Health Insurance Ownership and Quality of Computed Tomography Requests: Experience from a Peripheral Referral Hospital in Cameroon

Background Health insurance ownership facilitates access and minimizes financial hardship after utilization of healthcare services such as computed tomography (CT). Understanding the rational utilization of CT by people with health insurance can help optimize the scheme and provide baseline information for a national universal health coverage program. Objective To assess the relationship between health insurance ownership and the appropriateness of requests for CT in a peripheral referral hospital in Cameroon. Methods A survey of CT users was conducted during which information on health insurance ownership was collected and the request forms for CT assessed for appropriateness using the American College of Radiologists (ACR) Appropriateness Criteria®. Results We consecutively enrolled 372 participants of which 167 (45%) were females. The median age (range) was 52 (18–92) years. Thirty-eight out of 370 participants reported having health insurance (10.3%; 95% confidence interval (CI): 7.2%–13.4%). Twenty-nine out of 352 CT scan requests (8.2%; 95% CI: 5.3–11.0) were judged to be “inappropriate.” The proportion of inappropriate scan requests was higher amongst people with health insurance compared to those without health insurance (18.4% vs. 7.0%; χ2 = 5.8; p=0.02). In the logistic regression analysis, health insurance ownership was associated to the appropriateness of CT requests in the univariate analysis only (OR = 0.33; 95% CI: 0.13–0.84; p=0.020). Conclusions Inappropriate requests for CT were low but nevertheless associated to health insurance ownership. The continuous sensitization and training of physicians would help minimize potential wasteful utilization of resources.


Introduction
e continuous development of advanced healthcare technology such as computed tomography (CT) has led to an increase in the cost of healthcare [1]. e utilization of multislice and extreme detector CT equipment with a very broad range of clinical applications has especially led to the price hike of CT procedures [1]. A high cost for healthcare services has the potential to limit access to care as people in need of the services may not be able to afford for them [2,3].
Affordability of multislice CT technology is therefore perceived to be a barrier to its utilization, especially in settings without any financial protection for health. However, the affordability barrier to CT utilization can be minimized by the implementation of financial protection schemes [3]. Health insurance ownership is therefore expected to protect CT users from impoverishment and financial hardship that could arise from its utilization [3].
In Cameroon, CT scans have been installed in recent years in some peripheral referral hospitals around the country so as to improve upon the geographical accessibility of CT technology. e improved geographical accessibility is expected to reduce the time to obtain CT and also minimize inconveniences linked to long-distance travel to other towns. ese peripheral health facilities which have been focusing on primary and secondary care now have to incorporate CT utilization during routine clinical practice with the potential for better health outcomes for patients. Nevertheless, CT, being a referral technology with a high radiation exposure, requires rational cost-effective utilization [1,[4][5][6]. According to a national survey, 96 to 98% of the general population do not have health insurance [7]. Whilst awaiting the implementation of a national universal health coverage program, available financial protection schemes for health include private insurance policies and community-based mutual and employer-provided insurance schemes.
Given the improved access to healthcare services conferred by health insurance ownership [8,9], it is important to explore the rational utilization of services such as CT by people who have subscribed to health insurance policies. e information generated can help restructure available private health insurance policies and also shape the planned implementation of a public national universal health coverage scheme. ere is paucity of information on the utilization of CT by people with health insurance. A pilot study reported an association between inappropriate CT requests and health insurance ownership in an urban sub-Saharan context [10]. Further studies to ascertain this relationship from different socioeconomic and geographical contexts are important so as to provide empirical evidence. In this study, we explore this relationship in a peripheral intermediatelevel referral hospital in Cameroon.

Materials and Methods
A cross-sectional survey of CTusers was conducted. is study was approved by the institutional ethics committee of the University of Yaoundé 1 (108/UY1/FMSB/VDRC/CSD). Administrative authorization for the study was also obtained from the South-West Regional Delegation for Public Health (R11/MINSANTE/SWR/RDPH/82/786). e study was conducted at the Medical Imaging Centre of Regional Hospital Limbe. Regional Hospital Limbe is an intermediate-level referral hospital in the Southwest Region of Cameroon with the capacity of 200 beds. Designated as a Category 3 referral health facility in the health system pyramid of Cameroon where the categories range from 1 (tertiary care) to 7 (primary care), its main role is to provide secondary care. e geographic location of this health facility is peripheral with respect to the country's political capital Yaoundé. In recent years, Regional Hospital Limbe has benefitted from the installation of a 16slice CT scanner in an effort by the government to improve on access to health technology.

Participants.
CT users aged 18 years and above who consented to participate in the study were consecutively enrolled between March 2018 and February 2019. Informed consent was written and was obtained either from the patient or the caregiver.
Sample size estimation was done using Cochran's method for surveys with the appropriateness of requests for CT (expressed as a categorical binary variable) being the primary outcome [11]. Given an alpha level of 0.05, a 5% error margin, and a population variance of 0.25, the calculated sample size was 385 participants. Informed by a pilot survey [10], we expected data collection to last for 12 months, period during which an expected 1614 CT scans would be performed as anticipated from the hospital records. e estimated sample size of 385 exceeded 5% of this population, and the Cochran correction formula was applied [11] to give a minimum return sample size of 310 participants. Anticipating a nonresponse rate of 20% [10], a total of 388 potential participants were invited to participate in the study, and enrolment ended when a sufficient sample size was attained. Figure 1 summarizes participant selection.

Data
Collection. Data were collected from March 2018 to February 2019 using standardized forms. ere was content validation of the included items on the data forms and pretesting through a pilot survey [10]. Data were collected on age, sex, educational achievement, marital status, occupation, socioeconomic status, health insurance ownership, and type of health insurance. Further information was gathered on the clinical indications for CT, anatomic region, and qualification of the referring healthcare provider. e appropriateness of CT requests was ascertained using the American College of Radiologists (ACR) Appropriateness Criteria ® [12], which are guidelines that have been developed in collaboration with specialists from various domains with an aim to help referring healthcare providers to request for the best available imaging modality for specific clinical scenarios. During this study, the request forms for CT were used to determine appropriateness, and the final categorization was consensual between two radiologists (JT and POZ). CT request forms without any clinical indication were excluded from the study, whilst those with insufficient information were ignored from the appropriateness analysis. A research assistant collected all the data under the supervision of the principal investigator (JT).

Data Analysis.
e data forms were transcribed onto a Microsoft Excel® spreadsheet and analyzed using Stata® 12 (StataCorp, Texas, USA). Continuous variables were summarized using the mean and standard deviation or median and range as appropriate. Categorical variables were summarized using frequencies, percentages, and 95% confidence intervals (CIs). Chi-squared tests were performed to compare proportions of inappropriate CT requests among categories of health insurance ownership. Univariate and multivariable logistic regression techniques were used to determine if any factors were associated with the appropriateness of requests for CT. For the multivariable modelling, covariates were entered as a block and included age, sex, educational achievement, socioeconomic status, health insurance ownership, and the qualification of the referring healthcare provider. Statistical tests were two tailed, and p values less than 0.05 were considered statistically significant. Model fit was assessed using the R 2 statistic. e data were presented using tables.

Participant Characteristics.
ree hundred and seventytwo participants were surveyed of which 167 (45%) were females. e median age (range) was 52  years. e demographic and socioprofessional characteristics of the participants are presented in Table 1. irty-eight out of 370 participants reported having health insurance (10.3%; 95% confidence interval (CI): 7.2%-13.4%). e reported health insurance types are presented in Table 2.

Appropriateness of CT Requests.
ree hundred and fifty-two CT requests could be categorized for appropriateness. Of these, 29 (8.2%; 95% CI � 5.3-11.0) were judged to be "inappropriate." Table 3 shows the categorization of CT appropriateness based on health insurance ownership. e proportion of inappropriate scan requests was higher amongst people with health insurance compared to those without health insurance (18.4% vs. 7.0%; χ 2 � 5.8; p � 0.02). Table 4 summarizes the relationship between health insurance ownership and CT request appropriateness.
In the logistic regression modelling, health insurance ownership was associated with the appropriateness of CT requests in the univariate analysis only (OR � 0.33; 95% CI: 0.13-0.84; p � 0.020). However, educational achievement beyond ordinary level was associated with the appropriateness of CT requests both in the univariate (OR � 0.42; 95% CI: 0.18-0.97; p � 0.043) and multivariable analysis (aOR � 0.35; 95% CI: 0.13-0.91; p � 0.032). e results of the logistic regression analysis are presented in Table 5.

Discussion
From the findings of this study, an estimated 10% of the respondents reportedly had health insurance ownership. is is higher than the 2-4% health insurance ownership of the general population following a national survey [7]. is difference could be explained by the fact that this survey focused on a subset of the population utilizing a particular health service and so not representative of the general population. e purchase of healthcare in Cameroon is essentially through direct out-of-pocket (OOP) payments. OOP payments lead to unequal access to care, whilst health insurance ownership minimizes access inequities [13][14][15][16][17]. Many governments around the world seek to improve access to care for its population through the implementation of a universal Excluded for incomplete information n = 5  health coverage (UHC) scheme [3].
e benefits of UHC cannot be overemphasized, which is the reason why the government of Cameroon has embarked upon the creation of such a scheme in the near future. e anticipation of a UHC scheme motivated the assessment of the rational utilization of CT by people with existing insurance schemes. e findings of this study support the fact that health insurance ownership confers better access to CT and also potentiates the likelihood of inappropriately using CT as health insurance ownership was independently associated to inappropriate requests for CT. It is likely that people with insurance may have some compulsion to access some health services even when these are not indicated just because they have subscribed to and contribute for such schemes. is may put pressure on healthcare providers to request for CT as they may also fear any potential litigation should they fail to request, and a serious condition is later detected.
Educational achievement beyond ordinary levels was also associated with inappropriate CT requests. Most of the insured participants had an insurance premium provided by the employer, and it is likely that a minimum level of education is required to get employed with a formal contract of employment. Furthermore, people with higher educational qualifications may be more demanding when accessing healthcare services with more pressure to obtain some services even when these might not be needed.
Inappropriate utilization of a health technology such as CT for whatever reason has to be given a serious consideration given that CT is associated to high exposure to ionizing radiation with the potential of radiation-induced cancers [4,5,[18][19][20][21]. e cost of CT is also considerable irrespective of whether payments are made OOP or through insurance [1], and wasteful utilization will strain the pooled resources that have been made available for the scheme [3]. If CT utilization by a smaller population with health insurance shows inappropriate utilization patterns, then it is likely that this effect might be multiplied with the extension of health insurance to a wider population.
Geographic accessibility is fundamental to accessing healthcare services [22,23]. e installation of CT in   Strategies to curb inappropriate utilization must therefore be envisaged while the government plans to implement a universal health coverage scheme. Firstly, given the absence of regulations as to who is qualified to request for imaging studies in the study setting, we suggest that CT requests should be approved only when prescribed by duly trained healthcare professionals. Also, consultations with radiologists should be encouraged if uncertain about the role of CT as more cost-effective imaging alternatives might be readily available [24]. Furthermore, continuous training and sensitization, the use of guidelines, and imaging decision support by referring physicians should be encouraged [25,26]. Finally, a tracking system for all requested CT studies with regular reviews to assess prescription trends and relevance will provide useful feedback on utilization.
CT utilization and health insurance ownership have not been given much attention in the medical literature. Bellolio et al. reported an increase in CT utilization among commercially insured patients but did not assess the appropriateness of CT utilization [27]. Becker et al. also reported inappropriate utilization of both CT and magnetic resonance imaging without any assessment with respect to health insurance ownership [28]. is paper focused on the specific relationship between the health insurance ownership and the appropriateness of CT utilization, providing opportunities for scaling up the use of CT by people with health insurance in resource-poor settings.

Limitations.
As limitations to this study, some CT scan requests could not be categorized for appropriateness due to insufficient clinical information on the forms. Also, reporting bias could have influenced the findings of this study.

Conclusions
e findings of this study support the fact that health insurance ownership, despite the proven benefits in minimizing access inequities, can be associated with inappropriate requests for CT in the study setting. is is perceived as an unintended consequence that can be checked by the continuous sensitization and training of physicians, providing them with other cost-effective alternatives to CT as appropriate and encouraging the use of guidelines when uncertain. ese measures, in our opinion, could help enhance the rational utilization of CT and reduce unnecessary exposure to ionizing radiation.
Data Availability e dataset on which the findings of this study are based is available at https://doi.org/10.17632/r4dmt58v3r.1.

Conflicts of Interest
e authors declare that they have no conflicts of interest.