Fit to play but goalless: Labour market outcomes in a cohort of public sector ART patients in Free State province, South Africa Frikkie Booysen Department of Economics / Centre for Health Systems Research & Development (CHSR&D) University of the Free State IAEN Pre-Conference Symposium 16-17 July 2010, Vienna
Acknowledgements The financial support of CIDA, DCI, DfID, IDRC, JEAPP, USAID, AUSAID, UNDP, The World Bank s Research Committee, and BNPP Programme HEARD: financial support to attend IAEN/IAS conference Patients in the ART programme who willingy sacrificed their time and energy to participate in this research, and frankly shared their views and experiences with the researchers. The management and health care staff of the Free State Department of Health and of several local municipalities, who facilitated access to the study participants.
Background Adverse macro- and microeconomic impacts of HIV and AIDS are relatively well documented Access to ART is expanding rapidly, although coverage remains sub-optimal In South Africa, mainly accessed through the public health care sector end 2009, approximately 1 million people on ART, of which 95% in the public sector (UNAIDS, 2009) how can ART ameliorate these adverse economic impacts of HIV and AIDS?
Data: CP cohort study Sampling frame Eligible and certified ready to commence HAART in 2004/05 CD4<200 and/or WHO stage 4 + clinical assessment Randomly sampled 80/district proportional to treatment/non-treatment numbers Xhariep = 44 patients only, census Follow-up interviews at approximately 6-9 month intervals Replaced from original sampling frame if lost to follow-up Written, informed consent Nursing sister at assessment site + enumerators Including permission to access patient files Received store gift voucher on completion of interview
Key outcomes Labour market outcomes: (a) Too ill to work (b) Labour force participation (c) Unemployment (d) Absorption TIME: TREATMENT CAREER Treatment outcomes: (a) Clinical markers CD4 count (copies/mm 3 ) RNA level (copies/ml) CD4 > 350 and RNA < 500 (b) Self-reported illness (c) Health-related quality of life EQ-5D EQ-VAS (d) Self-reported side-effects (e) Recent hospitalisation
Key questions How do labour market outcomes vary by treatment duration and/or treatment responses? Are treatment dynamics significant, independent predictors of labour market outcomes?
Table 2: Subjective, self-reported outcomes, by treatment duration Treatment duration Outcome Pre-ART 0-3 months 3-12 months 12-24 months 24-36 months >36 months Feeling ill (%) 39.3 (4.3) 26.9 (3.2) 26.9 (2.3) 19.7 (1.8) 19.5 (2.1) 15.3 (10.4) *** Sample (n) EQ-5D (0-1) 0.706 (0.034) 0.818 (0.020) 0.816 (0.014) 0.833 (0.012) 0.854 (0.013) 0.895 (0.013) *** 1,714 EQ-VAS (%) 63.1 (2.0) 66.5 (1.4) 67.7 (1.0) 70.0 (0.9) 71.4 (1.1) 74.9 (1.3) *** 1,710 Side effects (%) 50.0 (15.0) 52.6 (4.0) 31.4 (2.5) 10.0 (1.4) 8.1 (1.4) 6.6 (1.7) *** 1,541 Severity of side effects (%): No side effects 50.0 47.7 68.5 89.9 92.8 93.9 *** 1,535 Non-disruptive side effe 8.3 4.0 2.3 0.7 0.0 0.0 Minor side effects 16.7 10.6 10.5 2.6 1.6 3.1 Severe side effects 25.0 37.8 18.7 6.8 5.6 3.1 Hospitalised (%) 11.2 (2.8) 8.9 (2.0) 13.2 (1.8) 6.5 (1.1) 6.8 (1.2) 9.0 (2.0) ** 1,705 Note: Standard errors reported in parentheses. Results for side effects only include patients on ARV treatment at the time. Results exclude those patients known to have interrupted their ARV treatment at some time or other during the study (n=27; N=130). Three asterisks denote differences that are statistically significant at the 1% level, while two asterisks denote differences that are statistically significant at the 5% level. Median values of all continuous variables also differ statistically significantly across treatment duration categories (p<0.001).
0 Probability.1.2.3.4.5.6.7.8.9 1 Figure 2a: Predicted probability of being too ill to work, by treatment outcomes and duration -12-6 0 6 12 18 24 30 36 42 48 Treatment duration (months) RNA < 500 CD4 >= 350 RNA < 500 and CD4 >= 350 Being too ill to work Note: Unadjusted predicted probabilities obtained from RE panel probit models. Includes all clinical markers for interviewed study participants. Data obtained from patient files. Results exclude those patients known to have interrupted their ARV treatment at some time or other during the study (n=27).
Figure 2b: Time trends in illness / disability among work force (%) 50 45 40 36.2 37.6 Cohort of public sector ART clients 35 32.2 31.8 32.6 30 28.2 25 22.1 20 15 10 5 LFS estimates, South Africa 5.2 5.2 4.9 4.8 4.8 5.0 4.8 15.0 6.0 0 Jul 04 - Dec 04 Jan 05 - Jun 05 Jul 05 - Dec 05 Jan 06 - Jun 06 Jul 06 - Dec 06 Jan 07 - Jun 07 Jul 07 - Dec 07 Jan 08 - Mar 08 Note: Outcomes for ART patients represent results for balanced panel only, i.e. patients observed in all six survey rounds. Provincial estimates of the numbers of ill / disabled persons in the work force are not available for all LFS survey years. Estimates are for unadjusted LFS data series.
0 Probability.1.2.3.4.5.6.7.8.9 1 Figure 3a: Predicted probability of participating in the labour force, by treatment outcomes and duration -12-6 0 6 12 18 24 30 36 42 48 Treatment duration (months) RNA < 500 CD4 >= 350 RNA < 500 and CD4 >= 350 Participating in the labour force Note: Unadjusted predicted probabilities obtained from RE panel probit models. Includes all clinical markers for interviewed study participants. Data obtained from patient files. Results exclude those patients known to have interrupted their ARV treatment at some time or other during the study (n=27).
Figure 3b: Time trends in labour force participation (%) 80 70 60 50 LFS estimates, Free State province 61.6 60.7 58.9 58.1 56.6 51.3 48.7 57.1 54.1 58.2 55.8 60.8 58.4 63.6 60.4 40 34.5 Cohort of public sector ART clients 30 20 Jul 04 - Dec 04 Jan 05 - Jun 05 Jul 05 - Dec 05 Jan 06 - Jun 06 Jul 06 - Dec 06 Jan 07 - Jun 07 Jul 07 - Dec 07 Jan 08 - Mar 08 Note: Outcomes for ART patients represent results for balanced panel only, i.e. patients observed in all six survey rounds. Estimates are for adjusted LFS data series (Statistics South Africa, 2007/08).
0 Probability.1.2.3.4.5.6.7.8.9 1 Figure 5a: Predicted probability of being absorped in the labour force, by treatment outcomes and duration -12-6 0 6 12 18 24 30 36 42 48 Treatment duration (months) RNA < 500 CD4 >= 350 RNA < 500 and CD4 >= 350 Being absorped in the labour force Note: Unadjusted predicted probabilities obtained from RE panel probit models. Includes all clinical markers for interviewed study participants. Data obtained from patient files. Results exclude those patients known to have interrupted their ARV treatment at some time or other during the study (n=27).
Figure 4: Time trends in unemployment rate (%) 90 80 70 71.1 63.9 Cohort of public sector ART clients 60 58.6 59.5 52.5 50 44.2 46.2 40 33.3 30 20 10 25.2 28.3 26.1 25.4 LFS estimates, Free State province 21.9 23.6 21.4 25.0 0 Jul 04 - Dec 04 Jan 05 - Jun 05 Jul 05 - Dec 05 Jan 06 - Jun 06 Jul 06 - Dec 06 Jan 07 - Jun 07 Jul 07 - Dec 07 Jan 08 - Mar 08 Note: Outcomes for ART patients represent results for balanced panel only, i.e. patients observed in all six survey rounds. Estimates are for adjusted LFS data series (Statistics South Africa, 2007/08).
Figure 5b: Time trends in absorption rate (%) 50 45 40 44.1 43.5 45.5 43.3 44.6 44.5 45.9 LFS estimates, Free State province 45.3 35 30 30.2 30.1 28.9 25 23.0 24.0 25.8 20 15 14.1 18.5 Cohort of public sector ART clients 10 5 0 Jul 04 - Dec 04 Jan 05 - Jun 05 Jul 05 - Dec 05 Jan 06 - Jun 06 Jul 06 - Dec 06 Jan 07 - Jun 07 Jul 07 - Dec 07 Jan 08 - Mar 08 Note: Outcomes for ART patients represent results for balanced panel only, i.e. patients observed in all six survey rounds. Estimates are for adjusted LFS data series (Statistics South Africa, 2007/08).
Table 3: Selected significant predictors of labour market outcomes Treatment duration t : [Comparison group = pre-art] 0-3 months -0.883-5.201 3-12 months -1.675 *** 1.666 *** 12-24 months -1.813 *** 1.559 *** 24-36 months -1.625 *** 1.793 *** >36 months -1.206 * 1.476 ** Self-reported illness t1 - t : [Comparison group = treatment t * not ill in either period] Not on ARV treatment t 1.054 *** -0.767 ** Treatment t * Fell ill t1 - t 0.957 *** -0.633 *** Treatment t * Not ill any longer t1 - t 0.750 *** -0.785 *** Treatment t * Remained ill t1 - t 1.019 *** -0.926 *** Health-related quality of life (EQ-5D) t1 - t : Treatment t * EQ-5D t1 - t -0.202 *** 0.234 *** Presence of self-reported side effects t1 - t : [Comparison group = treatment t * still no side effects] Treatment t * Continued side effects 1.010 *** -0.806 ** Education: [Comparison group = no education] Diploma/degree 2.621 *** Dwelling type t : [Comparison group = formal dwelling] Traditional dwelling -0.540 ** Access to employment network t1 - t : Being too ill to work Participating in the labour force what matters most for securing employment, is education, place of residence, previous employment and Being absorped in the labour force Treatment duration and subjective health outcomes matter for labour force participation, but NOT for securing employment [Comparison group = no other employed person in hhold in either period] Household included employed person in both periods links to the labour market t1 - t 1.179 *** Employed when first tested HIV-positive 1.096 *** Note: Results are for pooled or random effects (RE) panel probit models. All models are statistically significant in respect of overall fit (p<0.001). Results are reported as marginal effects of type dydx. Adjusted for gender, age, race, education, dwelling, marital status, dependency ratio, employment status at first HIVpositive test, access to disability grant, breadwinner status, access to inter-household employment networks, self-reported stigmatisation, district, follow-up duration, and month and year of interview. Three asterisks denote differences statistically significant at the 1% level, two and one asterisk significance at 5% and 10% levels respectively.
what then of social assistance? Social assistance: disability grant coverage has grown rapidly, but have since stabilized in terms of expenditure and number of beneficiaries Nattrass (2006/07) points out the dilemma of trading off income and wealth and the resultant threats to ART s long-term sustainability, BUT is this just qualitative anecdote or stated preferences OR can Venakataramani et al s (2009) results from Khayalitsha be generalized to the larger ART programme?
Figure 6: DG Access in CP6 Cohort, by Survey Round 90 80 70 60 50 40 30 20 10 0 58.2 51.0 43.8 DG income per month: Mean R805 Median R820 IQR R780 R870 69.7 62.7 55.7 74.3 67.5 60.7 Composition of income: DG income represents ~75% of total monthly individual income per month 79.7 78.7 73.4 72.3 67.0 65.8 Transitions in DG access: Over the study period, 167 clients gained access to a DG grant between subsequent periods, whereas 109 lost access to the DG grant between subsequent periods 1 2 3 4 5 6 82.2 76.0 69.9
Table 4: Transitions in DG Access and Adherence and Treatment Outcomes Missed a clinic visit in past month Did not miss a dose(s) in past week EQ-5D EQ-VAS CD4 RNA CD4>350 and RNA<500 Transitions in DG access: Gained versus no grant 0.855 [pooled] - 0.015 [pooled] -0.333 [pooled] -10.264 [FE] - 1.736 [FE] Lost versus kept grant 2.213 *** [pooled] 0.325 * [pooled] -0.074 * [FE] -5.623 ** [FE] 24.292 [FE] - 0.942 [FE], BUT what explains these effects? Model type logistic logistic linear linear linear linear Logistic Note: Provisional results only. Results are for Fixed Effects (FE) or Pooled regression models. Where no coefficients are reported, neither the FE nor the pooled model performed adequately in terms of overall fit. Results are adjusted for age, education, place of residence and survey round. Asterisks denote 1%, 5% and 10% levels of statistic significance, respectively.
(1) Methodological: Limitations Observations of clinical (dates of facility visits) and labour market outcomes (interview dates) are not synchronised Limited information regarding actual timing and/or duration of the observed labour market outcomes Counterfactual unclear in absence of comparative samples of HIVnegative and/or HIV-positive persons not on ART Potential attrition and selection bias in socio-demographics and key clinical and labour market outcomes affects generalisability (2) Econometric: Poor overall fit of regression models for transitions in labour market outcomes: poor specification and/or unobservables? Over -adjusting for time: treatment effects may dissolve or vanish if adjusting for temporal dimensions of survey data? Endogeneity of selected explanatory variables requires instrumentation
Key findings (1) Significant increases in labour force participation, due to improvements in self-reported health rather than clinical markers, BUT public sector ART patients are worse off compared to representative SA or FS populations, with exception of illness/disability and labour force participation Sustainable, effective treatment therefore grows the army of potential workers, but does not secure or buy someone a job what does? Education, household links to labour market, and being employed at first HIV+ test significantly associated with employment and absorption rather than treatment duration or outcomes, BUT significantly worse employment prospects in informal settlements and rural areas ( traditional dwellings )
Key findings (2) Yet, for now a high proportion of this cohort are still receiving disability grants ( suboptimal targeting?), although theoretically these grants should fall away in the presence of treatment success No comparable income-support for unemployed adults, which means the disincentive effect(s) of disability grants are likely to remain, unless, amongst others, we have successful urban and especially rural development strategies pro-poor, labour intensive growth strategies large(r)-scale public works programmes job creation/pro-poor growth strategies further education and training (FET) basic income grant (BIG)
Key findings (3) On another level, it is not clear what the implications are for the economic argument in favour of providing free or subsidized treatment, this following the global economic and financial crisis? direct benefits for the economy from public sector ART relatively limited, BUT evidence of the indirect benefits or potential positive (generational) externalities of provision of ARV treatment (e.g. time allocation, childrens schooling and health, prevention benefits) may help to strengthen our case?
Key findings (4) Our preliminary findings suggest that well targeted, short-term conditional transfers represent potential cost-effective enablers in ART or other programmes, but we agree with the received wisdom that disability or chronic disease grants are not sustainable in the long-term, what then? According to Russell et al (2007: 344), considerable challenges remain for people who are trying to live with HIV as a manageable chronic condition. ART programmes need to seek convergence with economic programmes that have expertise in livelihood support and promotion [and] social protection initiatives. The future for those on ART depends not only on the provision of medicine but also on economic and social support for rebuilding lives and livelihoods.