Why first-level health workers fail to follow guidelines for managing severe disease in children in the Coast Region, the United Republic of Tanzania
Nicholas D Walter a, Thomas Lyimo b, Jacek Skarbinski c, Emmy Metta b, Elizeus Kahigwa b, Brendan Flannery d, Scott F Dowell e, Salim Abdulla b & S Patrick Kachur c
a. Epidemic Intelligence Service, Centers for Disease Control and Prevention, 1600 Clifton Road NE, Atlanta, GA 30333, United States of America (USA).
b. Ifakara Health Research and Development Centre, Dar es Salaam, United Republic of Tanzania.
c. Malaria Branch, Centers for Disease Control and Prevention, Atlanta, GA, USA.
d. Respiratory Diseases Branch, Centers for Disease Control and Prevention, Atlanta, GA, USA.
e. Coordinating Office of Global Health, Centers for Disease Control and Prevention, Atlanta, GA, USA.
Correspondence to Nicholas D Walter (e-mail: firstname.lastname@example.org).
(Submitted: 29 January 2008 – Revised version received: 14 May 2008 – Accepted: 26 May 2008 – Published online: 28 November 2008.)
Bulletin of the World Health Organization 2009;87:99-107. doi: 10.2471/BLT.08.050740
Most of the 10 million childhood deaths occurring yearly take place in developing countries, where first-level outpatient health facilities are the primary source of health care.1,2 WHO’s integrated management of childhood illness (IMCI) strategy provides evidence-based guidelines for managing ill children in health facilities lacking sophisticated diagnostic equipment. Health workers use IMCI guidelines to assess children’s condition and classify illness on the basis of simple clinical symptoms and signs.3 The classifications guide treatment and referral.
Adopted in over 100 countries, IMCI improves health worker performance4–10 and may lower mortality.11 However, research has shown that many health workers do not adhere to IMCI guidelines,10,12–16 particularly for the management of severe illness.7,16,17 Adherence is difficult to study, and the reasons that health workers do not follow IMCI guidelines are unclear.10,12,18–21 Decision-making may be shaped by economic, patient-related, training, professional and organizational factors.20,22,23 Understanding non-adherence will help programmes improve IMCI implementation.
We assessed the reasons for non-adherence to IMCI guidelines for the case management of severely ill children at first-level health facilities. To understand how decisions about treatment and referral were made, we evaluated the management of children health workers considered severely ill.
The aim was to prepare for an intervention to improve survival among severely ill children in the Coast Region of the United Republic of Tanzania. With the assistance of the Child Health Unit in the Ministry of Health and Social Welfare (MHSW), we selected four contiguous districts in the Coast Region – Kisarawe, Kibaha urban, Kibaha rural and the south-western portion of Bagamoyo – because they were among the first to implement IMCI (in 2000) and because relatively good roads allow referral care and limit supply shortages (Fig. 1). The combined population is approximately 314 000.24 The site is primarily rural but includes periurban areas. Malaria is endemic; transmission occurs throughout the year. Mortality among children aged < 5 years is 126 per 1000 live births.25 For administrative and surveillance purposes, health facilities report the number of patient visits each month on forms precoded with specific diagnoses (e.g. malaria) through the Health Management Information System (HMIS).
Fig. 1. Health districts, health facilities and study design in study of health worker management of childhood illness, the United Republic of Tanzania, 2006
We included all functioning first-level health facilities (formal, non-hospital setting) who attend to children aged < 5 years. The MHSW, district and regional medical officers, community leaders and health workers identified 64 health facilities, two of which were inaccessible due to flooding (Fig. 1). We distinguished dispensaries (one or two-room clinics) from health centres (larger facilities with overnight beds).
Study design and data collection
We conducted retrospective and prospective case reviews of severely ill children, performed community follow-up, and administered questionnaires to the parents of these children. We also administered questionnaires and conducted qualitative interviews among health workers (Fig. 1).
We conducted retrospective case reviews at all health facilities during 6 weeks in October and November 2006. During visits conducted without advance notice between 09:00 and 13:00 on weekdays, we explained the purpose of the study and invited all health workers caring for children to participate. Using the routine clinical register, health workers identified up to five recent patients aged < 5 years whom they had attended and believed had experienced severe, potentially life-threatening illness. Health worker recall was prompted with a list of IMCI danger signs: loss of consciousness; lethargy; convulsions; inability to drink, eat or nurse; and/or the vomiting of everything consumed. We abstracted age, diagnoses, treatment and referral data from the register and used an open-ended question to prompt health workers’ recall of presenting clinical features. We also inventoried supplies and medications needed to implement IMCI.26,27
We conducted prospective case reviews during 8 weeks between November 2006 and January 2007 in all eight health centres and 10 dispensaries that were selected based on a probability proportionate to the number of children attended in 2005. Health workers identified all children considered severely ill during the prospective study period and entered demographics, age, diagnosis, treatment and referral data into a study register. To minimize register entries and maintain congruence with retrospective case reviews, they also recalled the presenting clinical features during weekly visits (information is missing for some children because some health workers were not located).
Children with fever and at least one IMCI danger sign were assigned the IMCI classification “very severe febrile disease”. The IMCI-recommended treatment for such children includes parenteral quinine and a parenteral broad-spectrum antibiotic (chloramphenicol alone, or benzylpenicillin with gentamicin). Hospital referral is also indicated.
In retrospective case reviews, information on health workers and children was linked; treatment and referral by health workers (whether or not trained in IMCI) were compared. In prospective reviews, we did not record a health worker’s identity to reduce the likelihood of eliciting socially desirable but false responses. Analysis was conducted using SUDAAN, version 9.0 (RTI International, Research Triangle Park, NC, United States of America), accounting for clustering at the health facility and health worker levels. We compared proportions using the χ² test. P < 0.05 was considered significant.
Follow-up of severely ill children
Seven to 14 days after children who were prospectively identified visited a health facility, we made three attempts to locate their parents or guardians. After obtaining informed consent, we administered a standardized questionnaire to parents detailing the child’s post-visit care and current status. Parents ranked different barriers to seeking hospital-based care as “not important”, “important” or “very important”. Those whose children had visited a hospital recalled travel costs and time, and the rest merely estimated them.
Using a principal-component analysis approach previously validated in adjacent health districts, we established a relative index of household socioeconomic status based on 23 questions assessing household asset ownership.28,29 In this technique, orthogonal linear combinations of household asset variables are extracted to generate a normally distributed index with a mean of zero that reflects long-term household wealth. This allows for discrimination between households by socioeconomic status. In prospective case review, we compared the prevalence of demographic and clinical characteristics among referred and non-referred children using multivariate logistic regression to calculate adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Factors predictive of referral at the P < 0.1 level in univariate analyses were included in the multivariate model, which used the PROC RLOGIST procedure based on generalized estimating equations. Models constructed with both unweighted data and data stratified by health facility type produced similar results; only the unweighted model is presented here.
Interviews in health facilities
We administered a standardized questionnaire on beliefs and practices surrounding the management of severe illnesses to 81 of 82 health workers. We also conducted qualitative interviews with 30 health workers in 18 facilities to gain data on complex multifaceted processes of decision-making, which are difficult to measure quantitatively.30,31 Qualitative research was independently designed and directed by an experienced social scientist (EM).
Qualitative and quantitative fieldwork was conducted concurrently and compared after completion of independent analyses. Interviews were conducted in Swahili by two experienced research assistants using a semi-structured interview guide, with probes for clarification. Interviews lasted about 45 minutes and were recorded, transcribed and systematically coded for content analysis.31,32 As additional data were reviewed, patterns emerged and codes were progressively expanded and refined. When further review did not generate additional hypotheses, the coding structure was finalized and applied to all transcripts, which were reviewed and coded by a single investigator (EM).
The institutional review boards of the Ifakara Health Research and Development Centre and of the Centers for Disease Control and Prevention (United States of America) approved this study.
Of the health facilities in this study, 74% were governmental. The WHO-recommended pre-referral parenteral medications that were available at governmental and nongovernmental health facilities, respectively, were benzylpenicillin (93% and 75%), chloramphenicol (86% and 30%), gentamicin (2% and 65%) and quinine (76% and 90%). Supplies were similar at dispensaries and health centres.
Fifty-nine (73%) health workers were clinical officers (2 years of post-secondary training) (Table 1). Health workers at governmental health facilities were more likely to have received IMCI training than those in nongovernmental health facilities (64% versus 18%, respectively; P = 0.0001). Overall, 52% of health workers reported completing an 11-day IMCI training course. Health workers often worked long hours in isolated settings; 70% of them provided emergency care on a 24-hour basis and 52% lived at the health facility. Only 26% reported a supervisory visit within the previous 6 months.
Table 1. Characteristics of health workers caring for severely ill children (n = 81) aged < 5 years at health facilities in the United Republic of Tanzania, 2006
Retrospective and prospective case reviews were completed for 297 and 205 severely ill children, respectively; results are combined, except where indicated (Table 2). Health workers recorded only specific diagnoses (not IMCI classifications) in registers. Nearly all children (478; 96%) were diagnosed with severe malaria or severe pneumonia; only 43 (9%) were diagnosed with both. Health workers consistently treated the specific diagnosis they assigned; occasionally they treated other conditions. Of 349 children diagnosed with severe malaria only, 333 (95%) received an antimalarial and 41 (12%) received an antibiotic; of 86 children diagnosed with severe pneumonia only, 9 (10%) received an antimalarial and all received an antibiotic; of 43 children diagnosed with both severe pneumonia and severe malaria, 38 (88%) received an antimalarial and all received an antibiotic.
Table 2. Characteristics of severely ill children aged < 5 years in retrospective and prospective case review, the United Republic of Tanzania, 2006
Of 409 severely ill children with full presenting clinical information, 240 (59%) met IMCI criteria for very severe febrile disease (Table 3). None received IMCI-appropriate therapy (parenteral broad-spectrum antibiotic and parenteral quinine). Of 47 (20%) who were given an antibiotic, 25 (53%) received benzylpenicillin, 1 (2%) received both benzylpenicillin and parenteral chloramphenicol, and 21 (45%) received either oral amoxicillin or co-trimoxazole. In retrospective case review, children treated by IMCI-trained health workers were more likely to receive both parenteral benzylpenicillin and quinine than those treated by non-IMCI-trained health workers (11% versus 0%, respectively; P < 0.001).
Table 3. Medication administered to children with very severe febrile disease (n = 240) in a retrospective and prospective case review according to health worker diagnosis, the United Republic of Tanzania, 2006a
Of 502 severely ill children in retrospective and prospective case review, 123 (25%) were referred for further treatment. In retrospective case review where provider data was available, children treated by IMCI-trained workers were more likely to be referred than those treated by health workers not trained in IMCI (38% versus 16%, respectively; P = 0.003) (Fig. 2). IMCI training remained predictive of referral after adjustment for health workers’ professional training (adjusted OR: 3.0; 95% CI: 1.7–5.4).
Fig. 2. Referral and health worker IMCI training in retrospective case review of severely ill children aged < 5 years, the United Republic of Tanzania, 2006
Of 205 children in the prospective case review, 38 (19%) were referred to the hospital (Fig. 3). Among the 166 located through community follow-up, referred children were more likely to reach the hospital than non-referred children: 61% (17/28) versus 2% (3/138), respectively; P < 0.01. Referred children were more likely to die than non-referred children (2% versus 18%, respectively; P < 0.001). Overall, 8 (5%) children died.
Fig. 3. Referral and mortality in prospective case review of severely ill children aged < 5 years, the United Republic of Tanzania, 2006
Transportation costs and availability were the barriers most frequently identified as “very important” by parents of children who did not reach the hospital (40% and 21%, respectively), yet transportation time to the hospital was similar among referred and non-referred groups (mean: 2.3 and 2.1 hours respectively, t-test result not significant). Travel costs were comparable as well (mean: 5500 and 4000 United Republic of Tanzania schillings respectively, t-test results not significant). “Waiting lines at the hospital”, “cost of treatment at the hospital” and “poor quality of service at the hospital” were considered “very important” by 17%, 13% and 9%, respectively. Lack of childcare for other children and need for husband’s authorization were each considered “very important” barriers by 1%.
The socioeconomic status index and the type of health facility where children were seen were not associated with referral in the univariate analysis (Table 4). In the multivariate analysis, two factors were associated with referral: diagnosis of severe anaemia (OR: 114; 95% CI: 12–1049) and lethargy or unconsciousness (OR: 4.8; 95% CI: 1.2–19). Although strongly associated with referral, a fatal outcome was not included in the multivariate analysis of predictors because it occurred after the fact.
Table 4. Univariate analysis of factors associated with referral among children located in community follow-up (n = 166), the United Republic of Tanzania, 2006
Health worker interviews
Among health workers, 71% considered reaching the hospital “easy” during the daytime, and 78% believed that referred children were likely to reach the hospital. Nonetheless, 64% reported that they commonly manage severely ill children without referral, and 91% agreed that “certain severe illnesses can be safely managed without referral”. Asked which conditions can be managed without referral, 68% responded severe malaria and 57% indicated severe pneumonia. Some health workers (24%) reported having withheld hospital referral because the parent of the child in question could not feasibly transport the child. No health workers reported withholding referral because they feared the parents would consider them incompetent. Only 5% reported ever withholding referral because the child’s condition appeared hopeless and further care seemed futile.
In qualitative interviews, health workers suggested that benzylpenicillin is fast-acting and effective, but chloramphenicol is unacceptably toxic. Barring medication shortages, health workers expressed confidence in their ability to safely manage severely ill children who do not have severe anaemia, severe dehydration or difficulty breathing. Health workers reported referring children to the hospital primarily for specific therapies not available at their health facility, such as blood transfusion, intravenous fluids or oxygen. They were confident in the quality of referral facilities but considered transportation costs an important barrier that kept children from actually receiving treatment at these facilities. They also reported that they frequently negotiated with parents of severely ill children about whether to refer them or to provide ongoing care through repeated visits to the health facility.
Health workers in the four Tanzanian districts studied rarely adhered to IMCI guidelines for the treatment and referral of severely ill children. They generally treated children according to a single, narrow diagnosis rather than a broad IMCI syndromic classification, and they rarely administered broad-spectrum antibiotics or referred severely ill children for hospital treatment.
Treatment based on a single narrow diagnosis
Health workers consistently administered rational therapy for the narrow diagnosis they made but rarely diagnosed or treated more than one condition. The imprecision of clinical diagnosis in similar settings is well established,33,34 and health workers’ confidence in their diagnostic accuracy probably contributes to missed opportunities to provide potentially life-saving therapies.
The narrow diagnoses used by health workers are the same as those administratively required for HMIS reporting. Rowe et al. have suggested that discrepancies between the diagnoses required for HMIS reporting and IMCI classifications may confuse health workers and contribute to poor adherence to IMCI guidelines.35 Our findings support this hypothesis. A quality-improvement programme in these districts will pilot an IMCI-based register in which both HMIS diagnoses and IMCI classifications are recorded.
Reluctance to administer chloramphenicol
Health workers often failed to give antibiotics when indicated and almost never administered recommended broad-spectrum antibiotic regimens. They considered chloramphenicol “too toxic” for use in children; despite its availability, chloramphenicol was administered to only one of 240 children with IMCI “very severe disease”. Instead, benzylpenicillin was the parenteral antibiotic of choice, despite its inadequate activity against important causes of severe illness (particularly Gram-negative sepsis).36 Health workers’ reluctance to administer chloramphenicol is of particular importance in light of a recent trial demonstrating the superiority of injectable ampicillin plus gentamicin over chloramphenicol for children aged 2–59 months with very severe pneumonia in low-resource settings.37 In our study, gentamicin was available in only one governmental health facility. Supplying gentamicin and encouraging its use with an injectable penicillin would probably be more efficacious and feasible than encouraging health workers in this setting to administer chloramphenicol.
Non-referral of severely ill children
Health workers overwhelmingly disagreed with the IMCI recommendation that all severely ill children be referred. Despite 5% mortality and a death rate equivalent to or greater than the in-hospital mortality recently documented in comparable settings in Kenya and the United Republic of Tanzania,38,39 health workers expressed confidence in their capacity to safely manage most cases of severe malaria and severe pneumonia without referral. Non-referral occurred even though reaching the hospital appeared generally feasible. Health workers’ confidence in the quality of referral care was notable given the documented low level of care provided in district-level hospitals in comparable Kenyan and Tanzanian settings.38,39
Health workers’ disagreement with IMCI referral guidelines represents an important challenge to IMCI implementation. Optimally, health workers would be trained and equipped to manage severe illness when referral is not feasible, to refer when possible, and to know how to distinguish between the two circumstances. Training and supervision programmes should reinforce the necessity of referring severely ill children when possible.
Our findings do provide some cause for optimism. Although adherence to IMCI guidelines was low, IMCI-trained health workers were more likely than those who lacked IMCI-training to refer severely ill children, and they are more likely to administer both a parenteral antibiotic and quinine to children with IMCI “very severe disease”. Despite low adherence to guidelines, health workers did identify and refer the most severely ill children, as evidenced by the nine-fold higher mortality among referred children. Finally, although therapy was inconsistent with IMCI guidelines, it was internally consistent, rather than arbitrary or irrational (i.e. health workers did not make one diagnosis and treat for another).
This study has limitations. Ideally, a study of adherence would include a probability sample of children with confirmed IMCI severe disease classifications. This study included only children thought to be severely ill by the health workers who examined them. Adherence to IMCI treatment and referral guidelines may have been even worse for severely ill children who were not labelled as such by health workers. We cannot verify that all children had severe disease according to IMCI classifications. Nonetheless, because they described IMCI danger signs in 79% of children without prompting and the mortality rate among children was high, we believe that health workers consistently recognized a subset of children more severely ill than others who attended the outpatient department.
Sampling techniques differed in retrospective and prospective case reviews, and this resulted in a slightly different mix of health facility types and ownership. Analyses were repeated with weighting for health facility type and ownership, but the primary findings were unchanged. Finally, this single site cannot be considered representative of a large and diverse country. While key health and socioeconomic indicators (high under-5 mortality, high malaria burden, low income and reliance on subsistence agriculture) are typical of many areas of the United Republic of Tanzania, these districts were selected because IMCI implementation was considered to be robust and referral care feasible. Our findings regarding conflicts between HMIS diagnoses and IMCI classification may be widely generalizable, but different organizational and economic constraints, as well as distance to referral care, may influence health worker adherence differently elsewhere in sub-Saharan Africa. We encourage local evaluation of reasons for health worker non-adherence. Determining which reasons are common across sites and might be more generally applicable would be of benefit to all programme countries.
Improving health workers’ performance is critical to putting evidenced-based interventions into practice.18,40 The large gap between IMCI guidelines and the practices of these Tanzanian health workers parallels similar gaps found in previous studies.13–15,19,41 We identified three reasons contributing to health workers’ non-adherence to IMCI guidelines: (i) the use of single, narrow diagnoses rather than IMCI classifications; (ii) the belief that chloramphenicol is unacceptably toxic, and (ii) the perception that referring severely ill children is often unnecessary. The United Republic of Tanzania MHSW recently initiated a reassessment of the national adaptation of IMCI guidelines; through this process, an attempt should be made to understand and overcome the reasons for health worker non-adherence. ■
We recognize the professionalism and dedication of fieldworkers and staff at the Ifakara Health Research and Development Centre. We thank Michael S Deming and Alexander K Rowe for their insightful critiques of this manuscript, and appreciate Elaine Scallan and Joseph E Logan for their generous editorial contributions.
Competing interests: None declared.
- Black RE, Morris SS, Bryce J. Where and why are 10 million children dying every year? Lancet 2003; 361: 2226-34 doi: 10.1016/S0140-6736(03)13779-8 pmid: 12842379.
- Jones G, Steketee RW, Black RE, Bhutta ZA, Morris SS. How many child deaths can we prevent this year? Lancet 2003; 362: 65-71 doi: 10.1016/S0140-6736(03)13811-1 pmid: 12853204.
- Gove S. Integrated management of childhood illness by outpatient health workers: technical basis and overview. Bull World Health Organ 1997; 75: 7-24 pmid: 9529714.
- Gouws E, Bryce J, Habicht JP, Amaral J, Pariyo G, Schellenberg JA, et al., et al. Improving antimicrobial use among health workers in first-level facilities: results from the multi-country evaluation of the Integrated Management of Childhood Illness strategy. Bull World Health Organ 2004; 82: 509-15 pmid: 15508195.
- Amaral J, Gouws E, Bryce J, Jorge A, Leite M, da Cunha ALA, et al., et al. Effect of Integrated Management of Childhood Illness (IMCI) on health worker performance in northeast-Brazil. Cad Saude Publica 2004; 20: s209-19 pmid: 15608935.
- El Arifeen S, Blum LS, Hoque DM, Chowdhury EK, Khan R, Black RE, et al., et al. Integrated Management of Childhood Illness (IMCI) in Bangladesh: early findings from a cluster-randomised study. Lancet 2004; 364: 1595-602 doi: 10.1016/S0140-6736(04)17312-1 pmid: 15519629.
- Armstrong Schellenberg J, Bryce J, de Savigny D, Lambrechts T, Mbuya C, Mgalula L, et al., et al. The effect of Integrated Management of Childhood Illness on observed quality of care of under-fives in rural Tanzania. Health Policy Plan 2004; 19: 1-10 doi: 10.1093/heapol/czh001 pmid: 14679280.
- Bryce J, Gouws E, Adam T, Black RE, Schellenberg JA, Manzi F, et al., et al. Improving quality and efficiency of facility-based child health care through Integrated Management of Childhood Illness in Tanzania. Health Policy Plan 2005; 20: i69-76 doi: 10.1093/heapol/czi053 pmid: 16306072.
- Bryce J, Victora CG, Habicht J-P, Black RE, Scherpbier RW, Multi-Country Evaluation (MCEITA).. Programmatic pathways to child survival: results of a multi-country evaluation of Integrated Management of Childhood Illness. Health Policy Plan 2005; 20: i5-17 doi: 10.1093/heapol/czi055 pmid: 16306070.
- Naimoli JF, Rowe AK, Lyaghfouri A, Larbi R, Lamrani LA. Effect of the Integrated Management of Childhood Illness strategy on health care quality in Morocco. Int J Qual Health Care 2006; 18: 134-44 doi: 10.1093/intqhc/mzi097 pmid: 16423842.
- Armstrong Schellenberg JR, Adam T, Mshinda H, Masanja H, Kabadi G, Mukasa O, et al., et al. Effectiveness and cost of facility-based Integrated Management of Childhood Illness (IMCI) in Tanzania. Lancet 2004; 364: 1583-94 doi: 10.1016/S0140-6736(04)17311-X pmid: 15519628.
- Eriksen J, Tomson G, Mujinja P, Warsame MY, Jahn A, Gustafsson LL. Assessing health worker performance in malaria case management of underfives at health facilities in a rural Tanzanian district. Trop Med Int Health 2007; 12: 52-61 pmid: 17207148.
- Font F, Quinto L, Masanja H, Nathan R, Ascaso C, Menendez C, et al., et al. Paediatric referrals in rural Tanzania: the Kilombero District Study – a case series. BMC Int Health Hum Rights 2002; 2: 4- doi: 10.1186/1472-698X-2-4 pmid: 11983024.
- Pariyo GW, Gouws E, Bryce J, Burnham G, Uganda IMCI Impact Study Team.. Improving facility-based care for sick children in Uganda: training is not enough. Health Policy Plan 2005; 20: i58-68 doi: 10.1093/heapol/czi051 pmid: 16306071.
- Zurovac D, Rowe AK. Quality of treatment for febrile illness among children at outpatient facilities in sub-Saharan Africa. Ann Trop Med Parasitol 2006; 100: 283-96 doi: 10.1179/136485906X105633 pmid: 16762109.
- Centers for Disease Control and Prevention. Health worker performance after training in Integrated Management of Childhood Illness – Western Province, Kenya, 1996-1997. MMWR Morb Mortal Wkly Rep 1998; 47: 998-1001 pmid: 9843326.
- Herman E. Final report: assessment of the quality of outpatient management of childhood illness in governmental and non-governmental health facilities – Bungoma District, Kenya. Atlanta: Centers for Disease Control and Prevention; 2002.
- Haines A, Kuruvilla S, Borchert M. Bridging the implementation gap between knowledge and action for health. Bull World Health Organ 2004; 82: 724-32 pmid: 15643791.
- Osterholt DM, Rowe AK, Hamel MJ, Flanders WD, Mkandala C, Marum LH, et al., et al. Predictors of treatment error for children with uncomplicated malaria seen as outpatients in Blantyre district, Malawi. Trop Med Int Health 2006; 11: 1147-56 doi: 10.1111/j.1365-3156.2006.01666.x pmid: 16903878.
- Radyowijati A, Haak H. Improving antibiotic use in low-income countries: an overview of evidence on determinants. Soc Sci Med 2003; 57: 733-44 doi: 10.1016/S0277-9536(02)00422-7 pmid: 12821020.
- Rowe AK, Onikpo F, Lama M, Deming MS. Risk and protective factors for two types of error in the treatment of children with fever at outpatient health facilities in Benin. Int J Epidemiol 2003; 32: 296-303 doi: 10.1093/ije/dyg063 pmid: 12714553.
- Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH, Abboud P, et al., et al. Why don’t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999; 282: 1458-65 doi: 10.1001/jama.282.15.1458 pmid: 10535437.
- Rowe AK, de Savigny D, Lanata C, Victora CG. How can we achieve and maintain high-quality performance of health care workers in low-resource settings? Lancet 2005; 366: 1026-35 doi: 10.1016/S0140-6736(05)67028-6 pmid: 16168785.
- 2002 population and housing census. Vol 7: Village and street statistics. Dar es Salaam: Tanzanian http://www.nbs.go.tz/Village_Statistics/Village_Statistics.htm [accessed on 31 October 2008].
- Tanzania demographic and health survey 2004–2005. Dar es Salaam: Tanzanian National Bureau of Statistics; 2005. Available from: http://www.measuredhs.com/pubs/pub_details.cfm?ID=566 [accessed 3 November 2008].
- Indicators for IMCI at first-level health facilities. World Health Organization; Geneva: 2006. Available from: http://www.who.int/imci-mce/Publications/im_p4.pdf [accessed 3 November 2008].
- Gouws E, Bryce J, Pariyo G, Armstrong Schellenberg J, Amaral J, Habicht JP. Measuring the quality of child health care at first-level facilities. Soc Sci Med 2005; 61: 613-25 doi: 10.1016/j.socscimed.2004.12.019 pmid: 15899320.
- Filmer D, Pritchett LH. Estimating wealth effects without expenditure data – or tears: an application to educational enrollments in states of India. Demography 2001; 38: 115-32 pmid: 11227840.
- Njau JD, Goodman C, Kachur SP, Palmer N, Khatib RA, Abdulla S, et al., et al. Fever treatment and household wealth: the challenge posed for rolling out combination therapy for malaria. Trop Med Int Health 2006; 11: 299-313 doi: 10.1111/j.1365-3156.2006.01569.x pmid: 16553910.
- Berkwits M, Aronowitz R. Different questions beg different methods. J Gen Intern Med 1995; 10: 409-10 doi: 10.1007/BF02599845 pmid: 7472693.
- Bradley EH, Curry LA, Devers KJ. Qualitative data analysis for health services research: developing taxonomy, themes, and theory. Health Serv Res 2007; 42: 1758-72 doi: 10.1111/j.1475-6773.2006.00684.x pmid: 17286625.
- Strauss A, Corbin J. Basics of qualitative research: grounded theory procedures and techniques. Newbury Park, CA: Sage Publications; 1990.
- Berkley JA, Maitland K, Mwangi I, Ngetsa C, Mwarumba S, Lowe BS, et al., et al. Use of clinical syndromes to target antibiotic prescribing in seriously ill children in malaria endemic area: observational study. BMJ 2005; 330: 995- doi: 10.1136/bmj.38408.471991.8F pmid: 15797893.
- Reyburn H, Mbatia R, Drakeley C, Carneiro I, Mwakasungula E, Mwerinde O, et al., et al. Overdiagnosis of malaria in patients with severe febrile illness in Tanzania: a prospective study. BMJ 2004; 329: 1212- doi: 10.1136/bmj.38251.658229.55 pmid: 15542534.
- Rowe AK, Hirnschall G, Lambrechts T, Bryce J. Linking the integrated management of childhood illness (IMCI) and health information system (HIS) classifications: issues and options. Bull World Health Organ 1999; 77: 988-95 pmid: 10680246.
- Berkley JA, Lowe BS, Mwangi I, Williams T, Bauni E, Mwarumba S, et al., et al. Bacteremia among children admitted to a rural hospital in Kenya. N Engl J Med 2005; 352: 39-47 doi: 10.1056/NEJMoa040275 pmid: 15635111.
- Asghar R, Banajeh S, Egas J, Hibberd P, Iqbal I, Katep-Bwalya M, et al., et al. Chloramphenicol versus ampicillin plus gentamicin for community acquired very severe pneumonia among children aged 2–59 months in low resource settings: multicentre randomised controlled trial (SPEAR study). BMJ 2008; 336: 80-4 doi: 10.1136/bmj.39421.435949.BE pmid: 18182412.
- English M, Esamai F, Wasunna A, Were F, Ogutu B, Wamae A, et al., et al. Assessment of inpatient paediatric care in first referral level hospitals in 13 districts in Kenya. Lancet 2004; 363: 1948-53 doi: 10.1016/S0140-6736(04)16408-8 pmid: 15194254.
- Reyburn H, Mwakasungula E, Chonya S, Mtei F, Bygbjerg I, Poulsen A, et al., et al. Clinical assessment and treatment in paediatric wards in the north-east of the United Republic of Tanzania. Bull World Health Organ 2008; 86: 132-9 doi: 10.2471/BLT.07.041723 pmid: 18297168.
- Mulholland K. Childhood pneumonia mortality – a permanent global emergency. Lancet 2007; 370: 285-9 doi: 10.1016/S0140-6736(07)61130-1 pmid: 17658399.
- Rowe AK, Onikpo F, Lama M, Cokou F, Deming MS. Management of childhood illness at health facilities in Benin: problems and their causes. Am J Public Health 2001; 91: 1625-35 doi: 10.2105/AJPH.91.10.1625 pmid: 11574325.