Factors associated with the awareness and ... - Wiley Online Library

DOI: 10.1111/j.1471-0528.2006.01042.x

General obstetrics

www.blackwellpublishing.com/bjog

Factors associated with the awareness and practice of evidence-based obstetric care in an African setting ATN Tita,a BJ Selwyn,b DK Waller,b AS Kapadia,b S Dongmoc a Center for Research in Women’s Health, Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of Alabama at Birmingham, Birmingham, AL, USA b School of Public Health, University of Texas Health Science Center at Houston, Houston, TX, USA c Bafut Health District, North West Provincial Delegation of Health, Bamenda, Cameroon Correspondence: Dr ATN Tita, Center for Research in Women’s Health, Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of Alabama at Birmingham, 618 20th Street South (OHB 458), Birmingham, AL 35233, USA. Email [email protected]

Accepted 20 June 2006.

Objective To identify the factors associated with important (‡50%) variation in awareness and practice of evidence-based obstetric interventions in an African setting where we have previously reported poor awareness and use of evidence-based reproductive interventions. Design Cross-sectional analysis of data from our Reproductive

Health Interventions Study. Setting North-west province, Cameroon, Africa. Population Health workers including obstetricians, other physicians,

midwives, nurses and other staff providing reproductive care. Main outcome measures Prevalence ratios (PR) of uniform

awareness and practice of four key evidence-based obstetric interventions from the World Health Organization Reproductive Health Library (WHO RHL): antiretrovirals to prevent motherto-child transmission of HIV/AIDS, antenatal corticosteroids for prematurity, uterotonics to prevent postpartum haemorrhage and magnesium sulphate for seizure prophylaxis. Methods Comparisons of descriptive covariates, applying logistic regression to estimate independent relationships with awareness and use of evidence-based interventions.

Results A total of 15.5% (50/322) of health workers were aware of all the four interventions while only 3.8% (12/312) reported optimal practice. Evidence-based awareness was strongly associated with practice (PR = 15.4; 96% CI: 4.3–55.0). Factors significantly associated with awareness were: attending continuing education, access to the WHO RHL, employment as an obstetrician/ gynaecologist and working in autonomous military or National Insurance Fund facilities. Controlling for potential confounding, working as an obstetrician was associated with increased awareness (adjusted prevalence odds ratio [aPOR] = 8.3; 95% CI: 1.3–53.8) as was median work experience of 5–15 years (aPOR = 2.0; 95% CI: 1.0–3.8). Internet access was associated with increased practice (aPOR = 3.4; 95% CI: 1.0–11.8). Other potentially important variations were observed, although they did not attain statistical significance. Conclusions Several factors including obstetric training and

continuous education positively influence evidence-based awareness and practice of key obstetric interventions. Confirmation and application of this information may enhance the effectiveness of programmes to improve maternal and perinatal outcomes. Keywords Evidence-based care, international health,

obstetric interventions.

Please cite this paper as: Tita A, Selwyn B, Waller D, Kapadia A, Dongmo S. Factors associated with the awareness and practice of evidence-based obstetric care in an African setting. BJOG 2006;113:1060–1066.

At the time of the study, A.T.N.T. was affiliated with Baylor College of Medicine and the University of Texas Health Science Center at Houston, Houston, TX, USA

countries.1 Decades of attention from international agencies through the Safe Motherhood Initiative and the Prevention of Maternal Death and Disability programmes have deepened awareness and appreciation of the problem. However, they are yet to significantly impact outcomes.2,3 Many strategies have been recommended, but evidence for actual effectiveness for many of these remains inconclusive.4 Improving access to

1060

ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology

Introduction Maternal and perinatal morbidity and mortality in developing countries remain extremely high compared with developed

Factors associated with evidence-based obstetrics

quality reproductive health services is among those strategies deemed promising.5 This should include the widespread use of interventions that have been proven to improve maternal and perinatal outcomes. A population-based reproductive health interventions study conducted in north-west Cameroon revealed poor awareness and use of important evidence-based obstetric interventions by health workers.6 In a qualitative substudy, these health workers perceived the key barriers to be deficient education and training in evidence-based care and the lack of the necessary supplies and materials. We conducted this analysis to identify factors that affect evidence-based awareness and practice. The primary objective was to identify descriptive characteristics that are associated with an important (at least a 50%) variation in awareness and practice of key evidence-based obstetric interventions.

Methods The Institutional Review Boards of Baylor College of Medicine and the University of Texas in Houston and the Provincial Delegate for Health in Cameroon approved our Reproductive Health Interventions Study. All participants gave informed consent. Key elements of the methodology, which has been previously described,6 are outlined below.

Design and participants We conducted a cross-sectional analysis of data from the Reproductive Health Interventions Study.6 This populationbased study was conducted in February 2004 in the anglophone north-west province of Cameroon. As part of that study, we surveyed health workers of all categories providing reproductive care in all public, missionary and private health facilities, including district and provincial hospitals. Reproductive health workers were defined as any formally trained or untrained staff providing intrapartum, antepartum or postpartum care. They included physicians, midwives, nurses (all categories) and occasionally untrained birth attendants such as nursing assistants, laboratory technicians or other lay workers.

Reproductive Health Library (WHO RHL), 2002.7 These interventions included antenatal corticosteroids for impending prematurity,8 antiretrovirals to prevent mother-to-child transmission of HIV/AIDS,9 uterotonics to prevent postpartum haemorrhage10 and magnesium sulphate for seizure prophylaxis in eclampsia.11–13 We considered these interventions vital to reducing maternal and perinatal morbidity and mortality in developing countries.14 ‘Evidence-based awareness’ was defined as awareness of all four component interventions. ‘Evidence-based practice’ was defined as optimal practice of all four component interventions. ‘Optimal practice’ was defined as use of an intervention ‘always’ or ‘greater than 50% of the time’ when indicated.

Exposure variables The influence of several characteristics (listed in Table 1) on evidence-based awareness and practice was explored. In addition, the influence of awareness was evaluated for the evidence-based practice outcome.

Sampling and sample size Interviews were sought with all reproductive care providers at district and provincial hospitals, all physicians and all heads of health centres or private clinics in the region. Workers in charge of the maternity units in these health centres and clinics were also targeted. If these unit heads were not available, another available health worker was enrolled. All 328 survey respondents presented in Table 1 were included in the analysis. They represented 91.4% of all targeted reproductive health workers and 92.2% of reproductive health units in the north-west province of Cameroon.6

Statistical analysis

These were estimates of the prevalence ratios (PR) of awareness and of practice of the composite of four evidence-based interventions selected from the World Health Organization

Microsoft Access (v. 2000), SPSS 11.0 (SPSS Inc., Chicago, IL, USA) and STATA 6.0 (STATA Corp., College Station, TX, USA) were used for data management and analysis. Crude PR and 95% confidence intervals of awareness and practice were determined for each descriptive characteristic. A priori, only characteristics associated with at least a 50% variation in awareness or practice (crude PR > 1.5 or 0.67 were then progressively removed, starting with the variable with the highest P value until the final parsimonious models were obtained. Variables were definitively kept out of the models if the log likelihood ratio test associated with their removal was not statistically significant at the 0.05 level. Logistic regression modelling, the most commonly used technique in the literature


1061

Instrument A 40-item, pre-tested, face-to-face, self-administered, anonymous questionnaire was utilised. Respondents were encouraged to add clarifying comments, which were taken into consideration in coding, thereby enhancing the validity of responses by 3% of participants. A total of 108 (32.9%) respondents wrote in comments.

Outcome measures

Tita et al.

Table 1. Characteristics of 328 participating reproductive health workers in north-west Cameroon Characteristic (N) Age (years)* (322) Gender (326): male Professional category (325) Obstetrician/gynaecologist GP or other physician Midwife Nurse Others (nursing assistants, lab technicians, lay staff) Years since training* (325) Any public health training (326): yes Type of health unit (328) Governmental Nongovernmental or missionary Lay private Other (military, parastatal) Location of health unit (328): provincial capital Faculty/preceptor status (318): yes Access to library (322): yes Access to internet or email (322): yes Number of times used health journal or database in past year* (309) Awareness of WHO RHL (324): aware Access to WHO RHL (325): yes Number of reproductive health workshops/conferences attended past 3 years* (320)

n

%

40.9 171

41 (20–72) 52.5

6 48 102 140 29 13.3 157

1.8 14.8 31.4 43.1 8.9 14 (0–42) 48.2

223 64 31 10 87 80 74 85 3.2 39 11 1.5

68.0 19.5 9.5 3.0 26.0 25.2 23.0 26.1 1 (0–51) 12.0 3.4 0 (0–20)

N, number of respondents with complete information about the characteristic; n, number of respondents positive for the stated characteristic. *For these continuous variables, n corresponds to the mean and % corresponds to median and (range).

Obstetric/gynaecologic training was associated with a significant increase in evidence-based awareness but not practice

when compared with midwives. Although not statistically significant, the point and interval estimates for other (generalist) physicians were consistent with an increase in evidence-based awareness relative to midwives. Nurses’ evidence-based awareness was comparable with that of the midwives, but they reported a five-fold reduction in practice compared with midwives, which was of borderline statistical significance. Attendance at two or more reproductive health educational events and access to the WHO RHL were associated with important increases in evidence-based awareness but not practice. Working at an autonomous military or social insurance fund facility was associated with increased awareness compared with that of government facilities. Working at missionary/nongovernmental organisation units was associated with increases in both awareness and practice based on predefined criteria, although these were not statistically significant at the 0.05 level. Other factors whose point and interval estimates showed crude associations with awareness or practice, without attaining statistical significance, included health workers’ age, work experience, access to a library or internet and journal use. Factors not associated with evidence-based awareness were gender, public health training, hospital location and self-reported status as a faculty or preceptor.

1062


for this kind of analysis, was the pre-specified method. However, because of concerns that the resultant prevalence odds ratios may be difficult to interpret and may overestimate the PR for common outcomes,16,17 we applied Poisson regression with robust variance to confirm the measures from logistic regression. Poisson regression with robust variance has been assessed for use with binary data18 and demonstrated to be a viable alternative for correctly estimating PR in crosssectional studies.17,19

Results A total of 15.5% of health workers with complete information (50/322) were aware of all the four vital interventions and 3.8% (12/312) reported appropriate practice of all. Estimates of prevalence of evidence-based awareness and practice and corresponding measures of association (PR) for each descriptive characteristic are presented in Table 2. Evidence-based awareness was associated with a 15-fold increase in practice (PR = 15.4; 95% CI: 4.3–55).

Crude relationships


Table 2. Awareness and practice of evidence-based obstetric care and crude PR by descriptive characteristics Characteristic

Evidence-based awareness Unaware Aware Age (years) ,35 35–44 45 Gender Female Male Profession Obstetrician/gynaecologist Other physician Midwives Nurses Others* Experience (years) 0–4.9 5–14.9 15 Public health training No Yes Access to library No Yes Internet/email access No Yes Journal/database use None One or more Aware of WHO RHL No Yes Access to WHO RHL No Yes Continuous education None Once Twice or more Type of health unit Governmental Nongovernmental organisation/missionary Lay private Other** Faculty/preceptor No Yes Location of unit All others Provincial capital

Awareness

Practice

Prevalence

PR (95% CI)

Prevalence

PR (95% CI)

N/A

N/A

1.2 18.0

1.0 15.4 (4.3–55.0)

21.6 15.9 11.2

1.0 0.73 (0.41–1.3) 0.52 (0.26–1.05)

2.8 6.2 1.1

1.0 2.2 (0.49–9.9) 0.39 (0.04–4.2)

13.8 17.3

1.0 1.25 (0.75–2.1)

3.4 4.2

1.0 1.25 (0.40–3.8)

66.7 22.9 14.0 13.2 6.9

4.8 (2.3–10.0) 1.64 (0.8–3.33) 1.0 0.95 (0.49–1.81) 0.49 (0.12–2.04)

16.7 4.3 7.1 1.5 0.0

2.36 (0.34–16.2) 0.60 (0.13–2.71) 1.0 0.21 (0.05–1.0) 0

14.1 22.2 11.7

1.0 1.57 (0.83–2.97) 0.83 (0.42–1.64)

6.5 3.4 2.3

1.0 0.52 (0.13–2.0) 0.35 (0.09–1.38)

17.5 13.6

1.0 0.78 (0.47–1.31)

5.0 2.6

1.0 0.52 (0.16–1.69)

14.0 21.9

1.0 1.57 (0.92–2.67)

3.8 4.3

1.0 1.14 (0.32–4.1)

14.0 20.2

1.0 1.45 (0.85–2.46)

2.6 7.3

1.0 2.8 (0.93–8.5)

12.6 19.0

1.0 1.51 (0.88–2.6)

5.3 3.0

1.0 0.57 (0.19–1.77)

14.7 23.1

1.0 1.57 (0.83–3.0)

3.6 5.6

1.0 1.53 (0.35–6.7)

14.9 36.4

1.0 2.43 (1.1–5.6)

3.7 9.1

1.0 2.49 (0.35–17.6)

11.4 15.4 23.1

1.0 1.35 (0.62–2.9) 2.03 (1.2–3.5)

2.6 7.8 3.8

1.0 2.98 (0.77–11.5) 1.46 (0.37–5.71)

12.4 21.9 16.7 40.0

1.0 1.77 (0.99–3.2) 1.35 (0.56–3.2) 3.2 (1.4–7.5)

3.3 6.6 0.0 11.1

1.0 2.0 (0.61–6.62) N/A 3.4 (0.47–24.8)

15.0 16.7

1.0 1.1 (0.61–1.9)

4.0 3.8

1.0 0.97 (0.27–3.48)

13.8 20.5

1.0 1.48 (0.87–2.5)

3.5 4.8

1.0 1.38 (0.43–4.46)

N/A, not applicable. *Nursing assistants, laboratory technicians and lay staff. **Autonomous military and National Insurance Fund units.


1063

Tita et al.

Independent relationships The parsimonious logistic regression models for evidencebased awareness and practice outcomes are presented in Table 3. Awareness was not included as a potential confounder in the model for evidence-based practice because as a presumed intermediate factor between other variables and practice, it’s inclusion, as expected, weakened the associations. The association between obstetric/gynaecologic training (compared with nonphysicians) and increased evidence-based awareness was confirmed: aPOR (95% CI) = 8.3 (1.3–53.8). The increase in awareness associated with other physician training, however, was not statistically significant. Median work experience between 5 and 15 years was also associated with an increase in awareness of borderline statistical significance. Access to the internet was associated with more than a three-fold increase in evidence-based practice but not awareness. Other important associations involving continuous education, work at private or missionary unit, access to the WHO RHL and use of journals did not attain statistical significance. The retrospective use of Poisson regression yielded the same parsimonious models for awareness and practice obtained using logistic regression. The magnitude Table 3. Important covariates of awareness and practice of key obstetric interventions adjusted for potential confounders (logistic regression model) Characteristic

Awareness aPOR (95% CI)

Profession Obstetrician/gynaecologist 8.3 (1.3–53.8) Other physician 1.8 (0.8–4.1) Work experience 5–14.9 years 2.0 (1.0–3.8) Access to WHO RHL Yes 2.0 (0.5–9.0) Continuous education event One or more 1.7 (0.9–3.2) Type of centre Private or missionary 1.6 (0.8–3.2) Age (years) Less than 35 N/A Internet access Yes N/A Used journal or database Yes N/A

Practice aPOR (95% CI)

N/A

N/A N/A 2.2 (0.6–8.1) 3.0 (0.8–11.7) 0.2 (0.03–1.3) 3.4 (1.0–11.8) 0.32 (0.1–1.2)

N/A, not retained in final model. aPOR is prevalence odds ratio adjusted for other confounders retained in the logistic regression model. Respective comparison categories include nonphysicians, experience of ,5 or 15 years, attendance at no educational event, age 35 years and above, no access to WHO RHL, governmental health unit, no access to internet and no use of health journal or database.

1064

of the associations were practically the same, except that for the relationship between obstetric training and awareness, adjusted PR (95% CI) = 3.6 (1.7–7.5), probably overestimated by logistic regression.

Discussion and conclusion This is the first study to our knowledge to quantify predictors of evidence-based awareness and practice in a developing world setting. Previous descriptions were based on anecdotes, speculation and qualitative data.20 Awareness was associated with a tremendous 15-fold increase in practice. Obstetric/ gynaecologic training and 5–15 years of work experience were associated with increased awareness but not with practice, while access to the internet was associated with increased practice only. In the crude analysis, attending reproductive health educational events, access to the WHO RHL and working at an autonomous military or National Insurance Funds unit were all associated with awareness of evidence-based obstetric care. We also noted factors whose estimates, while not attaining statistical significance, were indicative of potentially important variations in awareness or practice of at least 50%. We acknowledge some limitations. Expected over-reporting of evidence-based practices likely biased our results. Nevertheless, given the likelihood that this over-reporting was nondifferential with respect to the descriptive characteristics, the actual relationships would be stronger than reported here. Other likely sources of bias include failure to account for unmeasured covariates such as those directly affecting patients or the broad health system. An example affecting practice is the lack of necessary supplies and materials elicited during qualitative interviews.6 The relationship between internet access and improved practice may therefore be confounded by better access to requisite supplies. Also, our statistical methods do not entirely eliminate random errors: the sample size was inadequate for a complete delineation of all relationships of interest, increasing the likelihood of type 2 errors. The assessment of multiple relationships increased the likelihood of chance findings (type 1 error). With respect to the target population, these errors are minimised by coverage of more than 90% of target health workers and giving some emphasis to associations that did not attain statistical significance. Furthermore, the paucity of outcome events (evidence-based practice) could lead to a wrong logistic regression model; however, our analysis strategy and a scrutiny of the crude relationships indicate that this is unlikely. The conduct of this study in only one of the ten provinces of Cameroon limits its generalisability, although based on our knowledge, we believe that the findings reflect the situation in the country. Finally, the cross-sectional design inherently does not measure temporality between covariates and outcomes, thereby limiting confirmatory interpretations.



Overall, programmes that aim to reduce maternal and perinatal morbidity and mortality in developing countries by increasing awareness and practice of evidence-based interventions must use multiple strategies. In addition to overcoming known barriers, such programmes would also benefit from emphasis on characteristics that may increase awareness and practice. In this light, the possible role of the descriptive characteristics we have explored in increasing awareness and practice needs confirmation in analytic studies. If confirmed, this information should be useful in tailoring projects and programmes to improve maternal and perinatal outcomes. For example, obstetricians who are primarily concentrated at university teaching hospitals may play a lead role in continuous education of reproductive health workers. j

The association between evidence-based awareness and practice, although intuitive, was remarkable for its strength. Based on a proportional relationship between awareness and practice of 13 evidence-based interventions and a consistent lag in the former, we previously concluded that evidencebased awareness increases practice.6 Qualitative studies also reveal that reproductive health clinicians in developing countries expect that increased evidence-based awareness would lead to better practice.6,20 Obstetric/gynaecologic training, although associated with the highest increase in awareness, did not translate into an important increase in practice at the multivariate level. The association of median work experience with increased evidence-based awareness depicts an inverse U-shaped curve peaking at about 5–15 years postgraduation. This suggests an increased learning curve soon after training that is neither sustained nor transformed into better practice. This relationship was independent of attendance at an educational event and may reflect loss of self-motivation to acquire new knowledge, identified by health workers as barrier to awareness in this setting.6 It could also represent a cohort effect with both most recent and least recent graduates being less instructed in current evidence-based practices. Attendance at a reproductive health educational event is a key modifiable characteristic that would likely yield increases in awareness and possibly practice, given the right conditions. Health workers in developing settings identified this as a key element for increasing evidence-based awareness.6,21 At the time the study was conducted, there was a continuing programme initiated by a missionary health organisation to prevent maternal-to-child transmission of HIV, involving training seminars and provision of supplies. This was credited in part for the reported increase in practice of the intervention between 2001 (3%)22 and 2004 (40%).6 Therefore, our suggestive results regarding missionary facilities, support anecdotes upholding that missionary organisations play an important role in providing effective health care in developing countries.23,24 That access to the WHO RHL may lead to increased awareness but not increased practice may be explained by limited supplies. It is unclear how access to the internet can independently affect practice without affecting awareness. It is likely a proxy for one or more factors, not examined in this study, associated with improved practice. Based on the results of our prior qualitative substudy of barriers,6 we speculate that workers with internet access are also more likely to have access to the supplies needed to implement evidence-based practices. In the same vein, after adjusting for potential confounders, working at autonomous military or social insurance fund facilities was no longer associated with a significant variation in evidence-based awareness. Teaching responsibilities (medical or nursing faculty/ preceptor) was associated with evidence-based awareness in a qualitative study.20 This was not an important characteristic in our univariate or multivariate analyses.

1 World Health Organization. The World Health Report 2005: Make Every Mother and Child Count. Geneva, Switzerland: WHO, 2005. 2 World Bank, Human Development Network. Safe Motherhood and the World Bank: Lessons from 10 Years of Experience. Washington DC: World Bank, 1999. 3 Maine D, Rosenfield A. The AMDD program: history, focus and structure. Int J Gynaecol Obstet 2001;74:99–104. 4 Bale JR, Stoll BJ, Lucas AO, editors. Improving Birth Outcomes: Meeting the Challenge in the Developing World. Washington DC: National Academies Press, 2003. 5 Gulmezoglu AM. Promoting standards for quality of maternal health care. Br Med Bull 2003;67:73–83. 6 Tita ATN, Selwyn BJ, Dongmo S, Waller DK, Kapadia AS, Dongmo S. Evidence-based reproductive health care in Cameroon: populationbased study of awareness, use and barriers. Bull World Health Organ 2005;83:895–903. 7 World Health Organization. Reproductive Health Library No 5. Geneva, Switzerland: WHO, 2002 [www.rhlibrary.com]. Accessed 12 December 2004. 8 Crowley P. Prophylactic corticosteroids for preterm birth (Cochrane Review). In: The Cochrane Library. Oxford, UK: Update Software; 2002. 9 Brocklehurst P, Volmink J. Antiretrovirals for reducing the risk of mother-to-child transmission of HIV infection (Cochrane Review). In: The Cochrane Library. Oxford, UK: Update Software; 2002. 10 Prendiville WJ, Elbourne D, McDonald S. Active versus expectant management in the third stage of labour (Cochrane Review). In: The Cochrane Library. Oxford, UK: Update Software; 2002. 11 Duley L, Henderson-Smart D. Magnesium sulphate versus diazepam for eclampsia (Cochrane Review). In: The Cochrane Library. Oxford, UK: Update Software; 2002. 12 Duley L, Henderson-Smart D. Magnesium sulphate versus phenytoin for eclampsia (Cochrane Review). In: The Cochrane Library. Oxford, UK: Update Software; 2002. 13 Duley L, Gulmezoglu AM. Magnesium sulphate versus lytic cocktail for eclampsia (Cochrane Review). In: The Cochrane Library. Oxford, UK: Update Software; 2002. 14 Global Health Council. Making childbirth safer through promoting evidence-based care. Technical report. Washington DC: Global Health Council; 2002. 15 Hosmer DW, Lemeshow S. Applied Logistic Regression, 2nd edn. New York, NY: John Wiley & Sons Inc., 2000.


1065

References

Tita et al.

16 Thompson ML, Myers JE, Kriebel D. Prevalence odds ratio or prevalence ratio in the analysis of cross sectional data: what is to be done? Occup Environ Med 1998;55:272–7. 17 Barros AJD, Hirakata VN. Alternatives for logistic regression in crosssectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Med Res Methodol 2003; 3:21 [www.biomedcentral.com/1471-2288/3/21]. Accessed 9 October 2005. 18 Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;159:702–6. 19 Behrens T, Taeger D, Wellmann J, Keil U. Different methods to calculate effect estimates in cross-sectional studies. A comparison between prevalence odds ratio and prevalence ratio. Methods Inf Med 2004; 43:505–9.

20 Geyoushi BE, Matthews Z, Stones RW. Pathways to evidence-based reproductive healthcare in developing countries. BJOG 2003;110:500–7. 21 Qian X, Smith H, Zhou L, Liang J, Garner P. Evidence-based obstetrics in four hospitals in China: an observational study to explore clinical practice, women’s preferences and provider’s views. BMC Pregnancy Childbirth 2001;1:1 [www.biomedcentral.com/14712393/1/1]. Accessed 12 December 2004. 22 World Health Organization. Coverage of Selected Health Services for HIV/AIDS Prevention and Care in Less Developed Countries in 2001. Geneva, Switzerland: WHO, 2002. 23 Hoover EL, Cole-Hoover G, Berry PK, Hoover ET, Harris B, Rageh D, et al. Private volunteer medical organizations: how effective are they? J Natl Med Assoc 2005;97:270–5. 24 Green A, Matthias A. Non-governmental Organizations and Health in Developing Countries. New York, NY: St Martin’s Press, 1997.

1066
