Objective. To assess the costs and benefits of various approaches to early detection of developmental disabilities.
Design. Cost-benefit analyses based on data from previously published studies of developmental screening tests.
Setting. General pediatric practices and day care centers.
Patients and Other Participants. A total of 247 parents and their 0- to 6-year-old children—103 from day care centers and 144 from pediatric practices.
Main Outcome Measures. Licensed psychological examiners administered a screening test of parents' concerns about children's development and one or two direct screening tests: the Denver-II and/or the Battelle Developmental Inventory Screening Test. For the day care sample, examiners also administered to each child measures of intelligence, adaptive behavior, and language. In the pediatric sample, children were administered additional assessments. At the same time, diagnostic measures were administered to a randomly selected subsample to make determinations about developmental status. Each screening method was evaluated for its short-term costs (administration, interpretation, diagnosis, and treatment) and long-term benefits (impact of early intervention on adult functioning as inferred from longitudinal studies by other researchers).
Results. When the long-term costs and benefits were considered, none of the approaches emerged as markedly superior to another. When viewing the short-term costs, the various screening approaches differed markedly. The use of parents' concerns was by far the least costly for physicians to administer and interpret.
Conclusion. Physicians can incur tremendous expenses when attempting to detect children with developmental problems. Although the benefits of early detection and intervention are substantial, physicians are not well-compensated for providing a critical service to society. Health policymakers and third-party payers must reconsider their minimal investment in early detection by health care providers. Nevertheless, our findings have encouraging implications for practice, because the use of parents' concerns as a screening technique offers substantial savings over and above other methods.
Approximately one in eight children have developmental disabilities. These include speech-language impairments, mental retardation, emotional and conduct disturbances, autism and related developmental disorders, physical and health impairments, and traumatic brain injury.1 All these disabilities interfere significantly with academic or life functioning. Early intervention with children who have special learning needs has been shown to improve family functioning, child behavior, and/or adult outcome, including socioeconomic status.2
Nevertheless, children's access to early intervention depends on early detection. Unfortunately, many disabilities are difficult to recognize; they are often subtle, rarely associated with dysmorphology or other obvious characteristics, or incompletely manifested during pediatric encounters. For this reason, the American Academy of Pediatrics' Committee on Children with Disabilities and Committee on Health Promotion recommend that physicians screen for developmental and emotional/behavioral problems at each of the 12 well-child visits routinely scheduled between 0 and 5 years.3,4 Physicians who attempt to do so, however, confront a vast array of screening and prescreening measures that vary considerably in costs and effectiveness.5 Little empirical evidence exists to guide physicians in their choice of identification methods.
We evaluate the costs of four approaches to early detection with developmental disabilities in pediatric settings. The analyses examine the costs of screening, of diagnostic evaluations, and of treatment for each of the following four approaches.
Parents' concerns. This approach relies on a 2-item questionnaire that elicits parents' concerns or appraisals of their child's developmental and behavioral status.
Direct screening. This approach relies on screening measures that use a combination of parental report and direct elicitation to assess children's skills in various developmental domains. This illustration examines two different screening devices, the Denver-II and the Batelle Developmental Inventory Screening Test (BDIST).
The third and fourth approaches use a combination of the above. Each detects probable developmental disabilities in two stages. In both approaches, parental concerns are assessed in the first stage. In the second stage, a formal screen is used to assess some children, depending on whether parental concerns were present.3 Two-stage positive screening. In this approach, children receive direct screening if and only if parents express a concern in the first stage. Only patients whose parents raise significant concerns and who have low score on direct screening, ie, who are positive on both measures, are referred for diagnostic evaluations and intervention if indicated. (These cases are marked with a check in Fig 1). In this approach parents' concerns are used as a prescreening technique through which a subset of patients in need of in-depth screening are identified.
4 Two-stage negative screening. In this approach, the parents' concerns questionnaire is again administered to all parents. In contrast to two-stage positive screening, direct screening tests are administered to all children whose parents' do not raise concerns. Children with positive results on either the parents' concerns questionnaire or direct screening are referred for diagnostic evaluations and intervention as indicated in Fig 2.
The costs of administration, interpretation, and evaluation still paint a rather incomplete picture. A more complete analysis of costs recognizes that each approach is likely to alter the time of treatment onset. Early detection moves children into treatment earlier, and the costs of this treatment represent a cost of early detection. However, the benefits of beginning treatment early are well-documented.2 Thus, ensuring that children receive needed treatment early in life may serve to reduce future treatment costs.
We examine each of these costs separately. This simplifies the discussion, but more importantly, the different costs are incurred by different individuals or stakeholders. The costs of administering and interpreting screens fall largely on physicians. The costs of evaluations as well as treatment can be borne by many sources (eg, schools/taxpayers, parents/third-party payors) depending on the type of testing and interventions needed. We explore the implications of each approach according to the costs incurred by the various stakeholders.
METHODS AND RESULTS
For greater clarity in presenting findings, the methods for each cost analysis are followed by the corresponding results.
Sampling and Screening
The data used in assessing costs were drawn from previously published studies6-9 and included two separate samples totaling 247 children between 0 and 6 years and their parent(s). The first sample included 103 families randomly selected from day care centers serving largely low-income children. The second sample was used to cross-validate the results and included 144 pediatric patients seeking well child care in teaching hospitals or private practices. In both samples, a licensed psychological examiner administered direct screening tests including (the Denver-II and/or the BDIST). A second examiner, blinded to the results of screening tests, elicited parents' concerns. Blinded to either concerns or direct screening results, examiners administered diagnostic tests to all children in the first sample (measures of intelligence, achievement, and language). In the second sample, a randomly selected subset of children was administered diagnostic measures and the remainder administered additional screening measures of academic skills, behavior, and articulation to facilitate the examiners' diagnostic impression. In both samples, diagnostic impressions or diagnostic test results were used to determine the extent to which each of the four approaches correctly identified those children with (and without) developmental disabilities.
Applying criteria for special education placement to the day care sample revealed that 19% of children had developmental disabilities and the remainder appeared to be developing normally. The ability of each of the four screening approaches to accurately identify children with and without disabilities is presented in Table 1. Ideally, the percentages of children accurately detected (true positives and true negatives) should be 80% or more.10Although none of the approaches meet both standards, some methods were close and were clearly more accurate than others.
Cost 1: Costs of Administration
The first of the four costs involves the time professional office staff (eg, a nurse, nurse practitioner, or physician's assistant) spend administering the screening tools. We assume that administering either of the two direct screens requires 30 minutes. We valued this time using an annual salary of $55 000 plus 54% overhead (that included the purchase prices of the screening tests and protocols).11 Assuming a 40-hour work week, this implies an hourly cost of $40.72. The half-hour spent administering the direct screening tests, therefore, can be valued at $20.36 (½ hour × $40.72/hour). Similarly, we assumed that administering the parents' concerns instrument takes 1 minute of professional time. This implies an administrative cost of $0.68 (Table2).
Calculating the administrative costs of the two-stage approaches is a bit more complicated. Under either approach, parental concerns are assessed for all children; this entails a minimum cost of $0.68. Some children receive direct screening as well. The fraction who do depends on the percentage of parents who express concerns about their child's development in the first stage. Under the two-stage positive approach, only children whose parents expressed concerns receive direct screening. As shown in the far right column in Table 1, 38% of children were positive on the parents' concerns measure (sum of true and false positives, N = 39). This produces an average cost of $8.39 [39 × $20.35 (cost of direct screening) + $.68 × 103 (costs of eliciting parents' concerns) divided by 103].
Under the two-stage negative approach, direct screens are administered only to those children whose parents had no concerns. As shown in the far right column in Table 1, 62% of children were negative on the parents' concerns measure (sum of true negatives and false negatives, N = 64). This group produces an average cost of $13.33 [64 × $20.36 + $0.68 × 103) divided by 103].
It is important to note that these costs do not depend on whether one uses the BDIST or the Denver-II but rather on the numbers of children whose parents raised or did not raise concerns. These costs also do not depend on whether the approach identifies the child as probably disabled or not. Such is not the case for the costs of interpreting the results of the various screening approaches as discussed below.
Costs 2: Costs of Interpreting Results for Each Detection Approach
Subsequent to any of the four screening approaches, the physician involved will have to interpret the results to parents. The time required depends on whether the screening approach indicates the child may or may not have a disability. We assume positive results require 15 minutes for interpretation; negative results, 5 minutes. We valued this time by calculating an hourly cost for pediatricians' time. This was based on an average salary plus overhead of $190 351 per year)11 and a 52-hour work week. This results in an hourly cost of $70.40.
Unlike administration costs, we assume that costs of interpreting the screening(s) depends on whether the results are positive or negative. Thus, the costs of interpreting the positive tests are $17.60 ($70.40/60 minutes × 15 minutes); negative tests costs $5.85 ($70.39/60 minutes × 5 minutes).
Each of the four approaches differ in the average costs of interpretation because of differences in the percentage of cases that are positive and negative. Table 3 shows the average interpretation costs for each approach according to the costs associated with the percentages of positive and negative cases.
Cost 3: Costs of Diagnostic Testing
Individuals identified as probably having developmental disabilities are referred for diagnostic evaluations by specialists (eg, a developmental psychologist, speech-language pathologist, developmental/behavioral pediatrician, social worker, occupational or physical therapist). We assigned an average cost of $800 to these evaluations.
The different approaches differ in the average costs of evaluation because they refer different percentages of children for evaluation. Under parental concerns, for example, 38% of children are identified as possibly having a developmental disability and are referred for evaluation. The average evaluation cost for children who are screened under this approach, therefore is $302.91 (38% × $800). We can calculate average evaluation costs for the other approaches in a similar fashion as shown in Table 3. As one would expect, those costs are highest for the approaches that refer the most children for evaluation. The two-stage negative approach identifies far more children as potentially disabled and so results in far more evaluations being conducted.
Including these costs changes the relative costliness of the different approaches. Although the cost of a single assessment is the same regardless of the approach used, the number of assessments given under each approach differs rather markedly and creates stark differences in the average cost of assessments. One can now see that parental concerns is no longer the least costly option. It is now more costly than the two-stage positive approach using either the Denver-II or the BDIST. This is because the only individuals who receive an assessment under the two-stage positive approach are those identified as potentially disabled by both parents' concerns and the BDIST or Denver-II. By creating two hurdles, the two-stage positive approach reduces the number of cases that are positive and thus lowers assessment costs. In contrast, the two-stage negative approach uses a much looser standard—individuals failing either the parental concerns questionnaire or in-depth screening now receive an assessment. Because of this many more assessments are given, and this raises the overall costs of the approach.
Cost 4: Costs of Treatment
The picture of costs thus far illustrated is still rather limited. Because the approaches differ in the number of children receiving assessments, they will differ in the number of children referred for treatment. This will mean that some children will receive treatment sooner than they otherwise would have; it may also mean that future treatment costs are reduced. Therefore, an assessment of the costs of the different approaches is not complete until the costs of treatment have been considered.
How will screening change the timing and intensity of treatment children receive? This depends on whether the child actually has a developmental disability and whether this is correctly identified. In essence the screening process divides children into four groups: false positive, true negative, false negative, and true positive. Note that we are able to separate true from false positives and true from false negatives because diagnostic evaluations were given to all children. Although this information would not be available in practice, it is essential to the calculations presented.
True negatives and false positives are correctly identified as not needing treatment and so receive none. These two groups differ in that only the false positives receive diagnostic evaluations. The other two groups, false negatives and true positives, comprise individuals who have a disability and require treatment. The two groups differ in whether their condition is identified (true positive) or not (false negative). We assume, based on efficacy studies of early intervention, that identifying early children with mild disabilities (eg, moderate speech-language impairments) reduces the intensity but may not eliminate the need for services later in life. For children with more substantial disabilities, early intervention produces modest improvements in adaptive, social, and behavioral skills that reduce the intensity and costliness of special services in adulthood (eg, early intervention students with mental retardation require substantial special education services through high school but as adults are more likely to either live independently or require less intensive and expensive group homes because they are better able to care for themselves).12 Concomitantly, later identification is assumed to require either more intensive or more extended services that are more expensive.13-15 To compare true positive and false negatives we assumed that false negatives would be identified 1 year later (because their problems would have intensified, increasing the chance of positive screening test results). Overall, these assumptions, although conservative, offer a basis for predicting the expenses associated with early vs late identification. It should be noted that these costs are estimates because children in the sample were not followed over time.
A second set of considerations in estimating treatment costs concerns the nature of children's disabilities. The types of treatment children are expected to receive depends on the type and degree of impairment. For example, we expect that a child with moderate to severe mental retardation requires lengthier and more intensive services than a child with a speech-language impairment. In viewing our sample, of the 19% with disabilities, subgroups of disabilities and the usual type of treatment are listed below. This represents a fairly typical incidence and distribution of disability types and treatments in relation to prevalence studies1,14,15 and census data from the Office of Special Education.16
Following are subgroups of disabilities and the usual treatment:
A total of 25% had moderate speech-language impairments and were expected to receive itinerant consultation services (ie, speech-language therapy and/or resource room services).
A total of 25% had severe speech-language impairments including learning disabilities and were expected to receive either itinerant consultation or self-contained special classes.
A total of 46% had mild mental retardation or other mild global developmental problems and were expected to receive self-contained or itinerant consultation during school. As adults, some will require a group home providing minimal supervision or other supervised living arrangement. (Group homes, of which there are typically several levels, have graduating costs in inverse proportion to each individuals' mastery of adaptive behavior skills, eg, dressing, toileting, personal hygiene, etc.17,18 More skilled individuals require less caretaking and hence cost less to serve).
A total of 4% had moderate to several mental retardation or other developmental disabilities of substantial severity such as autism. These students were expected to require self-contained special classes and as adults to not only need group home programs but programs with greater levels of supervision.
Again using the findings of early intervention and special education census data,2,12,13 we developed hypothetical treatment profiles for each of the four disability groups as shown in Table 4. These profiles differ depending on whether the child's disabilities was identified early or not and by the type of disability.
The costs of services in Table 4 vary depending on whether nondisabled peers would have or would not have entered or exited school. After matriculation and before graduation (between 5 and 18 years) costs of special services are in excess of educational costs that would have been incurred by other children, ie, $4485 for itinerant consultation and $6479 for self-contained. Before age 5 and after age 18, the expenses of itinerant consultation are $9469 and self-contained programs costs $11 463.16 For students requiring group homes, we assigned values of $17 676, $26 994, and $36 212 for three levels of group home services.17 We assumed that group home costs were entirely in excess of nondisabled peers, ignoring for simplicity any participation by nondisabled groups in federally funded programs such as Aid to Families with Dependent Children, Pell Grants for postsecondary education, etc.
Table 5 creates cost profiles and benefits for services received. Early intervention has conflicting effects on costs; intervening early increases costs as children receive services earlier. Those services, however, may reduce the need for services later, and this works to reduce costs.2,17,18 Cells marked with a “(+)” represent age/impairment groups where early intervention raises treatment costs; those marked with a “(-)”, represent where it lowers treatment costs.
Tables 4 and 5 make it apparent that treatment costs are spread over time. Because dollars in the future are worth less than dollars in the present, all costs must be converted into net present values. We did so using a discount rate of 4%. Total treatment costs and benefits are summarized on the bottom lines of Table 4 and compared with the costs of the average child in the sample.
Table 6 shows the costs associated with each screening model when incorporating the costs of treatment. The two-stage negative approach is somewhat less costly than other approaches.
Considering Costs and Benefits Under Other Sets of Assumptions
Because the analyses we presented are partially based on several key assumptions, it is important to consider how those assumptions affect our results. Four assumptions seem particularly important: the discount rate used to value future costs, the costs of diagnostic evaluations, the mix of disabilities among children, and the time frame used to consider treatment costs. Table 7 presents cost estimates under alternative assumptions. In panel A are estimates of total costs under alternative discount rates. We cut the discount rate in half (to 2%) and doubled it (to 8%). Doing so reveals two things. First, we can see that raising the discount rate lowers total costs (because it lowers the value of future costs). Similarly, lowering the discount rate raises total costs. Second, we can see that changing the discount rate does not alter the relative costliness of the different screening models. The two-stage positive screening approach is still the most costly and the two-stage negative screening approach is still the least costly. In panel B we raised the cost of the diagnostic evaluation in order to include occupational therapy, magnetic resonance imaging's genetic testing or other expensive procedures. This also has no effect on the results. In panel C we recalculated based on a population with fewer severe disabilities. The effect on the results was inconsequential. Panel D limits the focus to only those costs incurred through age 18. This produces a slight change in the results, but one which is consistent with other intervention studies; the benefits are not fully visible until adulthood.2
We cross-validated the results to our second sample (144 children from pediatric practices and teaching hospital general pediatrics clinics all of whom received the parents' concerns questionnaire and the BDIST, plus diagnostic evaluations and/or additional assessments). This population differed in terms of the accuracy of parental concerns. In the day care sample, parental concerns about children's health status was highly related to excessive developmental concerns. In the pediatric sample, this phenomena was understandably not observed probably because pediatricians had addressed medical issues. As a consequence, there was a significant improvement in the accuracy of parental concerns. This produced a substantial reduction in the costs for several of the approaches, as shown in Table 8.
We evaluated various approaches to early detection. We consider different cost criteria for evaluating the alternatives. Our results demonstrate that specific statistical properties of screening approaches (ie, sensitivity and specificity) should be considered in the context of the expenses they imply. Sensitivity and specificity mean little apart from the costs associated with the different outcomes. When the long-term costs and benefits of early detection and early intervention are factored together, none of the approaches emerge as markedly superior to another. Two-stage negative screening has a slight advantage and two-stage positive has a slight disadvantage but these differences may not be ecologically very meaningful. Given that the long-term costs and benefits associated with the various approaches are generally equivalent, the short-term costs take on greater significance. The various approaches varied substantially in terms of costs to physicians for both administration and interpretation. In the short-term, two-stage negative was more costly to physicians although the single-stage approach involving parents' concerns was by far the least costly. However, different capitation arrangements may effect the actual costs to physicians under each approach. If referrals for further evaluations are made within a health maintenance organization and physicians hold some financial responsibility for diagnostic work-ups, the two-stage negative approach is even more costly. The two-stage positive approach is less costly and the use of parents' concerns falls somewhere in between. If, in contrast, referrals are made to public schools or to outside agencies, the use of parents' concerns is the least costly to physicians. Thus, in making sense of the data presented, physicians will need to consider their referral practices and capitation arrangements when selecting an approach to developmental detection.
One question raised by the findings is how the benefits of early intervention affect the findings. If early intervention produces even more dramatic changes in adult outcome (eg, if individuals with mild mental retardation do not require supervised living arrangements), negative two-stage screening, with its substantially higher rates of sensitivity, continues to be the most cost-effective approach. If early intervention does not make a difference in the long-term costs of services, then the most cost-effective approach to early detection is for physicians to do nothing at all. However, the strength and volume of studies supporting early intervention suggest that this argument is specious.2,12,13 Further, we used substantially more conservative estimates than found in other early intervention research (eg, the Perry Project projected a life-time value for early intervention of $100 000 per student).2 Our more cautious figures reflect population differences: the Perry Project involved children at-risk for school failure due to impoverished environments and our sample included children with known disabilities as well as those who were environmentally at risk. Even so, the economic value of early detection and intervention was easily supported.
Implications For Policy
Society has made a tremendous investment in cradle-to-grave special education programs for individuals with disabilities. The cost-effectiveness of these programs—the extent to which they produce a cost-savings in dollars and human potential—is predicated on early intervention, which, in turn, is entirely dependent on early detection. Although formal child-find programs exist in most states under the Individuals with Disabilities Education Act, the onus and expense of child-find still falls on health care providers. Reimbursement for these expenses is minimal either through Medicaid's Early and Periodic Screening, Diagnosis, and Treatment (in which the average well-visit of which developmental screening is only one small part, is reimbursed at the national average of $33) or because many third-party payors do not cover well visits at all.19 This leads to a quite observable disjuncture between the costs to society and the costs to physicians. It is in society's best interest for physicians to detect children with disabilities. It is in physicians' best interest, at least financially, to do nothing at all toward early detection. Thus, if health care providers are to serve as an instrument of society, they should be reimbursed appropriately for the costs of implementing an important and effective social policy. Health policymakers and third-party payors must reconsider the currently minimal and short-sighted investment in the first critical step toward beneficial early intervention services: screening for childhood disabilities by primary health care providers.
Implications for Practice
Despite the fact that physicians can incur substantial costs when screening and interpreting results and can expect only limited reimbursement, the findings still provide some encouraging suggestions for early detection in primary care settings. The use of parents' concerns as a screening technique offers substantial short-term savings for physicians, at least in terms of administration and interpretation costs. In the long-term, two-stage negative appears slightly more beneficial and can be made less expensive to administer (and perhaps to interpret) by using in-depth screening measures that rely on parental report, such as the Child Development Inventories. These measures offer substantial relief from the costs associated with direct screening tests (such as the Batelle or the Denver-II) because they usually can be self-administered in waiting or examination rooms, sent home with families, administered via interview by office staff, etc.
This study was supported by the Joe Kennedy Research Foundation and Vanderbilt University Medical School, Department of Pediatrics.
- Received January 8, 1996.
- Accepted July 15, 1996.
Reprint requests to (F.P.G.) Division of Child Development, Department of Pediatrics, Vanderbilt University School of Medicine, 2100 Pierce Ave, Nashville, TN 37232–3573.
- BDIST =
- Batelle Developmental Inventory Screening Test
- Yeargin-Allsopp M,
- Murphy CC,
- Oakley GP,
- Sikes RK
- ↵Barnett WS, Escobar CM. Economic costs and benefits of early intervention. In: Meisels SJ, Shonkoff JP, eds. Handbook of Early Childhood Intervention. Cambridge, England: Cambridge University Press; 1990:560–582
- American Academy of Pediatrics, Committee on Children With Disabilities
- American Academy of Pediatrics, Committee on Practice and Ambulatory Medicine
- Glascoe FP,
- Martin ED,
- Humphrey S
- Glascoe FP
- Glascoe FP
- Glascoe FP,
- Byrne KE,
- Westbrook AG
- Glascoe FP,
- Byrne KE,
- Chang B,
- Strickland B,
- Ashford L,
- Johnson K
- ↵Lichtenstein R, Ireton H. Preschool Screening. Identifying Young Children with Developmental and Educational Problems. Orlando, FL: Grune & Stratton; 1984
- ↵Medical Group Management Association. 1993 Cost Survey of the Medical Group Management Association. Englewood, CO: Medical Group Management Association; 1993
- Ramey CT,
- Ramey SL
- Ramey CT,
- Bryant DM,
- Wasik BH,
- Sparling JJ,
- Fendt KH,
- LaVange LM
- Algozzine B,
- Korinek L
- Halgren DW,
- Clarizio HF
- ↵Council for Exceptional Children. Statistical Profile of the Special Education in the United States (from the 15th Annual Report to Congress). Reston, VA: Council for Exceptional Children; 1993
- Blackorby J,
- Wagner M
- McManus M,
- Flint S,
- Kelly R
- Copyright © 1997 American Academy of Pediatrics