International Journal for Quality in Health Care Advance Access originally published online on January 19, 2006
International Journal for Quality in Health Care 2006 18(2):134-144; doi:10.1093/intqhc/mzi097
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Review Article
Effect of the Integrated Management of Childhood Illness strategy on health care quality in Morocco
1 Department of Health, Nutrition, Population, The World Bank, Washington, DC, 2 Global Immunization Division, National Center of Immunization, 3 Division of Parasitic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, GA, USA and 4 Service de Protection de la Santé de l’Enfant, Division de la Santé Maternelle et Infantile, Ministère de la Santé, Rabat, Morocco
Objective. To evaluate an intervention to promote health workers’ use of the World Health Organization’s Integrated Management of Childhood Illness clinical guidelines and to identify other factors influencing quality of care received by Moroccan children.
Setting. Public outpatient health facilities.
Design. Cross-sectional survey of consultations with sick children under 5 years old at facilities in two intervention and two comparison provinces in April 2000 (6–12 months after intervention). Consultations were observed, children’s caretakers and health workers were interviewed, and children were re-examined by a ‘gold standard’ study clinician.
Study participants. Probability sample of 467 consultations (97.9% participation) performed by 101 health workers in 62 facilities.
Intervention. Health workers received in-service training with job aids and a follow-up visit with feedback 4–6 weeks after training.
Main outcome measures. Index of overall guideline adherence (mean percentage of recommended tasks that were done per child) and the percentage of children requiring antibiotics correctly prescribed antibiotics.
Results. Quality of care was better in intervention provinces, according to the adherence index (79.7 versus 19.5%, P < 0.0001), correct prescription of antibiotics (60.8 versus 31.3%, P = 0.0013), and other indicators. Multivariate modeling revealed a variety of factors significantly associated with quality, including health worker attributes (pre-service training, residence in government-subsidized housing, sex, and opinions) and child/consultation attributes (child’s age and temperature, number of chief complaints, and caretaker type).
Conclusions. Exposure to the intervention was strongly associated with adherence to the guidelines and correct prescribing of antibiotics 6–12 months after exposure. Many other factors may influence health worker performance.
Keywords: adherence to clinical practice guidelines, Integrated Management of Childhood Illness, Morocco, child health services
Address reprint requests to Joseph F. Naimoli, The World Bank, 1818 H Street, NW, Washington, DC 20433, USA. E-mail: jnaimoli{at}worldbank.org
Accepted for publication November 28, 2005.
To reduce child mortality in developing countries, the World Health Organization (WHO) and other partners developed the Integrated Management of Childhood Illness (IMCI) strategy in the mid-1990s [1]. The strategy promotes health workers’ use of evidence-based clinical guidelines in outpatient settings to identify and treat common causes of child deaths (e.g. pneumonia, diarrhea, and malaria) and to counsel parents on ways to prevent disease and promote growth. To implement the guidelines, WHO recommends in-service training with job aids and a follow-up visit with feedback after training. As of December 2002, 97 countries were implementing IMCI (T. Lambrechts, WHO, personal communication), and there is considerable interest in quantifying the strategy’s effectiveness. Results from a multi-country evaluation have demonstrated that the strategy can improve the quality of care at health facilities [2–4] and seems to reduce child mortality [5].
In 1997, the Moroccan Ministry of Health began implementing the guidelines in a pilot project. An end-of-project evaluation was mounted to guide expansion in Morocco and contribute information more generally on the strategy’s effectiveness. Although the primary objective was to assess the effect of the intervention on health care quality, we were also concerned that other factors might influence guideline use. Few methodologically rigorous studies of the determinants of guideline use have been conducted in developing countries [6–10].
 |
Methods |
|---|
 Top Methods ResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
The IMCI intervention
The Ministry implemented the intervention according to WHO recommendations, except that clinical training lasted 12 days instead of 11. From May 1997 to December 1998, the Ministry adapted WHO’s generic guidelines, training materials, and job aids (a chart booklet and wall chart summarizing the guidelines; a two-page form where health workers record patient clinical findings, disease classifications, and treatments; and a mother’s card to help give counseling messages) and developed a model of patient flow. The Ministry selected two provinces (Agadir-Ida ou Tanane and Meknes el Menzeh) as pilot sites. Both had a relatively high population density and child mortality rate and were judged to have adequate infrastructure (e.g. functional health facilities) and motivated provincial health management teams. Thirty-two physician facilitators were trained through two ‘training of trainers’ courses, which included 12 days of clinical training and 5 days on facilitation techniques. Between February 1999 and February 2000, 282 health workers were trained through 12 clinical courses. Training included didactic sessions, role-play, practice with videos of patients and supervised clinical practice in health facilities. Health workers were visited 4–6 weeks after training to reinforce new skills, solve problems, and provide feedback to health workers and provincial health management teams.
Study design
We conducted a cross-sectional, stratified, cluster survey of outpatient facilities in the two pilot provinces (intervention group) and in two similar provinces (Tetouan and Larache) (comparison group). We selected comparison sites using the same criteria as the pilot sites. A cluster was defined as all ill children seen at a health facility during one working day (i.e., Monday–Friday) during the study period. In both study groups, public sector facilities offering outpatient curative services to ill children under 5 years old were eligible for inclusion. We selected 32 (54%) of 59 facilities in the intervention group and 32 (58%) of 55 facilities in the comparison group by systematic sampling [11]. Each facility was visited once for an entire day, and visit dates were selected by systematic sampling. The study period was April 10–21, 2000, which was about 6–12 months after training and follow-up visits.
Data collection
We trained eight survey teams (four persons each) in data collection methods. Training concluded once surveyor and study investigator agreement of 90% on practice results was achieved (1 week). Survey visits to health facilities were unannounced. Surveyors were not supervisors of facilities visited and were not blinded to intervention or comparison area. Survey teams arrived at facilities before opening (8:30 am) and remained until closing (6:30 pm). Surveyors collected data using five methods: (i) checklist-guided direct observation of clinical consultations; (ii) standardized, structured exit interviews with caretakers (usually the mother); (iii) re-examination of patients by a study clinician to obtain a ‘gold standard’ clinical evaluation; (iv) checklist-guided audit of facility supplies and equipment; and (v) standardized, structured interviews with health workers at day’s end. After ill children were re-examined, the study clinician provided appropriate medications free of charge to any child in need of treatment. Data collection instruments were adapted from those used in Benin by a team from the Centers for Disease Control and Prevention, who contributed to the development of the original WHO instruments [7].
Definitions
Morocco’s adaptation of IMCI guidelines served as the standard of health care quality for our study, as it was the Ministry’s intention to implement them nationally [12]. The integrated guidelines were similar to disease-specific guidelines that existed before the pilot project. For our descriptive analysis (Table 1), we used WHO indicators of appropriate guideline use and functional capacity of health facilities [13]. For our multivariate analysis to evaluate the intervention and identify other factors that influence quality, we used two outcome measures: an index of overall guideline adherence (continuous variable from 0 to 100%), and the appropriateness of antibiotic prescription (correct or incorrect) among children requiring antibiotics. We defined the adherence index as the percentage of recommended clinical tasks completed for each child. The index included 32 equally weighted items (Table 2). Not all tasks were needed for every child. Antibiotic prescription was considered correct if the health worker prescribed all antibiotics indicated by the guidelines (first- or second-line antibiotic was acceptable), as determined by the ‘gold standard’ study clinician. Dosage was not considered, because our primary concerns related to choice of medicine.
|
|
Analysis
Data were double entered and verified using EpiInfo [14]. Analyses were restricted to sick children 2–59 months old coming for an initial consultation for the presenting problem. Descriptive analyses in Tables 1 and 3 were performed with SUDAAN, which accounts for the clustered sampling design and sampling weights (weight = 1/probability of selection) [15].
|
For the multivariate analysis of the adherence index, which had a bimodal distribution (Figure 1), we created a series of ‘bivariate’ linear regression models (i.e. models contained the study group variable (IMCI or non-IMCI) plus the variable(s) for one other factor). This approach ensured that the residuals were approximately normally distributed. To adjust for confounding, factors with a P-value <0.10 from the bivariate analysis were entered into a multivariate model. Logical interactions were tested and included if the P-value was <0.05. Modeling was performed with SUDAAN, which uses generalized estimating equations to adjust for correlation of treatment quality among children seen by the same health worker. An exchangeable correlation structure was used. Models were weighted (weight = 1/probability of selection). For the multivariable analysis of correct antibiotic prescription, we created a series of univariate logistic regression models with SUDAAN (as above). To adjust for confounding, factors with a P-value <0.10 from the univariate analysis were entered into a multivariate model. Logical interactions were tested and included if the P-value was <0.05. Factors that did not initially meet the entrance criteria for the multivariate analysis (i.e., univariate P-value 
0.10) were then entered into the multivariate model, one at a time, and retained if their addition to the model changed the odds ratio (OR) of any already-included factor by >20%.
|
In the analysis of both outcomes, we excluded variables for consultation duration and caseload. Although they met the entrance criteria for multivariate analyses, they were strongly correlated with the intervention variable (Table 4). We also excluded the variable for correct diagnosis, because we believed it was in the causal pathway between other independent variables and treatment quality. All hypothesis testing and confidence interval (CI) estimation were done with an alpha level of 0.05.
|
Ethical approval
The study was considered evaluation rather than research and therefore did not require approval by the research committee on ethics in Morocco or CDC’s institutional review board.
|
Results |
|---|
|
TopMethodsResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
Enrollment
Of the 64 health facilities in the study, 62 received at least one sick child during the survey (31 intervention, 31 comparison). Of 523 ill children under 5 years old brought to a facility, surveyors enrolled and observed consultations for 512 (i.e., 97.9% participation). Of these, 467 were 2–59 months old and had come for an initial consultation for the presenting illness (242 intervention, 225 comparison). The 467 consultations were performed by 101 health workers (53 intervention, 48 comparison). All intervention group health workers had received IMCI training.
Descriptive results
Health facilities and workers in both study groups were similar in many respects. Intervention group health facilities, however, were more likely to be dispensaries, have vaccines available, and (as expected) have job aids (Table 3). Intervention group health workers were less likely to be physicians and more likely to be nursing aides.
Children in the intervention group received better quality care than children in the comparison group, according to a wide variety of indicators (Table 1). Intervention group health workers, however, did not always follow the new guidelines.
Multivariate analyses
Adherence index The multivariate analysis of overall adherence to the guidelines included 467 sick children seen by 101 health workers in 62 facilities. The mean percentage of recommended tasks performed was 52.2% [79.4% intervention (95% CI = 74.7–84.0%), 20.9% comparison (95% CI = 17.3–24.5%)] (Figure 1). The final model included 12 factors (of 56 examined) and was based on 445 children (22 children excluded because of missing values for predictors) (Table 5). Adherence was strongly associated with the intervention. In addition, adherence was significantly better for nurses (compared with physicians), female health workers, health workers living in government-subsidized housing, health workers who did not report lack of supervision as a problem, younger children, children with more complaints and children accompanied by mothers. The multiple R2 for the final model was 83.6%.
|
Prescribing practice The multivariate analysis of the appropriateness of antibiotic prescribing included 187 sick children seen by 75 health workers in 51 health facilities. The final model included seven factors (of 54 examined) and was based on 180 children (seven children excluded because of missing values for predictors) (Table 6). Correct prescribing was strongly associated with exposure to the intervention. Correct prescribing was also significantly better for physicians, children with a high fever, younger children, and children with more complaints reported by the child’s caretaker. In addition, univariate results for factors excluded from the multivariate model revealed that prescribing was significantly better when the illness was correctly diagnosed, consultation times were longer, and caseloads lower (although the last two factors were strongly related to study group) (Table 4).
|
|
Discussion |
|---|
|
TopMethodsResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
Effect of the IMCI intervention on quality of care
The major finding of this study was that exposure to the intervention was strongly associated with both overall adherence to the guidelines and correct prescribing of antibiotics 6–12 months after exposure. Health workers in both study groups had previously been exposed, through pre- and in-service training, to the majority of clinical tasks brought together by the IMCI guidelines (Table 1). Consequently, we believe that our two outcome variables are reasonable measures for assessing differences between the two groups. Results for many performance indicators were remarkably similar to those from IMCI studies with a similar design from Tanzania, Uganda, and Brazil [3,4]. Systematic reviews of studies in developing countries with more rigorous designs (e.g. randomized controlled trials), none of which evaluated the IMCI strategy, revealed that educational interventions had mixed results, ranging from <10% improvement in guideline adherence to >25% improvement [16]. Our results also indicated, however, that health workers in the intervention group did not always follow the guidelines, which suggests that the intervention may be necessary but insufficient to achieve superior levels of quality.
Other predictors of quality of care
Child’s age There may be several reasons why younger children received better quality care. If they are perceived to be more vulnerable, then health workers might follow the guidelines more carefully. Alternatively, older children, particularly those with mild disease, may simply have been more difficult to manage, because they often are more wakeful and bustling during consultations. In Benin (before IMCI), Rowe and colleagues found that younger children were more likely to be vaccinated and correctly treated for pneumonia [7], but younger children with malaria were less likely to be treated with an antimalarial [6]. A study of Kenyan children with malaria found that younger children were less likely to be treated with a non-recommended antimalarial [10].
Child’s symptoms One explanation for the positive relationship between the number of main symptoms mentioned by the parent (an indicator of case complexity, the amount of information readily available to the health worker, and perhaps perceived illness severity), and our two outcomes is that health workers perform better when more information is available. Alternatively, health workers may have perceived the guidelines to be more relevant for cases with greater complexity or severity. Although the actual number of symptoms was not associated with either outcome, a caretaker’s description of the illness may influence the health worker’s response. Or, perhaps, health workers perceive parents who mention multiple symptoms as more demanding, which may result in a more comprehensive consultation. The association between the child’s temperature and prescribing practice probably reflects health workers’ perception that high fever is a sign of serious illness, which should be treated with antimicrobials. Similar findings about temperature and prescribing behavior have been reported from studies in Benin and the Central African Republic [6–8].
Health worker type Interestingly, health workers with a higher level of pre-service training (physicians) were less likely to adhere to guidelines overall but more likely to prescribe antibiotics to children who needed them. Studies of the management of malaria in Kenya [10] and malaria and diarrhea in Benin [6,7] found that health workers with higher pre-service training were less likely to prescribe recommended treatments. The association of higher pre-service training and lower adherence may be explained by these health workers thinking that guidelines are inferior to clinical experience and professional judgement. Some may even believe that guidelines are not actually intended for physicians, or that guidelines are ‘suggestions’, not recommendations. Nurses (and to a greater extent, nursing aides), however, may be more comfortable with guidelines, because they may perceive them to be based on information and expertise they do not possess. Also, nurses and aides are probably trained and socialized to follow medical orders, and adhering to guidelines is a similar behavior.
Latent health worker factors This category included factors that reflected health worker motivation and opinions about work-related problems, and the results show several interesting tendencies. First, government-subsidized housing for health workers was associated with better adherence. Surprisingly little research has been done in low-income countries on the impact of salary and other job benefits on performance. Second, we found lower adherence associated with health workers reporting that a lack of supervision was a problem. This association might reflect the usefulness of supervision; however, when examined directly, supervision was not significantly associated with either adherence (results not shown) or correct antibiotic prescribing (Table 4). Third, we found significant univariate associations with other reported problems (lack of equipment and medicines) that disappeared when included in the final multivariate models. Although such results could simply be due to confounding, another possibility is that they were in the causal pathway between other factors and the outcomes.
|
Limitations |
|---|
|
TopMethodsResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
Our study had several important limitations. First, we only assessed clinical behavior in the short-term, and we cannot predict how performance might evolve after the first year. Second, provinces, health facilities, and workers were not selected randomly to receive the intervention, and differences in the health system between the study groups could explain some of the observed effect of the intervention. Also, without a pre-intervention measurement of performance, we were unable to control for differences between the two groups before IMCI was implemented. Third, our method of observation may have introduced observer error and misclassification of children’s symptoms (e.g. if respiratory rate changed from the consultation to the study clinician’s re-examination), which may underestimate health care quality [7]. The method was also susceptible to subject reactivity to the presence of an outside observer, which may overestimate quality [17,18].
Fourth, although no association might exist between adherence to the guidelines and latent characteristics of health workers, the possibility that measurement error may explain our finding cannot be ruled out. Experience in measuring such characteristics in developing countries is very limited [7,19,20], and the questions we asked may have been misinterpreted. Furthermore, the questions posed may not have captured the most important latent factors underlying performance in these settings. Finally, statistical associations do not prove causality, and there is little additional evidence available to help confirm our results. More in-depth, qualitative investigations of health worker practices, similar to the kind conducted in Nepal, Benin, and the United States [7,21,22] are needed to clarify complex relationships.
|
Conclusions |
|---|
|
TopMethodsResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
There are several important implications of this study for policy-making, program implementation, and further research. First, some kind of continuing education on the use of these guidelines, combined with on-the-job feedback, seems to improve adherence. The challenge that Morocco and many other countries face, however, is how to scale up such interventions effectively. Second, characteristics of participants in the clinical consultation may evoke different responses from health workers. Although innate characteristics of mothers and children cannot be changed, these factors may be addressed, in part, by refining the recommended intervention, particularly by emphasizing the importance of adequate history taking for all children, and by improving communication with clients throughout the clinical consultation. Third, variation in adherence by health worker type suggests a possible conflict, or at least some inconsistencies, between the integrated case management approach and other clinical rule systems already in use. Introducing these guidelines to medical students, for instance, may increase their legitimacy in the eyes of future practitioners and their supervisors.
We also learned an important methodological lesson. We suspected that adherence would be greater in the intervention group than in the comparison group and therefore were not surprised by the bimodal distribution of the adherence index. Univariate analyses for all factors (except study group), however, led to a violation of the normality assumption for linear regression. This required an adjustment in the first stage of our modeling procedure—from univariate to bivariate modeling—to avoid this violation. Future researchers should be alert to this possibility, particularly when the guidelines and intervention are new.
Finally, we believe that research on guideline use can be enhanced by the use of multiple indicators of quality. In our study, the prescribing practice outcome provides readers, particularly clinicians, with an easily identifiable point of entry into a complex set of guidelines for a critical behavior that bears a direct relationship to the ultimate clinical outcome of the patient. These indicators can alert managers to major medical errors occurring within a health system and permit comparisons across health systems. Their inconvenience is that they say nothing about the process used in making the treatment decision. In contrast, our adherence index provides a single number to represent a complex set of clinical behaviors and allows investigators to use all their data. Such variables are appealing to those with an interest in the policy and programmatic implications of guideline use. Consequently, our work may also contribute to that of other researchers who are exploring how best to construct, analyze and use these indices, particularly for IMCI [23].
|
Acknowledgements |
|---|
|
TopMethodsResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
We thank the Moroccan survey teams for their assistance with data collection; Fatiha Bouricha, Nora Baallal, Dr N. Diouri, and the firm of ‘Santé Consultant’ for their assistance with data management; Dr Samantha Rowe for her suggestions on the statistical analysis; Dr Tyane, Dr A. Zerrari, Dr H. Chekli, Ms Michele Moloney-Kitts, and Ms Helene Rippey for their sponsorship and overall direction of this research; Dr Theo Lippeveld and the Moroccan-based staff of John Snow Inc. for their technical inputs and assistance in the conduct of all field-based research operations; Dr Colette Geslin for her assistance in the training of the surveyors; Dr Michael Reich, Dr Heather Palmer, and Dr Winnie Chi-Man Yip for their technical advice and guidance at all stages of the research; and Dr James Heiby and Mr Bob Emrey for their encouragement and sustained support, without which this study would not have been completed.
The findings, interpretations, and conclusions expressed in this article are entirely those of the authors and do not represent the views of the World Bank, its Executive Directors, or the countries they represent.
The evaluation was funded by the US Agency for International Development, Morocco.
Preliminary findings were presented at the Ministry of Health, Rabat, Morocco, July 2001; at the United States Agency for International Development, Washington, DC, August 2001; and at the Harvard School of Public Health, Boston, MA, September 2001.
|
References |
|---|
|
TopMethodsResultsDiscussionLimitationsConclusionsAcknowledgementsReferences |
|---|
- Gove S. Integrated Management of Childhood Illness by outpatient health workers: technical basis and overview. The WHO Working Group on Guidelines for Integrated Management of the Sick Child. Bull World Health Organ 1997; 75 (suppl. 1): 7–24.
- Arifeen SE, Blum LS, Hoque DME et al. Integrated Management of Childhood Illness (IMCI) in Bangladesh: early findings from a cluster-randomized study. Lancet 2004; 364: 1595–1602.[CrossRef][Medline]
- Gouws E, Bryce J, Habicht JP 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: 509–515.[Medline]
- Tanzania IMCI Multi-Country Evaluation Health Facility Survey Study Group. The effect of Integrated Management of Childhood Illness on observed quality of care of under-fives in rural Tanzania. Health Policy Plann 2004; 19: 1–10.[Medline]
- Armstrong Schellenberg JRM, Adam T, Mshinda H et al. Effectiveness and cost of facility-based Integrated Management of Childhood Illness (IMCI) in Tanzania. Lancet 2004; 364: 1583–1594.[CrossRef][Medline]
- 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.
[Abstract/Free Full Text] - 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: 1625–1635.
[Abstract/Free Full Text] - Rowe AK, Hamel MJ, Flanders WD, Doutizanga R, Ndoyo J, Deming MS. Predictors of correct treatment of children with fever seen at outpatient health facilities in the Central African Republic. Am J Epidemiol 2000; 151: 1029–1035.
[Abstract/Free Full Text] - Paredes P, Pena M, Flores-Guerra E et al. Factors influencing physicians’ prescribing behavior in the treatment of childhood diarrhoea: knowledge may not be the clue. Soc Sci Med 1996; 42: 1141–1153.[CrossRef][ISI][Medline]
- Zurovac D, Rowe AK, Ochola SA et al. Predictors of the quality of health worker treatment practices for uncomplicated malaria at government health facilities in Kenya. Int J Epidemiol 2004; 33: 1080–1091.
[Abstract/Free Full Text] - Lohr S. Sampling: Design and Analysis. Boston, MA: Duxbury Press, 1999: 42–43, 159–161.
- Kingdom of Morocco. Prise en Charge Integree des Maladies de L’enfant. Rabat: Ministry of Health, 1999.
- World Health Organization. IMCI Planning Guide: Gaining Experience with the IMCI Strategy in a Country. Geneva: World Health Organization, 1999.
- Dean AG, Dean JA, Coulombier D et al. Epi Info, Version 6: A Word Processing, Database, and Statistics Program for Public Health on IBM-Compatible Microcomputers. Atlanta, GA: Centers for Disease Control and Prevention, 1995.
- SUDAAN. SUDAAN User’s Manual Release 8.0.0. Research Triangle Park, NC: Research Triangle Institute, 2001.
- Rowe AK, de Savigny D, Lanata CF, Victora CG. How can we achieve and maintain high-quality performance of health workers in low-resource settings? Lancet 2005; 366: 1026–1035.[CrossRef][ISI][Medline]
- Foster SL, Bell-Dolan DJ, Burge DA. Behavioral observation. In Bellack AS, Hersen M, eds. Behavioral Assessment: A Practical Handbook, third edition. New York: Pergamon Press, 1988: 119–160.
- Rowe AK, Lama M, Onikpo F, Deming MS. Health workers’ perceptions of how an observer influences their practices during consultations with ill children, Benin. Trop Doct 2002; 32: 166–167.[ISI][Medline]
- Bennett S, Miller-Franco L, Kanfer R, Stubblebine P. The Development of Tools to Measure the Determinants and Consequences of Health Worker Motivation in Developing Countries. Bethesda, MD: Partnerships for Health Reform Project, Abt Associates, Inc, 2000: Major Applied Research 5, Technical Paper 2.
- Miller-Franco L, Bennett S, Kanfer R, Stubblebine P. Health Worker Motivation in Jordan and Georgia: A Synthesis of Results. Bethesda, MD: Partnerships for Health Reform Project, Abt Associates, Inc., 2000: Major Applied Research 5, Technical Paper 3.
- Aitken JM. Voices from the inside: managing district health services in Nepal. Int J Health Plann Manage 1994; 9: 309–340.[ISI][Medline]
- Bradley EH, Holmboe ES, Mattera JA, Roumanis SA, Radford MJ, Krumholz HM. A qualitative study of increasing B-blocker use after myocardial infarction. Why do some hospitals succeed? JAMA 2001; 285: 2604–2611.
[Abstract/Free Full Text] - Bryce J, Victora CG, Habicht JP, Vaughan JP, Black RE. The multi-country evaluation of the Integrated Management of Childhood Illness Strategy: lessons for the evaluation of public health interventions. Am J Public Health 2004; 94: 406–415.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||