The African cigarette price data landscape: an overview of gaps and opportunities

Abstract

Introduction

Africa has the lowest cigarette taxes of any region. Price data are needed to produce evidence that informs tax policy change.

Objective

To locate and describe datasets that provide information on cigarette prices in African countries and to identify opportunities for expanding the region’s cigarette price information base.

Methods

Three data repositories and relevant literature were searched to locate free datasets that provide cigarette prices for African countries. 13 categories of information were extracted from each dataset, including the type of data collected (self-reported vs observational). The number of outliers and summary statistics, for the price of 20 cigarette sticks, was calculated using the most recent data available from each data collection project identified in a country. T-tests were conducted to compare these statistics for the 14 countries with both self-reported and observational price data available.

Results

A total of 131 datasets, covering 39/47 African countries, contain information on cigarette prices. Most datasets (N=111/131) provide self-reported information. Only 11 countries have used the three large-scale surveys that repeatedly collect information on smoking behaviour across Africa (Multiple Indicator Cluster Survey, Demographic and Health Survey and STEPwise) to collect information on cigarette prices. No active projects that collect self-reported price data sample from the youth (aged<15). Prices collected through retail observations exhibit fewer outliers than self-reported prices (p<0.05).

Conclusion

The tobacco-control community should lead the process of using existing data collection efforts to enhance Africa’s cigarette price information base. Increasing observational data collection efforts at retailers could improve our understanding of the prices paid for cigarettes in the region.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Cigarette taxes in Africa are the lowest among all regions. Cigarette prices are a key input for generating evidence to support increases in tobacco taxation.

• No study has surveyed the African cigarette price data landscape.

WHAT THIS STUDY ADDS

• This paper offers the tobacco-control community a comprehensive overview of the free datasets that contain information on cigarette prices in African countries and provides a description of the features of these datasets.

• It highlights gaps in existing price data collection efforts and demonstrates that a broader base of cigarette price information for the region can be achieved through collaborative efforts led by the tobacco-control community.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study serves as a reference point for researchers seeking a roadmap of the different sources of cigarette price data available for African countries.

• This study highlights the scope that exists for leveraging existing data collection efforts to improve current understanding of the prices paid for cigarettes in the region and identifies the role that the tobacco-control community can play in this process.

• This study also helps researchers and funders identify countries where the information gaps exist, and where skills and resources are most needed to obtain a comprehensive understanding of the cigarette price landscape in the African region.

Introduction

The number of tobacco users in Africa is expected to increase.¹ Factors underpinning this expectation include rapid population growth and targeted marketing by the tobacco industry to secure new customers in relatively low-prevalence markets.² Research from across the globe,^{3 4} and specifically from Africa,⁵,⁷ demonstrates that excise tax increases are a powerful tool for reducing tobacco use. At the policy level, the WHO Framework Convention on Tobacco Control (WHO FCTC) Article 6 Guidelines encourage Parties to raise taxes on tobacco products consistently to reduce their affordability over time.⁸

Despite this, countries classified as African by the WHO African Regional Office (WHO-AFRO) are the poorest performers in the area of tobacco taxation.⁹ The region’s average cigarette excise tax incidence on a 20-pack of the most-sold cigarette brand is 27.2%—the lowest of any WHO country grouping.⁹ Although not a necessary condition for securing tax policy change, policy-makers typically require local evidence before increasing tobacco taxes.¹⁰ Local price data are needed to produce this evidence.

For most countries, consistent sources of free cigarette price data are provided, biennially, by the WHO⁹ and the FCTC Convention Secretariat.¹¹ The FCTC Convention Secretariat provides the retail price of the most-sold brand of imported and domestic cigarettes available at the most widely used point of sale in capital cities.¹¹ The WHO provides the retail price of a 20-pack of cigarettes for three market segments (most sold, cheapest and most expensive). These prices are collected from supermarkets/hypermarkets, and convenience stores, in capital cities.⁹

The features of the African cigarette market present challenges for the application of these prices in tobacco-control research on the continent. These features include both the popularity of single-stick sales and purchases of cigarettes at informal retailers or through street vendors.¹²,¹⁵ Research from Africa shows that differences in the packaging type and the retail outlet where cigarettes are purchased have a significant impact on the unit cost of each cigarette for the buyer.¹⁴ This suggests a disconnect between the cigarette prices reported by leading health agencies and the prices typically paid for cigarettes in the region. An overview of the broader range of datasets that provide cigarette prices for African countries, along with a description of the features of these datasets, may therefore be useful to researchers.

This study’s purpose is threefold. First, it identifies and describes free datasets that provide information on cigarette prices in any of the 47 WHO-AFRO countries. Second, it summarises the price estimates derived from these datasets to enhance understanding of the prices currently paid for cigarettes in the region and to explore differences in price estimates obtained through various data collection methods (eg, self-reported prices collected through surveys and prices collected through observation at cigarette retailers). Finally, it uses observations made during the dataset search process to identify opportunities for improving the frequency and coverage of cigarette price data collection in the region.

As the first study to survey the African cigarette price data landscape, this paper makes three contributions. First, it provides an overview of the cigarette price data available for countries in the region and may thus serve as a valuable reference for researchers who need to use these data in their work. Second, this paper draws attention to those African countries where little is known about cigarette prices, highlighting where skills and resources are most needed to obtain a comprehensive understanding of the regional cigarette price landscape. Third, by interrogating a range of datasets to locate those that contain information on cigarette prices, this paper identifies the potential for multisectoral collaboration to expand the region’s cigarette price information base.

Methods

Search strategy

To ensure a replicable dataset search process, this study used selected aspects of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines for systematic reviews: defining eligibility criteria, specifying information sources and presenting the search strategy.¹⁶

To be eligible for inclusion, datasets needed to contain information on manufactured cigarette prices (retail level or self-reported) or provide information on both the amount spent on manufactured cigarettes and the number of cigarettes purchased, during a specified time frame. Only datasets collected in WHO-AFRO countries were considered.¹⁷ These 47 countries and their subregional location are listed in online supplemental appendix 1. Given the documented resource constraints facing the African tobacco-control community,¹⁸ datasets were only eligible for inclusion if they could be accessed free of charge. No time frame restriction was imposed in the search process.

Three data repositories were searched initially on 7 November 2023 and subsequently on 5 July 2024, to locate datasets meeting the inclusion criteria: the World Bank Microdata Library,¹⁹ the WHO Non-Communicable Disease Microdata Repository²⁰ and the DataFirst Open Data Portal.²¹ Unique search strategies, documented in online supplemental appendix 2, were adopted for each database. Broadly, each repository was searched for datasets that included any information on cigarettes in WHO-AFRO countries. The codebooks of the datasets returned were then examined to ascertain if information that allows the derivation of cigarette prices was present in the dataset.

The PubMed and EconLit databases were also searched initially on 31 October 2023 and subsequently on 5 July 2024, to find studies that use cigarette price data collected in African countries (search strings are presented in online supplemental appendix 3). The abstracts of the studies returned were screened for mentions of cigarette prices/expenditures or related concepts, such as ‘price elasticity’ and ‘tax simulation model’. Relevant studies were read to extract the source of cigarette price data used. Reference list searches were performed on each study that underwent full-text review to locate additional datasets. For all datasets located in studies, a Google Search was performed to ascertain whether other waves/rounds of data meeting the inclusion criteria had been collected under the same project banner.

Data analysis and synthesis

To be classified as an individual dataset, the dataset needed to be the result of a unique data collection process that occurred during a clearly defined time frame. By implication, each round of data gathered under the banner of a multiround data collection project is counted as a single dataset.

For each dataset meeting the inclusion criteria, the following information was extracted: (1) the dataset name and wave/round; (2) the year of data collection; (3) the country/countries covered; (4) the subregion/s in which these countries are located; (5) the type of cigarette price data collected (table 1); (6) a description of the cigarette price data collection universe; yes/no indicators showing whether each dataset provides information on (7) the cigarette brand, (8) the retail-outlet type, (9) the cigarette-packaging type associated with the price observations; and (10) whether the prices are nationally representative. National representativity was assessed based on whether it was claimed in the sampling procedure.

Table 1. Categorisation scheme employed to classify the type of cigarette price data collected in Africa^*.

Type of cigarette price data collected	Characteristics
Broad category: data based on self-reported information64
Individual-level self-reported data provided by cigarette smokers	This type of cigarette price data is collected through surveys. These surveys can be self-administered, implemented face to face or conducted online. In these surveys, respondents indicate the cost of/expenditure on/price paid for cigarettes purchased, and the number of cigarettes purchased, at a specified point in time (eg, the most recent purchase), or over a specified period (eg, the past week).
Constrained individual-level self-response data provided by cigarette smokers	This type of cigarette price data is collected through surveys. These surveys can be self-administered, implemented face to face or conducted online. In these surveys, respondents are asked to provide an indication of the price that they usually pay for a specified cigarette packaging type (eg, a pack of 20 cigarettes). Constraints on self-reported cigarette prices take the form of predefined price points or predefined price ranges, from which respondents can choose. As a hypothetical example, in response to a question such as: ‘How much do you usually pay for a pack of 20 cigarettes?’, response options that produce a constrained self-reported cigarette price by providing predetermined price points are shown in A below. Hypothetical response options that produce a constrained self-reported cigarette price by providing predetermined price ranges are shown in B below.
	A. Price points	B. Price ranges
	US$3 US$4 US$5 More than US$5	US$3–US$3.99 US$4–US$4.99 US$5–US$5.99 More than US$5.99
Household-level self-response data	This type of cigarette price data is collected through face-to-face surveys in which the survey administrator asks the household head about the value of (or expenditure on) cigarettes purchased by all household members over a specified period (eg, the past week or the past 30 days) and the number of cigarettes corresponding with the reported value/expenditure. These surveys collect information on a range of household expenditure items, not just cigarettes.
Community-level self-response data	This type of cigarette price data is provided by a small group of selected community members. It relies on respondents’ collective memory of cigarette prices observed at retailers within the locality that they reside in. This mode of price collection is typically conducted as part of a community questionnaire implemented in parallel with a national-level household survey. One questionnaire is administered for a village, group of villages, or urban location in each sample enumeration area selected for the household survey. Respondents selected to participate in this questionnaire comprise a group of at least five people residing in the same locality. Respondents typically include people such as the village head and their partner, school principals, leaders of religious groups, health professionals or long-term residents. Price collection relies on respondents’ collective memory of the prices charged for a range of items in the community that they live in, not just cigarette prices. For cigarettes specifically, respondents are only asked to supply the price of one cigarette packaging type and brand. Retailers are not visited to collect these prices.
Retail-level self-response data provided by cigarette retailers	This type of cigarette price data is collected through face-to-face interviews with cigarette retailers. During these interviews, retailers are asked to provide the price at which they sell the cigarette packaging type/s and brand/s of interest to the researchers.
Broad category: observational data64
Retail-level prices collected through direct observations made at cigarette retailers	This type of price data is collected based on direct observations made at retail outlets that sell cigarettes. Prices are recorded at the time of observation by trained/informed data collectors.

^*This categorisation scheme was developed solely based on observations made during the search for datasets that collect information on cigarettes prices in African countries. It is not intended to be a representation of all possible modes through which cigarette price data can be collected. The broad categories of the types of cigarette price data provided in the table above, and a more general discussion of the different modes of cigarette price data collection, can be found in volume 12 of the International Agency for Research on Cancer Handbook of Cancer Prevention.⁶⁴ The datasets that produced the categorisations shown in this table are the subject of the results section of this paper.

Because of the popularity of single-stick purchases in the region, for each dataset that reports the packaging type associated with the price observations, a record was kept of whether the dataset (11) provides the prices of single-stick cigarettes specifically. Similarly, for each dataset that reports the retail-outlet type associated with the price observations, a record was made of whether the dataset (12) allows users to estimate prices paid at informal retail outlets (eg, street vendors and spaza shops). For datasets based on individual-level or household-level surveys, a record was made of whether (13) information on respondents’ cigarette consumption is also available.

To provide an overview of Africa’s current cigarette price landscape, the most recent year of data available from each data collection project implemented in a country is used to calculate summary statistics, and the coefficient of variation (CoV), for the price of 20 cigarettes. Reported prices are based on derived per stick prices that account for all cigarette packaging types included in a particular dataset, multiplied by 20. Derived prices are presented in constant 2019 international dollars (I$).²² Prices based on self-reported data were weighted by the sampling weights provided in the dataset, and by cigarette consumption, when this information was provided. Details on how the per stick cigarette price was derived, for each dataset used in the analysis, are documented in online supplemental appendix 4.

The summary statistics presented are based on ‘cleaned’ prices in which outliers have been removed. Identifying outliers in each dataset followed two steps. First, extreme outliers—defined as price observations greater than I$30 for 20 cigarettes—were removed. These prices are implausibly high in the African context.²³ Following this, prices falling outside of 3 SDs of the mean price were removed.

Across all datasets, 2% of price observations (N=1080) were flagged as outliers: 6% of these were 3 SDs below the mean price, 41% were extreme outliers and 54% were prices higher than 3 SDs from the mean price. Overwhelmingly, the outliers identified across all datasets appear to be instances of a recorded packaging type that does not seem plausible in relation to the corresponding price/cost/expenditure. It is therefore assumed that observations flagged as outliers predominantly reflect issues of misreporting/data-capture errors, as opposed to ‘true’ prices. For this reason, the proportion of outliers is taken to provide an indication of data reliability. The smaller the proportion of outliers, the more reliable the price data, and vice versa.

T-tests were used to compare differences in the proportion of outliers, and the main summary statistics for countries with both self-reported and observational cigarette price data available.

Results

The search process returned 131 datasets meeting the inclusion criteria (figure 1). Because the data repository searches started by identifying datasets that include any information on cigarettes, insights about the scope for expanding African cigarette price data collection can be derived by examining the reasons for dataset exclusion (figure 1, left-hand panel).

Figure 1. Results of data repository and literature search. WHO GRTE stands for the WHO Global Report on the Tobacco Epidemic. Datasets that were removed because they only contain ‘miscellaneous’ information about cigarettes consist of datasets that only include the following information: information relating to income earned from cigarette sales or expenditure on cigarette production (n=3); information on healthcare worker advice regarding cigarette use during pregnancy (n=3); information on whether a cash grant was spent on cigarettes, among other products (n=5); information on a ‘cigarette cup’ as a unit of measurement in questions about tobacco crop production (n=3); information on the value of in-kind payments in the form of cigarettes for farm labour (n=1); respondents are asked to indicate what they are saving money for and cigarettes are given as an option (n=1); respondents are asked about their beliefs regarding AIDS transmission through cigarette sharing (n=1); cigarettes are used in a hypothetical example to elicit a response on what respondents think the best course of action is in a morally ‘difficult’ situation (n=1); information on the retail price per kilogram of cigarette tobacco (n=3); a dataset that compiles cigarette price indices, not actual retail prices (n=1).

Of the 559 datasets that contained any information on cigarettes, 324 were excluded because the information that they contained focused almost exclusively on smoking behaviour. Of these, 217 (67%) were from 3 self-response surveys implemented repeatedly in most African countries to monitor health behaviours: WHO STEPwise (STEPS) (N=43),²⁴ Multiple Indicator Cluster Surveys (MICS) (N=69)²⁵ and Demographic and Health Surveys (DHS) (N=105).²⁶

The 22 datasets excluded because they asked respondents how much they think a specified cigarette packaging-type costs are from Global Youth Tobacco Surveys (GYTS) implemented in African countries since 2012.²⁷ These datasets are excluded because they provide respondents’ perceptions of the prices paid for cigarettes, not the actual prices paid. Prior to 2012, GYTS datasets provided constrained self-response data on the price that students usually paid for a specified cigarette packaging type (typically a 20-pack of cigarettes).²⁸

Figures2 3 summarise the characteristics of the 131 datasets that met the inclusion criteria. This information is provided, by dataset name, in online supplemental appendix 5. Because 13 datasets provide information on cigarette prices in more than one country, the 131 datasets included in this study contain information on cigarette prices gathered through 210 country-specific data collection rounds/waves in 39 of Africa’s 47 countries (figure 2).

Figure 2. Number of times information on cigarette prices has been collected in each WHO-AFRO country, N=210. Map generated using the Power Map tool in Excel based on the countries covered in the 131 datasets provided in online supplemental appendix 5. Country classification is based on the WHO-AFRO regional grouping.¹⁷ A full list of these countries, and their subregional location within Africa, is provided in online supplemental appendix 1. No datasets meeting the inclusion criterion were found for three countries located in Central Africa (Central African Republic, Gabon, Sao Tome and Principe); three countries in Western Africa (Guinea, Guinea-Bissau, Liberia) and two Eastern African countries (Rwanda and South Sudan). AFRO, African Regional Office.

Figure 3. Characteristics of datasets that collect information on cigarette prices in WHO-AFRO countries, grouped by data type (N=131). Only one dataset that provides cigarette price information could not be classified because the method used to acquire the prices was not specified. These are time series average cigarette prices for the most popular price category in Ghana and Zimbabwe, collected by Philip Morris, and made public through the Master Settlement Agreement.^{65 †}Indicates that the sample’s representativity could not be assessed due to a lack of information about how the prices were acquired (only applied to the Phillip Morris dataset).⁶⁵ *Indicates that the criteria were assessed based on individual-level and household-level self-response data only. AFRO, African Regional Office.

Of these 210 country-specific data collection rounds/waves, 107 (51%) are from countries located in Southern Africa, 55 (26%) are from countries in Eastern Africa, 34 are from countries in Western Africa (16%) and 12 (6%) are from countries located in Central Africa (figure 2). Algeria, which has two datasets meeting the inclusion criteria, is the only WHO-AFRO country located in Northern Africa.

Across all 131 datasets, 92 (70%) provide information on the brand corresponding with each price observation (figure 3). 97 datasets (74%) provide information on the type of retail outlet at which the prices were paid/observed. Of these, 95 datasets (98%) provide information that allows researchers to derive estimates for the prices paid at informal retailers, specifically. Across all 131 datasets, 121 (92%) provide information on the packaging type associated with each price observation. Of these datasets, only 56 (46%) provide information that allows for single-stick prices to be derived. Across all 131 datasets, 74 (56%) provide prices collected from nationally representative samples.

Datasets based on self-reported information dominate the sample (N=111/131, 85%), with individual-level self-response data provided by cigarette smokers being the most common type (figure 3). Other than the constrained self-response data collected in the pre-2012 waves of the GYTS (N=48), no datasets that collect self-reported information on cigarette prices among the youth (aged <15) were found (online supplemental appendix 5). Of the 100 datasets that are based on either individual-level or household-level self-response data, 93 (93%) also collect information on respondents’ cigarette consumption (figure 3).

Although prices derived from household-level self-response data (N=15) predominantly come from national-level surveys,²⁹,³² they are also found in datasets developed to assess the impact of small-scale cash transfer programmes in Niger³³ and Madagascar³⁴ (online supplemental appendix 5).

Prices collected through observation at retailers make up 14% of datasets (N=19/131) (figure 3). 12 of these datasets come from the African Cigarette Prices (ACP) Project (online supplemental appendix 5). This is a multicountry data collection project with 12 rounds, specifically implemented to collect prices across multiple cigarette packaging types, brands and retail outlets in African countries.³⁵ However, cigarette prices based on observations at retailers are also collected through separate ‘community’ questionnaires in recent rounds of national household-level surveys implemented in Ethiopia,³⁶ Ghana³⁷ and Tanzania³⁸,⁴⁰ (online supplemental appendix 5). A community questionnaire is also implemented in waves 3 and 5 of the Malawi Integrated Household Survey,^{41 42} but these surveys collect community-level self-response data.

Table 2 provides descriptive statistics for the derived price of 20 cigarettes in constant 2019 I$ for the most recent year of data available from each unique data collection project implemented in a country (N=52). Across all 131 datasets, 98 represent the only or the most recent year of data collected. Despite this, 46 of these 98 datasets are excluded from table 2. 38 out of these 46 datasets are excluded because the parameters of interest cannot be constructed from these data (eg, constrained self-response data). Reasons for all exclusions are documented in online supplemental file appendix 6.

Table 2. Most recent year of data: number of outliers and summary statistics for the derived price of 20 cigarette sticks (N=52).

Data collection project	Dataset name	Country	Year	N: prices before removing outliers	N: prices after removing outliers (% of outliers in the sample)	N: usable observations*	Mean (95% CI)	Median	SD	CoV
Type of data: Individual-level self-response data
Global Adult Tobacco Survey66	GATS: Botswana	Botswana	2017	464	427 (7.9)	302	9.17 (8.77 to 9.57)	9.72	3.52	0.38
	GATS: Cameroon	Cameroon	2013	367	363 (1.1)	275	2.00 (1.89 to 2.11)	2.34	0.93	0.46
	GATS: Ethiopia	Ethiopia	2016	498	481 (3.4)	405	2.27 (2.17 to 2.38)	2.72	1.06	0.47
	GATS: Kenya	Kenya	2014	408	406 (0.5)	326	2.84 (2.75 to 2.94)	3.07	0.88	0.31
	GATS: Nigeria	Nigeria	2012	374	361 (3.5)	288	2.89 (2.75 to 3.03)	3.21	1.24	0.43
	GATS: Senegal	Senegal	2015	184	182 (1.1)	165	3.01 (2.77 to 3.24)	2.23	1.55	0.51
	GATS: South Africa	South Africa	2021	1382	1362 (1.4)	1178	3.93 (3.82 to 4.05)	2.85	2.02	0.51
	GATS: Tanzania	Tanzania	2018	288	286 (0.7)	221	3.39 (3.21 to 5.68)	2.67	1.36	0.40
	GATS: Uganda	Uganda	2013	390	382 (2.1)	322	2.83 (2.70 to 2.96)	2.07	1.19	0.42
International Tobacco Control Evaluation Project (ITC) Project67,69	ITC: Kenya (Wave 2)	Kenya	2018	716	707 (1.3)	624	3.26 (3.17 to 3.35)	2.39	1.16	0.36
	ITC: Mauritius (Wave 3)	Mauritius	2011	533	527 (1.1)	527	6.96 (6.87 to 7.05)	6.82	1.05	0.15
	ITC: Zambia (Wave 2)	Zambia	2014	615	590 (4.1)	556	3.82 (3.70 to 3.93)	3.42	1.42	0.37
National surveys not specific to a tobacco control project43 44 70	DHS: Cameroon 2018	Cameroon	2018	661	644 (2.6)	615	2.43 (2.33 to 2.52)	2.18	1.17	0.48
	MICS: Cameroon 2014	Cameroon	2014	590	580 (1.7)	577	2.29 (2.22 to 2.35)	2.28	0.80	0.35
	National Income Dynamics Study (wave 5)	South Africa	2017	3940	3659 (7.1)	3565	4.10 (4.01 to 4.19)	3.98	2.81	0.69
National surveys not specific to a tobacco control project45,52	STEPS: Algeria 2017	Algeria	2017	754	739 (2.0)	734	5.90 (5.77 to 6.04)	5.76	1.84	0.31
	STEPS: Benin 2015	Benin	2015	176	169 (4.0)	162	1.65 (1.55 to 1.74)	1.29	0.60	0.36
	STEPS: Botswana 2014	Botswana	2014	453	383 (15.5)	338	9.35 (9.04 to 9.66)	9.12	2.94	0.31
	STEPS: Cabo Verde 2020	Cabo Verde	2020	304	300 (1.3)	242	5.93 (5.78 to 6.08)	6.38	1.18	0.20
	STEPS: Eswatini 2014	Eswatini	2014	169	150 (11.2)	136	6.99 (6.47 to 7.50)	7.17	3.04	0.44
	STEPS: Ethiopia 2015	Ethiopia	2015	463	443 (4.3)	409	2.09 (2.00 to 2.18)	1.74	0.92	0.44
	STEPS: Lesotho 2012	Lesotho	2012	169	144 (14.8)	101	8.52 (8.03 to 9.01)	10.10	2.48	0.29
	STEPS: Zambia 2017	Zambia	2017	312	305 (2.2)	251	3.97 (3.60 to 4.34)	2.96	2.98	0.75
Tobacco-specific survey not under the banner of a long-term project71 72	Gambia Tobacco Survey	The Gambia	2017	792	781 (1.4)	712	3.49 (3.41 to 3.58)	4.14	1.14	0.33
	Township survey: Round 2	South Africa	2018	1929	1913 (0.8)	1913	4.56 (4.42 to 4.70)	4.57	2.43	0.53
Type of data: household-level self-response data
Subnational surveys not specific to a tobacco control project33 34	Human Development Cash Transfer—Impact Evaluation (Midline survey)†	Madagascar	2018	608	595 (2.1)	595	2.56 (2.48 to 2.64)	1.99	1.00	0.39
	Safety Nets Project Impact Evaluation: Baseline Survey†	Niger	2012	159	150 (5.7)	150	1.83 (1.68 to 1.97)	2.24	0.88	0.48
Type of data: community-level self-response data
National survey not specific to a tobacco control project41	Malawi Fifth Integrated Household Survey†	Malawi	2019	72	72 (0)	72	1.28 (0.95 to 1.60)	0.18	1.40	1.10
Type of data: Retail-level prices collected at cigarette retailers by trained data collectors and recorded at the time of observation
African Cigarette Prices Project (ACP)35	ACP Project: round 10	Botswana	2021	1841	1814 (1.5)	1814	9.71 (9.64 to 9.78)	10.60	1.52	0.16
	ACP Project: round 10	Cameroon	2021	66	62 (6.1)	62	1.97 (1.86 to 2.08)	2.03	0.42	0.21
	ACP Project: round 10	Chad	2021	1508	1506 (0.1)	1506	2.66 (2.62 to 2.70)	2.05	0.80	0.30
	ACP Project: rounds 2 and 3‡	Eswatini	2016§	667	667 (0)	667	6.88 (6.77 to 7.00)	6.88	1.54	0.22
	ACP Project: round 10	Ethiopia	2021	481	473 (1.7)	473	3.63 (3.51 to 3.74)	3.66	1.28	0.35
	ACP Project: round 5	Ghana	2018	276	276 (0)	276	3.31 (3.16 to 3.46)	3.32	1.27	0.38
	ACP Project: round 7	Kenya	2019	160	160 (0)	160	3.95 (3.78 to 4.11)	4.09	1.08	0.27
	ACP Project: rounds 11 and 12‡	Lesotho	2022	2941	2939 (0.1)	2939	6.65 (6.55 to 6.75)	7.49	2.76	0.42
	ACP Project: rounds 8 and 9‡	Madagascar	2019	306	305 (0.3)	305	3.63 (3.51 to 3.76)	3.78	1.12	0.31
	ACP Project: rounds 7–9‡	Malawi	2019¶	1848	1815 (1.8)	1815	2.50 (2.46 to 2.55)	2.10	1.00	0.40
	ACP Project: round 2	Mauritius	2016	243	232 (4.5)	232	8.62 (8.46 to 8.77)	8.41	1.22	0.14
	ACP Project: round 9	Mozambique	2020	497	497 (0)	497	4.41 (4.24 to 4.58)	4.29	1.96	0.44
	ACP Project: round 11	Namibia	2022	1293	1291 (0.2)	1291	5.47 (5.40 to 5.54)	5.41	1.28	0.24
	ACP Project: round 9	Nigeria	2019**	400	399 (0.3)	399	2.21 (2.13 to 2.28)	2.24	0.72	0.32
	ACP Project: rounds 11 and 12‡	South Africa	2022	9031	8912 (1.3)	8912	5.14 (5.10 to 5.17)	5.19	1.51	0.29
	ACP Project: round 9	Tanzania	2020	878	878 (0)	878	5.30 (5.21 to 5.40)	5.00	1.45	0.27
	ACP Project: round 7	Uganda	2019	90	87 (3.3)	87	5.06 (4.54 to 5.57)	3.24	2.43	0.48
	ACP Project: rounds 7 and 8‡	Zambia	2019	332	328 (1.2)	328	2.64 (2.46 to 2.82)	2.10	1.68	0.64
	ACP Project: rounds 11 and 12‡	Zimbabwe	2022	13 352	13 163 (1.4)	13 163	1.96 (1.94 to 1.99)	1.24	1.66	0.85
National survey not specific to a tobacco control project36 40	Ethiopia Socioeconomic Panel Survey (wave 5)	Ethiopia	2022	370	363 (1.9)	363	2.37 (2.66 to 2.48)	1.87	1.05	0.44
	Tanzania National Panel Survey (wave 5)†	Tanzania	2021	484	480 (0.8)	480	3.11 (3.02 to 3.12)	2.50	0.96	0.31
Sub-national survey not specific to a tobacco control project73	Impact Evaluation: Feeder Roads†	Ghana	2009	416	413 (0.7)	413	2.15 (2.04 to 2.27)	2.07	1.20	0.56
Type of data: Retail-level self-response data provided by cigarette retailers
Tobacco-specific survey not under the banner of a long-term project74	Sierra Leone Tobacco Survey: round 2†	Sierra Leone	2021	486	479 (1.4)	479	1.33 (1.31 to 1.36)	1.34	0.27	0.20
Sub-national survey not specific to a tobacco control project75	Township Micro-enterprise Survey: rounds 4 and 5†	South Africa	2013	224	224 (0)	224	6.06 (5.78 to 6.34)	5.94	2.12	0.35

Prices reported in this table are based on the price of a single cigarette stick, derived in a manner that accounts for the full range of manufactured cigarette packaging types that are recorded in each dataset. Details of how the price per single cigarette was derived are documented, for each dataset listed in this table, in online supplemental appendix 4. All prices are expressed in constant 2019 international dollars. The superscript, (f), detailed below, provides an indication of those datasets for which the per stick cigarette prices could only be derived based on a single cigarette packaging type (eg, if the survey only collected information on the price paid for a box of 20 cigarettes).

^*In the datasets based on individual-level self-reported information presented in this table, not all respondents who provided information that can be used to derive cigarette prices reported on their cigarette consumption behaviour. Thus, the difference between the number of observations given in the column labelled ‘N: Prices after removing outliers (% change in sample size)’ and the column labelled ‘N: Usable observations’, is the number of individuals who provided information that allowed the price that they paid for their most-recent cigarette purchase to be derived but did not provide information that allowed the number of cigarettes that they smoke each day/week to be calculated.

^†Only per stick cigarette prices based on one cigarette packaging type could be derived from this dataset. The statistics presented here thus do not capture the full range of packaging types in which cigarettes can be purchased. Details are provided in online supplemental appendix 4.

^‡Despite being collected in different ACP rounds, these data were collected in the same calendar year in this country. See online supplemental appendix 5 for the calendar months and years in which a specific ACP round took place. The ACP project usually conducts price collection twice a year, though three rounds of data collection took place in 2019.³⁷ ACP prices are not necessarily collected in the same countries in each round.

^§2022 is the most recent year of data collection in Eswatini under this data collection project. However, CPI data for the country are not available for 2022, making rebasing of the CPI to get the 2022 prices into 2019 prices, impossible. Therefore, 2016, which is the second most-recent round of ACP data collection in the country, is used as the most recent year in the current study.

^¶Despite 2021 being the most recent year for which cigarette price data are available for Malawi through the ACP Project, the data used here are from 2019 because this year offers a much larger sample size (N=1848) than both 2021 (N=96) and 2020 (N=42).

^**The most recent year of data collection in Nigeria was 2020. The data used here are from 2019 because this offers a larger sample size than 2020 (N=400 in 2019 and N=266 in 2020).

CoV, coefficient of variation; CPI, Consumer Price Index; DHS, Demographic and Health Survey; MICS, Multiple Indicator Cluster Survey.

The results presented below analyse two subgroups from table 2. Table 2 is reproduced, for each subgroup analysed, in online supplemental appendix 7A,B.

For 14 countries, cigarette prices could be derived based on both self-reported information (N=23) and prices collected through observation at cigarette retailers (N=16) (online supplemental appendix 7A). Among these data, the average percentage of price outliers in the sample is lower, by 2.5 percentage points, for prices derived based on observations at retailers than it is for prices derived based on self-reported information (p<0.05, online supplemental appendix 7A).

For 12 of the 14 countries with more than one type of cigarette price data available, the self-reported data (N=20) and the observational data (N=13) are more comparable because they account for all cigarette packaging types, any brand and any retail-outlet type. For this subsample, differences in the average mean and median prices, by data type, are not statistically significant (online supplemental appendix 7B). However, the average proportion of outliers is lower, by 2.8 percentage points, for prices collected at retailers than it is for self-reported prices (p<0.05, online supplemental appendix 7B). Furthermore, the average CoV obtained using self-reported data is 10.3 percentage points higher than that obtained using retail-level prices (p<0.05, online supplemental appendix 7B). This signifies that self-reported price information indicates a wider distribution of cigarette prices than is suggested by prices collected through observations at retailers.

Discussion

Results of this study indicate that although Africa’s cigarette price data landscape is populated, gaps exist, as do opportunities for expanding the scope and coverage of cigarette price data collection in the region.

Of the 324 datasets excluded in the data repository search because the information that they contained focused exclusively on smoking behaviour, 217 (67%) come from 1 of 3 self-response surveys implemented repeatedly in most African countries: STEPS,²⁴ MICS²⁵ and DHS.²⁶ Results showed that only eleven African countries⁴³,⁵⁴ have leveraged these repeated surveys that collect information on cigarette smoking behaviour to also collect data that can be used to derive cigarette prices. This suggests that the attention paid to monitoring cigarette use in Africa is not echoed in the monitoring of a key parameter that can curb its use—prices.

One reason for this may be that questions relating to cigarette prices are only collected through ‘optional modules’ in these surveys, while questions regarding cigarette smoking behaviour are collected in ‘core questionnaires’.^{24 25 55} Ensuring that information on cigarette prices is collected in these large-scale data-collection efforts will require a process of multi-sectoral collaboration that only the tobacco-control community has an obvious incentive to initiate.

The tobacco-control community should lobby governments to implement these optional modules and inform governments of (1) the importance of these data for supporting them, the local policy-makers, in their decision-making processes on matters of tobacco taxation¹⁰ and (2) the economies of scope realised through leveraging pre-existing data collection frameworks, that already collect information on cigarette smoking behaviour, to collect information on the prices that respondents pay for the cigarettes that they report smoking.⁵⁶

However, this does not represent a stand-alone solution for expanding Africa’s cigarette price information base. While most DHS, STEPS and MICS claim national representativity,⁵⁷,⁵⁹ the data collected are only representative of the populations covered in these surveys. The DHS, for example, targets women aged 15–49 and men aged 15–59.⁶⁰ Consequently, cigarette price data collected from this survey reflect only the prices paid by a nationally representative sample of people in this age cohort.

Results of this study show that no current projects that collect self-reported information on cigarette prices sample from people younger than fifteen. Yet, research using GYTS data from 22 African countries shows that 54% of cigarette smokers, aged 12–16, reported purchasing cigarettes from commercial vendors in the month preceding the survey.⁶¹ While no studies provide a direct comparison of the cigarette purchasing styles of younger adolescents and adults in African countries, international research suggests that adolescents typically purchase smaller cigarette packaging sizes than their adult counterparts.^{62 63} Research from African countries shows, further, that loose cigarettes are more commonly sold by street vendors than in formal retail stores, and that differences in the place of purchase and cigarette packaging size affect the unit cost of each cigarette for the buyer.¹⁴

Taken together, these findings indicate the need for additional data collection, outside of leveraging existing survey frameworks, to close the information gap on the prices that African youth pay for cigarettes. The expansion of efforts to collect data at retailers, if accompanied by a record of the unit of cigarette packaging, the type of retail outlet and the brand associated with each price observation, could help fill this information gap.⁶¹ This study’s dataset search process revealed that only one observational price data collection project—the ACP Project—provides this type of information for multiple cigarette packaging types, brands and retail-outlet types.³⁵

However, this project’s prices suffer the limitation that they are disproportionately sampled from Southern African countries,³⁵ where most non-ACP Project data collection already takes place. Contributing to the project’s disproportionate sampling of one African subregion is its position at a South African university, and its fieldworker recruitment strategy, which draws on registered students from African countries to collect price data in their home countries during the academic vacation period.³⁵ A further limitation of this project’s data is that the prices are not based on a nationally representative sample of cigarette retailers within countries.³⁵

Research has identified achieving adequate geographical coverage as a primary challenge for observational approaches to price data collection.⁶⁴ However, the ACP Project’s position at an African tertiary institution has not yet been leveraged to expand price data collection within countries, or across countries, through partnerships, and knowledge sharing, with universities located in different subregions on the continent.

Support for such collaboration is reflected in this study’s findings. Results indicate that prices collected through observations at retailers contain, on average, proportionately fewer outliers than self-reported information. Given this, proactivity in efforts to expand observational price data collection can improve our understanding of the extent of misreporting in the self-response surveys that currently dominate Africa’s cigarette price data landscape. The end result is a deeper insight into the cigarette prices in African countries, and the acquisition of the data required to generate evidence that can aid in securing public health gains.¹⁰

This study is subject to limitations. First, while efforts were made to sense-check the prices flagged as outliers, it is conceivable that the rules for identifying outliers may not perfectly account for local realities of what reasonably high (or low) prices may be. It is thus possible that true outliers may have been retained, and that valid observations may have been dropped, with implications for some of the study’s results. Relatedly, comparisons of the number of outliers obtained under different data collection methods could be biased by the fact that most of the observed data are from the ACP Project.

Additionally, because some surveys sampled different segments of the adult population, sample-specific biases in the prices reported may have influenced the summary statistics calculated and skewed any related analysis. Another limitation is that the included datasets provide price data collected in different years. While using constant international dollars facilitates cross-country comparisons, biases in the analysis may have resulted from the different time periods in which prices were collected.

Although some datasets located through the data repository searches were reobserved in studies found in the literature search, an additional limitation of this study is that no analysis of data ‘usage’ was conducted. Future research could examine the extent to which the freely available price data identified in this study is being used in tobacco-control research on the continent.

Conclusion

Africa’s cigarette price data landscape is populated, but gaps exist. A broader base of cigarette price information for the region could be achieved through proactive efforts by the tobacco-control community. These efforts should include lobbying governments to leverage existing surveys to collect cigarette price data and exploring options to extend the coverage of price collection at the retail level through partnerships across the region. Expanding observational price data collection efforts would address the existing information gap concerning the prices that young people pay for cigarettes and improve understanding of the degree of misreporting in self-response surveys conducted in African countries.

Data availability statement

Data are available in a public, open access repository. Data may be obtained from a third party and are not publicly available.