costs found in another study. The issue of the generalisability of the study results to other settings was not explicitly addressed, although extensive sensitivity analyses were carried out. These enhance the external validity of the analysis. The authors noted that a possible limitation of the analysis was the lack of robust clinical estimates with which to populate the decision model. This introduced uncertainty into the model.
Implications of the study The study results suggest that a further evaluation of some model inputs, including prevalence data, would help draw clear conclusions from the comparative analysis of the two screening settings.
Source of funding Funded by the UK Department of Health.
Bibliographic details Grill E, Uus K, Hessel F, Davies L, Taylor R S, Wasem J, Bamford J. Neonatal hearing screening: modelling cost and effectiveness of hospital- and community-based screening. BMC Health Services Research 2006; 6(14) Other publications of related interest Grill E, Hessel F, Siebert U, et al. Comparing the clinical effectiveness of different newborn hearing strategies. A decision analysis. BMC Public Health 2005;5:12.
Davis A, Hind S. The newborn hearing screening programme in England. Int J Pediatr Otolaryngol 2003;67 Suppl 1:S193-6.
Keren R, Helfand M, Homer C, et al. Projected cost-effectiveness of statewide universal newborn hearing screening. Pediatrics 2002;110:855-64.
Indexing Status Subject indexing assigned by NLM MeSH Community Health Services /economics; Cost-Benefit Analysis; England; Female; Health Care Costs; Hearing Disorders /congenital /diagnosis; Hearing Tests /economics /methods; Hospital Costs; Humans; Infant, Newborn; Male; Markov Chains; Neonatal Screening /economics /methods; Outcome and Process Assessment (Health Care) /economics; Predictive Value of Tests; Wales AccessionNumber 22006000632 Date bibliographic record published 31/08/2006 Date abstract record published 31/08/2006 |
|
|
|