Bulletin of the World Health Organization

Clinical management of children with fever: a cross-sectional study of quality of care in rural Zambia

Karsten Lunze a, Godfrey Biemba b, J Joseph Lawrence c, William B MacLeod d, Kojo Yeboah-Antwi d, Kebby Musokotwane e, Toyin Ajayi f, Simon Mutembo g, Chilunga Puta h, Duncan Earle h, Rick Steketee i & Davidson H Hamer d

a. Department of Medicine, Boston University Medical School, 801 Massachusetts Ave, Boston MA 02119, United States of America (USA).
b. Zambian Centre for Applied Health Research and Development, Lusaka, Zambia.
c. Global Health Corps Fellowship, Zambian Centre for Applied Health Research and Development, Lusaka, Zambia.
d. Center for Global Health and Development, Boston University School of Public Health, Boston, USA.
e. Ministry of Health, Lusaka, Zambia.
f. Commonwealth Care Alliance, Boston, USA.
g. Southern Provincial Medical Office, Ministry of Health, Choma, Zambia.
h. PATH, Lusaka, Zambia.
i. Malaria Control and Elimination Partnership in Africa Program, PATH, Seattle, USA.

Correspondence to Karsten Lunze (email: karsten.lunze@post.harvard.edu).

(Submitted: 20 January 2016 – Revised version received: 10 January 2017 – Accepted: 10 January 2017 – Published online: 05 April 2017.)

Bulletin of the World Health Organization 2017;95:333-342. doi: http://dx.doi.org/10.2471/BLT.16.170092


In Africa, historically, fever in children younger than 5 years has been considered to be due to malaria and has been treated immediately with antimalarial drugs without laboratory confirmation to avoid treatment delays.1 The recent reduction in malaria-related morbidity and mortality2 has prompted a change in clinical management of children presenting with fever.35 In 2010, the World Health Organization (WHO) updated its guidelines from presumptive antimalarial treatment of children with fever to recommend diagnostic testing for malaria, by either microscopy or malaria rapid diagnostic test, to ensure rational use of antimalarials.6

Reflecting global trends, Zambia has recently made progress in reducing its malaria-associated disease burden by about 7% annually from 2000 to 2010,7,8 resulting in over 50% reduction in malaria deaths.9

Health-care services in Zambia are delivered through a five-level hospital system, which includes three tertiary hospitals, 21 provincial or general hospitals and 85 district hospitals, employing specialist staff. These hospitals receive referrals from urban and rural health centres and community-based health posts at the lowest levels of the health-care system throughout the country. Health care in Zambia faces chronic challenges with human resources and inadequate supplies of drugs, equipment and medical supplies. Starting in 1996, the Zambian Ministry of Health adopted an integrated management of childhood illness (IMCI) approach to managing the causes of childhood febrile illness, including malaria, viral or bacterial infections.10 The health ministry has made several adaptations to the generic IMCI guidelines issued by WHO and the United Nations Children’s Fund (UNICEF), to include guidelines for human immunodeficiency virus (HIV), diarrhoea management, use of malaria rapid diagnostic test for classifying fever and newborn care. In the country, the principles of the IMCI guidelines are incorporated into pre-service curricula and training programmes and the approach is implemented in all districts. However, less than 10% of the districts have the minimum level of 80% health workers trained in IMCI as required by the ministry of health.11 As a result, health workers in Zambia face the challenge of managing febrile illness when malaria test results are negative.12

With this study, we aimed to evaluate current practices and standards of care for childhood febrile illness at different levels of the health-care system, in Southern Province, Zambia.


Study site

In 2013, Southern Province had an estimated population of about 1.6 million.13 It is one of 10 provinces in Zambia and is divided into 13 districts.13 Malaria transmission in the province ranges from hypoendemic (less than 10% of children aged 2–9 years are parasitemic) to holoendemic (more than 75% of children aged 2–9 years are parasitemic). In recent years, Southern Province has been the focus of intensive malaria control activities using reactive case detection14 and mass malaria test and treat strategies, with the aim of creating malaria-free areas within the province.15 In 2012, in Southern Province, 44% of children younger than 5 years slept under a treated bednet and 22% of all households reported indoor residual spraying in the previous 12 months.16

Study design

We conducted two types of cross-sectional surveys; one for facilities and one for health workers. Fundamental aspects of the survey design and execution were described previously.17 The facility survey was used to record level of staffing, health services provided by the facility, availability and adequacy of medical equipment, availability of basic drugs and supplies and availability of treatment charts and guidelines. We also reviewed outpatient, inpatient and laboratory or malaria rapid diagnostic test registers to document the number of children consulted at the facility, proportions with febrile illness and proportions of confirmed cases of malaria who presented with a febrile illness in the last month. The health worker survey was used to assess respondent’s training, length of service, access to national guidelines, use of wall charts, algorithms/decision charts or other job aids for managing illnesses, and their practice and knowledge on management of neonatal and child illnesses. In addition to the surveys, we observed the health workers’ interactions with febrile children and their caregivers to assess history taking and examination, investigations performed or ordered to aid diagnosis and actions taken, including treatments and counselling. We recorded the observation by using a structured data collection tool that was designed to assess all relevant steps in taking a history and examining a child, as described in the national IMCI guidelines. We also conducted brief structured exit interviews with caregivers (lasting less than five minutes) to collect de-identified information on demographic characteristics, chief complaint, counselling and drug dispensing practices. During the exit interview, we also reviewed the child’s clinic card and documented the care received including diagnosis made, investigations ordered and results and treatment given. These tools were adapted from previously used tools,17 and were pilot tested in a health centre outside Southern Province. All survey instruments are available from the authors.

The Institutional Review Boards of Boston University Medical Center, ERES Converge, (a local institutional review board in Lusaka, Zambia), the Zambian Ministry of Health and the Southern Provincial Medical Office approved the study.


Our sampling frame included three districts in Southern Province with differing levels of malaria transmission: Kazungula (low), Kalomo (moderate) and Siavonga (high). In the study area, none of the three districts had a level 3 (tertiary) hospital. Therefore, we sampled from level 2 (district hospitals) and level 1 (health centres), which in Zambia serve the highest volume of febrile children.

All health workers attending to febrile children and caregivers 18 years or older were eligible to be included in the study. Health workers caring for febrile children with complications requiring referral, and their caregivers; as well as caregivers unable or unwilling to provide written informed consent were excluded from the study.

Data collection

We collected the data from November to December 2013, a time of year in Zambia when malaria incidence starts to increase. We obtained written informed consent from all study participants in English or Chitonga, the local language in Southern Province.

Study outcomes and analysis

The primary outcome was the proportion of children with febrile illness (temperature above 37.5 °C or history of subjective fever in the previous 48 hours) managed according to the guidelines issued by the health ministry.10 The guidelines include identification of seven signs and symptoms of febrile illness: four danger signs for childhood illnesses in outpatient settings (poor oral intake, vomiting everything, convulsions and decreased level of consciousness); ensuring appropriate combined treatment of all major illnesses; strengthening the counselling of caregivers; and referring severely ill children.10 For our outcome, we defined appropriate management based on Zambia’s Integrated technical guidelines for frontline health workers18 and the Guidelines for the diagnosis and treatment of malaria in Zambia19 (Fig. 1). While both of these sets of guidelines are based on the IMCI guidelines, there is greater depth in the malaria guidelines.19

Fig. 1. Flowchart showing the development of the study’s primary outcome, appropriate fever management
Fig. 1. Flowchart showing the development of the study’s primary outcome, appropriate fever management
Note: Fever management best practice evaluation algorithm derived from the Integrated technical guidelines for frontline health workers18 and the Guidelines for diagnosis and treatment of malaria in Zambia.19

Secondary outcomes included: (i) proportion of children presenting with positive blood smear or rapid diagnostic test result and with medical conditions most commonly associated with non-malaria febrile illness; (ii) proportion of health facilities with numbers and cadres of staff as stated in ministry of health guidelines and established targets needed to manage childhood febrile illness; (iii) proportion of health facilities with essential equipment, supplies and drugs to manage childhood febrile illness; (iv) proportion of health workers in-service and pre-service training in the management of childhood febrile illness, such as IMCI and malaria.

We calculated frequencies for categorical variables and means and standard errors for continuous variables.


We surveyed 24 health facilities (comprising four hospital-affiliated urban health centres and 20 rural health centres) and 53 health workers (12 physicians or clinical officers, 26 nurses, eight environmental health technicians and seven other staff) who provided care for febrile children. Table 1 summarizes the characteristics of the surveyed facilities. We also observed the assessment and management of 161 febrile children, with mean age of 24.1 months. Almost half (49.7%; 80 children) were male.

Diagnostic and treatment capacity

While all 24 surveyed facilities had diagnostic capacity for malaria, 83.3% (20) lacked other diagnostic equipment such as a functional X-ray machine and 75% (18) lacked an instrument for counting respiratory rates. Twenty-three facilities (95.8%) lacked visual job aids, such as wall charts and diagnostic and treatment guidance. While all health facilities had artemether-lumefantrine for treatment of non-severe malaria, only 92% (22) stocked antibiotics for pneumonia, but drug stock-outs were common.


Table 2 presents the characteristics of the surveyed health workers, training they received and the results of their skill assessment. Over half (50.9%; 27 health workers) had recent formal training in either malaria case management, IMCI, or integrated community case management. Less than one third (32.1%; 17 health workers) had access to IMCI guidelines, while 24.5% (13) had access to both integrated community case management as well as malaria treatment guidelines.

Chief complaints

Of the 161 febrile children observed, most presented with symptoms and findings that were consistent with upper respiratory and gastrointestinal infections. The leading chief complaint was fever at 89.4% (144 children) followed by cough at 49.1% (79 children), diarrhoea at 24.8% (40 children), vomiting at 12.4% (20 children) and rash at 9.9% (16 children). Exit interviews revealed that non-pneumonia respiratory infections accounted for the highest proportion of diagnoses with 43% (69 children), followed by diarrhoea with 17% (27 children) and malaria with 10% (16 children).

Skills assessments

The health workers assessed 82.6% of the surveyed children (133) for at least one danger sign, 47.2% (76) for two or more danger signs and 8.1% (13) for all four danger signs (Table 3). None of the 161 children were assessed for all seven signs and symptoms of febrile illness, and only 14.3% (23) were assessed for at least four. While health workers measured temperature in 94% of the children (147 of the total 156 children where temperature was checked), none of the health workers took a medical history and none completed all physical examination tasks as prescribed. Only 14.3% (23 health workers) did more than half of the required fever management tasks.

Diagnosis and treatment

Table 4 and Table 5 present the health-care workers’ diagnosis and treatment of pneumonia and malaria respectively, as reported in exit interviews with caregivers. Only 27.8% (22 of the 79 children with diagnosis of cough or difficulty breathing) had their respiratory rate counted and 47.4% (9 of the 19 children with diagnosis of pneumonia) received appropriate antibiotic treatment. In contrast, over prescription of drugs was common. We found 45.8% (11 of the 19 children with a diagnosis of pneumonia) were inappropriately treated with amoxicillin, while 12.9% (11 of the 85 children with negative malaria rapid diagnostic test or microscopy results) were inaccurately classified as having malaria.

Of all febrile children 57.1% (92) were tested for malaria using rapid diagnostic test or microscopy of which 8.3% (7) tested positive for malaria. All seven children appropriately received artemisinin-based combination therapy. Of 26 (13%) children who had negative rapid diagnostic test results and who received amoxicillin, only one received amoxicillin appropriately.


This study assessed the quality of clinical management of childhood fever at lower levels of the health system (health centres and second level hospitals), which typically manage the majority of childhood febrile disease burden. Upper respiratory tract infections accounted for almost half of febrile children visits and only 10% of children were diagnosed with malaria. While this was not an etiology study, clinical diagnoses were similar to those found in recent childhood fever etiology studies conducted in Africa. For example, in a large Tanzanian study, most (62%; 625 children who presented to outpatient care for fever) were diagnosed with acute respiratory infections, while only 11% (105 children) were diagnosed with malaria.20

In our study, we observed some deviations from practice as recommended by national IMCI guidelines for management of childhood febrile illness. These include inadequate assessment for danger signs, such as limited evaluations to determine fever etiology, over- and under-diagnosis of malaria and inadequate pneumonia management. The health workers evaluated danger signs related to poor oral intake in about half of all children with fever, assessed convulsions or lethargy in about a third, and counted respiratory rate in 28% of cases of suspected pneumonia. These findings contrast with one study conducted in rural Ghana, where the respiratory rate was counted in 4% of children presenting with cough or difficult breathing.21 Upper respiratory tract infections are highly prevalent in Africa and are predominantly non-pneumonia infections that are likely to be of viral etiology.20 In our study, the majority of health workers did not appropriately evaluate and classify children with pneumonia and upper respiratory tract infections, which highlights an important case management challenge. In a recent study of patients with febrile illness presenting to primary health-care facilities in Zambia, we found high proportions of these patients who received overtreatment with antibiotics (61% and 75%, respectively) for those diagnosed with diarrhoea and upper respiratory tract infections,12 indicating a need to improve diagnostic capacity for non-malaria causes of febrile illness and limit antibiotic use to patients with definite bacterial infections. As malaria prevalence declines, high-quality care for febrile children requires consistent use of malaria diagnostic tests and formulation of a differential diagnosis based on symptoms and physical signs.22,23

Despite increasing availability of malaria rapid diagnostic tests in Zambia8 and health-care worker training in their use, little over half of febrile children in this study were tested for malaria with either rapid diagnostic test or microscopy. Of those tested, 13% of children with a negative test result were still classified as having malaria. Managing non-malaria fever continues to pose a challenge globally.24 WHO estimates that in the WHO African Region, over 60% of children with fever receive a blood test (rapid diagnostic test or microscopy) at public facilities.25 In a recent nationwide study in Zambia on fever management, among over 700 patients where half of the children were younger than 5 years, 75% were tested for malaria, and testing was associated with reduced antibiotic prescribing.12 In a Ghanaian study on childhood fever, where malaria rapid diagnostic tests were fully available, less than half of febrile children were given a malaria test. In the same study, 45% of artemisinin-based combination therapy prescriptions conformed to guidelines and were appropriately used to treat malaria.26

In our study, all patients who tested positive using microscopy or a rapid diagnostic test received artemisinin-based combination therapy, in contrast to only one with a negative test. This indicates a higher level of compliance with rapid diagnostic testing results than previously observed in a study in Zambia, where 36% of those with a negative rapid diagnostic test result were prescribed an antimalarial.17 Adherence to rapid diagnostic test results has varied widely in studies conducted in Africa.27 While adherence to positive results tends to be appropriate, compliance with negative results still needs to improve.28

Withholding antimalarial drugs from patients with a negative rapid diagnostic test result seems a safe approach, even for children living in areas highly endemic for malaria.1 While the use of malaria rapid diagnostic testing has reduced the consumption of antimalarial drugs, in some settings, it has increased the use of antibiotics,29 replacing the problem of misuse of antimalarial drugs with antibiotic overtreatment.12 The clinical algorithm for IMCI is designed to reduce unnecessary antibiotic treatment, however, in most cases, not adequately following the IMCI guideline still leads to some overtreatment because of the poor specificity of respiratory rate to diagnose pneumonia in patients with cough or difficult breathing.30 While the clinical overlap of pneumonia with malaria makes their management at peripheral facilities challenging, evidence exists that with effective training, provision of job aids, and regular supportive supervision, community health workers are able to correctly classify malaria and pneumonia and provide appropriate treatment.31

In our study, less than half of children diagnosed with pneumonia received the appropriate amoxicillin treatment, yet it was also often prescribed in children without pneumonia. According to WHO guidelines, about 80% of children with respiratory symptoms do not need antibiotics and can receive supportive care for cough and cold.24 A hospital trial of 900 children in Pakistan showed that even severe pneumonia can be managed safely at community level without antibiotics.32 Several studies in Pakistan have shown that community health workers can recognize chest indrawing and treat children in the community. 33,34 Training community health workers in the management of severe pneumonia can reduce treatment failure by half and can lead to substantial household savings. A study in Pakistan showed that costs were 1.40 United States dollars (US$) for cases managed by a health worker versus US$ 7.60 for treatment costs without. 35 As a result of these and other studies, current IMCI guidelines recommend that pneumonia with chest indrawing but no other danger signs be managed at the peripheral level.24

Our study further documented that few providers had the knowledge and skills to identify danger signs; make the right diagnosis; and initiate appropriate treatment following national guidelines. The results illustrate the need to enhance the availability and awareness of current diagnosis and treatment guidelines that providers will be able to use. District and facility managers can support health-care workers by providing structural support, through systematic coaching and supportive supervision, to recognize signs, symptoms and pertinent history to correctly categorize and treat febrile illness, such as measuring respiratory rate. Access to guidelines and job aids as well as structural and organizational changes such as providing triage facilities dedicated to managing children in larger facilities are ways to provide that support.

Our study has limitations. First, the relatively small numbers of health facilities and health workers included in the study limits the generalizability of our results beyond Southern Province. However, our findings are consistent with challenges related to clinical assessment and management of children with fever reported elsewhere in Zambia12 and other African countries.20 Second, the assessment of appropriateness of health worker assessments relied on maternal recall of symptoms and we did not independently verify disease etiologies in our study. The total number of children presenting with fever recorded in our study was less than anticipated. Finally, our observations of health workers were subject to the reactive effects of study arrangements, potentially modifying their behaviours when aware of being observed. However, the Hawthorne effect – i.e. health-care providers changing their behaviour when being observed – has been found to have limited influence on health workers’ practices during consultations with ill children in similar health-care settings in Benin.36

In conclusion, this study revealed various health systems challenges that may have contributed to a substantial proportion of children not receiving early and appropriate treatment for febrile illness. To improve management of febrile illness in Zambia, interventions need to focus on improving identification of severe illness and strengthening diagnostic capacity (both clinical and laboratory or point-of-care tests). In addition, health workers needs to improve prescribing practices for antibiotic therapy and to enhance the supply chain for better access to IMCI commodities, such as malaria rapid diagnostic tests, antimalarials and antibiotics. Health workers in Zambia need improved guidance and training to appropriately use and respond to malaria test results and to effectively manage childhood febrile illness.

Tools for the management of childhood fevers at peripheral health facilities are available in the form of international and national guidelines. The Zambian health system needs innovative approaches to improve their application in the management of childhood febrile illness. Improvement will ensure appropriate treatment and referral of patients with non-malaria febrile illness, especially for respiratory tract infections. It will also minimize unnecessary use of antimalarial drugs, thus reducing morbidity, mortality and related health-care costs.


We thank the Southern Province medical office, the DCMO and all study participants.


Funding for this study was provided by PATH’s Malaria Control and Elimination Partnership in Africa (MACEPA) through a grant from the Bill & Melinda Gates Foundation (GAT 1756–06232-SUB).

Competing interests:

None declared.