Main Article Content
Despite the implementation of full-day kindergarten (FDK) in several Canadian provinces, there is little evidence on the long-term outcomes associated with this program. Our objective was to use population-level linked data sources from Manitoba, Canada, to determine whether FDK results in better long-term academic outcomes and reduced inequities in outcomes.
Using data held in the Manitoba Centre for Health Policy Data Repository we examined provincial reading and numeracy assessments in grades 3, 7, and 8 and a performance index in grade 9 for students in two Manitoba school divisions between 1999-2012. In School Division A (SDA), FDK is targeted in the lowest SES schools; in School Division B (SDB) FDK was gradually introduced universally. SDA FDK students were matched using propensity scores to students in an adjacent school division with similar socioeconomic status (SES) but no FDK; in SDB a stepped-wedge design was used. Logistic regressions accounted for confounders including classroom effects and sex. Gamma sensitivity analyses were used to assess sensitivity of results to unmeasured confounding. The Kakwani Progressivity Index (KPI) determined how FDK affected equity.
There were 224-544 children in FDK and 869-1923 non-FDK matches in SDA, depending on the outcome examined; numbers in SDB ranged from 335-707 (FDK) and 222-475 (non-FDK). Including interactions, 35 comparisons were examined in SDA and 24 in SDB. None of the outcomes examined in SDB showed statistically significant effects of FDK that were robust to unmeasured confounding. In SDA there were only 3 statistically significant and robust findings of benefits of FDK, all related to math. Comparisons of KPIs for FDK and non-FDK children in both school divisions demonstrated inequities in outcomes associated with SES, however there were no significant differences in equity between the FDK and non-FDK children for any of the outcomes.
Our findings indicate no apparent benefits of universal FDK, and limited benefits from targeted FDK, specifically long-term improvements in numeracy for low-income girls. No reductions in inequity were found. Decisions regarding FDK implementation should weigh the costs of this program against the limited long-term academic benefits.
In the absence of whole-of-population data regarding depression in Australia, antidepressant supply identified through pharmaceutical data has been used by some as an indicator of depression. This approach has been criticised on the basis that up to 30% of antidepressants are prescribed for indications other than depression, including anxiety disorders, insomnia and pain. This study examines whether the identification of patients treated for depression can be improved by refining this antidepressant-based indicator via a series of pre-determined algorithms.
Pharmaceutical Benefits Scheme (PBS) and Medicare Benefits Scheme (MBS) records were linked to follow-up questionnaires completed between September 2012 and December 2014 by participants of the 45 and Up Study - a cohort study of residents of New South Wales, Australia, aged 45 years and older. After exclusions, 58,425 participants were included in the analyses. According to the basic antidepressant-based indicator, the supply of any antidepressant (Anatomical Therapeutic Chemical classification (ATC) code beginning with N06A) in the 30 days prior to the survey completion date was considered indicative of depression treatment in the last month. This algorithm was refined to: i) exclude tricyclic antidepressants (ATC code N06AA), which are commonly prescribed for insomnia and pain; and ii) re-categorise as ‘not treated for depression’ those antidepressant recipients who were also supplied an anxiolytic or sedative (ATC codes beginning with N05B and N05C) in the 12 months prior to the survey. Self-reported receipt of treatment for depression in the last month, from the questionnaire data, was used as a gold standard.
Results: The basic antidepressant-based indicator returned a sensitivity (Sn) of 59.9%, a positive predictive value (PPV) of 43.4% and a specificity (Sp) of 94.7%. When refined algorithm i) was applied, the PPV and Sp increased to 51.8% and 96.5% respectively, while Sn decreased to 54.6%. Refined algorithms ii) yielded similar PPVs and Sps to algorithm i) while Sns were lower. Further refinements to the indicator will be explored using primary care (MBS) data. Although MBS data do not contain diagnoses, they do contain indicators of when certain mental health services were provided, allowing for algorithms in which the prescription of antidepressants for mental health reasons is distinguished from their prescription for physical health problems.
The algorithms developed in this study can be applied to identify depression in future research based on Australian administrative health data.
We acknowledge the Commonwealth Department of Human Services for supplying PBS and MBS data.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.