Department of Health
Private Bag X828
Pretoria 0001
Republic of South Africa
Compiled by : Oumiki Khumisi (NDoH), Yogan Pillay, Jon Rohde (EQUITY), Calle Hedberg, Norah Stoops (HISP)
3.1 Training Information and Software Use
3.2 Emerging Cadre of Health Professionals
4.1 Data Quality
4.2 Utilisation Rates
4.3 Antenatal Care
4.4 Growth Monitoring
4.5 Diarrhoea
4.6 Immunisation
4.7 Cases Treated as Sexually Transmitted Infections
4.8 Male Urethral Discharge Incidence
4.9 Condom Utilisation Rate
One key element in Government health policy has been to establish Health Districts, provide integrated Primary Health Care to all citizens, and shift the emphasis in public health care from curative to preventive services. Decentralised Primary Health Care is also inextricably linked to Local Government and the Demarcation process.
To ensure that the above is achieved there was a need to gather, analyse and use information to be able to evaluate how good or bad the Department of Health is performing.
The information on the PHC was previously collected through the use of a PHC form based on data compiled from registers.
The current report is based on the District Health Information System Software developed by the Health Information System Programme based at the University of the Western Cape.
One vital component in this transformation of the South African health care system was and is to develop and maintain efficient and effective DHS Information Systems.
The methods, processes and software tools developed by the Health Information Systems Programme (HISP) was in February 1998 adopted by a National Health District Management indaba held in Port Elizabeth, and later endorsed by the National Health Information Systems of South Africa (NHISSA). HISP, a collaborative R&D programme between four universities and the national/provincial health administrations in South Africa, Mozambique, and India, started as a pilot project in four health districts around Cape Town in 1994-97.
The District Health Information System (DHIS) Software developed by HISP was adopted by the Western Cape Metropole from July 1998, by the Eastern Cape from November 1998. A decision was then taken that all provinces adopt the DHIS Software in early 1999. Gauteng, Kwa-Zulu Natal and Mpumalanga came on-board as the last during 2000. The costs of HISP research and development activities, as well as the national rollout process, have during the last 5 years been around R 10 million.
One of the top priorities of the Department of Health is the accelerated delivery of Primary Health Care (PHC) through the District Health System, which ties in with the broader municipality reform and decentralisation of service delivery. Another priority is the expansion of the district-based health information system to monitor and evaluate the impact of PHC delivery on the status of the health of the people of South Africa. These priorities are reflected in the Health Sector Strategic Framework, 1999-2004, also known as the Ten Point Plan.
The National and Provincial Departments of Health have during the last 3 years worked closely with the Health Information Systems Programme (HISP) and several other partners (the EQUITY project, the Health Systems Trust, CHESS) to implement a District Health Management Information System in all health districts in South Africa. Even if the bulk of work done has been institutional and human resource development 60-70% of total effort is training of managers and health workers it has been tightly linked to rolling out the District Health Information System (DHIS) Software from HISP as a core module for capturing, validating, analysing, and presenting routine data/information.
The aim is to report on the rollout process and to briefly present key national/provincial Primary Health Care indicators linked to the National Essential Data Set (EDS) for PHC, based on national data coverage from nearly all PHC facilities in the country (95%) for the financial year 2000-2001. Further aims are to ask national managers and provincial heads of health to
- Identify immediate priority areas for more in-depth investigations of available data/information in order to provide a basis for interventions to accelerate delivery (ref. Recent similar decision after presentation of EPI indicators);
- Specify the format, scope, and frequency of future PHC reporting to the Provincial Health Restructuring Committee (PHRC) and Minister and Members of Executive Council (MINMEC).
Note that the Department in early 2000 decided to use the DHIS software also for capturing, analysing, and distributing routine monthly data/information for all public and private hospitals in the country. Data input coverage here has also greatly improved currently at 92% but reporting on the hospital system is outside the scope of this report. Other current uses of the DHIS software e.g. for Emergency Medical Services and the Prevention of Mother To Child Transmission research sites are also not covered here.
Nearly all stakeholders in the public health sector during the last three years have accepted and implemented the first core, crucial principle for a District Health Management System: Development of an integrated but flexible routine information system based on Essential Data Sets at all administrative levels. The term Essential Data Set (previously often called Minimum Data Set) means acceptance of very stringent criteria for including data elements to be collected:
- They must be required for calculation of one or more indicators linked to defined targets
- They must be need to know, i.e. crucial for decision-making and service delivery
- They must be useful for managers at all levels, most especially at the level of data collection, not only the provincial and national levels as was common in the past.
The term integrated should be understood as representing both a major break with our legacy of a fragmented multitude of data sets and information systems, often linked to vertical programmes or racially based authorities, and the fact that the Essential Data Set at any administrative level is included in the Essential Data Sets at all subordinate levels. The term flexible means that each administrative level, while providing the essential data needed for the level above, is free to add data elements and indicators considered essential for their local environment. And it means that Essential Data Sets should be reviewed and revised regularly to ensure that every single data element collected actually is being used for monitoring and/or decision-making.
Several provinces started defining their first Essential Data Sets (EDS) in 1997 and 1998. The results from a survey of existing data sets, combined with recommendations from a National Health Information System for South Africa (NHIS/SA) Committee, resulted in a national Essential Data Set for PHC being adopted by the NHIS/SA in April 1999. This EDS for Primary Health Care is still in use and contains 20 compulsory and 18 optional items.
All nine provinces have during the last three years developed their own Essential Data Sets for both Primary Health Care and hospitals. Almost all provinces are also currently using or are about to implement version 2 of their EDS, having revised the data set after gaining experience on what is useful or not or after getting new stakeholders (e.g. Local Authorities or vertical Programme Managers) on-board. Some revisions have been quite radical for instance, the Western Cape replaced 37 out of 47 data elements in their first revision in 1998 and Mpumalanga cut the number of elements from 380 to 120 in late 2001 (some of the 120 are completely new). Only 5-6 data elements were changed during the second revision of the Western Cape EDS in 2000, clearly indicating steady progress towards a 100% useful data set. Similar positive trends are seen in other provinces.
2.1 Data Flows
Facilities collect data daily, usually through tally sheets, tick-registers and similar paper-based tools. A few PHC facilities in wealthier metropolitan areas have computerised electronic patient record systems, but most of these are outdated and/or ill suited for purposes beyond simple administrative tasks like printing labels etc so paper-based systems are used in parallel to provide usable data for management.
The data from tally sheets and tick-registers are summarised monthly, together with data from registers e.g. malnutrition and delivery, or data derived from stock control (e.g. condoms in stock are counted once a month). The summary sheets are forwarded in 90% of all districts to the district office for capture into the DHIS software, since few PHC facilities have computers. It should be noted, though, that a growing number of districts are providing each fixed PHC facility with a computer, basic software like Windows and Office, the DHIS software (which is free), and if possible a dial-up or direct connection to the government network with email etc. The facility then captures their own routine data, validates it, prints out reports and graphs for their own use, and submits the data electronically to the district office. District reports indicate that this is not only a very cost-effective and low-risk way of increasing computerisation and computer literacy among health workers it also radically increases the staff commitment towards data quality and data use as a tool to improve and monitor service delivery.
After routine data has been captured or collated at district level, the sub-set that belongs to the provincial EDS will be submitted electronically, either directly or via a regional office. Provinces are expected to submit National EDS data items to the National DoH on a monthly basis, but there are still delays and interruptions in this data flow for a range of reasons. We expect these problems to be completely ironed out before the end of 2002. Some of the interruptions in the data flow occur at or below the provincial level, often linked to re-organisation or staff turnover (many districts and even provinces have very few or no staff dedicated to health information). Nevertheless, the table below shows that we have reached 95% national data coverage, which is as a major milestone.
As can be seen from the table 1 below, the national average Data Input Coverage for the 2000-01 year is approximately 95% and all provinces are at 90% or above. It should also be noted that the data sets collected and used in this report include data from all the (old) 173 health districts. Some districts are clearly lagging behind others in terms of quality and timeliness of submission, but data is at least flowing from all of them. Provinces that started the HISP/DHIS Software rollout in 2000 seem to be catching up with provinces that started early (e.g. Western and Eastern Cape), indicating that these provinces are benefiting from the experience gained elsewhere.
Table 1. Data Input Coverage, Essential Data Sets Data Elements, and Extrapolation Factors for 1998 2001
Calendar years
Years
Province Essential Data 1998 1999 2000-2001 Extrapolation factors Population 2000 Elements Data received Data received EC
76
98%
99%
100%
1
6,829,517
FS
79
No data
No data
93%
1.08
2,777,182
GP
1023 (317*)
No data
No data
96%
1.04
7,831,627
KZN
79
No data
Pilot areas
90%
1.11
8,928,063
MPU
380 (120**)
50%
60%
91%
1.10
3,027,111
NC
78
40%
70%
98%
1.02
871,193
NP
60
No data
No data
93%
1.08
5,505,198
NW
91
Pilot areas
Pilot areas
98%
1.02
3,550,131
WC
97
99% (Jul->)
99%
99%
1.01
4,185,837
RSA
49
21%
33%
95%
43,505,859
* Gauteng has amalgamated many data sets - no facility collects over 317 elements
** Mpumalanga has recently reduced / improved their data set radically.Data from private health providers (e.g. HASA, NGOs) is also on the increase, even if most provinces are concentrating on improving timeliness and quality of public sector PHC data before expanding to the multitude of private health providers.
It must be stressed that whereas all districts and provinces are striving to reach 100% data input coverage, additional data beyond 95% will have limited impact on the value of most indicators at provincial or national levels if appropriate extrapolation factors are used. The factors used in several indicator tables below are specified in table 1 above. The column `Population 2000` represents mid-year 2000 figures, estimated based on Census 96 and the official growth model published by StatsSA in July 2001.
Work on developing DHIS Software commenced in the Western Cape Metropole (now City of Cape Town) in early 1998 by the HISP. This offered a viable district health information system that was welcomed by provinces. Western Cape decided to roll it out to the whole province from July 1998, and the Eastern Cape commenced with implementation from November 1998.
Figure 1: The Routine Monthly Data Module in the DHIS.
A decision to rollout the DHIS Software nationally was taken in April 1999 by the NHIS/SA Committee and ratified by the PHRC. Since the PHRC decision, the DHIS Software has been implemented in all nine provinces. The direct costs of HISP research and development activities, as well as the national rollout process, from 1996-2001 cost around R10 million. Key funding agencies have been NORAD and the Norwegian University Council, through the five universities involved with HISP, and USAID through the MSH/EQUITY project.
Funding for continuing the national (and international) expansion and consolidation of HISP/DHIS Software is partially secured, with the Norwegian University Council allocating approximately R 10 mill to HISP for 2002-2006 with tentatively 20% used for activities in SA and the National DoH awarding the EU-funded tender of around R 3.6 mill to support DHIS establishment in 13 rural District Municipalities during the next 16-21 months. USAID/EQUITY has also indicated they will continue to support the rollout process during 2002-2003, and several provinces have expressed willingness to cover a larger share of the rollout and training costs through Service Level Agreements with HISP or their partners.
Technically, the core modules in the software are written in Visual Basic for Applications (VBA), the programming language used by Microsoft Access. Data analysis and charts are in most cases done through an Access-based `Report Generator` in addition to pivot tables and chart capabilities in Microsoft Excel. The free desktop GIS viewer ArcExplorer is used to design and display/print health maps. The DHIS Software runs on all standard computers that can run Windows 95/98/NT/2000/XP and Microsoft Office, but hardware requirements related to memory and hard disk space increase with increased size of data sets. The figure below shows the core Access/Excel/Web modules currently in use in SA.
Figure 2: DHIS Software Application Structure (version 1.2, Service Release 2)
For health districts, HISP recommends a standard hardware and software package, including an A3 colour printer for wall graphs and small maps, all of which currently cost around R 15,000. Several provinces have purchased such or similar packages for all districts. Other provinces still relying on old or outdated computers, like KwaZulu-Natal, are actively looking for funds to provide each district with at least one such computer system appropriate for routine data analysis and management.
The DHIS Software itself is free and released into the public domain as so called "Open Source" software, where the software is distributed for free and complete with all source code. This approach can be seen as an extension of basic academic principles for international sharing of knowledge. Any user is free to utilise, reproduce, or adapt the software as they see fit as long as it is not abused for commercial gain (e.g. through selling the software to unsuspecting customers).
3.1 Training in Information and Software Use
The success in the implementation of the software can largely be attributed to on-going training and support offered by the trainers from HISP and their partners, who conducted over a thousand days of training during 2001. Much of the training has been tailored to the needs of the specific provinces using their own data. Training has been aimed at all levels of health workers, from data and administrative clerks, nurses, Information Officers, various level of supervisors, district managers, programme managers and staff from district, regional, provincial and national level. Key technical people from national DoH, one from each province, and CHESS also started the first round of training in DHIS software design during 2001 the objective here is both to have technically skilled trouble-shooters readily available within the departments and to develop capacity for DHIS software design and modifications, thus increasing future sustainability. All training is constantly being revised and improved through feedback from provinces and participants.
The UWC School of Public Health offers 7-8 different one-week courses during the Summer and Winter Schools to augment the training conducted in provinces. Some of these are directly aimed at DHIS Software users, like the DHIS Intermediate Course, the DHIS Advanced Course, the DHIS Use for Managers, and the course in Geographic Information System (GIS) for health. Others are focusing on training PHC and Hospital managers in using information (goals, targets, indicators) for management and planning.
Major progress has also been made during 2001 in writing up manuals, curricula and similar course materials. Even if most materials developed are aimed at practical training as part of the national rollout, they will in the future also form components of formal degree-yielding education paths within the fields of health information systems and/or medical informatics. UWC and University of Natal are among those universities working towards offering such studies.
3.2 Emerging Cadre of Health Information Professionals
The Eastern Cape was the first province to formally appoint District Information Officers as a key member of each health district management team. Other provinces are following suit, usually with a mixture of temporary and permanent positions related to health information work. Several provinces have also expanded or re-organised their provincial health information teams, which increasingly are requested to support development and capacity building at district or facility level instead of being directly responsible for data capture and analysis as in the past. Considerable work has been done to define new scopes of work through new or revised job descriptions as well as developing the human resources through training and practice.
Primary capture of data has almost disappeared at the provincial level and it is increasingly disappearing at the regional level. Even if the bulk of data currently is captured at the district level, there can be no doubt about where we are headed: Towards data capture, validation and initial analysis and use at facility level, where it belongs. Computerisation at facility level is also one step towards the long-term goal of linked, patient-based data sets that is used in daily management of patients as well as in generating aggregated data for management and planning.
Three points should be kept in mind when considering the current National Essential Data Set and the indicators derived from it in this report:
- The National EDS for Primary Health Care is still the first version fully implemented in the country it will be revised during 2002, allowing for additional key indicators/elements to be included (many of these are already captured by many provinces).
- The data is sometimes incomplete and often of unreliable quality. Many statistical and numerical tools are available for temporary adjustments (e.g. using extrapolation to adjust for missing data), and the resulting information is often more than clear enough to guide further investigations, interventions or management decisions.
- The real value of the DHIS Software lies in its ability to `drill down` from provincial averages to District Municipalities, Local Municipalities or even individual facilities. In almost all cases, drilling down to the next level will reveal highly relevant differences and disparities vital for planning effective interventions and optimise impact with scarce resources.
Point 2 and 3 are crucial when considering the merits of a district based routine information systems with its unavoidable shortcomings in terms of quality and comprehensiveness compared to data obtained from household surveys like the Demography and Health Survey. Surveys might provide more accurate estimates for average immunisation coverage, but only routine information systems can be used to accurately target interventions in health facilities, to monitor the performance of facilities or districts/municipalities, to provide rapid and meaningful feedback to staff and other stakeholders, or to foster a sustainable local information culture and local staff `ownership` towards data and information.
4.1 Data Quality
Improving data quality is a complex, time-consuming process. The DHIS Software has a range of built-in tools (see figure 6) for data validation, identification of missing data, identification of outliers, interpolation through linear regression and so forth. These tools play a major role in improving data quality for data sets already collected and captured, and most of the districts and provinces will continue to partially rely on these tools for improving data quality.
Figure 3: DHIS graph popup for Louis Trichardt Clinic for Jul 00 Jul 01
In the longer term, though, high data quality will depend on two crucial factors:
- That staff and managers at facility and district levels analyse and use their own data themselves for local management on a regular basis;
- That staff and managers at facility and district levels know that their data is used on a regular basis by national and provincial managers, by community and media organisations, by researchers, or by the public in general. Regular feedback from higher levels is a major motivational factor for many local staff.
Current data sets, even with significant quality problems, do however reflect to a significant extent the reality on the ground and can thus act as a guide to assessing the health status of the communities being served.
We have seen major improvement in data quality in every province since the introduction of the DHIS Software, partly due to its automatic tools for monitoring and addressing data errors and partly due to systematic training and feedback. The data quality tends to improve faster in areas where health trained staff (nurses, environmental officers, etc) are responsible for the data processing and handling. Data entry clerks generally do not perceive good quality data as their responsibility. The appointment of full time dedicated District Information Officers, preferably with a strong health background; make a substantial difference in the whole development and implementation of a district health information system.
With several provinces starting province-wide EDS data collection from April 2000, the year 2000-2001 (April 2000 to March 2001) was a natural choice for this report. It is important to note the following caveats:
Firstly, experience and analysis of the various provincial data sets show a clear negative correlation between number of essential data elements collected and data quality. The more items collected, the worse the quality. Indications are that 40-70 data elements are optimal in the initial stages; with a gradual increase possible as submissions and data quality reach sufficient levels.
Secondly, substantially more data items are available from several provinces at lower levels making more information available at the provincial or district level. Furthermore, the lower the level, the more differences become evident and disparities more clear. As data is aggregated at higher levels, differences become smoothed over and the averages converge.
Thirdly, missing data reports throughout the year result in estimates being used for the missing months in order to make figures comparable between provinces for the year. A number of data records in each province have been filled in or corrected using a range of interpolation tools available in the DHIS Software. This makes use of regression equations to "correct" outlying or missing data. As an example, around 3.5% of the complete Eastern Cape data set has been filled in using interpolation tools (trend values based on linear regression). The percentage will vary from province to province, but it is statistically improbable that the errors introduced through this method are more than 0.5% - and it is highly probable that leaving these gaps in the data would reduce indicator accuracy and make trend analysis more difficult in the longer term. This is done after all other avenues for obtaining the data or the correct values have been explored.
Figure 4: Population pyramid for South Africa for 1996
Source: Statistics South Africa, 1996
Fourthly, population denominators were estimated using Census 1996 data and the latest version of the StatsSA growth model. The population denominators used are mid-year estimates for 2000. These have been calculated using the raw Census 1996 data combined with the exponential growth factors and adjustment factors for the unknown age group published by StatsSA on 2 July 2001. It should be noted that several experts on demography have suggested that the Census 1996 might have under-estimated the number of children under-5 years. The under 5 years group is about 250,000 less than the 5-9 and 10-14 years groups as can be seen in Figure 7. Experts doubt as to whether fertility levels could have dropped that much.
4.2 Utilisation Rates
The Utilisation Rate indicator is the rate at which services are utilised by the target population, represented as the average number of PHC visits per person per year in the target population. It is used to determine overall utilisation patterns; it is particularly relevant for the move towards equity in the health sector. It has been established that utilisation depends on a number of factors such as accessibility, acceptably and appropriateness of services, as well as the legacy of apartheid with its gross inequity in resources and personnel.
Table 2. Utilisation Rate for provinces for 2000 2001
Province
PHC headcount
Total Pop
Visits/year
Data Elements headcount
Eastern Cape
14,339,786
6,829,517
2.1
2
Free State
5,069,882
2,777,182
1.8
2
Gauteng
10,649,014
7,831,627
1.4
17
Kwa-Zulu Natal
15,315,661
8,928,063
1.7
2
Mpumalanga
3,953,523
3,027,111
1.3
8
Northern Cape
1,931,154
871,193
2.2
3
Northern Province
10,984,360
5,505,198
2.0
4
North West
8,237,237
3,550,131
2.3
3
Western Cape
11,426,477
4,185,837
2.7
2
National PHC models generally calculate a need for 3-3.5 visits per capita per year. It can be noted from Table 2 that PHC utilisation is between 1.3 and 2.7 visits per capita per year, roughly half the expected value for the estimated full range of PHC services.
A more interesting picture, from an equity perspective, will always be the differences within each province. As can be observed, the rates for Gauteng and Mpumalanga are very low. This is most likely a result of the confusion around the definition of headcount and procedures for calculating them, as they use multiple categories to determine headcount (age break-down, gender break-down, clinic/home visits break-down, totals). Recent figures from an HR survey in Gauteng confirm the inaccuracy, and the discrepancy between child headcount and children weighed in Mpumalanga shows the same. This should be resolved with simpler headcount definitions as used in the other provinces. The low utilisation rates in Kwa-Zulu Natal (KZN) and the Free State are, on the other hand, not so easily explained and further investigations are needed.
4.3 Antenatal Care
"Antenatal Coverage" indicates how accessible ANC services are to pregnant women in general. This indicator shows the percentage of pregnant women who have a first antenatal visit (booking).
Table 3 ANC coverage of pregnant women by province
Pregnancies
1st ANC visits
(Pop <1 year)
Rate
Eastern Cape
111,235
150,368
74.0%
Free State
51,055
51,631
98.9%
Gauteng
124,327
147,363
84.4%
Kwa-Zulu Natal
143,276
196,745
72.8%
Mpumalanga
63,058
67,856
92.9%
Northern Cape
14878
18,565
80.1%
Northern Province
93,995
135,474
69.4%
North West
84,855
80,135
105.9%
Western Cape
75,547
82,223
91.9%
The numerator data (antenatal first visits) used to calculate antenatal coverage below has been adjusted using the extrapolation factors outlined in table 1. The estimated number of children under 1 year is used as a proxy for the number of pregnant women (denominator).
Figure 5: Antenatal Coverage Chart 2000-2001 (extrapolated)
Several of these provincial average ANC coverage rates are clearly higher than we can reasonably expect, since the antenatal clients using private care in most cases are not included. Some antenatal care provided by hospitals is also not included. Significant and systematic errors on the high side in the numerator data is theoretically possible but unlikely, so the high values are most likely caused by too low estimates for the number of pregnant women (i.e. under-estimation of children under 1 year in the Census 96 data).
The "Antenatal Care per Antenatal Client" indicator is a measure of the quality of maternal care. The national target is that all pregnant women should have at least 3 antenatal visits per pregnancy. As table 4 below shows, all provinces have average numbers above 3, though substantial numbers may not receive this minimum. Special annual surveys in Eastern Cape show that even with an average of 3.4 visits, only some 80% of ANC clients have 3 or more visits.
Table 4: Antenatal visits per antenatal client 2000-2001
Province
ANC all visits
1st ANC visits
Rate
Eastern Cape
434,620
128,689
3.4
Free State
225,498
54,883
4.1
Gauteng
455,530
129,507
3.5
Kwa-Zulu Natal
757,505
180,046
4.2
Mpumalanga
223,972
67,965
3.3
Northern Cape
63,033
15,427
4.1
Northern Province
489,043
140,136
3.5
North West
334,502
86,608
3.9
Western Cape
353,354
83,536
4.2
However, such averages hide important factors such as intra-provincial differences as is evident in the map below. Firstly, risk pregnancies are tentatively 15-25% of the total and should normally have many more follow-up visits than the minimum three. Secondly, provincial averages hide significant sub-province differences such as depicted in figure 8. It comes as no surprise that the developed urban areas, such as the Nelson Mandela Metro, on average have more than twice the number of antenatal visits per antenatal client as compared to the rural areas in Ukhahlamba or Oliver Tambo District Municipalities. Nevertheless, the DHIS Software can now be used to monitor the situation at each facility by Local Municipality, or by District Municipality.
Figure 6: Map of the Eastern Cape showing average ANC visits per ANC client for each of the seven District Municipalities
It can be confidently concluded that Antenatal coverage and ANC visits per client on average are close to national/provincial targets, but local variations and possible census under-estimation require attention and action.
4.4 Growth Monitoring
The provinces collect different data elements, sometimes with different definitions to assess nutrition. For instance, the Western Cape defines growth faltering ("Not gaining weight under 5 years new") as failure to gain weight through two consecutive weighings, whereas the other provinces count the child the first time weight gain is lacking. Non-weight gain in a given month is a far more sensitive index of deficient child growth than waiting for malnutrition to be diagnosed. All provinces are collecting numbers of severe malnutrition, although it appears that some facilities report a given malnourished child each time she appears, rather than once i.e., the first time the child is newly diagnosed. Obviously this overstates the problem and is the likely reason that the numbers are so high in Gauteng and KwaZulu-Natal.
The weighing of children is also problematic as a number of facilities do not conduct this assessment on all children under-five years of age. There are numerous cases where even severe malnutrition is diagnosed and reported without weight being recorded! The differences as seen in the table below, with regard to both child utilisation rates and weighing coverage, are so large that they should be regarded with suspicion pending further investigation.
Table 4: Growth monitoring and malnutrition for children under 5 years 2000-2001
Province
Headcount
Children
Not gaining
Underweight
New severe
Population
under 5 y
weighed
weight
for age
malnutrition
under 5 y
Eastern Cape
2,490,089
1,957,234
57,878
Not coll.
9,674
828,989
Free State
739,227
427,427
20,634
7,146
1,928
267,754
Gauteng
2,013,241
1,180,704
Not coll.
Not coll.
28,432
704,719
Kwa-Zulu Natal
3,395,894
1,547,608
Not coll.
Not coll.
41,730
1,037,806
Mpumalanga
401,839
570,811
12,645
12,945
152
358,718
Northern Cape
382,477
22,200
696
4,303
853
93,328
Northern Province
2,724,301
1,432,609
39,477
Not coll.
6,702
732,498
North West
2,152,932
Not coll.
10,940
Not coll.
12,832
400,667
Western Cape
2,659,182
595,770
10,215
23,455
3,303
406,286
Notes: Underweight for age means less than 3rd centile but above 60% expected weight for age (EWA). Severe malnutrition is under 60% EWA, kwashiorkor, marasmus and other cases of clinical malnutrition.
Western Cape collected data on `children weighed` only from Oct-2000 (half the year). The figures in Mpumalanga showing more children weighed than headcount of children are most likely a result of widespread errors due to too many and confusing headcount variables (8) collected in that province.
Nutrition measures vary, and routine weighing of children under-5 years is not yet regular. This system will allow early detection of growth faltering and intervention before malnutrition appears. Reported cases of severe malnutrition number nearly 100,000, alarmingly high, but over reporting in at least 2 provinces (Gauteng and KZN) seems likely.
4.5 Diarrhoea
Diarrhoea incidence reflects lack of access to adequate water and sanitation. While only a small fraction of estimated cases seek care at PHC facilities, the incidence of diarrhoea among children under five years is normally regarded as a good indicator of socio-economic development. The relatively low incidence rates seen for the Free State, Gauteng, and the Western Cape in the graph below are therefore to be expected.
The graph depicts two sources of data: the DHIS for 2000-2001, and the South African Demography and Health Survey from 1998. These two incidence rates are not directly comparable. The DHIS reflects cases seen and diagnosed at PHC facilities in 2000-2001, adjusted with the extrapolation factors in table 1. The SADHS 1998 figures were the number of children with diarrhoea per 1,000 children under 5 years during the last two weeks before the survey, multiplied by a factor of 2.17 to get a monthly equivalent. Nevertheless, a broad correlation between the two data sources can be observed. These data indicate that only around half of the children having diarrhoea seek treatment in the PHC system
Figure 7: Diarrhoeal Incidence Rate Chart, DHIS 2000-2001 and SADHS-98
The high rate for KZN is probably related to the cholera outbreak, but the relatively high rates seen in Northern Province and the North West might be indicative of poverty and increased vulnerability to communicable diseases.
It is, as usual, highly revealing to drill down to another level. For instance, the main district municipalities in the North West have incidence rates varying from 185 per 1,000 in Southern District Municipality to 285 per 1,000 in Bojanala Platinum District Municipality. Drilling down further to e.g. Category B Municipality is likely to reveal even larger disparities, thus allowing resources to be target to areas of greatest need and/or the greatest potential for rapid improvement.
4.6 Immunisation
The national target for EPI is to fully immunise 90% of all children before their first birthday. As can be seen from the table below, the DHIS data provides a mixed picture compared to the Demographic and Health Survey in 1998. The first row for each province is the number reported for the year, the second row calculates coverage rates for some key `indicator doses`, and the third row the coverage rates resulting from SADHS 1998. Even counting in several potential sources of error outlined below, the data indicates improved coverage of primary immunisation when considering that vaccinations done by the private sector have been partially captured only in the Western Cape.
Table 5: Immunisation coverages for 2000-2001 (extrapolated numerators)
OPV 1st
OPV 3rd
Measles 1st
Immun fully <1
Pop <1 yr
Eastern Cape
133,317
115,247
103,537
93,995
150,368
Coverage
88.7%
76.6%
68.9%
62.5%
SADHS-98
86.9%
61.3%
75.4%
52.6%
Free State
51,049
47,642
42,842
41,356
51,631
Coverage
98.9%
92.3%
83.0%
80.1%
SADHS-98
96.9%
72.6%
80.8%
67.8%
Gauteng
142,611
133,956
124,322
111,118
147,363
Coverage
96.8%
90.9%
84.4%
75.4%
SADHS-98
92.8%
80.8%
84.4%
72.4%
Kwa-Zulu Natal
211,033
192,432
175,687
162,564
196,745
Coverage
107.3%
97.8%
89.3%
82.6%
SADHS-98
87.7%
59.6%
82.5%
49.5%
Mpumalanga
57,398
53,855
45,585
43,776
67,856
Coverage
84.6%
79.4%
67.2%
64.5%
SADHS-98
90.3%
75.9%
83.7%
67.2%
Northern Cape
17,643
16,032
14,802
14,096
18,565
Coverage
95.0%
86.4%
79.7%
75.9%
SADHS-98
92.7%
85.5%
90.5%
80.8%
Northern Province
128,247
113,074
101,678
99,059
135,474
Coverage
94.7%
83.5%
75.1%
73.1%
SADHS-98
94.5%
83.6%
80.4%
74.9%
North West*
87,851
79,958
68,640
68,250
80,135
Coverage
109.6%
99.8%
85.7%
85.2%
SADHS-98
91.2%
70.8%
87.0%
60.6%
Western Cape
73,773
68,693
73,549
70,641
82,223
Coverage
89.7%
83.5%
89.5%
85.9%
SADHS-98
91.7%
72.5%
83.7%
64.2%
--------
--------
--------
--------
--------
--------
South Africa
902,923
820,889
750,642
704,855
930,360
Coverage
97.1%
88.2%
80.7%
75.8%
SADHS-98
91.0%
72.1%
82.2%
63.4%
It should be noted that the "Immunised fully under 1 year new" figures for some provinces were far out, yielding coverage rates of 150-300%. These have been conservatively reset to the values for Measles 1st dose at 9 months (usually the last dose), but actual figures are most likely lower depending on the occurrence of vaccine coverage shortfall for the other antigens. Immunised fully rates for provinces like KZN and NP should be treated with care they are definitely on the optimistic side. The gross errors for this data element have occurred in nearly all provinces nurses often count all children whose vaccination card shows Primary Course completed, instead of counting them only once when they receive the last dose and have proof of all other doses. Experience from the Eastern Cape indicates that these gross errors largely disappear after 6-18 months of training and feedback.
Finally, it should be noted that reporting of `BCG at birth` vaccinations currently is unsatisfactory, and the numbers are much lower than OPV 1st dose. The main reason for this is that such vaccinations are mostly given in hospitals, and many hospitals have not reported PHC data satisfactorily. With both PHC and hospital routine data captured into the DHIS, this is being resolved.
4.7 Cases treated as Sexually Transmissible Infections
High rates of Sexually Transmissible Infections are fuelling the HIV/AIDS pandemic. About half of the cases treated as Sexually Transmitted Infections, using a Syndromic Case Management of STIs approach, are not actually STIs but other genito-urinary tract infections, particularly in women. The rear bars in the graph below show the HIV-positive rates from the last published antenatal HIV survey (October 2000). Whereas it might be expected that KZN is high, it is surprising to see the difference between Mpumalanga relatively low Cases Treated as STI Rate but high HIV rate and Northern Province and the North West Province, with relatively high Cases treated as STI rate compared to the HIV rates.
Figure 8: Cases treated as STI for DHIS 2000-01 and HIV-ANC 2000
A number of reports, such as the one contained in the evaluation on the comprehensive management in the private sector and the workplace by the Centre for Health Policy, from many urban areas of South Africa indicate that a large percentage of patients suffering from STI-type infections prefer treatment in the private sector, probably because its perceived as having better patient confidentiality. Nonetheless, the STI treatment rate per 1000 population is high throughout the country.
4.8 Male Urethral Discharge Incidence
The Male urethral discharge incidence rate is often used as a more direct measure for Sexually Transmissible Infections, since such cases are invariably STIs. The figures from SADHS 1998 have been included for reference purposes in the table below, as well as the ratio between Male Urethral Discharge incidence and Case Treated as STIs.
Note again that the DHIS and SADHS figures are not directly comparable, since the DHIS data is cases treated at PHC facilities and the SADHS data is men reporting painful urination / discharge in the three months preceding the survey (multiplied by 4 to give annualised figures). Men with repeated infections would for instance be counted only once in the SADHS survey but each time they appear for treatment at a health facility. Some of the SADHS rates are also difficult to accept is it reasonable, for instance, that the rate of Male urethral discharge (MUD) should be nearly four times higher in Mpumalanga than in Gauteng given their proximity? The DHIS, on the other hand, indicates nearly similar levels of MUDs in Mpumalanga and Gauteng if the larger private sector in Gauteng is taken into account.
Table 6: Male Urethral Discharge Incidence Rate 2000-01
Province
MUD
Male pop >=15
DHIS Rate
SADHS
MUD/STI
Eastern Cape
82,691
1,800,743
45.9
552
31.5%
Free State
32,605
932,524
35.0
532
29.5%
Gauteng
85,830
2,995,363
28.7
192
24.7%
Kwa-Zulu Natal
117,576
2,588,763
45.4
664
21.3%
Mpumalanga
30,027
927,112
32.4
684
26.7%
Northern Cape
3,799
280,688
13.5
244
19.9%
Northern Province
77,139
1,363,888
56.6
380
27.2%
North West
52,108
1,141,838
45.6
264
27.7%
Western Cape
35,099
1,427,959
24.6
192
29.5%
The pattern observed for Cases treated as STI is repeated here. Note also that the Male urethral discharge cases in general are between 20-30% of all cases treated as STIs using the Syndromic Case Management of STIs approach.
4.9 Condom Utilisation Rate
Condoms are procured by the national Department of Health and then distributed to the provinces and other distribution points (NGOs etc) around the country. Around 290 million condoms were purchased in 2000. Gauteng is the only province that uses small funds to procure condoms in addition to those supplied by DoH.
Research has shown that sexually active people on average are having sex 100-120 times per year (every 3-4 days), and that only around 60% of all condoms distributed are actually used for its intended purpose. 200 condoms are thus usually regarded as equivalent to a "full protection year" when viewed as a sole contraceptive method. In the SADHS, only 1.5% of all married women 15-49 years admitted they had sex with more than one partner during the last 12 months before the survey completely different from many other surveys.
Table 7: Condom Distribution Rates for 2000-01
Province
Condoms issued
Male pop >=15
Rate
Eastern Cape
11,327,141
1,800,743
6.3
Free State
3,686,506
932,524
4.0
Gauteng
Not collected
2,995,363
Kwa-Zulu Natal
7,066,069
2,588,763
2.7
Mpumalanga
3,451,673
927,112
3.7
Northern Cape
620,538
280,688
2.2
Northern Province
7,480,202
1,363,888
5.5
North West
5,289,420
1,141,838
4.6
Western Cape
6,842,081
1,427,959
4.8
In the SADHS survey, around 8.0% of respondents reported using a condom during their last sexual encounter, equivalent to 300-400 million condoms per annum. This fits reasonably well with the number of condoms distributed, and there are also few reasons to expect that respondents would lie about this (less sensitive than e.g. questions about multiple sex partners).
Further investigations are necessary to verify that the low numbers for condom distribution by PHC facilities are correct the table above indicates only 15-20% of total volume and that most of them are thus distributed via other outlets (NGOs etc) that do not yet submit distribution data for capture into the DHIS. Hospitals are also often negligent in reporting on condom distribution this seems to be regarded as a self-help thing and not a "real" service provided.
In view of the above it is recommended that:
- There be a full roll-out of the DHIS Software to all Local Government health facilities to improve the rendering of health services
- Standardisation of definitions for consistency and maintenance of data flow for comparison purposes
- The principle of "not collecting data if it is not used, to apply, and managers to be advised to focus on smaller data elements that are manageable
- Health workers not to be overloaded with the collection of unnecessary data (data elements to be limited to those that are used for planning and monitoring)
- Inculcate the culture of information use by all levels, including managers
- Inter- intra provincial differences should be used by managers to monitor progress in accessibility, equity and quality of PHC services
- Find a means of monitoring or assessing services offered by other care providers (Private Sector, Traditional Healers, etc.) for data completeness.
PHC facilities throughout the country are now providing monthly service data using uniform definitions that enable comparisons of service load and the calculation of key coverage indicators for most major PHC activities.
Data quality as well as timeliness of submission must be and will be improved further, but almost all districts have reached level 1 (capture, validation, and submission of data), some have reached level 2 (regular production of reports to managers and feedback to reporting facilities), and a few have reached level 3 (systematic use of information to improve management decisions).
Large data requirements in some provinces tend to confuse staff and contribute to poor data quality. Experience has shown that a smaller data set, 40-70 items, is adequate to calculate key indicators for programme managers and substantially improve the quality of the information derived. Mpumalanga has radically reduced and improved their data set from early 2002, and there seems to be a strong drive in Gauteng towards radically reducing their very large data set.
A careful reassessment of the data items collected and adherence to the standard essential data set would enable the DHIS PHC data to monitor the major aspects of the PHC package to see what use rates are occurring in each population served by each facility and each district. However, great care should be used NOT to overload the capacity of workers to capture and report data items. The present data set provides reliable data on the most widely used and essential services and to add many data fields could result in a decline in timeliness and quality of data.
While this report aggregates data for entire provinces, the DHIS Software provides disaggregated data and indicators for provinces, District Councils (Health Districts), sub-districts and individual facilities, making identification of problem areas readily apparent and management intervention possible.
Data completeness for the last financial year reached an unprecedented 95% for the public sector, and most data problems that can be picked up and fixed using computerised tools have been addressed. Quality of data now needs to be improved from the collection point upwards, through systematic training in local use of data/information and regular feedback from district, provincial, and national levels.
Review of data over the past year shows:
Antenatal coverage and number of visits is reaching targets, but local variation requires attention and action.
Immunisation coverage has improved significantly since the 1998 SADHS.
Nutrition measures vary, and routine weighing of under-5s is not yet regular. A strong system will allow early detection of growth faltering and intervention before malnutrition appears. Reported cases of severe malnutrition number nearly 100,000, alarmingly high, but over-reporting in at least 2 provinces seems likely.
Diarrhoea reflects continued problems with access to clean water and proper sanitation. In addition and only a small fraction of estimated cases seek care at PHC facilities.
STI rates are high, confirmed by specific male urethral infections and condom issuance rates are less than 5/man per year.
PHC utilisation is between 1.3 and 2.7 visits per capita per year, roughly half the expected value for the estimated full range of PHC services.
Extensive data and indicators are available in each province to provide more detailed assessment of PHC service levels, quality and coverage of all or most program areas. These can be readily displayed for different districts or selected facilities enabling managers to identify problem areas and take corrective action.
7.1 Department of Health. 1966. Restructuring the National Health System for Universal Primary Health Care. Department of Health: Pretoria.
7.2 Department of Health. 1998. South African Demographic Health Survey Preliminary Report. Department of Health: Pretoria.
7.3 Department of Health. Health Sector Strategic Framework, 1999-2004. Department of Health: Pretoria.
7.4 Health Systems Trust. 1996. South African Health Review. Kwik Kopy Printing: Durban.
7.5 Statistics South Africa. 2001. Statistical Release P0302 of 2/7-2001. Statistics South Africa: Pretoria.
7.6 The Centre for Health Policy. 1998. STD Management in the Private Sector. WITS: Johannesburg.
