Are quality medicines affordable? Evidence from a large survey of medicine price and quality in Indonesia

Abstract

Background

Since Indonesia implemented one of the world’s largest single-payer health insurance schemes in 2014, the price of many common medicines has fallen dramatically. Industry groups warn unsustainably low prices threaten quality, while the government says medicines remain overpriced. We investigate the relationship between the price and quality of essential medicines and the affordability of medicines paid for out of pocket.

Methods

We bought over 1000 samples of five common prescription medicines—allopurinol, amlodipine, amoxicillin, cefixime and dexamethasone—online and from randomly selected pharmacies and health facilities in four regions across Indonesia. We recorded the price paid and tested samples for quality using high-performance liquid chromatography. We compared prices with the median and lowest prices for each medicine, tested for correlation between quality and price, and calculated affordability compared with the district minimum wage.

Results

Medicines available in the public procurement system were less likely to fail quality testing than other brands/varieties (4.2% vs 8.3%) but the difference was not statistically significant (p=0.086). There was no other relationship between quality and price, or branded status. Branded generic medicines sold at a large variety of price points, from 0.3 to 18.6 times the median price for the medicine and dose (IQR: 0.9–5.0, median 1.4). Unbranded generics traded in a narrower range (range 0.1–2.6; IQR 0.6–1.0, median 0.8). Medicines were most expensive in the region with the lowest wages, but even there, medicines selling at the 25th percentile of available prices cost a maximum of 0.7% of 1 day’s wage for a course.

Conclusion

In every study district, we found that Indonesian patients working at the minimum wage could access affordable, quality-assured versions of all studied essential medicines. More expensive brands were also widely available, but there was no relationship between price and quality.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• We tested over 1000 samples of five common medicines collected from dispensing points across Indonesia and showed that widely available, cheap medicines met the same standards as more expensive equivalents.

• Despite the universal availability of quality-assured free and affordable medicines in all study sites, our mystery shoppers were frequently charged many multiples of the lowest prices for branded versions of the products, including by hospitals.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• We recommend that medicine pricing policies address the incentive structures that drive health providers to encourage patients to pay relatively high prices for branded products.

• Procurement policies promoting the use of low-cost, unbranded generics should be accompanied by efforts to raise public confidence in the quality of free and affordable medicines.

Introduction

In Sustainable Development Goal 3.8, the nations of the world aspire to ensure ‘access to safe, effective, quality and affordable essential medicines and vaccines for all’.¹ In fact, many nations are still far from meeting that goal. According to the United Nations, the proportion of households spending over 10% of their budget on health is increasing, numbering one billion people in 2019.¹ WHO medicine pricing policy guidelines recommend a number of policies intended to increase the affordability of medicines, including promoting the use of unbranded generic medicines.² While global discussions about fair pricing for medicines are careful to consider the cost of quality assurance,^{3 4} recent, well-publicised cases of cost-cutting in production facilities for lower-priced medicines have raised concerns about the quality of these medicines, feeding into a narrative, long promoted by manufacturers of more expensive branded medicines, that equates affordable prices with questionable quality.^{5 6}

The Indonesian health system

The relationship between medicine pricing and quality has been a topic of active debate in Indonesia, the world’s fourth most populous nation, and home to one of the world’s largest single-payer health insurance schemes. The mandatory scheme—Jaminan Kesehatan Nasional or JKN—was introduced in 2014. At the end of 2023, it was reported that over 95% of Indonesians are registered as JKN participants.⁷

Registered users are entitled to free care at all levels for most health conditions. All medicines on the national formulary should also be provided to insured patients for free. Employers or individuals pay a monthly premium; the state covers premia for the poor.

Between 2014 and 2022, most medicines for JKN patients were procured through an online platform known as e-Catalogue. The Ministry of Health (MoH) estimated demand volumes and set a ceiling price for each medicine on the procurement list. With exceptions for products with three or fewer suppliers, any company with a valid market authorisation for an unbranded generic version of a listed product could bid at or below that price, provided they undertook to supply demand (at least up to the MoH’s estimate). The lowest bidder in each province won the contract to supply the medicine to that province for one (later two) years; winning prices were published online.8,10 Between 2014 and 2022, the price of most commonly used medicines fell dramatically.

Pharmaceutical industry actors reported that the ceiling prices imposed by the government for publicly procured medicines were irrationally low. When the cost of active ingredients, labour or other inputs rose, slim margins were eroded. Manufacturers were reluctant to supply at a loss, leading to stock-outs.11,14

In some ways, those stockouts worked in favour of healthcare providers. Before JKN, most patients would pay health facilities out of pocket for any medicines received, contributing to profitability (and in rarer cases sustainability) of service provision. Now, medicines supplied to public primary health centres by the district health bureau are paid for out of general taxation. In other settings, JKN reimburses dispensing health facilities for medicines for chronic conditions, paying the e-catalogue price plus a small margin for handling. All other medicines are expected to be covered by the hospital or other health provider from claims based on a patient’s diagnosis.¹⁵ Health facilities are thus incentivised to encourage patients to pay out of pocket for medicines or brands not covered by JKN.^{14 16} While the maximum retail price of many off-patent medicines marketed by domestic manufacturers using brand names (which we refer to as branded generics) fell steeply between 2014 and 2020,¹⁷ the government continues to focus attention on prices charged to patients paying out-of-pocket, pointing out that some brands sell for 3 to 5 as much as in neighbouring Malaysia.¹⁸ This rhetoric has fuelled parliamentary concerns that medicines are over-priced in Indonesia.^{19 20}

Insured patients are potentially exposed to these prices because, even when free medicines are in stock, many choose to pay out of pocket to avoid paperwork and long waiting times at public services, or because they distrust the quality of the free medicines provided.21,24 In a nationally representative survey of 345 000 households conducted in 2023, 80% of people who reported any symptoms of illness in the preceding month said they had self-treated.²⁵ Just 14.4% of the estimated US$3.6 billion spent on medicines in Indonesia in 2022 was channelled through public procurement.^{26 27} The actual amounts charged to patients in different healthcare settings have, to our knowledge, only been recorded in one smaller earlier study in East Java. This found that Indonesian patients were charged widely varying prices for different branded and unbranded versions of cardiovascular and anti-diabetic medicines, but did not consider affordability. All samples passed quality tests; thus, no relationship was found between their price and their pharmacopeial quality. The authors of that study, which collected 204 samples, suggested that pharmaceutical companies may be protecting investment in quality assurance for very low-priced unbranded generic products by selling a branded version of the same product at a higher price. Other studies that have detailed price and quality data for essential medicines are rare, and none has also calculated affordability.

In this study, we investigated the affordability of out-of-pocket medicines to patients earning the minimum wage in different areas of Indonesia. We also tested the hypothesis that cross-subsidisation might contribute to quality assurance by looking for differences in quality between low-priced unbranded generics that had no apparent potential for cross-subsidisation and those sold by companies that also sold higher-priced brands.

Methods

This study uses data collected as part of STARmeds, an investigation of medicine quality in Indonesia, combined with open-access medicine registration data published by the Indonesian national medicine regulator, Badan Pengawasan Obat dan Makanan (BPOM), and statistical data published by Statistics Indonesia.^{28 29}

Study settings and data collection

Detailed methods for the STARmeds study, describing the selection of medicines and sampling areas, construction of the sample frame, and fieldwork and medicines testing, along with all of the data collection forms and laboratory protocols, are published elsewhere.³⁰

Briefly, we purposively chose seven sampling areas to reflect Indonesia’s geographic, economic and demographic diversity. These were the Greater Jakarta region, and a large city and a more remote rural district in each of Western, Central and Eastern Indonesia (respectively Medan/Labuhan Batu; Surabaya/Malang; Kupang/Timor Tengah Selatan). Within each area, we listed, verified and randomly selected outlets: pharmacies; over-the-counter medicine shops; hospitals; health centres; and private doctors and midwives.

We sampled five medicines—allopurinol, amlodipine, amoxicillin, cefixime and dexamethasone—in tablet or capsule form as available. We also collected dry syrup amoxicillin. We aimed to buy medicines at a variety of price points; each retail outlet was visited by two different mystery shoppers, each with a personalised sample frame targeting broadly cheaper or more expensive medicines. They signalled their desired price point using phrases such as ‘Do you have anything a bit more affordable?’ or ‘Is this the best brand you have?’. In most cases, shoppers bought two different medicines at each pharmacy, in clinically plausible combinations. They carried prescriptions, but were instructed not to use them unless the pharmacist specifically asked for a prescription.

We also sampled from all registered apps offering instant delivery of medicines using geo-positioning, as well as from online stores of registered physical pharmacies. (The STARmeds study also conducted purposive sampling of medicines sold by unauthorised sellers online. These samples (n=186) were difficult to find; they are not considered in this paper because they do not reflect products easily accessible to most patients.)

At healthcare facilities, we sampled overtly, buying all available versions of the target medicine at the prices paid by patients.

Data related to price and product details were entered into a form preloaded onto the shoppers’ mobile phones using the open-source KoboCollect software.³¹

Quality testing

For budgetary reasons, we excluded some samples from testing, using a pre-determined triage system (see study repository). For all tested samples, we identified the active ingredient and performed assays (quantifying the active ingredient as a percentage of the labelled dose). All tablets or capsules were tested for dissolution (the percent of labelled active ingredient dissolving in a given time interval). For medicines with doses below 25 mg (amlodipine and dexamethasone), we also tested for uniformity of content of tablets in the same sample. A sample is a single active ingredient, brand, formulation and dose collected at a single time and place. PT Equilab International, an ISO/IEC 17025: 2017-certified private laboratory in Jakarta, performed the testing between April and November 2022, with real-time supervision from the STARmeds chemist.

Samples were tested using the United States Pharmacopeia reference standards, in accordance with the USP 43 NF 38 monographs.³² There was no USP monograph for cefixime capsules, so we followed Farmakope Indonesia VI’s adoption of Chinese Pharmacopeia standards.³³ Assay testing was done by high-performance liquid chromatography (HPLC -UV Waters, Aliance 2695 with UV Detector 2489), using columns as specified by USP, while dissolution was by Spectrophotometer-UV/VIS (Shimadzu UV-1800) with the exception of amoxicillin tablets and dexamethasone, where dissolution was tested by HPLC. Full laboratory validation and testing protocols are provided in the study repository.³⁰

A sample was considered out of specification if it failed any quality test it was subjected to. The field data form and the laboratory data were merged on barcode using Stata 18.0 software. Stata 18.0 was also used for reproducible cleaning, coding and to generate descriptive statistics and graphs. The dataset and code for analysis are provided in the STARmeds repository.³⁴

In analyses involving quality, we included 1088 samples for which we had laboratory data. Because the quality of the same brand is unlikely to vary by dose, these include 57 samples of doses that were not specifically targeted in our sampling frame (eg, in quality analyses, we included 22 samples of amlodipine 10 mg, as well as the 191 samples of the targeted 5 mg dose of amlodipine).

Price variation

When investigating the variation in prices paid by patients, we included 1003 samples. These comprised the seven combinations of active ingredients specifically targeted in data collection, whether or not we tested them for quality; we excluded non-targeted doses because small numbers did not give a fair picture of price variation across that product. We also excluded medicines provided free through the public health system.

In order to report comparable measures across medicines with different base price points, we created two standardised ratios. The first ratio was price-to-median, calculated in order to minimise distortions caused by outliers. Here, we calculated the ratio of each individual sample’s price per smallest counting unit to the median price of all samples of the same medicine (active ingredient, dose and formulation, regardless of brand). The smallest counting unit was a single tablet or capsule for all medicines except amoxicillin dry syrup, which was priced in 5 mL units. The second ratio was price-to-minimum, which provides an at-a-glance view of the total range of prices for a product. Here, we compared each sample price to the lowest price paid for any sample of the same medicine.

When comparing prices charged by different outlets for the identical product (the same market authorisation holder, dose, formulation and brand), we calculated the ratio of the individual sample to the lowest price paid for that specific brand.

In analyses involving quality, we included samples provided for free to insured patients. We priced these at a ‘retail equivalent’ price of the public procurement price plus a margin of 28%, the amount permitted by pharmaceutical regulations in public facilities.

Affordability

We adapted methodology developed by WHO/HAI to estimate medicine affordability.³⁵ We compared the price of each medicine with the minimum wage in the district of sampling as follows:

• Considering that low-income patients are generally able to ask for and obtain cheaper brands, we set the price at the 25th percentile of all prices paid for that active ingredient, formulation and dose in the specific sampling area.

• We calculated the number of units needed for a recommended course of treatment of each medicine per month, using defined daily doses recommended by the WHO and duration of treatment recommended by the Indonesian National Medicines Formulary 2021 for the most commonly indicated condition for each dose and formulation;

• We multiplied the unit cost of each medicine by the estimated number of units needed for a full course of treatment for an acute condition, or a 30.4 day’s supply of medicine for chronic diseases.

Medicines were considered affordable if the required course cost 1 day’s wage or less. Because minimum wages are set monthly, covering holidays and days off, we calculated 1 day’s wages as the monthly minimum wage divided by 30.4.

We also compared the cost of a course of treatment with the cost of a month’s supply of other commonly consumed commodities: the staple food (rice), a common convenience product (instant noodles) and a non-essential item (cigarettes). We used the national average per capita expenditure on these items from data reported in the 2022 round of Indonesia Social and Economic Survey (SUSENAS), a regular survey of over 345 000 households, which is representative at the district level.³⁶ Statistics Indonesia calculates per capita expenditure as household expenditure on the commodity divided by the total number of people in the household, regardless of how many people in the household consumed the commodity in question.

Potential for cross-subsidisation and quality effects

To identify the potential for cross-subsidies in product manufacturing or sales, we downloaded BPOM’s database of medicines registered in the Indonesian market, for all versions of the study medicines. We ordered the list by active ingredient, formulation type, market authorisation holder and branded status. Any unbranded generic medicine that had a branded equivalent registered to the same market authorisation holder was considered potentially cross-subsidised. Other unbranded generics were labelled ‘unbranded only’.

We compared the prevalence of out-of-specification samples between ‘unbranded only’ generics and unbranded generics that were paired by the market authorisation holder with a brand, testing for the significance of difference with a Pearson’s χ2 test.

Ethics

This substudy was approved by the Health Research Ethics Committee, Faculty of Medicine, University of Indonesia (protocol number 20-09-0999). Ethical approvals for the larger STARmeds study are posted in the study repository, where we also describe reporting of suspect products and other issues with ethical implications. Patients were not involved in the design or communication of this study.

Results

Description of samples

The distribution of the 1003 samples included in analyses of price variation and affordability, by medicine, dose, formulation, geographical area of sampling and source is provided in online supplemental table 1. Online supplemental table 2 provides the same data for the 1088 samples included in analyses involving quality.

Medicine price variations

We compared prices paid by different patients across all versions of the same medicines, regardless of brand (figure 1). We further compared prices charged by different sellers for the identical product (i.e, the same brand; figure 2).

Figure 1. Ratio of sample price to median price for the same medicine, dose and formulation; by active ingredient.

Figure 2. Variation in prices paid for the identical product at different outlets; showing 20 highly variable products.

Figure 1 compares the price of each individual sample with the median for all samples of that medicine. The cheapest products cost as little as one tenth of the median price, with the most expensive selling at over ten times the median price for most of the medicines and doses in the study. Although branded medicines were more expensive in general, all study medicines included unbranded generics priced higher than the lowest-priced branded product.

Online supplemental figure 1 shows the same data on a log scale, which provides greater clarity of detail at the lower ranges. More details of minimum prices and relative ratios are provided in table 1.

Table 1. Details of sample prices by medicine, dose and formulation.

Medicines	Number of samples	Lowest price(Rp)	Medianprice(Rp)	Ratio of price paid to median for all samples of this type			Ratio of price paid to lowest for any sample of this type
				Median	Range	IQR	Median	Range	IQR
Allopurinol 100 mg	190	104.4	400	1	0.3–10.6	0.8–2.5	3.8	1–40.7	2.9–9.6
Allopurinol 300 mg	57	240	800	1	0.3–11.2	0.7–1.3	3.3	1–37.3	2.4–4.2
Amlodipine 5 mg	190	84	782.5	1	0.1–18.6	0.6–1.8	9.3	1–173.2	5.4–16.7
Amoxicillin 500 mg	172	292	800	1	0.1–6	0.7–1.4	2.7	1–20.9	1.7-11.6
Amoxicillin syrup 125 mg	53	216.7	583.3	1	0.4–4.9	0.7–1.4	2.7	1–13	1.9–3.8
Cefixime 100 mg	150	464	2500	1	0.2–16.1	0.6–9	5.4	1–86.7	3.4–48.5
Dexamethasone 0.5 mg	191	87.5	290	1	0.3–5.2	0.7–1.4	3.3	1–17.1	2.3–4.6
All unbranded generic*	420	84	500	0.8	0.1–2.6	0.6–1.0	3.4	1–23.8	2.3–5.4
All branded generic*	583			1.4	0.3–18.6	0.9–5.0	5.5	1–173.2	2.9–16.2
All medicines*	1003			1	0.1–18.6	0.7–1.9	4.1	1–173.2	2.5–11.4

Average exchange rate for period of the study: USD1= 14,870

^*Ratios by brand status, and across all medicines, are calculated against the median and lowest price of all samples of the same medicine type

IQR, Interquartile Range; Rp, Rupiah.

Across all medicines, unbranded generics sold at a median of 0.8 times the mid-point price for their medicine and dose, while for branded generics, the median ratio was 1.4 times the median price. Online supplemental figure 2 depicts the range of price-to-median ratios across all medicines, by type of outlet. The cheapest and most expensive medicines are generally found at hospitals.

If we compare the most expensive products with the cheapest (a practice which tends to spotlight outliers), the smallest relative difference is a multiple of 13 for amoxicillin syrup, while at the other extreme, a patient paid an astonishing 173 times as much for the most expensive amlodipine, compared with the cheapest price. Table 1 also provides details of the range of ratios compared with the lowest price for each medicine and dose.

Prices varied within as well as between brands, because different outlets charged different amounts for the identical product. Across all products with five or more samples, the highest priced seller charged an average of 2.8 times more than the lowest-priced seller for the same unbranded generic and 1.6 times more for the same branded product. Figure 2 illustrates the differences for the 10 branded and 10 unbranded products with the widest price variation within the same medicine, dose, formulation and brand. In extreme cases, patients paid between five and 13 times more at some outlets than at others for the identical product.

Price and quality

In this analysis, we include 1088 samples subjected to any chemical testing, sampled from physical outlets or the online stores of registered pharmacies. Full details of quality testing are reported elsewhere,³⁷ and the sample-level data with all laboratory results are available in the study repository.³⁰

Here, we focus only on the relationship between price and quality. Of the 1088 samples included in the quality analysis, 84 failed any pharmacopeial test, for an unweighted prevalence of 7.7% out-of-specification samples.

Figure 3 plots test outcomes against price (relative to the medicine’s median) for branded and unbranded medicines. In logistic regression, there was no relationship between price and pharmacopeial quality (OR of failure by unit of price ratio 0.98, 95% CI 0.90 to 1.05, p>0.52), including after controlling for differences in medicine, district, or source (OR 1, 95% CI 0.91 to 1.1, p>0.91).

Figure 3. Quality testing outcome and price of each sample relative to the median for the medicine and dose, by branded status.

There was no significant difference in quality between branded and unbranded products (8.3% vs 7% failed any test, p=0.41) including after controlling for differences in medicine, district or source. Medicines available free to patients in the public insurance system were less likely to fail testing than medicines paid for out of pocket, but the difference was not statistically significant at the 95% confidence level (4.2% vs 8.3%, p=0.086).

Affordability of medicines

A course of medicines priced at the 25th percentile of prices paid locally cost more relative to minimum wages in remote Eastern Indonesia than in other study sites. The relative cost was between two and three times higher in remote eastern areas and rural districts than it was in large cities. This was both because medicines were more expensive in East Nusa Tenggara than in other regions, and because the minimum wage was lower in those districts. In the district with the highest relative costs, a treatment course of the most expensive medicine in our study cost 2.3% of the monthly minimum wage or 0.7% days’ wages (details are provided in online supplemental figure 3).

The absolute amount paid for a course of medicines at the 25th percentile of sample prices across all sites ranged from 4200 rupiah (about USD 0.28 at the 2022 average exchange rate of USD 1=IDR 14,870) to 34 200 rupiah (USD 2.3). We compared spending on medicines at the 25th percentile with average monthly per capita spending on other commodities reported by Statistics Indonesia. Monthly per capita spending on the staple food, rice, was twice as high as spending on the most expensive course of medicine in the study, and 15 times as much as the cheapest. Spending on instant noodles, a common, low-priced convenience food, was two-thirds of the average course-of-treatment spend for the medicines in our study. Indonesian households spent five times as much per person (including the non-smokers) buying cigarettes, on average, as they did on the average course of treatment in the study (shown in figure 4).

Figure 4. Per course/monthly costs of study medicines and monthly per capita spending on other commodities.

Potential for cross-subsidisation in the Indonesian medicine market

In our study, we collected 246 samples of unbranded generic products whose market authorisation holder also registered a branded equivalent (the same active ingredient and formulation), and 254 samples of unbranded products that had no registered branded equivalent. In order to test the hypothesis that the quality of cheaper unbranded products might be protected through cross-subsidisation with more expensive branded products, we compared the quality of these two groups of unbranded generics (see table 2).

Table 2. Relationship between cross-subsidy with sample INN and quality testing.

Paired status of unbranded sample	# of unique products*		# of study samples
	Indonesian market	STARmeds sample	Passed all tests	Failed any test (%)	Total
Unbranded only	214 (40.4%)	142	238	16 (6.3%)	254
MA holder also has branded version	316 (59.6%)	60	227	19 (7.7%)	246

^*Unique products: one active ingredient, formulation and brand/authorization-holder for allopurinol, amlodipine, amoxicillin, cefixime and dexamethasone

INN, International Nonproprietary Name; MA, Market Authorisation.

There was no statistically significant difference in testing failure rates between unbranded products that were not potentially cross-subsidised by a branded version and those that were (p=0.533).

Discussion

Ensuring access to quality-assured medicines is a core challenge for governments aspiring to provide universal health coverage to citizens. Affordable prices (to the public system and to patients paying out of pocket) are central to securing that access.

Since introducing a mandatory public health insurance scheme in 2014, Indonesia has introduced policies aimed at reducing the price of medicines procured by the state.¹⁷ While manufacturers complained prices had become unsustainably low, potentially threatening quality,³⁸ politicians continued to focus attention on the obstacle to access represented by higher-priced products.³⁹ This led to conflicting public narratives about medicine price and quality, though the perception that publicly procured medicines are of low quality dominated.^{40 41} The perception was reinforced in 2022, when several low-cost paediatric syrups made in Indonesia, including one provided free to insured patients, were found to be contaminated,⁴² allegedly contributing to some 200 child deaths.⁴³

In this large study of medicine quality in the Greater Jakarta region and six other rural and urban districts across Indonesia, we collected information on the actual prices paid by patients paying out of pocket for five common prescription medicines. We also tested samples for quality. We found a great diversity of both branded and unbranded products, selling at a wide array of price points; for all medicines and doses targeted in our study, the most expensive products sold for between five and 19 times the median price (and between 13 and 173 times the price of their cheapest retail equivalents). Leaving aside outliers and considering just the IQR, samples at the 75th percentile sold for multiples of 4.6 times more than those at the 25th percentile. This is a wider range than that identified in studies of price ranges in other countries, including Ireland and China,^{44 45} and is consistent with studies that show that mark-ups on insulin were far higher in Indonesia than in India or Ghana, especially in the private sector.⁴⁶

Britton et al suggested that the introduction of national health insurance contributed to the segmentation of the Indonesian pharmaceutical market, with multinational companies concentrating their sales on branded medicines, while domestic companies focused on unbranded generics.¹¹ However, we found that domestic companies were themselves segmenting the market by selling both branded and unbranded versions of the same medicine at different price points.

Although we identified more out-of-specification medicines than were found in a recent study of cardiovascular medicines in East Java, our findings concurred that there is no evidence of any relationship between price and quality, meaning that patients choosing the cheapest unbranded medicines are not disadvantaged.⁴⁷ Our findings did not, however, support those authors’ speculation that pharmaceutical companies may protect investment in assuring the quality of lower-priced medicines by cross-subsidising through the sale of high-priced brands of the same product. In our study, there was no difference in the quality of unbranded generics that were unique, compared with those that were matched with a more expensive brand.

Accessing free medicines in Indonesia can be time-consuming and inconvenient; this likely explains why so many patients choose to pay for common medicines out of pocket, even when free alternatives are available. However, it is less clear why people are willing to pay such a high premium for some brands when they could choose to buy versions that meet the same pharmacopeial specifications more cheaply. We were not able to interview patients; a major limitation of our study. But we believe at least part of the answer must be that, whatever the facts, Indonesian patients continue to associate low cost with low quality. This phenomenon has been widely reported elsewhere,⁴⁸ including in the descriptively titled study ‘This body does not want free medicines’, an examination of South African patients’ views on unbranded generic medicines procured through state systems.⁴⁹ Our study adds to a body of evidence suggesting that policy-makers should accompany cost-containing procurement policies with strategies to manage consumer perceptions of medicine quality if they wish to increase acceptance of lower-priced medicines.

Similarly, policy-makers should consider how procurement regulations impact on the people and institutions that control patient access to medicines. Again, we did not directly interview healthcare providers, but we note that the highest mark-up on branded medicines was often imposed by hospitals. Previous studies in Indonesia found that physicians, pharmacists and other healthcare professionals influence patients’ choice of medicines, often nudging them towards brands that will maximise revenues from medicine sales, including through incentive payments from pharmaceutical sales teams.^{13 14 47} This is not a universal phenomenon.⁵⁰ However, the dynamic is also at play in the USA and China, which, like Indonesia, have mixed health systems in which the sale of medicines has historically been a profit centre.^{48 51}

Though patients in many countries appear to associate low prices with low quality medicines, there is no strong evidence published to support those beliefs. The Indonesian study of cardiovascular medicines,⁴⁷ and another in India,⁵² found no substandard medicines, despite wide price variations. A 2024 study in three Sub-Saharan African countries compared prices with international reference prices from 2015.⁵³ Price ratios for originator brands and medicines imported from countries with stringent regulatory authorities were reported to be between two and three times higher than equivalents made in other countries. No samples of this more expensive type failed testing, compared with 21% of medicines made elsewhere, suggesting that higher prices might indeed indicate more investment in quality assurance in this context. However, when all samples in the study were entered into a regression model, the authors found no relationship between price and quality.

It is concerning that medicine prices in our study were highest in the poorer Eastern regions of Indonesia where incomes are lowest, likely a consequence of distance from the centres of pharmaceutical production in Java and high transport costs across the archipelago. We note also that patients tended to pay more when buying medicines from doctors and midwives. These health workers are often the principal source of medicines in remote rural areas of Indonesia. However, they are not allowed to buy medicines from distributors. This means they buy at higher retail prices, adding an additional mark-up when selling on to their patients. We believe this restriction should be reconsidered so that patients in remote areas are not doubly disadvantaged.

Overall, however, we found that Indonesian patients in most settings could access affordable, quality-assured versions of the commonly prescribed medicines in our study if they chose to. These were available at retail and from health facilities for patients who either chose to pay out of pocket or were obliged to because they were not correctly insured, or because public facilities were out of stock. For every medicine and in every district, a course of treatment could easily be obtained for less than 1 day’s wages. For even the most expensive medicine in the study, this is about half as much as households spend per person on another necessary commodity, rice, and just 40% of what the average household spends on cigarettes, a discretionary product.

Indonesia continues to face challenges in delivering health services evenly across the vast, geographically challenging and economically diverse archipelago, with access to hospitals and specialist doctors a particular challenge.⁵⁴ This clearly affects access to medicines that are delivered in tertiary settings.⁵⁵ Further, changes to the medicine procurement system implemented from January 2023, which eliminated consolidated buying through transparent, single-winner tenders for some medicines, are likely to affect the pharmaceutical market in unpredictable ways. We note also that, in common with virtually all academic studies of medicine quality, we could not afford to test for impurities, meaning that our results may understate the true number of substandard samples.

Conclusion

Our results suggest that at the time of the study, public procurement policies and market dynamics, coupled with relatively strong regulatory oversight of domestic production and the supply chain, were functioning well in the world’s fourth most populous country. Many patients apparently preferred not to queue up for free medicines at public facilities, continuing instead to pay out of pocket for common essential medicines. However, even those working at the minimum wage in remote areas could access a course of quality-assured medicine across four therapeutic classes for the equivalent of a few hours’ wage. Meanwhile, Indonesians with greater means could, if they chose, spend far more to obtain branded products of similar quality.

Data availability statement

Data are available in a public, open access repository.