Ethics statement

Formal permission was taken from MIS, DGHS to retrieve data from GRS website. As the data was retrieved from the anonymous records of GRS, no ethical clearance was sought. Informed consent was not required as the data collection procedure did not involve connecting participants.

Data source, study population, and duration

This study was conducted based on data available from the GRS of the MOHFW, Bangladesh. GRS is an online platform that collects and stores messages sent usually through short messaging service (SMS) to specified phone numbers from hospitals around Bangladesh [9]. These SMS are sent by both the healthcare providers and seekers. The data stored can be freely accessed through the following link: https://app.dghs.gov.bd/complaintbox/. The Management Information System (MIS), Directorate General of Health Services (DGHS) is responsible for the management and maintenance of GRS. After taking formal permission from MIS, we retrieved data stored between 1^st January 2023–8^th August 2023 from the site. A total of 11604 messages were retrieved. After discarding five messages with missing data, we analyzed 11,599 messages for this study.

Data structure of GRS system

Data analysis

We retrieved and curated the data from GRS platform in Microsoft Excel 365. The closure or response time, i.e., the duration between message reception of the message and closure, was calculated by subtracting the date and time of the reception of the message from the date and time of closure and was converted into hours. Normality of the data was checked using a histogram. As the response time was highly skewed, we normalized the variable by taking the natural log, essentially assuming a log-normal distribution. Bivariate analysis was carried out using the Chi-square test for categorical variables. The Kruskal-Wallis test was used to compare the closure time data across groups of more than two categories. Post-hoc analysis was carried out using the Rank Sum test. The Kaplan-Meier Failure curve was generated to compare different types of closure across types of feedback. The log-rank test was done to statistically test the differences. Choropleth maps for the distribution of different types of feedback across districts were created using ArcGIS 10. The population data for each district was retrieved from the preliminary reports of the latest (2022) census by the Bangladesh Bureau of Statistics [11] for normalization of messages per million of the population of a district. Pie charts showing different categories and sub-types of messages, which are not available from the online tabular list, were retrieved from the graphical analysis page of the GRS dashboard [12]. All other data analysis was carried out in Stata Version 17.

Main findings of this study

Our study revealed that over a period of seven months, the GRS online portal received more than ten thousand messages. Of which, nearly two-thirds were complaints. The majority of complaints were related to the health workforce, health infrastructure, and service utilization. The absence of existing service providers, waste management problems, and the absence of basic amenities were the most prevalent complaints received. All of these complaints mirror the health system constraints revealed by many previous studies [3,4,13–17]. The availability of a competent and adequate health workforce is a critical challenge faced by the health system of Bangladesh [3,15,16]. Ensuring a continuous supply of efficient health workers and their timely presence in the offices in a background of widespread corruption (10) had been a major challenge for many years [15]. Although the Government of Bangladesh (GoB) has a well-formulated health workforce strategy to tackle these issues [15], an appreciation of the issues arising at service delivery points through exploration, recognition, and resolution is required to ensure the presence of health workers in the facilities. For instance, one study revealed that doctors’ attendance could be improved by various incentives, including functional securities, higher payments, a supportive environment, timely training, and promotion decisions [18]. However, disciplinary actions for absence are disliked by and considered a negative incentive by the doctors.

The absence of basic facilities, including essential medicines, diagnostic facilities, and logistic supply, has been a constraint in organizing non-communicable disease service delivery in primary health care [16]. Failure to handle the COVID-19 crisis at the initial stages was an eye-opener with respect to the persistent weakness and inadequacy of facilities for health delivery in developing countries like Bangladesh [13]. The feedback of the GRS system reveals the continuation of this problem in Bangladesh’s health sector.

Improper management of medical waste, as reflected in the public opinions recorded by GRS, is another crucial problem in Bangladesh. A huge amount of waste is generated each day by hospitals. However, these are poorly managed. Poor medical management activities in district-level health facilities are evident from a recent study by Sujon and colleagues [19]. Waste is not segregated at the source or after collection. Improper waste collection, handling, storage, transportation, and disposal practices lead to various health hazards and environmental risks [20,21]. Although waste management strategies are outlined in the environmental assessment and action plan of the Health, Population, and Nutrition Sector Development Program (HPNSDP) [22] of Bangladesh, proper implementation of those strategies would require strict regulation, supervision, monitoring, and evaluation.

Our study reveals that there is a geographic variation in complaints, both in terms of volumes and rates. Healthcare seekers vary in their use of the GRS system in different districts and divisions. Therefore, prioritization and selective allocation of resources would help in the resolution and improvement of deficiencies in health care delivery.

Our analysis also revealed that the median response time for each type of message was reasonable, and complaints were given priority during resolution. However, the overall percentage of complaints resolved by the MIS, MOHFW seems very low. One reason for the low resolution is that the mechanism of retrieval of the final state of the resolution when a complaint is forwarded to other ministries is still not in place. Most of the ‘major’ complaints are forwarded to other departments of the health ministry or other ministries. But given the definitions of resolution [10], it is difficult to know from the GRS system how many of the forwarded complaints were resolved. Another reason for such a low resolution could be systemic bottlenecks, like the volume of feedback reaching the point of capacity saturation. However, as the GRS system is still being developed, the MIS has planned to increase the proportion of complaints being resolved by the department itself incrementally over the years. Nevertheless, our analysis revealed that a portion of all ranks of complaints are resolved. We found that minor and moderate complaints were more likely to be resolved than major complaints, resolutions tend to be slow, with the probability of resolution increasing after 54 hours had passed. This indicates a relatively sluggish response from the system, which should be expedited through necessary steps like training and posting of dedicated human resources for the system. The types of resolution are not specified in the GRS system. Hence, how the resolution occurs is not clear from the dashboard. As the system remains open to the public and may work as an important medium for improvement in public relations, the system should have this information presented in its graphical dashboard to increase transparency and demonstrate accountability.

Assessment of the GRS system structure and recommendations

The GRS system structure stores and categorizes messages after their individual assessment by the assigned MIS officers. Currently, the process from complaint or compliment submission to resolution is not yet fully digitized. While individual complaints are digitally tracked and subsequently revalidated by the Central Management Information System (MIS) team to ensure proper verification, the system primarily authenticates the complainant and assesses the source of the grievance. Internal documentation is used to categorize complaints by severity, which are then forwarded to the designated focal point at the respective facility. However, the subsequent action taken in response to the complaint—along with documented feedback from the focal person, typically the head of the institution—is not yet integrated into the same digital workflow. Efforts are currently underway to incorporate this final stage, which will complete the feedback loop of the digital GRS. Nevertheless, this is a tedious process where continuous training of the designated officers would be required, and an adequate supply of trained manpower needs to be ensured. Moreover, the probability of misclassification and ranking of messages due to subjective bias cannot be ignored. The system would be more informative when the closure types are further stratified based on follow-up information from other departments. However, this would require a cross-linkage with respective departments to ensure automated encoding of closure status and types. We recommend automation in the process of message classification, notification to specific departments, and closure status, thereby increasing the efficiency of the feedback system. This is possible through tokenization, classification, and sentiment analysis of message contents. However, this would require some preliminary studies to decipher the varieties of transcripts in which the messages are sent and find a way to make a uniform lexicon of words specific to public opinions regarding health services in the context of Bangladesh.

Implications for practice

GRS is an innovative way to organize and manage a large amount of feedback coming from healthcare seekers as well as providers. This could be an excellent management solution to understand the needs, identify the gaps, and locate the shortcomings of the healthcare system of a country. It can reduce feedback management workload, inefficiency, and incoordination. However, enrichment of the system through the creation of a separate department with adequate manpower and resources would allow enhanced cooperation among ministries to provide comprehensive and sustainable management of the challenges faced by the healthcare system.

Implications for research

The GRS system of the Directorate General of Health Services of Bangladesh creates an opportunity for various health systems and implementation research. Future research can explore many interesting areas, including whether GRS-based information can be used for differential allocation of manpower and healthcare resources, as a monitoring tool for quality healthcare, and as an evaluation metric for healthcare satisfaction at the user level.

Implications for policy

The government of Bangladesh can take evidence-based strategies to enhance healthcare delivery using people’s feedback from GRS. As the system records and categorizes complaints from health seekers with its a map of origin, it can prioritize and allocate resources to solve the problems of specific locations. This system can be adopted by countries around the world, particularly those with a large population and limited resources.

Limitation and strengths

We had to limit our analysis of messages for the first eight months because of the unavailability of previous data on the freely accessible pages. An in-depth analysis of message contents was beyond the scope of this study. Nevertheless, this was one of the first studies using information from the GRS of the MOHFW of Bangladesh.