Social Science & Medicine 113 (2014) 1e4
Contents lists available at ScienceDirect
Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed
Health worker performance in rural health organizations in low- and middle-income countries: Do organizational factors predict non-task performance? Rohan Jayasuriya a, *, Upali W. Jayasinghe b, Qian Wang a a b
School of Public Health and Community Medicine, University of New South Wales, Sydney, NSW 2052, Australia Research Centre for Primary Health Care and Equity, University of New South Wales, Sydney, NSW 2052, Australia
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 December 2012 Received in revised form 4 March 2014 Accepted 30 April 2014 Available online 2 May 2014
Health worker (HW) performance is a critical issue facing many low- and middle-income countries (LMICs). The aim of this study was to test the effects of factors in the work environment, such as organizational culture and climate, on HW non-task performance in rural health work settings in a LMIC. The data for the study is from a sample of 963 HWs from rural health centres (HCs) in 16 of the 20 provinces in Papua New Guinea. The reliability and validity of measures for organizational citizenship behaviour (OCB), counterproductive work behaviour (CWB) and work climate (WC) were tested. Multilevel linear regression models were used to test the relationship of individual and HC level factors with non-task performance. The survey found that 62 per cent of HCs practised OCB “often to always” and 5 percent practised CWB “often to always”. Multilevel analysis revealed that WC had a positive effect on organizational citizenship behaviour (OCB) and a negative effect on CWB. The mediation analyses provided evidence that the relationship between WC and OCB was mediated through CWB. Human resource policies that improve WC in rural health settings would increase positive non-task behaviour and improve the motivation and performance of HWs in rural settings in LMICs. Ó 2014 Published by Elsevier Ltd.
Keywords: Papua new Guinea Organizational citizenship behaviour Counterproductive work behaviour Work climate Rural health workers Rural health centres No-task performance
1. Introduction Health workers are central to the delivery of health care and their performance determines how well the system functions. The problem of inadequate HW performance in LMIC is an important issue that has garnered attention worldwide (Rowe et al., 2005; WHO, 2006). Rotundo and Sackett (2002: 66) defined job performance as “those actions and behaviours that are under the control of the individual and contribute to the goals of the organization”. However, shortcomings in facilities, medicines, supplies, equipment, that are not under the control of HWs, impose constraints on job performance in LMIC settings (WHO, 2006; Jayasuriya et al., 2012). Both Viswesvaran and Ones (2000) and Rotundo and Sackett (2002) concluded that three broad dimensions of job performance can be distinguished, namely (i) task performance, (ii) organizational citizenship behaviour (OCB), and (iii) counterproductive work behaviour (CWB). Individual job performance entails
* Corresponding author. E-mail address: [email protected] (R. Jayasuriya). http://dx.doi.org/10.1016/j.socscimed.2014.04.042 0277-9536/Ó 2014 Published by Elsevier Ltd.
both a task aspect (productivity or doing the job) and a non-task behavioural aspect, such as helping out colleagues, treating people respectfully, and doing the work efficiently and effectively. So far to our knowledge, the only evidence of the existence of non-task performance of HWs in LMICs has come from qualitative studies that highlight HW performance problems (Dieleman et al., 2003; Lindelow and Serneels, 2006; Razee et al., 2012). There is some evidence that a positive work (organizational) climate impacts on HW motivation and consequently on the performance of HW in LMICs (Perry et al., 2005). Organizational climate is a collective phenomenon and it “represents shared perceptions among unit employees regarding their perceptions of the work environment”. Although it is an individual’s perception, it is a property of the unit (Kuenzi and Schminke, 2009: 639). Overall, a strong relationship exists between WC and the attitudes of personnel towards their job (Kuenzi and Schminke, 2009). Studies have shown that individuals perform OCB as a reaction to positive treatment at work (Rioux and Penner, 2001). West and Richter (2011: 262) in a review of research on team climate, stated that “although many studies suggest climate has a direct effect on team and individual level outcomes, more research
2
R. Jayasuriya et al. / Social Science & Medicine 113 (2014) 1e4
is needed to investigate the moderating or mediation role of (team) climate”. The aim of this study was to test the effects of organizational culture and climate on HW non-task performance in rural health work settings. We present the methods and findings and discuss the implications of the findings for policies for strengthening human resource management in LMICs. 2. Methods 2.1. Sample This paper is based on data collected in a broader study examining rural HW motivation and performance in Papua New Guinea (PNG) and followed a qualitative study (Razee et al., 2012). The data were collected when rural HWs attended national competency training provided to HWs for the introduction of a new malaria diagnosis and treatment protocol, over a 6-month period in late 2010 and early 2011. HWs were given a self-administered survey in English. Anonymity of respondents was ensured as names and other identifiable information were not collected. Completed questionnaires were placed in sealed envelopes by the respondents, collectively packaged by the provincial level trainers and mailed to researchers at the University of New South Wales. Ethical clearance was obtained from the Medical Research Advisory Committee of PNG and the Human Research Ethics Committee at the University of New South Wales. We collected 1302 completed questionnaires from HWs of rural HCs in 16 of the 20 provinces. For this study, we excluded questionnaires where the HC code was incomplete; therefore, the sample for this study was 963 HWs from 148 HCs . 2.2. Development and validation of measures A well-accepted conceptualization of OCB distinguishes OCB that benefits the individual (OCBI) and that which benefits the organization (OCBO) (Williams and Anderson, 1991). For this study, the OCB items were selected from the OCBI dimension, to reflect the individual level, as the theory tested was about individual HW performance. Using the measures of Lee and Allen (2002) and Franco et al. (2004), we identified five items to measure OCBI, that were also relevant to the setting and the findings of the qualitative study. Items for CWB were selected from Bennett and Robinson’s (2000) dimension of organizational deviance behaviour and from our qualitative findings. For both measures, respondents were asked to state the frequency of behaviours of their co-workers (on a scale from never ¼ 1 to always ¼ 6), to reduce socially desirable responses in self-ratings and to evoke more truthful answers. Items and scales used in the study are in appendix A of the web attachment. The use of cross (mixed) level models where group level measures (e.g. organizational culture, WC) are used to predict individual level measures (e.g. OCB) is common in organizational research literature (Klein et al., 2001). A measure of work group climate developed by Perry et al. (2005), which has been tested in many LMICs and found to be reliable (Cronbach alpha ¼ 0.81) and valid was used. The type of organization (government or church) was used as a proxy for organizational level culture. Internal consistency reliability (Cronbach alpha) for the measures was 0.76, 0.74 and 0.65 for OCB, CWB and WC respectively. We tested the construct validity of the scales using a three-factor confirmatory factor analysis (CFA). The model fit was adequate with no cross loadings across scales. The model fit indices were X2 ¼ 215.7 (df 62), GFI ¼ 0.96, AGFI ¼ 0.95, CFI ¼ 0.93, RMSEA ¼ 0.053. The
correlations between scales were OCB and CWB ¼ 0.30; CWB and WC ¼ 0.22 and OCB and WC ¼ 0.51. All were significant (p < 0.01). 2.3. Data analysis List-wise deletion of 71 cases (as they contained missing values for most variables of interest) was followed by missing value imputation using the SPSS expectation maximization method. Ten cases from the dataset were deleted as they belonged to work teams with fewer than three staff, leaving 882 cases from 145 HCs for analysis. Multilevel linear regression models allow both level 1 and level 2 independent variables in the model and the proportion of variability attributed to each level of analysis to be determined (Rasbash et al., 2005). We tested a multilevel model with OCB (at individual level) as the dependent variable. We included both individual level (age, gender, profession and CWB) and HC level (work climate and type of organization) variables as independent variables. The significance of the fixed and random parameter variance estimates was assessed using the Wald joint c2 test statistic (Rasbash et al., 2005). We used 2 log likelihood, known as the “change in the deviance”, which uses a Chi-square distribution to test whether the difference between the models is statistically significant (Table 2) (Jayasinghe et al., 2008). All multilevel models were performed with MLwiN version 2.25 (Rasbash et al., 2005). To assess mediating effects, a product of coefficient test appropriate for multilevel data was used (Krull and MacKinnon, 2001). This tests (1) estimates the effect of the average WC on changes in the mediator CWB (a coefficient) by regressing changes in the mediator onto the average WC; (2) estimates the independent
Table 1 Demographic and organizational behavioural characteristics of respondents from government-owned and church-owned health centres at unit level.
Age group 13 Nurseb percentage 0e40 41e60 61e100 OCBc
Contents lists available at ScienceDirect
Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed
Health worker performance in rural health organizations in low- and middle-income countries: Do organizational factors predict non-task performance? Rohan Jayasuriya a, *, Upali W. Jayasinghe b, Qian Wang a a b
School of Public Health and Community Medicine, University of New South Wales, Sydney, NSW 2052, Australia Research Centre for Primary Health Care and Equity, University of New South Wales, Sydney, NSW 2052, Australia
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 December 2012 Received in revised form 4 March 2014 Accepted 30 April 2014 Available online 2 May 2014
Health worker (HW) performance is a critical issue facing many low- and middle-income countries (LMICs). The aim of this study was to test the effects of factors in the work environment, such as organizational culture and climate, on HW non-task performance in rural health work settings in a LMIC. The data for the study is from a sample of 963 HWs from rural health centres (HCs) in 16 of the 20 provinces in Papua New Guinea. The reliability and validity of measures for organizational citizenship behaviour (OCB), counterproductive work behaviour (CWB) and work climate (WC) were tested. Multilevel linear regression models were used to test the relationship of individual and HC level factors with non-task performance. The survey found that 62 per cent of HCs practised OCB “often to always” and 5 percent practised CWB “often to always”. Multilevel analysis revealed that WC had a positive effect on organizational citizenship behaviour (OCB) and a negative effect on CWB. The mediation analyses provided evidence that the relationship between WC and OCB was mediated through CWB. Human resource policies that improve WC in rural health settings would increase positive non-task behaviour and improve the motivation and performance of HWs in rural settings in LMICs. Ó 2014 Published by Elsevier Ltd.
Keywords: Papua new Guinea Organizational citizenship behaviour Counterproductive work behaviour Work climate Rural health workers Rural health centres No-task performance
1. Introduction Health workers are central to the delivery of health care and their performance determines how well the system functions. The problem of inadequate HW performance in LMIC is an important issue that has garnered attention worldwide (Rowe et al., 2005; WHO, 2006). Rotundo and Sackett (2002: 66) defined job performance as “those actions and behaviours that are under the control of the individual and contribute to the goals of the organization”. However, shortcomings in facilities, medicines, supplies, equipment, that are not under the control of HWs, impose constraints on job performance in LMIC settings (WHO, 2006; Jayasuriya et al., 2012). Both Viswesvaran and Ones (2000) and Rotundo and Sackett (2002) concluded that three broad dimensions of job performance can be distinguished, namely (i) task performance, (ii) organizational citizenship behaviour (OCB), and (iii) counterproductive work behaviour (CWB). Individual job performance entails
* Corresponding author. E-mail address: [email protected] (R. Jayasuriya). http://dx.doi.org/10.1016/j.socscimed.2014.04.042 0277-9536/Ó 2014 Published by Elsevier Ltd.
both a task aspect (productivity or doing the job) and a non-task behavioural aspect, such as helping out colleagues, treating people respectfully, and doing the work efficiently and effectively. So far to our knowledge, the only evidence of the existence of non-task performance of HWs in LMICs has come from qualitative studies that highlight HW performance problems (Dieleman et al., 2003; Lindelow and Serneels, 2006; Razee et al., 2012). There is some evidence that a positive work (organizational) climate impacts on HW motivation and consequently on the performance of HW in LMICs (Perry et al., 2005). Organizational climate is a collective phenomenon and it “represents shared perceptions among unit employees regarding their perceptions of the work environment”. Although it is an individual’s perception, it is a property of the unit (Kuenzi and Schminke, 2009: 639). Overall, a strong relationship exists between WC and the attitudes of personnel towards their job (Kuenzi and Schminke, 2009). Studies have shown that individuals perform OCB as a reaction to positive treatment at work (Rioux and Penner, 2001). West and Richter (2011: 262) in a review of research on team climate, stated that “although many studies suggest climate has a direct effect on team and individual level outcomes, more research
2
R. Jayasuriya et al. / Social Science & Medicine 113 (2014) 1e4
is needed to investigate the moderating or mediation role of (team) climate”. The aim of this study was to test the effects of organizational culture and climate on HW non-task performance in rural health work settings. We present the methods and findings and discuss the implications of the findings for policies for strengthening human resource management in LMICs. 2. Methods 2.1. Sample This paper is based on data collected in a broader study examining rural HW motivation and performance in Papua New Guinea (PNG) and followed a qualitative study (Razee et al., 2012). The data were collected when rural HWs attended national competency training provided to HWs for the introduction of a new malaria diagnosis and treatment protocol, over a 6-month period in late 2010 and early 2011. HWs were given a self-administered survey in English. Anonymity of respondents was ensured as names and other identifiable information were not collected. Completed questionnaires were placed in sealed envelopes by the respondents, collectively packaged by the provincial level trainers and mailed to researchers at the University of New South Wales. Ethical clearance was obtained from the Medical Research Advisory Committee of PNG and the Human Research Ethics Committee at the University of New South Wales. We collected 1302 completed questionnaires from HWs of rural HCs in 16 of the 20 provinces. For this study, we excluded questionnaires where the HC code was incomplete; therefore, the sample for this study was 963 HWs from 148 HCs . 2.2. Development and validation of measures A well-accepted conceptualization of OCB distinguishes OCB that benefits the individual (OCBI) and that which benefits the organization (OCBO) (Williams and Anderson, 1991). For this study, the OCB items were selected from the OCBI dimension, to reflect the individual level, as the theory tested was about individual HW performance. Using the measures of Lee and Allen (2002) and Franco et al. (2004), we identified five items to measure OCBI, that were also relevant to the setting and the findings of the qualitative study. Items for CWB were selected from Bennett and Robinson’s (2000) dimension of organizational deviance behaviour and from our qualitative findings. For both measures, respondents were asked to state the frequency of behaviours of their co-workers (on a scale from never ¼ 1 to always ¼ 6), to reduce socially desirable responses in self-ratings and to evoke more truthful answers. Items and scales used in the study are in appendix A of the web attachment. The use of cross (mixed) level models where group level measures (e.g. organizational culture, WC) are used to predict individual level measures (e.g. OCB) is common in organizational research literature (Klein et al., 2001). A measure of work group climate developed by Perry et al. (2005), which has been tested in many LMICs and found to be reliable (Cronbach alpha ¼ 0.81) and valid was used. The type of organization (government or church) was used as a proxy for organizational level culture. Internal consistency reliability (Cronbach alpha) for the measures was 0.76, 0.74 and 0.65 for OCB, CWB and WC respectively. We tested the construct validity of the scales using a three-factor confirmatory factor analysis (CFA). The model fit was adequate with no cross loadings across scales. The model fit indices were X2 ¼ 215.7 (df 62), GFI ¼ 0.96, AGFI ¼ 0.95, CFI ¼ 0.93, RMSEA ¼ 0.053. The
correlations between scales were OCB and CWB ¼ 0.30; CWB and WC ¼ 0.22 and OCB and WC ¼ 0.51. All were significant (p < 0.01). 2.3. Data analysis List-wise deletion of 71 cases (as they contained missing values for most variables of interest) was followed by missing value imputation using the SPSS expectation maximization method. Ten cases from the dataset were deleted as they belonged to work teams with fewer than three staff, leaving 882 cases from 145 HCs for analysis. Multilevel linear regression models allow both level 1 and level 2 independent variables in the model and the proportion of variability attributed to each level of analysis to be determined (Rasbash et al., 2005). We tested a multilevel model with OCB (at individual level) as the dependent variable. We included both individual level (age, gender, profession and CWB) and HC level (work climate and type of organization) variables as independent variables. The significance of the fixed and random parameter variance estimates was assessed using the Wald joint c2 test statistic (Rasbash et al., 2005). We used 2 log likelihood, known as the “change in the deviance”, which uses a Chi-square distribution to test whether the difference between the models is statistically significant (Table 2) (Jayasinghe et al., 2008). All multilevel models were performed with MLwiN version 2.25 (Rasbash et al., 2005). To assess mediating effects, a product of coefficient test appropriate for multilevel data was used (Krull and MacKinnon, 2001). This tests (1) estimates the effect of the average WC on changes in the mediator CWB (a coefficient) by regressing changes in the mediator onto the average WC; (2) estimates the independent
Table 1 Demographic and organizational behavioural characteristics of respondents from government-owned and church-owned health centres at unit level.
Age group 13 Nurseb percentage 0e40 41e60 61e100 OCBc
