From the Department of Pediatrics, Maulana Azad Medical College,
New Delhi 110 002, India.

Reprint requests: Dr. H.P.S. Sachdev, Professor and Incharge, Division of Clinical
Epidemiology, Department of Pediatrics, Maulana Azad Medical College,
New Delhi 110 002, India. E-mail: jiap@ren.nic.in

Manuscript received: December 31, 1998; Initial review completed: March
5,1999;                  Revision accepted: March 11, 1999.

---

Objective: To evaluate the utility of the "WHO/UNICEF algorithm for integrated
management of childhood illness (IMCI) between the age of 2 months to 5 years.
Design: Prospective observational.
Setting: The Outpatient Department and Emergency Room of a medical college hospital.
Methods: 203 children presenting to Outpatient Department (n=101) or Emergency Room(n=102) were assessed and classified as per `IMCI' algorithm and treatment required was identified. A detailed evaluation with all relevant investigations was also done for these subjects. The final diagnoses made and therapies instituted on this basis served as `gold standard'. The diagnostic and therapeutic agreements between the `gold standard' and the IMCI and vertical (on the basis of primary presenting complaint) algorithms were computed.

Results: More than one illness was present in 135 (66.5%) of subjects as per `gold standard'. The mean (SD) numbers of morbidities as per the gold standard and IMCI- low and high malaria risks were 2.1 (1.1), 1.8 (1.0) and 2.2 (1.1), respectively. Subjects having any referral criteria as per IMCI module had a greater co-existence of illnesses (mean 2.6 vs. 1.6 illnesses per child, respectively). The referral criteria proved useful in predicting hospitalization and a combination of hospitalization and observation; their sensitivity and specificity were 81% and 69% and 74% and 85%, respectively. IMCI algorithms covered majority (92%) of the recorded illnesses. A total agreement with IMCI (malaria low risk) was found in 129 (64%) cases while in 43 (22%) cases, there was partial agreement.
Corresponding figures for vertical (split IMCI) program were 93 (46%; p<0.001) and 41 (25%). The difference was primarily due to underdiagnoses (30%). Diagnostic discordance of IMCI algorithm and gold standard was evident for the cough category due to underdiagnosis of bronchial asthma and bronchiolitis and an overdiagnosis of pneumonia whereas the discordance for fever was due to an overdiagnosis of malaria. Identical results were found for broad treatment categories. The IMCI algorithm had a provision for preventive services of immunization (16.3% possibility of availing missed opportunities) and
feeding advice.

Conclusions: There is a sound scientific basis for adopting the IMCI approach since: (i) co-existence of morbidities is frequent; (ii) severe illness is assessed with good sensitivity and specificity; and (iii) the IMCI algorithm is diagnostically and therapeutically superior to the vertical disease specific algorithms. The generic IMCI algorithm needs adaptation to reflect the regional morbidity profile.

Key words: Integrated Management of Childhood Illness, Morbidity, Under five.

---

 

Subjects and Methods

Results

Discussion

References

In an attempt to reduce under-five morbidity and mortality in the developing world and to improve health-workers’performance in managing childhood illnesses, World Health Organization (WHO) and United Nations Children’s Fund (UNICEF) have developed an ‘Integrated Management of Childhood Illness’ (IMCI) approach(1,2). Instead of the earlier focus on single disease management protocols (for example, diarrheal diseases or acute respiratory infections), this approach is targeted at the principal causes of morbidity and mortality in an integrated manner and also addresses the overall health of the child. The recommended algorithms consider the integrated management of childhood illness in two age groups, namely one week up to two months and two months up to five years.

Data from the African subcontinent(3-6) focusing largely on paramedical personnel has validated the feasibility and utility of adopting the IMCI approach. Currently 60 countries are at different stages of implementation of the IMCI approach (Hans Troedsson, Department of Child and Adolescent Health and Development, World Health Organization, Geneva, personal communication). India too is in the process of introducing this strategy. However, before widespread implementation, the generic ‘IMCI’ algorithms require careful adaptation to reflect the epidemiological and cultural characteristics of the country.

There is a paucity of published experience with the ‘proposed approach’ in India.Further, there is scarce quantification of the upper range of expectations from this approach, namely the agreement between the ‘gold standard’ and a pediatric resident following the algorithms. The present study was therefore designed to generate relevant information in this context for the proposed algorithm for children between the ages of two months up to five years.
 
 

Subjects and Methods

Results

Discussion

References

Subjects and Methods

The study was performed at the Outpatient Department and Emergency Room between May, 1996 to January, 1997. Both Outpatient Department and Emergency Room settings were utilized so that illnesses of various types and severity could be evaluated. In order to minimize the bias resulting from the ‘Dengue’ epidemic encountered in Delhi, patients were not recruited for a period of about 3 weeks during the outbreak (September, 1996). All subjects presenting to the Outpatient Department or Emergency Room of the hospital for the first time for a fresh episode of any illness, who were aged between 2 months to 5 years were eligible for enrolment in the study. The recruited subjects were selected from the eligible cases in a randomized stratified manner (Outpatient Department and Emergency Room). For the Outpatient randomization, on a fixed OPD day of every week, the first recruited subject was selected by draw of lots from three numbers (namely 1,2,3). Subsequently every third child fulfilling the entry criteria was selected. A similar randomization was done for the patients who presented to the Emergency Room. The rationale for randomizing every third child was based on the morbidity load of the pediatric services and a pretest to evaluate the viability of this randomization procedure.

For the children recruited in the study, the WHO/UNICEF algorithm for ‘Integrated Management of Childhood Illness’ was referred to. Every study child was assessed and classified according to these guidelines, treatment steps identified and recorded in a proforma. A pediatric postgraduate trainee (DS) performed this assessment during the second year of the residency program. The study subjects were then assessed, examined and managed according to the protocol of the treating unit under the supervision of faculty and/or pediatric senior residents. All relevant investigations (including blood counts, chest radiograph, stool examination, blood cultures, lumbar puncture, etc.) were performed based on the history and detailed clinical examination. A thin ‘peripheral smear’ was made for each patient presenting with a history of fever or axillary temperature of 37.5°C or more, and examined for the presence of malarial parasite in addition to morphology and differential leukocyte counts. A chest radiograph was taken for every patient with a history of ‘difficult breathing’ or findings of respiratory distress (tachypnea or chest indrawing) or chest examination abnormalities. An otosco-pic examination of the ears was performed in every study subject. Based on the detailed clinical evaluation and relevant investigations, final diagnoses were made and therapies instituted. These diagnoses and treatments were considered as the ‘gold standard’.

The study children were either admitted, observed for a period of not more than 24 hours or sent home after initial evaluation, depending upon the nature and severity of illness. Each unimmunized or incompletely immunized sick child was immunized and dietary therapy/ advice was given to every malnourished child. Hospitalized children were followed up till discharge or death. Other children were asked to report for routine follow up (3 to 7 days later) to determine the final outcome.

Three categories of possible diagnoses and treatments were therefore available for each recruited study subject, namely, ‘gold standard’, ‘IMCI algorithm’ and ‘solitary vertical algorithm’. The last refers to a single algorithm being followed in a child on the basis of the primary presenting complaint stated by the mother or the relative. For this purpose, the IMCI algorithm was split into solitary vertical components on the basis of presenting complaints like cough or difficult breathing, loose stools, fever, etc.

A sample size of 203 was calculated to be sufficient to detect a difference of 5% in diagnostic agreement from the gold standard with 90% power and an alpha of 0.05. The data was entered and analyzed with the Epi-Info version 5.00 software. The diagnostic and therapeutic agreements between the ‘gold standard’ and the ‘IMCI’ and ‘vertical’ algorithms were computed. Other standard statistical tests performed included Student’s ‘t’ test, Chi-square test, Fischer’s exact test, Odds Ratio (OR), sensitivity, specificity, positive predictive value and negative predictive value.
 

Subjects and Methods

Results

Discussion

References

Results

A total of 203 children (101 from Out- patient Department and 102 from Emergency Room) were evaluated, out of which 78 (38.4%) were hospitalized, 44 (21.7%) were kept under observation for a period not more than 24 hours and 81 (39.9%) were sent back after initial evaluation. Two thirds (66.5%) of the children had two or more co-existent morbidities as per the ‘gold standard’ diagnoses. In comparison to the ‘gold standard’, the ‘IMCI’ module assuming a low malaria risk zone documented a slightly lower number of co-existent morbidities whereas with the assumption of high malaria risk zone, the total morbidities documented were higher. The mean (SD) numbers of morbidities in these three groups were 2.1 (1.1), 1.8 (1.0) and 2.2 (1.1) respectively. Children requiring referral as per IMCI algorithm had significantly greater co-existence of morbidities (2.5 ± 1.1 vs. 1.6 ± 0.8, p <0.001). Thus the co-existence of morbidities was significantly higher in those children who had been assessed to have a relatively severe condition.

The utility of the ‘General Danger Signs’ (convulsions, lethargy or unconsciousness, inability to drink or breastfeed and vomiting everything) and other ‘Referral Criteria’ outlined in IMCI module is summarized in Table I. These computations were done for predicting admissions (n=78) and a combination of both admission and subjects kept under observation (n=122). The latter analysis was also important for operational implications for para-medical personnel since even the trained pediatricians were uncertain that the observation subjects could be sent away immediately and they preferred to evaluate the response to supervised therapy before taking a decision. The specificity for ‘General Danger Signs’ was higher (87% and 95%) while the sensiti-vity was low (39% and 34%). However, when all the referral criteria (including ‘General Danger Signs’) were considered, the sensiti-vity increased substantially (81% and 69%) while the specificity declined marginally (74% and 85%).

Table I__Utility of ‘General Danger Signs’ and ‘Referral Criteria’ as a Predictor of Hospital Admissions/Observations.

Parameter                 Sensitivity       Specificity    Positive-Predictive   Negative-Predictive                     OR+

                                        (%)                 (%)                      Value (%)                      Value (%)                  (95% CI)

General       Admitted           39                87                     65                                    69                 4.3
Danger       (n = 78)                                                                                                                                                         (2.0-9.1)
Signs          Admitted
(GDS)          plus under
                   observation           34                95                     91                                    49                 10.1
                      (n=122)                                                                                                                (3.3-34.9)

Referral       Admitted          81               74                    66                                     86               11.7 Criteria        (n = 78)                                                                                                                                             (5.6-24.1) including     Admitted
GDS             plus under
                     observation         69               85                    88                                     65                12.7

The comparison of morbidity profile observed in these 203 subjects as per the gold standard and IMCI algorithm is depicted in Table II. It is apparent that majority (92%) of the recorded illnesses are covered by the IMCI algorithms. Broad discordance is apparent in the presenting symptoms of ‘cough’ and ‘fever’ (especially high malaria risk). The diagnostic discordance is evident for the ‘cough’ category, especially pertaining to underdiagnosis of bronchial asthma and bronchiolitis and an overdiagnosis of pneumonia whereas the discordance for fever was primarily related for an overdiagnosis of malaria assuming high risk of this disease.

Pneumonia                              34                                   38                                      38
Severe pneumonia
/very severe disease –             -                                     51                                      51
No Pneumonia/
cough or cold (URI)               62                                   49                                       49
Bronchial asthma                   42                                     –                                         –
Bronchiolitis                           9                                      –                                         –
Tuberculosis                           2                                      –                                         –
Empyema                                 1                                     –                                         –
Laryngotracheobronchitis        1                                      –                                         –
Acute ear infection                  5                                      2                                        2
Chronic ear infection             10                                      7                                        7

Diarrheal Diseases

Acute diarrhea                      51                                    55                                       55
Dysentery                             10                                      6                                         6
Persistent diarrhea                 3                                      3                                         3

Malnutrition

Very low weight                    36                                  35                                        35
(<3 Z score)
Severe malnutrition                 6                                    6                                          6
(clinically marasmus) –
Marasmic kwashiorkor           7                                     9                                          9

Anemia (Hb <10g/dl)           53                                  54                                         54

Fever
Malaria                                   9                                     7                                         89
Very severe febrile disease    –                                    34                                        34
Meningitis                               6                                     –                                          –
Measles                                   1                                     3                                         3
Dengue hemorrhagic fever       2                                     –                                         –
Sepsis                                      3                                     –                                         –
Enteric fever                            2                                     –                                         –
Others+                                   51                                    0                                         0

The diagnostic agreement was calculated between ‘gold standard’ and the ‘IMCI’ module. For this purpose, referral to a higher center was considered as a diagnostic match if the patient was admitted or kept under observation and a diagnostic mismatch (over- diagnosis) if they were neither hospitalized nor kept under observation. A total agreement was considered if the case required referral as per ‘IMCI’ or ‘Vertical’ algorithms and was actually admitted or kept under observation or all the diagnoses made by algorithms matched the gold standard. A total disagreement was considered if no diagnosis considered by the algorithms was made by the ‘gold standard’. Cases not fitting in the above two categories were defined as partial agreement. Table III summarizes the diagnostic agreement of the ‘IMCI’ with the ‘gold standard’ assuming both low and high malaria risks. Considering low malaria risk, there was a total agreement on all diagnoses in a single patient in 63.5% of subjects. On considering malaria as high risk, the mismatch increased by nearly 10%. The mismatch was more commonly of an under-diagnosis (26.1%) rather than overdiagnosis (19.7%). Moreover of the 36.5% of subjects in which there was any mismatch, 21.2% of subjects had partial diagnostic agreement with the ‘gold standard’. Thus only 15.3% of subjects had total disagreement with the ‘gold standard’ (malaria as low risk).

The performance of vertical programs was compared with that of ‘IMCI’ and the ‘gold standard’. For this analysis, each component of the ‘IMCI’ was considered as an individual vertical program and the primary presenting complaint stated by the mother/relative decided the selection of vertical program module. Table IV presents the results of such an analysis. It is evident that the ‘IMCI’ algorithms were markedly superior to the vertical programs. This difference was primarily related to underdiagnosis (missing illnesses) of the magnitude of 30%. A similar analysis was done regarding agreement for treatments (broad groups) identified/given as per these three regimens. A total agreement with ‘gold standard’ was seen in 63.5% and 45.8% for ‘IMCI’ and ‘Vertical’ programs, respectively (p<0.001). For these calculations, immunization essentially a preventive component was not considered. However, with the ‘IMCI’ algorithm, 31% of the non referred cases would have received immunization (this works out to be 16.3% of the total sample of 203 cases) whereas vertical programs do not have separate provision for promoting immunization.

Number of illnesses  Mean (SD)      2.1        (1.0)              1.8          (1.0)           1*

Total Agreement                           203        (100)           129         (63.5)+         93      (45.8)

Partial Agreement                          0             (0)              43           (21.2)          51       (25.1)

Total Disagreement                        0             (0)              31           (15.3)+       59        (29.1)

Underdiagnosis                               0             (0)              53           (26.1)        115      (56.6)

Overdiagnosis                                 0             (0)              40           (19.7)        32        (15.8)

Discussion
 

The major rationale for propagating the ‘IMCI’ approach is that a single diagnosis for a sick child is often inappropriate as it identifies only the most apparent problem, and can lead to an associated and potentially life threatening condition being overlooked (1,7,8). The current investigation reaffirms that co-existence of illnesses is a rule rather than exception with two thirds of children having more than one illness. Similar observations were recorded in earlier studies from Uganda(9), Bangladesh(10) and Kenya(4). Another important finding was that the number of morbidities was higher in those children who had been assessed to have a relatively severe condition (means of 2.5 vs. 1.6 illness/child). Data from Bangladesh(10) also showed that patients with more than one major provisional diagnosis were more often assessed to require admission than those with one major diagnosis. There is thus a sound scientific basis for adopting the ‘IMCI’ approach. In fact, in routine day to day practice also, pediatricians are expected to offer an integrated package of preventive, promotive and curative services as opposed to a single symptom or illness based therapy.

An important aspect of the ‘IMCI’ algorithm is the strength of the referral criteria. Weakness in this area can easily undermine the confidence of the para-medical personnel and the community for this proposed health intervention. The ‘IMCI’ guidelines are designed to be highly sensitive on the referral of patients with a possible severe illness, and ideally they would exclude those whose illness does not require care at a hospital(11). To achieve adequate sensitivity in detecting severely ill children who require referral, criteria must be used that inevitably lead to some children being referred unnecessarily. In our study the proportion of children with any of the referral criteria was 47% (Table V). In studies from Kenya, Gambia, Bangladesh, Uganda and Ethiopia, this figure ranged from 7% to 53% (3-5,9,10,12). The present series and the study from Bangladesh(10) had documented a higher percentage of cases (47% and 53%) with any of the referral criteria in comparison to the other reports (7% to 28%). This is primarily a reflection of the cohort selection; in the current investigation, the cohort comprised an equal number of Emergency Room and Outpatient subjects whereas in Bangladesh(10), the children presenting to Emergency Room were preferentially enrolled thus documenting an even higher proportion of patients requiring referral as per ‘IMCI’ algorithm. Other studies(3-5,9,12) enrolled patients presenting only to the Outpatient Department who were less likely to have severe morbidity.The substantially lower rate of referral in Ethiopia(5) may also be explained by the fact that this was the only study using health centers not attached to a hospital whereas other investigations(3,4,9,12) were conducted in Outpatient Departments of district or regional hospitals.

 
We compared the ‘IMCI’ recommendation for referral with the judgement of pediatrician on the need for hospitalization (and a combination of hospitalization and observation) and found a reasonably good sensitivity (81% and 69%, respectively) and specificity (74% and 85%, respectively). Our investigation had evaluated two other recommended criteria for referral (in contrast to earlier studies), namely, fever coming daily for more than 7 days and cough more than 30 days. Such subjects are referred for further assessment and might not require urgent attention at higher level. However, when these two criteria (in nine subjects; all for fever coming daily for more than seven days) were excluded, there was only a marginal change in sensitivity (77% and 65%, respectively) and specificity (78% and 90%, respectively). Table V compares the utility of these referral criteria in this and earlier studies. Children with any of the referral criteria had mean Odds Ratio (OR) ranging from 3.2 to 199.3 (except one study 3.2-12.7) for predicting hospitalization. The sensitivity and specificity of the referral criteria to predict a higher level of care ranged from 41% to 86% and 64% to 97%, respectively. When children were assessed by pediatricians as compared to trained health workers (both using ‘IMCI’ algorithm), the sensitivity was much better (69% to 86% vs. 41% to 66%) without much compromise in specificity (64% to 97% vs. 79% to 98%). This difference is only to be expected in view of limited experience of health workers in recognizing clinical signs. Inaccurate observation of clinical signs by health workers, especially chest indrawing and bipedal edema, resulting in missed referrals have been documented earlier(4-6). Emphasis therefore needs to be placed on the accurate recognition of signs based on sufficient clinical practice during training. The performance of health workers improves over weeks of observation and it was estimated that average health workers may require several months to incorporate thoroughly and efficiently the ‘IMCI’ process in their daily practice(5).

Regarding overall performance of these algorithms, the clinical experience is limited and mostly restricted to the African subcontinent. Most of these studies have focussed on paramedical personnel whose performances have been compared to clinicians who sometimes had access to radiology and laboratory results. These studies have focussed on the specific disease components of the algorithms(3-5,9). In these studies, the sensitivity and specificity for diagnosing pneumonia was found to vary from 76% to 97% and 49% to 89%, respectively. Similar figures for diarrheal dehydration, malaria and malnutrition were 25%-76% and 96%-99%, 87%-100% and 0%-8%, and 89%-96% and 66%-90%, respectively. The specificity in diagnosing malaria was found to be extremely low if microscopy was not used. Based on these findings, the guidelines were modified to exclude malaria treatment if a fever is accompanied by a running nose, measles or another apparent cause of fever. Studies have also compared the performance of various types of health workers after adequate training (MCH aides, rural medical aides and medical-assistants)(6).

Unfortunately there is scarce quantification of the utility of the ‘IMCI’ approach as a whole in comparison to a ‘gold standard’ and its advantages (if any) over the individual vertical programs. In the present study with ‘IMCI’ approach, considering low malaria risk zone, there was a total agreement with ‘gold standard’ in all diagnoses and prescribed broad categories of treatments in a single patient in 63.5 per cent of subjects, if appropriate referral was considered a diagnostic match. On considering malaria as high risk, the mismatch increased by nearly 10%. Thus a low malaria risk ‘IMCI’ algorithm is appropriate for this geographic area. The ‘IMCI’ approach also proved to be superior to the vertical disease specific algorithms both for diagnostic and therapeutic purposes. The diagnostic discordance was evident for the cough category especially pertaining to underdiagnosis of bronchial asthma and bronchiolitis and an overdiagnosis of pneumonia. In settings where asthma is common and drugs are available to manage it, the guidelines could be accordingly adapted to improve efficiency. The ‘IMCI’ algorithm also focus-ses on the provision of preventive services like immunization and feeding advice for every child which tend to get ignored with disease specific vertical algorithms. In the current study, there was a possibility of missed opportunities for immunization which were effectively covered by the ‘IMCI’ in 16.3% of subjects. It would be prudent to recall that the present study quantified the utility of the ‘IMCI’ algorithm on the basis of an assessment undertaken by a resident undergoing pediatric postgraduate training and the efficacy is likely to diminish in the hands of trained para-medical personnel. In this context, it would be desirable to evaluate the actual performance of trained para-medical personnel in the true field-setting. The ‘IMCI’ algorithm is complex and more difficult to comprehend than vertical disease-specific a lgorithms and sometimes presents a challenge to health workers(5).

It would be prudent to highlight that the study was conducted in outpatient and emergency services setting of an urban tertiary care center. Patients recruited from this setting are certainly different from those seen in peripheral health facilities. For example, bronchial asthma may probably have been over-represented (atmospheric pollution in Delhi is high) and this will affect the "agreement" between the ‘gold standard’ and the ‘IMCI’ algorithm. Guidelines for bronchial asthma can be added to the ‘IMCI’ approach in the process of adaptation to make the algorithm more relevant in places like Delhi where asthma is a major problem. Similarly other relevant data from different geographic regions of the country will prove useful for the adaptation process.

In conclusion, there is a sound scientific basis for adopting the ‘IMCI’ approach since: (i) co-existence of morbidities is a rule rather than exception for sick under five children; (ii) the algorithm provides good sensitivity and specificity for assessing severe illness; and (iii) the‘IMCI’ algorithm is superior to the vertical disease specific algorithms both for diagnostic and therapeutic purposes. It is however important to carefully adapt the generic ‘IMCI’ algorithm to reflect the local morbidity profile.

Contributors: HPSS coordinated the study (particularly its design and interpretation) and drafted the paper; he will act as the guarantor for the paper. DS participated in the data collection, and also helped in drafting the paper.

Funding: None.

Conflict of interest: None.

References