The mother-child interaction and clinical judgment during acute pediatric illnesses

~!i!!iii~i!i!!i ~

e mother-child interaction and clinical judgment g acute pediatric illnesses

Paul McCarthy, MD, Kimberly Freud~gman, PhD, Domenic Cicchetti, PhD, Linda Mayes, MD,

Jorge Lopez Benitez, MD, Sadek Salloum, ~D, Michael Baron, MD, Howard Fink, AID,

Robert Anderson, MD, and Robert LaCamera, AID

Objectives: For acutely ill children living in less than optimal environments, mothers and pediatricians may have a heightened perception of illness severity, a lower specificity of clinical judgments, and a tendency to over-utilize resources. We examined the mother-child interaction in order to understand the

relation of less optimal environments to clinical judgment and resource use.

S tudy design: At the 2-week and 6-, 15-, and 24-month well child visits of 516 children, the mother-well child interaction was assessed by using the Biringen's Emotional Availability Scales (EAS). Data were gathered regarding maternal depression and sense of competence, infant temperament, maternal social support, life events, the home environment, and demographics. At ill visits, the mother-i//child interaction was assessed by using the EAS, and mothers and pediatricians independently assessed illness severity using the Acute

Illness Observation Scales. Resource use during the illness was evaluated.

Results: One thousand nine hundred eight-three acute illnesses were assessed. A less optimal mother-child interaction was significantly (P < .05 for all comparisons) associated with poorer reliability of mothers' judgments, lower specificity of mothers' judgments (71% vs 85%) and pediatricians' judgments (92% vs 97%), and greater use of resources (eg, for hospitalizations, 2.6% of visits vs 0.7%). Adverse maternal, infant, and demographic characteristics were associated with a less optimal mother-well child (r = 0.68)

and mother-ill child (r = 0.80) interaction, a heightened perception of illness severity, and greater resource use.

Conclusion- Less optimal environments adversely affect the mother-child interaction; a poor mother-child interaction is correlated with low specificity of clinical judgment and over-utilization of resources. (J Pediavr 2000; 136:809-17)

From The Department of Pediatrics and the Yale Child Study Center, Yale University, School of Medicine, and Yale-New Haven Children's Hospital, New Haven, Connecticut. *The following pediatricians and nurse practitioners also participated in the study: Angela Crowley, PNP, Mary Ann Davidson, PNP, Janet Geiger, MD, Liesel Gould, MD, Karen Haddad, PNP, Margaret [keda, AID, Melinda Mahabee-Gittens, MD, Cynthia Mann, AID, Deborah Navedo, PNP, and Carter Stilson, MD. Supported by Grant No. RO1HD26575-01 from the National Institute of Child Health and

Human Development. Submitted for publication Mar 17, 1999; revisions received Aug 11, 1999, and Jan 10, 2000; accepted Jan 31, 2000. Reprint requests: Paul L. McCarthy, MD, Department of Pediatrics, Yale University School of Medicine, 333 Cedar St, New Haven, CT 06520-8064. Copyright © 2000 by Mosby, Inc. 0022-3476/2000/$12.00 + 0 91211106224 doi: 10.1067/mpd.2000.106224

Adverse maternal characteristics (eg, higher levels of anxiety) and social milieu (eg, living in the inner city) are associated with less reliable judgments by mothers about the severity of acute illness in their children, with perceptions by mothers and pediatricians that an illness is more severe leading to lowered specificity of clinical judgments and greater use of health resources during the illness (unpublished data). 1'2

We hypothesized that a key to Under- standing the effects of these adverse characteristics is the mother-child interaction. Such adverse characteristics powerfully affect the mother-well child interaction. 3"17 Domains fundamental to the mother-well child interaction-- crying, response of that crying to parent comforting, providing stimuli to main- tain a child in an optimal functional state J a r e also fundamental to assessing the degree of illness severity in a child with an acute illness. 18"20 A cross- age continuity has been reported for the mother-well child interaction. 21 It is reasonable to believe that such a continuity can also be documented between the mother-well child and mother-ill child interactions. These considerations

809

MCCARTHY ET AL THE JOURNAL OF PEDIATRICS JUNE 2000

led to the following conceptual framework. Impoverished environments adversely affect both the mother-well child and mother-ill child interaction, leading to a less optimal interactive state during illness. This, in turn, re- suits in the child being perceived as more ill, consequently leading to poorer specificity of clinical judgments and greater use of health resources. This conceptual framework allows applica- tion of techniques used to assess the mother-well child interaction to the mother ill child interaction and examination of the effect of those maternal, infant, and social milieu characteristics that powerfully affect the mother-well child interaction on the mother-ill child interaction and, hence, clinical judgment. If more powerful determi- nants of clinical judgment and resource use can be identified, then intervention strategies can be focused more appropriately.

STUDY HYPOTHESES Hypothesis I

The mother-well child interaction will be correlated with the mother-ill child interaction.

A less optimal mother-well child interaction will be correlated with a less optimal mother-ill child interaction.

Hypothesis H A less optimal mother-well child and

mother-ill child interaction will be correlated with: (1) poorer reliability and specificity of mothers' judgments during illness, (2) poorer specificity of pediatricians' judgments during illness, (3) more test ordering by pediatricians during illness, and (4) greater use of medical resources by pediatricians and by mothers during illness.

Hypothesis III Adverse maternal (depression, par-

enting sense of competence), infant (temperament), and social milieu (quality of home environment, social sup- ports, life events) characteristics will be

associated with a less optimal mother- child interaction when the child is well and when the child is ill, and hence, with poorer reliability and specificity of mothers' judgment and poorer specific@ of pediatricians' judgment during illness and greater use of tests and resources during illness.

METHODS Study Population and Design

Patients were enrolled at a medical center primary care center, the urban private practice of R.A. and R.L. (A- L), and the suburban private practice of M.B. and H.F. (B-F). These sites were chosen to ensure that the study participants represented a broad range of sociodemographic characteristics. Previous experience indicated, for example, that mothers in the PCC versus those in a suburban practice would be younger, less likely to have college de- grees, and more likely to be receiving federal assistance. 1'2 Study participants gave informed written consent, and the study protocol was approved by the Human Investigation Commit- tee of the Yale University School of Medicine.

Mother-Child Interaction At the 2-week and 6-, 15-, and 24-

month well child visits, pediatricians or nurse practitioners scored 3 domains (Maternal Sensitivity, Maternal Hos- tility, and Child Responsivity) of Birin- gen's Emotional Availability Scale, a measure of the mother-child interaction. 22"24 EAS scoring was done after a 2-minute observation period (during which the mother played with the child), after the history and the physical examination. Pediatricians and nurse practitioners were trained to score the EAS reliably (see Results), In order to test the convergent validity of the EAS scoring, videotapes of the well child care visit were made in a 20% subsample (videotaped cohort) and scored by the method of Bornstein by a blinded observer. 2s'26

Maternal, Infant, and Social Milieu Characteristics

Within 2 weeks of the well child care visit, a child developmentalist (K.E), blinded to EAS data, visited the home and scored the Caldwell H O M E Scales 2z and the Supplemental H O M E for Impoverished Familles. 28 Mothers filled out the Beck Depression Invento- ry, 29 the Parenting Social Support Index, 3° the Rothbart Infant Tempera- ment Questionnaire, 31 Parenting Sense of Competence Scale, a2 and a revised version of the Life Events Question- naire of Abidenaa; demographic data were recorded. A research associate gathered and collated these latter data completed by mothers; hence, the in- vestigators were unaware of these results. Pediatricians were also unaware of H O M E and SHIF data.

Clinical Judgment and Resource Use

During acute illness visits to the pediatrician, the mother and the pediatrician (or nurse practitioner) observed the child for 2 minutes and then independently scored the Acute Illness Observation Scales. 2° The AIOS are composed of 6 items, which assess by observation an ill child's degree of toxi- city; adequate levels of reliability and validity of the AIOS have been demonstrated. 1,2,20 After the observation period, history, and physical examination, the pediatrician scored the 3 domains of the EAS. Tests ordered, prescriptions written, follow-up appointments made, and hospital admissions were recorded on a standardized form. Mothers were provided a teaching videotape of the AIOS. For the videotaped cohort, ill visits were videotaped and analyzed by the method of Bornstein. Data about test ordering by pediatricians during the illness, prescriptions written, follow-up appointments made, and hospitalizations were gathered 2 weeks after the ill visit by review of standardized forms, phone logs, patient charts, and hospital medical records. Mothers' use of the emergency department was also

810

THE JOURNAL OF PEDIATRICS VOLUME 136, NUMBER 6

MCCARTHY ET AL

analyzed in this review. All participants in the study, with the exception of the principal investigator (EM.), were blinded to study hypotheses.

Data Analytic Techniques

HYPOTHESIS I. The correlation of the mother-well child EAS total score to the mother-il l child EAS total score was studied by using Pearson r correlation with re calculation. The re value, the proportion of variation explained in the outcome by the predictor variables, is derived by squaring the r correlation and multiplying by 100%. The re values of 1% to 8%, 9% to 24%, and 25% to 48% indicate, respectively, a small, medium, and large explanation of variation in the outcome. 34 For this analysis, the correlation of the mean of the EAS total scores for the individual mother-child dyad at all of their well visits to the mean of EAS total scores at all ill visits for the same dyad was studied. Thus the mother-infant dyad is the unit of analysis.

HYPOTHESIS II. Mothers ' reliability of AIOS scoring, that is, level of agreement with the pediatrician, was assessed by using the weighted kappa statistic. The percent of ill visits in which the z of kappa was above the median were then compared when the ill EAS total score was below the median. 55 A z of kappa above the median indicated perfect agreement.

The specificity of mothers' and pediatricians' AIOS scoring was calculated by constructing 2 x 2 tables. A positive test result was an AIOS score > 10; a negative test resuk was a score _<10.1'2'20 In the 2

x 2 tables, test positivity or negativity was assessed in relation to whether a serious illness was present. The definition of a serious illness 18'20'36'37 was the pres-

ence of a significant positive laboratory test result including isolation of bacteria from a normally sterile body fluid, isolation of a pathogen from stool, aseptic cerebrospinal fluid pleocytosis, pneu- monia demonstrated by chest x-ray

film, hypoxemia during a lower respi- ratory tract infection, or abnormalities of serum electrolytes, usually during acute gastroenteritis. Of children with a significant positive laboratory test result during an acute illness, 97% to 98% will have one or several of the above defined abnormalities. 1

Specificity is the proportion of patients who do not have a serious illness (ie, who do not have a significant positive laboratory test result) who are scored as test negative (are given an AIOS score <10). If patients who do not have a serious illness are perceived as having a serious illness and thus scored as test positive (are given an AIOS score > 10), then specificity is re- duced; it was hypothesized that a less optimal mother-child interaction would have this effect on perception of illness severity, thus reducing specificity.

Specificity was examined when EAS total scores during ill visits were below the median. In this analysis, unlike that in hypotheses I, the ill visit is the unit of analysis.

The use of the following resources was examined during itl visits when the ill EAS score for that visit was below the median: (1) diagnostic tests, (2) prescriptions, (3) hospitalizations, and (4) scheduled follow-up visits. For this analysis, each of these variables was scored as present or absent. A similar analysis was done when the well EAS total score temporally closest to the ill visit was below the median.

Mothers ' use of the ED was examined only in the PCC cohort. In this cohort, the ED at the Yale-New Haven Children's Hospital was readily available. To examine differences in reliability (z of kappa), specificity, and use of resources when EAS total scores (both well and ill) were below the median, the ;(2 statistic was used.

HYPOTHESIS III. Results relating to hypothesis III were studied by using multivariate techniques. 38 The independent variables were (1) total score or, when appropriate, subscale scores for

life events and maternal, infant, and social milieu characteristics and (2) demographic data including maternal age, years of education, employment, father living in the home, parity, federal assistance, race/ethnic group (African American, white, Hispanic, other), and sex. Analysis of intercorrelations among independent variables revealed that demographic data were intercorrelated (eg, maternal education and employment), as were Caldwell H O M E and SHIF data. Regarding other maternal, infant, and social milieu data, and as in our previous research, 2° a minimal uni- variate r value between a given pair of predictor variables of beyond 0.60 was used as a criterion for deleting that variable in the pair that had a lower correlation with the dependent variable; none of the paired r values met this criterion for collinearity, so that none of these maternal, infant, and social milieu predictor variables required deletion.

The outcomes sequentially examined were (1) AIOS scores of mothers and pediatricians and (2) use (present, absent) of tests, prescriptions, and hospitalizations and recommendation of follow-up visits. Mothers ' use of the ED (quantitated as 0, 1, 2, 3, 4, 5, and ->6 visits) in the PCC cohort was also examined. In the multivariate analyses, the Pearson r statistic was used with r e calculation. This is a multiple regression technique, available as part of data analytic capacity of SAS software (SAS Institute, Cary, NC). This technique, known as r-squared, evaluates and presents in rank order the following information: the ordered r-squared values of each of the 1~ input variables, taken alone; followed by the actual, or best (meaning correct), set of ordered input variables, taken two at a time; then followed by the ordered set of the best triad of input variables, with the process proceeding up to the presentation of the r-squared value of all 1~ predictor variables taken together. In order to distinguish meaningful from trivial values of r-squared, for any com- bination of predictors (eg, the best set

811


Table/. Sociodemographic characteristics*

of 10 input variables), the process was terminated when a given number of predictors failed to consistently add at least an additional 1% to the value of r- squared. This process of termination is consistent with the method of Cohen. 34 For this analysis, mean scores for maternal, infant, social milieu, and life events data for each mother-infant dyad were used for the predictor variables; the correlations of these mean predictor scores of each mother-infant dyad to the data for each ill visit for the same dyad was then studied.

RESULTS Study Population

The study began enrolling mothers in February 1995 and was completed

in March 1998. For the total study sample, 380 mothers were asked to participate, 316 enrolled, and 226 (72%) remained at study completion; these proportions did not vary significantly by s tudy site. Numbers enrolled at each study site were: PCC, 151; A-L, 111; and B-E 54. Data gathered from patients who subse- quently dropped from the study were used in the analyses of study hypotheses. Sixty-three mother-lnfant pairs were enrolled in the videotaped cohort.

Table I outlines demographic data of the cohorts at each study site. These data indicate that the study populations at the 3 study sites differed and represented a spectrum of sociodemographic characteristics.

Reliability and Convergent Validity of EAS Scoring

The EAS scoring of study pediatricians and nurse practitioners on a sample of well and ill children not enrolled in the study was reliable as compared with the scoring of K.F. (weighted kappa, KV¢, was 0.61, 0.63, and 0.71 for Maternal Sensitivity, Maternal Hostili- ty, and Infant Responsivity, respectively, which represents good chance cor- rected agreement)fi 9 Further, EAS scoring by pediatricians at the 6-month well child visit had high intercorrelations with scoring by the method of Bornstein of videotapes recorded at the same visit when the 2 methods were compared by a blinded observer. 4° Fi- nally, EAS scores (all 3 domains of the EAS and total score) at well visits of all 316 patients were highly correlated with pediatricians' global assessment of the mother-child interaction scored at the same visit on a 5-point Likert scale. 41

Frequency of Scoring Well EAS, Maternal, Infan~ and Social Milieu Characteristics and Ill Visits: Extent of Health Resource Use

The mother-well child interaction in the study population was assessed by using the EAS on 937 separate occa- sions. The mean number of EAS assessments for each mother-infant dyad was similar at the 3 study sites: PCC, 2.9; A-L, 3.0; and B-F, 3.2.

For the 316 patients, there was a mean of 2.4 home visits for the Cald- well H O M E / S H I F assessment for each mother-infant dyad. The total number of completed instruments for maternal, infant, and social milieu characteristics for 316 patients were: BDI, 806; Parenting Sense of Competence, 776; RIBQ, 803; Parenting Social Sup- port Index, 800; and Life Events Ques- tionnaire of Abiden, 789. The mean number of instruments completed per mother-infant dyad was higher in the suburban B-F practice than in the PCC or urban A-L practice.

Table II outlines the mean scores of maternal, infant, and social milieu char-

812

TNE JOURNAL OF PEDIATRICS VOLUME 136, NUMBER 6

MCCARTHY El- AL

acteristies at the 3 study sites. The

mean scores were derived by determin- ing the mean of each individual sub-

ject's scores and then taking a mean for each of the 5 study sites. PCC mothers as compared with mothers at the private practice sites had lower, less optimal scores on the H O M E and S H I F

and higher, less optimal scores on the BDI. There were no meaningful differ-

ences in mean scores of other characteristics among the 3 study populations.

There was a total of 2444 illness

episodes in the cohort . O f this total, 1985 (81%) were assessed in the s tudy protocol. The mean number of ill visits per patient assessed was quite similar across the 5 sites: PCC, 6.4; A-

L, 5.5; and B-E 6.3. The EAS and A IO S were completed by the s tudy pediatr ic ians in 1958 (98%) of the

1983 illness episodes. Mothers completed the A I O S in 1885 (96%) of these episodes.

In these 1985 ill visit episodes, 50 serious illnesses, meeting the criteria out-

lined in the Methods section, occurred. Regarding medical resource use, in

1426 of the 1983 visits at least one prescription was given, in 556 at least one diagnostic test was done, in 34 a hospital admission occurred, and in 522 follow-up appointments were made.

Analysis of Study Hypotheses

HYPOTHESIS I. The correlat ion of the mother-wel l child interaction to the

mother-il l child interaction was inves- tigated. The Pearson r correlation was

0.65 (r ~ 42.2%). Although mean well EAS scores did not differ by age, a sec- ond correlational analysis was done, which controlled for age and in which the ill visit, ra ther than the mother- infant pair, was the unit of analysis. The Pearson r correlat ion of the ill EAS score during a specific ill visit to the well EAS score of that mother- infant dyad that was scored temporally closest to the ill visit was studied. The r correlat ion between the well and ill EAS scores was 0.44.

Table II. Mean scores of maternal, infant, and social milieu characteristics by study site

TablelII. Relation of ill EAS score above and below the median to selected measures

Thus the mother-wel l child interaction and the mother- i l l child interac-

tion are correlated, and for individual mother-infant pairs, the quality of the well interaction predicts the quality of the ill interaction.

HYPOTHESIS II. A less optimal mother- child interaction during illness (an ill EAS total score below the median)

significantly affected the specificity, reliability, and use of resources in the di-

rection hypothesized (Table III). The mean ill EAS scores (+_ SD) of mother- infant dyads below and above the median were t6.6 _+ 1.12 and 19.5 _+ 0.73, respectively. A well EAS total score

below the median similarly affected the use of resources (Table IV); contrary to our hypothesis, a well EAS total score

813

McCARTHY Er AL THE JOURNAL OF PEDIATRICS JUNE 2000

Tab/e/E. Relation of well EAS above and below the median to resource use

Tab/e V. Relation of maternal, infant, social milieu, life events, and demographic data to the mother-well child interaction

Table VI. Relation of well EAS scores and maternal, infant, social milieu, life events, and demographic data to ill EAS scores

~ llv~ in hom~

below the median did not affect specificity or reliability (data not shown).

HYPOTHESIS I I I . The predictor variables (maternal, infant, social milieu, life events, and demographic data) had an r correlation of 0,68 and r 2 of 46% to the mother-well child interaction

(well EAS total score) (Table V). The SHIE H O M E (Restriction), H O M E (Maternal Responsiveness), and BDI accounted for most of the variance explained in the well EAS.

The correlation of the mother-well child interaction, maternal, infant, social milieu, and demographic data to

the mother-ill child interaction (ill EAS total score) was also studied. The predictor variables had an r of 0.80 (t 2 64%) with the ill EAS (Table VI). The well EAS total score, H O M E (Daily Stimulation), and H O M E (Maternal Responsiveness) accounted for most of the variance explained in the ill EAS.

For the preceding 2 analyses, in 87% of instances the direction of correlation was in the hypothesized direction; that is, adverse characteristics, such as depression, were associated with less optimal, lower EAS scores, and always in the hypothesized direction for the following predictor variables: HOME, SHIF, BDI, EAS, Demographic, RIBQ-Diztress to Approach, RIBQ- Distress to Limitation, and Parenting Social Support Index-Number of Peo- ple in Support Network.

Multivariate techniques were used to calculate separate models for the relation of the predictor variables (maternal, infant, social milieu, demographic, and ill EAS total score) to the following outcomes: pediatricians' AIOS scoring, mothers' AIOS scoring, prescription use, diagnostic test use, hospitalizations, recommendation of a follow-up visit, and mothers' use of the ED. In Table VII, the r correlation of all these predictor variables to the outcome llst- ed and the proportion of variation explained in the outcome by the predictor variables (r e ) is noted. The r e or prediction of variation in the outcome represents, according to the criteria of Cohen, 34 medium effect sizes for AIOS scores, use of prescriptions, diagnostic tests, and hospitalizations and large effect sizes for recommendation of a follow-up visit and use of the ED. As anticipated (because of the high in- tercorrelation between--on the one hand--maternal, infant, social milieu, and demographic data a n d - o n the other hand--EAS scores), maternal, infant, social milieu, and demographic data were components of all of the models in Table VII. In 82% of instances, the correlation of independent variable to the selected outcome was in

814


the direction hypothesized; that is, adverse characteristics, such as depression, were associated with higher AIOS scores. The r correlations of all HOME, SHIE BDI, RIBQ-Distress to Approach and RIBQ-Distress to Limitation, EAS, illness severity, and demographic data to the selected outcomes were always in the hypothesized direction.

In all of these analyses, illness severity (serious illness present, yes/no) as a predictor variable contributed modest- ly at best to the r2; for example, illness severity accounted for only 6%, 1%, and 1% of the variance explained in, respectively, pediatricians' AIOS score, mothers' AIOS score, and test ordering.

If demographic data were removed from the predictor variables in Table VII, the prediction of outcomes in Table VII by maternal, infant, and social milieu characteristics alone would be somewhat less but still robust (eg, for pediatricians AIOS' score 17.6% vs 21.2% in Table VII). Thus adverse maternal, infant, and social milieu data, independent of demographic data, are important predictors of clinical judgments and resource use.

To control for age in the analyses of hypothesis III, the maternal, infant, social milieu, and life events data that were scored temporally closest to the ill visit were used. The correlations noted in Table VII continued to be demonstrated but generally with less- ening of effect; for example, P for pediatricians' AIOS score was 17%, for mothers' AIOS scores 14%, and for hospitalizations 15% versus (Table VII) 21%, 23%, and 22%, respectively. Predicting use of the ED improved in the age-adjusted analysis versus the non-age-adjusted analysis (32% vs 29%, respectively).

DISCUSSION

For mother-infant dyads, the interaction when the child is well predicts the

MCCARTHY ET AL

Table VII. r Correlations and r 2 of maternal, infant, social milieu, life events, demographic and ill EAS data to selected outcomes

quality when the child is sick. Less optimal interactions during illness reduce significantly mothers' reliability and pediatricians' and mothers' specificity and increase use of resources. The mother- child interaction and the maternal, infant, and social milieu characteristics used in this study were more robust in predicting use of resources than were predictors used in a previous study (unpublished data). 1,2 Correlations in this report were in the direction hypothesized; adverse measures were associated with higher perceptions of illness severity, lower specificity of judgments, and greater resource use.

Acute illnesses in children in the first 2 to 3 years of life are often caused by viral infections. Only 2% to 3% represent serious illnesses such as pneumo- nia, bacteremia, meningitis, or urinary tract infection. 42 Children in this age group with a serious illness may not have typical physical examination findings. 43 Hence, observation of the child's state of well-being, most impor- tantly the manner in which the child responds to stimuli either from the parent or the physician, becomes a critical component of the evaluation of severity of illness. Our previous work has outlined 6 key observation items (the AIOS), which, when used for observational assessment of a child with an acute illness, are reliable and also specific and sensitive for serious illnesses. 2° Because the 6 items in the AIOS represent data that could potentially be assessed by parents, especially stimu-

lus-response data, a randomized, controlled trial was conducted, which compared the observational assessments of parents taught to use these scales with those of parents using global assessment of severity of illness. 1,2

As compared with the observational assessments of pediatricians made simultaneously with the parents and before the history and physical examination in over 700 acute illness episodes, the observational assessments of parents in the control group, and to a lesser extent, in the intervention group were characterized by perceiving a child as more ill. This heightened perception of illness lowered the reliability and specificity of parents' clinical judgments. 1 It was further found that adverse maternal (eg, higher levels of state and trait anxiety) and social milieu (eg, 1Mng in the inner city) characteristics exacerbated the tendency of mothers to perceive their child as more ill, thus reducing specificity. 2 A post hoc analysis showed that such adverse char- acteristlcs were also correlated with pediatricians perceiving the child as more ill and ordering more tests--9% of the variance in perception of degree of illness, that is, AIOS scores (vs 21% in this report) and 6% of the variance in test ordering (vs 22% in this report).

Similar findings were recently reported by McConnechie et a144 in Rochester. When comparing hospitalization rates for inner-city and suburban children while controlling for severity of illness, it was found that

815


hospitalizations for less serious illnesses, such as gastroenteritis, which are at

the discretion of the physician, were 3 times as frequent in inner-city children compared with suburban children. Such discret ionary admissions represented 59% of admissions in the study population. In the 2 populations, there were no differences in hospitalization rates for illnesses, such as congenital anomalies, in which less physician discretion was involved. The authors questioned whether clinical judgment was being applied in different ways in the 2 populations. Further, E D use

rates for children in Medicaid man- aged care programs, who often live in

impoverished environments, are 400 to 600 E D visits per 1000 patients per year as compared with rates of 75 to 180 per 1000 patients p e r y e a r for children with commercial insurance. 45 The

adverse maternal, infant, social milieu, life events, EAS, and demographic data (the predic tor variables used in

this report) accounted for 22% of the variance in hospital admissions and

29% of E D use. Hence, our data pro- vide an explanatory model for the in- creased hospitalization rates and E D use in mother- infant dyads in inner- city, stressed environments.

The model outlined here explains the findings of previous studies. It also presents the oppor tuni ty to s tudy interventions focused on those factors that can be identified at well child care

visits, which powerful ly influence the perception of illness and use of health care resources. A previous s tudy of

such interventions for families with ill children was able to reduce inappro- priate use of follow-up care by half. 46

There are limitations to this study. Not all patients had all planned assess-

ments performed. This was sometimes due to difficulty in contacting patients by phone or by letter because of changes in phone numbers or address- es. This was seen more often in urban

patients. However, in general, the number of assessments for each moth-

er-infant dyad was robust (eg, a mean

of 3 EAS evaluations ra ther than the planned 4). Attr i t ion also resulted in incomplete data gathering on enrolled patients. The re tent ion of s tudy sub- jects over 30 months of follow-up was reasonable, 72%. Attr i t ion rates did not differ by study site. We attempted to compensate for attrition bias by including data on patients who dropped from the study in our analysis. We plan

to analyze further the issue of attrition bias by using a method developed by one of u s 47 t o assure that results do not

differ between those completing and those not completing the study.

Our s tudy repor ts a correlat ion between the mother -wel l child and mother- i l l child interaction, between the mother-ill child interaction and the specificity of judgments and use of resources, and between less optimal environments and the mother-child interaction. Our s tudy hypotheses did not address the issue of whether the mother-child interaction mediates the

effects of environment on clinical judg-

ment. The question concerning possi-

ble mediator effects is a complex one and requires the design of a follow-up investigation that tests such a hypothesis directly. 48

We thank Donald Showalter for assistance in statistical analyses; the office staffs at partici- pating practices for their cooperation; and Paula Fackelman, Ilaria Borghese, and Ricar- da Tomlin for their help with data entry.

REFERENCES 1. McCarthy PL, Sznajderman S, Lust-

man-Findling K, Baron MA, Fink HD, Czarkowski N, et al. Mother's clinical judgment: a randomized trial of the Acute Illness Observation Scales. J Pediatr 1990;116:200-6.

2. McCarthy PL, Cicchetti DV, Sznaj- derman SD, Forsyth BC, Baron MA, Fink HD, et al. Demographic clinical and psychosocial predictors of the reliability of mothers' clinical judgments. Pediatrics 1991;88:1041-6.

3. Radke-Yarrow M, Cummings EM, Kuczynski L, Chapman M. Patterns of attachment in two- and three-year olds in normal families and families with

parental depression. Child Dev 1985; 56:884-93.

4. Donovan WL, Leavitt LA. Maternal self-efficacy and infant attachment: in- tegrating physiology, perceptions and behavior. Child Dev 1989;60:460-72.

5. Zuckerman BS, Beardslee WR. Ma- ternal depression: a concern for pediatricians. Pediatrics 1987;79:110-7.

6. Donovan WL. Maternal learned helplessness and physiologic response to infant crying. J Pers Soc Psychol 1981;40:919-26.

7. Donovan WL, Leavitt LA. Simulating conditions of learned helplessness: the effects of interventions and attribu- tions. Child Dev 1985;56:594-603.

8. Campbell SBG. Mother-infant interaction as a function of maternal ratings of temperament. Child Psychiatry Hum Dev 1979;10:67-76.

9. Klein PS. Behavior of Israeli mothers toward infants in relation to infants' perceived temperament. Child Dev 1984;55:1212-8.

10. Bates, JE, Maslin CA, Frankel KA. Attachment security, mother-child interaction and temperament as predictors of behavior problem ratings at age three years. In: Bretherton I, Waters E, editors. Growing points in attachment theory and research. Monogr Soc Res Child Dev 1985;50(1-2, Serial No. 209):167-93.

11. Mangelsdorf S, Gunnar M, Kesten- baum R, Lang S, Andreas D. Infant proneness to-distress temperament, maternal personality and mother-infant attachment: associates and good- ness of fit. Child Development 1990; 61:820-31.

12. Crnic KA, Greenberg MT, Ragozin AS, Robinson NM, Basham RB. Ef- fect of stress and social support on mothers and premature and full term infants. Child Dev 1983;54:20%17.

13. Crockenberg SB. Infant irritability, mother responsiveness, and social support influences on the security of infant-mother attachment. Child Dev 1981;52:857-65.

14. Cutrona CE, Troutman BR. Social support, infant temperament and parenting self-efficacy: a mediatlonal model of postpartum depression. Child Dev 1986;57:1507-18.

15. Larson CP. Efficacy of prenatal and post partum home visits on child health and development. Pediatrics 1980;66: 191-7.

16. Field TM, Widmayer SM, Stringer S, Ignatoff E. Teenage, lower class black mothers and their preterm infants: an

816


MCCARTHY ET AL

intervention and developmental follow up. Child Dev 1980;51:426-36.

17. Barrera ME, Rosenbaum PL, Cun- ningham CE. Early home intervention with low birth weight infants and their parents Child Dev 1986;57:20-33.

18. McCarthy PL, Jekel J E Stashwick CA, Spiesel SZ, Dolan TE Sharpe MR, et al. Further definition of history and observation variables in assessing febrile children. Pediatrics 1981;67: 687-93.

19. McCarthy PL, Cicchetti DV. Diagnos- tic styles of attending pediatricians, residents and nurses in evaluating febrile children. Clin Pediatr 1982;21: 534-7.

20. McCarthy PL, Sharpe MR, Spiesel SZ, Dolan TE Forsyth BW, DeWitt TG, et al. Definition of valid and reliable observation scales to identify serious illness in febrile children. Pedi- atrics 1982;70:802-9.

21. Pettit GS, Bates JE. Continuity of individual differences in mother-infant relationships from 6-13 months. Child Dev 1984;55:729-39.

22. Biringen Z, Robinson JA. Emotional availability in mother-child interactions: a reconceptualization for research. Am J Orthopsychiatry 1991; 61:258-71.

23. Biriugen Z, Robinson JA, Erode RN. Emotional availability scales. Denver: University of Colorado Health Sci- ences Center; 1990.

24. Biringen Z. Direct observation of maternal sensitivity and dyadic interactions in the home: relation to maternal thinking. Dev Psychol 1990;26:278-84.

25. Bornstein MH, Tamis-LeMonda CS, Tal J, Ludemann P, Toda S, Rahn CW, et al. Maternal responsiveness in infants in three societies: The United States, France and Japan. Child Dev 1992;63:808-21.

26. Bornstein MH, Tamis-LaMonda CS. Activities and interactions of mothers and their firstborn infants in the first six months of life: covariation, stabili- ty, continuity, correspondence, and

prediction. Child Dev 1990;61:1206- 17.

27. Caldwell B, Bradley R. Home Obser- vation for Measurement of the Envi- ronment. Little Rock: University of Arkansas at Little Rock; 1984.

28. Ertem 70, Forsyth BWC, Avni-Singer A J, Damour LK, Cicchetti DV. Devel- opment of a supplement to the Home Scale for children living in impoverished urban environments. Dev Behav Pediatr 1997; 18:322-8.

29. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psy- chiatry 1961;4:561-9.

30. Telleen S, Herzog A, Kilbane TL. Im- pact of a family support program on mothers' social support and parenting stress. Am J Orthopsychiatry 1989;59: 410-9.

31. Rothbart MK. Measurement of temperament in infancy. Child Dev 1981; 52:569-78.

32. Gibaud-Wallston JA. Self-esteem and situational stress: factors related to race, SES, and levels of perceived social support. J Clin Child Psychol 1978;9:13-6.

33. Abidin RR. Parenting Stress Index. Charlottesville (VA): Pediatric Psy- chology Press; 1983.

34. Cohen J. Statistical power analyses for the behavioral sciences. Hillsdale (N J): Erlbaum; 1988. p. 79-81.

35. Cicchetti DV, Showalter D, McCarthy PL. A computer program for calculat- ing subject-by-subject kappa or weighted kappa coefficients. Educ Psychol Meas 1990;50:153-8.

36. Baskin MN, Fleisher GR, O'Rourke EJ. Outpatient management of febrile infants 28 to 90 days of age with intra- muscular Ceftriaxone [abstract]. Am J Dis Child 1990;144:419.

37. Crain EE Shelov SP. Febrile infants: predictors of bacteremia. J Pediatr 1982;101:686-9.

38. Cohen J, Cohen P. Applied multiple regression/correlational analysis for the behavioral sciences. 2nd ed. Hills-

dale (NJ): Lawrence Erlbaum Associ- ates; 1983.

39. Cicchetti DV, Sparrow SS. Develop- ing criteria for establishing interrater reliability of specific items: applica- tions to assessment of adaptive behavior. Am J Ment Retard 1981;86: 127-37.

40. Freudigman KA, Borghese IE Mc- Carthy PL. A comparison of two methods of scoring the maternal welt-child interaction. Poster presented at Pedi- atric Academic Societies; 1997 May 2- 6; Washington, DC.

41. Freudigman KA, Borghese IE McCarthy PL. Practitioners' judgments of maternal well-child interaction. Platform Presentation, Pediatric Academic Societies; 1997 May 2-6; Washington, DC.

42. McCarthy PL. Acute infectious illness in children. Compr Ther 1988;14:51-7.

43. Samson JH. Febrile seizures and purulent meningitis, JAMA 1969;210: 1918-9.

44. McConnechie KM, Roghmann K J, Liptak GS. Socioeconomic variation in discretionary and mandatory hospitalization of infants: an ecologic analysis. Pediatrics 1997;99:774-84.

45. Public Sector Contracting Report. Marietta (GA): National Health Infor- mation; 1996;2:17-25.

46. Nelson EW, Van Cleve S, Swartz MK, Kessen W, McCarthy PL. Improving the use of early follow-up care after emergency department visits: a randomized trial. Am J Dis Child 1991; 145:440-4.

47. Cicchetti DV, Nelson L. Reexamining threats to the reliability and validity of putative brain-behavior relationships: new guidelines for assessing the effect of patients lost to follow up. J Clin Exp Neuropsychol 1994; 16:339-43.

48. Baron RM, Kenny DA. The modera- tor-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. J Pers Soc Psychol 1986;51: 1173 -82.

817

The mother-child interaction and clinical judgment during acute pediatric illnesses

Documents

Transcript of The mother-child interaction and clinical judgment during acute pediatric illnesses