Assessing Client Satisfaction with Health Service Delivery under the National Health Insurance Scheme: The Case of Komfo Anokye Teaching Hospital

,


Introduction
Ghana has undertaken quite a number of reforms in its health sector, some of which have been implemented in the area of health fnancing. Te previous reforms resulted in reduced healthcare utilization and inequities, leading to a signifcant milestone in 2004 when Ghana joined a number of countries such as Rwanda, Nigeria, Tanzania, and Kenya in implementing a social health insurance known as the National Health Insurance Scheme (NHIS). Tis initiative sought to pool and share health risks across the populace and minimize direct payment at service delivery points [1].
Ghana's health insurance scheme, implemented over 15 years, has been touted as one of the best performers in terms of implementation in Sub-Saharan Africa (SSA) and one worthy of emulation by other countries in the subregion (United States Agency for International Development) [2]. Tis health insurance scheme remains a powerful health fnancing tool for achieving universal healthcare coverage (UHC) and fnancial risk protection against all healthcare costs [3].
Although the NHIS has made signifcant strides in healthcare utilization, this was achieved at the cost of increased demand for healthcare with a resultant strain on staf and resources, infrastructural improvement, and fnally client satisfaction [4]. Client satisfaction as an outcome of healthcare delivery and an indicator of healthcare quality is a signifcant determinant of enrolment in the scheme and, hence, its sustainability. Ghana's healthcare system has witnessed fuctuations in enrolment rates between 2010 and 2017, with the stagnation of active membership. In 2010, 2015, and 2017, the enrolment rates of the NHIS were 33% (8.2 million), 41% (11.3 million), and 35% (10.3 million), respectively, among the general population [5]. Client dissatisfaction with services rendered by healthcare providers has been purported as one of the reasons for the poor enrolment in the scheme [6].
Several empirical studies have been conducted on the NHIS in Ghana. Tese studies have sought to determine the trend in enrolment into the NHIS [5] and to determine factors infuencing quality healthcare delivery in implementing the NHIS [7], to name a few. A very limited number of studies have sought to address clients' satisfaction with health service delivery under the NHIS. Te few existing ones, such as Fenny et al. [4,8], focused primarily on primary health care and thus limited their study population to households.
Prior to this study, some studies ostensibly showed that the insured were dissatisfed with treatment received in public health facilities for reasons such as long waiting times and discrimination on the part of the staf [1]. Terefore, this study sought to assess clients' level of satisfaction with health service delivery at the tertiary health care level and then determine if any diference existed in the level of satisfaction with service delivery between the insured and uninsured.
Te desire for advancement in the health sector and the efectiveness of the NHIS remains a motivation for this study. Evidence from this study can inform policy directions to strengthen healthcare delivery at the facility level.

Study Design and Selection of Facility.
Te study adopted a cross-sectional study design. It was conducted at the outpatient department of the Komfo Anokye Teaching Hospital. Tis facility was purposively chosen because it is the major tertiary health facility serving the country's western, northern, and middle zones. Tus, it could give us a more representative sample of the general public.

Sampling and Sample
Size. Tis study focused on insured and uninsured clients in the outpatient department who visited the hospital during the data collection period and was part of a bigger study.
A sample size of 300 participants was considered for the study. In determining the sample size for this study, an original sample size of 385 was obtained using Cochran's formula for unknown populations (N � z 2 pq/e 2 , z at 95% confdence level � 1.96; sampling error, e � 5%; and maximum variability, p � 50%) [9]. Adopting the modifed Cochran's formula, the original sample size of 385 obtained was then adjusted using an estimated patient population of 1000 at a 5% signifcance level (p value <0.05). Tis yielded a patient sample size of 278, rounded of to 300.
Stratifed and purposive sampling techniques were then utilized in selecting the study participants. Patients' insurance status and demographic records were ascertained from the outpatient registers from fve conveniently selected consulting rooms at the general outpatient department. Tey were then stratifed based on their insurance status into insured and uninsured. Tis was followed by a purposive selection of 150 insured and 150 uninsured participants using inclusion and exclusion criteria.

Inclusion.
Adults between the ages of 18 and 70 years who were of sound mind and gave informed consent for participation were included in the study.

Exclusion.
Individuals excluded from the study included those who had been triaged for in-patient admission based on the severity of their health condition, patients at the OPD who were in pain or had some discomfort for which reason could not participate in the study, and lastly privately insured patients.

Data Collection and Study Instrument.
Data collection spanned a period of 2 months from August 2019 to September 2019. Study data were collected and managed using REDCap© electronic data capture tools hosted at Komfo Anokye Teaching Hospital [10,11]. Te mobile app version of REDCAP was installed on the phones of the data collectors and access, given to them for data capturing. Prior to the data collection, verbal consent was sought from prospective participants. Privacy and confdentiality were ensured. Study participants were interviewed in a private room where others could not overhear ongoing discussions. Confdentiality was maintained by eliminating participants' identity from the data collection tool and keeping responses under password protection. Te questionnaire consisted of two sections: the frst section covered the patient's demographic information while the second section assessed fve areas of client satisfaction, adopted from existing literature [4,12], mainly, communication, physical environment, responsiveness, service availability, and communication on medications.

Measurement of Variables.
Clients' satisfaction was assessed using fve constructs consisting of (i) communication (ii) physical environment, (iii) responsiveness, (iv) service availability, and (v) communication on medications. Variables assigned to each item were scored in a fve-point Likert scale ordinal response (1: very unsatisfactory, 2: unsatisfactory, 3: neutral, 4: satisfactory, and 5: very satisfactory). Te Likert scale enabled the researcher to rate and rank the patients' satisfaction levels.
To assess the levels of client satisfaction for each construct, the mean ratings of the various indicators under each construct were added to obtain the total quality index. Te total quality index was then divided by the number of variables to obtain the mean rating for each construct. Te following operational range defnitions were set from existing literature: low level of satisfaction (1-2.33), moderate level of satisfaction (2.34-3.67), and high-level satisfaction (>3.67) [13,14]. Te range was obtained using the formula: (highest Likert scale point − lowest Likert scale point/3). Te constant 3 was used because the means were classifed into three, i.e., low, moderate, or high. A value of 1.33 was obtained using the formula above (5 − 1/3). To defne the ranges for each score, we ensure a diference of 1.33, i.e., low score � 1-2.33, moderate score � 2.34-3.67, and high score >3.67 [14].

Data
Analysis. Data analysis was performed using the Statistical Package for Social Science (SPSS) version 20 software after the data set was cleaned, extracted onto an Excel sheet, and coded. A mean score ranking analysis with summary statistics (mean) and standard deviation were used to ascertain the average responses of clients on a Likert scale in answering the frst objective: to determine clients' level of satisfaction with health service delivery under Ghana's NHIS. Tis was followed by an independent sample t-test of equality of means to determine the diference in the level of satisfaction between the insured and uninsured, the second objective.

Data Validation and Quality.
Te questionnaire was initially reviewed by an expert, and data collectors were trained to ensure quality control. Tis was followed by a pilot test survey on a subset of the intended population. A total of 20 clients were interviewed in the pilot survey. Te results of the pilot study were excluded from the fnal study data. A factor and Cronbach's alpha (CA) analysis was performed to determine the validity of the variables under each construct and the reliability of the data set. Te questionnaire was revised based on errors observed and fne-tuned till a higher CA of 0.84 was achieved for the fnal study.

Ethics Statement.
Ethical approval was obtained from the Committee on Human Research and Publication Ethics (CHRPE) of the School of Medical Sciences (SMS), Kwame Nkrumah University of Science and Technology (KNUST), and permission to carry on the survey was obtained from the health facility. Informed consent was obtained from each participant verbally, and a detailed explanation of the purpose of the study was rendered. Participation in the study was entirely voluntary with no incentives given to the participants. Furthermore, details concerning ethical approval have been attached to this fle.

Sociodemographic Profle of Participants.
Te mean age of the study participants was about 40 years. Females formed the majority in both groups (53.3% of the uninsured and 50.6% of the insured). More than half of the insured (55.3%) were single while less than half of the uninsured were single (44.7%). A majority of the respondents had received basic (primary and junior high) education, i.e., 43% for the uninsured and 35.3% for the insured. However, more insured than uninsured had received tertiary education. Christians were predominant in both groups, i.e., 75.3% among the uninsured and 82.0% among the insured (see Table 1).

Results from the Mean Score Ranking Analysis of Client Satisfaction with Health Service Delivery under NHIS.
From Table 2 and using the operational range defnitions set by and adopted from the existing literature (low level of satisfaction (1-2.33), moderate level of satisfaction (2.34-3.67), and high-level satisfaction (>3.67)) [13,14], it can be deduced that insured and uninsured clients were satisfed with "Physical Environment" with mean scores of 3.8 and 3.69, respectively. Tese mean score values were above 3.67, indicating a high level of satisfaction. Communication, service availability, responsiveness, and communication on medications constructs showed a moderate level of satisfaction for both insured and uninsured with mean scores between 2.34 and 3.67.
Te least rated factor by clients with health delivery under NHIS was "Responsiveness to service delivery," with a mean of 2.50 and 2.51 for the insured and uninsured, respectively.

Independent Sample t-Test on the means of the Insured and
Uninsured. In probing further to reveal whether a diference existed between the two groups, Table 3 shows that the mean satisfaction level of the insured clients in relation to communication of 3.51 was statistically diferent from the uninsured client's mean score of 3.15. Tus, the insured and the uninsured clients were statistically diferent in their satisfaction with communication (πdif = 0.36; p < 0.05). Similarly, the mean satisfaction level of the insured clients with service availability of 3.39 was statistically diferent from the uninsured client's mean score of 3.06 (πdif = 0.33; p < 0.05). However, with the high-level of satisfaction among the insured and uninsured with "Physical environment," there was no statistically signifcant diference observed (πdif = 0.11; p > 0.05). Te insured and uninsured groups of clients were generally satisfed with the hospital's physical environment. Further, the mean satisfaction level of the insured clients in relation to responsiveness in service delivery of 2.59 was not statistically diferent from the uninsured clients' mean of 2.51 (πdif = 0.08; p > 0.05). Lastly, the mean satisfaction level of the insured clients regarding communication on medication of 2.75 was not statistically diferent from the uninsured client's mean of 2.71 (πdif = 0.04; p > 0.05).

Discussion
Te study was conducted to assess clients' satisfaction level with healthcare delivery under the NHIS and to determine if any diference existed in the level of satisfaction between the insured and uninsured. Te discussion has been structured as follows: client satisfaction with healthcare delivery under the NHIS and the diference in satisfaction between insured and uninsured clients.

Client Satisfaction with Healthcare Delivery under the NHIS.
Findings from the study showed a higher level of satisfaction with the physical environment wherein healthcare is delivered. A mean score of 3.80 and 3.69 at 95% confdence level and p < 0.05 for insured and uninsured clients, respectively, demonstrated this. Tis fnding is important because the environment within which healthcare is received has a signifcant implication on the health and wellbeing of the people receiving care and on treatment  outcomes [15,16]. Tus, regardless of the insurance status, all patients are entitled to healthcare in a safe and serene environment. Financial risk protection should, therefore, not be provided at the expense of the health and safety of clients. In a somewhat similar study conducted in an NHIS clinic in a tertiary facility in Nigeria, insured clients expressed a higher level of satisfaction with the cleanliness and sanitation of the NHIS clinic (π = 3.25; s.d. = 0.78) compared to that of the general hospital environment (π = 2.43; s.d = 0.86) [17]. Tis study did not, however, assess the perception of those who did not attend the NHIS clinic.
Other client satisfaction variables, namely, communication (πins � 3.51; πunins � 3.15), service availability (πins � 3.39; πunins � 3.06), communication on medication (πins � 2.75; πunins � 2.71), and responsiveness (πins � 2.59; π unins � 2.51) showed a moderate level of satisfaction by both insured and uninsured clients. In a similar study by Fenny et al. [4] with a sample size twice as much as this study, more than half of the insured and uninsured clients expressed a moderate to high level of satisfaction with waiting time (52.6% insured satisfed; 51.8% uninsured satisfed), friendliness of the staf (57.5% insured satisfed; 54.1% uninsured satisfed), and laboratory services (58.2% insured satisfed; 63.6% uninsured satisfed). On the contrary, the majority of the insured (69.4%) expressed dissatisfaction with services at the dispensary. It can be inferred that overall, healthcare provider services under the NHIS have proven favorable to clients whether insured or uninsured. Although these fndings are contrary to those of Dalinjong and Laar in 2012, regional and healthcare provider disparities could account for this contradictory fnding.

Te Diference in Satisfaction between Insured and Uninsured Clients.
Tough scarcely the focus, the diference in satisfaction levels between insured and uninsured clients is a very important factor in increasing enrolment in the NHIS and has an implication on policy direction for the sustainability of the NHIS. It can be reasoned that a signifcant diference in the level of satisfaction between the two groups, with the insured being more satisfed than the uninsured can be an incentive for enrolment. Tis is because the propensity to enroll on the scheme is infuenced by clients' perception of the scheme [8]. Tis perception is partly a result of their experiences with the health services provided.
Tis study unraveled a signifcant diference in the mean score for communication (πdif � 0.36; p value <0.05) and service availability (πdif � 0.33; p value <0.05) between the insured and uninsured. Te insured were more satisfed with how providers related to them and how readily available services were. Tis can be attributed to the fact that health providers are likely to experience less stress and frustration providing services to the insured who are entitled to a wide range of services compared to the uninsured who are not and who may not be able to pay out of pocket for these services. Tis eventually afects how health providers communicate and relate with these clients.
Constructs such as the physical environment (πdif =0.11; p value =0.057), responsiveness (πdif = 0.08; p value =0.24), and communication on medications (πdif = 0.04; p value =0.614) showed no signifcant diference in mean scores between the two groups. In terms of the physical environment, both insured and uninsured receive care under the same sanitary conditions; hence, we do not expect to observe a signifcant diference in the mean scores.
Te observation of no signifcant diference in the time spent at the waiting area and obtaining medications was contrary to fndings from a study conducted in Bolgatanga in Ghana by Dalinjong and Laar in 2012, where a signifcant diference was observed (p value =0.003). Factors such as the shortage of health professionals, lengthy and cumbersome procedures, lack of appointment schedules, and increased demand for specialized healthcare nonetheless accounted for the long waiting time experienced by both the insured and uninsured [8,12].
Lastly, the study reveals a similar experience by both the insured and uninsured at the dispensary regarding communication on drug use and side efects. Similar to this study, fndings from the study by Fenny et al. in 2014 revealed no signifcant diference between client satisfaction ratings by insured and uninsured at a 5% signifcance level. Te poor rating of this construct by both groups, could be attributed to the high OPD attendance, which translates into a high demand for medications juxtaposed with inadequate drug dispensers and pharmacists. Since communication on medication afects Advances in Public Health health outcomes, the requisite steps need to be taken to ensure that the OPD attendance is well managed to facilitate an efective dispenser-patient time.

Limitations of the Study
External validity was low because the study was conducted only on outpatients and in Komfo Anokye Teaching Hospital. Considering the use of non-probability (purposive) sampling technique in this study, generalization must be done with caution as in-patients' satisfaction with care may difer markedly. Arguably, the OPD is the gateway to most of the hospital's services and thus remains an important area to sample respondents. Furthermore, client satisfaction is more subjective than an objective measure of quality and could be infuenced by sociodemographic factors like sex, marital status, age, and culture. Tese variables were, however, not adjusted for, in the study. Finally, a cross-sectional study of this nature only gives a snapshot of the event without a basis to establish causal relationships.

. Conclusion
Patient satisfaction is an essential indicator for measuring quality health care. Tough some studies previously projected client dissatisfaction with healthcare delivery under the NHIS, fndings from this study have proven to be contradictory to these perceptions. However, the disparity in clients' perceived satisfaction across the various studies could be attributed to structural and procedural diferences across diferent health facilities and individual factors such as sex, age, marital status, and level of education.
In spite of the generally acceptable level of satisfaction observed for all the client satisfaction constructs, the study identifes the need for healthcare managers to prioritize and channel more efort and resources towards improving responsiveness in the delivery of care. It advocates policy initiatives that seek to address this gap. Furthermore, it challenges health service providers to be modest in communicating with clients regardless of their socioeconomic status or insurance status. Health managers in countries with similar healthcare systems can, in the same vein, utilize these fndings in prioritizing and making decisions regarding quality improvement.
A further study is recommended to focus on in-patients' satisfaction with healthcare within the context of the implementation of the NHIS.

Data Availability
Te secondary data used to support the fndings of this study are included within the article. Te feld data collected and used in this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.