Modelling the health economic impact of managing cow milk allergy in South Africa

Abstract

Objective: To quantify the health economic impact of managing cow milk allergy (CMA) in South Africa, from the perspective of healthcare insurers in both the private and public sectors and parents/carers of CMA sufferers.

Methods: A decision model depicting the management of CMA in South Africa was constructed, using information obtained from interviews with paediatric specialists in the private and public sectors with relevant clinical experience. The model was used to estimate the expected 12-monthly levels of healthcare resource use and corresponding costs (at 2007/08 prices) attributable to managing CMA sufferers following an initial consultation with a paediatrician.

Results: The expected 12-monthly cost incurred by an insurer attributable to managing a CMA sufferer following an initial consultation with a paediatrician was estimated to be R2,430.4 (€202.0) and R1,073.7 (€89.0) in the private and public sector, respectively. The expected 12-monthly cost incurred by parents/carers following an initial consultation with a paediatrician was estimated to be R43,563.1 (€3,634.0) and R24,899.9 (€2,076.9) in the private and public sector, respectively. The time taken for a CMA sufferer to be put on an appropriate diet and achieve symptom resolution was estimated to be 24 days in the private sector and 18 days in the public sector. The total cost to manage an annual cohort of 18,270 newly diagnosed infants with CMA in South Africa in the first year following presentation to a paediatrician was estimated to be R22.1 (€1.8) million for healthcare insurers and R489.1 (€40.8) million for parents/carers. The expected costs to insurers were driven by visits to general paediatricians and prescriptions for dermatological drugs in both the private and public sectors. The expected costs to parents/carers were driven by over-the-counter (OTC) purchases of clinical nutrition preparations.

Limitations: The intolerance rates were derived from a study among 1,000 infants with CMA in the UK, healthcare resource use was not collected prospectively and the study period was limited to 1 year following presentation to a paediatrician and does not consider the impact of CMA in subsequent years. However, most children outgrow this form of allergy by the time they reach 2 years of age.

Conclusion: CMA imposes a substantial socio-economic burden in South Africa, especially on parents/carers of CMA sufferers. Any strategy that reduces this burden should potentially lead to higher compliance with clinicians’ recommendations, thereby improving health outcomes associated with treatment and should also release healthcare resource use for alternative use.

• amino acid formula
• cost
• cow milk allergy
• extensively hydrolysed formula
• resource use
• South Africa
• soy

Introduction

The South African healthcare system is predominantly state-funded with free access to basic primary and preventative care for all. The state-funded health service covers ∼90% of the population, and the private sector serves the remainder[1]. However, the state-funded health service only accounts for ∼41% of all health expenditure, with the remaining 59% due to private expenditure through prepaid insurance plans or out-of-pocket payments[1].

The public healthcare system consists of three tiers referred to as the Uniform Patient Fee Schedule (UPFS)[2]. Tier level H1 covers patients/parents who are on low incomes and who only pay what they can afford, although parents of children <6 years of age receive free paediatric healthcare on this Tier. Tier level H2 covers patients/parents who can afford to pay more than those on H1. Tier level H3 covers patients/parents with moderate income who either have insurance or can afford to pay in cash, although the insurance is usually the most basic cover with many exclusions.

Primary care facilities in the public sector are available to the entire population, however major barriers to care include physical distance, loss of income, waiting times, cost of transportation and inconvenient hours[3]. Consequently, there are numerous private healthcare insurers offering various healthcare plans. However, only 20% of the population has private medical insurance[4]. The type of costs reimbursed by an insurer depends on the details of the policy held, but most policies have a cap on usage of healthcare services and the costs reimbursed. For example, most policies restrict the number of physician visits, procedures and tests reimbursed. However, the cost of ethical pharmaceuticals and hospital admissions are fully reimbursed, even for those with a basic insurance package. Notwithstanding this, clinical nutrition preparations are not routinely reimbursed either in the public or private sector[4].

Although South Africa is classified as a middle-income country and spends about 8.6% of its gross domestic product on healthcare, the majority of the South African population has inadequate access to healthcare services and tackling epidemics of HIV/AIDS and tuberculosis are still the largest healthcare challenges for the country[1]. Nevertheless, food allergies including cow milk allergy (CMA) are treatable conditions that with appropriate management could potentially release scarce healthcare resources for alternative use within the healthcare system.

CMA is an adverse reaction to protein in cow's milk, involving the immune system. Its incidence in infancy in western industrialised countries has been estimated at 2–3%[5]. Symptoms of CMA generally appear within the first few months of life. An infant can experience symptoms either very quickly after feeding (rapid onset) or not until 7–10 days after consuming cow's milk (slower onset)[6–8]. The slower-onset reaction is more common and symptoms may include loose stools (possibly containing blood), vomiting, refusing food, irritability, colic, skin rashes and failure to thrive. This type of reaction is more difficult to diagnose because the same symptoms may occur with other health conditions[5]. Several tests can be used to inform a diagnosis of CMA including a stool test, radioallergosorbent test (RAST), skin prick test and patch testing ^8,9. Most children will outgrow this form of allergy by 2 years of age.

Nutritional formulas for infants with CMA include soy-based milk which contain proteins found in soybeans, extensively hydrolysed formulas (eHF) which contain milk proteins which have been pre-digested, and elemental formulas, (such as Neocate), which contain only amino acids in the absence of any protein and thus are not allergenic[10–14]. Partially hydrolysed formulas are not recommended for use in infants with CMA since they are not considered truly hypoallergenic and can still provoke significant allergic reactions[15].

Mothers of affected infants who are being breast-fed are advised to eliminate dairy products from their diet under the close supervision of a dietician, because a strict diet must be followed to ensure adequate intake of nutrients while eliminating cow milk protein. In many European countries, and in Australia and New Zealand, mothers of bottle-fed infants are often advised to feed either a soy formula or an eHF to those >6 months of age, but to initially feed only an eHF to those <6 months of age. Those infants who cannot tolerate soy would generally be switched to eHF and those who cannot tolerate eHF would generally be switched to an elemental formula. Irrespective of whether an affected infant is managed in the public or private sector, clinical nutrition preparations are bought by parents as an over-the-counter product, since they are not routinely reimbursed by any of the insurers.

In South Africa there are no guidelines on how to diagnose and manage CMA and it therefore varies from one clinician to the other. Against this background, this study assessed the resource implications and budget impact of such fragmented current practice from the perspectives of healthcare insurers and parents/carers of CMA sufferers in both the private and public sectors.

Methods

Study objective

The study objective was to quantify healthcare resource use and assess the budget impact of managing CMA in South Africa. This study was conducted from the perspective of (1) healthcare insurers in both the private and public sectors and (2) parents/carers of CMA sufferers.

Decision model

A computer-based decision model (Figures 1 and 2) was constructed using TreeAge Pro 2007 depicting the treatment patterns and associated resource use attributable to managing newly diagnosed infants <1 year of age at the time of presentation with the symptoms of CMA. The decision model was used to simulate the management of the annual cohort of new CMA sufferers over the first 12 months following an initial consultation with a paediatrician in the public and private sectors.

Figure 1. Decision model depicting the management of CMA in the public sector in South Africa. Numbers denote the probability that a patient would follow a particular path. eHF, extensively hydrolysed formula; AAF, amino acid formula.

Figure 2. Decision model depicting the management of cow milk allergy in the private sector in South Africa. Numbers denote the probability that a patient would follow a particular path. eHF, extensively hydrolysed formula; AAF, amino acid formula.

Model inputs – clinical outcomes

Based on the findings of a 1-year follow-up study the authors have undertaken among 1,000 infants with CMA in the UK, it was estimated that 9% of infants would be intolerant to soy and 29% to an eHF[16]. Moreover, the literature review could not identify any publications reporting intolerance rates to such unconventional diets as boiled cow's milk, evaporated cow's milk and goat's milk. Therefore, it was assumed that all infants with CMA would be intolerant to these diets.

Model inputs – resource use

A systematic literature search was performed using available databases, however no publications were found on the management of CMA in South Africa. Therefore, mean values of resource use were estimated using information obtained from interviews with 16 paediatric specialists comprising general paediatricians (n = 12), a paediatric dermatologist (n = 1), a paediatric allergist (n = 1), a paediatric gastroenterologist (n = 1) and a paediatric pulmonologist (n = 1) in the private sector. Three paediatric specialists in the public sector with relevant clinical experience of managing CMA were also interviewed. The interviews used a structured questionnaire and interviewees were asked objective questions about diagnosis, patient management, resource utilisation, unit costs, use of clinical nutrition preparations, use of prescribed drugs and time to symptom resolution based on their own practice together with general questions about the infrastructure of the healthcare system in South Africa. The information elicited from the interviewees was inputted into an Excel spreadsheet and analysed using descriptive statistics to obtain mean parameter values, as outlined below.

Management of CMA sufferers in the private sector

The 12 interviewed general paediatricians collectively saw 1,330 CMA sufferers in a year, of whom 34% were new patients. The four paediatric specialists saw a further 630 CMA sufferers each year of whom 82% were new patients. It was estimated that 51% of patients were self-referred, 21% were referred by GPs, 17% were referred by other consultant paediatricians, 8% were referred by dieticians and the remaining 4% were referred via Accident and Emergency (A&E). Moreover, 57% of CMA sufferers seen in the private sector were of Caucasian origin, 18% were of African origin and 26% were of mixed race. Approximately 26% of patients present with a combination of eczema and gastrointestinal (GI) symptoms, 28% with GI symptoms alone, 29% with eczema symptoms alone, 2% with urticaria alone, 4% with urticaria and other symptoms, 2% with failure to thrive and <1% with urticaria alone.

Irrespective of presenting symptoms, a diagnosis would be based on history and presentation of clinical signs and symptoms. In addition, patients may receive a RAST (65% of patients), IgE screen (15% of patients), skin prick test (13% of patients), stool test (12% of patients), patch test (8% of patients), Fx5 food allergy screen (8% of patients), cellular allergen stimulation test (CAST) (4% of patients), barium X-ray (1% of patients), pH measure test (1% of patients) and a biopsy (<1% of patients). None of the interviewed clinicians performed a double-blind placebo-controlled challenge, although the paediatricians would instruct dietary elimination of cow's milk from an infant's diet, but a full open challenge would only be performed by 20% of the interviewed paediatricians (i.e., 80% of paediatricians did not re-challenge sufferers following dietary elimination of cow' milk). The clinicians explained that most parents did not want their baby to be re-challenged following symptom resolution and either did not have time for additional visits or wished to keep their insurance claim to a minimum. Moreover, it was estimated that an open challenge would require two additional visits to a paediatrician, but apart from the cost of the visits it did not involve the use of any other healthcare resources. The period of initial elimination was ∼20 days and it would start from the initial visit. Approximately 88% of the interviewees used an eHF and the other 12% used an AAF formula during the open challenge.

All suspected bottle-fed CMA sufferers not undergoing elimination/provocation would have their treatment started at the first consultation. Approximately 57% of parents would be advised to treat their child with soy, 38% with eHF and 5% with goat's milk. Parents of infants who are intolerant to soy would be advised to try an eHF (62% of cases) or an AAF (28% of cases) or a lactose-free diet (10% of cases). Parents of infants who are intolerant to the first eHF would be advised to try another eHF (42% of cases) or an AAF (45% of cases) or soy (13% of cases). In addition to advising on the use of clinical nutrition preparations, 1% of parents would be advised to feed their infant evaporated milk or boiled cow's milk.

Following the start of a diet, an estimated 73% of symptom-free infants would have one follow-up visit with their paediatrician within 7 weeks of the initial consultation and 90% of patients who remain symptomatic following the start of a diet would have a follow-up visit within 3 weeks of the initial consultation. Approximately 17% of patients who remain symptomatic would have ∼1 consultation with a paediatric specialist. Of these, 9% would be referred to an allergist, 5% to a dermatologist, 2% to a gastroenetrologist and 1% to a pulmonologist. The referring paediatrician would subsequently manage the patient. Patients would generally have two visits with their referring paediatrician before a referral to a specialist. All subsequent dietary advice would be conducted over the telephone since none of the interviewees charged for telephone consultations. Paediatricians managed approximately 98% of patients, albeit by telephone and the remaining 2% would be managed by their GP. Additionally, ∼38% of breast-feeding mothers and ∼21% of bottle-feeding mothers would be referred to a dietician for two visits.

CMA sufferers presenting with GI symptoms would also be prescribed an antihistamine (30% of patients), a proton pump inhibitor (26% of patients), an H₂-antagonist (12% of patients) and/or an anti-spasmodic (8% of patients). Patients presenting with urticaria would generally be prescribed an antihistamine (40% of patients) and/or a steroid (40% of patients). Patients presenting with eczema would generally be prescribed an antihistamine (63% of patients), a corticosteroid (79% of patients), an emollient (18% of patients), a benzodiazepine (15% of patients) and/or a macrolide (5% of patients). Patients either received repeat prescriptions for up to 6 months, or were prescribed sufficient medication to last until the next physician visit. Patients generally took medication for symptomatic relief of GI symptoms for up to 6 months, but up to 12 months for symptomatic relief of dermatological symptoms.

An estimated 5% of CMA sufferers were hospitalised for diagnostic reasons for a mean 4 days. While in hospital, patients underwent a range of tests including, but not limited to, blood tests (100% of patients), biochemistry tests (75% of patients), pH testing (30% of patients), stool tests (25% of patients), urine swab and culture (25% of patients), phadiotop (25% of patients), skin prick tests (25% of patients), bacteriology (25% of patients) and barium X-rays (13% of patients). The costs associated with medications and hospitalisations were fully reimbursed by the insurers.

Management of CMA sufferers in the public sector

The three interviewed paediatricians collectively saw 252 patients with CMA in a year. Of these ∼33% were other paediatric referrals, ∼27% were GP referrals, ∼27% were referred from community health centres/day clinics and ∼13% were referred from A&E departments. Approximately 5% of CMA sufferers managed in the public sector were of Caucasian origin, 75% of African origin and 20% of mixed race.

An estimated 87% of sufferers presented with eczema alone, 10% with a combination of eczema and GI symptoms, 2% with GI symptoms alone, <1% with GI and respiratory symptoms, <1% with failure to thrive and <1% with urticaria.

In all cases a diagnosis would be based on history and presentation of clinical signs and symptoms. In addition, patients would receive a RAST (88% of patients), CAST (33% of patients), IgE screen (26% of patients), skin prick test (5% of patients), swab for culture (5% of patients) and biopsy (2% of patients). All the paediatricians advised dietary elimination of cow's milk, however none of the interviewees performed a full elimination/provocation procedure. The reasons for not conducting provocation included parents’ unwillingness to allow their child to be re-challenged following symptom resolution as well as time costs and costs associated with additional visits. None of the interviewees performed a double-blind placebo-controlled cow's milk challenge.

All sufferers would have an initial consultation with a paediatrician during which treatment would be initiated. Parents of an estimated 93% and 7% of sufferers would be advised to feed with soy and eHF, respectively. The parents of an infant intolerant to initial treatment with soy would be advised to switch to an eHF and to switch to soy if they were intolerant to initial treatment with eHF. The parents of an infant intolerant to second-line treatment with eHF or soy would be advised to try another eHF before trying an AAF. Parents were advised to keep their infant on a clinical nutrition preparation up to 1 year of age. However, most parents started treatment with soy irrespective of the clinicians’ guidance because of cost.

Following the start of a diet, ∼50% of symptom-free infants would have ∼1 follow-up visit within 2 weeks of the initial consultation and all symptomatic patients would have ∼2 follow-up visits every 20 days. A small number of infants who remained symptomatic would be referred to a paediatric allergist (<1% of infants) or a paediatric gastroenterologist (∼4% of infants). These patients were expected to have two visits with the specialist after which they would be discharged to the referring paediatrician for subsequent management. The waiting time to see a paediatric specialist was ∼3 weeks. Additionally, an estimated two-thirds of breast-feeding mothers and bottle-feeding mothers would be referred to a dietician for a mean two visits, who would advise the breast-feeding mothers to supplement their diet with a combination of vitamins and minerals.

CMA sufferers presenting with eczema would generally be prescribed a corticosteroid (100% of patients), an antihistamine (100% of patients), an emollient (51% of patients) and an antibiotic (8% of patients). All patients presenting with GI symptoms would generally be prescribed a proton pump inhibitor. The cost of medication was fully reimbursed for those covered by insurance. Approximately 50% of patients (those with more severe symptoms) would have to see their paediatrician to get a repeat prescription. The other 50% of patients would receive a sufficient number of repeat prescriptions to last until the next clinician visit. Patients generally took medication for symptomatic relief of GI symptoms for up to 6 months but up to 12 months for symptomatic relief of dermatological symptoms.

An estimated 2% of CMA sufferers would be hospitalised for a mean 5 days. While in hospital, patients would undergo a range of tests including, but not limited to, blood tests (50% on patients), biochemistry tests (50% of patients), pH testing (50% of patients), swab for culture (50% of patients), wet wraps (50% of patients) and endoscopy/biopsy (50% of patients).

Model outputs

The time taken for CMA sufferers to be put on an appropriate diet and achieve symptom resolution was estimated depending on whether they were initially seen in the public or private sector.

The model estimated the expected 12-monthly levels of healthcare resource use attributable to managing a new CMA sufferer following initial consultation with a paediatrician. Unit resource costs in South African Rand at 2007/08 prices (Table 1) obtained from the interviewees were applied to the resource utilisation estimates within the model. This enabled an estimation of the expected 12-monthly cost of healthcare resource use associated with managing a new CMA sufferer following initial consultation with a paediatrician, from the perspective of healthcare insurers and parents/carers of CMA sufferers in both the private and public sectors. The analysis assumed that all costs of healthcare resource use would be reimbursed by insurers with the exception of clinical nutrition preparations in both the private and public sectors. Consequently, it was assumed that the costs of clinical nutrition preparations would be paid by parents/carers of CMA sufferers in both the private and public sectors.

Table 1.  Unit resource costs at 2007/08 prices.

Resource	Unit cost in South African Rand* at 2007/08 prices in the
	Private sector	Public sector
Barium test	R275.00	–
Can of amino acid formula (400 g)	R492.00	R492.00
Can of cow's milk formula	R12.00	–
Can of extensively hydrolysed formula (400 g)	R235.00	R235.00
Can of evaporated milk	R12.00	–
Can of goat's milk	R12.00	–
Can of lactose free extensively hydrolysed formula	R235.00	–
Can of soy (400 g)	R119.00	R119.00
Cellular allergen stimulation test (CAST)	R248.00	R64.00
Dietician visit	R366.00	R41.00
Endoscopy	R2,431.00	R729.00
Fx5 allergy screen	R137.00	–
Follow-up allergist visit	R280.00	R74.00
Follow-up gastroenterologist visit	R430.00	R114.00
Follow-up general paediatrician visit	R347.00	R92.00
Follow-up paediatric surgeon visit	R617.00	–
Follow-up pulmonologist visit	R617.00	–
Follow-up dermatologist visit	R270.00	–
Hospital stay per day	R1,405.00	R820.00
Immunoglobulin E (IgE) screen	R155.00	R40.00
Initial allergist visit	R525.00	R139.00
Initial dermatologist visit	R428.00	–
Initial gastroenterologist visit	R70.00	R185.00
Initial general paediatrician visit	R408.00	R108.00
Initial paediatric surgeon visit	R813.00	–
Initial pulmonologist visit	R813.00	–
Patch test	R81.00	–
pH measure test	R275.00	–
Radioallergosorbent test (RAST)	R323.00	R84.00
Skin prick test	R81.00	R60.00
Stool test	R163.00	–
Swab	R81.00	R21.00

*R12 = ∼€1.

Budget impact of managing CMA in South Africa

The number of infants born in South Africa was estimated to be 775,452 per annum[1]. The estimated incidence of CMA among infants is 2.5%[5]. Hence, the annual number of new CMA sufferers in South Africa was estimated to be 18,270 infants. By accounting for the estimated 9% of the population with private medical insurance, it was estimated that ∼1,820 infants with CMA would be managed in the private sector and the remaining 16,450 in the public sector. The 12-monthly cost of healthcare resource use associated with managing an annual cohort of 18,270 new CMA sufferers following initial consultation with a paediatrician at 2007/08 prices was estimated from the perspective of healthcare insures and parents/carers of CMA sufferers in both the private and public sectors.

Sensitivity analyses

In order to assess uncertainty within the model, probabilistic sensitivity analysis was undertaken using Monte Carlo iterations (5,000 iterations of the model) by simultaneously varying the probabilities, resource use values and unit costs. Probabilities were varied randomly according to a beta distribution by assuming a 20% standard deviation around the mean. Resource use and unit cost estimates were varied randomly according to a lognormal distribution by assuming a 20% standard deviation around the mean. The 20% variability in each parameter was considered reasonable since it more than covered the variance in clinicians’ responses to individual questions. Additionally, deterministic sensitivity analyses were performed to identify how the results would change by varying different assumptions in the model (as described in Table 5).

Results

Resource implications and expected healthcare costs

The expected 12-monthly cost of healthcare resource use attributable to managing a CMA sufferer following an initial consultation with a paediatrician was estimated to be R2,430 (95% CI: R2,029; R2,978) in the private sector and R1,074 (95% CI: R956; R1,237) in the public sector. Additionally, the expected 12-monthly cost incurred by parents/carers of a CMA sufferer following an initial consultation with a paediatrician was estimated to be R43,563 (95% CI: R32,469; R58,277) and R24,899 (95% CI: R14,859; R38,002) for an infant managed in the private and public sector, respectively (Table 2).

Table 2.  Expected 12-monthly costs (South African Rand* at 2007/08 prices) per cow milk allergy sufferer following an initial consultation with a paediatrician.

Resource	Expected 12-monthly cost (South African Rand at 2007/08 prices) per cow milk allergy sufferer following an initial consultation with a paediatrician. (Percentage of total cost in parenthesis).
	Private sector				Public sector
	Cost to parents/carers		Cost to healthcare insurers		Cost to parents/carers		Cost to healthcare insurers
Amino acid formula	R14,389.40	(33%)	–		R552.20	(2%)	–
Extensively hydrolysed formula	R24,071.10	(55%)	–		R5,429.40	(22%)	–
Soy	R4,882.10	(11%)	–		R18,918.30	(76%)	–
Goat's milk	R10.80	(<1%)	–		–		–
Evaporated milk	R0.20	(<1%)	–		–		–
Lactose-free milk	R208.90	(<1%)	–		–		–
Cow's milk	R0.60	(<1%)	–		–		–
Diagnostic tests	–		R293.30	(12%)	–		R123.10	(11%)
General paediatrician visits	–		R785.00	(32%)	–		R211.70	(20%)
Specialist outpatient visits	–		R105.20	(4%)	–		R9.20	(1%)
Dietician visits	–		R269.80	(11%)	–		R59.30	(6%)
Hospital admissions	–		R337.60	(14%)	–		R92.80	(9%)
Hospital tests	–		R75.10	(3%)	–		R10.10	(1%)
Gastrointestinal drugs	–		R82.50	(3%)	–		R4.30	(<1%)
Dermatologicals	–		R477.90	(20%)	–		R563.20	(52%)
Antihistamines	–		R3.95	(<1%)	–		–
TOTAL	R43,563.10	(100%)	R2,430.35	(100%)	R24,899.90	(100%)	R1,073.70	(100%)

*R12 = ∼€1.

The primary cost driver associated with managing an infant in the private sector is expected to be the cost of visits to a general paediatrician, accounting for 32% of the healthcare cost. The secondary cost driver is expected to be the cost of dermatological prescriptions, accounting for a further 20%. In contrast, the cost of dermatological prescriptions is expected to be the primary cost driver in the public sector followed by the cost of general paediatrician visits, accounting for 52% and 20% of the expected healthcare cost, respectively.

Of the costs incurred by parents/carers of an infant managed in the private sector, ∼55%, ∼33% and ∼11% is expected to be due to the purchase of eHF, AAF and soy preparations, respectively. In contrast, ∼76%, ∼22% and ∼2% of the costs incurred by parents/carers of an infant managed in the public sector is expected to be due to the purchase of soy, eHF and AAF preparations, respectively.

The time taken for a CMA sufferer to be put on an appropriate diet and achieve symptom resolution was estimated to be 24 (95% CI: 23; 25) days in the private sector and 18 (95% CI: 16; 19) days in the public sector.

Budget impact of managing CMA

The expected 12-monthly levels of healthcare resource use attributable to managing an annual cohort of 18,270 newly diagnosed infants with CMA over the first year following initial consultation with a paediatrician is shown in Table 3, from which it can be seen that utilisation of healthcare resources is expected to be slightly less intensive on a per patient basis in the public sector. For example, in the public sector there are fewer hospital admissions (2 vs. 5% in the private sector). However, there are more visits to dieticians in the public sector (1.5 vs. 0.7 per patient in the private sector).

Table 3.  Expected healthcare resource use (95% confidence intervals) attributable to managing 18,270 new cow milk allergy sufferers following initial consultation with a paediatrician.

Resource	Expected healthcare resource use (95% confidence intervals) attributable to managing new cow milk allergy sufferers following initial consultation with a paediatrician.
	16,450 new cow milk allergy sufferers in the public sector	1,820 new cow milk allergy sufferers in the private sector
General paediatrician visits	35,021 (31,177; 40,342)	3,970 (3,306; 4,852)
Paediatric allergist visits	74 (66; 85)	220 (183; 269)
Paediatric dermatologist visits	–	122 (102; 149)
Paediatric gastroenterologist visits	888 (790; 1,023)	40 (33; 48)
Paediatric surgeon visits	–	4 (3; 5)
Paediatric pulmonologist visits	–	20 (17; 24)
Hospital admissions	329 (293; 379)	92 (77; 112)
Hospital days	1,645 (1,464; 1,895)	317 (264; 388)
Paediatrician visits while in hospital	1,645 (1,464; 1,895)	309 (257; 377)
Dietician visits	23,918 (21,293; 27,552)	1,344 (1,120; 1,644)

The total expected cost to manage an annual cohort of 1,820 infants with CMA in the first year following presentation to a paediatrician in the private sector was estimated to be R4.4 (95% CI: R3.7; R5.4) million for healthcare insurers and R79.5 (95% CI: R62.7; R106.1) million for the parents/carers of CMA sufferers (Table 4). Additionally, the total expected cost to manage an annual cohort of 16,450 infants with CMA in the first year following presentation to a paediatrician in the public sector was estimated to be R17.7 (95% CI: R15.7; R20.4) million for healthcare insurers and R409.6 (95% CI: R240.0; R625.2) million for parent/carers of CMA sufferers (Table 4). Hence, the total cost to manage 18,270 newly diagnosed infants with CMA in South Africa in the first year following presentation to a paediatrician was estimated to be R22.1 (95% CI: R19.4; R25.8) million for healthcare insurers and R489.1 (95% CI: R302.7; R731.2) million for parents/carers of CMA sufferers in both private and public sectors (Table 4).

Table 4.  Expected cost of healthcare resource use (95% confidence intervals) attributable to managing 18,270 new cow milk allergy sufferers following initial consultation with a paediatrician.

Resource	Expected cost (South African Rand*, million at 2007/08 prices) of healthcare resource use attributable to managing new cow milk allergy sufferers following initial consultation with a paediatrician. (95% confidence intervals in parenthesis).
	16,450 new cow milk allergy sufferers in the public sector		1,820 new cow milk allergy sufferers in the private sector		18,270 new cow milk allergy sufferers in both the public and private sectors
	Cost to parents/carers	Cost to healthcare insurers	Cost to parents/carers	Cost to healthcare insurers	Cost to parents/carers	Cost to healthcare insurers
Clinical nutrition
Amino acid formula	R9.08 (R5.32; R13.86)	–	R26.25 R20.72; R35.03)	–	R35.33 (R26.04; R36.71)	–
Extensively hydrolysed formula	R89.32 (R52.33; R136.32)	–	R43.91 R34.66; R58.61)	–	R133.22 (R87.00; 174.53)	–
Soy	R311.22 (R182.35; 474.99)	–	R8.90 (R7.03; R11.89)	–	R320.13 (R189.38; 82.74)	–
Goat's milk	R0.01 (R0.01; R0.01)	–	R0.02 (R0.02; R0.03)	–	R0.02 (R0.02; R0.02)	–
Evaporated milk	R0.01 (R0.01; R0.01)	–	R0.01 (R0.01; R0.01)	–	R0.01 (R0.001; R0.001)	–
Lactose-free milk	R0.01 (R0.001; R0.01)	–	R0.38 (R0.30; R0.51)	–	R0.38 (R0.30; R0.33)	–
Cow's milk	R0.01 (R0.001; R0.01)	–	R0.01 (R0.01; R0.01)	–	R0.01 (R0.01; R0.01)	–
Diagnostic tests	–	R2.03 (R1.80; R2.33)	–	R0.54 (R0.45; R0.65)	–	R2.56 (R2.25; R2.99)
Paediatric clinician visits
General paediatrician	–	R3.48 (R3.10; R4.01)	–	R1.43 (R1.19; R1.75)	–	R4.91 (R4.29; R5.76)
Paediatric allergist	–	R0.01 (R0.01; R0.01)	–	R0.10 (R0.08; R0.12)	–	R0.11 (R0.09; R0.13)
Paediatric dermatologist	–	R0.01 (R0.01; R0.01)	–	R0.05 (R0.04; R0.06)	–	R0.05 (R0.04; R0.06)
Paediatric gastroenterologist	–	R0.14 (R0.13; R0.17)	–	R0.02 (R0.02; R0.03)	–	R0.17 (R0.15; R0.20)
Paediatric surgeon	–	R0.01 (R0.01; R0.01)	–	R0.01 (R0.01; 0.01)	–	R0.01 (R0.01; R0.01)
Paediatric pulmonologist	–	R0.01 (R0.01; R0.01)	–	R0.02 (R0.01; R0.02)	–	R0.02 (R0.01; R0.02)
Dietician	–	R0.98 (R0.87; R1.12)	–	R0.49 (R0.41; R0.60)	–	R1.47 (R1.28; R1.73)
Admissions
Hospital stay	–	R1.53 (R1.36; R1.76)	–	R0.62 (R0.51; R0.75)	–	R2.14 (R1.87; R2.51)
Hospital tests	–	R0.17 (R0.15; R0.19)	–	R0.14 (R0.11; R0.17)	–	R0.30 (R0.26; R0.36)
Medication
Gastrointestinal drugs	–	R0.07 (R0.06; R0.08)	–	R0.15 (R0.13; R0.18)	–	R0.22 (R0.19; R0.27)
Dermatologicals	–	R9.27 (R8.25; R10.67)	–	R0.87 (R0.73; R1.07)	–	R10.14 (R8.97; R11.74)
Antihistamines	–	R0.01 (R0.01; R0.01)	–	R0.01 (R0.01; R0.01)	–	R0.01 (R0.01; R0.01)
TOTAL	R409.62 (R240.00; R625.17)	R17.66 (R15.73; R20.35)	R79.46 (R62.73; 106.06)	R4.43 (R3.69; R5.42)	R489.09 (R302.74; R731.24)	R22.10 (R19.42; R25.77)

*R12 = ∼€1.

Sensitivity analyses

Deterministic sensitivity analyses (Table 5) demonstrated that the greatest area of sensitivity surrounds the impact of the acquisition costs of the clinical nutrition preparations on the expected cost to parents/carers of a newly diagnosed CMA sufferer. For example, as the acquisition cost of eHF is varied the expected cost to parents/carers ranges from R38,758 to R48,388 in the private sector and from R23,813 to R25,986 in the public sector. Similarly, as the acquisition cost of soy is varied the expected cost to parents/carers ranges from R42,588 to R44,517 in the private sector and from R24,084 to R28,556 in the public sector.

Table 5.  Sensitivity analyses.

Parameter	Base-case value	Effect on the expected 12-monthly cost for managing a new cow milk allergy sufferer following an initial consultation with a paediatrician
Unit cost per can of amino acid formula ranges from R390 to R590	R492	The cost to parents/carers ranges from: R42,588 to R44,517 per patient in the private sector and R21,084 to R28,556 per patient in the public sector.
Unit cost per can of extensively hydrolysed formula ranges from R90 to R590	R235	The cost to parents/ranges from: R38,758 to R48,388 per patient in the private sector and R23,8139 to R25,986 per patient in the public sector.
Unit cost per can of soy ranges from R95 to R142	R119	The cost to healthcare insurers ranges from: R38,758 to R48,388 per patient in the private sector and R23,8139 to R25,986 per patient in the public sector.
Intolerance to soy ranges from 0.05 to 0.15	0.09	The cost to healthcare insurers ranges from: R2,419 to R2,420 per patient in the private sector and R1,070 to R1,080 per patient in the public sector. The cost to parents/carers ranges from: R31,286 to R33,506 per patient in the private sector and R23,990 to R26,261 per patient in the public sector.
Intolerance to extensively hydrolysed formula ranges from 0.10 to 0.40	0.29	The cost to healthcare insurers ranges from R2,414 to R2,422 per patient in the private sector and R1,072 to R1,075 per patient in the public sector. The cost to parents/carers ranges from: R30,594 to R33,366 per patient in the private sector and R24,896 to R25,010 per patient in the public sector.
Unit cost of initial consultation with a paediatrician ranges from R326 to R489 in the private sector	R408	The cost to healthcare insurers in the private sector ranges from R2,350 to R2,511 per patient.
Unit cost of initial consultation with a paediatrician ranges from R86 to R129 in the public sector	R108	The cost to healthcare insurers in the public sector ranges from R1,052 to R1,095 per patient.
Unit cost of a follow-up consultation with a paediatrician ranges from R277 to R416 in the private sector	R347	The cost to healthcare insurers in the private sector ranges from R2,354 to R2,50 per patient.
Unit cost of a follow-up consultation with a paediatrician ranges from R73 to R110 in the public sector	R92	The cost to healthcare insurers in the public sector ranges from R1,053 to R1,094 per patient.
Unit cost of a dietician consultation ranges from R292 to R439 in the private sector	R366	The cost to healthcare insurers in the private sector ranges from R2,376 to R2,485 per patient.
Unit cost of a dietician consultation ranges from R22 to R59 in the public sector	R41	The cost to healthcare insurers in the public sector ranges from R1,061 to R1,084 per patient.
Unit cost of hospital stay per day ranges from R1124 to R1686 in the private sector	R1,405	The cost to healthcare insurers in the private sector ranges from R2,382 to R2,479 per patient.
Unit cost of hospital stay per day ranges from R656 to R984 in the public sector	R820	The cost to healthcare insurers in the public sector ranges from R1,057 to R1,090 per patient.
Number of follow-up consultations with a paediatrician ranges from 1 to 3	2	The cost to healthcare insurers ranges from: R2,393 to R3,080 per patient in the private sector and R1,062 to R1,245 per patient in the public sector.

*R12 = ∼€1.

Notwithstanding this, the expected cost to parents/carers is most sensitive to changes in the acquisition costs of AAF. As the acquisition cost of AAF is varied the expected cost to parents/carers ranges from R40,676 to R46,440 in the private sector and from R24,789 to R25,009 in the public sector. Clearly, reducing the acquisition cost of clinical nutrition preparations has the potential to reduce the cost incurred by parents/sufferers. Moreover, the use of a particular clinical nutrition preparation is based on affordability and therefore the advice given to parents/carers by a clinician may well reflect their ability to pay rather than ‘best practice’.

Additionally, the expected costs to parents/carers in both the private and public sectors were found to be sensitive to the probability of being intolerant to soy. This is because those who are intolerant to soy would be switched to a more expensive formula. However, the expected costs to parents/carers were only sensitive to the probability of a CMA sufferer in the private sector being intolerant to eHF, but not in the public sector. This is due to the higher probability of a CMA sufferer who is intolerant to eHF being switched to a more expensive formula in the private sector than in the public sector and a higher reliance on soy to manage CMA in the public sector.

Sensitivity analyses also demonstrated that the expected costs to healthcare insurers is sensitive to the cost of consultations with a general paediatrician and the number of follow-up visits with general paediatricians in the private sector, but not in the public sector. This could be explained by the relatively low unit cost of clinician visits in the public sector compared to the private sector.

The expected costs were found to be insensitive to changes in any of the other model's inputs. Additionally, the length of time for a CMA sufferer to be put on appropriate diet and achieve symptom resolution was not affected by changes to any of the model's inputs.

Discussion

There were no published studies on the management of CMA in South Africa. Hence, the current study examined the health economic impact of managing CMA in South Africa using a decision-modelling approach that incorporated clinical outcomes obtained by the authors in their previous study on CMA in the UK[16] and resource utilisation estimates derived from interviews with clinicians involved in managing CMA in South Africa. Decision analysis is a systematic approach to decision making under uncertainty that uses mathematical relationships to define a series of possible consequences that flow from a set of options being evaluated[17]. Hence, a key advantage of using this approach is that it allows for variability and uncertainty in model inputs, as depicted by the Monte Carlo simulations.

The budget impact analyses revealed that CMA imposes a relatively modest burden on South African healthcare insurers. The total expected cost to manage an annual cohort of 1,820 CMA sufferers in the first year following presentation to a paediatrician was estimated to be R4.4 million in the private sector and R17.7 million to manage 16,450 sufferers in the public sector. However, the cost incurred by parents/carers of 1,820 sufferers in the private sector in the first year following presentation to a paediatrician was estimated to be R79.5 million and R409.6 million for parents/carers of 16,450 sufferers in the public sector. Hence, the total cost to manage 18,270 newly diagnosed CMA infants with CMA in South Africa in the first year following presentation to a paediatrician was estimated to be R22.1 million for healthcare insurers and R489.1 million for parents/carers. This analysis indicates that despite the relatively low incidence, CMA represents a significant overall cost to South African society. Moreover, the estimated burden of CMA in South Africa is comparable to that found in other countries[16],[18],[19]. However, the distinctive reimbursement practices of the South African healthcare system result in a disproportionate burden being placed on parents/carers of CMA sufferers compared to healthcare insurers. Parents/carers of CMA sufferers incur ∼96% of the expected healthcare costs of managing this allergy, which is accounted for entirely by the acquisition costs of clinical nutrition preparations in both the private and public sectors.

It was also found that the expected costs incurred by both insurers and parents/carers are higher in the private sector than the public sector. This is explained by a greater number of hospital admissions, hospital tests, outpatient visits, drugs for GI-related problems and higher usage of more expensive clinical nutrition preparations in the private sector. Moreover, the time for a CMA sufferer to be put on an appropriate diet and achieve symptom resolution was estimated to be longer in the private sector due to (1) a larger number of patient management pathways in the private sector, (2) a greater probability of a CMA sufferer being referred to a specialist paediatrician in the private sector and (3) a higher probability of sufferers being prescribed medications for symptomatic relief of CMA in the private sector, since these are reimbursed by insurers and thus free to parents/carers. All these serve to prolong the time it takes to achieve symptom resolution.

Clinical nutrition preparations are OTC products in South Africa. Hence, use of clinical nutrition preparations is driven by parents/carers’ affordability and is price elastic. Parents are the decision makers although clinicians may influence that decision. In accordance with the above, it was found that clinicians’ recommendations regarding the use of a particular clinical nutrition preparation are not always followed by parents/carers of CMA sufferers as they often select the cheapest available option. This arrangement is reflected in the disproportionately high usage of prescribed medications to relieve the symptoms of CMA in South Africa compared to Western European countries, in which the authors have previously reported that medication for symptomatic relief of CMA is generally not prescribed unless severe GI symptoms are present[16],[18],[19].

Notwithstanding the above, approximately 19% of the interviewed physicians in the private sector said it was possible to obtain reimbursement for a clinical nutrition preparation from an insurer. However, for a clinical nutrition preparation to be reimbursed CMA sufferers have to undergo a large number of tests and an unambiguous diagnosis of CMA has to be established. Moreover, once a diagnosis of CMA has been confirmed, the reimbursement of a clinical nutrition preparation is restricted to a review every three months or reimbursement is only for a limited period of time or there is a limit on the number of cans that would be reimbursed. Consequently, the analysis assumes no reimbursement of clinical nutrition preparations either in the private or public sector.

The budget impact analyses have assumed that all CMA sufferers would be detected and treated. However, it must be noted that in South Africa there are racial, gender and provincial disparities which are some of the primary challenges facing the Authorities and access to healthcare services, especially in the public sector, is inadequate[3]. Moreover, healthcare facilities are frequently physically inaccessible to the poor, to rural populations and other vulnerable members of society[3]. Therefore, the estimates of costs reported in this study may have been overestimated. Nevertheless, the results are indicative of a socio-economic burden that needs addressing.

This study is subject to a number of other limitations that could have confounded the findings. The intolerance rates to soy and eHF were based on a 1-year follow-up study among 1,000 infants with CMA who were managed in the community in the UK and not on South African infants. Healthcare resource use was not collected prospectively, but estimated from interviews with healthcare professionals with experience of managing CMA. Sixteen private paediatricians were interviewed and this should be a sufficiently large sample to accurately assess treatment patterns and healthcare resource use in the private sector. There is less confidence in the generalisability of information obtained from the small sample of public sector paediatricians. Nevertheless these were the best estimates available at the time. The model used resource use estimates for the ‘average patient’ and does not take into account such factors as age, gender, co-morbidities and other disease-related factors. Also, the estimated resource use may not be representative of the whole country since healthcare professionals who agreed to participate in the study were from major cities. However, the inherent variability and uncertainty surrounding the model inputs has been assessed by probabilistic sensitivity analyses.

Changes in the quality of life and improvements in general well-being of sufferers and their parents/carers as well as parents/carers’ preferences have been excluded in the model. Also excluded are changes in patients’ behaviour. Consequently, our study may have underestimated some of the impact of the treatment strategies that were evaluated. Also the period of the study was limited to one year following presentation to a paediatrician and does not consider the impact of CMA in subsequent years or management in the community by family practitioners. However, most children outgrow this form of allergy by the time they reach 2 years of age[5–7].

The present study also identified the current absence of guidelines on diagnosis and management of CMA in South Africa. Consequently, it varies greatly between different clinicians. The management of CMA also varies between the public and private sectors. Interestingly, the management of CMA in the public sector, possibly because of the small sample of interviewees, was found to be more consistent between clinicians compared to the disparities between clinicians in the private sector, where there was use of a wider variety of unorthodox diets to treat CMA, such as evaporated milk, boiled milk and goat's milk. Hence, there is an apparent need for a consensus on the management of CMA sufferers by clinicians, irrespective of the sector in which a patient is managed. Clinical guidelines for the management of CMA are being used or are in the process of being developed in some European countries and Australia[20–24]. However, before healthcare organisations adopt guidelines, they have an onus of responsibility to validate them and ensure they are clinically effective and cost-effective.

Conclusion

Within the aforementioned limitations, the model indicates that CMA imposes a substantial socio-economic burden in South Africa, especially on parents/carers of CMA sufferers. Any strategy that reduces this burden should potentially lead to higher compliance with clinicians’ recommendations, thereby improving health outcomes associated with treatment and should also release healthcare resources for alternative use within the system.

Transparency

Declaration of funding: This study was sponsored by SHS International Ltd, Liverpool, UK and Nutricia Middle East & Africa, Dubai, UAE. However, neither SHS International Ltd nor Nutricia had any input into the study design or data analysis. Neither did they review the results or have any input into writing this manuscript.

Declaration of financial/other relationships: E.S. and J.F.G. have disclosed that they have no relevant financial relationships.

Acknowledgements: The authors would like to thank the following for their contributions: Dr M. Ahmed, general paediatrician, Fourways Hospital, Johannesburg; Dr A. Bristow, general paediatrician, N1 City Hospital, Cape Town; Dr K. Eastrand, general paediatrician, Sunninghill Hospital, Johannesburg; Dr F. French, general paediatrician, Kloof Hospital, Pretoria; Dr M. Groenwald, paediatrician, 1 Military Hospital, Pretoria; Dr R. Heitner, general paediatrician, Morningside Mediclinic, Johannesburg; Dr L. Jedeikin, general paediatrician, Vincent Palotti Hospital, Cape Town; Dr S. Kling, general paediatrician and intensivist, Tygerberg Hospital, Cape Town; Dr J. Lund, general paediatrician, Constantiaberg Mediclinic, Cape Town; Prof C. Motala, paediatric allergist, Red Cross Children's Hospital, Cape Town; Dr S. Ponde, paediatric pulmonologist, Sunninghill Hospital, Johannesburg; Prof P. Potter, paediatric allergist, Lung Institute, Red Cross Children's Hospital and Groote Schuur Hospital, Cape Town; Dr N. Ranchod, general paediatrician, Olivedale Hospital, Johannesburg; Dr C. Schuman, general paediatrician, Femina Hospital, Pretoria; Dr P. Sinclair, general paediatrician, Vincent Palotti Hospital, Cape Town; Dr P. Snyman, general paediatrician, Wilgers Hospital, Pretoria; Dr H. Steinman, general paediatrician, Food and Allergy Consulting & Testing Services, Cape Town; Dr M. Zuckermann, paediatric gastroenterologist, Donald Gordon Medical Institute, Johannesburg; and Dr A. Wewege, general paediatrician, Constantiaberg Mediclinic, Cape Town.

Notes

*Neocate, SHS International Ltd, Liverpool, UK