Kithure et al. 2021 — Heavy metals and physico-chemical parameters in Nairobi PET-bottled drinking water
This 2021 study from the University of Nairobi Department of Chemistry analysed seven anonymised brands (A–G) of commercial PET-bottled drinking water purchased from Nairobi supermarkets for seven heavy metals (Sb, Cd, Cr, Cu, Mn, Pb, Zn) by atomic absorption spectrometry and three physico-chemical parameters (pH, electrical conductivity, total dissolved solids). Sb, Cd, and Zn were below detection limit in all seven brands. Cu, Pb, Cr, and Mn were detected in subsets of brands at levels the authors compared to Kenya Bureau of Standards (KEBS), National Environment Management Authority (NEMA), and WHO drinking water limits, with Pb exceeding cited KEBS/WHO limits in two brands and Mn exceeding cited limits in one brand. The paper reports concentrations in units of “mg/g” throughout, which is physically implausible for water samples; method-section calibration standards and limits of detection are reported in mg/L–equivalent units (µg/mL), so the reported concentrations are almost certainly mg/L mis-labelled. The wiki reports the source’s verbatim numbers with this caveat documented in ## Verification notes.
Key numbers
Sample-by-sample heavy metal concentrations from Table 2 (paper labels units “mg/g”; see verification notes for the unit-discrepancy assessment):
- Antimony (Sb): BDL in all seven samples (A–G).
- Cadmium (Cd): BDL in all seven samples (A–G).
- Zinc (Zn): BDL in all seven samples (A–G).
- Copper (Cu): A 0.05 ± 0.01; B 0.02 ± 0.03; C BDL; D 0.03 ± 0.03; E BDL; F BDL; G BDL.
- Lead (Pb): A 0.003 ± 0.01; B BDL; C BDL; D BDL; E BDL; F 0.32 ± 0.19; G 0.14 ± 0.07.
- Chromium (Cr, total): A BDL; B BDL; C BDL; D BDL; E 0.66 ± 0.03; F BDL; G BDL.
- Manganese (Mn): A BDL; B BDL; C BDL; D BDL; E BDL; F BDL; G 0.93 ± 0.53.
If the reported units are actually mg/L (the most defensible reading; see verification notes), the highest detected values correspond to:
- Pb sample F: 320 ppb (± 190 ppb).
- Pb sample G: 140 ppb (± 70 ppb).
- Cr sample E: 660 ppb (± 30 ppb).
- Mn sample G: 930 ppb (± 530 ppb).
- Cu sample A: 50 ppb (± 10 ppb).
AAS method detection limits (Table 1, in µg/mL ≡ mg/L): Sb 0.0016; Cu 0.027; Cd 0.008; Cr 0.030; Zn 0.020; Pb 0.017; Mn 0.050. Calibration correlation coefficients ≥ 0.9990 for all seven analytes.
Physico-chemical parameters (Table 4, four of seven samples analysed):
- pH: A 6.72 ± 0.15; B 6.63 ± 0.10; C 7.03 ± 0.23; D 7.32 ± 0.01. (Note: paper text describes sample D as below the NEMA 6.5–8.5 minimum but the tabulated value 7.32 is in range; an internal inconsistency in the discussion of pH outliers — see verification notes.)
- Total dissolved solids (TDS, mg/L): A 26.80 ± 0.50; B 63.10 ± 0.33; C 100.7 ± 0.08; D 98.40 ± 0.19. All within NEMA/KEBS/WHO 1500 mg/L ceiling.
- Electrical conductivity (EC, µS/cm): A 74.50 ± 0.05; B 125.70 ± 0.70; C 89.40 ± 0.93; D 197.70 ± 0.02. All within NEMA 25–2500, KEBS 30–2500, WHO 26–2500 µS/cm ranges cited by the authors.
Regulatory limits cited by the authors in Table 3 (paper units “mg/g”; the wiki does not propagate these as factual regulatory values — see verification notes for the discrepancy against actual WHO/KEBS drinking-water guideline values):
- Cr — KEBS 0.05; WHO 0.06.
- Mn — KEBS 0.1; WHO 0.2.
- Pb — KEBS 0.05; WHO 0.04.
- Cd — KEBS 0.005; WHO 0.004.
- Zn — KEBS 5.0; WHO 6.0.
- Sb — KEBS — ; WHO 0.006.
- Cu — KEBS 0.1; WHO 2.00.
Methods (brief)
Sampling: seven brands of PET-bottled drinking water purchased from major Nairobi supermarkets in triplicate during 2019–2020, transported to University of Nairobi laboratories, digested and refrigerated at 4 °C pending analysis.
Digestion: 100 mL water samples digested with 10 mL aqua regia (HCl:HNO₃ 3:1, analar grade, Sigma-Aldrich) plus 1 mL perchloric acid (Fisher Scientific) in conical flasks on a hot plate at 65 °C for 2.5 h. Cooled, filtered through Whatman No. 1 papers into 50 mL volumetric flasks, diluted with double-distilled water, transferred to nitric-rinsed specimen bottles, refrigerated until AAS analysis.
Instrument: the paper names two distinct AAS instruments in different sections — the abstract identifies “Atomic Absorption Spectrometer (AAS) Spectra AA-10” while the Results section names “computerized AA-7000 Shimadzu Atomic Absorption spectrophotometer”. The wiki cannot resolve which instrument actually produced the reported data; flagged in verification notes. AAS hollow-cathode lamps and operating wavelengths reported as Sb 228.8 nm, Cd 228.8 (paper text actually lists 228.8 for Sb and 357.9 for the next position which corresponds to neither Cd nor a standard wavelength — see verification notes for the wavelength-attribution problem), Cr 324.7, Cu 279.5, Pb 218 nm. Single 100 mg/L standard stock diluted into calibration series; 1 mL concentrated HNO₃ matrix-matched to each standard.
Physico-chemical parameters: pH by glass-electrode pH-meter (50 mL aliquot, 25 °C water bath, two-point calibration with pH 4.0 and pH 10 buffers) — only four samples analysed for pH/EC/TDS, not all seven. EC measured on 50 mL aliquots after calibration against 0.005 M KCl (654 µmho cm⁻¹) at 25 °C. TDS computed from EC by multiplying by a conversion factor of 0.7.
Speciation: Cr reported as total chromium (no Cr(VI)/Cr(III) separation). No arsenic measured. No mercury measured.
Limitations the source itself does not acknowledge but should: (a) reporting units stated as “mg/g” for aqueous samples are physically implausible at the reported magnitudes (0.93 mg/g manganese in water would be ~0.1% solids); (b) pH/EC/TDS subset (n=4) is not aligned with the heavy-metal subset (n=7) and the four samples are not identified by the same letters in both tables (Table 4 labels A–D, Table 2 labels A–G); (c) only single-time-point composites per brand are reported, with no within-brand temporal or batch variability assessment; (d) no procedural blank or certified reference material results are reported.
Implications
Certification: The verbatim values reported by this source would, if accepted as mg/L, place lead in two of seven Nairobi-supermarket PET-bottled brands at 140–320 ppb — well above any drinking-water regulatory ceiling cited in the paper. The wiki carries the values as the source reports them and flags the multiple data-integrity concerns in verification notes.
Courses: Useful as a teaching case for evaluating peer-reviewed drinking-water occurrence papers — the layered data-integrity concerns (unit mis-labelling, dual instrument identification, regulatory-limit transcription errors) illustrate the kind of close reading that distinguishes A-tier from B-tier reporting even within the peer-reviewed corpus.
App: Not directly applicable; bottled drinking water as a category is downstream of ingredient-level contamination modelling.
Microbiome: Not addressed.
Verification notes
This paper exhibits several internal data-integrity issues that the wiki documents rather than attempts to resolve:
-
Concentration units inconsistent with sample matrix. All Table 2 metal concentrations and the Table 3 regulatory comparison values are labelled “mg/g”. For aqueous samples this is physically implausible — 0.93 mg/g manganese would correspond to roughly 0.1% solids in water. Method-section calibration standards are described as “100 mg/L standard solution” (mg/L) and limits of detection (Table 1) are reported as µg/mL (≡ mg/L). The most defensible reading is that all “mg/g” labels are typographic errors for “mg/L”. The wiki reports the source’s verbatim “mg/g” notation in
## Key numbersand additionally states the mg/L-equivalent interpretation as ppb values for downstream comparability, without overwriting what the source published. -
Two distinct AAS instruments named. The abstract identifies the instrument as “Atomic Absorption Spectrometer (AAS) Spectra AA-10” (a Varian instrument). The Results section identifies it as “computerized AA-7000 Shimadzu Atomic Absorption spectrophotometer” (a different vendor and platform). The body description of operation (flame, acetylene gas) is consistent with either. The wiki documents both without choosing.
-
Regulatory limits cited do not match actual WHO/KEBS drinking-water guidelines. Table 3 cites WHO Pb at “0.04 mg/g” — actual WHO drinking-water provisional guideline for Pb is 0.010 mg/L. WHO Cd is cited at “0.004 mg/g” — actual WHO is 0.003 mg/L. WHO Cu cited at “2.00 mg/g” — actual WHO is 2.0 mg/L (consistent if mg/g is read as mg/L). The wiki does not propagate the Table 3 figures as authoritative regulatory values; downstream regulation pages source WHO/KEBS limits from the agencies directly, not from this paper.
-
pH discussion contradicts the pH table. The narrative around Figure 3 states that sample D has the lowest pH and falls below the NEMA 6.5–8.5 minimum, with implications for taste and dissolved metal speciation. The tabulated pH value for sample D in Table 4 is 7.32 ± 0.01, which is within range. The lowest tabulated pH is sample B at 6.63 ± 0.10, which is within both the KEBS 6.0–9.0 and NEMA 6.5–8.5 ranges cited by the authors. The narrative appears to describe a different sample than the table reports — no sample in Table 4 actually falls below either cited pH minimum.
-
AAS wavelength listing incomplete. The methods section lists operating wavelengths “228.8 nm, 357.9 nm, 324.7 nm, 279.5 nm and 218 nm respectively” for “Sb, Cd, Cr, Cu, Mn and Pb” (six analytes named, five wavelengths given; Zn is omitted). The 357.9 nm value matches neither cadmium’s standard 228.8 nm AAS line nor any other named analyte’s standard line; either a wavelength is mis-attributed or an analyte/wavelength pair was dropped during typesetting.
-
pH/EC/TDS sample set is not aligned with heavy-metals sample set. Table 4 reports physico-chemical parameters for four samples labelled A–D, while Table 2 reports heavy metals for seven samples labelled A–G. The paper does not state whether Table 4 samples A–D correspond to Table 2 samples A–D, or whether they are a different subset relabelled.
-
Brand-firewall posture is compliant. The paper explicitly anonymises the seven brands as Sample A–G “for privacy purposes” and the wiki preserves that anonymisation. No brand names appear in the source.
-
Procedural blanks and reference materials not reported. No procedural-blank results, certified reference material (CRM) recoveries, or spike-recovery data are reported. Without these, the wiki cannot independently verify whether the reported Pb concentrations of 320 ppb (sample F) and 140 ppb (sample G) — if read as mg/L — reflect actual product contamination, contamination introduced during 100 mL × triplicate aqua regia/HClO₄ digestion, or instrument drift.
-
Audit application (2026-05-30). Audit subagent (general-purpose, fresh context) flagged three ⚠ findings, all verified and applied: (a) Check 1 — corrected verification note #4 which had incorrectly stated sample B’s pH of 6.63 was below the NEMA 6.5–8.5 minimum (it is within range); (b) Check 5 — removed cross-literature comparison from
## Implications(“broadly inconsistent with the global bottled-water literature”) because cross-source comparison belongs in the Part 9 synthesis pass, not the source page; (c) Check 5 — removed synthesis-direction instruction (“downstream synthesis should weight this source with substantial caveat or treat it as discovery-only”) for the same reason. The audit also revised note #8 to drop the same cross-literature claim, retaining only verifiable arithmetic on this paper’s reported values and the procedural-blank gap.
The wiki reports the source’s findings as published while documenting these concerns; the page is B-tier in evidence-fitness terms because of the cumulative data-integrity issues, not because the source is non-peer-reviewed.
Wiki pages updated on ingest
Page history
The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.