Abstract
Background: South Africa has a dual high burden of HIV and non-communicable diseases (NCDs). In a response to the dual burden of these chronic diseases, the National Department of Health (NDoH) introduced a pilot of the Integrated Chronic Disease Management (ICDM) mod‐
el in June 2011 in selected Primary Health Care (PHC) facilities, one of the first of such efforts by an African Ministry of Health. The main aim of the ICDM model is to leverage the successes of the innovative HIV treatment programme for NCDs in order to improve the quality of
chronic disease care and health outcomes of adult chronic disease patients. Since the initiation of the ICDM model, little is known about the quality of chronic care resulting in the effectiveness of the model in improving health outcomes of chronic disease patients.
Objectives: To describe the chronic disease profile and predictors of healthcare utilisation (HCU) in a rural population in a South African municipality; and assess quality of care and effectiveness of the ICDM model in improving health outcomes of chronic disease patients receiving treatment in PHC facilities.
Methods: An NDoH pilot study was conducted in selected health facilities in the Bushbuckridge municipality, Mpumalanga province, north‐ east South Africa, where a part of the population has been continuously monitored by the Agincourt Health and Socio-Demographic
Surveillance System (HDSS) since 1992. Two main studies were conducted to address the two research objectives. The first study was a situation analysis to describe the chronic disease profile and predictors of healthcare utilisation in the population monitored by the Agincourt
HDSS. The second study evaluated quality of care in the ICDM model as implemented and assessed effectiveness of the model in improving health outcomes of patients receiving treatment in PHC facilities. This second study had three components: (1) a qualitative and (2) a quantitative evaluation of the quality of care in the ICDM model; and a (3) quantitative assessment of effectiveness of the ICDM model in improving patients‘ health outcomes. The two main studies have been categorised into three broad thematic areas: chronic disease profile and predictors of healthcare utilisation; quality of care in the ICDM model; and changes in patients‘ health outcomes attributable to the ICDM model.
In the first study, a cross-sectional survey to measure healthcare utilisation was targeted at 7,870 adults 50 years and over permanently residing in the area monitored by the Agincourt HDSS in 2010, the year before the ICDM model was introduced. Secondary data on healthcare
utilisation (dependent variable), socio-demographic variables drawn from the HDSS, receipt of social grants and type of medical aid (independent variables) were analysed. Predictors of HCU were determined by binary logistic regression adjusted for socio-demographic vari‐
ables.
The quantitative component of the second study was a cross-sectional survey conducted in 2013 in the seven PHC facilities implementing the ICDM model in the Agincourt sub-district (henceforth referred to as the ICDM pilot facilities) to better understand the quality of care in
the ICDM model. Avedis Donabedian‘s theory of the relationships between structure, process, and outcome (SPO) constructs was used to evaluate quality of care in the ICDM model exploring unidirectional, mediation, and reciprocal pathways. Four hundred and thirty-five (435) proportionately sampled patients ≥ 18 years and the seven operational managers of the PHC facilities responded to an adapted satisfaction questionnaire with measures reflecting structure (e.g. equipment), process (e.g. examination) and outcome (e.g. waiting time) constructs.
Seventeen dimensions of care in the ICDM model were evaluated from the perspectives of patients and providers. Eight of these 17 dimensions of care are the priority areas of the HIV treatment programme used as leverage for improving quality of care in the ICDM model: supply of critical medicines, hospital referral, defaulter tracing, prepacking of medicines, clinic appointments, reducing patient waiting time, and coherence of integrated chronic disease care (a one-stop clinic meeting most of patients‘ needs). A structural equation model was fit to operationalise Donabedian‘s theory using patient‘s satisfaction scores.
The qualitative component of the second study was a case study of the seven ICDM pilot facilities conducted in 2013 to gain in-depth perspectives of healthcare providers and users regarding quality of care in the ICDM model. Of the 435 patients receiving treatment in the pilot facilities, 56 were purposively selected for focus group discussions. An in-depth interview was conducted with the seven operational managers within the pilot facilities and the health manager of the Bushbuckridge municipality. Qualitative data were analysed, with MAXQDA 2
software, to identify 17 a priori dimensions of care and emerging themes. In addition to the emerging themes, codes generated in the qualitative analysis were underpinned by Avedis Donabedian‘s SPO theoretical framework.
A controlled interrupted time-series study was conducted for the 435 patients who participated in the cross-sectional study in the ICDM pilot facilities and 443 patients proportionately recruited from five PHC facilities not implementing the ICDM model (Comparison PHC facilities in the surrounding area outside the Agincourt HDSS) from 2011-2013. Health outcome data for each patient were retrieved from facility records at 30-time points (months) during the study period. We performed autoregressive moving average (ARMA) statistical modelling to account for autocorrelation inherent in the time-series data. The effect of the ICDM model on the control of BP (<140/90 mmHg) and CD4 counts (>350 cells/mm3) was assessed by controlled segmented linear regression analysis.
Results: Seventy-five percent (75%) of the 7,870 eligible adults 50+ responded to the health care utilization survey in the first study. All 5,795 responders reported health problems, of whom 96% used healthcare, predominantly at public health facilities (82%). Reported health
problems were: chronic non-communicable diseases (41% - e.g. hypertension), acute conditions (27% - e.g. flu), other conditions (26% - e.g. musculoskeletal pain), chronic communicable diseases (3% e.g. HIV and TB) and injuries (3%). Chronic communicable (OR=5.91, 95% CI: 1.44,24.32) and non-communicable (OR=2.85, 95% CI: 1.96, 4.14) diseases were the main predictors of healthcare utilisation.
Out of the 17 dimensions of care assessed in the quantitative component of the quality of care study, operational managers reported dissatisfaction with patient waiting time while patients reported dissatisfaction with the appointment system, defaulter-tracing of patients and
waiting time. The mediation pathway fitted perfectly with the data (coefficient of determination=1.00). The structural equation modeling showed that structure correlated with process (0.40) and outcome (0.75). Given structure, process correlated with outcome (0.88). Patients‘ perception of availability of equipment, supply of critical medicines and accessibility of care (structure construct) had a direct influence on the ability of nurses to attend to their needs, be professional and friendly (process construct). Patients also perceived that these process dimensions directly influenced coherence of care provided, competence of the nurses and patients‘ confidence in the nurses (outcome construct). These structure-related dimensions of care directly influenced outcome-related dimensions of care without the mediating effect of process factors.
In the qualitative study, manager and patient narratives showed inadequacies in structure (malfunctioning blood pressure machines and staff shortage); process (irregular prepacking of drugs); and outcome (long waiting times). Patients reported anti-hypertension drug stock-
outs; sub-optimal defaulter-tracing; rigid clinic appointments; HIV-related stigma in the community resulting from defaulter-tracing activities; and government nurses‘ involvement in commercial activities in the consulting rooms during office hours. Managers reported simultaneous treatment of chronic diseases by traditional healers in the community and thought there was reduced HIV stigma because HIV and NCD patients attended the same clinic.
In the controlled-interrupted time series study the ARMA model showed that the pilot facilities had a 5.7% (coef=0.057; 95% CI: 0.056,0.058; P<0.001) and 1.0% (coef=0.010; 95% CI: 0.003,0.016; P=0.002) greater likelihood than the comparison facilities to control patients‘ CD4 counts and BP, respectively. In the segmented analysis, the decreasing probabilities of controlling CD4 counts and BP observed in the pilot facilities before the implementation of the ICDM model were respectively reduced by 0.23% (coef = -0.0023; 95% CI: -0.0026,
-0.0021; P<0.001) and 1.5% (Coef= -0.015; 95% CI: -0.016, -0.014; P<0.001).
Conclusions: HIV and NCDs were the main health problems and predictors of HCU in the population. This suggests that public healthcare services for chronic diseases are a priority among older people in this rural setting. There was poor quality of care reported in five of the eight priority areas used as leverage for the control of NCDs (referral, defaulter tracing, prepacking of medicines, clinic appointments and waiting time); hence, the need to strengthen services in these areas. Application of the ICDM model appeared effective in reducing the decreasing trend in controlling patients‘ CD4 counts and blood pressure. Suboptimal BP control observed in this study may have been due to poor quality of care in the identified priority areas of the ICDM model and unintended consequences of the ICDM model such as work overload, staff shortage, malfunctioning BP machines, anti-hypertension drug stock-outs, and HIV-related stigma in the community. Hence, the HIV programme should be more extensively leveraged to improve the quality of hypertension treatment in order to achieve optimal BP control in the nationwide implementation of the ICDM model in PHC facilities in South Africa and, potentially, other LMICs.
el in June 2011 in selected Primary Health Care (PHC) facilities, one of the first of such efforts by an African Ministry of Health. The main aim of the ICDM model is to leverage the successes of the innovative HIV treatment programme for NCDs in order to improve the quality of
chronic disease care and health outcomes of adult chronic disease patients. Since the initiation of the ICDM model, little is known about the quality of chronic care resulting in the effectiveness of the model in improving health outcomes of chronic disease patients.
Objectives: To describe the chronic disease profile and predictors of healthcare utilisation (HCU) in a rural population in a South African municipality; and assess quality of care and effectiveness of the ICDM model in improving health outcomes of chronic disease patients receiving treatment in PHC facilities.
Methods: An NDoH pilot study was conducted in selected health facilities in the Bushbuckridge municipality, Mpumalanga province, north‐ east South Africa, where a part of the population has been continuously monitored by the Agincourt Health and Socio-Demographic
Surveillance System (HDSS) since 1992. Two main studies were conducted to address the two research objectives. The first study was a situation analysis to describe the chronic disease profile and predictors of healthcare utilisation in the population monitored by the Agincourt
HDSS. The second study evaluated quality of care in the ICDM model as implemented and assessed effectiveness of the model in improving health outcomes of patients receiving treatment in PHC facilities. This second study had three components: (1) a qualitative and (2) a quantitative evaluation of the quality of care in the ICDM model; and a (3) quantitative assessment of effectiveness of the ICDM model in improving patients‘ health outcomes. The two main studies have been categorised into three broad thematic areas: chronic disease profile and predictors of healthcare utilisation; quality of care in the ICDM model; and changes in patients‘ health outcomes attributable to the ICDM model.
In the first study, a cross-sectional survey to measure healthcare utilisation was targeted at 7,870 adults 50 years and over permanently residing in the area monitored by the Agincourt HDSS in 2010, the year before the ICDM model was introduced. Secondary data on healthcare
utilisation (dependent variable), socio-demographic variables drawn from the HDSS, receipt of social grants and type of medical aid (independent variables) were analysed. Predictors of HCU were determined by binary logistic regression adjusted for socio-demographic vari‐
ables.
The quantitative component of the second study was a cross-sectional survey conducted in 2013 in the seven PHC facilities implementing the ICDM model in the Agincourt sub-district (henceforth referred to as the ICDM pilot facilities) to better understand the quality of care in
the ICDM model. Avedis Donabedian‘s theory of the relationships between structure, process, and outcome (SPO) constructs was used to evaluate quality of care in the ICDM model exploring unidirectional, mediation, and reciprocal pathways. Four hundred and thirty-five (435) proportionately sampled patients ≥ 18 years and the seven operational managers of the PHC facilities responded to an adapted satisfaction questionnaire with measures reflecting structure (e.g. equipment), process (e.g. examination) and outcome (e.g. waiting time) constructs.
Seventeen dimensions of care in the ICDM model were evaluated from the perspectives of patients and providers. Eight of these 17 dimensions of care are the priority areas of the HIV treatment programme used as leverage for improving quality of care in the ICDM model: supply of critical medicines, hospital referral, defaulter tracing, prepacking of medicines, clinic appointments, reducing patient waiting time, and coherence of integrated chronic disease care (a one-stop clinic meeting most of patients‘ needs). A structural equation model was fit to operationalise Donabedian‘s theory using patient‘s satisfaction scores.
The qualitative component of the second study was a case study of the seven ICDM pilot facilities conducted in 2013 to gain in-depth perspectives of healthcare providers and users regarding quality of care in the ICDM model. Of the 435 patients receiving treatment in the pilot facilities, 56 were purposively selected for focus group discussions. An in-depth interview was conducted with the seven operational managers within the pilot facilities and the health manager of the Bushbuckridge municipality. Qualitative data were analysed, with MAXQDA 2
software, to identify 17 a priori dimensions of care and emerging themes. In addition to the emerging themes, codes generated in the qualitative analysis were underpinned by Avedis Donabedian‘s SPO theoretical framework.
A controlled interrupted time-series study was conducted for the 435 patients who participated in the cross-sectional study in the ICDM pilot facilities and 443 patients proportionately recruited from five PHC facilities not implementing the ICDM model (Comparison PHC facilities in the surrounding area outside the Agincourt HDSS) from 2011-2013. Health outcome data for each patient were retrieved from facility records at 30-time points (months) during the study period. We performed autoregressive moving average (ARMA) statistical modelling to account for autocorrelation inherent in the time-series data. The effect of the ICDM model on the control of BP (<140/90 mmHg) and CD4 counts (>350 cells/mm3) was assessed by controlled segmented linear regression analysis.
Results: Seventy-five percent (75%) of the 7,870 eligible adults 50+ responded to the health care utilization survey in the first study. All 5,795 responders reported health problems, of whom 96% used healthcare, predominantly at public health facilities (82%). Reported health
problems were: chronic non-communicable diseases (41% - e.g. hypertension), acute conditions (27% - e.g. flu), other conditions (26% - e.g. musculoskeletal pain), chronic communicable diseases (3% e.g. HIV and TB) and injuries (3%). Chronic communicable (OR=5.91, 95% CI: 1.44,24.32) and non-communicable (OR=2.85, 95% CI: 1.96, 4.14) diseases were the main predictors of healthcare utilisation.
Out of the 17 dimensions of care assessed in the quantitative component of the quality of care study, operational managers reported dissatisfaction with patient waiting time while patients reported dissatisfaction with the appointment system, defaulter-tracing of patients and
waiting time. The mediation pathway fitted perfectly with the data (coefficient of determination=1.00). The structural equation modeling showed that structure correlated with process (0.40) and outcome (0.75). Given structure, process correlated with outcome (0.88). Patients‘ perception of availability of equipment, supply of critical medicines and accessibility of care (structure construct) had a direct influence on the ability of nurses to attend to their needs, be professional and friendly (process construct). Patients also perceived that these process dimensions directly influenced coherence of care provided, competence of the nurses and patients‘ confidence in the nurses (outcome construct). These structure-related dimensions of care directly influenced outcome-related dimensions of care without the mediating effect of process factors.
In the qualitative study, manager and patient narratives showed inadequacies in structure (malfunctioning blood pressure machines and staff shortage); process (irregular prepacking of drugs); and outcome (long waiting times). Patients reported anti-hypertension drug stock-
outs; sub-optimal defaulter-tracing; rigid clinic appointments; HIV-related stigma in the community resulting from defaulter-tracing activities; and government nurses‘ involvement in commercial activities in the consulting rooms during office hours. Managers reported simultaneous treatment of chronic diseases by traditional healers in the community and thought there was reduced HIV stigma because HIV and NCD patients attended the same clinic.
In the controlled-interrupted time series study the ARMA model showed that the pilot facilities had a 5.7% (coef=0.057; 95% CI: 0.056,0.058; P<0.001) and 1.0% (coef=0.010; 95% CI: 0.003,0.016; P=0.002) greater likelihood than the comparison facilities to control patients‘ CD4 counts and BP, respectively. In the segmented analysis, the decreasing probabilities of controlling CD4 counts and BP observed in the pilot facilities before the implementation of the ICDM model were respectively reduced by 0.23% (coef = -0.0023; 95% CI: -0.0026,
-0.0021; P<0.001) and 1.5% (Coef= -0.015; 95% CI: -0.016, -0.014; P<0.001).
Conclusions: HIV and NCDs were the main health problems and predictors of HCU in the population. This suggests that public healthcare services for chronic diseases are a priority among older people in this rural setting. There was poor quality of care reported in five of the eight priority areas used as leverage for the control of NCDs (referral, defaulter tracing, prepacking of medicines, clinic appointments and waiting time); hence, the need to strengthen services in these areas. Application of the ICDM model appeared effective in reducing the decreasing trend in controlling patients‘ CD4 counts and blood pressure. Suboptimal BP control observed in this study may have been due to poor quality of care in the identified priority areas of the ICDM model and unintended consequences of the ICDM model such as work overload, staff shortage, malfunctioning BP machines, anti-hypertension drug stock-outs, and HIV-related stigma in the community. Hence, the HIV programme should be more extensively leveraged to improve the quality of hypertension treatment in order to achieve optimal BP control in the nationwide implementation of the ICDM model in PHC facilities in South Africa and, potentially, other LMICs.
Original language | English |
---|---|
Qualification | Master of Philosophy |
Awarding Institution |
|
Supervisors/Advisors |
|
Award date | 11 Nov 2016 |
Publication status | Published - 2016 |
Externally published | Yes |
Keywords
- Adults; predictors; healthcare utilisation; HIV; Chronic non-communicable diseases (NCDs); Primary Health Care (PHC); Health Outcomes; Integrated Chronic Disease Management (ICDM) Model; Avedis Donabedian; Quality of care; Structural equation model; Interrupted Time-Series; Segmented regression; multilevel regression; Agincourt HDSS study site; South Africa.