Adherence to management guidelines in acute respiratory infections and diarrhoea in children under 5 years old in primary health care in Botswana

Abstract

Objective. To evaluate health care providers’ adherence to management guidelines for acute respiratory infection and diarrhoea in children under 5 years old in Botswana primary health care.

Design. Cross-sectional prospective field survey. Data collection was carried out through observation of consecutive consultations at 30 randomly assigned clinics and health posts in three purposely chosen districts.

Study participants. This study comprises 185 cases of acute respiratory infection and 85 cases of diarrhoea.

Main measures. Criteria for acceptable standards of history taking and physical examination for acute respiratory infection and diarrhoea were defined as well as criteria for categorizing the appropriateness of antibiotic prescription. The percentage of oral dehydration salts provided in cases of diarrhoea was calculated.

Results. Acute respiratory infection and diarrhoea accounted for 270 (including 15 missing cases) of all main diagnoses (n = 539). In 262 cases (97%) health care providers were nurses or enrolled nurses; in 3% family welfare educators. Acceptable history taking, physical examination, and both combined in acute respiratory infection was found in 113 (63%), 32 (18%), and 28 (16%), and in diarrhoea in 45 (58%), 26 (34%) and 20 (26%) cases, respectively. Antibiotics were prescribed in 76 of 255 (30%) cases. Prescription was assessed as inappropriate in 56 of 76 (74%) of all cases; in 41 of 52 (79%) cases with acute respiratory infection, in none of the pneumonia cases, and in all 15 cases of diarrhoea. Oral rehydration salts were prescribed in 74 (87%) of the diarrhoea cases.

Conclusions. Health care providers’ adherence to guidelines on history taking was suboptimal in acute respiratory infection and diarrhoea but poor on examination in both conditions. A high level of inappropriate antibiotic prescription was found in acute respiratory infection and diarrhoea. Overall, there is considerable scope for improving diagnostic and therapeutic management of these major childhood diseases in Botswana primary health care.

Introduction

In low-income countries acute respiratory infection and diarrhoea account for 27% and 23%, respectively, of childhood mortality and are leading causes of childhood morbidity [1,2]. In acute respiratory infections, it is assumed that 80–90% of deaths are caused by pneumonia [1].

In both developed [3] and developing [4,5] countries, inappropriate management and prescription of medications by health workers is common, as well as misclassification of illness by health workers. A study in Kenya showed that only 8% correctly diagnosed severe pneumonia [6].

Overprescription and abuse of antibiotics in the treatment of acute respiratory infections [7,8] and diarrhea [9] is a worldwide problem, potentially leading to widespread antibiotic resistance.

Poor access to medical services, and inadequate referral and treatment are important determinants in infant mortality of acute respiratory infections [10]. Standard case management has been shown to reduce case fatality rates of acute respiratory infections and diarrhoea significantly, if families recognize signs of possible pneumonia and dehydration [8]. In order to reduce mortality attributed to acute respiratory infections and diarrhoea, the World Health Organization (WHO) has launched the Integrated Management of Childhood Illness (IMCI) strategy in many developing countries, including Botswana in 1998. Botswana reduced its estimated infant mortality rate from 97 in 1971 to 45 per 1000 live births in 1991 [11]. This reduction has been attributed to Botswana’s steady socio-economic and infrastructural development, its political stability, and its commitment to providing comprehensive public primary health care services to the majority of its population.

Most of the curative and preventive health care services in Botswana are organized and provided by trained nurses. Health posts, which represent the lowest level of institutional primary health care, are supervised by clinics, which provide more comprehensive services. Nurses are also the main drug prescribers. Primary health care services, including medicines, are free for children under the age of 5 years. During the 1990s the government focused on improving the quality of services, including promoting rational drug prescription.

Of all outpatient morbidity in under-5s in 1998, diarrhoeal diseases comprised 5% and acute respiratory infection 23% [12]. Of all in-patient morbidity and mortality, intestinal infectious diseases comprised 17% and 5%; acute respiratory infection 30% and 32%, respectively [12].

This study is part of a drug utilization study in primary health care in Botswana, which focuses on the quality of the diagnostic process [13], adherence to treatment guidelines [14], and the quality of dispensing [15].

As management standards differ from country to country, each country needs to define its own standard. Management standards for important childhood illnesses are described in nationally adopted WHO algorithms for the syndromic management of acute respiratory infections [16] and diarrhoea [17]. Adherence to algorithms is an important element of the quality of health care, as it assures rational case management.

The aim of this study was to evaluate health care providers’ adherence to national diagnostic and antibiotic treatment guidelines for acute respiratory infections and diarrhoea in children below the age of 5 years in Botswana primary health care.

Methods

Sampling took place at two levels: a purposeful selection was made of three of the country’s 22 health districts, representing an urban, a semi-urban, and a rural setting. Within each of these districts 10 primary health care facilities were randomly selected (Table 1). The target number of consecutive consultations per health facility was relative to 1996 attendance. The surveyors stayed at the selected health facility until the target number was reached. The patient encounter was the unit of sampling and study.

Inclusion criteria: all consultations, both first visits and controls; oral consent for participation in the study by the parent. Exclusion criteria: weight control and vaccination of healthy children; antenatal control and family planning in healthy women; direct observed treatment of tuberculosis.

At clinics consultations are carried out by nurses (3–5 years of training), at health posts by a nurse (3–4 years), or enrolled nurse (2 years). When there were staff shortages, family welfare educators (3 months training) sometimes carry out consultations. Only one health worker carried out ill-child consultations at each of the health facilities.

The survey was carried out by a team of two nurses and a pharmacy technician, who were trained before and during the pilot survey. The first nurse observed the consultation process with minimal interference. The pharmacy technician observed the dispensing process. The second nurse interviewed the caretaker about her/his knowledge of the medicines dispensed. The nurse–surveyor, an experienced family nurse practitioner, recorded the main diagnosis, which the consulting nurse entered in the outpatient register. This register is compatible with the International Classification of Health Problems in Primary Care [18]. Only the main diagnosis is entered by the consulting nurse in this register. For each consultation a general check-list, comprising 25 items, was filled in by the nurse–surveyor [14]. Up to three additional symptom diagnoses were recorded. For acute respiratory infection and diarrhoea specific check-lists were designed, reflecting the items of history and examination listed in national WHO-based algorithms [16,17]. Both check-lists comprised 12 items each. Since nurses in Botswana are not familiar with auscultation of the thorax, this item was not included in the acute respiratory infection check-list.

Criteria for acceptable adherence to guidelines of history taking and physical examination were defined and are given in Table 2. For diagnosing pneumonia it is essential that the child is examined for respiratory rate and chest indrawing. Using these two simple danger signs, health workers have been able to identify pneumonia [8,19]. Acute respiratory infection is categorized as follows: ‘very severe disease’ (not able to drink, stridor in calm child); ‘severe pneumonia’ (chest indrawing); ‘pneumonia’ (no chest indrawing but fast breathing); and ‘no pneumonia but cough and/or cold’ (no chest indrawing and no fast breathing). The first two categories need to be referred to hospital, the third given home care and antibiotics, and the fourth home care only.

In diarrhoea the nurse should ask for the presence of blood in the stools and assess dehydration by carrying out the skin pinch test and/or taking the weight of the child. According to the algorithm, the degree of dehydration is supposed to be classified as mild, moderate, or severe. All cases with diarrhoea should be given oral rehydration salts. If clinical mismanagement of acute respiratory infection or diarrhoea cases posed a serious threat to the health of the child, the consulting nurse was informed before the child left the health facility.

Categorization of the quality of antibiotic prescriptions was as follows: (i) appropriate: according to the acute respiratory infection and diarrhoea algorithms [16,17]; in acute respiratory infection a temperature of >37.5°C should be present; (ii) doubtful: as in (i) in those cases where the temperature had not been taken; and (iii) inappropriate: not recommended by the algorithms [16,17], and/or a temperature of ≤37.5°C.

Research and ethical clearance was obtained from the Ministry of Health in Botswana and the Regional Committee for Medical Research Ethics in Norway. The survey was carried out from June to November 1998.

A Data Set Dictionary was designed and all data from 2994 questionnaires were entered into EPI-Info 6. Conversion of the Epi-Info data set to Statistical Package for the Social Sciences [20] (SPSS) was done, but both Epi-Info and SPSS were used in the analysis of the data. Confidence intervals were calculated for levels of adherence to diagnostic guidelines. One-level and multilevel logistic regression analysis of factors independently associated with the defined acceptable standard of diagnosis was performed in Epi-Info, SPSS, and MLwiN (http://multilevel.ioe.ac.uk/index.htlm), using the following independent variables: geographical district, type of health facility, sex, and designation of health worker.

Results

Materials

The data comprise 539 encounters of children under 5 years old, comprising 18% of all 2994 consultations (country-wide 15% [15]). The mean age of this cohort was 1.70 years. Acute respiratory infection comprised 185 (34%), and diarrhoea 85 (16%) cases (Table 3). In eight and seven cases of acute respiratory infection and diarrhoea, respectively, no check-lists had been filled in.

Staff and referrals

Of 270 cases, 231 (86%) were carried out by nurses; 31 (11%) by enrolled nurses, and eight (3%) by family welfare educators. None of the diarrhoea cases was referred. Of four cases with acute respiratory infection (2.2%), two were referred to the hospital and two to the Senior District Medical Officer. However, none of these was cases diagnosed as pneumonia.

Adherence to guidelines on history taking and physical examination

Table 4 gives the number of items checked out in the consultations for acute respiratory infection and diarrhoea. In acute respiratory infection the most frequent symptoms asked for were duration of cough (97%) and fever (64%), and the most frequent signs looked for were temperature (88%) and chest indrawing (16%). The respiratory rate was registered in 11 (6%) only. In diarrhoea the symptoms most frequently asked for were duration (82%) and the presence of blood in the stools (83%). The temperature was registered in 88% of the cases. The skin pinch was carried out in one-third of the cases and the weight taken in only 14%.

In acute respiratory infection and diarrhoea, an acceptable standard of history taking was reached in 63% and 58%, of examination in 18% and 34%, and of both history and examination in 16% and 26%, respectively (Table 2).

One-level logistic regression analysis showed a significant association with the district [Gaborone: OR = 4.7 (95% CI: 2.0–11.2), P < 0.0001; Ngami: OR = 1.7 (95% CI: 0.6–4.4), P = 0.3] and type of health facility [clinics: OR = 0.3 (95% CI: 0.1–0.7), P = 0.009]. In the multilevel logistic regression analysis, the single health facility was used as level 2. However, while tendencies remained, the effects that were found to be statistically significant in the one-level logistic regression were now non-significant.

Antibiotic prescribing

Antibiotics were prescribed in 61 of 178 (34%) cases with acute respiratory infection and in 15 of 77 (19%) diarrhoea cases.

Acute respiratory infections.

In 41 (79%) of the non-pneumonia cases, but in none of the pneumonia cases, antibiotic prescription was inappropriate. The antibiotics most commonly prescribed were phenoxymethylpenicillin and amoxicillin.

Diarrhoea.

The nurse–surveyor assessed 83 (98%) of the diarrhoea cases as mild, one (1%) as moderate, and one (1%) as severe. No diarrhoea cases were referred. All antibiotic prescriptions were assessed as inappropriate (Table 5). Co-trimoxazol was the most frequently prescribed antibiotic. In 24 (28%) cases metronidazole was prescribed. Oral rehydration salts were prescribed in 74 (87%) of the cases.

Patient care indicators

Of 1087 drugs prescribed, 993 (91%) were dispensed. The average consulting time was 4 minutes (95% CI: 3.8–4.2).

Discussion

This survey shows that health care providers’ adherence to guidelines on history taking was suboptimal, but poor on examination in both acute respiratory infection and diarrhoea cases. The level of inappropriate antibiotic prescription was high.

Limitations of the study

Limitations.

One of the weaknesses of this study is the risk of observer bias. We tried to minimize this by not announcing the visits by the research team. Even though the process of being observed by a surveyor may have resulted in improved performance, we assume this to be unlikely, because health workers in Botswana are used to being visited by health officials, in this case by a nurse who was one of their own.

During the consultation there was only minimal interference by the surveyors. Only in cases where clinical mismanagement posed a serious threat to the health of the child, was information to correct this given to the consulting nurse before the child left the health facility. This, together with observer bias may have resulted in overestimation of adherence to management guidelines.

The survey was carried out in the period June–November 1998. This period covers about 4 months of the winter and 2 month of the summer season. This may have resulted in over-representation of acute respiratory infections, which are more prevalent during the winter season, and towards under-representation of diarrhoeal diseases, which are more prevalent during the summer season.

Among the additional diagnoses were 12 cases of acute respiratory infection or diarrhoea. Six patients with an acute respiratory infection or diarrhoea as the main diagnosis also had an additional diagnosis of diarrhoea or acute respiratory infection, respectively. Since no check-list was filled in these cases, we do not know the quality of diagnosis of 18 cases. However, missing information from these cases (7%) is unlikely to have affected our main results.

Validity.

The reason why we opted for a disproportionate sample was based on a compromise between available staff and transport resources, time frame, and access to the health facilities. We opted for quite a big sample size in order to allow for meaningful statistical analyses. Conversely, this choice may have affected representativity: health facility sample fraction by district differs as well as the number of encounters relative to the district population. However, the encounter, not prevalence of diagnoses, was the subject of study. Similar age distribution and prevalence of acute respiratory infection and diarrhoea, compared with national outpatient figures [12], support our belief that this study is representative of primary health care in Botswana.

Since verification of the diagnosis against a gold standard was not the subject of this survey, the level of uncertainty about the validity of the diagnosis and treatment is considerable.

The WHO-based algorithm for acute respiratory infections has a specificity of only 80% [21]. Cough and tachypnoea may be interpreted as pneumonia but are also caused by severe anaemia or malaria [22]. The IMCI algorithm appears to be diagnostically and therapeutically superior to vertical disease-specific algorithms [23].

In assessing the quality of primary health care, it has been shown that the direct check-list-based observation method has a better overall balance of sensitivity and specificity compared with interview of the caretaker, or review of the patient’s clinical record [24].

We regard the external validity of our findings to be high, since our data represent unselected, consecutive encounters, the majority of which were being assessed by the nursing staff (97%). This is the case in all rural areas of Botswana.

Only one experienced nurse—a family nurse practitioner with 5 years of training—observed all consultations. Although we believe this enhanced consistency in observation, there is also a risk of observation bias. Ascertaining particular observations by the consulting nurse may be difficult, such as her listening for wheezing and stridor. However, chest indrawing can only be noticed reliably by undressing the child. Therefore, we did not include wheezing and stridor in the criteria for acceptable adherence.

Adherence to guidelines on history taking and examination

Acute respiratory infection.

While adherence to guidelines on history taking was suboptimal, clinical examination of cases with acute respiratory infection was poor. According to the acute respiratory infections’ algorithm, all children with cough or difficult breathing should have their respiratory rate checked and be examined for chest indrawing, since both signs represent the single most specific signs for ruling out pneumonia in resource-poor countries [25].

The failure to check the respiratory rate and examine for chest indrawing may be one of the main reasons why so few cases of pneumonia had been diagnosed and so few cases of acute respiratory infection referred to a higher level. Although the algorithm categorizes acute respiratory infection into four groups according to severity, this was, however, not done by the nurses of the study populations. This also may have contributed to the non-referral of any of the pneumonia cases. Careful training of primary health workers in the assessment of respiratory rate and chest indrawing is essential if these clinical findings are to be used as reliable indicators in the management of cases with acute respiratory infection.

Diarrhoea.

In diarrhoea, adherence to guidelines on history taking was fair but poor on examination. This was related mainly to non-compliance to carry out the skin pinch test for dehydration and to take the weight of the child. Although nurses’ adherence to monthly weighing in general is excellent [26], the fact that most of the diarrhoea cases were mild and thus relatively easy to manage may explain the low level of adherence.

The cornerstone in the treatment of diarrhoea is the provision of oral rehydration salts and continued feeding. Despite the finding that most of the cases were mild, the oral rehydration salt prescription rate was high. We assume that health care providers’ familiarity with treatment guidelines and the excellent availability of oral rehydration salts at clinics and health posts contributed to this.

We could not find a significant association with relevant explanatory variables in the multilevel logistic regression analysis. This may be related to the relatively small number of encounters. However, we have in a previous paper [14] shown that for all age groups (n = 2994), clinic staff, who are better qualified than health post staff, adhered significantly better to national treatment guidelines than health post staff. All nurses (including enrolled nurses) have been trained in the management of acute respiratory infection and diarrhoea since the introduction of these programmes.

Adherence to national treatment guidelines

Acute respiratory infection.

The level of inappropriate antibiotic prescription in acute respiratory infection was too high. In general, only a minority of all lower respiratory tract infections should be treated with antibiotics; upper respiratory infections should not. Most respiratory tract infections are self-limiting and antibiotic treatment only slightly modifies their course—according to systematic reviews of randomized controlled trials [27].

Diarrhoea.

All antibiotic prescription in the diarrhoea cases, which mostly were not severe, was assessed as inappropriate. In mild cases of diarrhoea antibiotic treatment has no effect and may enhance the development of resistant strains [28]. However, prescription of metronidazole was assessed as appropriate because of the lack of laboratory facilities for diagnosing amoebiasis and giardiasis, both of which are prevalent in Botswana.

Improving the quality of management

The quality of management of acute respiratory infection and diarrhoea probably depends on a variety of factors, such as health workers’ knowledge, case management skills, motivation, supervision, the existence of incentives for health workers, the provision of necessary supplies and essential drugs, laboratory facilities, and the validity of the algorithm [23]. To improve the quality of management, strategies which target the specific obstacles identified should be applied. While some studies have been unable to show improvement of health workers’ performance by guidelines only [29], others have shown an effect by using clinical training units for the integrated management of sick children [30].

Conclusion

In Botswana primary care, health care providers’ adherence to guidelines on history taking was suboptimal in acute respiratory infection and diarrhoea, but poor on physical examination. The level of inappropriate antibiotic prescription seems too high. There is considerable scope for improving diagnostic and therapeutic management of these important childhood illnesses.

Acknowledgement

Our thanks go to the nursing staff at the health facilities for their hospitality towards the field team and for their co-operation. We acknowledge the contributions of the field surveyors who were part of the study team. We thank Ingvild Dalen, statistician at the Institute of General Practice and Community medicine, University of Oslo, for her statistical analyses; and Per Fugelli, professor of Social Medicine at the Institute of General Practice and Community Medicine, University of Oslo, who was one of the initiators of the Drug Utilization Study and who has contributed to the design of the survey. This study is part of the 1998 Botswana Drug Utilization Study, which has been carried out under a 4-year Health Sector Agreement between Norway (Institute of General Practice and Community Medicine, University of Oslo) and Botswana (Health Research Unit, Ministry of Health), which co-operate on health systems research. The study is financed jointly by both institutions.

References

Author notes

¹University of Oslo, Department of General Practice and Community Medicine, Oslo, Norway, and ²University of Botswana, Department of Population Studies, Gaborone, Botswana