How prepared are South Africa’s public hospitals to accurately claim costs from the country’s planned National Health Insurance (NHI) scheme?  

Not by a long shot, our study has revealed. 

This level of readiness is crucial because it will strongly influence how well the NHI, which attempts to address the uneven access to healthcare among the country’s rich and poor, will operate. 

Our research showed that staff of state healthcare facilities in 10 of the 11 NHI pilot districts that the health department launched in 2012 were frequently unable to provide the data that administrators would need to calculate accurate reimbursement amounts for services rendered.  

We found that one in three hospital patients across 45 public facilities were discharged from hospital without the necessary documentation. Moreover, only 15% of patient records contained the diagnosis codes that insurers require to calculate the cost of claims. 

The facilities we assessed included seven tertiary, 10 regional and 28 district hospitals across nine provinces. We looked at about 6,000 patient records of people who were admitted in March and July 2015.  

Our investigation revealed there was a substantial mismatch in the age of patients among different types of hospital records, affecting facilities’ ability to provide information for accurate claims, as insurance claims often vary with the age of patients. 

In some cases, the attending physician’s signature was missing from patient records, rendering such documents invalid and unusable for the processing of insurance reimbursements.   

What do patient records have to do with billing systems? 

The information that health insurers such as the NHI need to work out the amount they need to reimburse a hospital, is contained in patient records. 

Why?  

Because health workers need to assign special codes to the diagnoses made and consequent services provided. Each service or procedure has a price attached to it (prices among facilities can differ, depending on the fee of the hospital or healthcare provider).  

For this study, we analysed the diagnosis coding only — insurers use these codes to ensure that the services and procedures performed are consistent with the diagnosis information; the diagnosis data is therefore an important component of the reimbursement system.  

The diagnosis coding system that South Africa uses in both the private and public health system is called ICD-10 — the 10th revision of the World Health Organization’s international classification of disease and related health problems (ICD).  

In the private sector, medical aid companies calculate reimbursement amounts by using the services and procedures data linked to the ICD-10 codes that healthcare facilities or providers fill out on patients’ records. The NHI will do the same, as it will essentially be a state-run medical aid that pools funds to buy a package of healthcare services for everyone in South Africa. 

Regional and tertiary hospitals that participate in the NHI will be paid fixed amounts for the hospital stays of patients, based on the patient’s diagnosis and required treatment. When a facility spends less than the fixed amount, it will make a profit, and it will lose money when it spends more than the set reimbursement price. 

So if we want to know how equipped public hospitals are to bill correctly for the services they render, we need to look at how well they document and code diagnoses and procedures so that NHI administrators have accurate data to work with when they calculate the necessary reimbursements. 

What the law says about discharge summaries  

Coding data is noted in patient records and discharge summaries. Discharge summaries document a patient’s hospital journey from start to finish and include the reason for their hospitalisation as well as all the different treatments they received and might need after they have been discharged. 

But how useful discharge summaries are depends on the quality of the information they contain; in other words, how accurately the clinical information on patient records is transferred to discharge summaries. 

The South African National Health Act says all patients should receive written discharge summaries when they are discharged from hospital and copies should be kept in the patient’s health record. 

That’s why our study’s findings that only 65% of inpatient health records contained discharge summaries, and a large proportion (85%) of the summaries didn’t contain any coded diagnoses, are significant. 

The majority of South Africa’s public hospitals have a problem with tracking the costs of patient care, this study found. This poses a problem for the country’s plans for a National Health Insurance as hospitals could operate at a loss and face reimbursement challenges. (Photo: vnsny.org / Wikipedia)

The pilot district in the North West performed the best, with 84% of patients receiving discharge summaries, while the pilot district in the Northern Cape did the worst: only one in three (35%) of patients left the hospital with a discharge summary. 

At a department level, obstetrics departments showed the best practices — they provided 85% of patients with discharge summaries — and medical (54%) and psychiatric (54%) departments performed the worst.  

Perhaps the most significant finding of our study was that despite the availability of codable clinical data in inpatient health records, only 835 out of 5,575 patient records (15%) contained any diagnosis data using the ICD-10 coding system. 

The consequences of missing or inaccurate discharge summaries  

Patients who return home without discharge summaries, or summaries with incomplete or incorrect coding, could face problems in relation to any future treatment.  

Without a discharge summary they will have no documentation to use as evidence of the services they received, and to provide information on outcomes of such services, for example laboratory test results. 

The absence of this information will also have considerable financial and resource management consequences for hospitals, because the data are required to determine what type of fixed amount a hospital participating in the NHI can claim for and whether the facility makes a profit or loss as a result.  

Because coding facilitates billing, the absence of codes will result in patients leaving treatment facilities before finalising the billing process, thus delaying and complicating the reimbursement process. 

Missing coded diagnoses and data inaccuracies reveal that existing routine health information systems in public hospitals in NHI pilot districts are not yet able to sufficiently support reimbursements, which will in turn have an impact on how efficiently the NHI is able to run.  

Our study is the only research which has assessed the quality of patient records and discharge summaries across provinces in South Africa — other studies have focused on single sites or specific types of hospitals. But even so, such research has produced similar results to our study: that discharge summaries are frequently missing and patient records inaccurately coded. 

Additionally, studies in other developing countries such as Ghana and Kenya, which have also implemented NHI-like schemes, show that one of the biggest challenges such schemes face is a delay in claims reimbursement caused by inaccurate or incomplete diagnosis coding. 

Can we fix the problem?  

Interventions to improve the quality and availability of discharge summaries are clearly needed at all government hospitals. 

One option is training courses, but although such workshops have significantly improved the quality of discharge sheets in, for instance, India, very few courses have been tried out in South Africa. One of the challenges with training workshops here is the lack of time that public healthcare workers have available to attend them. Unlike the private healthcare sector, which relies on dedicated coders, government doctors are expected to code the diagnoses of all hospital patients themselves, on top of their clinical duties in an already under-resourced environment. 

Evidence shows that interventions to improve the quality of discharge summaries work best if two or more strategies are combined, for instance training and support.  

One such intervention has been tested in the internal medicine department of a tertiary Western Cape hospital. The “support” that was provided was an electronic discharge summary application — the electronic Continuity of Care Record (eCCR) — to help doctors with ICD-10 coding. The electronic system automatically integrated ICD code browsers and notes and basic coding rules. 

Doctors who participated in the study also received training on how to use the app, the basics of ICD coding and they had access to a case manager designated to support them as well as an online ICD coding course.  

The results? The training and support package increased the completeness of discharge summaries by 38%, but didn’t make much of a difference to how accurately the summaries were coded. A reason for this, the researchers pointed out, could be that discharge summaries are often the responsibility of junior clinicians, who “besides not having been trained to expert level in ICD coding are still learning to diagnose and manage complex cases in a tertiary hospital”.  

The study’s authors concluded that appointing expert coders, at least during the initial stages of the implementation of an insurance scheme, and conducting repeated training that will progressively improve coding, may be needed to ensure that discharge summaries are of a sufficient quality.  

During the Covid-19 pandemic, the need for accurate coding increased even further — not only for billing purposes, but also to have accurate records of Covid deaths and the types of conditions people developed because of Covid. The South African Medical Research Council (SAMRC) has helped the health department to implement new ICD-10 coding, developed by the WHO, for conditions associated with Covid. The SAMRC is also helping to develop a roadmap for health services to shift from the current ICD-10 coding standard to a more recent version called ICD-11. 

One of the most important lessons we’ve learned is that some of the hospitals in our study that performed relatively well with the completion of discharge summaries and diagnosis coding had management teams which recognised the importance of accurate records and that required clinicians to maintain accurate records. Perhaps the solution lies in learning from their experiences. DM/MC

This story was produced by the Bhekisisa Centre for Health Journalism. Sign up for the newsletter.

Dr Edward Nicol is a specialist scientist and head of the Health Information System Division of the SAMRC’s Burden of Disease Research Unit. Dr Lyn Hanmer is a specialist scientist and co-head of the WHO Collaborating Centre for the Family of International Classifications at the SAMRC. Professor Debbie Bradshaw is a chief specialist scientist and co-head of the WHO Collaborating Centre for the Family of International Classifications at the SAMRC.