Does pharmaceutical information systems data inform decision-making in public healthcare? Utility of a national system in a limited resource setting.

BACKGROUND: Globally, weak pharmaceutical information systems (PIS) negatively affect universal health coverage and outcomes. Few studies in sub-Saharan Africa qualitatively and quantitatively assess drivers and utility of data from PIS in public healthcare. METHODS: A nationwide cross-sectional descriptive study interviewed PIS focal persons in all 14 regions of Namibia. The primary outcome was extent and predictors of utility of PIS data. The extent of utility of PIS data was determined using descriptive statistics and predictors by logistic regression in SPSSv24 or thematic analysis for qualitative data. RESULTS: The study recruited 58 key informants at facility-based 56 (96.6%) and national 2 (3.4%) levels. Of the 56 facility-based respondents, 29 (51.8%) were female and 27 (48.2%) pharmacists. The mean age and PIS work experience were 33.5 ± 7.6years and 4.5 ± 3.3years respectively. The utility level of PIS data was 34 (60.7%) (target >80%). A total of 103 uses of PIS data were cited; of which 38 (36.9%) were informing decisions on rational medicine use, 27 (26.2%) on pharmaceutical stock management and 24 (23.3%) on strengthening pharmacy workforce. The utility of PIS data significantly decreased with lack of systems on routine reporting by health facility in-charge (cOR = 0.25, 95%CI: 0.06,0.90, p = 0.035). Longer work experience (cOR = 1.05, 95%CI: 0.88,1.25, p = 0.58), formal consultations (cOR = 1.29, 95%CI: 0.14,11.54, p = 0.82), and availability of feedback systems (cOR = 1.08, 95%CI: 0.33,3.56, p = 0.89) appeared to increase utility of PIS data. Two thematic drivers of utility of PIS data were programmatic "feedback and action on PIS; structures,technical supportfor PIS discussion"; technical "training/technical capacity of staff; tools and resources for data collection and utilization"; and human-resource "staff availability and workload; attitude and commitment". CONCLUSION: The nationwide study shows sub-optimal utility of PIS data in public healthcare in Namibia, which negatively affects delivery of pharmaceutical services. This calls for action to enhance capabilities for utilization of automated real-time pharmaceutical information decision support systems to enhance real-time analysis and feedback on medicines data in resource-limited settings.

Optimizing data quality of pharmaceutical information systems in public health care in resource limited settings.

BACKGROUND: Robust pharmaceutical management information systems (PMIS) strengthen healthcare planning and delivery. Few nationwide studies in resource limited settings in Africa validate the data quality of PMIS in public healthcare. OBJECTIVE: To determine predictors and quality of data in a nationwide PMIS database in Namibia. METHODS: A population-level analysis of the quality of data i.e. completeness, accuracy and consistency in a nationwide PMIS database, 2007-2015. Data quality of the PMIS was determined by three domains, completeness, accuracy and consistency. Data completeness was determined by level of missing data in SPSSv25, with acceptable level set at <5%. Data accuracy was determined by proportion of PMIS indicators with extreme outliers. Data consistence was determined by patterns of missingness, i.e. random or systematic. Predictors of data quality were determined using logistic regression modelling. RESULTS: A total of 544 entries and 12 indicators were registered in the PMIS at 38 public health facilities. All the PMIS indicators had missing data and 50% (n = 6) had inaccurate data i.e. extreme values. The data for most PMIS indicators (75%, n = 12) were consistent with the pattern of missing completely at random (MCAR, i.e. missingness <5%). Incompleteness of PMIS data was highest for average number of prescriptions 6%, annual expenditure per capita for pharmaceuticals 5% and population per pharmacist's assistant 5%. The main predictors of poor quality of PMIS data were year of reporting of PMIS data (p = 0.035), level of health facility (p < 0.001), vital reference materials available at the pharmacy (p = 0.002), and pharmacists' posts filled (p = 0.013). CONCLUSIONS: The data quality of PMIS in public health care in Namibia is sub-optimal and widely varies by reporting period, level of health facility and region. The integration of data quality assurance systems is required to strengthen quality of PMIS data to optimize quality of PMIS data in public health care.

Validity of World Health Organisation prescribing indicators in Namibia's primary healthcare: findings and implications.

OBJECTIVE: World Health Organization/International Network of Rational use of Drugs (WHO/INRUD) indicators are widely used to assess medicine use. However, there is limited evidence on their validity in Namibia's primary health care (PHC) to assess the quality of prescribing. Consequently, our aim was to address this. DESIGN, SETTING, PARTICIPANTS AND INTERVENTIONS: An analytical cross-sectional survey design was used to examine and validate WHO/INRUD indicators in out-patient units of two PHC facilities and one hospital in Namibia from 1 February 2015 to 31 July 2015. The validity of the indicators was determined using two-by-two tables against compliance to the Namibian standard treatment guidelines (NSTG). The receiver operator characteristics for the WHO/INRUD indicators were plotted to determine their accuracy as predictors of compliance to agreed standards. A multivariate logistic model was constructed to independently determine the prediction of each indicator. MAIN OUTCOMES AND RESULTS: Out of 1243 prescriptions; compliance to NSTG prescribing in ambulatory care was sub-optimal (target was >80%). Three of the four WHO/INRUD indicators did not meet Namibian or WHO targets: antibiotic prescribing, average number of medicines per prescription and generic prescribing. The majority of the indicators had low sensitivity and/or specificity. All WHO/INRUD indicators had poor accuracy in predicting rational prescribing. The antibiotic prescribing indicator was the only covariate that was a significant independent risk factor for compliance to NSTGs. CONCLUSION: WHO/INRUD indicators showed poor accuracy in assessing prescribing practices in ambulatory care in Namibia. There is need for appropriate models and/or criteria to optimize medicine use in ambulatory care in the future.