Cost of illness of non-communicable diseases in private and public health facilities in Nigeria: a qualitative and quantitative approach

¹Department of Community Medicine, Federal Teaching Hospital, Ido-Ekiti, Nigeria, ²Department of Community Medicine, Afe Babalola University, Ado Ekiti, Nigeria, ³Department of Community Medicine, University of Medical Sciences (UNIMED) Ondo, Nigeria, ⁴Department of Community Medicine, Ekiti State University, Ado-Ekiti, Nigeria, ⁵Department of Paediatrics, College of Medicine & Health Sciences, Afe Babalola University, Ado-Ekiti, Nigeria, ⁶NHIS Unit, Ekiti State University Teaching Hospital, Ado-Ekiti, Nigeria

^&Corresponding author
Tope Michael Ipinnimo, Department of Community Medicine, Federal Teaching Hospital, Ido-Ekiti, Nigeria

Introduction    

Non-communicable diseases (NCDs) impose a huge economic burden on countries causing and perpetuating poverty while hindering economic development [1]. They are estimated to cause a global cumulative economic loss of $47 trillion between 2011 and 2030 [1,2]. The economic cost of NCDs from 2015 to 2030 in Costa Rica, Jamaica and Peru was estimated at US$81.96 billion, US$18.45 billion and US$477.33 billion respectively [3]. In Nigeria, the cost of NCDs in the year 2005 was estimated at US$400 million and had risen to about US$8 billion by 2015, a more than twentyfold rise in ten years [4]. The cost of illness, also known as the total cost of managing NCDs is the sum of all the costs components which include direct costs such as medications cost, indirect costs or productivity losses and the intangible cost [5].

The direct and indirect costs of NCDs in public and private health facilities (HF) have a detrimental effect on healthcare outcomes in addition to the serious economic impact on the patients accessing healthcare in these facilities [5,6]. In Bangladesh, the public HF patients pay about US$57.54 for a visit more than the private HF patients, most of which was accounted for by the patients´ income loss [5]. However, some other studies revealed that medical care in public HF was less costly than in private HF [6-8]. In Nigeria, the mean direct cost of chronic illness was estimated at US$137.72 per patient per month and the indirect cost was 24% loss of productivity [9]. Furthermore, the mean monthly direct cost of treating hypertension and diabetes were $104.89 and $159.76 per patient respectively [9].

The cost of illness (COI) could be affected by several factors. In the Netherlands, the main factors increasing healthcare costs were old age and disabling conditions resulting from NCDs [10]. In addition, male gender although being a female patient predicts a higher average direct cost [5], lower socio-economic status/low-income quintile of patients or their household [5,11-13], increased duration of disease [14-17], increased severity of disease [7,11,14-21], and the presence or increased number of co-morbidities were found to increase NCDs COI [16,17]. In Nigeria, male gender, having both hypertension and diabetes as well as an employer paying for care increased the cost of follow-up care by ₦1,132, ₦1,091 and ₦3,735 respectively [22]. COI studies measure the economic burden of disease [23] and translate the adverse effects of diseases into monetary terms, the universal language of decision-makers and the policy arena. Estimates from these studies could be used to provide the basis for policies as well as the economic framework for evaluating health programs and interventions [23,24]. There are presently very few studies on COI of NCDs in Nigeria and none of these studies compares the costs in private and public HF to the best of our knowledge. Previous studies on public HF have demonstrated a high COI [25]. However, little is known about the cost in the private HF, despite being the dominant provider of secondary healthcare in Nigeria [26]. Additionally, these HF differ in their administration and management as well as in their budget and mode of financing. This study, therefore, aims to estimate and compare the COI of NCDs and its predictors in private and public HF in Ado-Ekiti, Nigeria.

Methods     

Study design and setting: this was a health facility-based comparative cross-sectional study that used a mixed method of data collection (quantitative and qualitative methods). The study was carried out in Ado-Ekiti, one of the major cities in Southwest, Nigeria. As the state capital of Ekiti State, inhabitants of the city and people from the environs receive healthcare in HF within the city. Ado-Ekiti has a public university teaching hospital, a general hospital, 28 comprehensive, primary and basic health centers, as well as a private university teaching hospital and one hundred and seven other registered private HF. The majority of the population seek healthcare in these HFs while others self-medicate, visit a traditional healer, or seek spiritual healings [27].

Ethical consideration: ethical approval for the study was sought and obtained from the Ethics and Research Review Committee of Federal Teaching Hospital, Ido-Ekiti. Written consent for the interview was obtained from all the respondents.

Participants and study size: the study included registered adult hypertensive and/or diabetic patients (patients with hypertension only, patients with diabetes only and patients with both diseases) accessing healthcare in private and public HF in Ado-Ekiti, southwest Nigeria. Patients who have accessed healthcare from both public and private HF in the last 3 months were excluded. A minimum sample size of 180 for each group was calculated using the formula for calculating sample size when comparing two means [28]: a 95% confidence interval, 90% power, standard deviation of the COI of NCDs in the general population from a previous study [9], mean COI of NCDs in public [25] and private HF [6] as well as a 10% non-response was assumed. A two-stage sampling technique was used to select eligible respondents. Eleven public and 39 private HF were selected by balloting in stage 1 while systematic random sampling technique was used to select eligible patients in stage 2.

Variables, data sources/ measurement and bias: a semi-structured, interviewer-administered questionnaire adapted from previous studies [5,9] was used to collect data from consented eligible patients after seeing the doctor. The questionnaire has 3 sections. Section A elicited socio-demographic and economic characteristics of the patients, section B elicited patients´ health status, characteristics of NCDs and other determinants of the COI and the last section collect data on patients´ monthly healthcare expenditure (direct and indirect cost) on NCDs. Direct costs are expenditures for which payments were made, while indirect costs are those for which resources were lost [24]. Direct costs collected from respondents included the cost of consultation, registration, drugs, consumables, investigations, accommodation, transportation, food as well as other payments made directly during the period of seeking healthcare. The entries were verified from payment receipts and HF records. The indirect cost was calculated using the human capital method, self-reported income or wage loss during the period of illness and healthcare in the HF were estimated [5]. These estimations were done for the patients and their caregivers. The weights and height of respondents were measured using a well-calibrated Omron HN289 digital scale and a portable stadiometer respectively and these were then subsequently used to determine BMI.

Quantitative data analysis and statistical methods: data entry and analysis were done with IBM SPSS Statistics for Window, Version 22.0 (IBM Corp., Armonk, N.Y., USA). All entries were checked twice to ensure correctness. Distributions of patients´ characteristics were presented using frequency tables and percentages. Costs were presented as an average with standard deviation in the local currency, Naira (₦) as well as in US dollar (US$), using the Central Bank of Nigeria exchange rate at the mid-point of the data collection year (2019). The chi-square test was used to compare patients´ characteristics between the two groups. Mann-Whitney U test (Wilcoxon 2 sample rank test) was used to compare the COI of NCDs between private and public HF in addition to assessing the association between patients´ characteristics and COI. The correlation coefficient was used to determine the association between COI and continuous variables such as age and income. The multiple linear regression technique was carried out to assess the predictors of COI of NCDs. A P-value of = 0.05 was taken as statistical significance.

Qualitative data (key informant interview): five key informant interviews (KII) were conducted with 5 HF administrative heads or their representatives. The five administrative heads or their representatives were purposively selected. Three were selected from private while 2 were from public HF. Data were collected with a KII guide that was designed by the researchers. Information was collected on the COI of NCDs and the factors affecting it. Participants were interviewed on how they charge patients, their experience with clients´ payment patterns and factors affecting the COI of NCDs in their HF. Interviews were conducted after appropriate introduction and consent had been taken from the participants. Interviews were face-to-face and each session lasted about 30 minutes. Notes were taken on paper, voice recording of the discussion with a digital voice recorder was also done. Data from the KII was analyzed using the ATLAS.ti software version 8.0 (this was based on using thematic content analysis). The audio recordings were transcribed verbatim. Analytical framework and codes were developed. The transcripts from the KII sessions were analyzed by focusing on recurrent, dominant and divergent opinions. Themes and sub-themes were generated to address the study objectives and results were presented in prose.

Results     

Quantitative findings

A total of 360 patients were approached in both private and public HF, of which 348 responded, giving a 96.7% response rate. Of the respondents, 173(49.7%) were in private while 175(50.3%) were in public HF. The mean age ± standard deviation of the respondents was 59.58±10.84 years (private, 59.5 ± 10.5years; public, 59.7±11.2years) (p=0.910) with the majority of them using paying for their healthcare out-of-pocket (OOP) (private, 90.2%; public, 94.3%) (p=0.152). The number of days of exercise (p < 0.001), level of education (p=0.003) and income (p=0.003) showed significant differences among respondents in the two groups. Details of respondents´ characteristics are in Table 1. Table 2 shows the monthly COI of NCDs among respondents. There was a statistically significant difference in the COI in private and public HF (private, ₦15,750.38 [US$43.75]; public, ₦13,283.37 [US$36.90]) (p < 0.001). The direct COI (private, ₦14,189.31 [US$39.41]; public, ₦9,544.11[US$26.51]) (p<0.001) and the indirect COI (private, ₦1,561.07 [US$4.34]; public, ₦3,739.26 [US$10.39]) (p < 0.001) also showed statistically significant difference. The indirect COI accounted for less than one-tenth of the COI in private HF but for almost a third in public HF (private, 9.9%; public, 28.2%). The factors that were significantly associated with COI of NCDs in private HF were age (r=0.151, p=0.047), payment method (p < 0.001) and in public HF were income (r=0.427, p < 0.001), socioeconomic status (p < 0.001) among others as shown in Table 3. Table 4 shows the result of multiple linear regression analysis of patients´ characteristics and COI of NCDs. In private HF, age, income, payment method, type of NCD, having 2 or more complications, exercise and admission were predictors of COI of NCDs. COI increased by ₦122.96 among respondents in private HF for each additional year in age while holding other variables constant (B=122.96; 95%CI=44.71-201.20; p=0.002). For every additional ₦1.00 rise in the income of private HF respondents, there was a ₦0.05 increase in their COI while holding other variables constant (B=0.05; 95%CI=0.03-0.08; p<0.01). Patients using OOP incurred ₦9,146.24 more than those using national health insurance scheme (NHIS) (B=9146.24; 95%CI=6960.55-11331.93; p < 0.001). The predictors in public HF were income, length of diagnosis, socioeconomic status, alcohol and admission. COI increased by ₦460.37 among respondents in public HF for each additional year in the length of diagnosis while holding other variables constant (B=460.37; 95%CI=77.34-843.41; p=0.019). Those drinking = 14 units of alcohol in a week paid ₦6,577.25 more than respondents drinking <14 units of alcohol in a week (B=6577.25; 95%CI=263.69-12890.81; p=0.041). Respondents with one admission incurred ₦36,353.05 more than those without admission (B=36353.05; 95%CI=29055.51-43650.58; p<0.001). Other characteristics were not found to predict the COI of NCDs in both groups.

Qualitative (Key Informant Interview) findings

The participants included two heads and three representatives of the head of HF. By profession, four are medical doctors while one is a nurse. The main issues discussed were the COI of NCDs and factors affecting it.

Cost of illness of NCD

Participants from both private and public HF mentioned that the COI of NCDs depends on the type of care received -either in-patient or out-patient care. Also, they agreed that the cost of in-patient care is higher than that of a clinic visit. This can be seen in an extracted excerpt below; “In a month a patient with any of those diseases will spend an average of ₦5,000 to ₦10,000 for an out-patient visit and for the in-patient care we would be looking at something far more than that, say about ₦30,000 to ₦50,000” (Representative of the head of a private HF) While speaking on if they consider waiting time and income forgone as part of cost, participants from private HF said patients do not wait for so long in their clinics and thus, not important to consider waiting time and income forgone in calculating the COI of NCDs. However, public HF participants mentioned that their patients wait for a long period, hence waiting time is important and should be considered while calculating COI. They also said that practice of such may be difficult as seen in the subsequent quotation; “Yes, they should be considered, although it might be difficult in reality. Sometimes patients spend the whole day waiting in the hospital without any compensation at the end, which is not too good” (Representative of the head of a public HF) On the way they charge their patients, participants in private HF mentioned that charging should be individualized. private HF individualized their charges depending on what the patients can afford and how much they can pay. This is seen in the extracted excerpts below; “We have several plans, silver plan, gold plan and platinum plan depending on patients paying ability.” (Representative of the head of a private HF) Although participants from public HF believed that patient charges should be individualized, however, the participants stated that they do not individualize their charges. A public HF participant is quoted below; “On a normal day costing is expected to be individualized, however patients in this hospital pay the same amount for most things, for example, the cost of consultation is the same either you are being treated by a consultant, resident or a house officer.” (Representative of the head of a public HF) All the participants described their experience with clients´ payment after treatment as not forthcoming. They mentioned that, some of the patients owe money and are usually not ready to pay. Many ended up bargaining in order to reduce their bills. This is seen in an extracted excerpt below; “Majority of patients are not ready to pay for the services they receive and almost all patients bargain including the millionaires even if it is just to give them ?500.” (Head of a private HF) The participants from both groups also noted that this clients´ payment pattern is common among patients paying OOP and believe that health insurance can help to prevent it as seen in an extracted excerpt below; “most patient don´t want to pay out of pocket and this brings the issue of health insurance which is very important because it reduces the burden of payment and out of pocket expenses.” (Representative of the head of a public HF).

Factors affecting the cost of illness of NCDs

Participants from both groups mentioned factors that could increase the COI of NCDs as comorbidities and presence of complication(s), need for admission and OOP. In addition, private HF participants mentioned that patients wanting special or private care could increase costs. Some of the participants were quoted saying; “Somebody that is coming with diabetic foot ulcer will have to pay for the care required for the ulcer in addition to controlling his blood sugar.” (Representative of the head of a private HF) “NHIS patients incur less than patients paying outside insurance because they pay only about 10% of their bill” (Representative of the head of a public HF) Both groups noted that NHIS, a healthy lifestyle and seeking medical intervention early would reduce the COI of NCDs. Also, participants in public HF said that costs will be reduced by time-specific clinic schedules. Some of the participants were quoted saying; “If people live healthy lifestyle there will be less diagnosis of hypertension and diabetes and even their complications.” (Head of a private HF) “We can schedule clinic visits by hours instead of all the patients coming by 7am in the morning including someone that will be seen by 4pm. This long waiting time is adding to the overall cost, so if we can schedule individual patients per time, it would reduce the cost.” (Representative of the head of a public HF) Participants from public HF mentioned that patients should not have a say in the amount they pay for their care. However, the participants from the private HF believed that although the patients may not have a say in how much they pay, their expectation is what will determine how much to pay as seen in the quote below; “They cannot determine how much they pay; they can only determine the kind of care they want. Some patients want you to give them the best drug or may want additional tests, this is their expectation and it will affect what they pay.” (Head of a private HF).

Discussion     

The average monthly COI of NCDs was significantly higher among patients in private (₦15,750.38 [US$43.75]) than in public HF (₦13,283.37 [US$36.90]). A higher COI in private than in public HF has been reported by previous studies [6-8]. These along with the finding in the present study contrast with the finding of a study in Bangladesh where public hospital patients pay more for their healthcare than private hospital patients [5]. The higher COI in private HF in this study could be due to the way private HF are financed in Nigeria. The majority of these HF are owned by individuals who are profit-oriented entrepreneur that sees it as an investment and a source of livelihood. The salaries of staff, running costs and other expenditures by these HF are usually from money generated from the patients unlike in public HF where there is reliance on finances from the state for their expenditures and running. Internally generated revenue from patients in public HF is just an addendum to the funding coming from the Government [29]. The direct COI was significantly higher for private than public HF, and in both makes up a huge proportion of the overall COI. This direct cost is similar to the findings in previous studies in Lagos and Kwara States, Nigeria [30,31]. However, it is lower than what was reported in studies conducted in other parts of the country [9,25,32]. In extremely low income of patients, this huge direct expenditure could make them cut down consumption of basic needs to meet up with payment or even skip appointments as long as they feel well but report back with complications [32] leading to a need for higher healthcare resources.

The indirect COI was significantly higher in public than in private HF. This is similar to the result obtained from a study in Bangladesh [5]. These indirect costs are primarily income losses by patients and their caregivers as a result of accessing healthcare. The higher indirect cost among patients in public HF could be attributed to the long queue and waiting time in these HF. Furthermore, the qualitative aspect of the study buttressed the fact that public HF patients wait longer to access healthcare than private HF patients. This waiting time is however not considered while calculating COI in both groups, even though public HF managers feel it should be considered. It was then suggested that the large indirect COI in public HF could be minimized by reducing the long waiting time through time-specific clinic appointments for patients attending these HF.

In terms of the predictors of COI of NCDs, it was found that COI increases by ₦122.96 for each additional year in the age of patients in private HF. This is consistent with reports from previous studies conducted in different parts of the world [5,11,16,17]. Old-age patients are usually more vulnerable and negatively affected by fees and associated spending on health [5]. Increasing income and higher socioeconomic status were also found to predict higher COI of NCDs. These results contrast some previous literature [5,11-13]. However, socioeconomic status has been found in another study not to be significantly associated with COI [22]. The finding in the present study may be due to the financial capability of the rich and high-income earners to pay for the luxury of care they want and can afford, which the poor may not be able to afford. Evidence from the qualitative aspect of the study shows that client expectation and what they can afford determines how much they pay for their health care. OOP was a positive predictor of higher COI of NCDs in private HF but not in public HF. OOP was also mentioned as a factor that could increase the COI of NCDs in the qualitative aspect of the study and NHIS was seen as a way out of it. This result is consistent with the finding of studies conducted in Nigeria and Bangladesh [5,22]. The lack of significant association between OOP and the COI of NCDs in public HF may be due to the relatively large part contributed by indirect cost to the COI which is not affected by health insurance. NHIS in Nigeria captures the components of the direct COI such as essential drugs, laboratory services and consultations but did not take into consideration the indirect cost [33].

As regards the effect of clinical variables, longer length of diagnosis in public, having both hypertension and diabetes, having two or more complications in private as well as admission in both groups were predictors of higher COI of NCDs. This has further collaborated in the qualitative aspect of the study where comorbidity, presence of complication(s) and need for admission were mentioned as factors that could raise the COI of NCDs. These findings are consistent with that of other studies that found a higher COI among patients with complications [11,14-22], patients with longer duration of illness or diagnosis [5,14-17], patients with comorbidity [16,17,22] and those with hospital admission [34]. These clinical variables are a pointer to increase disease severity and a patient with more severe disease will consume more healthcare resources. Studies conducted in different parts of the world have also documented a direct association between higher COI and increase disease severity [11,14,15,17,20]. Modifiable risk factors of NCDs can increase the risk of developing the disease and sometimes the risk of complications and severity. The effect of modifiable risk factors has not been evaluated on the COI of NCDs in previous studies [5,14,34]. However, this study was able to show that NCD patients with controlled risk factors (exercising =3 days/week in private and consuming <14 units/week of alcohol in public) experienced a significant reduction in their COI. This suggests that lifestyle modification will not only reduce the morbidity and mortality due to NCDs but would also reduce the economic burden of the disease.

The strength of this study lies in the mixed data collection method - a qualitative and quantitative approach. This method allows COI to be explored from the professionals´ point of view and not only from that of the researcher. Data were also collected from both the supply and demand side of the healthcare system which gave a more encompassing view of the economic impact of NCDs. However, the study design being cross-sectional could limit the extent to which inferences can be drawn concerning the causal association among variables. Future research could employ a longitudinal study technique and as well look at the burden of intangible costs of NCDs which was not assessed in this study.

Conclusion     

This study has shown that the COI of NCDs was higher in private than in public HF, however, there was a higher indirect cost in public HF due to long waiting time. Predictors of COI of NCDs identified were income and admission in both groups. In addition, age, OOP, type of NCDs, having 2 or more complications and exercise were predictors in private HF while socioeconomic status, length of diagnosis and alcohol were identified in public HF. It is therefore recommended that public HF should employ policies that would minimize the waiting time of patients to reduce the huge indirect cost. Individuals should also engage in lifestyle modifications such as the consumption of a safe level of alcohol and adequate physical exercise. Other identified predictors should be appropriately modified to achieve a reduction of COI due to NCD.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Tope Michael Ipinnimo conceptualized and designed the study. Material preparation, data collection and analysis were performed by all authors. The first draft of the manuscript was written by Tope Michael Ipinnimo and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Acknowledgments     

We would like to appreciate everyone that participated in this study.

Tables     

Table 1: socio-demographic and economic characteristics of respondents

Table 2: monthly cost of illness of NCDs among respondents

Table 3: socio-demographic, economic and clinical characteristics associated with cost of illness of NCDs

Table 4: multiple linear regression relating cost of illness of NCDs to predictor variables

References