Table 2 Privatisation-interrupted time series analysis
Author and interventionStudy designReported outcomes
Evans34 UKPrivatisation of UK rail industryChange in public investment not reported.Evaluation covers 1967–2002. Intervention occurred in 1996ITS of routinely collected data on fatalities resulting from train crashes—that is, collisions, derailments and the run-overs—on national rail system (range 11–119 per 5 years). Note: Data exclude the larger number of fatalities that result from people being run over on train linesMethods appraisal*: 1, 4, 5, 6, 7, 8Author states that no statistically significant change in fatalities from train crashes occurred after privatisation, but does not report details of statistical tests (ie, no CI or p values reported) (see fig 3)
Evans,35 UKDeregulation and privatisation of buses in UK, excluding London. Deregulation of buses in Scotland. Change in public investment not reported.Evaluation covers 1981–91. Intervention occurred in 1986ITS of routinely collected data on injuries (all severities) involving buses outside London, including non-bus-users injured in bus accidents (n (raw figures) = 9078–10 552 per annum). Data taken from the national road accident databaseMethods appraisal*: 2, 5,6, 7, 8Annual bus injuries (ie, number of people injured) decreased per 100 M vehicle km and increased per 100 M passenger journeys, reflecting an increase in buses, but a fall in passenger numbers. Author states statistically significant increase in the number of buses involved in multivehicle injury accidents (7.5%) and a decrease (5.6%) in the number of buses involved in single-vehicle injury accidents post-1987: no CI or p values reported (see fig 4)
Astrop,36 UKDeregulation and privatisation of buses in UK, excluding London. Deregulation of buses in Scotland.Change in public investment not reportedEvaluation covers 1979–89. Intervention occurred in 1986ITS of routinely collected data on number. of injury accidents (all severities) involving buses (n (raw figures) = approx 12 000–15 000 per annum). Data taken from the national road accident database: estimated from charts. Includes London, which was not exposed to interventionMethods appraisal*: 2, 5, 6, 8No statistical tests reported. Overall decline in incidence of bus injury accidents (ie, number of incidents causing injury to one or more people) per 100 M km before and after privatisation. Small increases in injury accidents per 100 M vehicle km immediately before and 3 years after intervention (see fig 5)
White,37 UKDeregulation and privatisation of buses in UK, excluding London. Deregulation of buses in Scotland. Change in public investment not reported.Evaluation covers 1966–90Intervention occurred in 1986ITS of routinely collected data on fatal/serious injury accidents involving buses in UK (n (raw figures) = approx 750–2100 per annum). Data taken from the national road accident database: estimated from charts. Includes London, which was not exposed to interventionMethods appraisal*: 2, 5, 6, 8Authors state that no “significant” (p value not given) change in fatal/serious injuries (ie, number of incidents involving buses causing fatal/serious injury to one or more people) occurred per 100 M passenger km afterintervention (see fig 6)
National Economic Research Associates,38 UKPrivatisation and deregulation of British Gas. Change in public investment not reported.Evaluation period 1978–95. Intervention occurred in 1986ITS of routinely collected data on gas explosions (range: 25–106 per annum) and fatalities (employees and public) due to gas explosions (range: 3–32 per annum). Data provided by employersMethods appraisal*: 1,2,5,7No statistical tests reported. No obvious change in trends on gas explosions and explosion fatalities associated with the company after privatisation. Apparent increase in the incidence of explosions and fatalities in the year before the intervention (see fig 7)
National Economic Research Associates,38 UKPrivatisation of most of the UK’s water industry. Change in public investment not reported.Evaluation covers 1986–95. Intervention occurred in 1989ITS of routinely collected data on major injuries to employees (range: 89–183 per 100 000 employees annum). Data supplied by employersMethods appraisal*: 1, 2, 6No statistical tests reported. No obvious change in incidence of non-fatal major injuries per 100 000 employees following privatisation. Apparent decrease in injuries in the year before the intervention (see fig 8)
National Economic Research Associates,38 UKPrivatisation of most of the UK’s electricity industry (excluding certain parts, notably nuclear power production). Change in public investment not reported.Evaluation covers 1983–95. Intervention occurred in 1990–1ITS of routinely collected data on major injuries to employees (range 94–249 injuries; rate of 92–174 per 100 000 employees per annum). All injuries for general public involving electricity supply industry (range 288–494 per annum). Data supplied from Department of Trade and Industry and Health and Safety ExecutiveMethods appraisal*: 1, 2, 6No statistical tests reported. No obvious change in trends on injuries to public and injuries per 100 000 employees following privatisation. Note there was a 25% decrease in the number of employees during the 3 years following privatisation (see fig 9)
HM Inspectorate of Mines,39 UKPrivatisation of 31 coal mines. Change in public investment not reported.Evaluation covers 1989–95. Interventions occurred 1993–5 (primarily 1994)ITS of routinely collected data on workplace injuries reported by employers to the Health and Safety Executive (range 270–5358 per annum). Approx no. of employees ranges from 69 900 (in 1989/1990) to 12 000 (1994, including contractors)Methods appraisal*: 6, 7No statistical tests reported. No obvious post-privatisation changes observable due to lack of data postintervention. Apparent increase in the incidence of major and fatal injuries per 1000 employees in the year before the intervention. Mass redundancies before and after privatisation (see fig 10)
  • ITS, interrupted time series.

  • *Methods appraisal: 1 = clearly defined intervention point; 2 = ⩽3 data points before and after intervention; 3 = no concurrent interventions; 4 = intervention unlikely to affect data collection; 5 = objective/blinded outcome measurement; 6 = reliable/accurate outcome measurement; 7 = datasets describe ⩽80% of participants; 8 = rational explanation for intervention effect; 9 = rationale for the number and spacing of data points described; 10 = appropriate time series analysis.