Background: The presence and influence of nongovernmental organizations (NGOs) in the landscape o... more Background: The presence and influence of nongovernmental organizations (NGOs) in the landscape of global health and development have dramatically increased over the past several decades. The distribution of NGO activity and the ways in which contextual factors influence the distribution of NGO activity across geographies merit study. This paper explores the distribution of NGO activity, using Bolivia as a case study, and identifies local factors that are related to the distribution of NGO activity across municipalities in Bolivia. Methods: The research question is addressed using a geographic information system (GIS) and multiple regression analyses of count data. We used count data of the total number of NGO projects across Bolivian municipalities to measure NGO activity both in general and in the health sector specifically and national census data for explanatory variables of interest. Results: This study provides one of the first empirical analyses exploring factors related to the distribution of NGO activity at the national scale. Our analyses show that NGO activity in Bolivia, both in general and health-sector specific, is distributed unevenly across the country. Results indicate that NGO activity is related to population size, extent of urbanization, size of the indigenous population, and health system coverage. Results for NGO activity in general and health-sector specific NGO activity were similar. Conclusions: The uneven distribution of NGO activity may suggest a lack of coordination among NGOs working in Bolivia as well as a lack of coordination among NGO funders. Coordination of NGO activity is most needed in regions characterized by high NGO activity in order to avoid duplication of services and programmes and inefficient use of limited resources. Our findings also indicate that neither general nor health specific NGO activity is related to population need, when defined as population health status or education level or poverty levels. Considering these results we discuss broader implications for global health and development and make several recommendations relevant for development and health practice and research.
Interval censored life history data arise when the events of interest are only detectable at peri... more Interval censored life history data arise when the events of interest are only detectable at periodic assessments. When interest lies in the occurrence of two such events, bivariate interval censored event time data are obtained. We describe how to fit a four state Markov model useful for characterizing the association between two interval censored event times. The approach treats the two events symmetrically and enables one to fit multiplicative intensity models which give estimates of covariate effects as well as relative risks characterizing the association between the two events. An EM algorithm is described for estimation for which the maximization step can be carried out with standard software. The method is illustrated by application to data from a trial of HIV patients where the events are the onset of viral shedding in the blood and urine among individuals infected with cytomegalovirus.
To investigate the hypotheses that the topography of the optic nerve head (ONH) significantly cha... more To investigate the hypotheses that the topography of the optic nerve head (ONH) significantly changes during the day in untreated primary open-angle glaucoma (uPOAG) and healthy volunteers; and that there is a significant association with diurnal variations of intraocular pressure (IOP) and mean ocular perfusion pressure (MOPP). Fourteen uPOAG and 14 age-matched normals were included. IOP, blood pressure, and ONH topography were measured between 7:00 AM and 10:00 PM. MOPP was calculated. A mixed-effect repeated measures analysis was done. Variance component analysis was done for glaucoma and normal groups separately based on the mixed-effect models. The cup volume, rim volume, and cup shape measure in the temporal (T) and temporal-inferior (TI) sectors were significantly different between the 2 groups (P<0.05). The highest variance component was owing to "patients" whereas "time" had the smallest contributed percentage. Cup volume (T and TI) and reference height (RH) showed a significant (P<0.001) diurnal change in uPOAG. Rim volume (T and TI) showed a significant diurnal change in normals (P≤0.01). There was no significant (P>0.05) association between the change in IOP, MOPP, and ONH topography in either group. There was a significant association between cup volume and RH in both groups (P<0.001,…
To investigate diurnal change and pattern of variation in intraocular pressure (IOP) and systolic... more To investigate diurnal change and pattern of variation in intraocular pressure (IOP) and systolic (SBP) and diastolic (DSP) blood pressures in a group with untreated primary openangle glaucoma (uPOAG) and compare it with an age-matched, normal group. METHODS. IOP, SBP, and DBP were measured in 14 patients with uPOAG and in 14 normal subjects, every hour between 7 AM and 10 PM and the mean ocular perfusion pressure (MOPP) was calculated. Mixed-effect linear models were used to analyze the repeated-measures data in which both fixed and random effects were included. The relative diurnal change was calculated as the percentage decrease from maximum. RESULTS. The uPOAG group had the higher IOP (P Ͻ 0.001) and lower MOPP (P ϭ 0.025). There was a significant diurnal change in IOP, SBP, DBP, and MOPP in both groups (P Ͻ 0.001). The pattern of diurnal variation in IOP (P ϭ 0.137), SBP (P ϭ 0.569), and DBP (P ϭ 0.937) was not significantly different between groups but was significantly different for MOPP (P ϭ 0.040). MOPP and IOP were most similar at 7 AM and 1 PM. Postprandial hypotension was significant for SBP, DBP, and MOPP (P Ͻ 0.001), but not IOP (P ϭ 0.388) in both groups. The relative change in MOPP was larger in the uPOAG group (38% vs. 26%, P Ͻ 0.001), but the change in IOP was similar (42% vs. 41%, P ϭ 0.786). There was a significant effect of DBP on IOP over the course of the day in the uPOAG group (P ϭ 0.011) but not in the normal group (P ϭ 0.733). CONCLUSIONS. The relative diurnal change in IOP was similar in both uPOAG and normal subjects but MOPP showed a significant difference. MOPP significantly decreased after lunch, and was at its lowest in uPOAG at 7 AM, when IOP was at its highest. A significant association was found between diurnal DBP and IOP in uPOAG. (Invest Ophthalmol Vis Sci. 2005;46: 561-567)
To examine the impact of diurnal variation in intraocular pressure (IOP) and mean ocular perfusio... more To examine the impact of diurnal variation in intraocular pressure (IOP) and mean ocular perfusion pressure (MOPP) on the variation in anterior optic nerve capillary blood flow (BF) in patients with untreated early primary open-angle glaucoma (uPOAG) and healthy volunteers. METHODS. Fourteen patients with uPOAG (age, 56.3 Ϯ 12 years [SD]; seven men) and 14 normal subjects (age, 57.6 Ϯ 9.9 years; five men) were examined. Diurnal IOP, systolic (SBP) and diastolic (DBP) blood pressures, and optic nerve head (ONH) topography were measured every hour; and diurnal BF was measured by flowmeter every 2 hours between 0700 and 2200 hours. A perfusion image analyzer was used to calculate the mean BF within the rim (mean rim flow, MRF). The local flow (LF) was calculated using the median and mean flow rates within a 10 ϫ 10-pixel window placed on the rim in the area of maximum topography fluctuation (MTF). The MOPP was then calculated. Mixed-effect linear models were used to analyze the repeated measures data in which both fixed and random effects were included. RESULTS. IOP, BP, and MOPP had significant diurnal variation (P Ͻ 0.040). LF measured at the sector of MTF significantly changed in patients with uPOAG (P ϭ 0.006) but not in normal subjects (P ϭ 0.660). MRF did not show significant diurnal change in either group (P ϭ 0.130, P ϭ 0.770). LF increased significantly after lunch in the uPOAG group (P ϭ 0.001). SBP had a significant effect on LF over the course of the day in the uPOAG group (P ϭ 0.043). The diurnal change in IOP, BP, and MOPP did not have a significant effect on MTF in either group. In uPOAG, the local flow, in areas of greatest topographical change, correlated inversely with IOP at 0700 hours (P Յ 0.002). CONCLUSIONS. The mean rim flow did not change during the day, implying that the anterior optic nerve capillary blood flow was autoregulated in both normal subjects and in patients with uPOAG, despite significant changes in IOP and MOPP. However, the regions of greatest diurnal change in rim topography (MTF) had significant diurnal change in capillary blood flow in patients with uPOAG but not in normal subjects. (Invest Ophthalmol Vis Sci. 2005;46:4581-4587)
Naive use of misclassified covariates leads to inconsistent estimators of covariate effects in re... more Naive use of misclassified covariates leads to inconsistent estimators of covariate effects in regression models. A variety of methods have been proposed to address this problem including likelihood, pseudo-likelihood, estimating equation methods, and Bayesian methods, with all of these methods typically requiring either internal or external validation samples or replication studies. We consider a problem arising from a series of orthopedic studies in which interest lies in examining the effect of a short-term serological response and other covariates on the risk of developing a longer term thrombotic condition called deep vein thrombosis. The serological response is an indicator of whether the patient developed antibodies following exposure to an antithrombotic drug, but the seroconversion status of patients is only available at the time of a blood sample taken upon the discharge from hospital. The seroconversion time is therefore subject to a current status observation scheme, or Case I interval censoring, and subjects tested before seroconversion are misclassified as nonseroconverters. We develop a likelihood-based approach for fitting regression models that accounts for misclassification of the seroconversion status due to early testing using parametric and nonparametric estimates of the seroconversion time distribution. The method is shown to reduce the bias resulting from naive analyses in simulation studies and an application to the data from the orthopedic studies provides further illustration.
In recent years there has been considerable research devoted to the development of methods for th... more In recent years there has been considerable research devoted to the development of methods for the analysis of incomplete data in longitudinal studies. Despite these advances, the methods used in practice have changed relatively little, particularly in the reporting of pharmaceutical trials. In this setting, perhaps the most widely adopted strategy for dealing with incomplete longitudinal data is imputation by the "last observation carried forward" (LOCF) approach, in which values for missing responses are imputed using observations from the most recently completed assessment. We examine the asymptotic and empirical bias, the empirical type I error rate, and the empirical coverage probability associated with estimators and tests of treatment effect based on the LOCF imputation strategy. We consider a setting involving longitudinal binary data with longitudinal analyses based on generalized estimating equations, and an analysis based simply on the response at the end of the scheduled follow-up. We find that for both of these approaches, imputation by LOCF can lead to substantial biases in estimators of treatment effects, the type I error rates of associated tests can be greatly inflated, and the coverage probability can be far from the nominal level. Alternative analyses based on all available data lead to estimators with comparatively small bias, and inverse probability weighted analyses yield consistent estimators subject to correct specification of the missing data process. We illustrate the differences between various methods of dealing with drop-outs using data from a study of smoking behavior.
Background: The presence and influence of nongovernmental organizations (NGOs) in the landscape o... more Background: The presence and influence of nongovernmental organizations (NGOs) in the landscape of global health and development have dramatically increased over the past several decades. The distribution of NGO activity and the ways in which contextual factors influence the distribution of NGO activity across geographies merit study. This paper explores the distribution of NGO activity, using Bolivia as a case study, and identifies local factors that are related to the distribution of NGO activity across municipalities in Bolivia. Methods: The research question is addressed using a geographic information system (GIS) and multiple regression analyses of count data. We used count data of the total number of NGO projects across Bolivian municipalities to measure NGO activity both in general and in the health sector specifically and national census data for explanatory variables of interest. Results: This study provides one of the first empirical analyses exploring factors related to the distribution of NGO activity at the national scale. Our analyses show that NGO activity in Bolivia, both in general and health-sector specific, is distributed unevenly across the country. Results indicate that NGO activity is related to population size, extent of urbanization, size of the indigenous population, and health system coverage. Results for NGO activity in general and health-sector specific NGO activity were similar. Conclusions: The uneven distribution of NGO activity may suggest a lack of coordination among NGOs working in Bolivia as well as a lack of coordination among NGO funders. Coordination of NGO activity is most needed in regions characterized by high NGO activity in order to avoid duplication of services and programmes and inefficient use of limited resources. Our findings also indicate that neither general nor health specific NGO activity is related to population need, when defined as population health status or education level or poverty levels. Considering these results we discuss broader implications for global health and development and make several recommendations relevant for development and health practice and research.
Interval censored life history data arise when the events of interest are only detectable at peri... more Interval censored life history data arise when the events of interest are only detectable at periodic assessments. When interest lies in the occurrence of two such events, bivariate interval censored event time data are obtained. We describe how to fit a four state Markov model useful for characterizing the association between two interval censored event times. The approach treats the two events symmetrically and enables one to fit multiplicative intensity models which give estimates of covariate effects as well as relative risks characterizing the association between the two events. An EM algorithm is described for estimation for which the maximization step can be carried out with standard software. The method is illustrated by application to data from a trial of HIV patients where the events are the onset of viral shedding in the blood and urine among individuals infected with cytomegalovirus.
To investigate the hypotheses that the topography of the optic nerve head (ONH) significantly cha... more To investigate the hypotheses that the topography of the optic nerve head (ONH) significantly changes during the day in untreated primary open-angle glaucoma (uPOAG) and healthy volunteers; and that there is a significant association with diurnal variations of intraocular pressure (IOP) and mean ocular perfusion pressure (MOPP). Fourteen uPOAG and 14 age-matched normals were included. IOP, blood pressure, and ONH topography were measured between 7:00 AM and 10:00 PM. MOPP was calculated. A mixed-effect repeated measures analysis was done. Variance component analysis was done for glaucoma and normal groups separately based on the mixed-effect models. The cup volume, rim volume, and cup shape measure in the temporal (T) and temporal-inferior (TI) sectors were significantly different between the 2 groups (P<0.05). The highest variance component was owing to "patients" whereas "time" had the smallest contributed percentage. Cup volume (T and TI) and reference height (RH) showed a significant (P<0.001) diurnal change in uPOAG. Rim volume (T and TI) showed a significant diurnal change in normals (P≤0.01). There was no significant (P>0.05) association between the change in IOP, MOPP, and ONH topography in either group. There was a significant association between cup volume and RH in both groups (P<0.001,…
To investigate diurnal change and pattern of variation in intraocular pressure (IOP) and systolic... more To investigate diurnal change and pattern of variation in intraocular pressure (IOP) and systolic (SBP) and diastolic (DSP) blood pressures in a group with untreated primary openangle glaucoma (uPOAG) and compare it with an age-matched, normal group. METHODS. IOP, SBP, and DBP were measured in 14 patients with uPOAG and in 14 normal subjects, every hour between 7 AM and 10 PM and the mean ocular perfusion pressure (MOPP) was calculated. Mixed-effect linear models were used to analyze the repeated-measures data in which both fixed and random effects were included. The relative diurnal change was calculated as the percentage decrease from maximum. RESULTS. The uPOAG group had the higher IOP (P Ͻ 0.001) and lower MOPP (P ϭ 0.025). There was a significant diurnal change in IOP, SBP, DBP, and MOPP in both groups (P Ͻ 0.001). The pattern of diurnal variation in IOP (P ϭ 0.137), SBP (P ϭ 0.569), and DBP (P ϭ 0.937) was not significantly different between groups but was significantly different for MOPP (P ϭ 0.040). MOPP and IOP were most similar at 7 AM and 1 PM. Postprandial hypotension was significant for SBP, DBP, and MOPP (P Ͻ 0.001), but not IOP (P ϭ 0.388) in both groups. The relative change in MOPP was larger in the uPOAG group (38% vs. 26%, P Ͻ 0.001), but the change in IOP was similar (42% vs. 41%, P ϭ 0.786). There was a significant effect of DBP on IOP over the course of the day in the uPOAG group (P ϭ 0.011) but not in the normal group (P ϭ 0.733). CONCLUSIONS. The relative diurnal change in IOP was similar in both uPOAG and normal subjects but MOPP showed a significant difference. MOPP significantly decreased after lunch, and was at its lowest in uPOAG at 7 AM, when IOP was at its highest. A significant association was found between diurnal DBP and IOP in uPOAG. (Invest Ophthalmol Vis Sci. 2005;46: 561-567)
To examine the impact of diurnal variation in intraocular pressure (IOP) and mean ocular perfusio... more To examine the impact of diurnal variation in intraocular pressure (IOP) and mean ocular perfusion pressure (MOPP) on the variation in anterior optic nerve capillary blood flow (BF) in patients with untreated early primary open-angle glaucoma (uPOAG) and healthy volunteers. METHODS. Fourteen patients with uPOAG (age, 56.3 Ϯ 12 years [SD]; seven men) and 14 normal subjects (age, 57.6 Ϯ 9.9 years; five men) were examined. Diurnal IOP, systolic (SBP) and diastolic (DBP) blood pressures, and optic nerve head (ONH) topography were measured every hour; and diurnal BF was measured by flowmeter every 2 hours between 0700 and 2200 hours. A perfusion image analyzer was used to calculate the mean BF within the rim (mean rim flow, MRF). The local flow (LF) was calculated using the median and mean flow rates within a 10 ϫ 10-pixel window placed on the rim in the area of maximum topography fluctuation (MTF). The MOPP was then calculated. Mixed-effect linear models were used to analyze the repeated measures data in which both fixed and random effects were included. RESULTS. IOP, BP, and MOPP had significant diurnal variation (P Ͻ 0.040). LF measured at the sector of MTF significantly changed in patients with uPOAG (P ϭ 0.006) but not in normal subjects (P ϭ 0.660). MRF did not show significant diurnal change in either group (P ϭ 0.130, P ϭ 0.770). LF increased significantly after lunch in the uPOAG group (P ϭ 0.001). SBP had a significant effect on LF over the course of the day in the uPOAG group (P ϭ 0.043). The diurnal change in IOP, BP, and MOPP did not have a significant effect on MTF in either group. In uPOAG, the local flow, in areas of greatest topographical change, correlated inversely with IOP at 0700 hours (P Յ 0.002). CONCLUSIONS. The mean rim flow did not change during the day, implying that the anterior optic nerve capillary blood flow was autoregulated in both normal subjects and in patients with uPOAG, despite significant changes in IOP and MOPP. However, the regions of greatest diurnal change in rim topography (MTF) had significant diurnal change in capillary blood flow in patients with uPOAG but not in normal subjects. (Invest Ophthalmol Vis Sci. 2005;46:4581-4587)
Naive use of misclassified covariates leads to inconsistent estimators of covariate effects in re... more Naive use of misclassified covariates leads to inconsistent estimators of covariate effects in regression models. A variety of methods have been proposed to address this problem including likelihood, pseudo-likelihood, estimating equation methods, and Bayesian methods, with all of these methods typically requiring either internal or external validation samples or replication studies. We consider a problem arising from a series of orthopedic studies in which interest lies in examining the effect of a short-term serological response and other covariates on the risk of developing a longer term thrombotic condition called deep vein thrombosis. The serological response is an indicator of whether the patient developed antibodies following exposure to an antithrombotic drug, but the seroconversion status of patients is only available at the time of a blood sample taken upon the discharge from hospital. The seroconversion time is therefore subject to a current status observation scheme, or Case I interval censoring, and subjects tested before seroconversion are misclassified as nonseroconverters. We develop a likelihood-based approach for fitting regression models that accounts for misclassification of the seroconversion status due to early testing using parametric and nonparametric estimates of the seroconversion time distribution. The method is shown to reduce the bias resulting from naive analyses in simulation studies and an application to the data from the orthopedic studies provides further illustration.
In recent years there has been considerable research devoted to the development of methods for th... more In recent years there has been considerable research devoted to the development of methods for the analysis of incomplete data in longitudinal studies. Despite these advances, the methods used in practice have changed relatively little, particularly in the reporting of pharmaceutical trials. In this setting, perhaps the most widely adopted strategy for dealing with incomplete longitudinal data is imputation by the "last observation carried forward" (LOCF) approach, in which values for missing responses are imputed using observations from the most recently completed assessment. We examine the asymptotic and empirical bias, the empirical type I error rate, and the empirical coverage probability associated with estimators and tests of treatment effect based on the LOCF imputation strategy. We consider a setting involving longitudinal binary data with longitudinal analyses based on generalized estimating equations, and an analysis based simply on the response at the end of the scheduled follow-up. We find that for both of these approaches, imputation by LOCF can lead to substantial biases in estimators of treatment effects, the type I error rates of associated tests can be greatly inflated, and the coverage probability can be far from the nominal level. Alternative analyses based on all available data lead to estimators with comparatively small bias, and inverse probability weighted analyses yield consistent estimators subject to correct specification of the missing data process. We illustrate the differences between various methods of dealing with drop-outs using data from a study of smoking behavior.
Uploads
Papers by Leilei Zeng