Discussion
For Pakistan, screening by universal US or by clinical examination and targeted US are both cost-effective options to consider for implementing a screening programme for DDH at the hospital level. Although both options are considered cost-effective, they require a very different annual investment at the hospital level. For an estimated 30 000 annual births at IHHN, the additional costs for screening by clinical examination and targeted US would amount to PPP$30 000 while implementing a universal US screening programme would require an additional investment of PPP$300 000 per year. In order to make an informed decision between the available options, several arguments must be taken into consideration: willingness-to-pay thresholds, equity and available resources. The national UHC package for Pakistan allocates PPP$447 502 towards childhood developmental screening programmes for the entire country,28 an amount that is unable to cover the costs of our proposed screening programme at this time.
Conservatively treated DDH without complications incurs no loss of DALYs, as it has no further impact on life after treatment other than the elimination of the effects of DDH. As stated earlier, the DALYs lost due to the potential development of osteoarthritis in middle age are not included in this analysis. The DW for hip osteoarthritis is estimated at 0.16537 and is, therefore, smaller than the DW of untreated DDH. It is understood that the impact of osteoarthritis in these patients is already included in the DW for DDH together with other complications and impacts.
High-income countries usually apply a willingness-to-pay threshold of PPP$50 000 per DALY averted, a threshold easily met for all options in this study.38 Little information is available about willingness-to-pay thresholds in LMICs, but data from Thailand and Malaysia show much lower thresholds of around PPP$20 000 per DALY averted through medical treatment and as low as PPP$2 000 per DALY averted for preventive strategies.30 38 When applying these thresholds, only clinical examination and targeted US fall below the threshold.
One of the reasons the costs of the clinical examination and targeted US model are so low is that the model included only a 59% turnout for the targeted USs, lowering the costs of treatment and follow-up to a minimum (table 1). Although we do not have data on the characteristics of the people who do and who do not attend the second-stage screening through US, the available data from clubfoot programmes across LMICs show that poverty is the main cross-cutting factor for dropping out of treatment.39 Therefore, we can assume that the poorer children in society have a higher risk of failing to follow-up with two-step screening, making universal US screening the more equitable option. It is also important to note that the rate of surgery goes up in the scenarios with a high screening attrition rate. This counterintuitive effect is due to the assumption in the model that all children with a failed conservative treatment for DDH or post-treatment sequelae will access surgery. Given that the access to surgery for DDH in children not screened for DDH, was modelled at 10% only, the number of children requiring surgical care for failed DDH inefficient screening programmes turns out to be higher than the number of children access surgical care for DDH on the absence of screening. This phenomenon should be seen as an effect of increased access to care, and thus as an increased utilisation of services and not as an indication of the inefficiency of screening.
Lastly, resource utilisation differs greatly between both options. Universal US screening creates the largest need for follow-up by trained orthopaedic surgeons in all the scenarios. Orthopaedic surgeons themselves can be considered a ‘resource’, a ‘resource’ that may not be readily available in the Pakistani health system due to known health workforce shortage challenges. This leads to other constraints in the health system planning than mere financial constraints and constraints that may not be alleviated with financial means in the short run. An important strength of our model is that it allows the use of different outcome variables. If the health system experiences other larger constraints than finances, such as a lack of surgeons/nurses/US machine, these outcomes can be used instead of cost to determine the most effective and feasible scenario to be implemented.
The population density and health facility distribution where USs can be done should also be considered to ensure that the programme can cover the target population. It must be feasible for patients and caregivers to visit health facilities. In areas where visits are feasible but cumbersome or where ability to retain access to patients is difficult, full screening before discharge may be preferable. It is also vital to contemplate how different aspects of each programme would be received by society from multiple perspectives. For example, having a screening programme that is completed before discharge could result in a longer stay at the maternity ward may not be practical in communities where there are high rates of informal labour, as increased hospital stay may result in loss of wages.40
Executing a high-quality cost-effectiveness analysis proves difficult in the absence of the required data. In this study, many assumptions had to be made. For instance, the degree to which the Pavlik harness is visible or impedes care is not taken into consideration in our analysis, even though compliance with the Pavlik harness may well be lower than that with the brace used to correct clubfoot.41 42 Nevertheless, clubfoot is the closest available proxy measure. It is, therefore, not unreasonable that initiation and adherence rates for DDH treatment may be lower than for clubfoot, although this may be offset by the shorter treatment course that is necessary, thereby generating lower drop-out rates.
We believe our model, in essence the underlying decision tree, is context-independent and can be used by other countries and settings to assess the most cost-effective screening modality taking into consideration financial or contextual constraints and societal preferences. Our model covers the entire patient journey from screening, to accessing treatment, adhering to treatment and completing treatment with or without complications. This patient journey in itself is similar across settings, however, with very different probabilities for a patient to follow one path over another. As such, in order to use this model in other settings than Pakistan, local/regional data collection or identification of proxies is necessary to run the model. The extensiveness of the model also stimulates policy-makers or physicians to reflect on certain probabilities that may not automatically be perceived as important or impactful in their setting. But it is important to remember that no health system achieves 100% screening rates, health system accessibility, adherence or positive outcomes, and thus every branch of our model has some level of impact on the final level of cost-effectiveness irrespective of where the model is being applied.
The main limitation of this study is the lack of data in current literature on DDH screening attendance rates, compliance rates and drop-out rates in Pakistan and middle-income countries in general. The use of proxies such as clubfoot and deafness screening data made this analysis possible, however, it is impossible to assess how appropriate these are as proxies for DDH screening. Additionally, the wide variety of reported prevalence rates and sensitivity and specificity rates for different screening modalities generate a non-negligible level of uncertainty in the model. Last but not least, our model considers only costs incurred at the level of the hospital/health system and assumes that trained personnel and material are available. It does not include direct and indirect costs incurred by the patient and their families.
Our results indicate that universal US screening would be the ideal screening method for DDH at IHHN and may be used to inform other similar studies or programmes globally. It should be cautioned that while these options may be the best DDH screening programmes for the IHHN in Karachi at this time, it may not be the best scenario for other cities or provinces in Pakistan. Resource availability, geography, infrastructure, treatment capacity, health system values and societal factors must all be considered when determining which screening protocol may be the best fit in other areas. The presence of personnel and a sufficient budget alone are not enough to determine availability, personnel must be able to administer the tests with adequate skills without being constrained by administrative or licensing regulations.
