Highly effective antiretroviral drug (ARV) interventions are available that can reduce perinatal HIV transmission rates among infants to less than 2%, even in resource-limited settings.1 Yet, although there is unprecedented commitment to eliminate new pediatric infection and some reductions have been achieved, more than 390,000 children worldwide still become infected with HIV each year.2,3 Only 50% of pregnant women are tested for HIV in the highest-burden countries in east and southern Africa, and only 68% of HIV-infected pregnant women receive some form of ARVs in these countries.4 Failure to identify pregnant women who are HIV infected and provide them with appropriate ARV interventions will result in continued HIV infections among children and deaths among mothers.4,5
In KwaZulu-Natal (KZN) Province, South Africa, antenatal HIV prevalence in 2010 ranged between 31.1% and 42.3% across 11 districts.6 Perinatal mother-to-child HIV transmission rates, estimated among infants aged 6 weeks, varied from 4.4%–10.1% in 2009 and HIV-attributable maternal mortality was estimated at 40.5% in the triennium from 2008 to 2010.7,8 Eligibility criteria for starting pregnant HIV-infected women on lifelong antiretroviral therapy (ART) were revised in South Africa in 2010 following an update of international recommendations.9 Despite this change in policy and the availability of key resources and medications, ART initiation rates of pregnant women have remained low. In 2008, 97.3% of women in South Africa attending antenatal care received an HIV test, 91.3% of women testing positive had been issued nevirapine and 69.8% of babies received nevirapine at delivery.10 Among all HIV-infected pregnant women eligible for ART by the 2008 guidelines, only 12.8% had visited an ARV clinic and only 8.7% had initiated ART.10
HIV-infected pregnant women with low CD4 counts and in need of ART are most likely to transmit the virus to their infants and die themselves. About 40% of HIV-infected mothers have CD4 counts <350/dL, and this subset of women account for 80% of infant transmissions and suffer 80% of HIV-associated maternal deaths.11 Between 2008 and 2010, institutional maternal mortality across KZN was estimated to be 192.3 per 100,000 live births, and the worsening of this rate over the past decade has been attributed in part to the HIV epidemic.8 Identifying eligible HIV-infected pregnant women and starting them on ART is therefore likely to have major public health benefits.
Simple continuous quality improvement (QI) health system interventions that use local data to inform health providers about their own performance have been used in other settings to improve the consistency and reliability of health care delivery, including HIV-related interventions.12–14 Using QI approaches, a district health management team (DHMT) in KZN undertook a 6-month intensive intervention—a “campaign”—to fast-track eligible HIV-infected pregnant women onto ART. Results from this district are contrasted with another district in KZN, which serves as a counterfactual.
Study Setting and Population
Ugu Health District in southwestern KZN has a population of 704,023 distributed across a mix of rural, peri-urban, and urban settings with the majority (76%) living in rural areas.15 The district has 65 health care facilities including 47 primary health care clinics, 14 mobile clinics, and 4 district hospitals. The DHMT is led by a District Health Manager with a coordinator responsible for delivery of the prevention of mother-to-child transmission (PMTCT) program. The District’s Chief Medical Officer and the District Information Officer along with other support staff from the District office made up the remaining members of the DHMT. Almost all were clinically trained (primarily nursing) with the Chief Medical Officer and District Manager trained as medical doctors.
A comparison district from within KZN was selected based on demographic and health system characteristics, to establish a counterfactual. Districts within KZN are subject to province-wide ARV policies and procedures, governance structures, health information systems, and health systems management and staffing practices. In addition, Umzinyathi District was most closely matched to Ugu District in terms of population size (483,572) and distribution (80% living in rural settings) and health system characteristics (38 primary health clinics, 9 mobile clinics, and 4 district hospitals) and was selected as the comparison. The DHMT structure and management systems are similar in the 2 districts. Antenatal HIV prevalence in Ugu was estimated to be 37.3% and in Umzinyathi was 31.7%.16
A health systems improvement team (20,000+) from the University of KZN (UKZN) has been supporting systematic efforts to improve the delivery of HIV care to pregnant women in Ugu since 2008. Through these and other related efforts of the DHMT and its partners, the district reported high levels of HIV testing, CD4 testing, and delivery of single- or dual-ARV therapy to pregnant women in the antenatal period.10,16 In mid-2009, however, few eligible HIV-infected pregnant women were being started on ART during the antenatal period. As a result, the DHMT requested the input of the 20,000+ team to design and guide the implementation of a campaign among district health workers based on QI principles, to fast-track eligible pregnant women onto ART.
The campaign borrowed from theoretical frameworks used to guide hospital patient safety campaigns that have been used over the past decade to improve delivery of evidence-based interventions in the United States and the United Kingdom. In those efforts, the conceptual model has 4 important elements: a galvanizing and time-bound aim; a set of clear, evidence-based healthcare interventions; a social system to organize the participation of numerous hospitals and healthcare organizations; and a measurement framework focused on the primary aim.17,18 In Ugu, the Campaign set a 6-month goal of developing a reliable district-wide system to find and initiate 90% of eligible women on ART during their pregnancy. Eligibility for ART was defined by the South African HIV treatment guidelines and policy as all HIV-infected pregnant women with a CD4 count of less than 350 cells/mm3 or those meeting WHO clinical stage 3 or 4.9 The campaign was titled, “Masibavikele,” the Zulu term for “Let’s protect them” referring to the HIV-infected mothers and their unborn children. The key components of the campaign are described in Table 1.
In April 2010, the South Africa Department of Health (SA DOH) introduced policies to expand nurse-initiated management of antiretroviral therapy and prioritization of pregnant HIV-infected women for initiation of ART.9 These occurred after the conclusion of the campaign and were not part of the study interventions.
The Masibavikele Campaign was launched in late September 2009 and completed in March 2010. Performance in the 2 districts was assessed in 3 time periods namely at baseline (6 months before September 2009), during the period of the intervention (6 months) and postintervention (6 months after March 2010).
Data Collection and Storage
Process data for the PMTCT cascade and ART initiations in Ugu district were captured using a campaign-specific clinic data collection tool (Fig. 1) and routinely submitted to the District Health Information System (DHIS), the standard public health information system used throughout South Africa to capture a range of primary care indicators. In Umzimyathi, the DHIS data capturing system was used similarly to collect and report data, but no campaign-specific clinic data collection tool was introduced as there was no concurrent campaign underway. Data on ART eligibility, referral, and initiation among pregnant women were captured and extracted from the DHIS for both Ugu and Umzinyathi to provide aggregate data for comparison. Data on ART eligibility and referral were captured at antenatal clinics, whereas data on ART initiation among pregnant women were collected at the ART initiation sites. De-identified data were stored in password-protected study databases and within the routine Provincial DHIS storage facility.
All data were initially entered in Microsoft Excel and crosschecked with the DHIS data store. These data were then imported into Stata version 11, 2011 (StatCorp, College Station, TX), for analysis. After routine distribution analysis, a 1-tailed, 2-sample equal variance t test was selected to compare mean ART referral and initiation rates in the preintervention, intervention, and postintervention periods in both the intervention and control districts, with an alpha of 0.05 considered significant.
Statistical process control (SPC) software, QI Macros (KnowWare, 2011), was used to develop SPC charts for analysis of month-to-month interpretation during the study period.19–21 SPC charts were used to compare referral and ART initiation rates before and after the start of the intervention in September 2009 in the intervention and control districts. We initially applied conventional run chart rules to detect signals in the continuously collected data and to determine if they were the likely cause of changes in referral or ART initiation rates.22 For both analyses presented below, we further applied control chart rules (Associates in Process Improvement) to define trends toward improvement.23 These include 1 or more values above or below the upper or lower control limits, 2 or more values above or below the 2σ control limits, 8 or move values above or below the median line, and a trend of 6 or more ascending or descending values.
Ethical approval was granted by both the UKZN (BF061/08) and Centers for Disease Control Institutional Review Boards. In addition, the KZN Department of Health and district management gave their consent for the study.
Referrals for ART
The average number of monthly referrals of pregnant women for ART in Ugu District increased from 78.7 (95% CI: 68.7 to 88.7) in the preintervention period to 140.2 (95% CI: 107.8 to 172.5) during the intervention period. In the postintervention sustainability period, referrals rose further to 188.2 (95% CI: 167.2 to 209.1). The mean ART referral rate in the intervention period was significantly different from the preintervention period (P = 0.002). The postintervention phase also showed a highly significant improvement from the preintervention phase (P < 0.001). Overall, this represented an increase in referrals among those eligible from 58.6% preintervention to 76.9% postintervention.
In the comparison district, the average number of monthly ART referrals among pregnant women remained essentially unchanged from 90.8 (95% CI: 74.2 to 107.5) from April to September 2009 to 98.5 (95% CI: 79.8 to 117.2) from April to September 2010. The mean ART referral rate in the period that corresponded to the “postintervention phase” was not significantly different from the period corresponding to the “preintervention phase” (P = 0.28).
The average number of pregnant women initiated on ART in Ugu each month increased from 20.7 (95% CI: 1.6 to 39.8) in the preintervention period to 67.2 (95% CI: 52.4 to 81.9) during the intervention period. In the postintervention period, the average number of pregnant women initiated on ART rose to 123.8 (95% CI: 107.9 to 139.8). The mean ART initiation rate in the preintervention period was significantly different from the intervention period (P = 0.002). The postintervention phase demonstrated highly significant improvement from both the preintervention and intervention phase (P < 0.001 in both comparisons). Overall, this represented an increase in ART initiations among those referred from 55.4% preintervention to 65.8% postintervention.
The increased rate for ART initiation occurred about 2 months after the campaign interventions were introduced and following increased numbers of HIV-infected women being referred to ART initiation sites. The increased rate of ART initiation was sustained in the period after February 2010 (Figs. 2A, B). These increases occurred about halfway into the 6-month campaign and ahead of the additional SA DOH initiatives that were introduced in April 2010. The fall-off in ART initiations in the last 2 months of the postintervention period was attributed to the backlog of women requiring ART having been addressed and, therefore, reaching a steady state of referrals and ART initiation.
In the control district, the average number of monthly ART initiations among pregnant HIV-infected women increased from 39.3 (95% CI: 28.4 to 50.3) between April and September 2009 to 54.5 (95% CI: 39.0 to 70.0) between April and September 2010. The mean ART initiation rate in the period that corresponded to the postintervention phase was not significantly different from the period corresponding to the preintervention phase (P = 0.07).
Figures 2A, B illustrate significant increases in the median referral and ART initiation rates in the intervention district using the API control chart rules described in the Methods section above. A similar analysis of data from the control district (Figs. 3A, B) shows no significant trend toward improvement in referrals or initiations on ART for pregnant women over the same time period.
A targeted brief ART campaign among health workers and led by the DHMT used QI health systems approaches to significantly improve access to ART for HIV-infected pregnant women across a large health district in KZN. During the same time period, rates of referral and ART initiations in a control district remained largely unchanged. While this as not a randomized study, data from the counterfactual district suggest that the changes in the intervention district were directly related to the campaign undertaken by the DHMT.
Campaign approaches in this study focused on generating momentum through strong public endorsement and advocacy from the district health managers and leaders, setting ambitious targets that reflected the need, gathering relevant data that tracked the steps of the ART referral pathway, feeding data back to front-line healthcare teams at monthly perinatal meetings, simplifying pre-ART preparation steps, and creating roving ART initiation teams. These gains have been sustained and, indeed, improved further since the end of the campaign. The latter part of the campaign was reinforced by major changes in SA DOH policy and guideline changes that were issued in April 2010 and that reinforced the value of, and opportunities for identifying and treating eligible HIV-infected pregnant women, and created greater public and provider awareness of the importance of this intervention.9
Campaigns that use QI methods in conjunction with public activation and alignment of stakeholders over a defined period of time have improved quality and access to health care services in other settings.24 The 100,000 Lives Campaign in the US, the Safer Healthcare Now Campaign in Canada, and the 1000 Lives Campaigns in Wales all involved a variety of infection prevention and patient safety interventions that are widely credited with saving many lives and reducing morbidity and mortality in hospitalized patients.25,26 Similar Campaigns have been conducted elsewhere in Europe and Australia.25 In South Africa, the Best Care...Always! Campaign is leading similar efforts to improve patient safety and infection prevention in 192 public and private hospitals.27
Quality improvement health systems strengthening methods have successfully improved access to HIV care and treatment services in other settings. In the inner-city of Johannesburg, QI efforts by the Reproductive Health Research Unit have surpassed National Strategic Plan targets for ART initiation within a shortened timeframe.28 In Umkhanyakude District in KZN, QI initiatives led by the Centre for Rural Health led to increased delivery of HIV treatment.13 In Cape Town, a district-led QI approach to increase access to ARVs for pregnant women resulted in significant reductions in mother-to-child transmission rates and improvements in ART access for all patients compared with other sites in the region.12 In Thailand and Haiti, using similar methods, HealthQual has shown improvement for pediatric and adult ARV delivery.29
The main limitation of this study is the lack of a randomized comparison district. A control district was identified that provided counterfactual outcomes. Although well matched for a number of demographic, economic, environmental, geographic, and health system management factors, the baseline starting performance for the 2 districts was, however, quite different—there was less room for improvement in the control district compared with the intervention district. Although the districts were well matched and their populations were demographically and economically similar, the design of the study did not allow capture of patient-level sociodemographic data to confirm these similarities between the actual clinic attendees between the 2 settings. In addition, the Ugu district health team had received training and support in QI methods for about 1.5 years from the UKZN team, which allowed them to rapidly and confidently lead the campaign and implement solutions that were identified in the course of the intervention, including redirecting resources such as doctors and nurses to mobile ART initiation teams. Finally, seasonal variation often results in fewer clinic visits around the holiday season in December and January because of shorter clinic work schedules and holiday closures. This effect is universal and clinic schedules in both intervention and control district would have been affected similarly.
There were multiple contextual factors that need to be considered when evaluating the campaign’s effectiveness. In the postintervention period, the Government of South Africa released new ART guidelines including prioritization of pregnant women for ART and empowering nurses to more comprehensively treat persons living with HIV.9,30 Another nongovernmental organization operating in Ugu District invested heavily in training nurses and doctors on these new guidelines. These changes may have contributed to the higher rates of ART initiations witnessed among pregnant women in the postintervention period in both Ugu and Umzinyathi. However, improvements in access to ART in Ugu district occurred before the new policies were enacted. Another key limitation of this study, as with all studies whose evaluation relies on data collected in public health information systems, is the potential inaccuracy and lack completeness of the DHIS.31 Although we were able to mitigate against this possibility in Ugu District, where we collected site specific data which was crosschecked with data collected in the DHIS, we were not able to conduct similar data verification in Umzinyathi. Data errors, therefore, remain a possibility in the control district.
The plausibility of the cause and effect relationship between the campaign and the increase in ART initiation rates is strengthened by the close association of the timing of the improvements in access with the onset and roll-out of the Campaign. The rapid timeline and predicted low incremental costs of the campaign (no personnel were added to the health system) suggest that this is a good model that could be replicated in similar resource-strained environments. The use of preexisting meetings and natural systems linkages increase the likelihood of sustainability. In addition, reliance and strengthening of the existing public health information system was a strong asset to the program.
Our results add to the literature on how QI health systems strengthening methods can improve performance of large-scale health programs. In this case, in addition to traditional QI methods that use data-driven front-line decision making to improve performance, a campaign strategy was shown to be effective in rapidly improving the performance of a key component of the PMTCT cascade. Although this district may have been optimally positioned for success, given its prior experience with QI methods and its highly functioning district management infrastructure and although the cost-effectiveness of the approach was not formally assessed, it does indicate the potential for the campaign approach to improve access to essential maternal, newborn, and child health interventions including lifelong ART.
The authors thank Jane Roessner for her review and comments on this manuscript and Teddy Svoronos for his research support.
1. World Health Organization. Antiretroviral Therapy for HIV Infection in Infants and Children: Towards Universal Access. Recommendations for a Public Health Approach. Geneva, IL: World Health Organization; 2010.
2. UNAIDS. Global Plan Towards the Elimination of New HIV Infections Among Children by 2015 and Keeping Their Mothers Alive. Geneva, IL: UNAIDS; 2011.
3. UNAIDS. UNAIDS World AIDS Day Report, 2011. How to Get to Zero: Faster, Smarter, Better. Geneva, IL: UNAIDS; 2011.
4. World Health Organization, UNICEF. Towards Universal Access: Scaling Up Priority HIV/AIDS Interventions in the Health Sector 2010 Progress Report. Geneva, Switzerland: World Health Organization; 2010.
5. United Nations. UN millennium declaration. In: Nations U, ed. A/RES/55/2, Section II. New York, NY; United Nations; 2000.
6. South African National Department of Health. The 2010 National Antenatal Sentinel HIV & Syphilis Prevalence Survey in South Africa. Pretoria, South Africa: National Department of Health; 2010.
7. Horwood C. KZN IMPACT Assessment. Durban: Centre for Rural Health, University of KZN; 2009.
8. National Committee on Confidential Enquires into Maternal Deaths (NCCEMD). Saving Mothers 2008-2010, 5th Report on the Confidential Enquires into Maternal Deaths in South Africa. Johannesburg, South Africa: South African Department of Health; 2012.
9. South African National Department of Health SANAC. Clinical Guidelines: PMTCT (Prevention of Mother-to-Child Transmission). Pretoria, South Africa: South African Department of Health; 2010.
10. Horwood C, Haskins L, Vermaak K, et al.. Prevention of mother to child transmission of HIV (PMTCT) programme in KwaZulu-Natal, South Africa: an evaluation of PMTCT implementation and integration into routine maternal, child and women's health services. Trop Med Int Health. 2010;9:992–999.
11. Kuhn L, Aldrovandi GM, Sinkala M, et al.. Potential impact of new WHO criteria for antiretroviral treatment for prevention of mother-to- child HIV transmission. AIDS. 2010;24:1374–1377.
12. Youngleson MS, Nkurunziza P, Jennings K, et al.. Improving a mother to child HIV transmission programme through health system redesign: quality improvement, protocol adjustment and resource addition. PLoS One. 2010;5:e13891.
13. Barker PM, McCannon CJ, Mehta N, et al.. Strategies for the scale-up of antiretroviral therapy in South Africa through health system optimization. J Infect Dis. 2007;196(suppl 3):S457–S463.
14. Berwick DM. Lessons from developing nations on improving health care. BMJ. 2004;328:1124–1129.
15. Statistics South Africa. General Household Survey 2007. Pretoria, South Africa: Statistics South Africa; 2008.
16. Day CBP, Haynes R, Smith J, et al., eds. District Health Barometer 2007/08. South Africa: Health Systems Trust; 2010.
17. Bisognano M, McCannon J, Botwinick L. A campaign to save 100,000 lives. Trustee. 2005;58:12–14.
18. Berwick DM, Calkins DR, McCannon CJ, et al.. The 100,000 lives campaign: setting a goal and a deadline for improving health care quality. JAMA. 2006;295:324–327.
19. Benneyan JC, Lloyd RC, Plsek PE. Statistical process control as a tool for research and healthcare improvement. Qual Saf Health Care. 2003;12:458–464.
20. Finison LJ, Finison KS. Applying control charts to quality improvement. J Healthc Qual. 1996;18:32–41.
21. Shewhart WA, Deming WE. Statistical Method From the Viewpoint of Quality Control. New York, NY: Dover; 1986.
22. Perla RJ, Provost LP, Murray SK. The run chart: a simple analytical tool for learning from variation in healthcare processes. BMJ Qual Saf. 2011;20:46–51.
23. Provost LP, Murray SK. The Health Care Data Guide: Learning from Data for Improvement. San Francisco, CA:Jossey-Bass; 2011.
24. Bradley EH, Nembhard IM, Yuan CT, et al.. What is the experience of national quality campaigns? Views from the field. Health Serv Res. 2010;45():1651–1669.
25. McCannon CJ, Hackbarth AD, Griffin FA. Miles to go: an introduction to the 5 Million Lives Campaign. Jt Comm J Qual Patient Saf. 2007;33:477–484.
26. McCannon CJ, Schall MW, Calkins DR, et al.. Saving 100,000 lives in US hospitals. BMJ. 2006;332:1328–1330.
28. Barker PM, Venter F. Setting district-based annual targets for HAART and PMTCT—a first step in planning effective intervention for the HIV/AIDS epidemic. S Afr Med J. 2007;97:916–917.
29. Lolekha R, Chunwimaleung S, Hansudewechakul R, et al.. Pediatric HIVQUAL-T: measuring and improving the quality of pediatric HIV care in Thailand, 2005-2007. Jt Comm J Qual Patient Saf. 2010;36:541–551.
30. Colvin CJ, Fairall L, Lewin S, et al.. Expanding access to ART in South Africa: the role of nurse-initiated treatment. S Afr Med J. 2010;100:210–212.
31. Mate KS, Bennett B, Mphatswe W, et al.. Challenges for routine health system data management in a large public programme to prevent mother-to-child HIV transmission in South Africa. PLoS One. 2009;4:e5483.