Internal protocol
Frozen
Frozen 8 Sept 2026, 11:17 am
Prospective study protocol · v1.0.0Pilot / method-development period
Spatiotemporal heterogeneity of ambient PM2.5 across Delhi-NCR: a prospective longitudinal observational study
A frozen study plan for the primary 2026–27 analysis. The early archive remains visible for method development, but it is explicitly excluded from the prespecified primary analysis period.
External registry
Prepared · not submitted
OSF registration still pending
Primary window
1 Oct 2026
through 31 Mar 2027 · Delhi local time
Baseline sites
Pending
Automatically freezes after primary start
Primary research question
How much persistent spatial heterogeneity in hourly ambient PM2.5 exists across Delhi-NCR after removing the concurrent network-wide temporal signal?
Primary outcome: Hourly PM2.5 concentration (µg/m³) at the physical monitoring site × Delhi-local hour level. Campus-linked comparisons remain a secondary public-facing application rather than the study's primary scientific endpoint.
Prospective separation
Pilot data are not primary-study data.
Pilot
6 Sept 2026 → 30 Sept 2026
Method development, source validation, QC design and exploratory inspection. Excluded from the primary prospective estimand.
Primary
1 Oct 2026 → 31 Mar 2027
Prespecified analysis window. It ends on schedule regardless of observed effect sizes; routine collection may continue afterward.
External registration
Prepared for OSF registration.
The internal database protocol is already immutable. That is useful for provenance, but it is not a substitute for an independent registry. The OSF preregistration should be submitted before the primary period starts.
Primary estimand
Remove the shared hourly signal before comparing place.
Rit = PM2.5it − medianj(PM2.5jt)
Site-hour residual
R_it = PM2.5_it - median_j(PM2.5_jt), where j indexes eligible physical monitoring sites reporting at hour t.
Site effect
For each eligible physical site i, the primary site-level effect is the median R_it across all included primary-period hours for that site.
Network heterogeneity
Primary network heterogeneity is summarized by the interquartile range and the 90th–10th percentile range of site-level effects across eligible physical sites.
Uncertainty
95% percentile confidence intervals from a local-calendar-day block bootstrap with 1,000 resamples and deterministic random seed 20261001. Confidence intervals are not reported until at least 30 represented primary-period days are available.
Eligibility
Define the denominator before looking at results.
Baseline cohort
At the primary-period start, baseline eligible physical sites are those marked active in the station registry and with at least one valid PM2.5 observation during the preceding 7 Delhi-local calendar days. The resulting baseline cohort is frozen before primary analysis.
Network hour
A primary network hour is included when at least 50% of baseline eligible physical sites have a valid PM2.5 value in that hour.
Site inclusion
A physical site is included in the primary site-level heterogeneity summary when it has a valid value in at least 70% of included primary network hours.
Missingness
Missing observations remain missing. No temporal interpolation, carry-forward, or substitution from neighboring monitors is permitted.
Quality control
No discretionary cleaning after the result appears.
Critical flags
Observations carrying a critical, observation-specific QC flag are excluded from the primary analysis and retained in the raw/provenance archive.
Review flags
Review-flagged observations remain in the primary analysis unless the flag itself indicates an invalid normalized value. A sensitivity analysis excludes review-flagged observations.
Outliers
No discretionary concentration outlier deletion is permitted. High but valid source values remain unless excluded by a prespecified QC rule.
Station identity
Station renames, operator changes, and coordinate shifts are retained as audit events. A material identity change may require a protocol amendment before combining pre- and post-change observations as one site.
Secondary analyses
Useful, but not allowed to become the primary endpoint after the fact.
- 1PM10 analyzed analogously as a secondary pollutant outcome.
- 2Delhi-local diurnal PM2.5 summaries using network-hour medians grouped by local clock hour.
- 3Station-to-station Pearson correlation on shared hourly PM2.5 observations, with same-physical-site pairs excluded from independence-sensitive summaries.
- 4Paired campus-reference monitor differences restricted to shared observed hours; campus analyses are secondary applications, not the primary scientific estimand.
- 5Delhi-only versus wider Delhi-NCR descriptive comparisons.
Sensitivity analyses
Results should survive reasonable alternate rules.
- 1Exclude all review-flagged observations.
- 2Repeat site inclusion using 60% and 80% completeness thresholds instead of 70%.
- 3Restrict geography to Delhi only.
- 4Use a deterministic single-feed-per-physical-site representation instead of within-site hourly median aggregation.
- 5Repeat heterogeneity summaries using mean site-hour residuals in place of medians.
Weather-adjusted extension
The model is still locked.
No weather-adjusted model will be fitted until at least 30 represented air-data days and 21 days of protocol-eligible locked-model meteorology exist.
The exact model formula, transformations, diagnostics, and inferential procedure must be frozen in a dated protocol amendment before the first weather-adjusted model is fitted. Weather-adjusted results are not part of the v1.0.0 primary estimand.
Interpretation boundary
Descriptive first.
The primary estimand is descriptive and interval-based; no binary null-hypothesis significance claim is designated as the study's primary conclusion.
The protocol does not infer causal local emission sources, personal exposure, or health effects from ambient spatial differences.
Ethics & scope
Environmental data, not patient data.
This protocol uses publicly available area-level environmental measurements and contains no individual-level patient or participant data. No claim of ethics exemption is made; any institutional determination required for submission will be sought. Any future linkage to individual or non-public health data requires separate ethics, privacy, and data-governance review.
Protocol identity
DAH-PM25-2026-27-v1
SHA-256 8e9fc3991c4fb49845f49aadcc1660ab79767b41ed8eca4a71537a903381c351