Evidence based well child careBMJ 2001; 323 doi: https://doi.org/10.1136/bmj.323.7317.846 (Published 13 October 2001) Cite this as: BMJ 2001;323:846
- Eugene Dinkevich, assistant professor of paediatricsa,
- Jordan Hupert, assistant professor of paediatricsb,
- Virginia A Moyer (), associate professor of paediatrics and epidemiologyc
- a Department of Pediatrics, State University of New York at Brooklyn, Brooklyn, New York, NY11203, USA
- b Division of General Pediatrics, University of Illinois at Chicago, Chicago, IL 60612, USA
- c Center for Population Health and Evidence Based Medicine, Department of Pediatrics, University of Texas at Houston, TX 77030, USA
- Correspondence to: V A Moyer
This is the first in a series of five articles
Your primary care paediatric practice has recently decided to review its preventive care practices before deciding which to include in a new computerised record system. You know that these practices vary considerably among group members, even as to how many check ups a child really needs. The value of some specific manoeuvres, such as the Adams forward bend test for scoliosis, for which adolescents are often referred from school, is doubted. You determine to find the best evidence for common preventive health interventions for children.
Fewer visits than in the standard schedules for children up to age 2 years are sufficient to detect physical abnormalities and psychosocial and developmental outcomes
Group care is as effective as individual care for routine checks
The Adams forward bend test is not accurate enough for screening for idiopathic scoliosis
Proving the value of check ups for healthy children and finding new and more effective ways to provide preventive care to all children remain major challenges
Routine checks on apparently healthy children are an important part of preventive services available for children. In developed countries outside the United States, paediatricians are trained to practise as hospital based specialists providing clinical care, while general practitioners and public health nurses are responsible for preventive care, including care of healthy children.1 In the United States, general paediatricians provide both preventive and clinical care and spend as much as 40% of their time checking healthy children.2
The major objective of these check ups is maintenance of health and prevention of disease. This is traditionally accomplished by repeated evaluations of healthy children under five heads: screening, health promotion, disease prevention, patient management, and follow up.
A routine check up includes history taking, physical examination, observation of parent-child interaction, and laboratory testing. All of these are forms of screening but little is known about their effectiveness at different ages. Guidelines developed by the American Academy of Pediatrics3 recommend the topics to review during history taking, but time is often short and topics that interest the paediatrician do not always interest the parents.4 The recommendations for screening physical examinations of the American Academy of Pediatrics, the Canadian Task Force on Periodic Health Examination,5 and the British Royal College of General Practitioners6 differ greatly. The American academy recommends a complete examination at each visit, while the Canadian and British agencies recommend only specific forms of physical examination on each occasion. The discrepancy between the recommendations and variability among practice styles are due to the scarcity of evidence linking the well child examination to measurable clinical outcomes.7
Health promotion and disease prevention include age specific counselling called anticipatory guidance. Although practices vary with different settings and paediatricians, a number of studies have attempted to measure the effectiveness of anticipatory guidance. Patient management and follow up are also important parts of well child care. There is a growing body of clinical evidence specifically on these components of the health supervision visit (see below).
The situation outlined above and the brief background overview suggest a number of questions about well child care and its effectiveness. You wish to use an evidence based approach, so you formulate three questions in a manner that maximises the yield from searching: each question includes the population, the intervention, and the outcome of interest. You look first for high quality systematic reviews and evidence based guidelines to answer your questions in one or more of the Cochrane Library, the Best Evidence database, Medline (Ovid), and PubMed Clinical Queries.
Question 1: In healthy children, does lowering the number of health supervision visits, compared with the standard recommended, result in adverse care outcomes?
Search: Cochrane Library: “well child care”; “well baby care”; “child health supervision”; Best Evidence: “well child care”; “well baby care”; “child health supervision”; Medline (Ovid): “number of well child visits”; “number of well baby visits”; “number of health supervision visits”
Question 2: Among normal full term babies, does group well child care offer advantages over individual well child care with respect to maternal-child interaction, child development, and health services utilisation?
Search: Medline (Ovid): “group well child care”; “group health supervision”
Question 3: In adolescents presenting for a routine examination, what is the diagnostic accuracy of the Adams forward bend test, compared to posteroanterior spine x ray with Cobb angle [the angle between lines drawn across the top of the uppermost and bottom of the lowermost of a group of vertebrae on x ray] determination, in diagnosing clinically significant idiopathic scoliosis?
Search: Cochrane Library: “scoliosis”; “cobb”; “spine”; Best Evidence: “scoliosis”; “cobb”; “spine”; PubMed: Clinical Queries→Diagnosis→Specificity: “scoliosis” and “forward bend”.
You find no reviews or citations in the Cochrane Library or the Best Evidence database. Your search of Medline yields eight documents for the first question, two of which seem pertinent. 8 9 The strategy for the search on group well child care brings up 12 citations, of which two look both relevant and methodologically sound. Two of the four studies of the Adams test are directly relevant to your question. You call the library and order all of the articles as well as the US Preventive Services Task Force report on scoliosis that one of your colleagues says he uses as the basis for not performing this test.
Summary of evidence
Frequency of health supervision visits
In a randomised controlled trial of healthy full term infants carried out in the United States from 1971 to 1973, investigators compared a schedule of three health supervision visits in the first year with the existing standard six visits (to either a paediatrician or a paediatric nurse practitioner). The three visit schedule included two additional visits to a nurse for immunisation, but no additional visits to a paediatrician or paediatric nurse practitioner.
The outcome variables included measurement of maternal knowledge of child rearing, maternal satisfaction with care, compliance with recommendations, and abnormalities detected or missed. A physician not involved in the main part of the study performed an independent physical examination at 15 months to detect any abnormality that might have been missed by the study paediatricians or nurse practitioners.
Of 297 babies enrolled, 246 (83%) completed the study. The reasons for withdrawal from the study were similar in the two groups. There were no significant differences between the two groups on any of the health measures evaluated by this study. Although these outcomes were only proxy measures of quality of care, and longer term outcomes are unknown, the authors concluded that the two schedules were equally effective in achieving objectives of well child care in the first year of life.
The other trial, in Canada, randomised healthy neonates to receive either 10 or five health supervision visits in the first two years of life. This more recent trial focused on traditional medical outcomes and on psychosocial and developmental outcomes, reflecting a shift in the objectives of well child care over several decades. Outcomes were measured with the Bayley scales of infant development and with the home observation for measurement of the environment (HOME) scale, which correlates with later school performance.10 In addition, the Hulka infancy questionnaire was used to assess maternal anxiety, and a standardised questionnaire was used to measure parental satisfaction with health care. This study also employed an independent physician to carry out a complete physical assessment of the children at the end of the study. Subjects were randomised to either the standard or the reduced visit groups.
The study was large enough to detect clinically important differences in the rate of undetected physical abnormalities between the groups. A total of 570 babies were enrolled and 466 (82%) completed the study. Dropout rates were similar in both groups. There were no statistically significant differences in the use of the emergency department, in any of the above outcome measures, or in the numbers of major or minor abnormalities found, and the study paediatricians had detected all abnormalities.
The results of the two studies were remarkably similar in that no clinically important differences were found between the children assigned to the reduced and the standard visit schedules. Because the authors used standardised assessments of development and home environment, you have confidence in the validity of the outcomes they chose to measure.
Group well child care
In group well child care, the provider facilitates discussion of child rearing issues with a group of parents of similarly aged children. Two randomised trials have compared the effectiveness of group and individual well child care for infants in the first year of life in households with middle and low incomes. Rice et al11 randomised patients in groups of four to assure similar ages for each well child care group, while Taylor et al12–14 randomised individual subjects. Study completion rate was 88% in the study by Rice et al, but 67% in the study by Taylor et al. Both studies used an intention to treat analysis, but owing to the nature of the study, neither the subjects nor the investigators were blind to the relevant intervention. In the study by Taylor et al, the same nurse practitioners provided care for both groups, so observer bias may have been introduced if the nurse practitioners treated the two groups differently. No report of concomitant interventions was given for either study, and the groups were similar in most respects at the beginning of the study. There were no significant differences between the groups in utilisation measures, maternal-child interaction, child development, or maternal outcomes. These two studies show that group well child care is as effective as individual care in low risk middle class and high risk socioeconomically disadvantaged families.
Adams forward bend test for scoliosis
The first of the two useful studies of the Adams test retrieved, by Cote et al, using a referral population at a university hospital, deals directly with your question.15 Two independent investigators examined 105 consecutively referred patients (87 girls) with a mean age of 15.5 (SD 4.8) years. All but two (with congenital scoliosis) had adolescent idiopathic scoliosis and 26 had already undergone some treatment for the condition. The gold standard for the diagnosis of scoliosis was a Cobb angle measurement of 20° on full spine x ray (determined by a third investigator). A positive forward bend test was defined as the appearance, to both examiners, of any trunk asymmetry.
The results show that a negative Adams test modifies the pretest probability significantly more than a positive test (table) You are concerned that the usual severity of scoliosis in your practice is likely to be quite different from that encountered in a referral clinic. An increased severity spectrum can change the likelihood ratios unpredictably, so likelihood ratios generated from a referred population should be applied to a primary care population only with caution.
The second study, by Goldberg et al, was designed to assess the conclusion of the US Preventive Services Task Force that no recommendation could be made either for or against screening for scoliosis (in particular, using the Adams forward bend test). 16 17 The study was carried out in primary and post-primary schools in Dublin, Ireland. Only girls aged 10-14 were included in this study because of very low incidence of clinically definite scoliosis in boys.18 Initial examinations with the Adams test were done at school; those who were positive were referred to the hospital scoliosis clinic for confirmation and, if appropriate, x ray examination. A substantially higher Cobb angle (≥40° at the time of diagnosis or subsequently) than in the study by Cote et al was used to define clinically significant scoliosis. Those who had negative results were followed up for four years. Of 8686 girls initially enrolled, 5179 (59%) were re-examined four years later. Only this cohort was used to assess the diagnostic characteristics of the screening test. As in the study by Cote et al, negative results by the Adams test were found to be more reliable for clinical purposes than a positive results (table). This was true even though the likelihood ratio for a positive Adams test was relatively high at 8.5. The key to understanding this apparent anomaly lies in considering both the prevalence of scoliosis in the study cohort and the severity of disease as defined by the investigators. The prevalence of curvature ≥40° in the Dublin school population who attended long term follow up was 0.1%. Given this low prevalence, patients with a positive Adams test would have a 1% chance of having significant scoliosis. It is doubtful that an increase in disease probability from 0.1% to 0.9% would cross a test or treatment threshold. In the case of a negative test, the upper end of the 95% confidence interval is not very different from 1. You conclude that the disease severity is likely to be similar to your setting and the Adams test does not appear adequate to confidently rule scoliosis 40° in or out.
Applying the evidence
At your next staff meeting, you report that you found little specific evidence for the overall effectiveness of well child care, and no evidence was found to support the current American Academy of Pediatrics recommendation for 20 visits by the 21st birthday. The two relevant studies concluded that a schedule with fewer visits had no detrimental effect on child health. In addition, group well child care was shown to be as effective as individual care. You also report that, for reasons of spectrum bias and small or imprecise likelihood ratios, neither of the two recent studies you reviewed about the Adams test provided sufficient evidence to recommend the test. You agree to review the literature periodically for newly published studies on this topic.
The search criteria given in these articles are intended to illustrate principles: they are not likely to be precisely replicable, as the literature is continually being updated. Readers interested in designing their own searches may find the explanatory chapter in the book from which this series has been taken useful.20
Well child care incorporates many screening tests (history and physical examination) and therapeutic interventions (anticipatory guidance, etc). Unfortunately almost no evidence is available to validate most of what goes to make up the health supervision visit. Recently, two Canadian physicians, Leslie and James Rourke, have attempted to develop an evidence based approach to well child care.19 The Rourke baby record, a health supervision guide for infants and young children, incorporates recommendations of the Canadian task force on periodic health examination which were based on the available evidence relevant to health screening in infants and young children. While much of the Rourke baby record is evidence based, much still has to rely on expert opinion. The record itself has yet to be evaluated in terms of its effects on clinically relevant outcomes